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

TL;DR

This paper details a student-led hackathon aimed at reproducing key MRI reconstruction methods, highlighting best practices for reproducibility and presenting outcomes from replicating influential research papers.

Contribution

It introduces a structured hackathon approach for reproducibility in MRI research and provides insights into replicating complex models and establishing reproducible workflows.

Findings

Successful reproduction of three influential MRI papers

Identification of best practices for reproducible codebases

Enhanced understanding of advanced MRI reconstruction methods

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.