TimeLesSeg: Unified Contrast-Agnostic Cross-Sectional and Longitudinal MS Lesion Segmentation via a Stochastic Generative Model

2026-05-08

AI Summary

This study addresses the challenges of automatic multiple sclerosis lesion segmentation, including variability in image contrast, domain shift across scanners, and inconsistent data structures between cross-sectional and longitudinal scans. The authors propose the first unified framework that leverages a single convolutional neural network to achieve contrast-agnostic lesion segmentation, accommodating both time-series and non-temporal inputs. Their approach incorporates lesion masks to model pathological priors and employs morphological operations to stochastically simulate lesion evolution. Domain randomization is further enhanced through Gaussian mixture modeling to improve robustness to contrast variations. Experiments demonstrate that the method outperforms existing contrast-invariant approaches on three public and two internal datasets under single-modality input, surpasses SAMSEG in longitudinal segmentation accuracy, and exhibits superior capability in capturing dynamic lesion burden changes compared to both SAMSEG and LST-AI.

Abstract

Multiple sclerosis (MS) expresses substantial clinical and radiological heterogeneity, which poses significant challenges for automatic lesion segmentation. The current deep learning-based SOTA is highly susceptible to changes in both distribution, e.g., changes in scanner; as well as the structure of inputs, evident in the current divide between cross-sectional and longitudinal approaches. We introduce TimeLesSeg, a unified contrast-agnostic framework designed to segment MS lesions regardless of the presence of a temporal dimension in its inputs, with a single convolutional neural network. Our approach models pathological priors through lesion masks, which are processed together with the current scan. Cross-sectional processing is enabled by exposing the model to training cases where no prior information is available, which are modeled with an empty mask, allowing it to operate seamlessly in both scenarios. To overcome the scarcity and inconsistency of longitudinal datasets, we propose a novel generative pipeline in which patterns of lesion evolution are simulated by stochastically deforming each individual lesion with morphological operations, producing realistic prior timepoints. In parallel, we achieve contrast agnosticism through Gaussian mixture model-based domain randomization, enabling the network to experience a wide spectrum of intensity profiles. Results on three publicly available and two in-house datasets show that TimeLesSeg outperforms the contrast-agnostic state of the art on single-modality inputs across overlap- and distance-based metrics. In longitudinal processing, our method outperforms SAMSEG, and captures lesion load dynamics more accurately than both the former and LST-AI. All source code related to the development of TimeLesSeg is available at https://github.com/NeuroADaS-Lab/TimeLesSeg.

Problem

• multiple sclerosis
• lesion segmentation
• cross-sectional
• longitudinal
• contrast-agnostic

Innovation

• contrast-agnostic
• longitudinal segmentation
• stochastic generative model
• domain randomization
• unified framework