Auxiliary Diagnosis System For Autistic Children Based On Eeg Indicators

Autism spectrum disorder(ASD)is usually a neurological disorder that begins in early development and seriously affects cognition,sensory movement,social emotional and social activities.The cause of autism is unknown.At present,the diagnosis of autism mainly relies on scale score and clinical behavioral symptom assessment,which is subjective to some extent.Therefore,it is particularly important to find objective biological indicators to assist diagnosis.As far as the current situation is concerned,the age of ASD children is getting smaller and smaller,early diagnosis and intervention are very meaningful.Based on the resting state electroencephalography(EEG)analysis technique,EEG of preschool children in different ages is included in this thesis.According to the two-year-old stage,power spectrum,brain complexity and function connection obtained from two different angles in the brain area and across the brain regions were studied to analyze the differences between children groups.The following mainly describes the research results of this paper:(1)Firstly,according to the five divided brain regions,the power spectrum and wavelet entropy in the brain region were calculated for 3-4 years old(36 ASD children and 40 typical development(TD)children)and 5-6 years old(41 ASD children and 50 TD children)children without significant difference in age and gender to analyze the difference between the two groups by using t-test and Bonferroni correction.The comparison found that the difference of the two algorithms were significant when analyzing the differences in the brain region between the two groups of children.(2)Secondly,according to the five divided brain regions and the ten brain regions divided into the left and right hemispheres,the functional connectivity matrix across the brain regions between the two groups of children in the above samples was calculated.By Bonferroni correction and SVM(support vector machine)classification and comparison,it is found that the wavelet consistency algorithm has good classification performance according to 10 brain regions,and the single-band classification accuracy can reach97.92%.Moreover,it was found that the wavelet coherence values between the frontallobe and other brain regions were significantly different.Through the significant connection between the channels,we found that the connection between the wavelet coherence value and the phase delay index in the brain region is not significant.(3)Based on the above research,this paper designs a set of EEG assessment system based on power spectrum?wavelet entropy and wavelet coherence for autistic children.The indicators of TD children with different ages are used as reference standards.The original EEG of the participant children were extracted feature indexes for comparison after pre-processing?signal extraction and feature calculation.Finally,the test results are obtained.To sum up,this study was aimed to look for the potential objective physiological indicators from the three perspectives of power spectrum?brain complexity in the brain region and function connection across brain regions,and designed an auxiliary diagnosis system based on these indicators for autistic children,which provided an important reference to clinical diagnosis and early intervention for autistic children.