Functional brain segmentation using inter-subject correlation in fMRI

摘要: The human brain continuously processes massive amounts of rich sensory in formation. To better understand such highly complex processes, modern neuroimaging studies are increasingly utilizing experimental setups that mimic daily-life situations. We propose a new exploratory data-analysis approach, functional segmentation intersubject correlation analysis (FuSeISC), to facilitate the magnetic resonance (fMRI) data sets collected these experiments. method provides type areas, not only characterizing areas display similar processing across subjects but also which is variable. tested FuSeISC using fMRI during traditional block design stimuli (37 subjects) as well naturalistic auditory narratives (19 subjects). identified spatially local and/or bilaterally symmetric clusters several cortical many known be types used prominent for spatial exploration large obtained stimuli, has other potential applications generation atlases including both lower- and higher-order areas. Finally, part FuSeISC, we criterion-based sparsification shared nearest-neighbor graph detecting noisy data. In our tests with synthetic data, this technique was superior well-known clustering methods, Ward's method, affinity propagation K-means++.