Sparsity and Low-Rank Based Methods for Capturing Common and Discriminative Features in Multi-Subject fMRI Data

Abstract

Data-driven analysis for functional magnetic resonance imaging (fMRI) data has played an important role for uncovering salient brain functional networks that are shared across multiple subjects. On the other hand, recent fMRI studies indicate that there is significant and consistent heterogeneity present across different subject groups and individuals. While independent component analysis (ICA) has been a major tool to perform data-driven analysis of fMRI data, dictionary learning (DL) approaches are increasingly receiving attention due to their modeling capability and flexibility. In this work, a series of DL algorithms are formulated and their efficacy in estimating good spatial maps is explored. First, multi-subject fMRI data analysis methods based on sparse DL are proposed. The component spatial maps are identified by exploiting the sparsity of the maps while the clusters of the subjects are pursued simultaneously by postulating that the fMRI volumes admit subspace clustering structures. Furthermore, a framework is developed to capitalize on the available class labels and capture not only the commonly shared components across the population, but also the unique components that contribute to discrimination. A systematic comparison with conventional ICA is performed based on real fMRI data consisting of healthy controls and patients with schizophrenia.