MARIO : Monitoring AMD progression in OCT (MIC Group 6)

Copyright German Cancer Research Center (DKFZ) and contributors. Please make sure that your usage of this code is in compliance with its license.

This repository contains the implementation of the 2nd place method from the 2024 MICCAI MARIO Challenge.

Getting Started

Follow these steps to replicate our results:

1. Setup and Install Dependencies

First, clone the repository and set up the necessary environment:

git clone [repository URL]
cd mario
conda env create -f environment.yml

2. Preprocess the Dataset

Our model requires the dataset to be in .npy format for efficiency. Please convert the files from .png to .npy before proceeding.

3. Setup Environment Variables

Configure the following environment variables according to your setup:

• SAVE_DIR_RESULTS: Directory where models, evaluations, and other outputs will be saved.
• DATA_DIR: Path to the MARIO dataset.

4. Select Correct Splits File

In main.py, line 175, insert the correct splits file:

• Use splits.json for a split on the Training set.
• Use splits_train_all.json for a split on the Training + Validation set.

5. Train the Model

Siamese Network

To train the Siamese Network, execute:

python main.py

AutoEncoder

For the AutoEncoder, follow these steps:

1. Run the following script to catalog all available images. Ensure the output file is saved in the directory where you intend to store your model results:

   python utils.find_images.py

2. Start the training process by running:

   python main_ae.py

Paper

If you use this code in you research, please cite the following paper (TODO)