├── BT_CODE
    ├── README.md
    ├── data_csv
    │   ├── BraTS.csv
    │   ├── OASIS.csv
    │   └── brats
    │   │   ├── test_0.csv
    │   │   ├── test_1.csv
    │   │   └── test_2.csv
    ├── dataload
    │   ├── augmentations.py
    │   └── dataload.py
    ├── evaluate.py
    ├── model
    │   └── unet3d.py
    ├── requirements.txt
    ├── resources
    │   ├── geodesic_polyhedron.nii.gz
    │   └── geodesic_polyhedron.vtp
    ├── train.py
    └── utils
    │   ├── __pycache__
    │       ├── loss.cpython-36.pyc
    │       ├── prelin_sphere.cpython-36.pyc
    │       └── vtk_itk.cpython-36.pyc
    │   ├── loss.py
    │   ├── metrics.py
    │   ├── prelin_sphere.py
    │   └── vtk_itk.py
└── README.md


/BT_CODE/README.md:
--------------------------------------------------------------------------------
 1 | We provide the official <b>PyTorch</b> implementation of Our Layer Decomposition
 2 | 
 3 | 
 4 | Take the implementation on brain tumor data as example:
 5 | ### 1. Install environment
 6 | ```bash
 7 | $ cd BT_CODE
 8 | $ pip install -r requirements.txt
 9 | ```
10 | ### 2. Preprocessing the pretraining dataset
11 | All volumes are already co-registered, skull stripped and resampled into a 1-mm isotropic image in atlas space.
12 | 
13 | ### 3. Train the layer decomposition model
14 | ```bash
15 | python train.py 
16 | ```
17 | 
18 | 


--------------------------------------------------------------------------------
/BT_CODE/data_csv/brats/test_0.csv:
--------------------------------------------------------------------------------
 1 | Dir_path,Flair_path,T1_path,T1ce_path,T2_path,Seg_path,HGG/LGG
 2 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_11_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_11_1/Brats18_2013_11_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_11_1/Brats18_2013_11_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_11_1/Brats18_2013_11_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_11_1/Brats18_2013_11_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_11_1/Brats18_2013_11_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_11_1/Brats18_2013_11_1_seg.nii.gz
 3 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_14_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_14_1/Brats18_2013_14_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_14_1/Brats18_2013_14_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_14_1/Brats18_2013_14_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_14_1/Brats18_2013_14_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_14_1/Brats18_2013_14_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_14_1/Brats18_2013_14_1_seg.nii.gz
 4 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_20_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_20_1/Brats18_2013_20_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_20_1/Brats18_2013_20_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_20_1/Brats18_2013_20_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_20_1/Brats18_2013_20_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_20_1/Brats18_2013_20_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_20_1/Brats18_2013_20_1_seg.nii.gz
 5 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_25_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_25_1/Brats18_2013_25_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_25_1/Brats18_2013_25_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_25_1/Brats18_2013_25_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_25_1/Brats18_2013_25_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_25_1/Brats18_2013_25_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_25_1/Brats18_2013_25_1_seg.nii.gz
 6 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_3_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_3_1/Brats18_2013_3_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_3_1/Brats18_2013_3_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_3_1/Brats18_2013_3_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_3_1/Brats18_2013_3_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_3_1/Brats18_2013_3_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_3_1/Brats18_2013_3_1_seg.nii.gz
 7 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_7_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_7_1/Brats18_2013_7_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_7_1/Brats18_2013_7_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_7_1/Brats18_2013_7_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_7_1/Brats18_2013_7_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_7_1/Brats18_2013_7_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_7_1/Brats18_2013_7_1_seg.nii.gz
 8 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAP_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAP_1/Brats18_CBICA_AAP_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAP_1/Brats18_CBICA_AAP_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAP_1/Brats18_CBICA_AAP_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAP_1/Brats18_CBICA_AAP_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAP_1/Brats18_CBICA_AAP_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAP_1/Brats18_CBICA_AAP_1_seg.nii.gz
 9 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABN_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABN_1/Brats18_CBICA_ABN_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABN_1/Brats18_CBICA_ABN_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABN_1/Brats18_CBICA_ABN_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABN_1/Brats18_CBICA_ABN_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABN_1/Brats18_CBICA_ABN_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABN_1/Brats18_CBICA_ABN_1_seg.nii.gz
10 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALU_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALU_1/Brats18_CBICA_ALU_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALU_1/Brats18_CBICA_ALU_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALU_1/Brats18_CBICA_ALU_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALU_1/Brats18_CBICA_ALU_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALU_1/Brats18_CBICA_ALU_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALU_1/Brats18_CBICA_ALU_1_seg.nii.gz
11 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AMH_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AMH_1/Brats18_CBICA_AMH_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AMH_1/Brats18_CBICA_AMH_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AMH_1/Brats18_CBICA_AMH_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AMH_1/Brats18_CBICA_AMH_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AMH_1/Brats18_CBICA_AMH_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AMH_1/Brats18_CBICA_AMH_1_seg.nii.gz
12 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANZ_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANZ_1/Brats18_CBICA_ANZ_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANZ_1/Brats18_CBICA_ANZ_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANZ_1/Brats18_CBICA_ANZ_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANZ_1/Brats18_CBICA_ANZ_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANZ_1/Brats18_CBICA_ANZ_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANZ_1/Brats18_CBICA_ANZ_1_seg.nii.gz
13 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOP_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOP_1/Brats18_CBICA_AOP_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOP_1/Brats18_CBICA_AOP_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOP_1/Brats18_CBICA_AOP_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOP_1/Brats18_CBICA_AOP_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOP_1/Brats18_CBICA_AOP_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOP_1/Brats18_CBICA_AOP_1_seg.nii.gz
14 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APZ_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APZ_1/Brats18_CBICA_APZ_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APZ_1/Brats18_CBICA_APZ_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APZ_1/Brats18_CBICA_APZ_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APZ_1/Brats18_CBICA_APZ_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APZ_1/Brats18_CBICA_APZ_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APZ_1/Brats18_CBICA_APZ_1_seg.nii.gz
15 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQG_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQG_1/Brats18_CBICA_AQG_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQG_1/Brats18_CBICA_AQG_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQG_1/Brats18_CBICA_AQG_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQG_1/Brats18_CBICA_AQG_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQG_1/Brats18_CBICA_AQG_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQG_1/Brats18_CBICA_AQG_1_seg.nii.gz
16 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQP_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQP_1/Brats18_CBICA_AQP_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQP_1/Brats18_CBICA_AQP_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQP_1/Brats18_CBICA_AQP_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQP_1/Brats18_CBICA_AQP_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQP_1/Brats18_CBICA_AQP_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQP_1/Brats18_CBICA_AQP_1_seg.nii.gz
17 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQU_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQU_1/Brats18_CBICA_AQU_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQU_1/Brats18_CBICA_AQU_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQU_1/Brats18_CBICA_AQU_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQU_1/Brats18_CBICA_AQU_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQU_1/Brats18_CBICA_AQU_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQU_1/Brats18_CBICA_AQU_1_seg.nii.gz
18 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARF_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARF_1/Brats18_CBICA_ARF_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARF_1/Brats18_CBICA_ARF_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARF_1/Brats18_CBICA_ARF_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARF_1/Brats18_CBICA_ARF_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARF_1/Brats18_CBICA_ARF_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARF_1/Brats18_CBICA_ARF_1_seg.nii.gz
19 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASA_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASA_1/Brats18_CBICA_ASA_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASA_1/Brats18_CBICA_ASA_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASA_1/Brats18_CBICA_ASA_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASA_1/Brats18_CBICA_ASA_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASA_1/Brats18_CBICA_ASA_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASA_1/Brats18_CBICA_ASA_1_seg.nii.gz
20 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASK_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASK_1/Brats18_CBICA_ASK_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASK_1/Brats18_CBICA_ASK_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASK_1/Brats18_CBICA_ASK_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASK_1/Brats18_CBICA_ASK_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASK_1/Brats18_CBICA_ASK_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASK_1/Brats18_CBICA_ASK_1_seg.nii.gz
21 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASV_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASV_1/Brats18_CBICA_ASV_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASV_1/Brats18_CBICA_ASV_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASV_1/Brats18_CBICA_ASV_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASV_1/Brats18_CBICA_ASV_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASV_1/Brats18_CBICA_ASV_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASV_1/Brats18_CBICA_ASV_1_seg.nii.gz
22 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATD_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATD_1/Brats18_CBICA_ATD_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATD_1/Brats18_CBICA_ATD_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATD_1/Brats18_CBICA_ATD_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATD_1/Brats18_CBICA_ATD_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATD_1/Brats18_CBICA_ATD_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATD_1/Brats18_CBICA_ATD_1_seg.nii.gz
23 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATV_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATV_1/Brats18_CBICA_ATV_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATV_1/Brats18_CBICA_ATV_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATV_1/Brats18_CBICA_ATV_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATV_1/Brats18_CBICA_ATV_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATV_1/Brats18_CBICA_ATV_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATV_1/Brats18_CBICA_ATV_1_seg.nii.gz
24 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUR_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUR_1/Brats18_CBICA_AUR_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUR_1/Brats18_CBICA_AUR_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUR_1/Brats18_CBICA_AUR_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUR_1/Brats18_CBICA_AUR_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUR_1/Brats18_CBICA_AUR_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUR_1/Brats18_CBICA_AUR_1_seg.nii.gz
25 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWG_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWG_1/Brats18_CBICA_AWG_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWG_1/Brats18_CBICA_AWG_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWG_1/Brats18_CBICA_AWG_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWG_1/Brats18_CBICA_AWG_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWG_1/Brats18_CBICA_AWG_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWG_1/Brats18_CBICA_AWG_1_seg.nii.gz
26 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXL_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXL_1/Brats18_CBICA_AXL_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXL_1/Brats18_CBICA_AXL_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXL_1/Brats18_CBICA_AXL_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXL_1/Brats18_CBICA_AXL_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXL_1/Brats18_CBICA_AXL_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXL_1/Brats18_CBICA_AXL_1_seg.nii.gz
27 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXO_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXO_1/Brats18_CBICA_AXO_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXO_1/Brats18_CBICA_AXO_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXO_1/Brats18_CBICA_AXO_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXO_1/Brats18_CBICA_AXO_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXO_1/Brats18_CBICA_AXO_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXO_1/Brats18_CBICA_AXO_1_seg.nii.gz
28 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYI_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYI_1/Brats18_CBICA_AYI_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYI_1/Brats18_CBICA_AYI_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYI_1/Brats18_CBICA_AYI_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYI_1/Brats18_CBICA_AYI_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYI_1/Brats18_CBICA_AYI_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYI_1/Brats18_CBICA_AYI_1_seg.nii.gz
29 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AZH_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AZH_1/Brats18_CBICA_AZH_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AZH_1/Brats18_CBICA_AZH_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AZH_1/Brats18_CBICA_AZH_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AZH_1/Brats18_CBICA_AZH_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AZH_1/Brats18_CBICA_AZH_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AZH_1/Brats18_CBICA_AZH_1_seg.nii.gz
30 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHK_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHK_1/Brats18_CBICA_BHK_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHK_1/Brats18_CBICA_BHK_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHK_1/Brats18_CBICA_BHK_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHK_1/Brats18_CBICA_BHK_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHK_1/Brats18_CBICA_BHK_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHK_1/Brats18_CBICA_BHK_1_seg.nii.gz
31 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_147_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_147_1/Brats18_TCIA01_147_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_147_1/Brats18_TCIA01_147_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_147_1/Brats18_TCIA01_147_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_147_1/Brats18_TCIA01_147_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_147_1/Brats18_TCIA01_147_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_147_1/Brats18_TCIA01_147_1_seg.nii.gz
32 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_190_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_190_1/Brats18_TCIA01_190_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_190_1/Brats18_TCIA01_190_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_190_1/Brats18_TCIA01_190_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_190_1/Brats18_TCIA01_190_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_190_1/Brats18_TCIA01_190_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_190_1/Brats18_TCIA01_190_1_seg.nii.gz
33 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_231_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_231_1/Brats18_TCIA01_231_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_231_1/Brats18_TCIA01_231_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_231_1/Brats18_TCIA01_231_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_231_1/Brats18_TCIA01_231_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_231_1/Brats18_TCIA01_231_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_231_1/Brats18_TCIA01_231_1_seg.nii.gz
34 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_390_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_390_1/Brats18_TCIA01_390_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_390_1/Brats18_TCIA01_390_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_390_1/Brats18_TCIA01_390_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_390_1/Brats18_TCIA01_390_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_390_1/Brats18_TCIA01_390_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_390_1/Brats18_TCIA01_390_1_seg.nii.gz
35 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_412_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_412_1/Brats18_TCIA01_412_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_412_1/Brats18_TCIA01_412_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_412_1/Brats18_TCIA01_412_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_412_1/Brats18_TCIA01_412_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_412_1/Brats18_TCIA01_412_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_412_1/Brats18_TCIA01_412_1_seg.nii.gz
36 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_460_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_460_1/Brats18_TCIA01_460_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_460_1/Brats18_TCIA01_460_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_460_1/Brats18_TCIA01_460_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_460_1/Brats18_TCIA01_460_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_460_1/Brats18_TCIA01_460_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_460_1/Brats18_TCIA01_460_1_seg.nii.gz
37 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_135_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_135_1/Brats18_TCIA02_135_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_135_1/Brats18_TCIA02_135_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_135_1/Brats18_TCIA02_135_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_135_1/Brats18_TCIA02_135_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_135_1/Brats18_TCIA02_135_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_135_1/Brats18_TCIA02_135_1_seg.nii.gz
38 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_179_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_179_1/Brats18_TCIA02_179_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_179_1/Brats18_TCIA02_179_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_179_1/Brats18_TCIA02_179_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_179_1/Brats18_TCIA02_179_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_179_1/Brats18_TCIA02_179_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_179_1/Brats18_TCIA02_179_1_seg.nii.gz
39 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_222_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_222_1/Brats18_TCIA02_222_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_222_1/Brats18_TCIA02_222_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_222_1/Brats18_TCIA02_222_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_222_1/Brats18_TCIA02_222_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_222_1/Brats18_TCIA02_222_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_222_1/Brats18_TCIA02_222_1_seg.nii.gz
40 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_290_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_290_1/Brats18_TCIA02_290_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_290_1/Brats18_TCIA02_290_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_290_1/Brats18_TCIA02_290_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_290_1/Brats18_TCIA02_290_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_290_1/Brats18_TCIA02_290_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_290_1/Brats18_TCIA02_290_1_seg.nii.gz
41 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_321_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_321_1/Brats18_TCIA02_321_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_321_1/Brats18_TCIA02_321_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_321_1/Brats18_TCIA02_321_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_321_1/Brats18_TCIA02_321_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_321_1/Brats18_TCIA02_321_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_321_1/Brats18_TCIA02_321_1_seg.nii.gz
42 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_370_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_370_1/Brats18_TCIA02_370_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_370_1/Brats18_TCIA02_370_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_370_1/Brats18_TCIA02_370_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_370_1/Brats18_TCIA02_370_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_370_1/Brats18_TCIA02_370_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_370_1/Brats18_TCIA02_370_1_seg.nii.gz
43 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_394_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_394_1/Brats18_TCIA02_394_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_394_1/Brats18_TCIA02_394_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_394_1/Brats18_TCIA02_394_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_394_1/Brats18_TCIA02_394_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_394_1/Brats18_TCIA02_394_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_394_1/Brats18_TCIA02_394_1_seg.nii.gz
44 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_473_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_473_1/Brats18_TCIA02_473_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_473_1/Brats18_TCIA02_473_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_473_1/Brats18_TCIA02_473_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_473_1/Brats18_TCIA02_473_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_473_1/Brats18_TCIA02_473_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_473_1/Brats18_TCIA02_473_1_seg.nii.gz
45 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_607_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_607_1/Brats18_TCIA02_607_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_607_1/Brats18_TCIA02_607_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_607_1/Brats18_TCIA02_607_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_607_1/Brats18_TCIA02_607_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_607_1/Brats18_TCIA02_607_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_607_1/Brats18_TCIA02_607_1_seg.nii.gz
46 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_138_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_138_1/Brats18_TCIA03_138_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_138_1/Brats18_TCIA03_138_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_138_1/Brats18_TCIA03_138_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_138_1/Brats18_TCIA03_138_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_138_1/Brats18_TCIA03_138_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_138_1/Brats18_TCIA03_138_1_seg.nii.gz
47 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_265_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_265_1/Brats18_TCIA03_265_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_265_1/Brats18_TCIA03_265_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_265_1/Brats18_TCIA03_265_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_265_1/Brats18_TCIA03_265_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_265_1/Brats18_TCIA03_265_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_265_1/Brats18_TCIA03_265_1_seg.nii.gz
48 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_419_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_419_1/Brats18_TCIA03_419_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_419_1/Brats18_TCIA03_419_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_419_1/Brats18_TCIA03_419_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_419_1/Brats18_TCIA03_419_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_419_1/Brats18_TCIA03_419_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_419_1/Brats18_TCIA03_419_1_seg.nii.gz
49 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_149_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_149_1/Brats18_TCIA04_149_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_149_1/Brats18_TCIA04_149_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_149_1/Brats18_TCIA04_149_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_149_1/Brats18_TCIA04_149_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_149_1/Brats18_TCIA04_149_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_149_1/Brats18_TCIA04_149_1_seg.nii.gz
50 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_361_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_361_1/Brats18_TCIA04_361_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_361_1/Brats18_TCIA04_361_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_361_1/Brats18_TCIA04_361_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_361_1/Brats18_TCIA04_361_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_361_1/Brats18_TCIA04_361_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_361_1/Brats18_TCIA04_361_1_seg.nii.gz
51 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_277_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_277_1/Brats18_TCIA05_277_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_277_1/Brats18_TCIA05_277_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_277_1/Brats18_TCIA05_277_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_277_1/Brats18_TCIA05_277_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_277_1/Brats18_TCIA05_277_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_277_1/Brats18_TCIA05_277_1_seg.nii.gz
52 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_165_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_165_1/Brats18_TCIA06_165_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_165_1/Brats18_TCIA06_165_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_165_1/Brats18_TCIA06_165_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_165_1/Brats18_TCIA06_165_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_165_1/Brats18_TCIA06_165_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_165_1/Brats18_TCIA06_165_1_seg.nii.gz
53 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_332_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_332_1/Brats18_TCIA06_332_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_332_1/Brats18_TCIA06_332_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_332_1/Brats18_TCIA06_332_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_332_1/Brats18_TCIA06_332_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_332_1/Brats18_TCIA06_332_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_332_1/Brats18_TCIA06_332_1_seg.nii.gz
54 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_105_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_105_1/Brats18_TCIA08_105_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_105_1/Brats18_TCIA08_105_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_105_1/Brats18_TCIA08_105_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_105_1/Brats18_TCIA08_105_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_105_1/Brats18_TCIA08_105_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_105_1/Brats18_TCIA08_105_1_seg.nii.gz
55 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_167_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_167_1/Brats18_TCIA08_167_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_167_1/Brats18_TCIA08_167_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_167_1/Brats18_TCIA08_167_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_167_1/Brats18_TCIA08_167_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_167_1/Brats18_TCIA08_167_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_167_1/Brats18_TCIA08_167_1_seg.nii.gz
56 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_242_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_242_1/Brats18_TCIA08_242_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_242_1/Brats18_TCIA08_242_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_242_1/Brats18_TCIA08_242_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_242_1/Brats18_TCIA08_242_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_242_1/Brats18_TCIA08_242_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_242_1/Brats18_TCIA08_242_1_seg.nii.gz
57 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_406_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_406_1/Brats18_TCIA08_406_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_406_1/Brats18_TCIA08_406_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_406_1/Brats18_TCIA08_406_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_406_1/Brats18_TCIA08_406_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_406_1/Brats18_TCIA08_406_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_406_1/Brats18_TCIA08_406_1_seg.nii.gz
58 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_15_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_15_1/Brats18_2013_15_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_15_1/Brats18_2013_15_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_15_1/Brats18_2013_15_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_15_1/Brats18_2013_15_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_15_1/Brats18_2013_15_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_15_1/Brats18_2013_15_1_seg.nii.gz
59 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_24_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_24_1/Brats18_2013_24_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_24_1/Brats18_2013_24_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_24_1/Brats18_2013_24_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_24_1/Brats18_2013_24_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_24_1/Brats18_2013_24_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_24_1/Brats18_2013_24_1_seg.nii.gz
60 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_8_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_8_1/Brats18_2013_8_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_8_1/Brats18_2013_8_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_8_1/Brats18_2013_8_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_8_1/Brats18_2013_8_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_8_1/Brats18_2013_8_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_8_1/Brats18_2013_8_1_seg.nii.gz
61 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_254_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_254_1/Brats18_TCIA09_254_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_254_1/Brats18_TCIA09_254_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_254_1/Brats18_TCIA09_254_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_254_1/Brats18_TCIA09_254_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_254_1/Brats18_TCIA09_254_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_254_1/Brats18_TCIA09_254_1_seg.nii.gz
62 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_428_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_428_1/Brats18_TCIA09_428_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_428_1/Brats18_TCIA09_428_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_428_1/Brats18_TCIA09_428_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_428_1/Brats18_TCIA09_428_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_428_1/Brats18_TCIA09_428_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_428_1/Brats18_TCIA09_428_1_seg.nii.gz
63 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_493_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_493_1/Brats18_TCIA09_493_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_493_1/Brats18_TCIA09_493_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_493_1/Brats18_TCIA09_493_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_493_1/Brats18_TCIA09_493_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_493_1/Brats18_TCIA09_493_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_493_1/Brats18_TCIA09_493_1_seg.nii.gz
64 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_130_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_130_1/Brats18_TCIA10_130_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_130_1/Brats18_TCIA10_130_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_130_1/Brats18_TCIA10_130_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_130_1/Brats18_TCIA10_130_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_130_1/Brats18_TCIA10_130_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_130_1/Brats18_TCIA10_130_1_seg.nii.gz
65 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_241_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_241_1/Brats18_TCIA10_241_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_241_1/Brats18_TCIA10_241_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_241_1/Brats18_TCIA10_241_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_241_1/Brats18_TCIA10_241_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_241_1/Brats18_TCIA10_241_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_241_1/Brats18_TCIA10_241_1_seg.nii.gz
66 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_282_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_282_1/Brats18_TCIA10_282_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_282_1/Brats18_TCIA10_282_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_282_1/Brats18_TCIA10_282_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_282_1/Brats18_TCIA10_282_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_282_1/Brats18_TCIA10_282_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_282_1/Brats18_TCIA10_282_1_seg.nii.gz
67 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_310_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_310_1/Brats18_TCIA10_310_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_310_1/Brats18_TCIA10_310_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_310_1/Brats18_TCIA10_310_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_310_1/Brats18_TCIA10_310_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_310_1/Brats18_TCIA10_310_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_310_1/Brats18_TCIA10_310_1_seg.nii.gz
68 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_351_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_351_1/Brats18_TCIA10_351_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_351_1/Brats18_TCIA10_351_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_351_1/Brats18_TCIA10_351_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_351_1/Brats18_TCIA10_351_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_351_1/Brats18_TCIA10_351_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_351_1/Brats18_TCIA10_351_1_seg.nii.gz
69 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_410_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_410_1/Brats18_TCIA10_410_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_410_1/Brats18_TCIA10_410_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_410_1/Brats18_TCIA10_410_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_410_1/Brats18_TCIA10_410_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_410_1/Brats18_TCIA10_410_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_410_1/Brats18_TCIA10_410_1_seg.nii.gz
70 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_449_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_449_1/Brats18_TCIA10_449_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_449_1/Brats18_TCIA10_449_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_449_1/Brats18_TCIA10_449_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_449_1/Brats18_TCIA10_449_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_449_1/Brats18_TCIA10_449_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_449_1/Brats18_TCIA10_449_1_seg.nii.gz
71 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_628_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_628_1/Brats18_TCIA10_628_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_628_1/Brats18_TCIA10_628_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_628_1/Brats18_TCIA10_628_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_628_1/Brats18_TCIA10_628_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_628_1/Brats18_TCIA10_628_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_628_1/Brats18_TCIA10_628_1_seg.nii.gz
72 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_639_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_639_1/Brats18_TCIA10_639_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_639_1/Brats18_TCIA10_639_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_639_1/Brats18_TCIA10_639_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_639_1/Brats18_TCIA10_639_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_639_1/Brats18_TCIA10_639_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_639_1/Brats18_TCIA10_639_1_seg.nii.gz
73 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_249_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_249_1/Brats18_TCIA12_249_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_249_1/Brats18_TCIA12_249_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_249_1/Brats18_TCIA12_249_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_249_1/Brats18_TCIA12_249_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_249_1/Brats18_TCIA12_249_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_249_1/Brats18_TCIA12_249_1_seg.nii.gz
74 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_480_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_480_1/Brats18_TCIA12_480_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_480_1/Brats18_TCIA12_480_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_480_1/Brats18_TCIA12_480_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_480_1/Brats18_TCIA12_480_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_480_1/Brats18_TCIA12_480_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_480_1/Brats18_TCIA12_480_1_seg.nii.gz
75 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_621_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_621_1/Brats18_TCIA13_621_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_621_1/Brats18_TCIA13_621_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_621_1/Brats18_TCIA13_621_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_621_1/Brats18_TCIA13_621_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_621_1/Brats18_TCIA13_621_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_621_1/Brats18_TCIA13_621_1_seg.nii.gz
76 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_633_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_633_1/Brats18_TCIA13_633_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_633_1/Brats18_TCIA13_633_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_633_1/Brats18_TCIA13_633_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_633_1/Brats18_TCIA13_633_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_633_1/Brats18_TCIA13_633_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_633_1/Brats18_TCIA13_633_1_seg.nii.gz
77 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_650_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_650_1/Brats18_TCIA13_650_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_650_1/Brats18_TCIA13_650_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_650_1/Brats18_TCIA13_650_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_650_1/Brats18_TCIA13_650_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_650_1/Brats18_TCIA13_650_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_650_1/Brats18_TCIA13_650_1_seg.nii.gz
78 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_10_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_10_1/Brats18_2013_10_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_10_1/Brats18_2013_10_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_10_1/Brats18_2013_10_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_10_1/Brats18_2013_10_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_10_1/Brats18_2013_10_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_10_1/Brats18_2013_10_1_seg.nii.gz
79 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_2_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_2_1/Brats18_2013_2_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_2_1/Brats18_2013_2_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_2_1/Brats18_2013_2_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_2_1/Brats18_2013_2_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_2_1/Brats18_2013_2_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_2_1/Brats18_2013_2_1_seg.nii.gz
80 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALN_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALN_1/Brats18_CBICA_ALN_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALN_1/Brats18_CBICA_ALN_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALN_1/Brats18_CBICA_ALN_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALN_1/Brats18_CBICA_ALN_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALN_1/Brats18_CBICA_ALN_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALN_1/Brats18_CBICA_ALN_1_seg.nii.gz
81 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APY_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APY_1/Brats18_CBICA_APY_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APY_1/Brats18_CBICA_APY_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APY_1/Brats18_CBICA_APY_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APY_1/Brats18_CBICA_APY_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APY_1/Brats18_CBICA_APY_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_APY_1/Brats18_CBICA_APY_1_seg.nii.gz
82 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQZ_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQZ_1/Brats18_CBICA_AQZ_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQZ_1/Brats18_CBICA_AQZ_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQZ_1/Brats18_CBICA_AQZ_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQZ_1/Brats18_CBICA_AQZ_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQZ_1/Brats18_CBICA_AQZ_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQZ_1/Brats18_CBICA_AQZ_1_seg.nii.gz
83 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATB_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATB_1/Brats18_CBICA_ATB_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATB_1/Brats18_CBICA_ATB_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATB_1/Brats18_CBICA_ATB_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATB_1/Brats18_CBICA_ATB_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATB_1/Brats18_CBICA_ATB_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATB_1/Brats18_CBICA_ATB_1_seg.nii.gz
84 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXJ_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXJ_1/Brats18_CBICA_AXJ_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXJ_1/Brats18_CBICA_AXJ_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXJ_1/Brats18_CBICA_AXJ_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXJ_1/Brats18_CBICA_AXJ_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXJ_1/Brats18_CBICA_AXJ_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXJ_1/Brats18_CBICA_AXJ_1_seg.nii.gz
85 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHB_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHB_1/Brats18_CBICA_BHB_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHB_1/Brats18_CBICA_BHB_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHB_1/Brats18_CBICA_BHB_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHB_1/Brats18_CBICA_BHB_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHB_1/Brats18_CBICA_BHB_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHB_1/Brats18_CBICA_BHB_1_seg.nii.gz
86 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_378_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_378_1/Brats18_TCIA01_378_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_378_1/Brats18_TCIA01_378_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_378_1/Brats18_TCIA01_378_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_378_1/Brats18_TCIA01_378_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_378_1/Brats18_TCIA01_378_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_378_1/Brats18_TCIA01_378_1_seg.nii.gz
87 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_171_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_171_1/Brats18_TCIA02_171_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_171_1/Brats18_TCIA02_171_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_171_1/Brats18_TCIA02_171_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_171_1/Brats18_TCIA02_171_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_171_1/Brats18_TCIA02_171_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_171_1/Brats18_TCIA02_171_1_seg.nii.gz
88 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_368_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_368_1/Brats18_TCIA02_368_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_368_1/Brats18_TCIA02_368_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_368_1/Brats18_TCIA02_368_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_368_1/Brats18_TCIA02_368_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_368_1/Brats18_TCIA02_368_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_368_1/Brats18_TCIA02_368_1_seg.nii.gz
89 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_133_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_133_1/Brats18_TCIA03_133_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_133_1/Brats18_TCIA03_133_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_133_1/Brats18_TCIA03_133_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_133_1/Brats18_TCIA03_133_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_133_1/Brats18_TCIA03_133_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_133_1/Brats18_TCIA03_133_1_seg.nii.gz
90 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_343_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_343_1/Brats18_TCIA04_343_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_343_1/Brats18_TCIA04_343_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_343_1/Brats18_TCIA04_343_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_343_1/Brats18_TCIA04_343_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_343_1/Brats18_TCIA04_343_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_343_1/Brats18_TCIA04_343_1_seg.nii.gz
91 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_603_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_603_1/Brats18_TCIA06_603_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_603_1/Brats18_TCIA06_603_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_603_1/Brats18_TCIA06_603_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_603_1/Brats18_TCIA06_603_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_603_1/Brats18_TCIA06_603_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_603_1/Brats18_TCIA06_603_1_seg.nii.gz
92 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_0_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_0_1/Brats18_2013_0_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_0_1/Brats18_2013_0_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_0_1/Brats18_2013_0_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_0_1/Brats18_2013_0_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_0_1/Brats18_2013_0_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_0_1/Brats18_2013_0_1_seg.nii.gz
93 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_402_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_402_1/Brats18_TCIA09_402_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_402_1/Brats18_TCIA09_402_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_402_1/Brats18_TCIA09_402_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_402_1/Brats18_TCIA09_402_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_402_1/Brats18_TCIA09_402_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_402_1/Brats18_TCIA09_402_1_seg.nii.gz
94 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_276_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_276_1/Brats18_TCIA10_276_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_276_1/Brats18_TCIA10_276_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_276_1/Brats18_TCIA10_276_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_276_1/Brats18_TCIA10_276_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_276_1/Brats18_TCIA10_276_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_276_1/Brats18_TCIA10_276_1_seg.nii.gz
95 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_442_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_442_1/Brats18_TCIA10_442_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_442_1/Brats18_TCIA10_442_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_442_1/Brats18_TCIA10_442_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_442_1/Brats18_TCIA10_442_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_442_1/Brats18_TCIA10_442_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_442_1/Brats18_TCIA10_442_1_seg.nii.gz
96 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_470_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_470_1/Brats18_TCIA12_470_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_470_1/Brats18_TCIA12_470_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_470_1/Brats18_TCIA12_470_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_470_1/Brats18_TCIA12_470_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_470_1/Brats18_TCIA12_470_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_470_1/Brats18_TCIA12_470_1_seg.nii.gz
97 | 


--------------------------------------------------------------------------------
/BT_CODE/data_csv/brats/test_1.csv:
--------------------------------------------------------------------------------
 1 | Dir_path,Flair_path,T1_path,T1ce_path,T2_path,Seg_path,HGG/LGG
 2 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_12_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_12_1/Brats18_2013_12_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_12_1/Brats18_2013_12_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_12_1/Brats18_2013_12_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_12_1/Brats18_2013_12_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_12_1/Brats18_2013_12_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_12_1/Brats18_2013_12_1_seg.nii.gz
 3 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_18_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_18_1/Brats18_2013_18_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_18_1/Brats18_2013_18_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_18_1/Brats18_2013_18_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_18_1/Brats18_2013_18_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_18_1/Brats18_2013_18_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_18_1/Brats18_2013_18_1_seg.nii.gz
 4 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_21_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_21_1/Brats18_2013_21_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_21_1/Brats18_2013_21_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_21_1/Brats18_2013_21_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_21_1/Brats18_2013_21_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_21_1/Brats18_2013_21_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_21_1/Brats18_2013_21_1_seg.nii.gz
 5 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_26_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_26_1/Brats18_2013_26_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_26_1/Brats18_2013_26_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_26_1/Brats18_2013_26_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_26_1/Brats18_2013_26_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_26_1/Brats18_2013_26_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_26_1/Brats18_2013_26_1_seg.nii.gz
 6 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_4_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_4_1/Brats18_2013_4_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_4_1/Brats18_2013_4_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_4_1/Brats18_2013_4_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_4_1/Brats18_2013_4_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_4_1/Brats18_2013_4_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_4_1/Brats18_2013_4_1_seg.nii.gz
 7 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAG_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAG_1/Brats18_CBICA_AAG_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAG_1/Brats18_CBICA_AAG_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAG_1/Brats18_CBICA_AAG_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAG_1/Brats18_CBICA_AAG_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAG_1/Brats18_CBICA_AAG_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAG_1/Brats18_CBICA_AAG_1_seg.nii.gz
 8 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABB_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABB_1/Brats18_CBICA_ABB_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABB_1/Brats18_CBICA_ABB_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABB_1/Brats18_CBICA_ABB_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABB_1/Brats18_CBICA_ABB_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABB_1/Brats18_CBICA_ABB_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABB_1/Brats18_CBICA_ABB_1_seg.nii.gz
 9 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABO_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABO_1/Brats18_CBICA_ABO_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABO_1/Brats18_CBICA_ABO_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABO_1/Brats18_CBICA_ABO_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABO_1/Brats18_CBICA_ABO_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABO_1/Brats18_CBICA_ABO_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ABO_1/Brats18_CBICA_ABO_1_seg.nii.gz
10 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALX_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALX_1/Brats18_CBICA_ALX_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALX_1/Brats18_CBICA_ALX_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALX_1/Brats18_CBICA_ALX_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALX_1/Brats18_CBICA_ALX_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALX_1/Brats18_CBICA_ALX_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ALX_1/Brats18_CBICA_ALX_1_seg.nii.gz
11 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANI_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANI_1/Brats18_CBICA_ANI_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANI_1/Brats18_CBICA_ANI_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANI_1/Brats18_CBICA_ANI_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANI_1/Brats18_CBICA_ANI_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANI_1/Brats18_CBICA_ANI_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANI_1/Brats18_CBICA_ANI_1_seg.nii.gz
12 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOD_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOD_1/Brats18_CBICA_AOD_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOD_1/Brats18_CBICA_AOD_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOD_1/Brats18_CBICA_AOD_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOD_1/Brats18_CBICA_AOD_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOD_1/Brats18_CBICA_AOD_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOD_1/Brats18_CBICA_AOD_1_seg.nii.gz
13 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOZ_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOZ_1/Brats18_CBICA_AOZ_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOZ_1/Brats18_CBICA_AOZ_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOZ_1/Brats18_CBICA_AOZ_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOZ_1/Brats18_CBICA_AOZ_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOZ_1/Brats18_CBICA_AOZ_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AOZ_1/Brats18_CBICA_AOZ_1_seg.nii.gz
14 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQA_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQA_1/Brats18_CBICA_AQA_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQA_1/Brats18_CBICA_AQA_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQA_1/Brats18_CBICA_AQA_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQA_1/Brats18_CBICA_AQA_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQA_1/Brats18_CBICA_AQA_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQA_1/Brats18_CBICA_AQA_1_seg.nii.gz
15 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQN_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQN_1/Brats18_CBICA_AQN_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQN_1/Brats18_CBICA_AQN_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQN_1/Brats18_CBICA_AQN_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQN_1/Brats18_CBICA_AQN_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQN_1/Brats18_CBICA_AQN_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQN_1/Brats18_CBICA_AQN_1_seg.nii.gz
16 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQQ_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQQ_1/Brats18_CBICA_AQQ_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQQ_1/Brats18_CBICA_AQQ_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQQ_1/Brats18_CBICA_AQQ_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQQ_1/Brats18_CBICA_AQQ_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQQ_1/Brats18_CBICA_AQQ_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQQ_1/Brats18_CBICA_AQQ_1_seg.nii.gz
17 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQV_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQV_1/Brats18_CBICA_AQV_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQV_1/Brats18_CBICA_AQV_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQV_1/Brats18_CBICA_AQV_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQV_1/Brats18_CBICA_AQV_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQV_1/Brats18_CBICA_AQV_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQV_1/Brats18_CBICA_AQV_1_seg.nii.gz
18 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARW_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARW_1/Brats18_CBICA_ARW_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARW_1/Brats18_CBICA_ARW_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARW_1/Brats18_CBICA_ARW_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARW_1/Brats18_CBICA_ARW_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARW_1/Brats18_CBICA_ARW_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ARW_1/Brats18_CBICA_ARW_1_seg.nii.gz
19 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASG_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASG_1/Brats18_CBICA_ASG_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASG_1/Brats18_CBICA_ASG_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASG_1/Brats18_CBICA_ASG_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASG_1/Brats18_CBICA_ASG_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASG_1/Brats18_CBICA_ASG_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASG_1/Brats18_CBICA_ASG_1_seg.nii.gz
20 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASN_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASN_1/Brats18_CBICA_ASN_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASN_1/Brats18_CBICA_ASN_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASN_1/Brats18_CBICA_ASN_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASN_1/Brats18_CBICA_ASN_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASN_1/Brats18_CBICA_ASN_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASN_1/Brats18_CBICA_ASN_1_seg.nii.gz
21 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASW_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASW_1/Brats18_CBICA_ASW_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASW_1/Brats18_CBICA_ASW_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASW_1/Brats18_CBICA_ASW_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASW_1/Brats18_CBICA_ASW_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASW_1/Brats18_CBICA_ASW_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASW_1/Brats18_CBICA_ASW_1_seg.nii.gz
22 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATF_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATF_1/Brats18_CBICA_ATF_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATF_1/Brats18_CBICA_ATF_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATF_1/Brats18_CBICA_ATF_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATF_1/Brats18_CBICA_ATF_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATF_1/Brats18_CBICA_ATF_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATF_1/Brats18_CBICA_ATF_1_seg.nii.gz
23 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUN_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUN_1/Brats18_CBICA_AUN_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUN_1/Brats18_CBICA_AUN_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUN_1/Brats18_CBICA_AUN_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUN_1/Brats18_CBICA_AUN_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUN_1/Brats18_CBICA_AUN_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AUN_1/Brats18_CBICA_AUN_1_seg.nii.gz
24 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AVG_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AVG_1/Brats18_CBICA_AVG_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AVG_1/Brats18_CBICA_AVG_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AVG_1/Brats18_CBICA_AVG_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AVG_1/Brats18_CBICA_AVG_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AVG_1/Brats18_CBICA_AVG_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AVG_1/Brats18_CBICA_AVG_1_seg.nii.gz
25 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWH_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWH_1/Brats18_CBICA_AWH_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWH_1/Brats18_CBICA_AWH_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWH_1/Brats18_CBICA_AWH_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWH_1/Brats18_CBICA_AWH_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWH_1/Brats18_CBICA_AWH_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AWH_1/Brats18_CBICA_AWH_1_seg.nii.gz
26 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXM_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXM_1/Brats18_CBICA_AXM_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXM_1/Brats18_CBICA_AXM_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXM_1/Brats18_CBICA_AXM_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXM_1/Brats18_CBICA_AXM_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXM_1/Brats18_CBICA_AXM_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXM_1/Brats18_CBICA_AXM_1_seg.nii.gz
27 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXW_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXW_1/Brats18_CBICA_AXW_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXW_1/Brats18_CBICA_AXW_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXW_1/Brats18_CBICA_AXW_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXW_1/Brats18_CBICA_AXW_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXW_1/Brats18_CBICA_AXW_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXW_1/Brats18_CBICA_AXW_1_seg.nii.gz
28 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYU_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYU_1/Brats18_CBICA_AYU_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYU_1/Brats18_CBICA_AYU_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYU_1/Brats18_CBICA_AYU_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYU_1/Brats18_CBICA_AYU_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYU_1/Brats18_CBICA_AYU_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AYU_1/Brats18_CBICA_AYU_1_seg.nii.gz
29 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BFB_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BFB_1/Brats18_CBICA_BFB_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BFB_1/Brats18_CBICA_BFB_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BFB_1/Brats18_CBICA_BFB_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BFB_1/Brats18_CBICA_BFB_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BFB_1/Brats18_CBICA_BFB_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BFB_1/Brats18_CBICA_BFB_1_seg.nii.gz
30 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHM_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHM_1/Brats18_CBICA_BHM_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHM_1/Brats18_CBICA_BHM_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHM_1/Brats18_CBICA_BHM_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHM_1/Brats18_CBICA_BHM_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHM_1/Brats18_CBICA_BHM_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_BHM_1/Brats18_CBICA_BHM_1_seg.nii.gz
31 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_180_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_180_1/Brats18_TCIA01_180_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_180_1/Brats18_TCIA01_180_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_180_1/Brats18_TCIA01_180_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_180_1/Brats18_TCIA01_180_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_180_1/Brats18_TCIA01_180_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_180_1/Brats18_TCIA01_180_1_seg.nii.gz
32 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_201_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_201_1/Brats18_TCIA01_201_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_201_1/Brats18_TCIA01_201_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_201_1/Brats18_TCIA01_201_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_201_1/Brats18_TCIA01_201_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_201_1/Brats18_TCIA01_201_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_201_1/Brats18_TCIA01_201_1_seg.nii.gz
33 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_235_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_235_1/Brats18_TCIA01_235_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_235_1/Brats18_TCIA01_235_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_235_1/Brats18_TCIA01_235_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_235_1/Brats18_TCIA01_235_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_235_1/Brats18_TCIA01_235_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_235_1/Brats18_TCIA01_235_1_seg.nii.gz
34 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_401_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_401_1/Brats18_TCIA01_401_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_401_1/Brats18_TCIA01_401_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_401_1/Brats18_TCIA01_401_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_401_1/Brats18_TCIA01_401_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_401_1/Brats18_TCIA01_401_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_401_1/Brats18_TCIA01_401_1_seg.nii.gz
35 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_429_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_429_1/Brats18_TCIA01_429_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_429_1/Brats18_TCIA01_429_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_429_1/Brats18_TCIA01_429_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_429_1/Brats18_TCIA01_429_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_429_1/Brats18_TCIA01_429_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_429_1/Brats18_TCIA01_429_1_seg.nii.gz
36 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_499_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_499_1/Brats18_TCIA01_499_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_499_1/Brats18_TCIA01_499_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_499_1/Brats18_TCIA01_499_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_499_1/Brats18_TCIA01_499_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_499_1/Brats18_TCIA01_499_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_499_1/Brats18_TCIA01_499_1_seg.nii.gz
37 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_151_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_151_1/Brats18_TCIA02_151_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_151_1/Brats18_TCIA02_151_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_151_1/Brats18_TCIA02_151_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_151_1/Brats18_TCIA02_151_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_151_1/Brats18_TCIA02_151_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_151_1/Brats18_TCIA02_151_1_seg.nii.gz
38 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_198_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_198_1/Brats18_TCIA02_198_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_198_1/Brats18_TCIA02_198_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_198_1/Brats18_TCIA02_198_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_198_1/Brats18_TCIA02_198_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_198_1/Brats18_TCIA02_198_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_198_1/Brats18_TCIA02_198_1_seg.nii.gz
39 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_274_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_274_1/Brats18_TCIA02_274_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_274_1/Brats18_TCIA02_274_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_274_1/Brats18_TCIA02_274_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_274_1/Brats18_TCIA02_274_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_274_1/Brats18_TCIA02_274_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_274_1/Brats18_TCIA02_274_1_seg.nii.gz
40 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_300_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_300_1/Brats18_TCIA02_300_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_300_1/Brats18_TCIA02_300_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_300_1/Brats18_TCIA02_300_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_300_1/Brats18_TCIA02_300_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_300_1/Brats18_TCIA02_300_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_300_1/Brats18_TCIA02_300_1_seg.nii.gz
41 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_322_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_322_1/Brats18_TCIA02_322_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_322_1/Brats18_TCIA02_322_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_322_1/Brats18_TCIA02_322_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_322_1/Brats18_TCIA02_322_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_322_1/Brats18_TCIA02_322_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_322_1/Brats18_TCIA02_322_1_seg.nii.gz
42 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_374_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_374_1/Brats18_TCIA02_374_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_374_1/Brats18_TCIA02_374_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_374_1/Brats18_TCIA02_374_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_374_1/Brats18_TCIA02_374_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_374_1/Brats18_TCIA02_374_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_374_1/Brats18_TCIA02_374_1_seg.nii.gz
43 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_455_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_455_1/Brats18_TCIA02_455_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_455_1/Brats18_TCIA02_455_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_455_1/Brats18_TCIA02_455_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_455_1/Brats18_TCIA02_455_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_455_1/Brats18_TCIA02_455_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_455_1/Brats18_TCIA02_455_1_seg.nii.gz
44 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_491_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_491_1/Brats18_TCIA02_491_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_491_1/Brats18_TCIA02_491_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_491_1/Brats18_TCIA02_491_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_491_1/Brats18_TCIA02_491_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_491_1/Brats18_TCIA02_491_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_491_1/Brats18_TCIA02_491_1_seg.nii.gz
45 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_608_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_608_1/Brats18_TCIA02_608_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_608_1/Brats18_TCIA02_608_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_608_1/Brats18_TCIA02_608_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_608_1/Brats18_TCIA02_608_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_608_1/Brats18_TCIA02_608_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_608_1/Brats18_TCIA02_608_1_seg.nii.gz
46 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_199_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_199_1/Brats18_TCIA03_199_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_199_1/Brats18_TCIA03_199_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_199_1/Brats18_TCIA03_199_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_199_1/Brats18_TCIA03_199_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_199_1/Brats18_TCIA03_199_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_199_1/Brats18_TCIA03_199_1_seg.nii.gz
47 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_338_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_338_1/Brats18_TCIA03_338_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_338_1/Brats18_TCIA03_338_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_338_1/Brats18_TCIA03_338_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_338_1/Brats18_TCIA03_338_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_338_1/Brats18_TCIA03_338_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_338_1/Brats18_TCIA03_338_1_seg.nii.gz
48 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_474_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_474_1/Brats18_TCIA03_474_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_474_1/Brats18_TCIA03_474_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_474_1/Brats18_TCIA03_474_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_474_1/Brats18_TCIA03_474_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_474_1/Brats18_TCIA03_474_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_474_1/Brats18_TCIA03_474_1_seg.nii.gz
49 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_192_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_192_1/Brats18_TCIA04_192_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_192_1/Brats18_TCIA04_192_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_192_1/Brats18_TCIA04_192_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_192_1/Brats18_TCIA04_192_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_192_1/Brats18_TCIA04_192_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_192_1/Brats18_TCIA04_192_1_seg.nii.gz
50 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_437_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_437_1/Brats18_TCIA04_437_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_437_1/Brats18_TCIA04_437_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_437_1/Brats18_TCIA04_437_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_437_1/Brats18_TCIA04_437_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_437_1/Brats18_TCIA04_437_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA04_437_1/Brats18_TCIA04_437_1_seg.nii.gz
51 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_444_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_444_1/Brats18_TCIA05_444_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_444_1/Brats18_TCIA05_444_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_444_1/Brats18_TCIA05_444_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_444_1/Brats18_TCIA05_444_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_444_1/Brats18_TCIA05_444_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_444_1/Brats18_TCIA05_444_1_seg.nii.gz
52 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_184_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_184_1/Brats18_TCIA06_184_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_184_1/Brats18_TCIA06_184_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_184_1/Brats18_TCIA06_184_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_184_1/Brats18_TCIA06_184_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_184_1/Brats18_TCIA06_184_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_184_1/Brats18_TCIA06_184_1_seg.nii.gz
53 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_372_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_372_1/Brats18_TCIA06_372_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_372_1/Brats18_TCIA06_372_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_372_1/Brats18_TCIA06_372_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_372_1/Brats18_TCIA06_372_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_372_1/Brats18_TCIA06_372_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA06_372_1/Brats18_TCIA06_372_1_seg.nii.gz
54 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_113_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_113_1/Brats18_TCIA08_113_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_113_1/Brats18_TCIA08_113_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_113_1/Brats18_TCIA08_113_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_113_1/Brats18_TCIA08_113_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_113_1/Brats18_TCIA08_113_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_113_1/Brats18_TCIA08_113_1_seg.nii.gz
55 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_218_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_218_1/Brats18_TCIA08_218_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_218_1/Brats18_TCIA08_218_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_218_1/Brats18_TCIA08_218_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_218_1/Brats18_TCIA08_218_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_218_1/Brats18_TCIA08_218_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_218_1/Brats18_TCIA08_218_1_seg.nii.gz
56 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_278_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_278_1/Brats18_TCIA08_278_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_278_1/Brats18_TCIA08_278_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_278_1/Brats18_TCIA08_278_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_278_1/Brats18_TCIA08_278_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_278_1/Brats18_TCIA08_278_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_278_1/Brats18_TCIA08_278_1_seg.nii.gz
57 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_436_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_436_1/Brats18_TCIA08_436_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_436_1/Brats18_TCIA08_436_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_436_1/Brats18_TCIA08_436_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_436_1/Brats18_TCIA08_436_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_436_1/Brats18_TCIA08_436_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_436_1/Brats18_TCIA08_436_1_seg.nii.gz
58 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_16_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_16_1/Brats18_2013_16_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_16_1/Brats18_2013_16_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_16_1/Brats18_2013_16_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_16_1/Brats18_2013_16_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_16_1/Brats18_2013_16_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_16_1/Brats18_2013_16_1_seg.nii.gz
59 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_29_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_29_1/Brats18_2013_29_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_29_1/Brats18_2013_29_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_29_1/Brats18_2013_29_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_29_1/Brats18_2013_29_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_29_1/Brats18_2013_29_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_29_1/Brats18_2013_29_1_seg.nii.gz
60 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_9_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_9_1/Brats18_2013_9_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_9_1/Brats18_2013_9_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_9_1/Brats18_2013_9_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_9_1/Brats18_2013_9_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_9_1/Brats18_2013_9_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_9_1/Brats18_2013_9_1_seg.nii.gz
61 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_255_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_255_1/Brats18_TCIA09_255_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_255_1/Brats18_TCIA09_255_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_255_1/Brats18_TCIA09_255_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_255_1/Brats18_TCIA09_255_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_255_1/Brats18_TCIA09_255_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_255_1/Brats18_TCIA09_255_1_seg.nii.gz
62 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_451_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_451_1/Brats18_TCIA09_451_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_451_1/Brats18_TCIA09_451_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_451_1/Brats18_TCIA09_451_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_451_1/Brats18_TCIA09_451_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_451_1/Brats18_TCIA09_451_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_451_1/Brats18_TCIA09_451_1_seg.nii.gz
63 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_103_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_103_1/Brats18_TCIA10_103_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_103_1/Brats18_TCIA10_103_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_103_1/Brats18_TCIA10_103_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_103_1/Brats18_TCIA10_103_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_103_1/Brats18_TCIA10_103_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_103_1/Brats18_TCIA10_103_1_seg.nii.gz
64 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_152_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_152_1/Brats18_TCIA10_152_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_152_1/Brats18_TCIA10_152_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_152_1/Brats18_TCIA10_152_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_152_1/Brats18_TCIA10_152_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_152_1/Brats18_TCIA10_152_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_152_1/Brats18_TCIA10_152_1_seg.nii.gz
65 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_261_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_261_1/Brats18_TCIA10_261_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_261_1/Brats18_TCIA10_261_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_261_1/Brats18_TCIA10_261_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_261_1/Brats18_TCIA10_261_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_261_1/Brats18_TCIA10_261_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_261_1/Brats18_TCIA10_261_1_seg.nii.gz
66 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_299_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_299_1/Brats18_TCIA10_299_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_299_1/Brats18_TCIA10_299_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_299_1/Brats18_TCIA10_299_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_299_1/Brats18_TCIA10_299_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_299_1/Brats18_TCIA10_299_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_299_1/Brats18_TCIA10_299_1_seg.nii.gz
67 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_330_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_330_1/Brats18_TCIA10_330_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_330_1/Brats18_TCIA10_330_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_330_1/Brats18_TCIA10_330_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_330_1/Brats18_TCIA10_330_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_330_1/Brats18_TCIA10_330_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_330_1/Brats18_TCIA10_330_1_seg.nii.gz
68 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_387_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_387_1/Brats18_TCIA10_387_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_387_1/Brats18_TCIA10_387_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_387_1/Brats18_TCIA10_387_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_387_1/Brats18_TCIA10_387_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_387_1/Brats18_TCIA10_387_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_387_1/Brats18_TCIA10_387_1_seg.nii.gz
69 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_413_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_413_1/Brats18_TCIA10_413_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_413_1/Brats18_TCIA10_413_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_413_1/Brats18_TCIA10_413_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_413_1/Brats18_TCIA10_413_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_413_1/Brats18_TCIA10_413_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_413_1/Brats18_TCIA10_413_1_seg.nii.gz
70 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_490_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_490_1/Brats18_TCIA10_490_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_490_1/Brats18_TCIA10_490_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_490_1/Brats18_TCIA10_490_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_490_1/Brats18_TCIA10_490_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_490_1/Brats18_TCIA10_490_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_490_1/Brats18_TCIA10_490_1_seg.nii.gz
71 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_632_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_632_1/Brats18_TCIA10_632_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_632_1/Brats18_TCIA10_632_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_632_1/Brats18_TCIA10_632_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_632_1/Brats18_TCIA10_632_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_632_1/Brats18_TCIA10_632_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_632_1/Brats18_TCIA10_632_1_seg.nii.gz
72 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_640_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_640_1/Brats18_TCIA10_640_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_640_1/Brats18_TCIA10_640_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_640_1/Brats18_TCIA10_640_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_640_1/Brats18_TCIA10_640_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_640_1/Brats18_TCIA10_640_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_640_1/Brats18_TCIA10_640_1_seg.nii.gz
73 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_298_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_298_1/Brats18_TCIA12_298_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_298_1/Brats18_TCIA12_298_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_298_1/Brats18_TCIA12_298_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_298_1/Brats18_TCIA12_298_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_298_1/Brats18_TCIA12_298_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA12_298_1/Brats18_TCIA12_298_1_seg.nii.gz
74 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_615_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_615_1/Brats18_TCIA13_615_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_615_1/Brats18_TCIA13_615_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_615_1/Brats18_TCIA13_615_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_615_1/Brats18_TCIA13_615_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_615_1/Brats18_TCIA13_615_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_615_1/Brats18_TCIA13_615_1_seg.nii.gz
75 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_624_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_624_1/Brats18_TCIA13_624_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_624_1/Brats18_TCIA13_624_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_624_1/Brats18_TCIA13_624_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_624_1/Brats18_TCIA13_624_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_624_1/Brats18_TCIA13_624_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_624_1/Brats18_TCIA13_624_1_seg.nii.gz
76 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_634_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_634_1/Brats18_TCIA13_634_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_634_1/Brats18_TCIA13_634_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_634_1/Brats18_TCIA13_634_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_634_1/Brats18_TCIA13_634_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_634_1/Brats18_TCIA13_634_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_634_1/Brats18_TCIA13_634_1_seg.nii.gz
77 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_653_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_653_1/Brats18_TCIA13_653_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_653_1/Brats18_TCIA13_653_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_653_1/Brats18_TCIA13_653_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_653_1/Brats18_TCIA13_653_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_653_1/Brats18_TCIA13_653_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_653_1/Brats18_TCIA13_653_1_seg.nii.gz
78 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_17_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_17_1/Brats18_2013_17_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_17_1/Brats18_2013_17_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_17_1/Brats18_2013_17_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_17_1/Brats18_2013_17_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_17_1/Brats18_2013_17_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_2013_17_1/Brats18_2013_17_1_seg.nii.gz
79 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAB_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAB_1/Brats18_CBICA_AAB_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAB_1/Brats18_CBICA_AAB_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAB_1/Brats18_CBICA_AAB_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAB_1/Brats18_CBICA_AAB_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAB_1/Brats18_CBICA_AAB_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AAB_1/Brats18_CBICA_AAB_1_seg.nii.gz
80 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANG_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANG_1/Brats18_CBICA_ANG_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANG_1/Brats18_CBICA_ANG_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANG_1/Brats18_CBICA_ANG_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANG_1/Brats18_CBICA_ANG_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANG_1/Brats18_CBICA_ANG_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ANG_1/Brats18_CBICA_ANG_1_seg.nii.gz
81 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQJ_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQJ_1/Brats18_CBICA_AQJ_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQJ_1/Brats18_CBICA_AQJ_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQJ_1/Brats18_CBICA_AQJ_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQJ_1/Brats18_CBICA_AQJ_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQJ_1/Brats18_CBICA_AQJ_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AQJ_1/Brats18_CBICA_AQJ_1_seg.nii.gz
82 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASE_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASE_1/Brats18_CBICA_ASE_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASE_1/Brats18_CBICA_ASE_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASE_1/Brats18_CBICA_ASE_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASE_1/Brats18_CBICA_ASE_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASE_1/Brats18_CBICA_ASE_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ASE_1/Brats18_CBICA_ASE_1_seg.nii.gz
83 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATX_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATX_1/Brats18_CBICA_ATX_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATX_1/Brats18_CBICA_ATX_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATX_1/Brats18_CBICA_ATX_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATX_1/Brats18_CBICA_ATX_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATX_1/Brats18_CBICA_ATX_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_ATX_1/Brats18_CBICA_ATX_1_seg.nii.gz
84 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXQ_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXQ_1/Brats18_CBICA_AXQ_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXQ_1/Brats18_CBICA_AXQ_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXQ_1/Brats18_CBICA_AXQ_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXQ_1/Brats18_CBICA_AXQ_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXQ_1/Brats18_CBICA_AXQ_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_CBICA_AXQ_1/Brats18_CBICA_AXQ_1_seg.nii.gz
85 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_150_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_150_1/Brats18_TCIA01_150_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_150_1/Brats18_TCIA01_150_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_150_1/Brats18_TCIA01_150_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_150_1/Brats18_TCIA01_150_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_150_1/Brats18_TCIA01_150_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_150_1/Brats18_TCIA01_150_1_seg.nii.gz
86 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_425_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_425_1/Brats18_TCIA01_425_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_425_1/Brats18_TCIA01_425_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_425_1/Brats18_TCIA01_425_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_425_1/Brats18_TCIA01_425_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_425_1/Brats18_TCIA01_425_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA01_425_1/Brats18_TCIA01_425_1_seg.nii.gz
87 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_226_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_226_1/Brats18_TCIA02_226_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_226_1/Brats18_TCIA02_226_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_226_1/Brats18_TCIA02_226_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_226_1/Brats18_TCIA02_226_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_226_1/Brats18_TCIA02_226_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_226_1/Brats18_TCIA02_226_1_seg.nii.gz
88 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_430_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_430_1/Brats18_TCIA02_430_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_430_1/Brats18_TCIA02_430_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_430_1/Brats18_TCIA02_430_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_430_1/Brats18_TCIA02_430_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_430_1/Brats18_TCIA02_430_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA02_430_1/Brats18_TCIA02_430_1_seg.nii.gz
89 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_296_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_296_1/Brats18_TCIA03_296_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_296_1/Brats18_TCIA03_296_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_296_1/Brats18_TCIA03_296_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_296_1/Brats18_TCIA03_296_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_296_1/Brats18_TCIA03_296_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA03_296_1/Brats18_TCIA03_296_1_seg.nii.gz
90 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_396_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_396_1/Brats18_TCIA05_396_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_396_1/Brats18_TCIA05_396_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_396_1/Brats18_TCIA05_396_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_396_1/Brats18_TCIA05_396_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_396_1/Brats18_TCIA05_396_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA05_396_1/Brats18_TCIA05_396_1_seg.nii.gz
91 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_205_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_205_1/Brats18_TCIA08_205_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_205_1/Brats18_TCIA08_205_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_205_1/Brats18_TCIA08_205_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_205_1/Brats18_TCIA08_205_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_205_1/Brats18_TCIA08_205_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/HGG/Brats18_TCIA08_205_1/Brats18_TCIA08_205_1_seg.nii.gz
92 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_28_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_28_1/Brats18_2013_28_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_28_1/Brats18_2013_28_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_28_1/Brats18_2013_28_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_28_1/Brats18_2013_28_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_28_1/Brats18_2013_28_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_2013_28_1/Brats18_2013_28_1_seg.nii.gz
93 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_620_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_620_1/Brats18_TCIA09_620_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_620_1/Brats18_TCIA09_620_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_620_1/Brats18_TCIA09_620_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_620_1/Brats18_TCIA09_620_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_620_1/Brats18_TCIA09_620_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA09_620_1/Brats18_TCIA09_620_1_seg.nii.gz
94 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_325_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_325_1/Brats18_TCIA10_325_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_325_1/Brats18_TCIA10_325_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_325_1/Brats18_TCIA10_325_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_325_1/Brats18_TCIA10_325_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_325_1/Brats18_TCIA10_325_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_325_1/Brats18_TCIA10_325_1_seg.nii.gz
95 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_629_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_629_1/Brats18_TCIA10_629_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_629_1/Brats18_TCIA10_629_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_629_1/Brats18_TCIA10_629_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_629_1/Brats18_TCIA10_629_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_629_1/Brats18_TCIA10_629_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA10_629_1/Brats18_TCIA10_629_1_seg.nii.gz
96 | /DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_623_1,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_623_1/Brats18_TCIA13_623_1_flair.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_623_1/Brats18_TCIA13_623_1_t1.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_623_1/Brats18_TCIA13_623_1_t1ce.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_623_1/Brats18_TCIA13_623_1_t2.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_623_1/Brats18_TCIA13_623_1_seg.nii.gz,/DATA/BraTS2018/MICCAI_BraTS_2018_Data_Training/LGG/Brats18_TCIA13_623_1/Brats18_TCIA13_623_1_seg.nii.gz
97 | 


--------------------------------------------------------------------------------
/BT_CODE/dataload/augmentations.py:
--------------------------------------------------------------------------------
  1 | import numpy as np
  2 | import torch
  3 | import copy
  4 | from skimage.transform import resize
  5 | from scipy.ndimage.interpolation import map_coordinates
  6 | import SimpleITK as sitk
  7 | import pickle
  8 | from scipy.ndimage import gaussian_filter
  9 | import random
 10 | import abc
 11 | import os
 12 | 
 13 | 
 14 | def resize_segmentation(segmentation, new_shape, order=3, cval=0):
 15 |     tpe = segmentation.dtype
 16 |     unique_labels = np.unique(segmentation)
 17 |     assert len(segmentation.shape) == len(new_shape), "new shape must have same dimensionality as segmentation"
 18 |     if order == 0:
 19 |         return resize(segmentation.astype(float), new_shape, order, mode="constant", cval=cval, clip=True, anti_aliasing=False).astype(tpe)
 20 |     else:
 21 |         reshaped = np.zeros(new_shape, dtype=segmentation.dtype)
 22 | 
 23 |         for i, c in enumerate(unique_labels):
 24 |             mask = segmentation == c
 25 |             reshaped_multihot = resize(mask.astype(float), new_shape, order, mode="edge", clip=True, anti_aliasing=False)
 26 |             reshaped[reshaped_multihot >= 0.5] = c
 27 |         return reshaped
 28 | 
 29 | 
 30 | class AbstractTransform(object):
 31 |     __metaclass__ = abc.ABCMeta
 32 | 
 33 |     @abc.abstractmethod
 34 |     def __call__(self, **data_dict):
 35 |         raise NotImplementedError("Abstract, so implement")
 36 | 
 37 |     def __repr__(self):
 38 |         ret_str = str(type(self).__name__) + "( " + ", ".join(
 39 |             [key + " = " + repr(val) for key, val in self.__dict__.items()]) + " )"
 40 |         return ret_str
 41 | 
 42 | 
 43 | def uniform(low, high, size=None):
 44 | 
 45 |     if low == high:
 46 |         if size is None:
 47 |             return low
 48 |         else:
 49 |             return np.ones(size) * low
 50 |     else:
 51 |         return np.random.uniform(low, high, size)
 52 | 
 53 | 
 54 | def augment_contrast(data_sample, contrast_range=(0.75, 1.25), preserve_range=True, per_channel=True):
 55 |     if not per_channel:
 56 |         mn = data_sample.mean()
 57 |         if preserve_range:
 58 |             minm = data_sample.min()
 59 |             maxm = data_sample.max()
 60 |         if np.random.random() < 0.5 and contrast_range[0] < 1:
 61 |             factor = np.random.uniform(contrast_range[0], 1)
 62 |         else:
 63 |             factor = np.random.uniform(max(contrast_range[0], 1), contrast_range[1])
 64 |         data_sample = (data_sample - mn) * factor + mn
 65 |         if preserve_range:
 66 |             data_sample[data_sample < minm] = minm
 67 |             data_sample[data_sample > maxm] = maxm
 68 |     else:
 69 |         for c in range(data_sample.shape[0]):
 70 |             mn = data_sample[c].mean()
 71 |             if preserve_range:
 72 |                 minm = data_sample[c].min()
 73 |                 maxm = data_sample[c].max()
 74 |             if np.random.random() < 0.5 and contrast_range[0] < 1:
 75 |                 factor = np.random.uniform(contrast_range[0], 1)
 76 |             else:
 77 |                 factor = np.random.uniform(max(contrast_range[0], 1), contrast_range[1])
 78 |             data_sample[c] = (data_sample[c] - mn) * factor + mn
 79 |             if preserve_range:
 80 |                 data_sample[c][data_sample[c] < minm] = minm
 81 |                 data_sample[c][data_sample[c] > maxm] = maxm
 82 |     return data_sample
 83 | 
 84 | 
 85 | def convert_3d_to_2d_generator(data_dict):
 86 |     shp = data_dict['data'].shape
 87 |     data_dict['data'] = data_dict['data'].reshape((shp[0], shp[1] * shp[2], shp[3], shp[4]))
 88 |     data_dict['orig_shape_data'] = shp
 89 |     shp = data_dict['seg'].shape
 90 |     data_dict['seg'] = data_dict['seg'].reshape((shp[0], shp[1] * shp[2], shp[3], shp[4]))
 91 |     data_dict['orig_shape_seg'] = shp
 92 |     return data_dict
 93 | 
 94 | 
 95 | def convert_2d_to_3d_generator(data_dict):
 96 |     shp = data_dict['orig_shape_data']
 97 |     current_shape = data_dict['data'].shape
 98 |     data_dict['data'] = data_dict['data'].reshape((shp[0], shp[1], shp[2], current_shape[-2], current_shape[-1]))
 99 |     shp = data_dict['orig_shape_seg']
100 |     current_shape_seg = data_dict['seg'].shape
101 |     data_dict['seg'] = data_dict['seg'].reshape((shp[0], shp[1], shp[2], current_shape_seg[-2], current_shape_seg[-1]))
102 |     return data_dict
103 | 
104 | 
105 | def get_range_val(value, rnd_type="uniform"):
106 |     if isinstance(value, (list, tuple, np.ndarray)):
107 |         if len(value) == 2:
108 |             if value[0] == value[1]:
109 |                 n_val = value[0]
110 |             else:
111 |                 orig_type = type(value[0])
112 |                 if rnd_type == "uniform":
113 |                     n_val = random.uniform(value[0], value[1])
114 |                 elif rnd_type == "normal":
115 |                     n_val = random.normalvariate(value[0], value[1])
116 |                 n_val = orig_type(n_val)
117 |         elif len(value) == 1:
118 |             n_val = value[0]
119 |         else:
120 |             raise RuntimeError("value must be either a single vlaue or a list/tuple of len 2")
121 |         return n_val
122 |     else:
123 |         return value
124 | 
125 | 
126 | def augment_gaussian_noise(data_sample, noise_variance=(0, 0.1)):
127 |     if noise_variance[0] == noise_variance[1]:
128 |         variance = noise_variance[0]
129 |     else:
130 |         variance = random.uniform(noise_variance[0], noise_variance[1])
131 |     data_sample = data_sample + np.random.normal(0.0, variance, size=data_sample.shape)
132 |     return data_sample
133 | 
134 | 
135 | def augment_gaussian_blur(data_sample, sigma_range, per_channel=True, p_per_channel=1):
136 |     if not per_channel:
137 |         sigma = get_range_val(sigma_range)
138 |     for c in range(data_sample.shape[0]):
139 |         if np.random.uniform() <= p_per_channel:
140 |             if per_channel:
141 |                 sigma = get_range_val(sigma_range)
142 |             data_sample[c] = gaussian_filter(data_sample[c], sigma, order=0)
143 |     return data_sample
144 | 
145 | 
146 | def augment_brightness_multiplicative(data_sample, multiplier_range=(0.5, 2), per_channel=True):
147 |     multiplier = np.random.uniform(multiplier_range[0], multiplier_range[1])
148 |     if not per_channel:
149 |         data_sample *= multiplier
150 |     else:
151 |         for c in range(data_sample.shape[0]):
152 |             multiplier = np.random.uniform(multiplier_range[0], multiplier_range[1])
153 |             data_sample[c] *= multiplier
154 |     return data_sample
155 | 
156 | 
157 | def augment_gamma(data_sample, gamma_range=(0.5, 2), invert_image=False, epsilon=1e-7, per_channel=False,
158 |                   retain_stats=False):
159 |     if invert_image:
160 |         data_sample = - data_sample
161 |     if not per_channel:
162 |         if retain_stats:
163 |             mn = data_sample.mean()
164 |             sd = data_sample.std()
165 |         if np.random.random() < 0.5 and gamma_range[0] < 1:
166 |             gamma = np.random.uniform(gamma_range[0], 1)
167 |         else:
168 |             gamma = np.random.uniform(max(gamma_range[0], 1), gamma_range[1])
169 |         minm = data_sample.min()
170 |         rnge = data_sample.max() - minm
171 |         data_sample = np.power(((data_sample - minm) / float(rnge + epsilon)), gamma) * rnge + minm
172 |         if retain_stats:
173 |             data_sample = data_sample - data_sample.mean() + mn
174 |             data_sample = data_sample / (data_sample.std() + 1e-8) * sd
175 |     else:
176 |         for c in range(data_sample.shape[0]):
177 |             if retain_stats:
178 |                 mn = data_sample[c].mean()
179 |                 sd = data_sample[c].std()
180 |             if np.random.random() < 0.5 and gamma_range[0] < 1:
181 |                 gamma = np.random.uniform(gamma_range[0], 1)
182 |             else:
183 |                 gamma = np.random.uniform(max(gamma_range[0], 1), gamma_range[1])
184 |             minm = data_sample[c].min()
185 |             rnge = data_sample[c].max() - minm
186 |             data_sample[c] = np.power(((data_sample[c] - minm) / float(rnge + epsilon)), gamma) * float(rnge + epsilon) + minm
187 |             if retain_stats:
188 |                 data_sample[c] = data_sample[c] - data_sample[c].mean() + mn
189 |                 data_sample[c] = data_sample[c] / (data_sample[c].std() + 1e-8) * sd
190 |     if invert_image:
191 |         data_sample = - data_sample
192 |     return data_sample
193 | 
194 | 
195 | def augment_linear_downsampling_scipy(data_sample, zoom_range=(0.5, 1), per_channel=True, p_per_channel=1,
196 |                                       channels=None, order_downsample=1, order_upsample=0, ignore_axes=None):
197 |     if not isinstance(zoom_range, (list, tuple, np.ndarray)):
198 |         zoom_range = [zoom_range]
199 | 
200 |     shp = np.array(data_sample.shape[1:])
201 |     dim = len(shp)
202 | 
203 |     if not per_channel:
204 |         if isinstance(zoom_range[0], (tuple, list, np.ndarray)):
205 |             assert len(zoom_range) == dim
206 |             zoom = np.array([uniform(i[0], i[1]) for i in zoom_range])
207 |         else:
208 |             zoom = uniform(zoom_range[0], zoom_range[1])
209 | 
210 |         target_shape = np.round(shp * zoom).astype(int)
211 | 
212 |         if ignore_axes is not None:
213 |             for i in ignore_axes:
214 |                 target_shape[i] = shp[i]
215 | 
216 |     if channels is None:
217 |         channels = list(range(data_sample.shape[0]))
218 | 
219 |     for c in channels:
220 |         if np.random.uniform() < p_per_channel:
221 |             if per_channel:
222 |                 if isinstance(zoom_range[0], (tuple, list, np.ndarray)):
223 |                     assert len(zoom_range) == dim
224 |                     zoom = np.array([uniform(i[0], i[1]) for i in zoom_range])
225 |                 else:
226 |                     zoom = uniform(zoom_range[0], zoom_range[1])
227 | 
228 |                 target_shape = np.round(shp * zoom).astype(int)
229 |                 if ignore_axes is not None:
230 |                     for i in ignore_axes:
231 |                         target_shape[i] = shp[i]
232 | 
233 |             downsampled = resize(data_sample[c].astype(float), target_shape, order=order_downsample, mode='edge',
234 |                                  anti_aliasing=False)
235 |             data_sample[c] = resize(downsampled, shp, order=order_upsample, mode='edge',
236 |                                     anti_aliasing=False)
237 | 
238 |     return data_sample
239 | 
240 | 
241 | def augment_mirroring(sample_data, sample_seg=None, axes=(0, 1, 2)):
242 |     if (len(sample_data.shape) != 3) and (len(sample_data.shape) != 4):
243 |         raise Exception(
244 |             "Invalid dimension for sample_data and sample_seg. sample_data and sample_seg should be either "
245 |             "[channels, x, y] or [channels, x, y, z]")
246 |     if 0 in axes and np.random.uniform() < 0.5:
247 |         sample_data[:, :] = sample_data[:, ::-1]
248 |         if sample_seg is not None:
249 |             sample_seg[:, :] = sample_seg[:, ::-1]
250 |     if 1 in axes and np.random.uniform() < 0.5:
251 |         sample_data[:, :, :] = sample_data[:, :, ::-1]
252 |         if sample_seg is not None:
253 |             sample_seg[:, :, :] = sample_seg[:, :, ::-1]
254 |     if 2 in axes and len(sample_data.shape) == 4:
255 |         if np.random.uniform() < 0.5:
256 |             sample_data[:, :, :, :] = sample_data[:, :, :, ::-1]
257 |             if sample_seg is not None:
258 |                 sample_seg[:, :, :, :] = sample_seg[:, :, :, ::-1]
259 |     return sample_data, sample_seg
260 | 
261 | 
262 | class Compose(AbstractTransform):
263 |     def __init__(self, transforms):
264 |         self.transforms = transforms
265 | 
266 |     def __call__(self, data_dict):
267 |         for t in self.transforms:
268 |             data_dict = t(data_dict)
269 |         return data_dict
270 | 
271 |     def __repr__(self):
272 |         return str(type(self).__name__) + " ( " + repr(self.transforms) + " )"
273 | 
274 | 
275 | class GaussianNoiseTransform(AbstractTransform):
276 |     def __init__(self, noise_variance=(0, 0.1), data_key="data", label_key="seg", p_per_sample=1):
277 |         self.p_per_sample = p_per_sample
278 |         self.data_key = data_key
279 |         self.label_key = label_key
280 |         self.noise_variance = noise_variance
281 |     
282 |     def  __call__(self, img_array):
283 |         if np.random.uniform() < self.p_per_sample:
284 |             img_array = augment_gaussian_noise(img_array, self.noise_variance)
285 |         return img_array
286 | 
287 |     # def __call__(self, data_dict):
288 |     #     for b in range(len(data_dict[self.data_key])):
289 |     #         if np.random.uniform() < self.p_per_sample:
290 |     #             data_dict[self.data_key][b] = augment_gaussian_noise(data_dict[self.data_key][b], self.noise_variance)
291 |     #     return data_dict
292 | 
293 | 
294 | class GaussianBlurTransform(AbstractTransform):
295 |     def __init__(self, blur_sigma=(1, 5), data_key="data", label_key="seg", different_sigma_per_channel=True,
296 |                  p_per_channel=1, p_per_sample=1):
297 |         self.p_per_sample = p_per_sample
298 |         self.different_sigma_per_channel = different_sigma_per_channel
299 |         self.p_per_channel = p_per_channel
300 |         self.data_key = data_key
301 |         self.label_key = label_key
302 |         self.blur_sigma = blur_sigma
303 |     
304 |     def  __call__(self, img_array):
305 |         if np.random.uniform() < self.p_per_sample:
306 |             img_array = augment_gaussian_blur(img_array,  self.blur_sigma,
307 |                         self.different_sigma_per_channel, self.p_per_channel)
308 |         return img_array
309 | 
310 |     # def __call__(self, data_dict):
311 |     #     for b in range(len(data_dict[self.data_key])):
312 |     #         if np.random.uniform() < self.p_per_sample:
313 |     #             data_dict[self.data_key][b] = augment_gaussian_blur(data_dict[self.data_key][b], self.blur_sigma,
314 |     #                                                              self.different_sigma_per_channel, self.p_per_channel)
315 |     #     return data_dict
316 | 
317 | 
318 | class BrightnessMultiplicativeTransform(AbstractTransform):
319 |     def __init__(self, multiplier_range=(0.5, 2), per_channel=True, data_key="data", p_per_sample=1):
320 |         self.p_per_sample = p_per_sample
321 |         self.data_key = data_key
322 |         self.multiplier_range = multiplier_range
323 |         self.per_channel = per_channel
324 |     
325 |     def  __call__(self, img_array):
326 |         if np.random.uniform() < self.p_per_sample:
327 |             img_array = augment_brightness_multiplicative(img_array,self.multiplier_range, self.per_channel)
328 |         return img_array
329 | 
330 |     # def __call__(self, data_dict):
331 |     #     for b in range(len(data_dict[self.data_key])):
332 |     #         if np.random.uniform() < self.p_per_sample:
333 |     #             data_dict[self.data_key][b] = augment_brightness_multiplicative(data_dict[self.data_key][b],
334 |     #                                                                             self.multiplier_range,
335 |     #                                                                             self.per_channel)
336 |     #     return data_dict
337 | 
338 | 
339 | class GammaTransform(AbstractTransform):
340 |     def __init__(self, gamma_range=(0.5, 2), invert_image=False, per_channel=False, data_key="data", retain_stats=False,
341 |                  p_per_sample=1):
342 |         self.p_per_sample = p_per_sample
343 |         self.retain_stats = retain_stats
344 |         self.per_channel = per_channel
345 |         self.data_key = data_key
346 |         self.gamma_range = gamma_range
347 |         self.invert_image = invert_image
348 |     
349 |     def  __call__(self, img_array):
350 |         if np.random.uniform() < self.p_per_sample:
351 |             img_array = augment_gamma(img_array,self.gamma_range,
352 |                                                             self.invert_image,
353 |                                                             per_channel=self.per_channel,
354 |                                                             retain_stats=self.retain_stats)
355 |         return img_array
356 | 
357 |     # def __call__(self, data_dict):
358 |     #     for b in range(len(data_dict[self.data_key])):
359 |     #         if np.random.uniform() < self.p_per_sample:
360 |     #             data_dict[self.data_key][b] = augment_gamma(data_dict[self.data_key][b], self.gamma_range,
361 |     #                                                         self.invert_image,
362 |     #                                                         per_channel=self.per_channel,
363 |     #                                                         retain_stats=self.retain_stats)
364 |     #     return data_dict
365 | 
366 | 
367 | class SimulateLowResolutionTransform(AbstractTransform):
368 |     def __init__(self, zoom_range=(0.5, 1), per_channel=False, p_per_channel=1,
369 |                  channels=None, order_downsample=1, order_upsample=0, data_key="data", p_per_sample=1,
370 |                  ignore_axes=None):
371 |         self.order_upsample = order_upsample
372 |         self.order_downsample = order_downsample
373 |         self.channels = channels
374 |         self.per_channel = per_channel
375 |         self.p_per_channel = p_per_channel
376 |         self.p_per_sample = p_per_sample
377 |         self.data_key = data_key
378 |         self.zoom_range = zoom_range
379 |         self.ignore_axes = ignore_axes
380 |     
381 |     def  __call__(self, img_array):
382 |         if np.random.uniform() < self.p_per_sample:
383 |             img_array = augment_linear_downsampling_scipy(img_array,
384 |                                                             zoom_range=self.zoom_range,
385 |                                                             per_channel=self.per_channel,
386 |                                                             p_per_channel=self.p_per_channel,
387 |                                                             channels=self.channels,
388 |                                                             order_downsample=self.order_downsample,
389 |                                                             order_upsample=self.order_upsample,
390 |                                                             ignore_axes=self.ignore_axes)
391 |         return img_array
392 | 
393 |     # def __call__(self, data_dict):
394 |     #     for b in range(len(data_dict[self.data_key])):
395 |     #         if np.random.uniform() < self.p_per_sample:
396 |     #             data_dict[self.data_key][b] = augment_linear_downsampling_scipy(data_dict[self.data_key][b],
397 |     #                                                                             zoom_range=self.zoom_range,
398 |     #                                                                             per_channel=self.per_channel,
399 |     #                                                                             p_per_channel=self.p_per_channel,
400 |     #                                                                             channels=self.channels,
401 |     #                                                                             order_downsample=self.order_downsample,
402 |     #                                                                             order_upsample=self.order_upsample,
403 |     #                                                                             ignore_axes=self.ignore_axes)
404 |     #     return data_dict
405 | 
406 | 
407 | class ContrastAugmentationTransform(AbstractTransform):
408 |     def __init__(self, contrast_range=(0.75, 1.25), preserve_range=True, per_channel=True, data_key="data",
409 |                  p_per_sample=1):
410 |         self.p_per_sample = p_per_sample
411 |         self.data_key = data_key
412 |         self.contrast_range = contrast_range
413 |         self.preserve_range = preserve_range
414 |         self.per_channel = per_channel
415 |     
416 |     def  __call__(self, img_array):
417 |         if np.random.uniform() < self.p_per_sample:
418 |             img_array = augment_contrast(img_array,contrast_range=self.contrast_range,
419 |                                                                preserve_range=self.preserve_range,
420 |                                                                per_channel=self.per_channel)
421 |         return img_array
422 | 
423 |     # def __call__(self, data_dict):
424 |     #     for b in range(len(data_dict[self.data_key])):
425 |     #         if np.random.uniform() < self.p_per_sample:
426 |     #             data_dict[self.data_key][b] = augment_contrast(data_dict[self.data_key][b],
427 |     #                                                            contrast_range=self.contrast_range,
428 |     #                                                            preserve_range=self.preserve_range,
429 |     #                                                            per_channel=self.per_channel)
430 |     #     return data_dict
431 | 
432 | 
433 | class MirrorTransform(AbstractTransform):
434 |     def __init__(self, axes=(0, 1, 2), data_key="data", label_key="seg"):
435 |         self.data_key = data_key
436 |         self.label_key = label_key
437 |         self.axes = axes
438 |         if max(axes) > 2:
439 |             raise ValueError("MirrorTransform now takes the axes as the spatial dimensions. What previously was "
440 |                              "axes=(2, 3, 4) to mirror along all spatial dimensions of a 5d tensor (b, c, x, y, z) "
441 |                              "is now axes=(0, 1, 2). Please adapt your scripts accordingly.")
442 | 
443 |     def __call__(self, data_dict):
444 |         data = data_dict.get(self.data_key)
445 |         seg = data_dict.get(self.label_key)
446 | 
447 |         for b in range(len(data)):
448 |             sample_seg = None
449 |             if seg is not None:
450 |                 sample_seg = seg[b]
451 |             ret_val = augment_mirroring(data[b], sample_seg, axes=self.axes)
452 |             data[b] = ret_val[0]
453 |             if seg is not None:
454 |                 seg[b] = ret_val[1]
455 | 
456 |         data_dict[self.data_key] = data
457 |         if seg is not None:
458 |             data_dict[self.label_key] = seg
459 | 
460 |         return data_dict
461 | 


--------------------------------------------------------------------------------
/BT_CODE/dataload/dataload.py:
--------------------------------------------------------------------------------
  1 | from __future__ import print_function
  2 | import cv2
  3 | import os
  4 | import torch
  5 | import numpy as np
  6 | import pandas as pd
  7 | from scipy import ndimage
  8 | import elasticdeform
  9 | 
 10 | import random
 11 | from typing import Optional, List, Callable, Any
 12 | from torch.utils.data import Dataset
 13 | from torchvision import datasets, transforms
 14 | from skimage.transform import resize
 15 | import SimpleITK as sitk
 16 | import torchio as tio
 17 | 
 18 | from utils.vtk_itk import pd_to_itk_image
 19 | from utils.prelin_sphere import *
 20 | from dataload.augmentations import * 
 21 | 
 22 | 
 23 | class Dataset_train(Dataset):
 24 |     def __init__(self,oasis_data_list,train_data_list,transform,crop_size):
 25 |         ployhedron_path = './resources/geodesic_polyhedron.vtp'
 26 |         reader = vtk.vtkXMLPolyDataReader()
 27 |         reader.SetFileName(ployhedron_path)
 28 |         reader.Update()
 29 |         self.input_poly_data = reader.GetOutput()
 30 | 
 31 |         self.itk_img_SetOrigin = (0.0, 0.0, 0.0)
 32 | 
 33 |         self.normal_file_path = np.asarray(oasis_data_list.iloc[:,0])
 34 |         self.datafile_path = np.asarray(train_data_list.iloc[:,0])
 35 |         self.flair_file_path = np.asarray(train_data_list.iloc[:,1])
 36 | 
 37 |         self.gen_file_path = np.concatenate((self.flair_file_path ,self.normal_file_path))
 38 |         self.data_len_gen = len(self.gen_file_path)
 39 |         self.data_len_normal = len(self.normal_file_path)
 40 |         self.data_len = len(self.datafile_path)
 41 |         self.crop_size = crop_size
 42 | 
 43 |         # self.trans = tio.RandomBlur()
 44 |         self.inten = True
 45 |         self.trans =tio.Compose((
 46 |                 tio.RandomElasticDeformation(num_control_points=7, locked_borders=2),
 47 |                 tio.RandomBlur()))
 48 | 
 49 |     def __getitem__(self,index):
 50 |         normal_index = random.randint(0,self.data_len_normal-1)
 51 |         normal_img_path = self.normal_file_path[normal_index]
 52 |         itk_img_normal = sitk.ReadImage(normal_img_path)
 53 |         itk_img_normal.SetOrigin(self.itk_img_SetOrigin)
 54 |         normal_array = sitk.GetArrayFromImage(itk_img_normal)
 55 |         # normal_array = self.aug_sample(normal_array)
 56 |         normal_mean = np.mean(normal_array[np.nonzero(normal_array)])
 57 | 
 58 |         gen_tumor_index = random.randint(0,self.data_len_gen-1)
 59 |         gen_img_path = self.gen_file_path[gen_tumor_index]
 60 |         itk_img_gen = sitk.ReadImage(gen_img_path)
 61 |         itk_img_gen.SetOrigin(self.itk_img_SetOrigin)
 62 |         gen_array = sitk.GetArrayFromImage(itk_img_gen)
 63 | 
 64 |         tumor_mask = self.tumor_array_generate(gen_array,itk_img_gen,self.input_poly_data)
 65 |         tumor_gen = self.tumor_generate(gen_array,tumor_mask,normal_mean,self.trans)
 66 |         normal_crop,tumor_crop,mask_crop = self.random_crop(normal_array,gen_array,tumor_gen,tumor_mask)
 67 |         mask_crop[normal_crop==0]=0
 68 |         mix_crop,mix_tumor = self.mix_array(normal_crop,tumor_crop,mask_crop)
 69 | 
 70 |         normal_crop = self.normalize_mri(normal_crop)
 71 |         tumor_norm = self.normalize_mri(mix_tumor)
 72 |         mix_crop = self.normalize_mri(mix_crop)
 73 | 
 74 |         x_normal = normal_crop[np.newaxis,:,:,:]
 75 |         x_tumor = tumor_norm[np.newaxis,:,:,:]
 76 |         x_train = mix_crop[np.newaxis,:,:,:]
 77 |         x_mask = mask_crop[np.newaxis,:,:,:]
 78 | 
 79 |         return (torch.tensor(x_train.copy(), dtype=torch.float),
 80 |                 torch.tensor(x_normal.copy(), dtype=torch.float),
 81 |                 torch.tensor(x_tumor.copy(), dtype=torch.float),
 82 |                 torch.tensor(x_mask.copy(), dtype=torch.float))
 83 |     
 84 | 
 85 |     def tumor_array_generate(self,img_array, itk_img, input_poly_data):
 86 |         brain_start_x,brain_end_x,brain_start_y,brain_end_y,brain_start_z,brain_end_z = self.crop_img(img_array)
 87 |         ratio_x = np.random.uniform(0.3,0.8)
 88 |         ratio_y = np.random.uniform(0.2,0.7)
 89 |         if np.random.random()< 0.5:
 90 |             ratio_z = np.random.uniform(0.24,0.44)
 91 |         else:
 92 |             ratio_z = np.random.uniform(0.56,0.76)
 93 |         
 94 |         tumor_center_x = int((brain_end_x-brain_start_x)*ratio_x + brain_start_x)
 95 |         tumor_center_y = int((brain_end_y-brain_start_y)*ratio_y + brain_start_y)
 96 |         tumor_center_z = int((brain_end_z-brain_start_z)*ratio_z + brain_start_z)
 97 |         center = [tumor_center_z,tumor_center_y,tumor_center_x]
 98 | 
 99 |         k = 4/3 * np.pi
100 |         volume = np.random.randint(7000, 200000)
101 |         radius = (volume / k) ** (1/3)
102 |         ratio = np.random.uniform(0.8, 1)
103 |         a = radius
104 |         b = radius / ratio
105 |         c = radius * ratio
106 | 
107 |         radii = c,b,a
108 |         angles = np.random.uniform(0, 180, size=3)
109 | 
110 |         octaves = np.random.randint(4, 8)
111 |         offset = np.random.randint(1000)
112 |         scale = 0.5
113 | 
114 |         output_poly_data = get_resection_poly_data(input_poly_data,offset,center,radii,angles,octaves,scale)
115 |         output_poly_stk = pd_to_itk_image(output_poly_data,itk_img)
116 |         output_poly_array = sitk.GetArrayFromImage(output_poly_stk)
117 |         output_poly_array[output_poly_array>0]=1
118 |         return output_poly_array
119 | 
120 | 
121 |     def crop_img(self,img_array):
122 |         zx = np.any(img_array, axis=(1,2))
123 |         start_slicex, end_slicex = np.where(zx)[0][[0, -1]]
124 |         zy = np.any(img_array, axis=(0,2))
125 |         start_slicey, end_slicey = np.where(zy)[0][[0, -1]]   
126 |         zz = np.any(img_array, axis=(0,1))
127 |         start_slicez, end_slicez = np.where(zz)[0][[0, -1]]
128 |         return start_slicex, end_slicex, start_slicey, end_slicey, start_slicez, end_slicez
129 |     
130 |     def aug_sample(self,x):
131 |         if random.random() < 0.5:
132 |             x = np.flip(x, axis=0)
133 |         if random.random() < 0.5:
134 |             x = np.flip(x, axis=1)
135 |         if random.random() < 0.5:
136 |             x = np.flip(x, axis=2)
137 |         return x
138 |     
139 |     def mix_array(self,normal,tumor,mask):
140 |         mix_ratio_value = random.uniform(0.2,0.8)
141 |         mix_mask = mask*mix_ratio_value
142 |         mix_array = np.multiply(mix_mask, tumor) + np.multiply((1-mix_mask), normal)
143 |         tumor[mask!=0]=mix_array[mask!=0]
144 |         return mix_array,tumor
145 |     
146 |     def normalize_mri(self,npimg):
147 |         image_nonzero = npimg[np.nonzero(npimg)]
148 |         if np.mean(npimg)==0:
149 |             print('tumor is 0!')
150 |             tmp =  npimg
151 |         elif np.std(image_nonzero) < 0.01: 
152 |             print('nonzero std is 0!')
153 |             tmp =  (npimg - np.min(npimg))/(np.max(npimg)-np.min(npimg))
154 |         else: 
155 |             tmp=(npimg - np.mean(image_nonzero))/ np.std(image_nonzero)
156 |         return tmp
157 | 
158 |     def tumor_generate(self,img_array,tumor_mask,normal_mean,trans):
159 |         tumor_gen = img_array.copy()[np.newaxis,:,:,:]
160 |         tumor_aug = np.array(trans(tumor_gen))[0]
161 |         tumor_aug[tumor_mask==0]=0
162 |         tumor_mean = np.mean(tumor_aug[np.nonzero(tumor_aug)])
163 |         if self.inten:
164 |             intensity_index = random.uniform(1.0,3.0)
165 |             tumor_aug = tumor_aug*(normal_mean*intensity_index/tumor_mean)
166 |         else:
167 |             tumor_aug  = tumor_aug
168 |         return tumor_aug
169 |     
170 |     def random_crop(self, img, gen, tumor, mask):
171 |         zx = np.any(img, axis=(1,2))
172 |         start_slicex, end_slicex = np.where(zx)[0][[0, -1]]
173 |         zy = np.any(img, axis=(0,2))
174 |         start_slicey, end_slicey = np.where(zy)[0][[0, -1]]
175 |         zz = np.any(img, axis=(0,1))
176 |         start_slicez, end_slicez = np.where(zz)[0][[0, -1]]
177 | 
178 |         normal_img = img[start_slicex:end_slicex, start_slicey:end_slicey,start_slicez:end_slicez]
179 |         
180 |         zx_gen = np.any(gen, axis=(1,2))
181 |         start_slicex_gen, end_slicex_gen = np.where(zx_gen)[0][[0, -1]]
182 |         zy_gen = np.any(gen, axis=(0,2))
183 |         start_slicey_gen, end_slicey_gen = np.where(zy_gen)[0][[0, -1]]
184 |         zz_gen = np.any(gen, axis=(0,1))
185 |         start_slicez_gen, end_slicez_gen = np.where(zz_gen)[0][[0, -1]]
186 |         tumor_img = tumor[start_slicex_gen:end_slicex_gen, start_slicey_gen:end_slicey_gen,start_slicez_gen:end_slicez_gen]
187 |         tumor_mask = mask[start_slicex_gen:end_slicex_gen, start_slicey_gen:end_slicey_gen,start_slicez_gen:end_slicez_gen]
188 | 
189 |         normal_img = resize(normal_img,self.crop_size,order=1,preserve_range=True)
190 |         tumor_img = resize(tumor_img,self.crop_size,order=0,preserve_range=True)
191 |         tumor_mask = resize(tumor_mask,self.crop_size,order=0,preserve_range=True)
192 |         return normal_img.astype(np.int16), tumor_img.astype(np.int16), tumor_mask.astype(np.int16)
193 | 
194 |     def __len__(self):
195 |         return len(self.datafile_path)
196 | 


--------------------------------------------------------------------------------
/BT_CODE/evaluate.py:
--------------------------------------------------------------------------------
  1 | import argparse
  2 | import os
  3 | import time
  4 | import logging
  5 | import sys
  6 | import math
  7 | import csv
  8 | import numpy as np
  9 | import pandas as pd
 10 | 
 11 | import SimpleITK as sitk
 12 | import torch
 13 | import torch.nn as nn
 14 | import torch.nn.functional as F
 15 | import torch.nn.parallel
 16 | import torch.backends.cudnn as cudnn
 17 | import torch.optim
 18 | from torch.optim.lr_scheduler import ReduceLROnPlateau
 19 | import torch.utils.data
 20 | from tensorboardX import SummaryWriter
 21 | from torchsummary import summary
 22 | from skimage.transform import resize
 23 | 
 24 | from utils.metrics import calc_all_metrics,calculate_metric_percase
 25 | from model import unet3d
 26 | 
 27 | 
 28 | parser = argparse.ArgumentParser(description='evaluate model path')
 29 | #dataset path
 30 | parser.add_argument('--train_csv_dir', default='./data_csv/BraTS.csv',help='dataset path')
 31 | parser.add_argument('--test_csv_dir', default='./data_csv/brats',help='dataset path')
 32 | parser.add_argument('--crop_size',default=[128,128,128])
 33 | parser.add_argument('--fold', default=0, type=int)
 34 | 
 35 | parser.add_argument('-j', '--num_workers', default=8, type=int, metavar='N',
 36 |                     help='number of data loading workers (default: 4)')
 37 | parser.add_argument('-b', '--batch_size', default=1, type=int,
 38 |                     metavar='N', help='mini-batch size (default: 2)')
 39 | 
 40 | parser.add_argument('--model_save_path', type=str, default='../model/best_valid_model.pt',
 41 |                     help='Name of Experiment')
 42 | parser.add_argument('--result_save_path', type=str, default='../result',
 43 |                     help='Name of Experiment')
 44 | parser.add_argument('--snapshot_path', type=str, default='fold_0',
 45 |                     help='Name of Experiment')
 46 | parser.add_argument('--gpu', type=str,default='0', help='gpu') 
 47 | args = parser.parse_args()
 48 | 
 49 | os.environ["CUDA_VISIBLE_DEVICES"] = args.gpu
 50 | torch.cuda.current_device()
 51 | torch.cuda._initialized = True
 52 | 
 53 | def normalize_mri(npimg):
 54 |     image_nonzero = npimg[np.nonzero(npimg)]
 55 |     tmp=(npimg - np.mean(image_nonzero))/ np.std(image_nonzero)
 56 |     return tmp
 57 | 
 58 | 
 59 | def dice_coef(input,target):
 60 |     """soft dice loss"""
 61 |     eps = 1e-7
 62 |     intersection = np.multiply(input,target).sum()
 63 |     return  (2. * intersection+eps) / (input.sum() +target.sum() + eps)
 64 | 
 65 | def reshape_tumor(img,crop_axis,crop_shape,img_shape):
 66 |     normal_img = resize(img,crop_shape,order=0,preserve_range=True)
 67 |     start_slicex,end_slicex,start_slicey,end_slicey,start_slicez,end_slicez = crop_axis
 68 |     output_img = np.zeros(img_shape)
 69 |     output_img[start_slicex:end_slicex, start_slicey:end_slicey,start_slicez:end_slicez] = normal_img
 70 |     return output_img
 71 | 
 72 | def crop_brain(img,seg):
 73 |     #  crop_brain
 74 |     zx = np.any(img, axis=(1,2))
 75 |     start_slicex, end_slicex = np.where(zx)[0][[0, -1]]
 76 |     zy = np.any(img, axis=(0,2))
 77 |     start_slicey, end_slicey = np.where(zy)[0][[0, -1]]
 78 |     zz = np.any(img, axis=(0,1))
 79 |     start_slicez, end_slicez = np.where(zz)[0][[0, -1]]
 80 |     normal_img = img[start_slicex:end_slicex, start_slicey:end_slicey,start_slicez:end_slicez]
 81 |     crop_seg = seg[start_slicex:end_slicex, start_slicey:end_slicey,start_slicez:end_slicez]
 82 |     normal_img = resize(normal_img,args.crop_size,order=1,preserve_range=True)
 83 |     crop_seg = resize(crop_seg,args.crop_size,order=0,preserve_range=True)
 84 |     return normal_img,crop_seg
 85 | 
 86 | def crop_brain_dice(img):
 87 |     zx = np.any(img, axis=(1,2))
 88 |     start_slicex, end_slicex = np.where(zx)[0][[0, -1]]
 89 |     zy = np.any(img, axis=(0,2))
 90 |     start_slicey, end_slicey = np.where(zy)[0][[0, -1]]
 91 |     zz = np.any(img, axis=(0,1))
 92 |     start_slicez, end_slicez = np.where(zz)[0][[0, -1]]
 93 |     normal_img = img[start_slicex:end_slicex, start_slicey:end_slicey,start_slicez:end_slicez]
 94 |     crop_axis = [start_slicex,end_slicex,start_slicey,end_slicey,start_slicez,end_slicez]
 95 |     crop_shape = normal_img.shape
 96 |     normal_img = resize(normal_img,args.crop_size,order=1,preserve_range=True)
 97 |     return normal_img,crop_axis,crop_shape
 98 | 
 99 | def test():
100 |     if not os.path.exists(args.result_save_path):
101 |         os.makedirs(args.result_save_path)
102 |     save_result_path = os.path.join(args.result_save_path,args.snapshot_path)
103 |     if not os.path.exists(save_result_path):
104 |         os.makedirs(save_result_path)
105 |     device=torch.device('cuda' if torch.cuda.is_available() else 'cpu')
106 |     print("Total CUDA devices: ", torch.cuda.device_count()) 
107 | 
108 |     test_dir = os.path.join(args.test_csv_dir,'test_'+str(args.fold)+'.csv')
109 |     test_data_list = pd.read_csv(test_dir)
110 |     flair_file_path = np.asarray(test_data_list.iloc[:,1])
111 |     seg_file_path = np.asarray(test_data_list.iloc[:,5])
112 | 
113 |     model = unet3d.UNet3D(n_class=3, act='relu')
114 |     model = nn.DataParallel(model, device_ids = [i for i in range(torch.cuda.device_count())])
115 |     model.to(device)
116 | 
117 |     checkpoint = torch.load(args.model_save_path)
118 |     state_dict = checkpoint['state_dict']
119 |     model.load_state_dict(state_dict)  
120 |     print('load successful') 
121 | 
122 |     for index in range(len(test_data_list)):
123 |         img_path = flair_file_path[index]
124 |         itk_img = sitk.ReadImage(img_path)
125 |         img_array = sitk.GetArrayFromImage(itk_img)
126 |         seg_path = seg_file_path[index]
127 |         itk_img_seg = sitk.ReadImage(seg_path)
128 |         seg_array = sitk.GetArrayFromImage(itk_img_seg)
129 | 
130 |         img_brain,img_seg = crop_brain(img_array,seg_array)
131 |         img_brain_norm = normalize_mri(img_brain)
132 |         img_array = img_brain_norm[np.newaxis,np.newaxis,:,:,:]
133 |         input_patch = torch.tensor(img_array.copy(), dtype=torch.float).to(device)
134 |         with torch.no_grad():
135 |             model.eval()
136 |             out_patch = model(input_patch)
137 |             output_normal = F.sigmoid(out_patch[0,0,:,:,:]).cpu().detach().numpy()
138 |             output_tumor = F.sigmoid(out_patch[0,1,:,:,:]).cpu().detach().numpy()
139 |             mask_tumor = F.sigmoid(out_patch[0,2,:,:,:]).cpu().detach().numpy()
140 |         output_save = sitk.GetImageFromArray((output_normal*255).astype(np.int16))
141 |         sitk.WriteImage(output_save,os.path.join(save_result_path,'output_normal_'+str(index)+'.nii.gz'))
142 |         output_save = sitk.GetImageFromArray((output_tumor*255).astype(np.int16))
143 |         sitk.WriteImage(output_save,os.path.join(save_result_path,'output_tumor_'+str(index)+'.nii.gz'))
144 |         output_save = sitk.GetImageFromArray((mask_tumor*255).astype(np.int16))
145 |         sitk.WriteImage(output_save,os.path.join(save_result_path,'output_mask_'+str(index)+'.nii.gz'))
146 |         output_save = sitk.GetImageFromArray((img_brain).astype(np.int16))
147 |         sitk.WriteImage(output_save,os.path.join(save_result_path,'input_'+str(index)+'.nii.gz'))
148 |         output_save = sitk.GetImageFromArray((img_seg).astype(np.int16))
149 |         sitk.WriteImage(output_save,os.path.join(save_result_path,'seg_'+str(index)+'.nii.gz'))
150 |         print(index)
151 | 
152 | 
153 | def dice_cacluate(gate):
154 |     if not os.path.exists(args.result_save_path):
155 |         os.makedirs(args.result_save_path)
156 |     save_result_path = args.result_save_path
157 |     csv_path = os.path.join(save_result_path,str(args.fold)+'_metric.csv')
158 |     dist_csv_col = [
159 |     'Index',
160 |     'Dice',
161 |     'Jaccard',
162 |     'Hausdorff',
163 |     'ASD',
164 |     'Sensitivity',
165 |     'Specificity'
166 |     ]
167 |     f_train = open(csv_path,'w+',newline='')
168 |     wf_train = csv.writer(f_train)
169 |     wf_train.writerow(dist_csv_col)
170 | 
171 |     device=torch.device('cuda' if torch.cuda.is_available() else 'cpu')
172 |     print("Total CUDA devices: ", torch.cuda.device_count()) 
173 | 
174 |     test_dir = os.path.join(args.test_csv_dir,'test_'+str(args.fold)+'.csv')
175 |     test_data_list = pd.read_csv(test_dir)
176 |     flair_file_path = np.asarray(test_data_list.iloc[:,1])
177 |     seg_file_path = np.asarray(test_data_list.iloc[:,5])
178 | 
179 |     model = unet3d.UNet3D(n_class=3, act='relu')
180 |     model = nn.DataParallel(model, device_ids = [i for i in range(torch.cuda.device_count())])
181 |     model.to(device)
182 | 
183 |     checkpoint = torch.load(args.model_save_path)
184 |     state_dict = checkpoint['state_dict']
185 |     model.load_state_dict(state_dict)  
186 |     print('load successful') 
187 | 
188 |     dice_sum = []
189 |     JC_sum = []
190 |     HD_sum = []
191 |     ASD_sum = []
192 |     SEN_sum = []
193 |     SPEC_sum = []
194 | 
195 |     for index in range(len(test_data_list)):
196 |         img_path = flair_file_path[index]
197 |         itk_img = sitk.ReadImage(img_path)
198 |         img_array = sitk.GetArrayFromImage(itk_img)
199 | 
200 |         seg_path = seg_file_path[index]
201 |         itk_img_seg = sitk.ReadImage(seg_path)
202 |         seg_array = sitk.GetArrayFromImage(itk_img_seg)
203 |         seg_array[seg_array>0]=1
204 |         img_shape = seg_array.shape
205 | 
206 |         img_brain,crop_axis,resize_shape = crop_brain_dice(img_array)
207 |         img_brain_norm = normalize_mri(img_brain)
208 |         img_array = img_brain_norm[np.newaxis,np.newaxis,:,:,:]
209 |         input_patch = torch.tensor(img_array.copy(), dtype=torch.float).to(device)
210 |         with torch.no_grad():
211 |             model.eval()
212 |             out_patch = model(input_patch)
213 |             output_normal = out_patch[0,0,:,:,:].cpu().detach().numpy()
214 |             output_tumor = F.sigmoid(out_patch[0,1,:,:,:]).cpu().detach().numpy()
215 |             mask_tumor = F.sigmoid(out_patch[0,2,:,:,:]).cpu().detach().numpy()
216 |         output_tumor[output_tumor>gate]=1
217 |         output_tumor[output_tumor<gate]=0
218 |         tumor_seg = reshape_tumor(output_tumor,crop_axis,resize_shape,img_shape)
219 | 
220 |         dice_index = dice_coef(tumor_seg,seg_array)
221 |         dice, jc, hd, asd, sen, spec = calculate_metric_percase(tumor_seg,seg_array)
222 |         dice_sum.append(dice_index) 
223 |         JC_sum.append(jc)
224 |         HD_sum.append(hd)
225 |         ASD_sum.append(asd)
226 |         SEN_sum.append(sen)
227 |         SPEC_sum.append(spec)
228 | 
229 |         result_wirte = []
230 |         result_wirte.append(index)
231 |         result_wirte.append(dice)
232 |         result_wirte.append(jc)
233 |         result_wirte.append(hd)
234 |         result_wirte.append(asd)
235 |         result_wirte.append(sen)
236 |         result_wirte.append(spec)
237 |         wf_train.writerow(result_wirte)
238 |         print(index,dice_index,result_wirte)
239 |     mean_dice = np.mean(np.array(dice_sum))
240 |     mean_JC = np.mean(np.array(JC_sum))
241 |     mean_HD = np.mean(np.array(HD_sum))
242 |     mean_ASD = np.mean(np.array(ASD_sum))
243 |     mean_SEN = np.mean(np.array(SEN_sum))
244 |     mean_SPEC = np.mean(np.array(SPEC_sum))
245 | 
246 |     result_wirte = []
247 |     result_wirte.append('mean')
248 |     result_wirte.append(mean_dice)
249 |     result_wirte.append(mean_JC)
250 |     result_wirte.append(mean_HD)
251 |     result_wirte.append(mean_ASD)
252 |     result_wirte.append(mean_SEN)
253 |     result_wirte.append(mean_SPEC)
254 |     wf_train.writerow(result_wirte)
255 |     f_train.close()
256 | 
257 | if __name__ == '__main__':
258 |     test()
259 |     dice_cacluate(gate=0.5)


--------------------------------------------------------------------------------
/BT_CODE/model/unet3d.py:
--------------------------------------------------------------------------------
  1 | import torch
  2 | import torch.nn as nn
  3 | import torch.nn.functional as F
  4 | from torchsummary import summary
  5 | import os 
  6 | 
  7 | class ContBatchNorm3d(nn.modules.batchnorm._BatchNorm):
  8 |     def _check_input_dim(self, input):
  9 | 
 10 |         if input.dim() != 5:
 11 |             raise ValueError('expected 5D input (got {}D input)'.format(input.dim()))
 12 |         #super(ContBatchNorm3d, self)._check_input_dim(input)
 13 | 
 14 |     def forward(self, input):
 15 |         self._check_input_dim(input)
 16 |         return F.batch_norm(
 17 |             input, self.running_mean, self.running_var, self.weight, self.bias,
 18 |             True, self.momentum, self.eps)
 19 | 
 20 | class LUConv(nn.Module):
 21 |     def __init__(self, in_chan, out_chan, act):
 22 |         super(LUConv, self).__init__()
 23 |         self.conv1 = nn.Conv3d(in_chan, out_chan, kernel_size=3, padding=1)
 24 |         self.bn1 = ContBatchNorm3d(out_chan)
 25 | 
 26 |         if act == 'relu':
 27 |             self.activation = nn.ReLU(out_chan)
 28 |         elif act == 'prelu':
 29 |             self.activation = nn.PReLU(out_chan)
 30 |         elif act == 'elu':
 31 |             self.activation = nn.ELU(inplace=True)
 32 |         else:
 33 |             raise
 34 | 
 35 |     def forward(self, x):
 36 |         out = self.activation(self.bn1(self.conv1(x)))
 37 |         return out
 38 | 
 39 | def _make_nConv(in_channel, depth, act, double_chnnel=False):
 40 |     if double_chnnel:
 41 |         layer1 = LUConv(in_channel, 32 * (2 ** (depth+1)),act)
 42 |         layer2 = LUConv(32 * (2 ** (depth+1)), 32 * (2 ** (depth+1)),act)
 43 |     else:
 44 |         layer1 = LUConv(in_channel, 32*(2**depth),act)
 45 |         layer2 = LUConv(32*(2**depth), 32*(2**depth)*2,act)
 46 | 
 47 |     return nn.Sequential(layer1,layer2)
 48 | 
 49 | class DownTransition(nn.Module):
 50 |     def __init__(self, in_channel,depth, act):
 51 |         super(DownTransition, self).__init__()
 52 |         self.ops = _make_nConv(in_channel, depth,act)
 53 |         self.maxpool = nn.MaxPool3d(2)
 54 |         self.current_depth = depth
 55 | 
 56 |     def forward(self, x):
 57 |         if self.current_depth == 3:
 58 |             out = self.ops(x)
 59 |             out_before_pool = out
 60 |         else:
 61 |             out_before_pool = self.ops(x)
 62 |             out = self.maxpool(out_before_pool)
 63 |         return out, out_before_pool
 64 | 
 65 | class UpTransition(nn.Module):
 66 |     def __init__(self, inChans, outChans, depth,act):
 67 |         super(UpTransition, self).__init__()
 68 |         self.depth = depth
 69 |         self.up_conv = nn.ConvTranspose3d(inChans, outChans, kernel_size=2, stride=2)
 70 |         self.ops = _make_nConv(inChans+ outChans//2,depth, act, double_chnnel=True)
 71 | 
 72 |     def forward(self, x, skip_x):
 73 |         out_up_conv = self.up_conv(x)
 74 |         concat = torch.cat((out_up_conv,skip_x),1)
 75 |         out = self.ops(concat)
 76 |         return out
 77 | 
 78 | class OutputTransition(nn.Module):
 79 |     def __init__(self, inChans, n_labels):
 80 | 
 81 |         super(OutputTransition, self).__init__()
 82 |         self.final_conv = nn.Conv3d(inChans, n_labels, kernel_size=1)
 83 |         # self.sigmoid = nn.Sigmoid()
 84 | 
 85 |     def forward(self, x):
 86 |         out = self.final_conv(x)# self.sigmoid(self.final_conv(x))
 87 |         return out
 88 | 
 89 | class UNet3D(nn.Module):
 90 |     # the number of convolutions in each layer corresponds
 91 |     # to what is in the actual prototxt, not the intent
 92 |     def __init__(self, n_class=1, act='relu'):
 93 |         super(UNet3D, self).__init__()
 94 | 
 95 |         self.down_tr64 = DownTransition(1,0,act)
 96 |         self.down_tr128 = DownTransition(64,1,act)
 97 |         self.down_tr256 = DownTransition(128,2,act)
 98 |         self.down_tr512 = DownTransition(256,3,act)
 99 | 
100 |         self.up_tr256 = UpTransition(512, 512,2,act)
101 |         self.up_tr128 = UpTransition(256,256, 1,act)
102 |         self.up_tr64 = UpTransition(128,128,0,act)
103 |         self.out_tr = OutputTransition(64, n_class)
104 | 
105 |     def forward(self, x):
106 |         self.out64, self.skip_out64 = self.down_tr64(x)
107 |         self.out128,self.skip_out128 = self.down_tr128(self.out64)
108 |         self.out256,self.skip_out256 = self.down_tr256(self.out128)
109 |         self.out512,self.skip_out512 = self.down_tr512(self.out256)
110 | 
111 |         self.out_up_256 = self.up_tr256(self.out512,self.skip_out256)
112 |         self.out_up_128 = self.up_tr128(self.out_up_256, self.skip_out128)
113 |         self.out_up_64 = self.up_tr64(self.out_up_128, self.skip_out64)
114 |         self.out = self.out_tr(self.out_up_64)
115 | 
116 |         return self.out
117 | 


--------------------------------------------------------------------------------
/BT_CODE/requirements.txt:
--------------------------------------------------------------------------------
  1 | absl-py @ file:///tmp/build/80754af9/absl-py_1615411197583/work
  2 | albumentations==0.4.5
  3 | appdirs==1.4.4
  4 | asn1crypto==1.2.0
  5 | astor==0.8.1
  6 | astroid==2.4.2
  7 | Augmentor==0.2.8
  8 | autopep8==1.5.3
  9 | backports.functools-lru-cache @ file:///tmp/build/80754af9/backports.functools_lru_cache_1605305165209/work
 10 | backports.tempfile @ file:///home/linux1/recipes/ci/backports.tempfile_1610991236607/work
 11 | backports.weakref==1.0.post1
 12 | batchgenerators==0.21
 13 | beautifulsoup4 @ file:///home/linux1/recipes/ci/beautifulsoup4_1610988766420/work
 14 | bleach==1.5.0
 15 | blessed==1.18.0
 16 | blessings==1.7
 17 | blinker==1.4
 18 | brotlipy==0.7.0
 19 | cached-property==1.5.2
 20 | cachetools==4.2.1
 21 | certifi==2020.12.5
 22 | cffi @ file:///tmp/build/80754af9/cffi_1613246439577/work
 23 | chainer==6.5.0
 24 | chainercv==0.13.1
 25 | chardet @ file:///tmp/build/80754af9/chardet_1607706739153/work
 26 | click @ file:///home/linux1/recipes/ci/click_1610990599742/work
 27 | cloudpickle @ file:///tmp/build/80754af9/cloudpickle_1598884132938/work
 28 | cmake==3.17.3
 29 | colorlog==4.0.2
 30 | conda==4.9.2
 31 | conda-build==3.0.27
 32 | conda-package-handling @ file:///tmp/build/80754af9/conda-package-handling_1603018143336/work
 33 | conda-verify==3.4.2
 34 | configparser==5.0.0
 35 | coverage @ file:///tmp/build/80754af9/coverage_1614614208500/work
 36 | cryptography @ file:///tmp/build/80754af9/cryptography_1615532099045/work
 37 | cycler==0.10.0
 38 | Cython @ file:///tmp/build/80754af9/cython_1614014843172/work
 39 | cytoolz==0.11.0
 40 | dask @ file:///tmp/build/80754af9/dask-core_1615055117017/work
 41 | dataclasses==0.8
 42 | decorator @ file:///home/ktietz/src/ci/decorator_1611930055503/work
 43 | Deprecated==1.2.12
 44 | easydict==1.9
 45 | efficientnet-pytorch==0.6.3
 46 | einops==0.3.0
 47 | elasticdeform==0.4.6
 48 | faiss==1.5.3
 49 | faiss-gpu==1.6.3
 50 | fcn==6.4.17
 51 | filelock @ file:///home/linux1/recipes/ci/filelock_1610993975404/work
 52 | flake8==3.8.2
 53 | flyingcircus==0.1.4.0
 54 | flyingcircus-numeric==0.1.1.2
 55 | future==0.18.2
 56 | gast @ file:///tmp/build/80754af9/gast_1597433534803/work
 57 | gdown==3.8.3
 58 | glob2 @ file:///home/linux1/recipes/ci/glob2_1610991677669/work
 59 | google-auth==1.28.0
 60 | google-auth-oauthlib==0.4.3
 61 | google-pasta==0.2.0
 62 | gpustat==0.6.0
 63 | grpcio @ file:///tmp/build/80754af9/grpcio_1614884176509/work
 64 | h5py==3.1.0
 65 | html5lib==0.9999999
 66 | humanize==3.7.0
 67 | idna @ file:///home/linux1/recipes/ci/idna_1610986105248/work
 68 | imageio @ file:///tmp/build/80754af9/imageio_1594161405741/work
 69 | imgaug==0.4.0
 70 | importlib-metadata==3.9.0
 71 | isort==4.3.21
 72 | Jinja2 @ file:///tmp/build/80754af9/jinja2_1612213139570/work
 73 | joblib @ file:///tmp/build/80754af9/joblib_1613502643832/work
 74 | jsonpatch==1.32
 75 | jsonpointer==2.1
 76 | kaggle==1.5.6
 77 | Keras==2.3.1
 78 | Keras-Applications==1.0.8
 79 | Keras-Preprocessing==1.1.2
 80 | kiwisolver @ file:///tmp/build/80754af9/kiwisolver_1612282412546/work
 81 | lazy-object-proxy==1.4.3
 82 | linecache2==1.0.0
 83 | Markdown @ file:///tmp/build/80754af9/markdown_1614363833670/work
 84 | MarkupSafe==1.1.1
 85 | matplotlib @ file:///tmp/build/80754af9/matplotlib-suite_1613407855456/work
 86 | mccabe==0.6.1
 87 | MedPy==0.4.0
 88 | mkl-fft==1.3.0
 89 | mkl-random==1.1.1
 90 | mkl-service==2.3.0
 91 | mock==4.0.3
 92 | mpmath==1.2.1
 93 | munch==2.5.0
 94 | networkx @ file:///tmp/build/80754af9/networkx_1598376031484/work
 95 | nibabel==2.5.1
 96 | NiftyNet==0.6.0
 97 | noise==1.2.2
 98 | nose==1.3.7
 99 | numeral==0.1.0.17
100 | numex==0.0.0.7
101 | numpy==1.19.5
102 | nvidia-ml-py3==7.352.0
103 | oauthlib==3.1.0
104 | olefile==0.46
105 | opencv-python==4.1.2.30
106 | opt-einsum==3.3.0
107 | packaging==20.4
108 | pandas==1.1.5
109 | patsy==0.5.1
110 | Pillow==7.1.2
111 | pip-conflict-checker==0.6.0
112 | pkginfo==1.7.0
113 | pretrainedmodels==0.7.4
114 | progress==1.5
115 | protobuf==3.15.6
116 | psutil==5.7.0
117 | pyasn1==0.4.8
118 | pyasn1-modules==0.2.8
119 | pycodestyle==2.6.0
120 | pycosat==0.6.3
121 | pycparser @ file:///tmp/build/80754af9/pycparser_1594388511720/work
122 | pydicom==1.4.2
123 | pyflakes==2.2.0
124 | pylidc==0.2.2
125 | pylint==2.5.3
126 | pymrt==0.0.3.4
127 | pynrrd==0.4.1
128 | pyOpenSSL @ file:///tmp/build/80754af9/pyopenssl_1608057966937/work
129 | pyparsing @ file:///home/linux1/recipes/ci/pyparsing_1610983426697/work
130 | PySocks @ file:///tmp/build/80754af9/pysocks_1605305763431/work
131 | python-dateutil @ file:///home/ktietz/src/ci/python-dateutil_1611928101742/work
132 | python-slugify==4.0.0
133 | pytk==0.0.2.1
134 | pytz @ file:///tmp/build/80754af9/pytz_1612215392582/work
135 | PyWavelets @ file:///tmp/build/80754af9/pywavelets_1601658315997/work
136 | PyYAML==5.4.1
137 | pyzmq==22.0.3
138 | raster-geometry==0.1.4.1
139 | requests @ file:///tmp/build/80754af9/requests_1608241421344/work
140 | requests-oauthlib==1.3.0
141 | retrying==1.3.3
142 | rsa==4.7.2
143 | ruamel-yaml==0.15.87
144 | ruamel.yaml.clib==0.2.0
145 | scikit-image==0.16.2
146 | scikit-learn @ file:///tmp/build/80754af9/scikit-learn_1592502882723/work
147 | scipy @ file:///tmp/build/80754af9/scipy_1597686625380/work
148 | seaborn==0.9.0
149 | segmentation-models-pytorch==0.1.0
150 | setuptools-scm==6.0.1
151 | Shapely==1.7.1
152 | SimpleITK==1.2.4
153 | six @ file:///tmp/build/80754af9/six_1605205335545/work
154 | soupsieve @ file:///tmp/build/80754af9/soupsieve_1614023247537/work
155 | SQLAlchemy==1.3.18
156 | statsmodels @ file:///tmp/build/80754af9/statsmodels_1614023745203/work
157 | sympy==1.8
158 | tensorboard==2.4.1
159 | tensorboard-plugin-wit==1.8.0
160 | tensorboardX==2.2
161 | tensorflow==1.13.1
162 | tensorflow-estimator==1.13.0
163 | tensorflow-gpu==1.13.1
164 | tensorrt==0.0.1
165 | termcolor==1.1.0
166 | text-unidecode==1.3
167 | tflearn==0.3.2
168 | thop==0.0.31.post2005241907
169 | threadpoolctl @ file:///tmp/tmp9twdgx9k/threadpoolctl-2.1.0-py3-none-any.whl
170 | toml==0.10.1
171 | toolz @ file:///home/linux1/recipes/ci/toolz_1610987900194/work
172 | torch==1.8.0+cu111
173 | torch-tb-profiler==0.1.0
174 | torchaudio==0.8.0
175 | torchcontrib==0.0.2
176 | torchfcn==1.9.6
177 | torchfile==0.1.0
178 | torchio==0.18.40
179 | torchstat==0.0.7
180 | torchsummary==1.5.1
181 | torchvision==0.9.0+cu111
182 | tornado @ file:///tmp/build/80754af9/tornado_1606942266872/work
183 | tqdm @ file:///tmp/build/80754af9/tqdm_1615925068909/work
184 | traceback2==1.4.0
185 | typed-ast==1.4.1
186 | typing==3.6.6
187 | typing-extensions @ file:///home/ktietz/src/ci_mi/typing_extensions_1612808209620/work
188 | unittest2==1.1.0
189 | urllib3 @ file:///tmp/build/80754af9/urllib3_1615837158687/work
190 | visdom==0.1.8.9
191 | vtk==9.0.1
192 | wcwidth==0.2.5
193 | webencodings==0.5.1
194 | websocket-client==0.58.0
195 | Werkzeug @ file:///home/ktietz/src/ci/werkzeug_1611932622770/work
196 | wget==3.2
197 | wrapt==1.12.1
198 | xlrd==1.2.0
199 | zipp @ file:///tmp/build/80754af9/zipp_1615904174917/work
200 | 


--------------------------------------------------------------------------------
/BT_CODE/resources/geodesic_polyhedron.nii.gz:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/xiaoman-zhang/Layer-Decomposition/d9ddef579b192cce5e2c8cb28d76d4b2cd367a95/BT_CODE/resources/geodesic_polyhedron.nii.gz


--------------------------------------------------------------------------------
/BT_CODE/train.py:
--------------------------------------------------------------------------------
  1 | import argparse
  2 | import os
  3 | import time
  4 | import logging
  5 | import sys
  6 | import numpy as np
  7 | import pandas as pd
  8 | 
  9 | import torch
 10 | import torch.nn as nn
 11 | import torch.nn.parallel
 12 | import torch.backends.cudnn as cudnn
 13 | import torch.optim
 14 | from torch.optim.lr_scheduler import ReduceLROnPlateau
 15 | import torch.utils.data
 16 | from tensorboardX import SummaryWriter
 17 | from torchsummary import summary
 18 | from skimage.transform import resize
 19 | import torch.nn.functional as F
 20 | import SimpleITK as sitk
 21 | 
 22 | 
 23 | from model import unet3d
 24 | from utils.loss import DiceLoss
 25 | from dataload.dataload import *
 26 | 
 27 | parser = argparse.ArgumentParser(description='Layer Decomposition for 3D Tumor Segmentation')
 28 | #dataset path
 29 | #oasis_csv_dir: data from OASIS (used as normal data)
 30 | #train_csv_dir: data from BraTS (can be added after warming up/ used for tumor generater)
 31 | parser.add_argument('--oasis_csv_dir', default='data_csv/OASIS.csv',help='dataset path')
 32 | parser.add_argument('--train_csv_dir', default='data_csv/BraTS.csv',help='dataset path')
 33 | parser.add_argument('--test_csv_dir', default='../DATA/data_csv/brats',help='dataset path')
 34 | parser.add_argument('--fold', default=0, type=int,
 35 |                     metavar='fold', help='train fold')
 36 | parser.add_argument('--transformation', default=0, type=int,
 37 |                     metavar='N', help='transformation  (default: 2)')
 38 | 
 39 | parser.add_argument('--crop_size',default=[128,128,128])
 40 | parser.add_argument('--load'  , type=bool,  default=False,  help='whether loaded')
 41 | parser.add_argument('-j', '--num_workers', default=4, type=int, metavar='N',
 42 |                     help='number of data loading workers (default: 4)')
 43 | parser.add_argument('--epochs', default=500, type=int, metavar='N',
 44 |                     help='number of total epochs to run')
 45 | parser.add_argument('-b', '--batch_size', default=2, type=int,
 46 |                     metavar='N', help='mini-batch size (default: 2)')
 47 | 
 48 | parser.add_argument('--lr', '--learning-rate', default=1e-2, type=float,
 49 |                     metavar='LR', help='initial learning rate')
 50 | parser.add_argument('--momentum', default=0.9, type=float, metavar='M',
 51 |                     help='momentum')
 52 | parser.add_argument('--weight-decay', '--wd', default=1e-4, type=float,
 53 |                     metavar='W', help='weight decay (default: 5e-4)')
 54 | parser.add_argument('--patience', default=40, type=int)
 55 | parser.add_argument('--patience_lr', default=10, type=int)
 56 | 
 57 | 
 58 | parser.add_argument('--model_path', type=str, default='model',
 59 |                     help='Name of Experiment')
 60 | parser.add_argument('--snapshot_path', type=str, default='train',help='Name of Experiment')
 61 | parser.add_argument('--model_name', type=str, default='model.pt',help='Name of Experiment')
 62 | parser.add_argument('--best_test_model_name', type=str, default='best_test_model.pt',help='Name of Experiment')
 63 | parser.add_argument('--best_valid_model_name', type=str, default='best_valid_model.pt',help='Name of Experiment')
 64 | parser.add_argument('--validation_path', type=str, default='model',
 65 |                     help='Name of Experiment')
 66 | parser.add_argument('--gpu', type=str,default='5', help='gpu') 
 67 | args = parser.parse_args()
 68 | 
 69 | os.environ["CUDA_VISIBLE_DEVICES"] = args.gpu
 70 | torch.cuda.current_device()
 71 | torch.cuda._initialized = True
 72 | 
 73 | def train():
 74 |     oasis_data_list = pd.read_csv(args.oasis_csv_dir)
 75 |     train_data_list = pd.read_csv(args.train_csv_dir)
 76 | 
 77 |     train_dataset = Dataset_train(oasis_data_list,train_data_list,args.transformation,args.crop_size)
 78 |     train_loader =  torch.utils.data.DataLoader(
 79 |             train_dataset, batch_size=args.batch_size, shuffle=True,
 80 |             num_workers=args.num_workers, pin_memory=True)
 81 |     
 82 |     device=torch.device('cuda' if torch.cuda.is_available() else 'cpu')
 83 |     print("Total CUDA devices: ", torch.cuda.device_count())
 84 | 
 85 |     model = unet3d.UNet3D(n_class=3, act='relu')
 86 |     model = nn.DataParallel(model, device_ids = [i for i in range(torch.cuda.device_count())])
 87 |     model.to(device)
 88 | 
 89 |     optimizer = torch.optim.Adam(model.parameters(), lr=args.lr, betas=(0.9, 0.999), eps=1e-08, weight_decay=3e-5) 
 90 |     scheduler = ReduceLROnPlateau(optimizer, factor = 0.1, patience = args.patience_lr, mode = 'min')
 91 | 
 92 |     criterion = nn.L1Loss()
 93 |     criterion_ce = nn.BCELoss()
 94 |     criterion_dice = DiceLoss()
 95 | 
 96 |     if args.load:
 97 |         snapshot_path = os.path.join(args.model_path, args.snapshot_path)
 98 |         checkpoint = torch.load(os.path.join(snapshot_path,'model_state.pt'))
 99 |         state_dict = checkpoint['state_dict']
100 |         model.load_state_dict(state_dict)   
101 |         optimizer.load_state_dict(checkpoint['optimizer_state_dict'])
102 |         intial_epoch= checkpoint['epoch']
103 |         print('load model from:',os.path.join(snapshot_path,args.model_name))
104 |     else:
105 |         intial_epoch=0
106 |     train_model(intial_epoch,model,optimizer,scheduler,criterion,criterion_ce,criterion_dice,train_loader,device)
107 | 
108 | def train_model(intial_epoch,model,optimizer,scheduler,criterion,criterion_ce,criterion_dice,train_loader,device):
109 |     snapshot_path = os.path.join(args.model_path, args.snapshot_path)
110 |     if not os.path.exists(snapshot_path):
111 |         os.makedirs(snapshot_path)
112 |     logging.basicConfig(filename=snapshot_path + "/log.txt", level=logging.INFO,
113 |                         format='[%(asctime)s.%(msecs)03d] %(message)s', datefmt='%H:%M:%S')
114 |     logging.getLogger().addHandler(logging.StreamHandler(sys.stdout))
115 |     logging.info(str(args))
116 |     writer = SummaryWriter(snapshot_path + '/log')
117 |     logging.info("{} itertations per epoch".format(len(train_loader)))
118 |     logging.info(optimizer.state_dict()['param_groups'][0]['lr'])
119 | 
120 |     scalar_step = intial_epoch*len(train_loader)
121 |     best_loss=10.0
122 |     best_valid_loss = 10.0
123 |     best_test_dice = 0.0
124 | 
125 |     num_epoch_no_improvement=0
126 |     # to track the training loss as the model trains
127 |     train_losses = []
128 |     # to track the average training loss per epoch as the model trains
129 |     avg_train_losses = []
130 |     avg_valid_losses = []
131 | 
132 |     for epoch in range(intial_epoch,args.epochs):
133 |         print('Epoch {}/{}'.format(epoch, args.epochs - 1))
134 |         print('-' * 10)
135 |         model.train()
136 |         for i, (input, target_normal,target_tumor,tumor_mask) in enumerate(train_loader):
137 |             input_var = input.to(device)
138 |             target_normal = target_normal.to(device)
139 |             target_tumor = target_tumor.to(device)
140 |             tumor_mask = tumor_mask.to(device)
141 | 
142 |             optimizer.zero_grad()
143 |             output = model(input_var)
144 |             #output three channel-normal/tumor/segmentation mask
145 |             output_normal = output[:,0:1,:,:,:]
146 |             output_tumor = output[:,1:2,:,:,:]
147 |             output_mask = F.sigmoid(output[:,2:3,:,:,:])
148 | 
149 |             loss_normal_all = criterion(output_normal,target_normal)
150 |             loss_tumor_all = criterion(output_tumor,target_tumor)
151 |             loss_mask = criterion_dice(output_mask,tumor_mask)
152 | 
153 |             if len(torch.unique(tumor_mask)) == 2:
154 |                 #Emphasis on normal part & tumor part
155 |                 loss_normal = criterion(output_normal[tumor_mask!=0],target_normal[tumor_mask!=0])
156 |                 loss_tumor = criterion(output_tumor[tumor_mask!=0],target_tumor[tumor_mask!=0])
157 |                 loss = loss_mask + loss_normal_all + loss_tumor_all + loss_normal + loss_tumor 
158 |                 writer.add_scalar('loss/loss_normal', loss_normal, scalar_step)
159 |                 writer.add_scalar('loss/loss_tumor', loss_tumor, scalar_step)
160 |                 print('epoch %d iteration %d :  loss: %f, loss_normal_all: %f,loss_tumor_all: %f, loss_normal: %f,loss_tumor: %f,loss_mask: %f' % (epoch,scalar_step, loss.item(),loss_normal_all.item(),loss_tumor_all.item(), loss_normal.item(),loss_tumor.item(),loss_mask.item()))
161 |             else:
162 |                 loss =  loss_mask + loss_normal_all + loss_tumor_all
163 |                 print('epoch %d iteration %d :  loss: %f, loss_normal_all: %f,loss_tumor_all: %f,loss_mask: %f' % (epoch,scalar_step, loss.item(),loss_normal_all.item(),loss_tumor_all.item(),loss_mask.item()))
164 |             
165 |             loss.backward()
166 |             optimizer.step()
167 |             loss = loss.float()
168 |             train_losses.append(round(loss.item(), 2))
169 |             writer.add_scalar('loss/train_loss', loss, scalar_step)
170 |             writer.add_scalar('loss/loss_normal_all', loss_normal_all, scalar_step)
171 |             writer.add_scalar('loss/loss_tumor_all', loss_tumor_all, scalar_step)
172 |             writer.add_scalar('loss/loss_mask', loss_mask, scalar_step)
173 |             scalar_step += 1
174 |             train_loss=np.average(train_losses)
175 |         writer.add_scalar('loss/train_loss_epoch', train_loss, epoch)
176 |         avg_train_losses.append(train_loss)
177 |         scheduler.step(train_loss)
178 |         test_dice, test_dice_mask, valid_loss = test(epoch,model,criterion,writer,device,gate=0.5)
179 |         writer.add_scalar('loss/test_dice', test_dice, epoch)
180 |         writer.add_scalar('loss/test_dice_mask', test_dice_mask, epoch)
181 |         writer.add_scalar('loss/valid_loss', valid_loss, epoch)
182 |         logging.info("Epoch {}, test_dice is {:.4f},test_dice_mask is {:.4f}, valid loss is {:.4f}".format(epoch+1,test_dice,test_dice_mask,valid_loss))
183 |         train_losses=[]
184 | 
185 |         if epoch < 30 and valid_loss < best_valid_loss:
186 |             logging.info("Epoch {}, saving best valid model, test_dice is {:.4f},test_dice_mask is {:.4f}, valid loss is {:.4f}".format(epoch+1,test_dice,test_dice_mask, valid_loss))
187 |             best_valid_loss = valid_loss
188 |             torch.save({
189 |                 'epoch': epoch+1,
190 |                 'state_dict' : model.state_dict(),
191 |                 'optimizer_state_dict': optimizer.state_dict()
192 |             },os.path.join(snapshot_path,args.best_valid_model_name))
193 |             print("Saving model ",os.path.join(snapshot_path,args.best_valid_model_name))
194 |         elif valid_loss < best_valid_loss:
195 |             logging.info("Epoch {}, saving best valid model, test_dice is {:.4f},test_dice_mask is {:.4f}, valid loss is {:.4f}".format(epoch+1,test_dice,test_dice_mask, valid_loss))
196 |             best_valid_loss = valid_loss
197 |             torch.save({
198 |                 'epoch': epoch+1,
199 |                 'state_dict' : model.state_dict(),
200 |                 'optimizer_state_dict': optimizer.state_dict()
201 |             },os.path.join(snapshot_path,'model_valid.pt'))
202 |             print("Saving model ",os.path.join(snapshot_path,'model_valid.pt'))
203 | 
204 |         
205 |         if test_dice_mask > best_test_dice:
206 |             logging.info("Epoch {}, saving best test model, test_dice is {:.4f},test_dice_mask is {:.4f}, valid loss is {:.4f}".format(epoch+1,test_dice,test_dice_mask, valid_loss))
207 |             best_test_dice = test_dice_mask
208 |             torch.save({
209 |                 'epoch': epoch+1,
210 |                 'state_dict' : model.state_dict(),
211 |                 'optimizer_state_dict': optimizer.state_dict()
212 |             },os.path.join(snapshot_path,args.best_test_model_name))
213 |             print("Saving model ",os.path.join(snapshot_path,args.best_test_model_name))
214 |         
215 |         if train_loss < best_loss:
216 |             logging.info("Epoch {}, Training loss decreases from {:.4f} to {:.4f}".format(epoch+1,best_loss, train_loss))
217 |             best_loss = train_loss
218 |             num_epoch_no_improvement = 0
219 |             #save model
220 |             torch.save({
221 |                 'epoch': epoch+1,
222 |                 'state_dict' : model.state_dict(),
223 |                 'optimizer_state_dict': optimizer.state_dict()
224 |             },os.path.join(snapshot_path,args.model_name))
225 |             print("Saving model ",os.path.join(snapshot_path,args.model_name))
226 |         else:
227 |             logging.info("Epoch {}, Training loss is {:.4f}, does not decrease from {:.4f}, num_epoch_no_improvement {}".format(epoch+1, train_loss,best_loss,num_epoch_no_improvement))
228 |             num_epoch_no_improvement += 1
229 |             logging.info(optimizer.state_dict()['param_groups'][0]['lr'])
230 |             torch.save({
231 |                 'epoch': epoch+1,
232 |                 'state_dict' : model.state_dict(),
233 |                 'optimizer_state_dict': optimizer.state_dict()
234 |             },os.path.join(snapshot_path,'model_state.pt'))
235 |         
236 |         if num_epoch_no_improvement >= args.patience or optimizer.state_dict()['param_groups'][0]['lr'] < 1e-6:
237 |             logging.info("Early Stopping")
238 |             torch.save({
239 |                 'epoch': epoch+1,
240 |                 'state_dict' : model.state_dict(),
241 |                 'optimizer_state_dict': optimizer.state_dict()
242 |             },os.path.join(snapshot_path,'model_final.pt'))
243 |             print("Saving model ",os.path.join(snapshot_path,'model_final.pt'))
244 |             break
245 |     torch.save({
246 |                 'epoch': epoch+1,
247 |                 'state_dict' : model.state_dict(),
248 |                 'optimizer_state_dict': optimizer.state_dict()
249 |             },os.path.join(snapshot_path,'model_final.pt'))
250 |     print("Saving model ",os.path.join(snapshot_path,'model_final.pt'))
251 | 
252 | 
253 | def test(epoch,model,criterion,writer,device,gate):
254 |     #three fold cross-validation
255 |     test_csv_dir = os.path.join(args.test_csv_dir,'test_'+str(args.fold)+'.csv')
256 |     test_data_list = pd.read_csv(test_csv_dir)
257 |     flair_file_path = np.asarray(test_data_list.iloc[:,1])
258 |     seg_file_path = np.asarray(test_data_list.iloc[:,5])
259 |     scalar_step_test = epoch*len(test_data_list)
260 | 
261 |     model.eval()
262 |     dice_sum = []
263 |     mask_dice_sum = []
264 |     valid_losses = []
265 | 
266 |     for index in range(len(flair_file_path)):
267 |         img_path = flair_file_path[index]
268 |         itk_img = sitk.ReadImage(img_path)
269 |         img_array = sitk.GetArrayFromImage(itk_img)
270 |         seg_path = seg_file_path[index]
271 |         itk_img_seg = sitk.ReadImage(seg_path)
272 |         seg_array = sitk.GetArrayFromImage(itk_img_seg)
273 |         seg_array[seg_array>0]=1
274 |         img_shape = seg_array.shape
275 |         img_brain,crop_axis,resize_shape = crop_brain_dice(img_array)
276 |         img_brain_norm = normalize_mri(img_brain)
277 |         img_array = img_brain_norm[np.newaxis,np.newaxis,:,:,:]
278 |         input_patch = torch.tensor(img_array.copy(), dtype=torch.float).to(device)
279 |         with torch.no_grad():
280 |             model.eval()
281 |             out_patch = model(input_patch)
282 |             output_normal_valid = out_patch[:,0:1,:,:,:]
283 |             output_tumor_valid  = out_patch[:,1:2,:,:,:]
284 |             output_mask = F.sigmoid(out_patch[:,2:3,:,:,:])
285 |             output_input_valid = torch.mul(output_tumor_valid,output_mask) + torch.mul(output_normal_valid,(1-output_mask))
286 |             valid_loss = criterion(output_input_valid,input_patch)
287 |             writer.add_scalar('valid_loss/valid_loss', valid_loss, scalar_step_test)
288 |             valid_loss = valid_loss.float()
289 |             valid_losses.append(round(valid_loss.item(), 2))
290 |             output_normal = out_patch[0,0,:,:,:].cpu().detach().numpy()
291 |             output_tumor = F.sigmoid(out_patch[0,1,:,:,:]).cpu().detach().numpy()
292 |             output_mask = output_mask[0][0].cpu().detach().numpy()
293 |         output_tumor[output_tumor>gate]=1
294 |         output_tumor[output_tumor<gate]=0
295 |         output_mask[output_mask>gate]=1
296 |         output_mask[output_mask<gate]=0
297 |         tumor_seg = reshape_tumor(output_tumor,crop_axis,resize_shape,img_shape)
298 |         mask_seg = reshape_tumor(output_mask,crop_axis,resize_shape,img_shape)
299 |         dice_index = dice_coef(tumor_seg,seg_array)
300 |         mask_dice_index = dice_coef(mask_seg,seg_array)
301 |         dice_sum.append(dice_index)
302 |         mask_dice_sum.append(mask_dice_index)
303 |     mean_dice = np.mean(np.array(dice_sum))
304 |     mean_dice_mask = np.mean(np.array(mask_dice_sum))
305 |     mean_valid_loss = np.average(valid_losses)
306 |     return mean_dice, mean_dice_mask,mean_valid_loss
307 |   
308 | def normalize_mri(npimg):
309 |     image_nonzero = npimg[np.nonzero(npimg)]
310 |     tmp=(npimg - np.mean(image_nonzero))/ np.std(image_nonzero)
311 |     return tmp
312 | 
313 | def dice_coef(input,target):
314 |     """soft dice loss"""
315 |     eps = 1e-7
316 |     intersection = np.multiply(input,target).sum()
317 |     return  (2. * intersection+eps) / (input.sum() +target.sum() + eps)
318 | 
319 | def reshape_tumor(img,crop_axis,crop_shape,img_shape):
320 |     normal_img = resize(img,crop_shape,order=0,preserve_range=True)
321 |     start_slicex,end_slicex,start_slicey,end_slicey,start_slicez,end_slicez = crop_axis
322 |     output_img = np.zeros(img_shape)
323 |     output_img[start_slicex:end_slicex, start_slicey:end_slicey,start_slicez:end_slicez] = normal_img
324 |     return output_img
325 |         
326 | def crop_brain_dice(img):
327 |     zx = np.any(img, axis=(1,2))
328 |     start_slicex, end_slicex = np.where(zx)[0][[0, -1]]
329 |     zy = np.any(img, axis=(0,2))
330 |     start_slicey, end_slicey = np.where(zy)[0][[0, -1]]
331 |     zz = np.any(img, axis=(0,1))
332 |     start_slicez, end_slicez = np.where(zz)[0][[0, -1]]
333 |     normal_img = img[start_slicex:end_slicex, start_slicey:end_slicey,start_slicez:end_slicez]
334 |     crop_axis = [start_slicex,end_slicex,start_slicey,end_slicey,start_slicez,end_slicez]
335 |     crop_shape = normal_img.shape
336 |     normal_img = resize(normal_img,args.crop_size,order=1,preserve_range=True)
337 |     return normal_img,crop_axis,crop_shape
338 | 
339 | if __name__ == '__main__':
340 |     train()


--------------------------------------------------------------------------------
/BT_CODE/utils/__pycache__/loss.cpython-36.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/xiaoman-zhang/Layer-Decomposition/d9ddef579b192cce5e2c8cb28d76d4b2cd367a95/BT_CODE/utils/__pycache__/loss.cpython-36.pyc


--------------------------------------------------------------------------------
/BT_CODE/utils/__pycache__/prelin_sphere.cpython-36.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/xiaoman-zhang/Layer-Decomposition/d9ddef579b192cce5e2c8cb28d76d4b2cd367a95/BT_CODE/utils/__pycache__/prelin_sphere.cpython-36.pyc


--------------------------------------------------------------------------------
/BT_CODE/utils/__pycache__/vtk_itk.cpython-36.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/xiaoman-zhang/Layer-Decomposition/d9ddef579b192cce5e2c8cb28d76d4b2cd367a95/BT_CODE/utils/__pycache__/vtk_itk.cpython-36.pyc


--------------------------------------------------------------------------------
/BT_CODE/utils/loss.py:
--------------------------------------------------------------------------------
 1 | #!/usr/bin/env python
 2 | # -*- coding: utf-8 -*-
 3 | 
 4 | import torch
 5 | import torch.nn as nn
 6 | import torch.nn.functional as F
 7 | import numpy as np
 8 | 
 9 | 
10 | class Dice_Coef(nn.Module):
11 |     def __init__(self):
12 |         super(Dice_Coef, self).__init__()
13 |      
14 |     def forward(self, predict, target,weights=False):
15 |         smooth = 1e-5 
16 |         total_dice=0
17 |         weight_sum = 0
18 |         target = make_one_hot_3d(target,2)
19 |         predict =  make_one_hot_3d(predict,2)
20 |         for i in range(target.shape[0]):
21 |             dice = dice_coef_binary_np(predict[i], target[i])
22 |             total_dice += dice
23 |         return total_dice/target.shape[0]
24 | 
25 | def dice_coef(input, target):
26 |     """soft dice loss"""
27 |     eps = 1e-7
28 |     intersection = np.multiply(input,target).sum()
29 |     return  (2. * intersection+eps) / (input.sum() +target.sum() + eps)
30 | 
31 | def make_one_hot_3d(npmask, num_classes):
32 |     w,h,d = npmask.shape
33 |     result = np.zeros((num_classes,w,h,d))
34 |     result[0]=1-npmask
35 |     result[1]=npmask
36 |     # result = result.scatter_(1, npmask, 1)
37 |     return result
38 | 
39 | def make_one_hot(npmask, num_classes):
40 |     """
41 |     Convert class index tensor to one hot encoding tensor.
42 |     Args:
43 |          input: A tensor of shape [N, 1, *]
44 |          num_classes: An int of number of class
45 |     Returns:
46 |         A tensor of shape [N, num_classes, *]
47 |     """
48 |     shape = np.array(npmask.shape)
49 |     shape[1] = num_classes
50 |     shape = tuple(shape)
51 |     result = torch.zeros(shape)
52 |     result = result.scatter_(1, npmask.cpu(), 1).cuda()
53 |     return result
54 | 
55 | def dice_coef_binary(predict, target):
56 |     """soft dice loss"""
57 |     eps = 1e-7
58 |     intersection = torch.sum(torch.mul(predict, target))
59 |     return  (2. * intersection+eps) / (torch.sum(predict) +torch.sum(target) + eps)
60 | 
61 | def dice_coef_binary_np(input, target):
62 |     """soft dice loss"""
63 |     eps = 1e-7
64 |     intersection = np.multiply(input,target).sum()
65 |     return  (2. * intersection+eps) / (input.sum() +target.sum() + eps)
66 | 
67 | class DiceLoss(nn.Module):
68 |     def __init__(self):
69 |         super(DiceLoss, self).__init__()
70 |      
71 |     def forward(self, predict, target,weights=False):
72 |         smooth = 1e-5
73 |         loss_record=[]
74 |         total_loss=0
75 |         weight_sum = 0
76 |         dice_loss = 1-dice_coef_binary(predict[:,0],target[:, 0])
77 |         return dice_loss
78 | 


--------------------------------------------------------------------------------
/BT_CODE/utils/metrics.py:
--------------------------------------------------------------------------------
  1 | #!/usr/bin/env python3
  2 | # -*- coding: utf-8 -*-
  3 | 
  4 | import numpy as np
  5 | from scipy import ndimage as ndi
  6 | 
  7 | from medpy import metric
  8 | 
  9 | def calculate_metric_percase(pred, gt):
 10 |     dice = metric.binary.dc(pred, gt)
 11 |     jc = metric.binary.jc(pred, gt)
 12 |     hd = metric.binary.hd95(pred, gt)
 13 |     asd = metric.binary.asd(pred, gt)
 14 |     sen = metric.sensitivity(pred, gt)
 15 |     spec = metric.specificity(pred, gt)
 16 |     return dice, jc, hd, asd, sen, spec
 17 | 
 18 | def mc_calc_localized_metrics(y_true, y_pred):
 19 |   """
 20 |   Calculate localized metrics. We care about accuracy near the knee joint
 21 |   more than along the femur/tibia bone shaft. Splits the true/predicted mask
 22 |   below the centroid and calculate metrics only for that region.
 23 |   
 24 |   Assumption: Label 1 = femur and 2 = tibia
 25 |   """
 26 |   dice_scores = [-1] # Keep placeholder for background score for compatibility
 27 |   localized_pred_masks = np.zeros_like(y_pred)
 28 |   for curr_class in range(1, 3):
 29 |     true_mask = np.zeros_like(y_true)
 30 |     pred_mask = np.zeros_like(y_pred)
 31 |     true_mask[np.where(y_true == curr_class)] = 1
 32 |     pred_mask[np.where(y_pred == curr_class)] = 1
 33 |     
 34 |     # Calculate centroid of ground truth mask
 35 |     centroid = ndi.center_of_mass(true_mask)
 36 |     centroid_y = centroid[1]
 37 |     
 38 |     # Keep desired volume (below center of mass)
 39 |     if curr_class == 1: # Femur keep below centroid
 40 |       true_mask[:,:int(centroid_y),:] = 0
 41 |       pred_mask[:,:int(centroid_y),:] = 0
 42 |     else: # Tibia keep above centroid
 43 |       true_mask[:,int(centroid_y):,:] = 0
 44 |       pred_mask[:,int(centroid_y):,:] = 0
 45 |     
 46 |     dice_score = dice_coef(true_mask, pred_mask)
 47 |     dice_scores.append(dice_score)
 48 |     # Return localized predicted masks as well
 49 |     localized_pred_masks[np.where(pred_mask == 1)] = curr_class
 50 |     
 51 |   return (dice_scores, localized_pred_masks)
 52 |     
 53 | def calc_volumetric_metrics(y_true, y_pred):
 54 |   """
 55 |   Convenience function to calculate only volumetric metrics which are faster
 56 |   to calculate and return as a list.
 57 |   Calculate DICE, VOE, and VD as percentages.
 58 |   """
 59 |   
 60 |   dice = dice_coef(y_true, y_pred)*100.
 61 |   VOE  = voe_vol_overlap_err(y_true, y_pred)*100.
 62 |   VD   = vd_rel_vol_diff(y_true, y_pred)*100.
 63 |   
 64 |   metrics = [dice,VOE,VD]
 65 |   return metrics
 66 |   
 67 | def calc_all_metrics(y_true, y_pred, voxel_spacing=None):
 68 |   """
 69 |   Convenience function to calculate all metrics and return as a list.
 70 |   DICE, VOE, and VD are returned as percentage.
 71 |   """
 72 |   
 73 |   dice = dice_coef(y_true, y_pred)*100.
 74 |   VOE  = voe_vol_overlap_err(y_true, y_pred)*100.
 75 |   VD   = vd_rel_vol_diff(y_true, y_pred)*100.
 76 |   
 77 |   sd1 = get_surf_dists(y_pred, y_true, voxel_spacing)
 78 |   sd2 = get_surf_dists(y_true, y_pred, voxel_spacing)
 79 |   
 80 |   n_pred = sd1.shape[0]
 81 |   n_true = sd2.shape[0]
 82 |   
 83 |   ASD = (np.sum(sd1) + np.sum(sd2)) / (n_true + n_pred)
 84 |   RSD = np.sqrt( (np.sum(sd1**2) + np.sum(sd2**2)) / (n_true + n_pred) )
 85 |   MSD = max(sd1.max(), sd2.max())
 86 |   
 87 |   metrics = [dice,VOE,VD,ASD,RSD,MSD]
 88 |   return metrics
 89 |   
 90 | def dice_coef(y_true, y_pred):
 91 |   """
 92 |   Calculate DICE coefficient score for predicted mask. Expects true and
 93 |   predicted masks to be binary arrays [0/1].
 94 |   
 95 |   # Arguments:
 96 |       y_true: numpy array of true targets, y_true.shape = [slices, H, W]
 97 |       y_pred: numpy array of predicted targets, y_pred.shape = [slices, H, W]
 98 |   # Returns:
 99 |       Scalar DICE coefficient for each label in the range [0 1]
100 |   """
101 |   intersection = np.sum(y_true * y_pred)
102 |   union = np.sum(y_true) + np.sum(y_pred)
103 |     
104 |   if union == 0:
105 |     union = 1
106 |   dice_score = (2*intersection)/union
107 |   return dice_score
108 | 
109 | def dice_coef_mc(y_true, y_pred, nb_classes):
110 |   """
111 |   Calculate DICE coefficient score for each class separately in case of multi
112 |   class segmentation.
113 |   
114 |   # Arguments:
115 |       y_true: numpy array of true targets, y_true.shape = [slices, H, W]
116 |       y_pred: numpy array of predicted targets, y_pred.shape = [slices, H, W]
117 |       nb_classes: number of different labels in true/predicted targets
118 |   # Returns:
119 |       list of DICE coefficients for each label in the range [0 1]
120 |   """
121 |   dice_scores = []
122 |   for curr_class in range(nb_classes):
123 |     true_mask = np.zeros_like(y_true)
124 |     pred_mask = np.zeros_like(y_pred)
125 |     true_mask[np.where(y_true == curr_class)] = 1
126 |     pred_mask[np.where(y_pred == curr_class)] = 1
127 |     
128 |     intersection = np.sum(true_mask * pred_mask)
129 |     union = np.sum(true_mask) + np.sum(pred_mask)
130 |     
131 |     if union == 0:
132 |       union = 1
133 |     dice_score = (2*intersection)/union
134 |     dice_scores.append(dice_score)
135 |     
136 |   return dice_scores
137 | 
138 | def voe_vol_overlap_err(y_true, y_pred):
139 |   """
140 |   Calculate Volume Overlap Error [VOE] for predicted mask. Expects true and
141 |   predicted masks to be binary arrays [0/1].
142 |                 DICE
143 |   VOE = 1 -  -----------
144 |               2 - DICE
145 |   # Arguments:
146 |       y_true: numpy array of true targets, y_true.shape = [slices, H, W]
147 |       y_pred: numpy array of predicted targets, y_pred.shape = [slices, H, W]
148 |   # Returns:
149 |       Scalar VOE in the range [0 1]
150 |   """
151 |   dsc = dice_coef(y_true, y_pred)
152 |   
153 |   return 1. - dsc/(2. - dsc)
154 | 
155 | def vd_rel_vol_diff(y_true, y_pred):
156 |   """
157 |   Calculate Relative Volume Difference [RVD] for predicted mask. Expects true and
158 |   predicted masks to be binary arrays [0/1].
159 |          |y_pred| - |y_true|
160 |   RVD = ----------------------
161 |               |y_true|
162 |   # Arguments:
163 |       y_true: numpy array of true targets, y_true.shape = [slices, H, W]
164 |       y_pred: numpy array of predicted targets, y_pred.shape = [slices, H, W]
165 |   # Returns:
166 |       Scalar RVD in the range [0 1]
167 |   """
168 |   
169 |   nb_true = np.float64(np.sum(y_true))
170 |   nb_pred = np.float64(np.sum(y_pred))
171 |   
172 |   if nb_true < 1:
173 |     return 0
174 |   
175 |   return (nb_pred - nb_true) / nb_true
176 | 
177 | def asd_avg_surf_dist(y_true, y_pred, voxel_spacing=None):
178 |   """
179 |   Calculate Average Surface Distance [ASD] for predicted mask. Expects true and
180 |   predicted masks to be binary arrays [0/1].
181 |   
182 |   # Arguments:
183 |       y_true: numpy array of true targets, y_true.shape = [slices, H, W]
184 |       y_pred: numpy array of predicted targets, y_pred.shape = [slices, H, W]
185 |       voxel_spacing: tuple with voxel spacing in 3 dimensions
186 |   # Returns:
187 |       Scalar ASD in mm
188 |   """
189 | 
190 |   sd1 = get_surf_dists(y_pred, y_true, voxel_spacing)
191 |   sd2 = get_surf_dists(y_true, y_pred, voxel_spacing)
192 |   
193 |   n_pred = sd1.shape[0]
194 |   n_true = sd2.shape[0]
195 |   
196 |   asd = (np.sum(sd1) + np.sum(sd2)) / (n_true + n_pred)
197 |   
198 |   return asd
199 | 
200 | def rsd_rms_surf_dist(y_true, y_pred, voxel_spacing=None):
201 |   """
202 |   Calculate Root Mean Square Surface Distance [RSD] for predicted mask.
203 |   Expects true and predicted masks to be binary arrays [0/1].
204 |   
205 |   # Arguments:
206 |       y_true: numpy array of true targets, y_true.shape = [slices, H, W]
207 |       y_pred: numpy array of predicted targets, y_pred.shape = [slices, H, W]
208 |       voxel_spacing: tuple with voxel spacing in 3 dimensions
209 |   # Returns:
210 |       Scalar RSD in mm
211 |   """
212 |   
213 |   sd1 = get_surf_dists(y_pred, y_true, voxel_spacing)
214 |   sd2 = get_surf_dists(y_true, y_pred, voxel_spacing)
215 |   
216 |   n_pred = sd1.shape[0]
217 |   n_true = sd2.shape[0]
218 |   
219 |   rsd = np.sqrt( (np.sum(sd1**2) + np.sum(sd2**2)) / (n_true + n_pred) )
220 |   
221 |   return rsd
222 | 
223 | def msd_max_surf_dist(y_true, y_pred, voxel_spacing=None):
224 |   """
225 |   Calculate Maximum Surface Distance [MSD] for predicted mask. Also known as
226 |   Symmetric Hausdorff Distance
227 |   Expects true and predicted masks to be binary arrays [0/1].
228 |   
229 |   # Arguments:
230 |       y_true: numpy array of true targets, y_true.shape = [slices, H, W]
231 |       y_pred: numpy array of predicted targets, y_pred.shape = [slices, H, W]
232 |       voxel_spacing: tuple with voxel spacing in 3 dimensions
233 |   # Returns:
234 |       Scalar MSD in mm
235 |   """
236 |   
237 |   sd1 = get_surf_dists(y_pred, y_true, voxel_spacing)
238 |   sd2 = get_surf_dists(y_true, y_pred, voxel_spacing)
239 |    
240 |   msd = max(sd1.max(), sd2.max())
241 |   
242 |   return msd
243 | 
244 | def get_surf_dists(target, reference, voxel_spacing=None):
245 |   """
246 |   Calculate surface distances of target with respect to reference
247 |   
248 |   # Arguments:
249 |       target: numpy array of target mask, target.shape = [slices, H, W]
250 |       reference: numpy array of reference mask, reference.shape = [slices, H, W]
251 |       voxel_spacing: tuple with voxel spacing in 3 dimensions
252 |   # Returns:
253 |       Array with target surface differences in mm
254 |   """
255 |   y_true = reference.astype(np.bool)
256 |   y_pred = target.astype(np.bool)
257 |   
258 |   # 8 connectivity in 3D, any voxel with at least one background voxel as
259 |   # neighbor is part of object boundary
260 |   connectivity_struc_elem = ndi.generate_binary_structure(3, 3)
261 |   
262 |   # Extract boundaries [binary XOR]
263 |   true_border = y_true ^ ndi.binary_erosion(y_true,
264 |                                             structure=connectivity_struc_elem,
265 |                                             iterations=1)
266 |   pred_border = y_pred ^ ndi.binary_erosion(y_pred,
267 |                                             structure=connectivity_struc_elem,
268 |                                             iterations=1)
269 |   # Distance transform, scipy dist. transform returns distance of every
270 |   # non-zero voxel to nearest background, 0, voxel. Hence need to invert
271 |   # border image where the boundary is now 0 and everywhere else is 1 leading
272 |   # to distances over entire image [not signed distance]
273 |   dt = ndi.distance_transform_edt(~true_border, sampling=voxel_spacing)
274 |   surf_dists = dt[pred_border]
275 | 
276 |   return surf_dists


--------------------------------------------------------------------------------
/BT_CODE/utils/prelin_sphere.py:
--------------------------------------------------------------------------------
  1 | """
  2 | Reference:
  3 | A self-supervised learning strategy for postoperative brain cavity segmentation simulating resections
  4 | """ 
  5 | import warnings
  6 | import numpy as np
  7 | from tqdm import tqdm
  8 | from noise import pnoise3, snoise3
  9 | import vtk
 10 | from vtk.util.numpy_support import numpy_to_vtk
 11 | from vtk.numpy_interface import dataset_adapter as dsa
 12 | 
 13 | 
 14 | 
 15 | def add_noise_to_sphere(poly_data, octaves, offset=0, scale=0.5):
 16 |     """
 17 |     Expects sphere with radius 1 centered at the origin
 18 |     """
 19 |     wrap_data_object = dsa.WrapDataObject(poly_data)
 20 |     points = wrap_data_object.Points
 21 |     normals = wrap_data_object.PointData['Normals']
 22 | 
 23 |     points_with_noise = []
 24 |     zipped = list(zip(points, normals))
 25 |     for point, normal in zipped:#tqdm(zipped):
 26 |         offset_point = point + offset
 27 |         noise = scale * snoise3(*offset_point, octaves=octaves)
 28 |         point_with_noise = point + noise * normal
 29 |         points_with_noise.append(point_with_noise)
 30 |     points_with_noise = np.array(points_with_noise)
 31 | 
 32 |     vertices = vtk.vtkPoints()
 33 |     with warnings.catch_warnings():
 34 |         warnings.simplefilter('ignore')
 35 |         points_with_noise_vtk = numpy_to_vtk(points_with_noise)
 36 |     vertices.SetData(points_with_noise_vtk)
 37 |     poly_data.SetPoints(vertices)
 38 | 
 39 |     return poly_data
 40 | 
 41 | 
 42 | def center_poly_data(poly_data):
 43 |     centerOfMassFilter = vtk.vtkCenterOfMass()
 44 |     centerOfMassFilter.SetInputData(poly_data)
 45 |     centerOfMassFilter.SetUseScalarsAsWeights(False)
 46 |     centerOfMassFilter.Update()
 47 |     center = np.array(centerOfMassFilter.GetCenter())
 48 | 
 49 |     transform = vtk.vtkTransform()
 50 |     transform.Translate(-center)
 51 | 
 52 |     transform_filter = vtk.vtkTransformPolyDataFilter()
 53 |     transform_filter.SetTransform(transform)
 54 |     transform_filter.SetInputData(poly_data)
 55 |     transform_filter.Update()
 56 | 
 57 |     poly_data = transform_filter.GetOutput()
 58 |     return poly_data
 59 | 
 60 | 
 61 | def transform_poly_data(poly_data, center, radii, angles):
 62 |     transform = vtk.vtkTransform()
 63 |     transform.Translate(center)
 64 |     x_angle, y_angle, z_angle = angles  # there must be a better way
 65 |     transform.RotateX(x_angle)
 66 |     transform.RotateY(y_angle)
 67 |     transform.RotateZ(z_angle)
 68 |     transform.Scale(*radii)
 69 | 
 70 |     transform_filter = vtk.vtkTransformPolyDataFilter()
 71 |     transform_filter.SetTransform(transform)
 72 |     transform_filter.SetInputData(poly_data)
 73 |     transform_filter.Update()
 74 | 
 75 |     poly_data = transform_filter.GetOutput()
 76 |     return poly_data
 77 | 
 78 | 
 79 | def compute_normals(poly_data):
 80 |     normal_filter = vtk.vtkPolyDataNormals()
 81 |     normal_filter.AutoOrientNormalsOn()
 82 |     normal_filter.SetComputePointNormals(True)
 83 |     normal_filter.SetComputeCellNormals(True)
 84 |     normal_filter.SplittingOff()
 85 |     normal_filter.SetInputData(poly_data)
 86 |     normal_filter.ConsistencyOn()
 87 |     normal_filter.Update()
 88 |     poly_data = normal_filter.GetOutput()
 89 |     return poly_data
 90 | 
 91 | 
 92 | def get_resection_poly_data(
 93 |         poly_data,
 94 |         offset,
 95 |         center,
 96 |         radii,
 97 |         angles,
 98 |         octaves=4,
 99 |         scale=0.5,
100 |         deepcopy=True,
101 |         ):
102 |     if deepcopy:
103 |         new_poly_data = vtk.vtkPolyData()
104 |         new_poly_data.DeepCopy(poly_data)
105 |         poly_data = new_poly_data
106 | 
107 |     poly_data = add_noise_to_sphere(poly_data, octaves=octaves, offset=offset)
108 |     poly_data = center_poly_data(poly_data)
109 | 
110 |     poly_data = transform_poly_data(poly_data, center, radii, angles)
111 |     poly_data = compute_normals(poly_data)
112 |     return poly_data
113 | 


--------------------------------------------------------------------------------
/BT_CODE/utils/vtk_itk.py:
--------------------------------------------------------------------------------
 1 | import SimpleITK as sitk
 2 | import vtk
 3 | from vtk.util import numpy_support
 4 | import numpy as np
 5 | 
 6 | def pd_to_numpy_vol(pd, spacing=[1.,1.,1.], shape=None, origin=None, foreground_value=255, backgroud_value = 0):
 7 |     if shape is None:
 8 |         bnds = np.array(pd.GetBounds())
 9 |         shape = np.ceil((bnds[1::2]-bnds[::2])/spacing).astype(int)+15
10 |     if origin is None:
11 |         origin = bnds[::2]+(bnds[1::2]-bnds[::2])/2
12 | 
13 |     #make image
14 |     extent = np.zeros(6).astype(int)
15 |     extent[1::2] = np.array(shape)-1
16 |     
17 |     imgvtk = vtk.vtkImageData()
18 |     imgvtk.SetSpacing(spacing)
19 |     imgvtk.SetOrigin(origin)
20 |     imgvtk.SetExtent(extent)
21 |     imgvtk.AllocateScalars(vtk.VTK_UNSIGNED_CHAR, 1)
22 |     
23 |     vtk_data_array = numpy_support.numpy_to_vtk(num_array=np.ones(shape[::-1]).ravel()*backgroud_value,  # ndarray contains the fitting result from the points. It is a 3D array
24 |                                                 deep=True, array_type=vtk.VTK_FLOAT)
25 |     
26 |     imgvtk.GetPointData().SetScalars(vtk_data_array)
27 |     
28 |     #poly2 stencil
29 |     pol2stenc = vtk.vtkPolyDataToImageStencil()
30 |     pol2stenc.SetInputData(pd)
31 |     pol2stenc.SetOutputSpacing(spacing)
32 |     pol2stenc.SetOutputOrigin(origin)
33 |     pol2stenc.SetOutputWholeExtent(extent)
34 |     pol2stenc.Update()
35 |     
36 |     #stencil to image
37 |     imgstenc = vtk.vtkImageStencil()
38 |     imgstenc.SetInputData(imgvtk)
39 |     imgstenc.SetStencilConnection(pol2stenc.GetOutputPort())
40 |     imgstenc.ReverseStencilOn()
41 |     imgstenc.SetBackgroundValue(foreground_value)
42 |     imgstenc.Update()
43 |     
44 |     ndseg = numpy_support.vtk_to_numpy(imgstenc.GetOutputDataObject(0).GetPointData().GetArray(0))
45 |     return ndseg.reshape(shape[::-1])
46 | 
47 | 
48 | def pd_to_itk_image(pd, ref_img, foreground_value=255, backgroud_value = 0):
49 |     ndseg = pd_to_numpy_vol(pd, spacing=ref_img.GetSpacing(), shape=ref_img.GetSize(), origin=ref_img.GetOrigin() )
50 |     segitk = sitk.GetImageFromArray(ndseg.astype(np.int16))
51 |     segitk.CopyInformation(ref_img)
52 |     return segitk
53 | 
54 | 
55 | 
56 | def rotate_img(img, rotation_center=None, theta_x=0,theta_y=0, theta_z=0, translation=(0,0,0), interp=sitk.sitkLinear, pixel_type=None, default_value=None):
57 |     if not rotation_center:
58 |         rotation_center = np.array(img.GetOrigin())+np.array(img.GetSpacing())*np.array(img.GetSize())/2
59 |     if default_value is None:
60 |         default_value = img.GetPixel(0,0,0)
61 |     pixel_type = img.GetPixelIDValue()
62 | 
63 |     rigid_euler = sitk.Euler3DTransform(rotation_center, theta_x, theta_y, theta_z, translation)
64 |     return sitk.Resample(img, img, rigid_euler, interp, default_value, pixel_type)
65 | 
66 | def rotate_polydata(pd, rotation_center, theta_x=0,theta_y=0, theta_z=0, translation=(0,0,0)):
67 |     rigid_euler = sitk.Euler3DTransform(rotation_center, -theta_x, -theta_y, -theta_z, translation)
68 |     matrix = np.zeros([4,4])
69 |     old_matrix=np.array(rigid_euler.GetMatrix()).reshape(3,3)
70 |     matrix[:3,:3] = old_matrix
71 |     matrix[-1,-1] = 1
72 | 
73 |     # to rotate about a center we first need to translate
74 |     transform_t = vtk.vtkTransform()
75 |     transform_t.Translate(-rotation_center)
76 |     transformer_t = vtk.vtkTransformPolyDataFilter()
77 |     transformer_t.SetTransform(transform_t)
78 |     transformer_t.SetInputData(pd)
79 |     transformer_t.Update()
80 | 
81 |     transform = vtk.vtkTransform()
82 |     transform.SetMatrix(matrix.ravel())
83 | 
84 |     transformer = vtk.vtkTransformPolyDataFilter()
85 |     transformer.SetTransform(transform)
86 |     transformer.SetInputConnection(transformer_t.GetOutputPort())
87 |     transformer.Update()
88 | 
89 |     # translate back
90 |     transform_t2 = vtk.vtkTransform()
91 |     transform_t2.Translate(rotation_center)
92 |     transformer_t2 = vtk.vtkTransformPolyDataFilter()
93 |     transformer_t2.SetTransform(transform_t2)
94 |     transformer_t2.SetInputConnection(transformer.GetOutputPort())
95 |     transformer_t2.Update()
96 | 
97 |     return transformer_t2.GetOutputDataObject(0)
98 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
 1 | ## Self-supervised Tumor Segmentation with Sim2Real Adaptation 
 2 | 
 3 | [Project page](https://xiaoman-zhang.github.io/Layer-Decomposition/)
 4 | 
 5 | This repo provides a demo for the IEEE JBHI 2023 "Self-supervised Tumor Segmentation with Sim2Real Adaptation" on the  brain tumor segmentation task.
 6 | 
 7 | To cite, please use:
 8 | 
 9 | ```
10 | @ARTICLE{10032792,
11 |   author={Zhang, Xiaoman and Xie, Weidi and Huang, Chaoqin and Zhang, Ya and Chen, Xin and Tian, Qi and Wang, Yanfeng},
12 |   journal={IEEE Journal of Biomedical and Health Informatics}, 
13 |   title={Self-supervised Tumor Segmentation with Sim2Real Adaptation}, 
14 |   year={2023},
15 |   volume={},
16 |   number={},
17 |   pages={1-13},
18 |   doi={10.1109/JBHI.2023.3240844}
19 |   }
20 | ```
21 | 
22 | #### Acknowledgement
23 | 
24 | Part of our code is borrowed from the following repositories.
25 | * [Resseg](https://github.com/fepegar/resseg-ijcars): "A self-supervised learning strategy for postoperative brain cavity segmentation simulating resections", International Journal of Computer Assisted Radiology and Surgery 2021
26 | 
27 | We thank to the authors for releasing their codes. Please also consider citing their works.
28 | 


--------------------------------------------------------------------------------