DLPipelines

A collection of deep learning pipeline templates focused on medical imaging, powered by PyTorch Lightning, Weights & Biases, and MONAI.

Overview

DLPipelines provides ready-to-use templates for building efficient and reproducible deep learning workflows in medical imaging. The repository combines the power of PyTorch Lightning for organized training loops, Weights & Biases for comprehensive experiment tracking, and MONAI for medical-imaging-specific components.

Features

• Standardized project structure for medical imaging deep learning projects
• Integration with PyTorch Lightning for clean and organized training code
• Automated experiment tracking and visualization with Weights & Biases
• MONAI-based data transformations and neural network architectures
• Configurable training pipelines with reproducible experiments
• Medical imaging specific data loading and preprocessing templates

Installation

We recommend using uv for fast and reliable Python package management.

# Install uv if you haven't already
pip install uv

# Clone the repository
git clone https://github.com/f10409/DLPipelines.git
cd DLPipelines

# Create a virtual environment and install dependencies using uv
uv sync

Note: Using uv significantly speeds up dependency resolution and package installation compared to traditional pip.

Support

For questions and support, please open an issue in the GitHub repository.