Research On The Neural Mechanism Of The Attention Deficit Hyperactivity Disorder Based On EEG And Brain MRI Analyses

Attention deficit hyperactivity disorder(ADHD) is a prevalently behavioral disorder syndrome found mainly in school-age population. It has critical effects on the patients' study and living. Nowadays, most doctors use inventories and questionnaires to diagnosis this disease with less objectivity index. This study aims to explore the neural mechanisms and clinical diagnosis of ADHD by electroencephalogram(EEG) and brain magnetic resonance imaging(MRI) analysis, combine with signal processing, bispectral analysis, image processing and deep learning methods.For EEG signal study of the first part, 16 children from ADHD group and Control group were recruited to perform improved visual-continuous performance test. EEG were recorded mainly under Go and NoGo conditions. Time domain, frequency domain and improved bispectrum effective energy analysis methods were proposed to figure out the quantitative EEG differences between ADHD and normal children. Results showed that:(1) Under the condition of Go, ADHD group had a significant lower P2-N2 peak-peak value than Control group;(2) Compared with Control group, ADHD group had significant lower spectral amplitude around 11 Hz under the condition of NoGo;(3) Whether under Go or NoGo condition, ADHD group had significant larger effective energy of bispectrum concentrated in ? wave frequency band than that of Control group;(4) Under both two conditions, ADHD group had significant high frequency position of bispectral peak than Control group.For brain MRI study of the second part, preprocessing with radial basis function(RBF) neural network, probabilistic neural network(PNN) and convolutional neural network(CNN) methods were proposed for ADHD clinical MR images classification. Results showed that:(1) The classification accuracy rate of RBF neural network reached to 72.41%;(2) The classification accuracy rate of PNN reached to 89.66%;(3) The classification accuracy rate of CNN reached to 86.21%. All neural networks achieved the automatic diagnosis of ADHD, and accuracy rate were satisfying. These results provide objective index for ADHD diagnosis. This can reduce the burden on clinician's diagnoses for a large number of clinical MR images can automatically recognize.Therefore, this article has comprehensively analysed the clinical signals of ADHD from one-dimensional EEG, two-dimensional images, time domain, frequency domain and spatial domain, and has discussed the neural mechanisms which cause to attention deficit. These research results provide fast, accurate, quantitative clinical diagnostic methods for ADHD.