EEG Microstate Analysis for Diagnosis of Children with ASD

The analysis of electroencephalogram (EEG) based on machine learning methods has been an effective tool for the diagnosis of children with autism spectrum disorders (ASD). In this paper, we introduce the microstate analysis of EEG to explore whether microstate features can serve as effective biomarkers for ASD diagnosis. After preprocessing raw EEG data, we generate template data by aggregating the EEG data of all the subjects. Afterwards, we segment the template data into five microstate prototypes, and use them to backfit all the preprocessed EEG data to generate a microstate stream. Later, we compute microstate metrics and use various classifiers for the classification. The experimental results show that the features extracted by EEG microstate analysis can distinguish between ASD children and typically developing children more effectively.