基于脑电图相位功能连接的孤独症谱系障碍识别OA

Objective To improve the diagnostic accuracy and specificity of autism spectrum disorder(ASD)based on the analysis of electroencephalogram(EEG)phase functional connectivity.Methods A total of 25 children with ASD(ASD group)and 25 typically developing children(control group)were recruited as participants,and their EEG data were collected.After preprocessing the EEG data,the δ(1 to 4 Hz),θ(4 to 8 Hz),α(8 to 13 Hz)and β(13 to 30 Hz)bands as well as their ASD-related characteristics were extracted.Two network properties(NP),and six-dimensional features of the spatial pattern of network topology(SPN)of the EEG connectivity map,and two-dimensional features including power spectral density(PSD)and sample entropy(SampEn)were calculated.Six classifiers based on machine learning algorithms were trained for classification experiments.The average performance metrics(accuracy,sensitivity,and specificity)of NP,SPN,PSD,and SampEn across the 4 EEG bands were computed to verify the classification efficacy of the extracted features for ASD.Results SPN achieved the optimal classification performance in the β band,and the stability of the classification results was significantly improved after merging all features.Analysis of low-proportion training data(5 participants selected from both the ASD group and the control group)showed that the model trained with merged features yielded the highest accuracy(83.33%),sensitivity(93.33%),and specificity(76.67%).Visualization analysis indicated that the functional connectivity characteristics of the ASD group and the control group differed significantly in the θ,β,and δ bands,which was conducive to the diagnostic identification of ASD.Conclusion The extraction based on EEG phase connectivity has a certain effect on ASD classification,and can serve as a potential electrophysiological indicator for ASD diagnostic identification.