Prostate MRI Classification

This repository contains code for classifying prostate MRI images using deep learning techniques.

Setup

1. Clone the repository:

   git clone https://github.com/KevinMathewT/prostate-mri-classification.git
   cd prostate-mri-classification

2. Install dependencies using Poetry:

   poetry install

3. Activate the virtual environment:

   poetry shell

Running the Project

Training the Model

1. Create a configuration file or use an existing one from the config/ directory. Example configurations may include hyperparameters, dataset paths, and training settings.

2. Run the training script with the chosen configuration file:

   accelerate launch -m train config/resnet50.yaml

   Replace path/to/config.yaml with the actual path to your configuration file.

3. The script will handle model initialization, training, validation, and logging based on the provided configuration.

Validation

You can load and evaluate trained models using code snippets provided in notebooks/ for further analysis and visualization.

Repository Structure

• config/: Contains YAML configuration files for specifying training parameters.
• loader/: Code for loading and preprocessing MRI datasets.
• model/: Definitions of model architectures used for classification.
• notebooks/: Jupyter notebooks for exploration and evaluation of model performance.
• criterion.py: Implements custom loss functions for the training process.
• engine.py: Core training and evaluation logic, including epochs and batch handling.
• optimizer.py: Optimizer configurations for training.
• train.py: Main training script, which integrates the components and runs the training pipeline.
• utils.py: Utility functions for setup, initialization, and debugging.

Feel free to explore and modify the repository to suit your requirements!