Recognition Method Of Attention Deficit/Hyperactivity Disorder Based On EEG Signal

The neurocognitive attention functions involve the cooperation of multiple brain regions,and the defects in the cooperation will lead to attention-deficit/hyperactivity disorder（ADHD）,which is one of the most common neuropsychiatric disorders for children.The current ADHD diagnosis is mainly based on subjective evaluation that is easily biased by the experience of the clinicians and lacks the support of objective indicators.Due to the complex connection between attention function and other cognitive functions,this study used EEG technology,designed attention network test experiments,collected a large number of event-related potential data,and proposed an event-related analysis based on time-frequency decomposition analysis and brain functional connectivity analysis.This method can realize the effective correlation analysis between attention function activity and EEG signal.The experimental results showed that each attention component has corresponding EEG activity.The enhancedθ band activity reflects the process of executive conflict,and the α inhibitory effect can distinguish between resting state and attentional states.And the brain functional connectivity results showed that the connectivity of alpha waves,delta waves and theta waves at long distances（frontal lobe to parietal or occipital lobe）are likely to be neuro markers for different cognitive states.In conclusion,the EEG signal analysis method proposed in this paper can effectively observe the relationship between EEG rhythm activity and attentional cognitive function,and provide a reliable processing flow for subsequent research.In this study,we also proposed a new CNN-LSTM model,which concatenates a convolutional neural network and a long short-term memory network,to identify a public EEG dataset containing 144 children.The results showed that our CNN-LSTM model can achieve up to 98.23% accuracy in the 5-fold cross-validation method,significantly outperforming other currently commonly used convolutional neural network models.This study also observes the features automatically extracted by the proposed model using convolutional visualization and saliency map methods,which can intuitively explain how the model discriminates different groups.The features extracted by this model were mainly located in the frontal and central brain regions,and there were significant differences in the time-segment mapping between the three different groups.The P300 and contingent negative variation（CNV）in the frontal lobe had the largest decrease in the healthy control（HC）group,and the ADD group had the smallest decrease.In the central area,only the HC group had a significant negative oscillation of CNV waves.The results of this study suggest that the CNNLSTM model can effectively identify children with ADHD and its subtypes.The visual features automatically extracted by this model could better explain the differences in the ERP response among different groups,which is more convincing than previous studies.And it could be used as more reliable neural biomarkers to help with more accurate diagnosis in the clinics.