Abstract
MRIscansacquiredduringclinicalfollow-up,orlongitudinal MRI,arecentraltomonitoringmultiplesclerosisforapatient,wherenew or enlarging lesions indicate disease activity and guide treatment planning. However, manual or automated identification of these changes is complicatedbytherarity,variability,andpoordelineationofthelesions. Furthermore, while generative AI has made strides in cross-sectional image synthesis for data augmentation, longitudinal synthesis remains underexplored. In this work, we formulate MS follow-up synthesis as a controlledconditionalimage-to-imagetranslationtask,whereabaseline image and spatial change mask are used to synthesise a follow-up scan. We propose Regulate, Modulate, and Differentiate (RMD), a backboneagnosticconditioningframeworkthatmitigatesbaselinereplicationusing randomnoisereplacement,modulatesgenerationusingdualFiLMmodules that capture global progression descriptors and spatial changes at differentscales,andsupervisessharpchangestructurewithaDifferenceof-Gaussians loss. RMD was evaluated across Pix2Pix, Gaussian diffusion,andBrownianBridgeDiffusionusinginternalheld-outandexternal MSSEG-2cohorts.Indownstreamnewlesiondetection,RMDimproved lesion-levelF andrecallforPix2PixandDiffusionacrossbothdatasets, 1 andforBBDMinternally.ReaderassessmentpreferredRMDovernaive conditioning (47% vs. 43%) and showed significant gains in image quality and anatomical plausibility. These results position RMD as a step toward robust progression-aware synthesis for longitudinal MS MRI.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/SASHIMI_026.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=eGMqpowX5F
BibTex
@InProceedings{MatPra_Regulate_MICCAISAT2026,
author = { Mathur, Prateek AND Banahan, Paul AND Burns, Jane AND Kelly, Brendan S. AND Killeen, Ronan AND MacMahon, Peter AND Lawlor, Aonghus},
title = { { Regulate, Modulate, Differentiate: Towards Modelling Subtle Changes in Longitudinal Multiple Sclerosis MRI } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17258},
month = {pending},
page = {pending}
}