├── .idea
    ├── .gitignore
    ├── inspectionProfiles
    │   └── profiles_settings.xml
    ├── misc.xml
    ├── modules.xml
    ├── preprocess.iml
    └── vcs.xml
├── CBAM.py
├── FreeSurfer+FSL.py
├── FreeSurfer.py
├── README.md
├── ants.py
├── antspy.py
├── crop_resize.py
├── data_augmentation.py
├── dcm2nii.py
├── image_combine_example.py
├── img2nii.py
├── jacobian.py
├── phi
    ├── .idea
    │   ├── NII.iml
    │   ├── deployment.xml
    │   ├── inspectionProfiles
    │   │   └── profiles_settings.xml
    │   ├── misc.xml
    │   ├── modules.xml
    │   └── workspace.xml
    ├── f_img.jpg
    ├── img.jpg
    ├── img1.jpg
    ├── img2.jpg
    ├── m_img.jpg
    ├── my_show_grid.py
    └── show_grid.py
├── resize_img.py
└── torch_datagenerators.py


/.idea/.gitignore:
--------------------------------------------------------------------------------
1 | # Default ignored files
2 | /shelf/
3 | /workspace.xml
4 | # Datasource local storage ignored files
5 | /../../../../../../:\Data\1.Work\02.Code\preprocess\.idea/dataSources/
6 | /dataSources.local.xml
7 | # Editor-based HTTP Client requests
8 | /httpRequests/
9 | 


--------------------------------------------------------------------------------
/.idea/inspectionProfiles/profiles_settings.xml:
--------------------------------------------------------------------------------
1 | <component name="InspectionProjectProfileManager">
2 |   <settings>
3 |     <option name="USE_PROJECT_PROFILE" value="false" />
4 |     <version value="1.0" />
5 |   </settings>
6 | </component>


--------------------------------------------------------------------------------
/.idea/misc.xml:
--------------------------------------------------------------------------------
1 | <?xml version="1.0" encoding="UTF-8"?>
2 | <project version="4">
3 |   <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.7" project-jdk-type="Python SDK" />
4 | </project>


--------------------------------------------------------------------------------
/.idea/modules.xml:
--------------------------------------------------------------------------------
1 | <?xml version="1.0" encoding="UTF-8"?>
2 | <project version="4">
3 |   <component name="ProjectModuleManager">
4 |     <modules>
5 |       <module fileurl="file://$PROJECT_DIR$/.idea/preprocess.iml" filepath="$PROJECT_DIR$/.idea/preprocess.iml" />
6 |     </modules>
7 |   </component>
8 | </project>


--------------------------------------------------------------------------------
/.idea/preprocess.iml:
--------------------------------------------------------------------------------
 1 | <?xml version="1.0" encoding="UTF-8"?>
 2 | <module type="PYTHON_MODULE" version="4">
 3 |   <component name="NewModuleRootManager">
 4 |     <content url="file://$MODULE_DIR$" />
 5 |     <orderEntry type="inheritedJdk" />
 6 |     <orderEntry type="sourceFolder" forTests="false" />
 7 |   </component>
 8 |   <component name="PyDocumentationSettings">
 9 |     <option name="format" value="PLAIN" />
10 |     <option name="myDocStringFormat" value="Plain" />
11 |   </component>
12 |   <component name="TestRunnerService">
13 |     <option name="PROJECT_TEST_RUNNER" value="pytest" />
14 |   </component>
15 | </module>


--------------------------------------------------------------------------------
/.idea/vcs.xml:
--------------------------------------------------------------------------------
1 | <?xml version="1.0" encoding="UTF-8"?>
2 | <project version="4">
3 |   <component name="VcsDirectoryMappings">
4 |     <mapping directory="$PROJECT_DIR$/show_nii" vcs="Git" />
5 |   </component>
6 | </project>


--------------------------------------------------------------------------------
/CBAM.py:
--------------------------------------------------------------------------------
 1 | import torch
 2 | import torch.nn as nn
 3 | 
 4 | 
 5 | class CBAM_Module(nn.Module):
 6 |     def __init__(self, dim, in_channels, ratio, kernel_size):
 7 |         super(CBAM_Module, self).__init__()
 8 |         self.avg_pool = getattr(nn, "AdaptiveAvgPool{0}d".format(dim))(1)
 9 |         self.max_pool = getattr(nn, "AdaptiveMaxPool{0}d".format(dim))(1)
10 |         conv_fn = getattr(nn, "Conv{0}d".format(dim))
11 |         self.fc1 = conv_fn(in_channels, in_channels // ratio, kernel_size=1, padding=0)
12 |         self.relu = nn.ReLU()
13 |         self.fc2 = conv_fn(in_channels // ratio, in_channels, kernel_size=1, padding=0)
14 |         self.sigmoid = nn.Sigmoid()
15 |         assert kernel_size in (3, 7), 'kernel size must be 3 or 7'
16 |         padding = 3 if kernel_size == 7 else 1
17 |         self.conv = conv_fn(2, 1, kernel_size=kernel_size, stride=1, padding=padding)
18 | 
19 |     def forward(self, x):
20 |         print("CBAM")
21 |         # Channel attention module:（Mc(f) = σ(MLP(AvgPool(f)) + MLP(MaxPool(f)))）
22 |         module_input = x
23 |         avg = self.fc2(self.relu(self.fc1(self.avg_pool(x))))
24 |         mx = self.fc2(self.relu(self.fc1(self.max_pool(x))))
25 |         x = self.sigmoid(avg + mx)
26 |         x = module_input * x
27 |         # Spatial attention module:Ms (f) = σ( f7×7( AvgPool(f) ; MaxPool(F)] )))
28 |         module_input = x
29 |         avg = torch.mean(x, dim=1, keepdim=True)
30 |         mx, _ = torch.max(x, dim=1, keepdim=True)
31 |         x = torch.cat((avg, mx), dim=1)
32 |         x = self.sigmoid(self.conv(x))
33 |         x = module_input * x
34 |         return x


--------------------------------------------------------------------------------
/FreeSurfer+FSL.py:
--------------------------------------------------------------------------------
 1 | import os
 2 | import glob
 3 | 
 4 | 
 5 | # 使用FreeSurfer对图像进行颅骨剥离
 6 | print("FreeSurfer start......\n")
 7 | # 图像坐在的目录
 8 | #------------------------图像路径需更改------------------------#
 9 | path ="/home/syzhou/zuzhiang/Dataset/MGH10/Heads"
10 | # 读取目录下的.img文件列表，*.img表示该目录下所有以.img结尾的文件
11 | #-----------------------图像后缀名需更改--- -------------------#
12 | files = glob.glob(os.path.join(path,"*.img"))
13 | # 输出路径
14 | #------------------------输出路径需更改------------------------#
15 | out_path="/home/syzhou/zuzhiang/MIP/FSL_img/MGH10_results"
16 | print("number: ",len(files))
17 | # 下面为freesurfer的环境配置命令
18 | a = "export FREESURFER_HOME=/home/syzhou/zuzhiang/freesurfer;"
19 | b = "source $FREESURFER_HOME/SetUpFreeSurfer.sh;"
20 | # 数据所在的目录
21 | c = "export SUBJECTS_DIR="+path+";"
22 | 
23 | for file in files:
24 |     # 将文件路径和文件名分离
25 |     filename = os.path.split(file)[1] # 将路径名和文件名分开
26 |     filename = filename.split(".")[0] # 去除所有扩展名   
27 |     #recon-all是颅骨去除的命令
28 |     # mri_convert是进行格式转换，从mgz转到nii.gz，只是为了方便查看
29 |     filename=filename[:] #根据扩展名的不同，这里需要做更改，只保留文件名即可
30 |     # 当前输出文件路径，以.nii.gz格式保存
31 |     cur_out_path=os.path.join(out_path,filename+".nii.gz")
32 |     print("file name: ",file)
33 |     cmd = a + b + c + "mri_watershed "+file+" "+ cur_out_path
34 |     #print(cmd,"\n")
35 |     os.system(cmd)
36 | 
37 | 
38 | # 使用FSL对图像和对应的label进行仿射对齐
39 | print("FSL start......\n")
40 | # fixed图像的路径
41 | #---------------去除头骨后的fixed图像名需更改-------------------#
42 | f_path= os.path.join(out_path,"g1.nii.gz")
43 | # moving图像的路径
44 | m_path=out_path
45 | # label的路径
46 | #-----------------------label路径需更改-----------------------#
47 | label_path="/home/syzhou/zuzhiang/Dataset/MGH10/Atlases"
48 | files=glob.glob(os.path.join(m_path,"*.nii.gz"))
49 | print("number: ",len(files))
50 | for file in files:
51 |     print("file: ",file)
52 |     # 根据图像名找到对应的label名
53 |     filename=os.path.split(file)[1]
54 |     filename = filename.split(".")[0] # 去除所有扩展名
55 |     #---------------------label后缀名需更改--------------------#
56 |     label=os.path.join(label_path,filename+".img")
57 |     # 下面分别是输出图像名/转换矩阵名/label名，
58 |     out_img=os.path.join(out_path,filename+"_img.nii.gz")
59 |     out_mat=os.path.join(out_path,filename+".mat")
60 |     out_label=os.path.join(out_path,filename+"_label.nii.gz")
61 |     # 如果当前文件和fixed图像一样则只将对应label格式转换为.nii.gz
62 |     if f_path==file:
63 |         convert="mri_convert " + label +" " + out_label
64 |         os.system(convert)
65 |         print("continue.........")
66 |         continue    
67 |     # 将moving图像对齐到fixed图像
68 |     flirt_img="flirt -in "+file+ " -ref "+f_path+" -out "+out_img+" -omat "+out_mat+ " -dof 12"
69 |     # 将上一步的仿射变换矩阵作用在图像对应的label上
70 |     flirt_label="flirt -in "+label+" -ref "+f_path+" -out "+out_label+" -init "+out_mat+" -applyxfm -interp nearestneighbour"
71 |     #print(flirt_img,"\n")
72 |     #print(flirt_label,"\n")
73 |     os.system(flirt_img)
74 |     os.system(flirt_label)
75 | 
76 | print("\n\nEnd")


--------------------------------------------------------------------------------
/FreeSurfer.py:
--------------------------------------------------------------------------------
 1 | import os
 2 | import glob
 3 | 
 4 | path = r"/home/syzhou/zuzhiang/Dataset/MGH10/Heads"
 5 | # 读取目录下的nii.gz文件
 6 | images = glob.glob(os.path.join(path,"*.img"))
 7 | # 下面为freesurfer的环境配置命令
 8 | a = "export FREESURFER_HOME=/home/syzhou/zuzhiang/freesurfer;"
 9 | b = "source $FREESURFER_HOME/SetUpFreeSurfer.sh;"
10 | # 数据所在的目录
11 | c = "export SUBJECTS_DIR="+path+";"
12 | 
13 | #images=['/home/syzhou/zuzhiang/Dataset/MGH10/Heads/1127.img']
14 | for image in images:
15 |     # 将文件路径和文件名分离
16 |     filename = os.path.split(image)[1] # 将路径名和文件名分开
17 |     filename = os.path.splitext(filename)[0] #将文件名和扩展名分开，如果为.nii.gz，则认为扩展名是.gz
18 |     # freesurfer环境配置、颅骨去除、未仿射对齐mpz转nii、仿射对齐、仿射对齐mpz转nii.gz格式
19 |     #recon-all是颅骨去除的命令
20 |     # mri_convert是进行格式转换，从mgz转到nii.gz，只是为了方便查看
21 |     # --apply_transform：仿射对齐操作
22 |     # 转格式
23 |     filename=filename[:] #根据扩展名的不同，这里需要做更改，只保留文件名即可
24 |     cur_path=os.path.join(path,filename) 
25 |     print("file name: ",cur_path)
26 |     cmd = a + b + c \
27 |           + "recon-all -parallel -i " + image + " -autorecon1 -subjid " + cur_path + "&&" \
28 |           + "mri_convert " +  cur_path + "/mri/brainmask.mgz " +cur_path + "/mri/"+filename+".nii.gz;"\
29 |           + "mri_convert " + cur_path + "/mri/brainmask.mgz --apply_transform " + cur_path + "/mri/transforms/talairach.xfm -o " + cur_path + "/mri/brainmask_affine.mgz&&" \
30 |           + "mri_convert " + cur_path + "/mri/brainmask_affine.mgz " + cur_path + "/mri/"+filename+"_affine.nii.gz;"
31 |     #print("cmd:\n",cmd)
32 |     os.system(cmd)


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
 1 | 数据很杂，就懒得逐一整理了，很多都是实验性的，可能存在问题，不过好久没弄也忘得差不多，有问题的话还请大家自行阅读代码解决吧，下面只把每个文件是做什么用的大体说一下：
 2 | 
 3 | - `phi`：
 4 |   - `my_show_grid.py`：先产生规则的网格图片，然后使用STN根据读入的形变场对其变形
 5 |   - `show_grid.py`：读入形变场的.nii文件，然后生成可视化的形变场图片
 6 | - `show_nii`：
 7 |   - `show_nii.py`：展示三维的.nii图像
 8 |   - `show_nii3.py`：同时展示三维的.nii格式的器官、器官分割mask、病灶分割mask图像
 9 | - `ants.py`：使用ANTs包内的SyN对图像进行配准
10 | - `antspy.py`：使用基于python版的ANTs——antspy对图像和其对应的标签同时进行配准
11 | - `CBAM`：CBAM注意力模块的实现
12 | - `crop_resize.py`：先找到包含脑部区域的最小矩形框，然后手动计算矩形框的大小（每一维应该是16的倍数），然后用已经注释掉的部分进行resize
13 | - `data_augmentation.py`：数据增强的代码，包括B样条采样、反转、平移、缩放、旋转、灰度值均衡化等，但B样条采样貌似有问题
14 | - `dcm2nii.py`：将dcm格式的三维图像转换为.nii格式的
15 | - `FreeSurfer+FSL.py`：先用FreeSurfer对脑部图像进行颅骨剥离，然后用FSL对图像和标签同时进行仿射对齐
16 | - `FreeSurfer.py`：用FreeSurfer对脑部图像进行颅骨剥离
17 | - `image_combine_example.py`：别人给的，没用过
18 | - `img2nii.py`：将.img格式的三维图像转化为.nii格式的
19 | - `jacobian.py`：计算形变场的雅克比行列式
20 | - `resize_img.py`：对图像进行resize
21 | - `tor_datagenerators.py`：基于pytorch的数据生成器
22 | 


--------------------------------------------------------------------------------
/ants.py:
--------------------------------------------------------------------------------
 1 | import os
 2 | import glob
 3 | 
 4 | path="/home/syzhou/zuzhiang/Dataset/LPBA40"
 5 | out_path="/home/syzhou/zuzhiang/MIP/ANTs"
 6 | f_name="/home/syzhou/zuzhiang/Dataset/LPBA40/1.nii.gz"
 7 | for i in range(2,41):
 8 |     m_name=os.path.join(path,str(i)+".nii.gz")
 9 |     out_name=str(i)+"m"
10 |     cmd= "antsRegistrationSyN.sh -d 3 -f " + f_name + " -m " + m_name + " -o " + out_name
11 |     print("cmd: ",cmd)
12 |     os.system(cmd)
13 | print("End")
14 | 


--------------------------------------------------------------------------------
/antspy.py:
--------------------------------------------------------------------------------
 1 | import os
 2 | import glob
 3 | import ants
 4 | import numpy as np
 5 | import SimpleITK as sitk
 6 | 
 7 | # ants图片的读取
 8 | f_img = ants.image_read("./data/f_img.nii.gz")
 9 | m_img = ants.image_read("./data/m_img.nii.gz")
10 | f_label = ants.image_read("./data/f_label.nii.gz")
11 | m_label = ants.image_read("./data/m_label.nii.gz")
12 | 
13 | '''
14 | ants.registration()函数的返回值是一个字典：
15 |     warpedmovout: 配准到fixed图像后的moving图像 
16 |     warpedfixout: 配准到moving图像后的fixed图像 
17 |     fwdtransforms: 从moving到fixed的形变场 
18 |     invtransforms: 从fixed到moving的形变场
19 | 
20 | type_of_transform参数的取值可以为：
21 |     Rigid：刚体
22 |     Affine：仿射配准，即刚体+缩放
23 |     ElasticSyN：仿射配准+可变形配准，以MI为优化准则，以elastic为正则项
24 |     SyN：仿射配准+可变形配准，以MI为优化准则
25 |     SyNCC：仿射配准+可变形配准，以CC为优化准则
26 | '''
27 | # 图像配准
28 | mytx = ants.registration(fixed=f_img, moving=m_img, type_of_transform='SyN')
29 | # 将形变场作用于moving图像，得到配准后的图像，interpolator也可以选择"nearestNeighbor"等
30 | warped_img = ants.apply_transforms(fixed=f_img, moving=m_img, transformlist=mytx['fwdtransforms'],
31 |                                    interpolator="linear")
32 | # 对moving图像对应的label图进行配准
33 | warped_label = ants.apply_transforms(fixed=f_img, moving=m_label, transformlist=mytx['fwdtransforms'],
34 |                                      interpolator="linear")
35 | # 将配准后图像的direction/origin/spacing和原图保持一致
36 | warped_img.set_direction(f_img.direction)
37 | warped_img.set_origin(f_img.origin)
38 | warped_img.set_spacing(f_img.spacing)
39 | warped_label.set_direction(f_img.direction)
40 | warped_label.set_origin(f_img.origin)
41 | warped_label.set_spacing(f_img.spacing)
42 | img_name = "./result/warped_img.nii.gz"
43 | label_name = "./result/warped_label.nii.gz"
44 | # 图像的保存
45 | ants.image_write(warped_img, img_name)
46 | ants.image_write(warped_label, label_name)
47 | 
48 | # 将antsimage转化为numpy数组
49 | warped_img_arr = warped_img.numpy(single_components=False)
50 | # 从numpy数组得到antsimage
51 | img = ants.from_numpy(warped_img_arr, origin=None, spacing=None, direction=None, has_components=False, is_rgb=False)
52 | # 生成图像的雅克比行列式
53 | jac = ants.create_jacobian_determinant_image(domain_image=f_img, tx=mytx["fwdtransforms"][0], do_log=False, geom=False)
54 | ants.image_write(jac, "./result/jac.nii.gz")
55 | # 生成带网格的moving图像，实测效果不好
56 | m_grid = ants.create_warped_grid(m_img)
57 | m_grid = ants.create_warped_grid(m_grid, grid_directions=(False, False), transform=mytx['fwdtransforms'],
58 |                                  fixed_reference_image=f_img)
59 | ants.image_write(m_grid, "./result/m_grid.nii.gz")
60 | 
61 | '''
62 | 以下为其他不常用的函数：
63 | 
64 | ANTsTransform.apply_to_image(image, reference=None, interpolation='linear')
65 | ants.read_transform(filename, dimension=2, precision='float')
66 | # transform的格式是".mat"
67 | ants.write_transform(transform, filename)
68 | # field是ANTsImage类型
69 | ants.transform_from_displacement_field(field)
70 | '''
71 | 
72 | print("End")
73 | 


--------------------------------------------------------------------------------
/crop_resize.py:
--------------------------------------------------------------------------------
 1 | import os
 2 | import glob
 3 | import numpy as np
 4 | import SimpleITK as sitk
 5 | 
 6 | 
 7 | # 获取可以包裹mask的最小bounding box
 8 | def get_bbox_from_mask(mask, outside_value=0):
 9 |     mask_voxel_coords = np.where(mask != outside_value)
10 |     minzidx = int(np.min(mask_voxel_coords[0]))
11 |     maxzidx = int(np.max(mask_voxel_coords[0])) + 1
12 |     minxidx = int(np.min(mask_voxel_coords[1]))
13 |     maxxidx = int(np.max(mask_voxel_coords[1])) + 1
14 |     minyidx = int(np.min(mask_voxel_coords[2]))
15 |     maxyidx = int(np.max(mask_voxel_coords[2])) + 1
16 |     return [[minzidx, maxzidx], [minxidx, maxxidx], [minyidx, maxyidx]]
17 | 
18 | 
19 | # 根据bbox截取图片
20 | def crop_to_bbox(image, bbox):
21 |     assert len(image.shape) == 3, "only supports 3d images"
22 |     # slice是切片函数，参数为：起始值，终止值，[步长]
23 |     resizer = (slice(bbox[0][0], bbox[0][1]), slice(bbox[1][0], bbox[1][1]), slice(bbox[2][0], bbox[2][1]))
24 |     return image[resizer]
25 | 
26 | 
27 | out_path = r"C:\Users\zuzhiang\Desktop"
28 | files = glob.glob(os.path.join(r"C:\Data\1.Work\03.DataSet\LPBA40\train\256-256-256", "*.nii.gz"))
29 | print("Number of images: ", len(files))
30 | 
31 | minx, miny, minz, maxx, maxy, maxz = 300, 300, 300, 0, 0, 0
32 | for file in files:
33 |     _, name = os.path.split(file)
34 |     print("Name: ", name)
35 |     old_img = sitk.ReadImage(file)
36 |     img_arr = sitk.GetArrayFromImage(old_img)
37 |     bbox = get_bbox_from_mask(img_arr)
38 |     if bbox[0][0] < minx:
39 |         minx = bbox[0][0]
40 |     if bbox[0][1] > maxx:
41 |         maxx = bbox[0][1]
42 |     if bbox[1][0] < miny:
43 |         miny = bbox[1][0]
44 |     if bbox[1][1] > maxy:
45 |         maxy = bbox[1][1]
46 |     if bbox[2][0] < minz:
47 |         minz = bbox[2][0]
48 |     if bbox[2][1] > maxz:
49 |         maxz = bbox[2][1]
50 |     print("bbox: ", bbox)
51 | print("\nminx: %d  maxx: %d  miny: %d  maxy: %d  minz: %d  maxz: %d" % (minx, maxx, miny, maxy, minz, maxz))
52 | 
53 | # # 根据以上6个值手工得到bbox
54 | # bbox = [[0, 240], [21, 229], [40, 216]]
55 | # print("bbox: ", bbox)
56 | # for file in files:
57 | #     _, name = os.path.split(file)
58 | #     print("Name: ", name)
59 | #     old_img = sitk.ReadImage(file)
60 | #     img_arr = sitk.GetArrayFromImage(old_img)
61 | #     img_arr = crop_to_bbox(img_arr, bbox)
62 | #     print("new shape: ", img_arr.shape)
63 | #     img = sitk.GetImageFromArray(img_arr)
64 | #     img.SetOrigin(old_img.GetOrigin())
65 | #     img.SetDirection(old_img.GetDirection())
66 | #     img.SetSpacing(old_img.GetSpacing())
67 | #     sitk.WriteImage(img, os.path.join(out_path, name))
68 | 


--------------------------------------------------------------------------------
/data_augmentation.py:
--------------------------------------------------------------------------------
  1 | import random
  2 | import numpy as np
  3 | import SimpleITK as sitk
  4 | import matplotlib.pyplot as plt
  5 | 
  6 | np.set_printoptions(threshold=np.inf)
  7 | 
  8 | 
  9 | def resample(image, transform):
 10 |     # Output image Origin, Spacing, Size, Direction are taken from the reference
 11 |     # image in this call to Resample
 12 |     reference_image = image
 13 |     interpolator = sitk.sitkCosineWindowedSinc
 14 |     default_value = -100.0
 15 |     return sitk.Resample(image, reference_image, transform,
 16 |                          interpolator, default_value)
 17 | 
 18 | 
 19 | # 数据增强函数，可以对2D/3D图像随机进行直方图均衡化、翻转、平移、放缩、旋转、灰度值归一化等操作
 20 | def data_augmentation(path=None, image=None, do_bspline=True, do_flip=False, do_trslt=True, do_scale=True, do_rota=True,
 21 |                       show_help_info=False):
 22 |     '''
 23 |     :param path: 图像路径
 24 |     :param image: 输入图像，当path为None使用
 25 |     :param do_bspline: 是否做B样条转换，只在为3D图像时使用，因2D时有问题（没效果）
 26 |     :param do_flip: 是否做翻转操作
 27 |     :param do_trslt: 是否做平移操作
 28 |     :param do_scale: 是否做缩放操作
 29 |     :param do_rota: 是否做旋转操作
 30 |     :param show_help_info: 是否显示帮助信息
 31 |     :return:
 32 |     '''
 33 |     if path is not None:
 34 |         image = sitk.ReadImage(path)
 35 |     image_arr = sitk.GetArrayFromImage(image)
 36 | 
 37 |     # 去除灰度直方图中最亮的0.2%的灰度值
 38 |     value, _, count = np.unique(image_arr, return_inverse=True, return_counts=True)
 39 |     # axis=0，按照行累加。axis=1，按照列累加。axis不给定具体值，就把numpy数组当成一个一维数组。
 40 |     # 累加即当前元素值，等于之前所有元素值之和+当前元素值
 41 |     quantiles = np.cumsum(count).astype(np.float32)
 42 |     quantiles /= quantiles[-1]
 43 |     # 找到灰度直方图中小于0.998的最大位置，并将大于其对应值的值设为该值
 44 |     pos = np.where(quantiles < 0.998)[0]
 45 |     val_max = value[np.max(pos)]
 46 |     image_arr[image_arr > val_max] = val_max
 47 | 
 48 |     # 计算窗宽和窗位
 49 |     min_val, max_val = np.min(image_arr), np.max(image_arr)
 50 |     window_width, window_center = max_val - min_val, (min_val + max_val) / 2
 51 | 
 52 |     dim = image.GetDimension()
 53 |     if show_help_info:
 54 |         print("dim: ", dim)
 55 |     transform = sitk.AffineTransform(dim)
 56 |     matrix = np.array(transform.GetMatrix()).reshape((dim, dim))
 57 |     if dim == 2:
 58 |         if do_flip:
 59 |             # 翻转（flip）
 60 |             if np.random.rand() > 0.5:
 61 |                 image = image[::-1]  # 沿x轴反转
 62 |                 if show_help_info:
 63 |                     print("x axis flip.")
 64 |             if np.random.rand() > 0.5:
 65 |                 image = image[:, ::-1]  # 沿y轴反转
 66 |                 if show_help_info:
 67 |                     print("y axis flip.")
 68 |         if do_trslt:
 69 |             # 平移（translation）
 70 |             x_trslt, y_trslt = np.random.randint(-20, 20, 2)  # 每个轴的平移范围在[-20,20]像素之间
 71 |             transform.SetTranslation((float(x_trslt), float(y_trslt)))
 72 |             if show_help_info:
 73 |                 print("x_trslt: ", x_trslt, "  y_trslt: ", y_trslt)
 74 |         if do_scale:
 75 |             # 缩放（scale）
 76 |             x_scale = 1.0 + random.uniform(-0.1, 0.1)  # 缩放范围为原来的[0.9,1.1]
 77 |             y_scale = 1.0 + random.uniform(-0.1, 0.1)
 78 |             # x_scale, y_scale表示原图与结果图的倍数关系，如scale为2时缩小为原来的0.5
 79 |             matrix[0, 0] = x_scale
 80 |             matrix[1, 1] = y_scale
 81 |             if show_help_info:
 82 |                 print("x_scale: ", x_scale, "  y_scale: ", y_scale)
 83 |         if do_rota:
 84 |             # 旋转（rotation）
 85 |             degree = np.random.randint(-15, 15)  # 旋转角度范围为[-15°,15°]
 86 |             radians = -np.pi * degree / 180.
 87 |             rotation = np.array([[np.cos(radians), -np.sin(radians)], [np.sin(radians), np.cos(radians)]])
 88 |             matrix = np.dot(rotation, matrix)
 89 |             if show_help_info:
 90 |                 print("degree: ", degree)
 91 | 
 92 |     elif dim == 3:
 93 |         if do_bspline:
 94 |             # B样条变换（B Spline）
 95 |             m=5
 96 |             spline_order = 3
 97 |             bspline = sitk.BSplineTransform(dim, spline_order)
 98 |             bspline.SetTransformDomainPhysicalDimensions(image.GetSize())
 99 |             mesh_size = [m, m, m]
100 |             bspline.SetTransformDomainMeshSize(mesh_size)
101 |             # Random displacement of the control points.
102 |             # [13,18]为变形的强度，值越大变形越大
103 |             originalControlPointDisplacements = np.random.random(len(bspline.GetParameters())) * np.random.randint(13,18)
104 |             bspline.SetParameters(originalControlPointDisplacements)
105 |             image = resample(image, bspline)
106 |         if do_flip:
107 |             # 翻转（flip）
108 |             if np.random.rand() > 0.5:
109 |                 image = image[::-1]  # 沿x轴反转
110 |                 if show_help_info:
111 |                     print("x axis flip.")
112 |             if np.random.rand() > 0.5:
113 |                 image = image[:, ::-1]  # 沿y轴反转
114 |                 if show_help_info:
115 |                     print("y axis flip.")
116 |             if np.random.rand() > 0.5:
117 |                 image = image[:, :, ::-1]  # 沿z轴反转
118 |                 if show_help_info:
119 |                     print("z axis flip.")
120 |         if do_trslt:
121 |             # 平移（translation）
122 |             x_trslt, y_trslt, z_trslt = np.random.randint(-20, 20, 3)  # 每个轴的平移范围在[-20,20]像素之间
123 |             transform.SetTranslation((float(x_trslt), float(y_trslt), float(z_trslt)))
124 |             if show_help_info:
125 |                 print("x_trslt: ", x_trslt, "  y_trslt: ", y_trslt, "  z_trslt: ", z_trslt)
126 |         if do_scale:
127 |             # 缩放（scale）
128 |             x_scale = 1.0 + random.uniform(-0.1, 0.1)  # 缩放范围为原来的[0.9,1.1]
129 |             y_scale = 1.0 + random.uniform(-0.1, 0.1)
130 |             z_scale = 1.0 + random.uniform(-0.1, 0.1)
131 |             # x_scale, y_scale, z_scale表示原图与结果图的倍数关系，如scale为2时缩小为原来的0.5
132 |             matrix[0, 0] = x_scale
133 |             matrix[1, 1] = y_scale
134 |             matrix[2, 2] = z_scale
135 |             if show_help_info:
136 |                 print("x_scale: ", x_scale, "  y_scale: ", y_scale, "  z_scale: ", z_scale)
137 |         if do_rota:
138 |             # 旋转（rotation）
139 |             x_dgr, y_dgr, z_dgr = np.random.randint(-15, 15, 3)  # 旋转角度范围为[-15°,15°]
140 |             x_rad = -np.pi * x_dgr / 180.
141 |             y_rad = -np.pi * y_dgr / 180.
142 |             z_rad = -np.pi * z_dgr / 180.
143 |             rotation = np.array([[np.cos(y_rad) * np.cos(z_rad), np.cos(y_rad) * np.sin(z_rad), -np.sin(y_rad)],
144 |                                  [-np.cos(x_rad) * np.sin(z_rad) + np.sin(x_rad) * np.sin(y_rad) * np.cos(z_rad),
145 |                                   np.cos(x_rad) * np.cos(z_rad) + np.sin(x_rad) * np.sin(y_rad) * np.sin(z_rad),
146 |                                   np.sin(x_rad) * np.cos(y_rad)],
147 |                                  [np.sin(x_rad) * np.sin(z_rad) + np.cos(x_rad) * np.sin(y_rad) * np.cos(z_rad),
148 |                                   -np.sin(x_rad) * np.cos(z_rad) + np.cos(x_rad) * np.sin(y_rad) * np.sin(z_rad),
149 |                                   np.cos(x_rad) * np.cos(y_rad)]])
150 |             matrix = np.dot(rotation, matrix)
151 |             if show_help_info:
152 |                 print("x_dgr: ", x_dgr, "  y_dgr: ", y_dgr, "  z_dgr: ", z_dgr)
153 | 
154 |     transform.SetMatrix(matrix.ravel())
155 |     # 以下两行是为了让图像的中心点为物体的中心点
156 |     center = image.GetOrigin() + np.array(image.GetSize()) * np.array(image.GetSpacing()) * 0.5
157 |     transform.SetCenter(center)
158 |     out = resample(image, transform)
159 | 
160 |     # 保持和原图像相同的窗宽和窗位，同时做了归一化
161 |     # 不加此步，则图像会在缩放和旋转的时候导致灰度值改变
162 |     out_arr = sitk.GetArrayFromImage(out)
163 |     min_window = float(window_center) - 0.5 * float(window_width)
164 |     out_arr = (out_arr - min_window) / float(window_width)
165 |     out_arr[out_arr < 0] = 0.
166 |     out_arr[out_arr > 1] = 1.
167 |     if dim == 2:
168 |         return out_arr, dim
169 |     elif dim == 3:
170 |         return sitk.GetImageFromArray(out_arr), dim
171 | 
172 | 
173 | if __name__ == "__main__":
174 |     path_2D = r"C:\Users\zuzhiang\Desktop\1.jpg"
175 |     path_3D = r"C:\Data\1.Work\02.Code\DDR\results\f_img.nii"
176 |     out_path_3D = r"C:\Users\zuzhiang\Desktop\out.nii"
177 |     # out, dim = data_augmentation(path_3D)
178 |     out, dim = data_augmentation(image=sitk.ReadImage(path_3D))
179 |     if dim == 2:
180 |         plt.imshow(out)
181 |         plt.title("result image")
182 |         plt.show()
183 |     elif dim == 3:
184 |         sitk.WriteImage(out, out_path_3D)
185 |     print("end")
186 | 


--------------------------------------------------------------------------------
/dcm2nii.py:
--------------------------------------------------------------------------------
 1 | import os
 2 | import SimpleITK as sitk
 3 | 
 4 | 
 5 | def dcm2nii(dcm, nii):
 6 |     # GetGDCMSeriesIDs读取序列号相同的dcm文件
 7 |     series_id = sitk.ImageSeriesReader.GetGDCMSeriesIDs(dcm)
 8 |     # GetGDCMSeriesFileNames读取序列号相同dcm文件的路径，series[0]代表第一个序列号对应的文件
 9 |     series_file_names = sitk.ImageSeriesReader.GetGDCMSeriesFileNames(dcm, series_id[0])
10 |     print(len(series_file_names))
11 |     series_reader = sitk.ImageSeriesReader()
12 |     series_reader.SetFileNames(series_file_names)
13 |     image3d = series_reader.Execute()
14 |     print("type: ",type(image3d))
15 |     sitk.WriteImage(image3d, nii)
16 | 
17 | 
18 | if __name__=="__main__":
19 |     '''
20 |     dcm对应的文件夹下有很多子文件夹，每个子文件夹是一套dicom图像，将所有的dicom
21 |     图像转换为.nii格式的图像
22 |     '''
23 |     dcm=input("dcm_dir:\n")
24 |     nii = input("nii_dir:\n")
25 |     for root,dirs,files in os.walk(dcm):
26 |         print(dirs)
27 |         break
28 |     for dir in dirs:
29 |         dcm_dir=dcm+"\\"+dir
30 |         if not os.listdir(dcm_dir): #若文件夹为空则不处理
31 |             continue
32 |         nii_file=nii+"\\"+dir+".nii" 
33 |         dcm2nii(dcm_dir,nii_file) #将dicom转为nii


--------------------------------------------------------------------------------
/image_combine_example.py:
--------------------------------------------------------------------------------
  1 | import numpy as np
  2 | import skimage.util as su
  3 | import matplotlib.pyplot as plt
  4 | 
  5 | 
  6 | def paste(img, canvas, i, j, method='replace', export_dtype='float'):
  7 |     """ paste `img` on `canvas` with its left-top corner at (i, j) """
  8 |     # check dtypes
  9 |     img = su.img_as_float(img)
 10 |     canvas = su.img_as_float(canvas)
 11 |     # check shapes
 12 |     if len(img.shape) != 2 or len(img.shape) != 3:
 13 |         if len(canvas.shape) != 2 or len(canvas.shape) != 3:
 14 |             assert AttributeError('dimensions of input images not all equal to 2 or 3!')
 15 |     # check channels
 16 |     # all grayscale image
 17 |     if len(img.shape) == 2 and len(canvas.shape) == 2:
 18 |         pass
 19 |     # `img` color image, possible with alpha channel; `canvas` grayscale image
 20 |     elif len(img.shape) == 3 and len(canvas.shape) == 2:
 21 |         c = img.shape[-1]
 22 |         if c == 3:
 23 |             canvas = np.stack([canvas]*c, axis=-1)
 24 |         if c == 4:
 25 |             canvas = np.stack([canvas]*(c-1)+[np.ones((canvas.shape[0], canvas.shape[1]))], axis=-1)
 26 |     # `canvas` color image, possible with alpha channel; `img` grayscale image
 27 |     elif len(img.shape) == 2 and len(canvas.shape) == 3:
 28 |         c = canvas.shape[-1]
 29 |         if c == 3:
 30 |             img = np.stack([img]*c, axis=-1)
 31 |         if c == 4:
 32 |             img = np.stack([img]*(c-1)+[np.ones((img.shape[0], img.shape[1]))], axis=-1)
 33 |     # all color image
 34 |     elif len(img.shape) == 3 and len(canvas.shape) == 3:
 35 |         if img.shape[-1] == 3 and canvas.shape[-1] == 4:
 36 |             img = np.concatenate([img, np.ones((img.shape[0], img.shape[1], 1))], -1)
 37 |         elif img.shape[-1] == 4 and canvas.shape[-1] == 3:
 38 |             canvas = np.concatenate([canvas, np.ones((canvas.shape[0], canvas.shape[1], 1))], -1)
 39 |         elif img.shape[-1] == canvas.shape[-1]:
 40 |             pass
 41 |         else:
 42 |             assert ValueError('channel number should equal to 3 or 4!')
 43 |     # get shapes
 44 |     h_i, w_i = img.shape[:2]
 45 |     h_c, w_c = canvas.shape[:2]
 46 |     # find extent of `img` on `canvas`
 47 |     i_min = np.max([0, i])
 48 |     i_max = np.min([h_c, i+h_i])
 49 |     j_min = np.max([0, j])
 50 |     j_max = np.min([w_c, j+w_i])
 51 |     # paste `img` on `canvas`
 52 |     if method == 'replace':
 53 |         canvas[i_min:i_max, j_min:j_max] = img[i_min-i:i_max-i, j_min-j:j_max-j]
 54 |     elif method == 'add':
 55 |         canvas[i_min:i_max, j_min:j_max] += img[i_min-i:i_max-i, j_min-j:j_max-j]
 56 |     else:
 57 |         raise ValueError('no such method!')
 58 |     # return `canvas`
 59 |     if export_dtype == 'float':
 60 |         return canvas
 61 |     elif export_dtype == 'ubyte':
 62 |         return su.img_as_ubyte(canvas)
 63 |     else:
 64 |         raise ValueError('no such data type for exporting!')
 65 | 
 66 | 
 67 | def combine_avg(imgs, num_w=10, strides=(10, 10), padding=5, bg_level_1=1.0, bg_level_2=1.0, export_dtype='float'):
 68 |     """ paste contents of `imgs` on a single image with `strides` """
 69 |     # dtypes check
 70 |     imgs = [su.img_as_float(img) for img in imgs]
 71 |     # shapes check
 72 |     shapes = [img.shape for img in imgs]
 73 |     if not all([len(s) == 2 or len(s) == 3 for s in shapes]):
 74 |         assert AttributeError('dimensions of imgs not all 2 or 3!')
 75 |     # find the shape of canvas
 76 |     n = len(imgs)
 77 |     num_h = (n - 1) // num_w + 1
 78 |     h = strides[0]*(num_h-1) + shapes[-1][0]
 79 |     w = strides[1]*(num_w-1) + shapes[-1][1]
 80 |     lt_poses = [(strides[0]*i, strides[1]*j) for i in range(num_h) for j in range(num_w) if i*num_w+j<n]
 81 |     # canvas initialization
 82 |     if all([len(s)==2 for s in shapes]):
 83 |         canvas = np.zeros((h, w), np.float)
 84 |         fig = np.zeros((h+2*padding, w+2*padding), np.float) + bg_level_2
 85 |     else:
 86 |         c = max([s[-1] for s in shapes if len(s)==3])
 87 |         canvas = np.zeros((h, w, c), np.float)
 88 |         fig = np.zeros((h+2*padding, w+2*padding, c), np.float) + bg_level_2
 89 |     # paste `imgs` on `canvas`, average overlapping areas (using ones matrix `counter` to record overlapping times)
 90 |     counter = np.zeros((h, w))
 91 |     for i, lt_pos in enumerate(lt_poses):
 92 |         canvas = paste(imgs[i], canvas, lt_pos[0], lt_pos[1], 'add')
 93 |         counter = paste(np.ones((imgs[i].shape[0], imgs[i].shape[1])), counter, lt_pos[0], lt_pos[1], 'add')
 94 |     canvas[counter<0.5] = bg_level_1
 95 |     counter[counter<0.5] = 1.0
 96 |     if len(canvas.shape) == 2:
 97 |         canvas = canvas / counter
 98 |     else:
 99 |         canvas = canvas / np.expand_dims(counter, -1)
100 |     # add padding, i.e. paste `canvas` on `fig`
101 |     fig = paste(canvas, fig, padding, padding)
102 |     # return
103 |     if export_dtype == 'float':
104 |         return fig
105 |     elif export_dtype == 'ubyte':
106 |         return su.img_as_ubyte(fig)
107 |     else:
108 |         raise ValueError('no such data type for exporting!')
109 | 
110 | 
111 | black = np.zeros((3, 3))
112 | white = np.ones((3, 3))
113 | b = combine_avg([black, white]*15, 5, (3, 3), 0)
114 | 
115 | plt.imshow(b)
116 | plt.show()
117 | 


--------------------------------------------------------------------------------
/img2nii.py:
--------------------------------------------------------------------------------
 1 | import numpy as np
 2 | import SimpleITK as sitk
 3 | import nibabel as nib
 4 | import cv2
 5 | 
 6 | np.set_printoptions(threshold=np.inf)
 7 | input_filename = "./test.img"
 8 | img=sitk.ReadImage(input_filename)
 9 | #将图片转化为数组
10 | img_arr=sitk.GetArrayFromImage(img)
11 | out = sitk.GetImageFromArray(img_arr)
12 | out.SetDirection(img.GetDirection())
13 | sitk.WriteImage(out, "./out.nii")
14 | print("end")
15 | 


--------------------------------------------------------------------------------
/jacobian.py:
--------------------------------------------------------------------------------
 1 | import math
 2 | import torch
 3 | import numpy as np
 4 | import torch.nn as nn
 5 | import SimpleITK as sitk
 6 | import torch.nn.functional as F
 7 | 
 8 | 
 9 | # RCN
10 | def jacobian_det(flow):
11 |     _, std = torch.std(torch.det(torch.det([
12 |         flow[:, :, 1:, :-1, :-1] - flow[:, :, :-1, :-1, :-1] +
13 |         torch.Tensor([1, 0, 0], dtype=torch.float32),
14 |         flow[:, :, :-1, 1:, :-1, :] - flow[:, :, :-1, :-1, :-1] +
15 |         torch.Tensor([0, 1, 0], dtype=torch.float32),
16 |         flow[:, :, :-1, :-1, 1:] - flow[:, :, :-1, :-1, :-1] +
17 |         torch.Tensor([0, 0, 1], dtype=torch.float32)
18 |     ], dim=-1)), dim=[2, 3, 4])
19 |     return std
20 | 
21 | 
22 | # FAIM
23 | def Get_Ja(flow):
24 |     '''
25 |     Calculate the Jacobian value at each point of the displacement map having
26 |     size of b*h*w*d*3 and in the cubic volumn of [-1, 1]^3
27 |     '''
28 |     D_y = (flow[:, 1:, :-1, :-1, :] - flow[:, :-1, :-1, :-1, :])
29 |     D_x = (flow[:, :-1, 1:, :-1, :] - flow[:, :-1, :-1, :-1, :])
30 |     D_z = (flow[:, :-1, :-1, 1:, :] - flow[:, :-1, :-1, :-1, :])
31 |     D1 = (D_x[..., 0] + 1) * ((D_y[..., 1] + 1) * (D_z[..., 2] + 1) - D_z[..., 1] * D_y[..., 2])
32 |     D2 = (D_x[..., 1]) * (D_y[..., 0] * (D_z[..., 2] + 1) - D_y[..., 2] * D_x[..., 0])
33 |     D3 = (D_x[..., 2]) * (D_y[..., 0] * D_z[..., 1] - (D_y[..., 1] + 1) * D_z[..., 0])
34 |     return D1 - D2 + D3
35 | 
36 | 
37 | def NJ_loss(ypred):
38 |     '''
39 |     Penalizing locations where Jacobian has negative determinants
40 |     '''
41 |     Neg_Jac = 0.5 * (torch.abs(Get_Ja(ypred)) - Get_Ja(ypred))
42 |     return torch.sum(Neg_Jac)
43 | 
44 | 
45 | # torch.random.manual_seed(0)
46 | path_m2f=r"C:\Data\1.Work\02.Code\preprocess\data\m2f_flow.nii.gz"
47 | path_f2m=r"C:\Data\1.Work\02.Code\My_VoxelMorph\Result\7_flow.nii" #bdr_flow.nii.gz
48 | m2f_flow=sitk.GetArrayFromImage(sitk.ReadImage(path_m2f))[np.newaxis,...]
49 | f2m_flow=sitk.GetArrayFromImage(sitk.ReadImage(path_f2m))[np.newaxis,...]
50 | print(f2m_flow.shape)
51 | m2f_flow=torch.from_numpy(m2f_flow)
52 | f2m_flow=torch.from_numpy(f2m_flow)
53 | print("flow_shape: ",f2m_flow.shape)
54 | print("shape: ",Get_Ja(m2f_flow).shape)
55 | print("m2f_NJ_loss: ",NJ_loss(m2f_flow).data.item())
56 | print("f2m_NJ_loss: ",NJ_loss(f2m_flow).data.item())
57 | 


--------------------------------------------------------------------------------
/phi/.idea/NII.iml:
--------------------------------------------------------------------------------
 1 | <?xml version="1.0" encoding="UTF-8"?>
 2 | <module type="PYTHON_MODULE" version="4">
 3 |   <component name="NewModuleRootManager">
 4 |     <content url="file://$MODULE_DIR$" />
 5 |     <orderEntry type="inheritedJdk" />
 6 |     <orderEntry type="sourceFolder" forTests="false" />
 7 |   </component>
 8 |   <component name="TestRunnerService">
 9 |     <option name="projectConfiguration" value="pytest" />
10 |     <option name="PROJECT_TEST_RUNNER" value="pytest" />
11 |   </component>
12 | </module>


--------------------------------------------------------------------------------
/phi/.idea/deployment.xml:
--------------------------------------------------------------------------------
 1 | <?xml version="1.0" encoding="UTF-8"?>
 2 | <project version="4">
 3 |   <component name="PublishConfigData">
 4 |     <serverData>
 5 |       <paths name="zuzhiang_B808">
 6 |         <serverdata>
 7 |           <mappings>
 8 |             <mapping local="$PROJECT_DIR$" web="/" />
 9 |           </mappings>
10 |         </serverdata>
11 |       </paths>
12 |     </serverData>
13 |   </component>
14 | </project>


--------------------------------------------------------------------------------
/phi/.idea/inspectionProfiles/profiles_settings.xml:
--------------------------------------------------------------------------------
1 | <component name="InspectionProjectProfileManager">
2 |   <settings>
3 |     <option name="USE_PROJECT_PROFILE" value="false" />
4 |     <version value="1.0" />
5 |   </settings>
6 | </component>


--------------------------------------------------------------------------------
/phi/.idea/misc.xml:
--------------------------------------------------------------------------------
1 | <?xml version="1.0" encoding="UTF-8"?>
2 | <project version="4">
3 |   <component name="JavaScriptSettings">
4 |     <option name="languageLevel" value="ES6" />
5 |   </component>
6 |   <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.7" project-jdk-type="Python SDK" />
7 | </project>


--------------------------------------------------------------------------------
/phi/.idea/modules.xml:
--------------------------------------------------------------------------------
1 | <?xml version="1.0" encoding="UTF-8"?>
2 | <project version="4">
3 |   <component name="ProjectModuleManager">
4 |     <modules>
5 |       <module fileurl="file://$PROJECT_DIR$/.idea/NII.iml" filepath="$PROJECT_DIR$/.idea/NII.iml" />
6 |     </modules>
7 |   </component>
8 | </project>


--------------------------------------------------------------------------------
/phi/.idea/workspace.xml:
--------------------------------------------------------------------------------
  1 | <?xml version="1.0" encoding="UTF-8"?>
  2 | <project version="4">
  3 |   <component name="ChangeListManager">
  4 |     <list default="true" id="6b91bcce-5732-40f7-b68a-1dce039bf4cd" name="Default Changelist" comment="" />
  5 |     <option name="EXCLUDED_CONVERTED_TO_IGNORED" value="true" />
  6 |     <option name="SHOW_DIALOG" value="false" />
  7 |     <option name="HIGHLIGHT_CONFLICTS" value="true" />
  8 |     <option name="HIGHLIGHT_NON_ACTIVE_CHANGELIST" value="false" />
  9 |     <option name="LAST_RESOLUTION" value="IGNORE" />
 10 |   </component>
 11 |   <component name="FileTemplateManagerImpl">
 12 |     <option name="RECENT_TEMPLATES">
 13 |       <list>
 14 |         <option value="Python Script" />
 15 |       </list>
 16 |     </option>
 17 |   </component>
 18 |   <component name="ProjectId" id="1dIhdVQ2bUpkk8hy3r87pwDZ0sx" />
 19 |   <component name="PropertiesComponent">
 20 |     <property name="WebServerToolWindowFactoryState" value="true" />
 21 |     <property name="last_opened_file_path" value="$PROJECT_DIR$/../ICNet" />
 22 |     <property name="nodejs_interpreter_path.stuck_in_default_project" value="undefined stuck path" />
 23 |     <property name="nodejs_npm_path_reset_for_default_project" value="true" />
 24 |     <property name="settings.editor.selected.configurable" value="configurable.group.appearance" />
 25 |   </component>
 26 |   <component name="RunDashboard">
 27 |     <option name="ruleStates">
 28 |       <list>
 29 |         <RuleState>
 30 |           <option name="name" value="ConfigurationTypeDashboardGroupingRule" />
 31 |         </RuleState>
 32 |         <RuleState>
 33 |           <option name="name" value="StatusDashboardGroupingRule" />
 34 |         </RuleState>
 35 |       </list>
 36 |     </option>
 37 |   </component>
 38 |   <component name="RunManager" selected="Python.zuzhiang">
 39 |     <configuration name="my_show_grid" type="PythonConfigurationType" factoryName="Python" temporary="true">
 40 |       <module name="NII" />
 41 |       <option name="INTERPRETER_OPTIONS" value="" />
 42 |       <option name="PARENT_ENVS" value="true" />
 43 |       <envs>
 44 |         <env name="PYTHONUNBUFFERED" value="1" />
 45 |       </envs>
 46 |       <option name="SDK_HOME" value="" />
 47 |       <option name="WORKING_DIRECTORY" value="$PROJECT_DIR$" />
 48 |       <option name="IS_MODULE_SDK" value="true" />
 49 |       <option name="ADD_CONTENT_ROOTS" value="true" />
 50 |       <option name="ADD_SOURCE_ROOTS" value="true" />
 51 |       <EXTENSION ID="PythonCoverageRunConfigurationExtension" runner="coverage.py" />
 52 |       <option name="SCRIPT_NAME" value="$PROJECT_DIR$/my_show_grid.py" />
 53 |       <option name="PARAMETERS" value="" />
 54 |       <option name="SHOW_COMMAND_LINE" value="false" />
 55 |       <option name="EMULATE_TERMINAL" value="false" />
 56 |       <option name="MODULE_MODE" value="false" />
 57 |       <option name="REDIRECT_INPUT" value="false" />
 58 |       <option name="INPUT_FILE" value="" />
 59 |       <method v="2" />
 60 |     </configuration>
 61 |     <configuration name="ncc_loss" type="PythonConfigurationType" factoryName="Python" temporary="true">
 62 |       <module name="NII" />
 63 |       <option name="INTERPRETER_OPTIONS" value="" />
 64 |       <option name="PARENT_ENVS" value="true" />
 65 |       <envs>
 66 |         <env name="PYTHONUNBUFFERED" value="1" />
 67 |       </envs>
 68 |       <option name="SDK_HOME" value="" />
 69 |       <option name="WORKING_DIRECTORY" value="$PROJECT_DIR$" />
 70 |       <option name="IS_MODULE_SDK" value="true" />
 71 |       <option name="ADD_CONTENT_ROOTS" value="true" />
 72 |       <option name="ADD_SOURCE_ROOTS" value="true" />
 73 |       <EXTENSION ID="PythonCoverageRunConfigurationExtension" runner="coverage.py" />
 74 |       <option name="SCRIPT_NAME" value="$PROJECT_DIR$/ncc_loss.py" />
 75 |       <option name="PARAMETERS" value="" />
 76 |       <option name="SHOW_COMMAND_LINE" value="false" />
 77 |       <option name="EMULATE_TERMINAL" value="false" />
 78 |       <option name="MODULE_MODE" value="false" />
 79 |       <option name="REDIRECT_INPUT" value="false" />
 80 |       <option name="INPUT_FILE" value="" />
 81 |       <method v="2" />
 82 |     </configuration>
 83 |     <configuration name="show_dvf" type="PythonConfigurationType" factoryName="Python" temporary="true">
 84 |       <module name="NII" />
 85 |       <option name="INTERPRETER_OPTIONS" value="" />
 86 |       <option name="PARENT_ENVS" value="true" />
 87 |       <envs>
 88 |         <env name="PYTHONUNBUFFERED" value="1" />
 89 |       </envs>
 90 |       <option name="SDK_HOME" value="" />
 91 |       <option name="WORKING_DIRECTORY" value="$PROJECT_DIR$" />
 92 |       <option name="IS_MODULE_SDK" value="true" />
 93 |       <option name="ADD_CONTENT_ROOTS" value="true" />
 94 |       <option name="ADD_SOURCE_ROOTS" value="true" />
 95 |       <EXTENSION ID="PythonCoverageRunConfigurationExtension" runner="coverage.py" />
 96 |       <option name="SCRIPT_NAME" value="$PROJECT_DIR$/show_dvf.py" />
 97 |       <option name="PARAMETERS" value="" />
 98 |       <option name="SHOW_COMMAND_LINE" value="false" />
 99 |       <option name="EMULATE_TERMINAL" value="false" />
100 |       <option name="MODULE_MODE" value="false" />
101 |       <option name="REDIRECT_INPUT" value="false" />
102 |       <option name="INPUT_FILE" value="" />
103 |       <method v="2" />
104 |     </configuration>
105 |     <configuration name="show_grid" type="PythonConfigurationType" factoryName="Python" temporary="true">
106 |       <module name="NII" />
107 |       <option name="INTERPRETER_OPTIONS" value="" />
108 |       <option name="PARENT_ENVS" value="true" />
109 |       <envs>
110 |         <env name="PYTHONUNBUFFERED" value="1" />
111 |       </envs>
112 |       <option name="SDK_HOME" value="" />
113 |       <option name="WORKING_DIRECTORY" value="$PROJECT_DIR$" />
114 |       <option name="IS_MODULE_SDK" value="true" />
115 |       <option name="ADD_CONTENT_ROOTS" value="true" />
116 |       <option name="ADD_SOURCE_ROOTS" value="true" />
117 |       <EXTENSION ID="PythonCoverageRunConfigurationExtension" runner="coverage.py" />
118 |       <option name="SCRIPT_NAME" value="$PROJECT_DIR$/show_grid.py" />
119 |       <option name="PARAMETERS" value="" />
120 |       <option name="SHOW_COMMAND_LINE" value="false" />
121 |       <option name="EMULATE_TERMINAL" value="false" />
122 |       <option name="MODULE_MODE" value="false" />
123 |       <option name="REDIRECT_INPUT" value="false" />
124 |       <option name="INPUT_FILE" value="" />
125 |       <method v="2" />
126 |     </configuration>
127 |     <configuration name="zuzhiang" type="PythonConfigurationType" factoryName="Python" temporary="true">
128 |       <module name="NII" />
129 |       <option name="INTERPRETER_OPTIONS" value="" />
130 |       <option name="PARENT_ENVS" value="true" />
131 |       <envs>
132 |         <env name="PYTHONUNBUFFERED" value="1" />
133 |       </envs>
134 |       <option name="SDK_HOME" value="" />
135 |       <option name="WORKING_DIRECTORY" value="$PROJECT_DIR$" />
136 |       <option name="IS_MODULE_SDK" value="true" />
137 |       <option name="ADD_CONTENT_ROOTS" value="true" />
138 |       <option name="ADD_SOURCE_ROOTS" value="true" />
139 |       <EXTENSION ID="PythonCoverageRunConfigurationExtension" runner="coverage.py" />
140 |       <option name="SCRIPT_NAME" value="$PROJECT_DIR$/zuzhiang.py" />
141 |       <option name="PARAMETERS" value="" />
142 |       <option name="SHOW_COMMAND_LINE" value="false" />
143 |       <option name="EMULATE_TERMINAL" value="false" />
144 |       <option name="MODULE_MODE" value="false" />
145 |       <option name="REDIRECT_INPUT" value="false" />
146 |       <option name="INPUT_FILE" value="" />
147 |       <method v="2" />
148 |     </configuration>
149 |     <recent_temporary>
150 |       <list>
151 |         <item itemvalue="Python.zuzhiang" />
152 |         <item itemvalue="Python.show_grid" />
153 |         <item itemvalue="Python.my_show_grid" />
154 |         <item itemvalue="Python.show_dvf" />
155 |         <item itemvalue="Python.ncc_loss" />
156 |       </list>
157 |     </recent_temporary>
158 |   </component>
159 |   <component name="SvnConfiguration">
160 |     <configuration />
161 |   </component>
162 |   <component name="TaskManager">
163 |     <task active="true" id="Default" summary="Default task">
164 |       <changelist id="6b91bcce-5732-40f7-b68a-1dce039bf4cd" name="Default Changelist" comment="" />
165 |       <created>1592118601422</created>
166 |       <option name="number" value="Default" />
167 |       <option name="presentableId" value="Default" />
168 |       <updated>1592118601422</updated>
169 |       <workItem from="1592118602897" duration="3626000" />
170 |       <workItem from="1592892744074" duration="2062000" />
171 |       <workItem from="1593055240133" duration="370000" />
172 |       <workItem from="1593063881195" duration="1809000" />
173 |       <workItem from="1593135989135" duration="2767000" />
174 |       <workItem from="1593241652713" duration="1065000" />
175 |       <workItem from="1593328320420" duration="65000" />
176 |       <workItem from="1594798812247" duration="616000" />
177 |       <workItem from="1594816182187" duration="4746000" />
178 |       <workItem from="1594969364069" duration="390000" />
179 |       <workItem from="1594970205363" duration="12189000" />
180 |       <workItem from="1595233323094" duration="2048000" />
181 |       <workItem from="1595296098035" duration="877000" />
182 |       <workItem from="1596098829588" duration="2103000" />
183 |     </task>
184 |     <servers />
185 |   </component>
186 |   <component name="TypeScriptGeneratedFilesManager">
187 |     <option name="version" value="1" />
188 |   </component>
189 |   <component name="com.intellij.coverage.CoverageDataManagerImpl">
190 |     <SUITE FILE_PATH="coverage/phi$3Dphi.coverage" NAME="3Dphi Coverage Results" MODIFIED="1593055288307" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
191 |     <SUITE FILE_PATH="coverage/phi$2Dphit.coverage" NAME="2Dphit Coverage Results" MODIFIED="1592900054945" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
192 |     <SUITE FILE_PATH="coverage/NII$nii.coverage" NAME="nii Coverage Results" MODIFIED="1592123037670" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
193 |     <SUITE FILE_PATH="coverage/phi$zuzhiang.coverage" NAME="zuzhiang Coverage Results" MODIFIED="1596103830854" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
194 |     <SUITE FILE_PATH="coverage/phi$show_dvf.coverage" NAME="show_dvf Coverage Results" MODIFIED="1594818888983" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
195 |     <SUITE FILE_PATH="coverage/phi$my_show_grid.coverage" NAME="my_show_grid Coverage Results" MODIFIED="1594984904231" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
196 |     <SUITE FILE_PATH="coverage/phi$nii.coverage" NAME="nii Coverage Results" MODIFIED="1592892912003" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
197 |     <SUITE FILE_PATH="coverage/phi$show_grid.coverage" NAME="show_grid Coverage Results" MODIFIED="1594984929594" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
198 |     <SUITE FILE_PATH="coverage/phi$ncc_loss.coverage" NAME="ncc_loss Coverage Results" MODIFIED="1593241663936" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
199 |   </component>
200 | </project>


--------------------------------------------------------------------------------
/phi/f_img.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/zuzhiang/MedicalImageProcess/4361482e453ff8f949f1df0383541b8f72f33a72/phi/f_img.jpg


--------------------------------------------------------------------------------
/phi/img.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/zuzhiang/MedicalImageProcess/4361482e453ff8f949f1df0383541b8f72f33a72/phi/img.jpg


--------------------------------------------------------------------------------
/phi/img1.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/zuzhiang/MedicalImageProcess/4361482e453ff8f949f1df0383541b8f72f33a72/phi/img1.jpg


--------------------------------------------------------------------------------
/phi/img2.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/zuzhiang/MedicalImageProcess/4361482e453ff8f949f1df0383541b8f72f33a72/phi/img2.jpg


--------------------------------------------------------------------------------
/phi/m_img.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/zuzhiang/MedicalImageProcess/4361482e453ff8f949f1df0383541b8f72f33a72/phi/m_img.jpg


--------------------------------------------------------------------------------
/phi/my_show_grid.py:
--------------------------------------------------------------------------------
 1 | import numpy as np
 2 | import SimpleITK as sitk
 3 | import matplotlib.pyplot as plt
 4 | 
 5 | import torch
 6 | import torch.nn as nn
 7 | import torch.nn.functional as F
 8 | 
 9 | 
10 | # 空间转换网络
11 | class SpatialTransformer(nn.Module):
12 |     # 1.生成网格grid；2.new_grid=grid+flow，即旧网格加上一个位移；3.将网格规范化到[-1,1]；4.根据新网格对原图进行采样
13 |     def __init__(self, size, mode='bilinear'):
14 |         """
15 |         Instiatiate the block
16 |             :param size: size of input to the spatial transformer block
17 |             :param mode: method of interpolation for grid_sampler
18 |         """
19 |         super(SpatialTransformer, self).__init__()
20 | 
21 |         # Create sampling grid
22 |         vectors = [torch.arange(0, s) for s in size]
23 |         grids = torch.meshgrid(vectors)
24 |         grid = torch.stack(grids)  # y, x, z
25 |         grid = torch.unsqueeze(grid, 0)  # add batch
26 |         grid = grid.type(torch.FloatTensor)
27 |         self.register_buffer('grid', grid)
28 | 
29 |         self.mode = mode
30 | 
31 |     def forward(self, src, flow):
32 |         """
33 |         Push the src and flow through the spatial transform block
34 |             :param src: the original moving image
35 |             :param flow: the output from the U-Net
36 |         """
37 |         new_locs = self.grid + flow
38 | 
39 |         shape = flow.shape[2:]
40 | 
41 |         # Need to normalize grid values to [-1, 1] for resampler
42 |         for i in range(len(shape)):
43 |             new_locs[:, i, ...] = 2 * (new_locs[:, i, ...] / (shape[i] - 1) - 0.5)
44 | 
45 |         if len(shape) == 2:
46 |             new_locs = new_locs.permute(0, 2, 3, 1)  # 维度置换，变为0,2,3,1
47 |             new_locs = new_locs[..., [1, 0]]
48 |         elif len(shape) == 3:
49 |             new_locs = new_locs.permute(0, 2, 3, 4, 1)
50 |             new_locs = new_locs[..., [2, 1, 0]]
51 | 
52 |         return F.grid_sample(src, new_locs, mode=self.mode)
53 | 
54 | 
55 | # 生成网格图片
56 | def create_grid(size, path):
57 |     num1, num2 = (size[0] + 10) // 10, (size[1] + 10) // 10  # 改变除数（10），即可改变网格的密度
58 |     x, y = np.meshgrid(np.linspace(-2, 2, num1), np.linspace(-2, 2, num2))
59 | 
60 |     plt.figure(figsize=((size[0] + 10) / 100.0, (size[1] + 10) / 100.0))  # 指定图像大小
61 |     plt.plot(x, y, color="black")
62 |     plt.plot(x.transpose(), y.transpose(), color="black")
63 |     plt.axis('off')  # 不显示坐标轴
64 |     # 去除白色边框
65 |     plt.gca().xaxis.set_major_locator(plt.NullLocator())
66 |     plt.gca().yaxis.set_major_locator(plt.NullLocator())
67 |     plt.subplots_adjust(top=1, bottom=0, left=0, right=1, hspace=0, wspace=0)
68 |     plt.margins(0, 0)
69 |     plt.savefig(path)  # 保存图像
70 |     # plt.show()
71 | 
72 | 
73 | if __name__ == "__main__":
74 |     out_path = r"C:\Users\zuzhiang\Desktop\new_img.nii"  # 图片保存路径
75 |     # 读入形变场
76 |     phi = sitk.ReadImage("./2D1.nii")  # [324,303,2]
77 |     phi_arr = torch.from_numpy(sitk.GetArrayFromImage(phi)).float()
78 |     phi_shape = phi_arr.shape
79 |     # 产生网格图片
80 |     create_grid(phi_shape, out_path)
81 |     img = sitk.GetArrayFromImage(sitk.ReadImage(out_path))[..., 0]
82 |     img = np.squeeze(img)[np.newaxis, np.newaxis, :phi_shape[0], :phi_shape[1]]
83 |     # 用STN根据形变场对网格图片进行变形
84 |     STN = SpatialTransformer(phi_shape[:2])
85 |     phi_arr = phi_arr.permute(2, 0, 1)[np.newaxis, ...]
86 |     warp = STN(torch.from_numpy(img).float(), phi_arr)
87 |     # 保存图片
88 |     warp_img = sitk.GetImageFromArray(warp[0, 0, ...].numpy().astype(np.uint8))
89 |     sitk.WriteImage(warp_img, out_path)
90 |     print("end")
91 | 


--------------------------------------------------------------------------------
/phi/show_grid.py:
--------------------------------------------------------------------------------
 1 | import matplotlib.pyplot as plt
 2 | import SimpleITK as sitk
 3 | import numpy as np
 4 | 
 5 | 
 6 | def grid2contour(grid, title):
 7 |     '''
 8 |     grid--image_grid used to show deform field
 9 |     type: numpy ndarray, shape： (h, w, 2), value range：(-1, 1)
10 |     '''
11 |     assert grid.ndim == 3
12 |     x = np.arange(-1, 1, 2.0 / grid.shape[1])
13 |     y = np.arange(-1, 1, 2.0 / grid.shape[0])
14 |     X, Y = np.meshgrid(x, y)
15 |     Z1 = grid[:, :, 0] + 2  # remove the dashed line
16 |     Z1 = Z1[::-1]  # vertical flip
17 |     Z2 = grid[:, :, 1] + 2
18 | 
19 |     plt.figure()
20 |     plt.contour(X, Y, Z1, 15, levels=50, colors='k') #改变levels的值，可以改变形变场的外貌
21 |     plt.contour(X, Y, Z2, 15, levels=50, colors='k')
22 |     plt.xticks(()), plt.yticks(())  # remove x, y ticks
23 |     plt.title(title)
24 |     plt.show()
25 | 
26 | 
27 | def show_grid():
28 |     img = sitk.ReadImage(r"C:\Users\zuzhiang\Desktop\7_flow.nii.gz")
29 |     img_arr = sitk.GetArrayFromImage(img)[:,:,0,:2]
30 |     img_shape = img_arr.shape
31 |     print("shape: ", img_shape)
32 | 
33 |     # 起点、终点、步长（可为小数）
34 |     x = np.arange(-1, 1, 2 / img_shape[1])
35 |     y = np.arange(-1, 1, 2 / img_shape[0])
36 |     X, Y = np.meshgrid(x, y)
37 |     regular_grid = np.stack((X, Y), axis=2)
38 |     grid2contour(regular_grid, "regular_grid")
39 | 
40 |     rand_field = np.random.rand(*img_shape[:2], 2)  # 参数前加*是以元组形式导入
41 |     rand_field_norm = rand_field.copy()
42 |     rand_field_norm[:, :, 0] = rand_field_norm[:, :, 0] * 2 / img_shape[1]
43 |     rand_field_norm[:, :, 1] = rand_field_norm[:, :, 1] * 2 / img_shape[0]
44 |     sampling_grid = regular_grid + rand_field_norm
45 |     grid2contour(sampling_grid, "sampling_grid")
46 | 
47 |     img_arr[..., 0] = img_arr[..., 0] * 2 / img_shape[1]
48 |     img_arr[..., 1] = img_arr[..., 1] * 2 / img_shape[0]
49 |     img_grid = regular_grid + img_arr
50 |     grid2contour(img_grid, "img_grid")
51 | 
52 | 
53 | if __name__ == "__main__":
54 |     show_grid()
55 |     print("end")
56 | 


--------------------------------------------------------------------------------
/resize_img.py:
--------------------------------------------------------------------------------
 1 | import os
 2 | import glob
 3 | import numpy as np
 4 | import SimpleITK as sitk
 5 | from skimage.transform import resize
 6 | import models.data_augmentation as da
 7 | 
 8 | path = r"C:\Data\1.Work\03.DataSet\LPBA40"
 9 | out_path = r"C:\Data\1.Work\03.DataSet\LPBA40_DA"
10 | #file_lst = glob.glob(os.path.join(path, '*/*bfc.nii'))
11 | file_lst=[r"C:\Data\1.Work\02.Code\DDR\fixed_LPBA40.nii"]
12 | num = len(file_lst)
13 | # for file in file_lst:
14 | #     img, dim = da.data_augmentation(file, do_bspline=False, do_trslt=True, do_scale=True, do_rota=True,
15 | #                                     show_help_info=True)
16 | #     _, name = os.path.split(file)
17 | #     print("path: ", os.path.join(out_path, name))
18 | #     sitk.WriteImage(img, os.path.join(out_path, name))
19 | 
20 | for file in file_lst:
21 |     # img, dim = da.data_augmentation(file, do_bspline=False, do_trslt=False, do_scale=False, do_rota=False,
22 |     # show_help_info=True)
23 |     img = sitk.ReadImage(file)
24 |     img_arr = sitk.GetArrayFromImage(img)
25 |     new_shape = [256, 128, 256]
26 |     new_img = resize(img_arr, new_shape, order=3, mode='constant', cval=0, preserve_range=bool)
27 |     # order几次样条插值，cval外部补零，保留原数据（否则被默认标准化、归一化之类的乱七八糟）
28 |     '''
29 |     The order of interpolation. The order has to be in the range 0-5:
30 |     0: Nearest-neighbor
31 |     1: Bi-linear (default)
32 |     2: Bi-quadratic
33 |     3: Bi-cubic
34 |     4: Bi-quartic
35 |     5: Bi-quintic
36 |     '''
37 |     new_img = sitk.GetImageFromArray(new_img)
38 |     new_img.SetOrigin(img.GetOrigin())
39 |     new_img.SetSpacing(img.GetSpacing())
40 |     new_img.SetDirection(img.GetDirection())
41 |     print("file: ", file)
42 |     #temp_path = r"C:\Users\zuzhiang\Desktop\out.nii"
43 |     sitk.WriteImage(new_img, file)
44 | 


--------------------------------------------------------------------------------
/torch_datagenerators.py:
--------------------------------------------------------------------------------
 1 | import glob
 2 | import numpy as np
 3 | import SimpleITK as sitk
 4 | import torch.utils.data as Data
 5 | 
 6 | '''
 7 | 通过继承Data.Dataset，实现将一组Tensor数据对封装成Tensor数据集
 8 | 至少要重载__init__，__len__和__getitem__方法
 9 | '''
10 | class Dataset(Data.Dataset):
11 |     def __init__(self, path):
12 |         # 初始化
13 |         self.files = glob.glob(path)
14 | 
15 |     def __len__(self):
16 |         # 返回数据集的大小
17 |         return len(self.files)
18 | 
19 |     def __getitem__(self, index):
20 |         # 索引数据集中的某个数据，还可以对数据进行预处理
21 |         # 下标index参数是必须有的，名字任意
22 |         img_arr = sitk.GetArrayFromImage(sitk.ReadImage(self.files[index]))[..., np.newaxis]
23 |         return img_arr


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