SRSC+: A Bilevel Sensitivity-Corrected Reconstruction Framework with Deep Priors for Parallel MRI

This repository provides the official implementation of SRSC+.

🔧 Requirements

The environment configuration is provided in environment.yml. You can create the environment using:

conda env create -f environment.yml

🚀 Getting Started

To run the reconstruction on test data:

python main.py

The code includes a set of Phantom data for quick testing.

📁 Project Structure

• main.py – Entry point for running the reconstruction.
• gen_mask.m – MATLAB script for generating sampling masks.
• environment.yml – Conda environment specification.
• data/ – Contains example Phantom test data.
• mask/ – Contains example undersampled mode.
• algorithm/SRSC.py – Our propose ADDL algorithm.
• utils/ – Supporting modules.

📦 Pretrained Models & Training Data & Undersampled Mode

Pretrained ADDL networks and additional test data are available at:

👉 Google Drive - SRSC Resources

• Please place the file net.pth in folder utils/checkpoint.
• Generate your own sampling patterns via gen_mask.m.
• For questions or suggestions, please contact me at: 2019151068@email.szu.edu.cn