Ripplet II transform and higher order cumulants from R-fMRI data for diagnosis of autism

摘要

Diagnose of autism spectral disorder (ASD) as a mental disorder by machine learning algorithms has attracted many attentions. Finding biomarkers from the rest state functional magnetic resonance imaging (R-fMRI) data is one of the common methods used for classifying ASD and normal healthy person (HP). This paper presents Eickhoff-Zilles (EZ) atlas to evaluate time courses for 20 ASDs and 16 HPs in 116 regions of interest (ROIs). To extract the effective features for classification, Ripplet II transform and higher order cumulants are proposed. Then, two sample t-test is employed to select the discriminative features for classification. After normalizing the selected feature vector, the data are classified by support vector machine (SVM). The results show that the proposed method achieves 91.67% accuracy which outperforms previous works.