Identifying Autistic Children Using Multi-Scale Entropies of Right Prefrontal Oxy-Hemoglobin Signals

This study aimed to identify children with autism spectrum disorders (ASD) by extracting features from prefrontal oxy-hemoglobin signals during an inhibitory control task. For this purpose, 46 ASD children and 32 typically developing (TD) children were recruited. For each participant, we utilized a four-channel portable functional Near-Infrared Spectroscopy (fNIRS) to collect prefrontal oxy-hemoglobin signals during a Go/NoGo task, and suggested multi-scale entropies (MSEs) of those signals as features for identification. Findings showed that children with ASD exhibited a bigger averaged MSE value in the right prefrontal lobe than TD children. The classifier with SVM achieved a maximal classification accuracy of 93.33% and an AUC value of 0.889.