A Master Class on Reproducibility: A Student Hackathon on Advanced MRI Reconstruction Methods

2026-01-26

AI Summary

This study addresses the reproducibility challenges in MRI reconstruction research by systematically reimplementing three representative methods—MoDL (Model-based Deep Learning), HUMUS-Net (a CNN–Transformer hybrid architecture), and a training-free physics-regularized approach for dynamic MRI—through an educational hackathon framework. By establishing a unified reproducibility pipeline, the work identifies critical implementation barriers and proposes a general reproducibility protocol alongside standardized coding practices. The project not only successfully replicates the core results of the original studies but also yields a transferable best-practice guide for reproducible research in medical imaging AI, offering methodological support to the broader community.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

We report the design, protocol, and outcomes of a student reproducibility hackathon focused on replicating the results of three influential MRI reconstruction papers: (a) MoDL, an unrolled model-based network with learned denoising; (b) HUMUS-Net, a hybrid unrolled multiscale CNN+Transformer architecture; and (c) an untrained, physics-regularized dynamic MRI method that uses a quantitative MR model for early stopping. We describe the setup of the hackathon and present reproduction outcomes alongside additional experiments, and we detail fundamental practices for building reproducible codebases.

Problem

• reproducibility
• MRI reconstruction
• scientific replication
• computational imaging
• open science

Innovation

• reproducibility
• MRI reconstruction
• hackathon
• unrolled networks
• physics-informed learning