Integrating genomic and resting State fMRI for efficient autism spectrum disorder classification

Abstract

Autism spectrum disorder (ASD) is a neurodevelopmental disorder with a complex clinical syndrome and difficult diagnosis. The fusion of multimodal data improves the accuracy of ASD diagnosis, benefiting from the complementary information contained in the multimodal data. The aim of this study was to diagnose ASD using predictors based on integrated resting state MRI (rs-fMRI) and genetic data. Our study used fMRI and gene expression data from 71 participants in the National Database for Autism Research (NDAR). T-test and SVM-RFE were used for feature reduction and optimal feature selection. To make the best use of the features extracted from each data source, we used the EasyMKL to establish a classification model for the prediction of ASD rather than simply concatenating feature matrices. The experimental results validate the effectiveness of our method. In addition, the imaging and genetic features we extracted have been shown to be significantly associated with ASD supported by previous studies.