https://github.com/sdsubhajitdas/brain-tumor-segmentation
Brain Tumor Segmentation done using U-Net Architecture.
https://github.com/sdsubhajitdas/brain-tumor-segmentation
brain-tumor-segmentation deep-learning image-segmentation pytorch tumor-segmentation unet unet-image-segmentation unet-pytorch
Last synced: about 1 year ago
JSON representation
Brain Tumor Segmentation done using U-Net Architecture.
- Host: GitHub
- URL: https://github.com/sdsubhajitdas/brain-tumor-segmentation
- Owner: sdsubhajitdas
- License: mit
- Created: 2019-02-01T18:19:24.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2023-07-21T04:01:11.000Z (almost 3 years ago)
- Last Synced: 2025-03-31T06:05:10.516Z (about 1 year ago)
- Topics: brain-tumor-segmentation, deep-learning, image-segmentation, pytorch, tumor-segmentation, unet, unet-image-segmentation, unet-pytorch
- Language: Jupyter Notebook
- Homepage:
- Size: 37.1 MB
- Stars: 278
- Watchers: 7
- Forks: 63
- Open Issues: 6
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Brain Tumor Segmentation
[](https://www.python.org/)
[](https://github.com/sdsubhajitdas)
[](https://github.com/Jeetu95/Brain-Tumor-Segmentation)
[](https://forthebadge.com)
[](https://opensource.org/licenses/MIT)
[](https://github.com/Jeetu95/Brain-Tumor-Segmentation/graphs/commit-activity)
[](https://github.com/Jeetu95/Brain-Tumor-Segmentation/issues)
This project uses [U-Net Architecture](https://arxiv.org/abs/1505.04597) to create segmentation masks for brain tumor images.
## Overview
- [Dataset Used](#Dataset-Used)
- [Data Augmentation](#Data-Augmentation)
- [Model Architecture](#Model-Architecture)
- [Training Process](#Training-Process)
### Dataset Used
Dataset used in this project was provided by Jun Cheng.
This dataset contains 3064 T1-weighted contrast-enhanced images with three kinds of brain tumor. For a detailed information about the dataset please refer to this [site](https://figshare.com/articles/brain_tumor_dataset/1512427).
Version 5 of this dataset is used in this project. Each image is of dimension ```512 x 512 x 1``` , these are black and white images thus having a single channel.
Some Data Samples
Original Image | Mask Image
:-------------------------:|:-------------------------:
 | 
### Data Augmentation
The basic forms of data augmentation are used here to diversify the training data.
All the augmentation methods are used from [Pytorch's](https://pytorch.org) [Torchvision](https://pytorch.org/docs/stable/torchvision/index.html) module.
- [Horizontally Flip](https://pytorch.org/docs/stable/torchvision/transforms.html#torchvision.transforms.functional.hflip)
- [Vertically Flip](https://pytorch.org/docs/stable/torchvision/transforms.html#torchvision.transforms.functional.vflip)
- [Rotation](https://pytorch.org/docs/stable/torchvision/transforms.html#torchvision.transforms.functional.rotate) Between 75°-15°
Code Responsible for augmentation
Each augmentation method has a probability of 0.5 and the order of application is also random. For Rotation Augmentation the degree of rotation is chosen randomly between 75°-15°.
### Model Architecture
The model architecture is depicted in this picture.
### Training Process
The model was trained on a [Nvidia GTX 1050Ti](https://www.geforce.com/hardware/desktop-gpus/geforce-gtx-1050-ti/specifications) 4GB GPU. Total time taken for model training was 6 hours and 45 minutes. We started with an initial learning rate of 1e-3 and reduced it by 85% on plateauing, final learning rate at the end of 100 epochs was 2.7249e-4.
Some graphs indicating Learning Rate & Loss Value over 100 epochs are given below.
Learning Rate Graph in Tensorboard.
Loss Graph Plotted in [Matplotlib](https://matplotlib.org)
Loss Graph Plotted in Tensorboard
To see the complete output produced during the training process check [this](logs/05-47-51_PM_on_May_20,_2019/training_output_log.txt)
## Installation
This project uses python3.
Clone the project.
```bash
git clone https://github.com/Jeetu95/Brain-Tumor-Segmentation.git
```
Install Pytorch from this [link](https://pytorch.org/get-started/locally/)
Use pip to install all the dependencies
```bash
pip install -r requirements.txt
```
To open the notebook
```bash
jupyter lab
```
To see logs in Tensorboard
```bash
tensorboard --logdir logs --samples_per_plugin images=100
```
To setup the project dataset
```bash
python setup_scripts/download_dataset.py
python setup_scripts/unzip_dataset.py
python setup_scripts/extract_images.py
```
## Usage
```bash
python api.py --file --ofp
python api.py --folder --odp
python api.py -h
```
```
Available api.py Flags
--file : A single image file name.
--ofp : An optional folder location only where the output image will be stored. Used only with --file tag.
--folder: Path of a folder containing many image files.
--odp : An optional folder location only where the output images will be stored. Used only with --folder tag.
-h, --help : Shows the help contents.
```
Some results generated by API are [here](images/API)
## Results
The mean [Dice Score](https://en.wikipedia.org/wiki/S%C3%B8rensen%E2%80%93Dice_coefficient) our model gained was 0.74461 in testing dataset of 600 images.
From this we can conclude that in our testing dataset our constructed mask has a similarity of about 74% with the original mask.
Some samples from our training dataset output are below. The top best results are [here](images).To see all the results click on this [Google Drive link](https://drive.google.com/drive/folders/1vwwUipaH9Yb0NLelv3lW-04E6WnVJ3nh?usp=sharing)
. | .
:-------------------------:|:-------------------------:
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
## Contributing
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
## License
[MIT License](LICENSE)