├── LICENSE ├── README.md ├── brain-tumor-segmentation └── README.md ├── figures └── unetr_architecture.png ├── metrics.py ├── results ├── 114.png ├── 124.png ├── 132.png ├── 135.png ├── 2.png ├── 21.png ├── 23.png ├── 26.png ├── 58.png ├── 6.png ├── 60.png ├── 68.png ├── 71.png ├── 75.png ├── 77.png ├── 85.png ├── 86.png ├── 9.png ├── 92.png └── README.md ├── test.py ├── train.py └── unetr_2d.py /LICENSE: -------------------------------------------------------------------------------- 1 | Apache License 2 | Version 2.0, January 2004 3 | http://www.apache.org/licenses/ 4 | 5 | TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 6 | 7 | 1. Definitions. 8 | 9 | "License" shall mean the terms and conditions for use, reproduction, 10 | and distribution as defined by Sections 1 through 9 of this document. 11 | 12 | "Licensor" shall mean the copyright owner or entity authorized by 13 | the copyright owner that is granting the License. 14 | 15 | "Legal Entity" shall mean the union of the acting entity and all 16 | other entities that control, are controlled by, or are under common 17 | control with that entity. For the purposes of this definition, 18 | "control" means (i) the power, direct or indirect, to cause the 19 | direction or management of such entity, whether by contract or 20 | otherwise, or (ii) ownership of fifty percent (50%) or more of the 21 | outstanding shares, or (iii) beneficial ownership of such entity. 22 | 23 | "You" (or "Your") shall mean an individual or Legal Entity 24 | exercising permissions granted by this License. 25 | 26 | "Source" form shall mean the preferred form for making modifications, 27 | including but not limited to software source code, documentation 28 | source, and configuration files. 29 | 30 | "Object" form shall mean any form resulting from mechanical 31 | transformation or translation of a Source form, including but 32 | not limited to compiled object code, generated documentation, 33 | and conversions to other media types. 34 | 35 | "Work" shall mean the work of authorship, whether in Source or 36 | Object form, made available under the License, as indicated by a 37 | copyright notice that is included in or attached to the work 38 | (an example is provided in the Appendix below). 39 | 40 | "Derivative Works" shall mean any work, whether in Source or Object 41 | form, that is based on (or derived from) the Work and for which the 42 | editorial revisions, annotations, elaborations, or other modifications 43 | represent, as a whole, an original work of authorship. For the purposes 44 | of this License, Derivative Works shall not include works that remain 45 | separable from, or merely link (or bind by name) to the interfaces of, 46 | the Work and Derivative Works thereof. 47 | 48 | "Contribution" shall mean any work of authorship, including 49 | the original version of the Work and any modifications or additions 50 | to that Work or Derivative Works thereof, that is intentionally 51 | submitted to Licensor for inclusion in the Work by the copyright owner 52 | or by an individual or Legal Entity authorized to submit on behalf of 53 | the copyright owner. For the purposes of this definition, "submitted" 54 | means any form of electronic, verbal, or written communication sent 55 | to the Licensor or its representatives, including but not limited to 56 | communication on electronic mailing lists, source code control systems, 57 | and issue tracking systems that are managed by, or on behalf of, the 58 | Licensor for the purpose of discussing and improving the Work, but 59 | excluding communication that is conspicuously marked or otherwise 60 | designated in writing by the copyright owner as "Not a Contribution." 61 | 62 | "Contributor" shall mean Licensor and any individual or Legal Entity 63 | on behalf of whom a Contribution has been received by Licensor and 64 | subsequently incorporated within the Work. 65 | 66 | 2. Grant of Copyright License. Subject to the terms and conditions of 67 | this License, each Contributor hereby grants to You a perpetual, 68 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 69 | copyright license to reproduce, prepare Derivative Works of, 70 | publicly display, publicly perform, sublicense, and distribute the 71 | Work and such Derivative Works in Source or Object form. 72 | 73 | 3. Grant of Patent License. Subject to the terms and conditions of 74 | this License, each Contributor hereby grants to You a perpetual, 75 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 76 | (except as stated in this section) patent license to make, have made, 77 | use, offer to sell, sell, import, and otherwise transfer the Work, 78 | where such license applies only to those patent claims licensable 79 | by such Contributor that are necessarily infringed by their 80 | Contribution(s) alone or by combination of their Contribution(s) 81 | with the Work to which such Contribution(s) was submitted. If You 82 | institute patent litigation against any entity (including a 83 | cross-claim or counterclaim in a lawsuit) alleging that the Work 84 | or a Contribution incorporated within the Work constitutes direct 85 | or contributory patent infringement, then any patent licenses 86 | granted to You under this License for that Work shall terminate 87 | as of the date such litigation is filed. 88 | 89 | 4. Redistribution. You may reproduce and distribute copies of the 90 | Work or Derivative Works thereof in any medium, with or without 91 | modifications, and in Source or Object form, provided that You 92 | meet the following conditions: 93 | 94 | (a) You must give any other recipients of the Work or 95 | Derivative Works a copy of this License; and 96 | 97 | (b) You must cause any modified files to carry prominent notices 98 | stating that You changed the files; and 99 | 100 | (c) You must retain, in the Source form of any Derivative Works 101 | that You distribute, all copyright, patent, trademark, and 102 | attribution notices from the Source form of the Work, 103 | excluding those notices that do not pertain to any part of 104 | the Derivative Works; and 105 | 106 | (d) If the Work includes a "NOTICE" text file as part of its 107 | distribution, then any Derivative Works that You distribute must 108 | include a readable copy of the attribution notices contained 109 | within such NOTICE file, excluding those notices that do not 110 | pertain to any part of the Derivative Works, in at least one 111 | of the following places: within a NOTICE text file distributed 112 | as part of the Derivative Works; within the Source form or 113 | documentation, if provided along with the Derivative Works; or, 114 | within a display generated by the Derivative Works, if and 115 | wherever such third-party notices normally appear. The contents 116 | of the NOTICE file are for informational purposes only and 117 | do not modify the License. You may add Your own attribution 118 | notices within Derivative Works that You distribute, alongside 119 | or as an addendum to the NOTICE text from the Work, provided 120 | that such additional attribution notices cannot be construed 121 | as modifying the License. 122 | 123 | You may add Your own copyright statement to Your modifications and 124 | may provide additional or different license terms and conditions 125 | for use, reproduction, or distribution of Your modifications, or 126 | for any such Derivative Works as a whole, provided Your use, 127 | reproduction, and distribution of the Work otherwise complies with 128 | the conditions stated in this License. 129 | 130 | 5. Submission of Contributions. Unless You explicitly state otherwise, 131 | any Contribution intentionally submitted for inclusion in the Work 132 | by You to the Licensor shall be under the terms and conditions of 133 | this License, without any additional terms or conditions. 134 | Notwithstanding the above, nothing herein shall supersede or modify 135 | the terms of any separate license agreement you may have executed 136 | with Licensor regarding such Contributions. 137 | 138 | 6. Trademarks. This License does not grant permission to use the trade 139 | names, trademarks, service marks, or product names of the Licensor, 140 | except as required for reasonable and customary use in describing the 141 | origin of the Work and reproducing the content of the NOTICE file. 142 | 143 | 7. Disclaimer of Warranty. Unless required by applicable law or 144 | agreed to in writing, Licensor provides the Work (and each 145 | Contributor provides its Contributions) on an "AS IS" BASIS, 146 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or 147 | implied, including, without limitation, any warranties or conditions 148 | of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A 149 | PARTICULAR PURPOSE. You are solely responsible for determining the 150 | appropriateness of using or redistributing the Work and assume any 151 | risks associated with Your exercise of permissions under this License. 152 | 153 | 8. Limitation of Liability. In no event and under no legal theory, 154 | whether in tort (including negligence), contract, or otherwise, 155 | unless required by applicable law (such as deliberate and grossly 156 | negligent acts) or agreed to in writing, shall any Contributor be 157 | liable to You for damages, including any direct, indirect, special, 158 | incidental, or consequential damages of any character arising as a 159 | result of this License or out of the use or inability to use the 160 | Work (including but not limited to damages for loss of goodwill, 161 | work stoppage, computer failure or malfunction, or any and all 162 | other commercial damages or losses), even if such Contributor 163 | has been advised of the possibility of such damages. 164 | 165 | 9. Accepting Warranty or Additional Liability. While redistributing 166 | the Work or Derivative Works thereof, You may choose to offer, 167 | and charge a fee for, acceptance of support, warranty, indemnity, 168 | or other liability obligations and/or rights consistent with this 169 | License. However, in accepting such obligations, You may act only 170 | on Your own behalf and on Your sole responsibility, not on behalf 171 | of any other Contributor, and only if You agree to indemnify, 172 | defend, and hold each Contributor harmless for any liability 173 | incurred by, or claims asserted against, such Contributor by reason 174 | of your accepting any such warranty or additional liability. 175 | 176 | END OF TERMS AND CONDITIONS 177 | 178 | APPENDIX: How to apply the Apache License to your work. 179 | 180 | To apply the Apache License to your work, attach the following 181 | boilerplate notice, with the fields enclosed by brackets "[]" 182 | replaced with your own identifying information. (Don't include 183 | the brackets!) The text should be enclosed in the appropriate 184 | comment syntax for the file format. We also recommend that a 185 | file or class name and description of purpose be included on the 186 | same "printed page" as the copyright notice for easier 187 | identification within third-party archives. 188 | 189 | Copyright [yyyy] [name of copyright owner] 190 | 191 | Licensed under the Apache License, Version 2.0 (the "License"); 192 | you may not use this file except in compliance with the License. 193 | You may obtain a copy of the License at 194 | 195 | http://www.apache.org/licenses/LICENSE-2.0 196 | 197 | Unless required by applicable law or agreed to in writing, software 198 | distributed under the License is distributed on an "AS IS" BASIS, 199 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 200 | See the License for the specific language governing permissions and 201 | limitations under the License. 202 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow 2 | This repository demonstrates the utilization of UNETR for brain tumor segmentation. 3 | 4 | ## Architecture 5 | 6 | | ![The block diagram of the Original UNETR model.](figures/unetr_architecture.png) | 7 | | :--: | 8 | | *The block diagram of the Original UNETR model.* | 9 | 10 | ## Dataset 11 | The dataset contains 3064 pairs of MRI brain images and their respective binary mask indicating tumor. 12 |

13 | Download the dataset: [Brain Tumor Segmentation](https://www.kaggle.com/datasets/nikhilroxtomar/brain-tumor-segmentation) 14 |
15 | Original Dataset: [Brain Tumor Segmentation](https://figshare.com/articles/dataset/brain_tumor_dataset/1512427) 16 | 17 | ## Results 18 | The sequence in the images below is `Input Image`, `Ground Truth` and `Prediction`.

19 | | ![](results/2.png) | 20 | | :--: | 21 | | ![](results/6.png) | 22 | | ![](results/19.png) | 23 | | ![](results/21.png) | 24 | | ![](results/68.png) | 25 | 26 | ## How to improve 27 | - Train on more epochs. 28 | - Increase the input image resolution. 29 | - Apply data augmentation. 30 | 31 | 32 | ## Contact 33 | For more follow me on: 34 | 35 | - YouTube 36 | - Facebook 37 | - Twitter 38 | - Instagram 39 | - Telegram 40 | -------------------------------------------------------------------------------- /brain-tumor-segmentation/README.md: -------------------------------------------------------------------------------- 1 | # Dataset 2 | -------------------------------------------------------------------------------- /figures/unetr_architecture.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/figures/unetr_architecture.png -------------------------------------------------------------------------------- /metrics.py: -------------------------------------------------------------------------------- 1 | import numpy as np 2 | import tensorflow as tf 3 | 4 | smooth = 1e-15 5 | def dice_coef(y_true, y_pred): 6 | y_true = tf.keras.layers.Flatten()(y_true) 7 | y_pred = tf.keras.layers.Flatten()(y_pred) 8 | intersection = tf.reduce_sum(y_true * y_pred) 9 | return (2. * intersection + smooth) / (tf.reduce_sum(y_true) + tf.reduce_sum(y_pred) + smooth) 10 | 11 | def dice_loss(y_true, y_pred): 12 | return 1.0 - dice_coef(y_true, y_pred) 13 | -------------------------------------------------------------------------------- /results/114.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/114.png -------------------------------------------------------------------------------- /results/124.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/124.png -------------------------------------------------------------------------------- /results/132.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/132.png -------------------------------------------------------------------------------- /results/135.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/135.png -------------------------------------------------------------------------------- /results/2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/2.png -------------------------------------------------------------------------------- /results/21.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/21.png -------------------------------------------------------------------------------- /results/23.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/23.png -------------------------------------------------------------------------------- /results/26.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/26.png -------------------------------------------------------------------------------- /results/58.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/58.png -------------------------------------------------------------------------------- /results/6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/6.png -------------------------------------------------------------------------------- /results/60.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/60.png -------------------------------------------------------------------------------- /results/68.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/68.png -------------------------------------------------------------------------------- /results/71.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/71.png -------------------------------------------------------------------------------- /results/75.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/75.png -------------------------------------------------------------------------------- /results/77.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/77.png -------------------------------------------------------------------------------- /results/85.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/85.png -------------------------------------------------------------------------------- /results/86.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/86.png -------------------------------------------------------------------------------- /results/9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/9.png -------------------------------------------------------------------------------- /results/92.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/nikhilroxtomar/Brain-Tumor-Segmentation-using-UNETR-in-TensorFlow/d3752891c438f8961bf1c680d4b431133f6cf586/results/92.png -------------------------------------------------------------------------------- /results/README.md: -------------------------------------------------------------------------------- 1 | # Results 2 | -------------------------------------------------------------------------------- /test.py: -------------------------------------------------------------------------------- 1 | import os 2 | os.environ["TF_CPP_MIN_LOG_LEVEL"] = "2" 3 | 4 | import numpy as np 5 | import cv2 6 | from tqdm import tqdm 7 | import tensorflow as tf 8 | from patchify import patchify 9 | from train import load_dataset, create_dir 10 | from metrics import dice_loss, dice_coef 11 | 12 | 13 | """ UNETR Configration """ 14 | cf = {} 15 | cf["image_size"] = 256 16 | cf["num_channels"] = 3 17 | cf["num_layers"] = 12 18 | cf["hidden_dim"] = 128 19 | cf["mlp_dim"] = 32 20 | cf["num_heads"] = 6 21 | cf["dropout_rate"] = 0.1 22 | cf["patch_size"] = 16 23 | cf["num_patches"] = (cf["image_size"]**2)//(cf["patch_size"]**2) 24 | cf["flat_patches_shape"] = ( 25 | cf["num_patches"], 26 | cf["patch_size"]*cf["patch_size"]*cf["num_channels"] 27 | ) 28 | 29 | 30 | if __name__ == "__main__": 31 | """ Seeding """ 32 | np.random.seed(42) 33 | tf.random.set_seed(42) 34 | 35 | """ Directory for storing files """ 36 | create_dir(f"results") 37 | 38 | """ Load the model """ 39 | model_path = os.path.join("files", "model.h5") 40 | model = tf.keras.models.load_model(model_path, custom_objects={"dice_loss": dice_loss, "dice_coef": dice_coef}) 41 | 42 | """ Dataset """ 43 | dataset_path = "brain-tumor-segmentation" 44 | (train_x, train_y), (valid_x, valid_y), (test_x, test_y) = load_dataset(dataset_path) 45 | 46 | print(f"Train: \t{len(train_x)} - {len(train_y)}") 47 | print(f"Valid: \t{len(valid_x)} - {len(valid_y)}") 48 | print(f"Test: \t{len(test_x)} - {len(test_y)}") 49 | 50 | """ Prediction """ 51 | for x, y in tqdm(zip(test_x, test_y), total=len(test_x)): 52 | """ Extracting the name """ 53 | name = x.split("/")[-1] 54 | 55 | """ Reading the image """ 56 | image = cv2.imread(x, cv2.IMREAD_COLOR) 57 | image = cv2.resize(image, (cf["image_size"], cf["image_size"])) 58 | x = image / 255.0 59 | 60 | patch_shape = (cf["patch_size"], cf["patch_size"], cf["num_channels"]) 61 | patches = patchify(x, patch_shape, cf["patch_size"]) 62 | patches = np.reshape(patches, cf["flat_patches_shape"]) 63 | patches = patches.astype(np.float32) 64 | patches = np.expand_dims(patches, axis=0) 65 | 66 | """ Read Mask """ 67 | mask = cv2.imread(y, cv2.IMREAD_GRAYSCALE) 68 | mask = cv2.resize(mask, (cf["image_size"], cf["image_size"])) 69 | mask = mask / 255.0 70 | mask = np.expand_dims(mask, axis=-1) 71 | mask = np.concatenate([mask, mask, mask], axis=-1) 72 | 73 | """ Prediction """ 74 | pred = model.predict(patches, verbose=0)[0] 75 | pred = np.concatenate([pred, pred, pred], axis=-1) 76 | 77 | """ Save final mask """ 78 | line = np.ones((cf["image_size"], 10, 3)) * 255 79 | cat_images = np.concatenate([image, line, mask*255, line, pred*255], axis=1) 80 | save_image_path = os.path.join("results", name) 81 | cv2.imwrite(save_image_path, cat_images) 82 | 83 | 84 | 85 | 86 | 87 | 88 | 89 | 90 | 91 | 92 | ## ... 93 | -------------------------------------------------------------------------------- /train.py: -------------------------------------------------------------------------------- 1 | import os 2 | os.environ["TF_CPP_MIN_LOG_LEVEL"] = "2" 3 | 4 | import numpy as np 5 | import cv2 6 | from glob import glob 7 | from sklearn.utils import shuffle 8 | import tensorflow as tf 9 | from tensorflow.keras.callbacks import ModelCheckpoint, CSVLogger, ReduceLROnPlateau, EarlyStopping 10 | from tensorflow.keras.optimizers import Adam, SGD 11 | from sklearn.model_selection import train_test_split 12 | from patchify import patchify 13 | from unetr_2d import build_unetr_2d 14 | from metrics import dice_loss, dice_coef 15 | 16 | 17 | """ UNETR Configration """ 18 | cf = {} 19 | cf["image_size"] = 256 20 | cf["num_channels"] = 3 21 | cf["num_layers"] = 12 22 | cf["hidden_dim"] = 128 23 | cf["mlp_dim"] = 32 24 | cf["num_heads"] = 6 25 | cf["dropout_rate"] = 0.1 26 | cf["patch_size"] = 16 27 | cf["num_patches"] = (cf["image_size"]**2)//(cf["patch_size"]**2) 28 | cf["flat_patches_shape"] = ( 29 | cf["num_patches"], 30 | cf["patch_size"]*cf["patch_size"]*cf["num_channels"] 31 | ) 32 | 33 | def create_dir(path): 34 | if not os.path.exists(path): 35 | os.makedirs(path) 36 | 37 | def load_dataset(path, split=0.1): 38 | """ Loading the images and masks """ 39 | X = sorted(glob(os.path.join(path, "images", "*.png"))) 40 | Y = sorted(glob(os.path.join(path, "masks", "*.png"))) 41 | 42 | """ Spliting the data into training and testing """ 43 | split_size = int(len(X) * split) 44 | 45 | train_x, valid_x = train_test_split(X, test_size=split_size, random_state=42) 46 | train_y, valid_y = train_test_split(Y, test_size=split_size, random_state=42) 47 | 48 | train_x, test_x = train_test_split(train_x, test_size=split_size, random_state=42) 49 | train_y, test_y = train_test_split(train_y, test_size=split_size, random_state=42) 50 | 51 | return (train_x, train_y), (valid_x, valid_y), (test_x, test_y) 52 | 53 | def read_image(path): 54 | path = path.decode() 55 | image = cv2.imread(path, cv2.IMREAD_COLOR) 56 | image = cv2.resize(image, (cf["image_size"], cf["image_size"])) 57 | image = image / 255.0 58 | 59 | """ Processing to patches """ 60 | patch_shape = (cf["patch_size"], cf["patch_size"], cf["num_channels"]) 61 | patches = patchify(image, patch_shape, cf["patch_size"]) 62 | patches = np.reshape(patches, cf["flat_patches_shape"]) 63 | patches = patches.astype(np.float32) 64 | 65 | return patches 66 | 67 | def read_mask(path): 68 | path = path.decode() 69 | mask = cv2.imread(path, cv2.IMREAD_GRAYSCALE) 70 | mask = cv2.resize(mask, (cf["image_size"], cf["image_size"])) 71 | mask = mask / 255.0 72 | mask = mask.astype(np.float32) 73 | mask = np.expand_dims(mask, axis=-1) 74 | return mask 75 | 76 | def tf_parse(x, y): 77 | def _parse(x, y): 78 | x = read_image(x) 79 | y = read_mask(y) 80 | return x, y 81 | 82 | x, y = tf.numpy_function(_parse, [x, y], [tf.float32, tf.float32]) 83 | x.set_shape(cf["flat_patches_shape"]) 84 | y.set_shape([cf["image_size"], cf["image_size"], 1]) 85 | return x, y 86 | 87 | def tf_dataset(X, Y, batch=2): 88 | ds = tf.data.Dataset.from_tensor_slices((X, Y)) 89 | ds = ds.map(tf_parse).batch(batch).prefetch(10) 90 | return ds 91 | 92 | 93 | if __name__ == "__main__": 94 | """ Seeding """ 95 | np.random.seed(42) 96 | tf.random.set_seed(42) 97 | 98 | """ Directory for storing files """ 99 | create_dir("files") 100 | 101 | """ Hyperparameters """ 102 | batch_size = 8 103 | lr = 0.1 104 | num_epochs = 500 105 | model_path = os.path.join("files", "model.h5") 106 | csv_path = os.path.join("files", "log.csv") 107 | 108 | """ Dataset """ 109 | dataset_path = "brain-tumor-segmentation" 110 | (train_x, train_y), (valid_x, valid_y), (test_x, test_y) = load_dataset(dataset_path) 111 | 112 | print(f"Train: \t{len(train_x)} - {len(train_y)}") 113 | print(f"Valid: \t{len(valid_x)} - {len(valid_y)}") 114 | print(f"Test: \t{len(test_x)} - {len(test_y)}") 115 | 116 | train_dataset = tf_dataset(train_x, train_y, batch=batch_size) 117 | valid_dataset = tf_dataset(valid_x, valid_y, batch=batch_size) 118 | 119 | """ Model """ 120 | model = build_unetr_2d(cf) 121 | model.compile(loss=dice_loss, optimizer=SGD(lr), metrics=[dice_coef, "acc"]) 122 | # model.summary() 123 | 124 | callbacks = [ 125 | ModelCheckpoint(model_path, verbose=1, save_best_only=True), 126 | ReduceLROnPlateau(monitor='val_loss', factor=0.1, patience=5, min_lr=1e-7, verbose=1), 127 | CSVLogger(csv_path), 128 | EarlyStopping(monitor='val_loss', patience=20, restore_best_weights=False) 129 | ] 130 | 131 | model.fit( 132 | train_dataset, 133 | epochs=num_epochs, 134 | validation_data=valid_dataset, 135 | callbacks=callbacks 136 | ) 137 | -------------------------------------------------------------------------------- /unetr_2d.py: -------------------------------------------------------------------------------- 1 | 2 | import os 3 | os.environ["TF_CPP_MIN_LOG_LEVEL"] = "2" 4 | 5 | from math import log2 6 | import tensorflow as tf 7 | import tensorflow.keras.layers as L 8 | from tensorflow.keras.models import Model 9 | 10 | def mlp(x, cf): 11 | x = L.Dense(cf["mlp_dim"], activation="gelu")(x) 12 | x = L.Dropout(cf["dropout_rate"])(x) 13 | x = L.Dense(cf["hidden_dim"])(x) 14 | x = L.Dropout(cf["dropout_rate"])(x) 15 | return x 16 | 17 | def transformer_encoder(x, cf): 18 | skip_1 = x 19 | x = L.LayerNormalization()(x) 20 | x = L.MultiHeadAttention( 21 | num_heads=cf["num_heads"], key_dim=cf["hidden_dim"] 22 | )(x, x) 23 | x = L.Add()([x, skip_1]) 24 | 25 | skip_2 = x 26 | x = L.LayerNormalization()(x) 27 | x = mlp(x, cf) 28 | x = L.Add()([x, skip_2]) 29 | 30 | return x 31 | 32 | def conv_block(x, num_filters, kernel_size=3): 33 | x = L.Conv2D(num_filters, kernel_size=kernel_size, padding="same")(x) 34 | x = L.BatchNormalization()(x) 35 | x = L.ReLU()(x) 36 | return x 37 | 38 | def deconv_block(x, num_filters, strides=2): 39 | x = L.Conv2DTranspose(num_filters, kernel_size=2, padding="same", strides=strides)(x) 40 | return x 41 | 42 | def build_unetr_2d(cf): 43 | """ Inputs """ 44 | input_shape = (cf["num_patches"], cf["patch_size"]*cf["patch_size"]*cf["num_channels"]) 45 | inputs = L.Input(input_shape) ## (None, 256, 3072) 46 | 47 | """ Patch + Position Embeddings """ 48 | patch_embed = L.Dense(cf["hidden_dim"])(inputs) ## (None, 256, 768) 49 | 50 | positions = tf.range(start=0, limit=cf["num_patches"], delta=1) ## (256,) 51 | pos_embed = L.Embedding(input_dim=cf["num_patches"], output_dim=cf["hidden_dim"])(positions) ## (256, 768) 52 | x = patch_embed + pos_embed ## (None, 256, 768) 53 | 54 | """ Transformer Encoder """ 55 | skip_connection_index = [3, 6, 9, 12] 56 | skip_connections = [] 57 | 58 | for i in range(1, cf["num_layers"]+1, 1): 59 | x = transformer_encoder(x, cf) 60 | 61 | if i in skip_connection_index: 62 | skip_connections.append(x) 63 | 64 | """ CNN Decoder """ 65 | z3, z6, z9, z12 = skip_connections 66 | 67 | ## Reshaping 68 | z0 = L.Reshape((cf["image_size"], cf["image_size"], cf["num_channels"]))(inputs) 69 | 70 | shape = ( 71 | cf["image_size"]//cf["patch_size"], 72 | cf["image_size"]//cf["patch_size"], 73 | cf["hidden_dim"] 74 | ) 75 | z3 = L.Reshape(shape)(z3) 76 | z6 = L.Reshape(shape)(z6) 77 | z9 = L.Reshape(shape)(z9) 78 | z12 = L.Reshape(shape)(z12) 79 | 80 | ## Additional layers for managing different patch sizes 81 | total_upscale_factor = int(log2(cf["patch_size"])) 82 | upscale = total_upscale_factor - 4 83 | 84 | if upscale >= 2: ## Patch size 16 or greater 85 | z3 = deconv_block(z3, z3.shape[-1], strides=2**upscale) 86 | z6 = deconv_block(z6, z6.shape[-1], strides=2**upscale) 87 | z9 = deconv_block(z9, z9.shape[-1], strides=2**upscale) 88 | z12 = deconv_block(z12, z12.shape[-1], strides=2**upscale) 89 | # print(z3.shape, z6.shape, z9.shape, z12.shape) 90 | 91 | if upscale < 0: ## Patch size less than 16 92 | p = 2**abs(upscale) 93 | z3 = L.MaxPool2D((p, p))(z3) 94 | z6 = L.MaxPool2D((p, p))(z6) 95 | z9 = L.MaxPool2D((p, p))(z9) 96 | z12 = L.MaxPool2D((p, p))(z12) 97 | 98 | ## Decoder 1 99 | x = deconv_block(z12, 128) 100 | 101 | s = deconv_block(z9, 128) 102 | s = conv_block(s, 128) 103 | 104 | x = L.Concatenate()([x, s]) 105 | 106 | x = conv_block(x, 128) 107 | x = conv_block(x, 128) 108 | 109 | ## Decoder 2 110 | x = deconv_block(x, 64) 111 | 112 | s = deconv_block(z6, 64) 113 | s = conv_block(s, 64) 114 | s = deconv_block(s, 64) 115 | s = conv_block(s, 64) 116 | 117 | x = L.Concatenate()([x, s]) 118 | x = conv_block(x, 64) 119 | x = conv_block(x, 64) 120 | 121 | ## Decoder 3 122 | x = deconv_block(x, 32) 123 | 124 | s = deconv_block(z3, 32) 125 | s = conv_block(s, 32) 126 | s = deconv_block(s, 32) 127 | s = conv_block(s, 32) 128 | s = deconv_block(s, 32) 129 | s = conv_block(s, 32) 130 | 131 | x = L.Concatenate()([x, s]) 132 | x = conv_block(x, 32) 133 | x = conv_block(x, 32) 134 | 135 | ## Decoder 4 136 | x = deconv_block(x, 16) 137 | 138 | s = conv_block(z0, 16) 139 | s = conv_block(s, 16) 140 | 141 | x = L.Concatenate()([x, s]) 142 | x = conv_block(x, 16) 143 | x = conv_block(x, 16) 144 | 145 | """ Output """ 146 | outputs = L.Conv2D(1, kernel_size=1, padding="same", activation="sigmoid")(x) 147 | 148 | return Model(inputs, outputs, name="UNETR_2D") 149 | 150 | if __name__ == "__main__": 151 | config = {} 152 | config["image_size"] = 512 153 | config["num_layers"] = 12 154 | config["hidden_dim"] = 64 155 | config["mlp_dim"] = 128 156 | config["num_heads"] = 6 157 | config["dropout_rate"] = 0.1 158 | config["patch_size"] = 1 159 | config["num_patches"] = (config["image_size"]**2)//(config["patch_size"]**2) 160 | config["num_channels"] = 3 161 | 162 | model = build_unetr_2d(config) 163 | model.summary() 164 | --------------------------------------------------------------------------------