Abstract
Diffusion MRI is widely used to explore brain microstructure, and numerous machine learning methods have been proposed to extract predictive information. Yet progress remains difficult to assess: studies rely on single datasets, heterogeneous pipelines, limited evaluation, and often lack accessible code, preventing meaningful comparison and reproducibility. We introduce DiffBench, a modular and transparent benchmarking framework for machine and deep learning in dMRI microstructure. The benchmark unifies multiple representative datasets within a standardized end-to-end pipeline and enables systematic comparison across tissue types, feature representations and models. It reveals differences between white and gray matter analyses, quantify the impact of diffusion microstructure feature choices, and provide a set of reference results for machine and deep learning prediction. This shows that standardized benchmarking is essential to move the field beyond proxy tasks and toward robust methods with real clinical relevance.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/cdmri_008.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=VPFBBCkZLO
BibTex
@InProceedings{JimGab_DiffBench_MICCAISAT2026,
author = { Jimenez, Gabriela Gomez AND Aggarwal, Himanshu AND Wassermann, Demian AND Dorent, Reuben},
title = { { DiffBench: A Benchmark Toward Reproducible Evaluation in Diffusion MRI Prediction } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17278},
month = {pending},
page = {pending}
}