FM-fMRI: Event Conditioned Flow Matching for Rest-to-Task fMRI Time-Series Synthesis

2026-05-25

AI Summary

Task-based fMRI data are costly to acquire and difficult to collect at scale, limiting their application in neuroscience and clinical settings. This work proposes an event-conditioned continuous-time flow matching model that, for the first time, applies flow matching to fMRI synthesis by generating task-evoked brain time series from resting-state fMRI guided by task event information. The approach enables flexible generation under heterogeneous task schedules and comprehensively evaluates synthesis quality through multiple metrics, including spectral characteristics, functional connectivity consistency, and distributional alignment. Evaluated on the Human Connectome Project (HCP) and the BioPoint autism dataset, the proposed method significantly outperforms baseline generative models—including diffusion models, GANs, and VAEs—in both generation fidelity and downstream autism classification performance.

Abstract

Task-based fMRI provides a direct readout of task-evoked neural dynamics, but it is expensive and difficult to acquire at scale, motivating rest-to-task synthesis from widely available resting-state fMRI (rsfMRI). We propose FM-fMRI, an event-conditioned flow-matching model that learns a continuous-time conditional vector field to generate task ROI time series from a subject's rsfMRI and the task event information. The formulation enables fast ODE-based sampling and flexible conditioning over heterogeneous event schedules. Rather than optimizing for pointwise reconstruction, we evaluated generated signals using complementary criteria that probe temporal and spectral structure, subject and group-level connectome consistency, and distributional alignment. On the public Human Connectome Project and internal BioPoint autism cohort, FM-fMRI achieves the strongest spectral and connectivity agreement and improved distribution-level matching over conditional diffusion, generative adversarial networks (GANs), and variational autoencoders (VAEs) baselines. Furthermore, we augment the BioPoint cohort by synthesizing task-fMRI ROI time series with our method, improving downstream autism classification and demonstrating practical utility in data-limited clinical settings. The code will be available on GitHub.

Problem

• rest-to-task fMRI synthesis
• task-based fMRI
• resting-state fMRI
• time-series generation
• neuroimaging data augmentation

Innovation

• flow matching
• rest-to-task fMRI synthesis
• event-conditioned generation
• connectome consistency
• ODE-based sampling