This repository contains code related to the paper Semi-Supervised Variational Autoencoder for Survival Prediction
We have implemented a semi-supervised variational autoencoder for modeling of 3D brain images and classification.
The code is designed for the BraTS dataset and can be used easily after you obtain the BraTS training data. To train a model simply run the data preparation script and then the main script.
python create_dataset.py --data_dir path/to/bratsdata --output_dir data/brats_19/
python main.py --data_dir_train data/brats_19/Train --data_dir_val data/brats_19/Validation
The code was tested with the versions provided in requirements.txt. To make sure you have them run
pip install -r requirements.txt