Classification and comparative analysis of control and migraine subjects using EEG signals

Abstract

Migraine is an incapacitating neurovascular disorder that disables the brain by a severe headache and dysfunction of the autonomic nervous system. There is no perfect diagnosis of migraine till date. Migraine diagnosis if replaced by electroencephalogram (EEG) modality could help in the diagnosis of the disease. Recent advances in EEG signal processing have led to multi-resolution, processing, and methods of feature extraction. In this study, a nonlinear parametric method is used to acquire EEG features of and are used for the classification of control and migraine subjects. This EEG classification is carried out by classifiers based on supervised classification methods—backpropagation used in artificial neural network (ANN) and the results are compared with a bilinear supervised classifier support vector machine (SVM). The classification results confirm that the methodology has a potential to classify EEG and can be used to detect EEG of migraine subjects and could thus further result in improved diagnosis of migraine.