INR-QSM

This repo contains Matlab codes for generating input data and python codes for running INR-QSM.

Here is the paper for INR-QSM:
A subject-specific unsupervised deep learning method for quantitative susceptibility mapping using implicit neural representation
https://www.sciencedirect.com/science/article/abs/pii/S1361841524000987

Brief introduction

A subject-specific unsupervised deep learning approach for QSM (quantitative susceptibility mapping) reconstruction using implicit neural representation (INR).

Feature

• A new signal representation scheme for QSM recon which maps the susceptbility value as an implicit function of coordinates
• The non-local phase effect is considered in patch-based QSM deep learning method
• The proposed method is subject-specific and unsupervised, indicating that it is free of generalization problems and the requirement of a large dataset, which are two issues in supervised deep learning methods
• Several acceleration strategies are adopted to accelerate training process

Setup

• STISuite 3.0
• PyTorch 1.11.0
• Python 3.8

Usage

Data preparation

1. Generate test data data.mat containing phi, msk, WG based on files in data_prep folder
2. Adjust the config.py by inputting correct voxel size, B0_dir, patch size, and other parameters

Training and prediction

3. Run main.py for generating INR-QSM output

Contact

Feel free to contact zhangming430424@gmail.com or mingzhang.bme@sjtu.edu.cn for questions/discussions/suggestions.