AI Insight
This study introduces a new mathematical framework called spatiotemporal variational tensor decomposition (ST-VTD) for analyzing brain imaging data from multiple subjects. The method combines tensor factorization with machine learning techniques, specifically using LSTM neural networks to model temporal patterns and specialized regularization for spatial patterns, allowing it to capture both shared patterns across subjects and individual variations. Testing on synthetic fMRI data showed the approach outperformed existing classical and probabilistic decomposition methods in recovering underlying brain activity patterns.
Why it matters
This advancement could improve researchers' ability to identify both common and individual-specific brain activity patterns across groups of people, with potential applications in understanding neurological conditions, cognitive processes, and individual differences in brain function. The method's ability to handle complex spatiotemporal variability may lead to more accurate interpretations of neuroimaging studies.
Understand the Science
arXiv:2607.22262v1 Announce Type: cross
Abstract: Modeling shared and subject-specific structure in multisubject spatiotemporal data remains challenging, particularly in neuroimaging, where both spatial and temporal patterns exhibit rich variability across subjects. Existing matrix and tensor decompositions provide interpretable factorizations, but rely on fixed multilinear structures or coupling schemes that may limit their flexibility in capturing complex variability. In this work, we introduce a spatiotemporal variational tensor decomposition (ST-VTD) framework that combines a tensor factorization generative model with structured priors to jointly represent spatial maps and temporal dynamics. Spatial factors are regularized to promote a low-rank structure inspired by the LL1 decomposition, while temporal factors are modeled using a learned Long short-term memory (LSTM)-based prior, enabling flexible and adaptive dynamics. Posterior inference is performed using an amortized variational formulation by unrolling iterations of an optimization algorithm, leading to an interpretable and parameter-efficient architecture. The proposed inference framework employs a warm-start strategy based on group independent component analysis, which we found to improve optimization performance. Experiments on a realistic synthetic functional MRI (fMRI) dataset demonstrate that the proposed approach significantly improves latent factor recovery compared with representative classical and probabilistic decomposition benchmarks.
Source: Variational Low-rank Tensor Decomposition for Multisubject Spatiotemporal Data Analysis