Brain Tumor Semantic Segmentation

Contributors

• Fuat Arslan
• Melih Berk Yılmaz

Files Description

• data_loading: This folder contains the code for loading the data from the dataset. It was adapted from the original repository of the nnU-Net project.
• net/utils.py: Contains utility functions for take configuration parameters from the config file.
• net/wrappers.py: Wrapper functions for training and testing the models.
• net/loss.py: Implementation of loss functions. Dice loss and cross entropy loss are implemented.
• net/networks.py: Contains neural network architectures used for image segmentation.
• train.py: Script for training the segmentation models on datasets.
• test.py: Script for evaluating the performance of the models on test data.
• config.yaml: Configuration file for the project. Contains parameters for training and testing the models.
• utils/args.py: Contains functions for parsing command line arguments. It was adapted from the original repository of the nnU-Net project.
• sample_out_viz.py: Script for visualizing the output of the models on sample images.

Installation

To set up this project, follow these steps:

git clone https://github.com/fuat-arslan/MRI-Segmentation.git
cd MRI-Segmentation

Train

python train.py --model_config config.yaml 

Test

python test.py --model_config path/to/config.yaml 

Put the config.yaml to the folder of the model for testing. Arange the test data path in the config.yaml file according to your data path.

Sample Output Visualization