Research On EEG Feature Extraction And Classification Algorithm Of Children With ADHD

Attention deficit hyperactivity disorder(ADHD)is called ADHD in China.It is a common type of mental disorder in childhood.In recent years,the number of patients has increased,so it has received more and more attention from society.However,the current pathogenesis It is not clear that the clinical diagnosis mainly depends on the way of observation and behavioral scale,which is subjective,so this article will study a set of objective indicators to diagnose children with ADHD.Non-invasive measurement of EEG can accurately record the EEG signal of the test subject.In this study,the measured EEG signal is analyzed from the aspects of information entropy,wavelet coherence,and power spectral density,and then the characteristics of these features are extracted.Research on classification algorithms of eigenvalues,through research on ADHD and EEG signals of normal children,can play a certain auxiliary role in clinical diagnosis.First,the resting eye closed EEG signals of 102 children aged 4-9 years were collected in the study,including 52 children with ADHD and 50 normal children.After screening the signals,the EEG signals of 46 ADHD children and 45 normal children were obtained.Then further preprocessing to remove artifacts such as power frequency,electrooculogram,myoelectricity,sweating,etc.,to obtain relatively pure EEG signals,and then extract features from three directions: information entropy,wavelet coherence,and power spectral density.Power spectrum It mainly analyzes the power differences in different frequency bands.The information entropy is mainly extracted from approximate entropy,sample entropy,sorting entropy,and multi-scale entropy.Wavelet coherence mainly analyzes the strength of the connection between the electrodes in the brain region and the trans-brain region.For the eigenvalues extracted from the information entropy and power spectrum,the fuzzy C-means clustering algorithm based on the kernel function is used for classification,and the support vector machine(SVM)is used to obtain the classification accuracy of a single channel and compare the eigenvalues.Classification accuracy of the multi-feature classification and found that multi-feature fusion classification is more effective.Finally,an effective classification model is established by the SVM algorithm.The experimental results showed that the energy in the ADHD group increased significantly in the theta band(4-8Hz),while the energy in the bate band(14-30Hz)decreased significantly.The classification accuracy of the whole brain is 71.25% and 74.25%,respectively.In terms of approximate entropy,sample entropy,and multiscale entropy,the ADHD group is significantly lower than the normal control group except for the T8 electrode,and the accuracy of the whole brain classification is 56.89%.,57.62%,65.26%,the combined classification accuracy is 73.36%;the accuracy of classification using wavelet coherence as features is 78.53%.All the above features are classified as a valid feature subset,and the classification accuracy of the final classification model reaches 93.48%.The research mainly takes the perspective of clinical application as the starting point,extracts the features that can effectively identify the EEG signals of children with ADHD,fuses the features to establish an efficient classification model,which provides an objective and effective method for the clinical diagnosis of ADHD.