Multimodal Neuroimaging Patterns Associated with Social Responsiveness Impairment in Autism: A Replication Study

Abstract

Multimodal fusion is an effective approach to discover covarying patterns of multiple imaging types impaired in brain diseases, such as Autism spectrum disorders (ASD). Compelling evidence has confirmed that social impairment is a primary and common characteristic of ASD. Here, based on a supervised learning strategy and independent component analysis, we used Social Responsiveness Scale (SRS) of participants as reference, to guide the 3-way MRI data fusion, aiming to identify the multimodal brain regions associated with social impairment in ASD. Results show high consistence on the brain patterns identified in two independent cohorts, suggesting that lower fALFF and ReHo values in the default mode network (DMN) and occipital cortex, as well as reduced cortical thickness in inferior frontal and superior temporal cortex were associated with social impairment in autism. Moreover, these multimodal brain regions were also detected in a repeatable manner, validating the robustness of the fusion with reference approach, demonstrating the ability of multimodal fusion to identify potential imaging markers of interest for mental disorders.