BraTS-Unet2D

Project description: Built for the brain tumour segmentation task, the data can be from any year of the BraTS challenge (the project used data from 2019) and model == Unet-2D

You can find BraTS data here:

Website:

1. BraTS 2021-Kaggle (https://www.kaggle.com/datasets/dschettler8845/brats-2021-task1)
2. BraTS 2019 (https://www.med.upenn.edu/cbica/brats2019/data.html)

1. data processing

Locate the data folder and run the data_split.py

Divide the dataset into trainset and testset and make a copy of it and save the divisions with the filename of the json file.

Run data_process.py

Images saved in npz format, Ground Truth saved as png

2. Get JSON file

Locate the home folder and run get_json.py

Get the json data files for train and test, data reading will be done based on these files.

3. train && test

Before you start running train, perhaps you need to confirm the args argument in train.py. AND then in terminal:

python train.py

OR in pycharm

RUN train

To avoid trouble, the test function has also been written into the train.py file, which you can easily find and make changes to.

Simple? Aha!

Reference

Papers:

1. U-Net: Convolutional Networks for Biomedical Image Segmentation (https://arxiv.org/abs/1505.04597)
2. The RSNA-ASNR-MICCAI BraTS 2021 Benchmark on Brain Tumor Segmentation and Radiogenomic Classification (https://arxiv.org/abs/2107.02314)