A mathematical analysis for the detection of epileptic seizure from EEG signals using wavelet transform with optimization algorithms and machine learning classifiers

Abstract

Epilepsy is a neurological disease caused by sudden abnormal of newlineelectrical discharge by the brain neurons. Around 1% of the world s population newlinesuffers from epilepsy disease. To save and protect the patient from this disease newlinerequires early detection. To predict the abnormalities during the epileptic newlineperiod, the brain signal should be detected and analyzed. The direct analysis of newlinebrain signal is very complex and time consuming and not accurate. Also not newlineprovide all the information about particular disease because many features are newlinehidden in time series. Due to technological developments different diagnostic newlinetools are available to record the brain signals such as Positron Emission newlineTomography (PET), Magnetic Resonance Imaging (MRI), Computer newlineTomography (CT), and Electroencephalogram (EEG). Out of these, EEG is newlinevery low cost and very comfort for patient. The raw brain signals have many newlineunrecognizable signal and hence not suitable for diagnosis of epileptic seizures. newlineNowadays using the developed mathematical based tools on signal newlineand image processing, the epileptic seizure is accurately detected at the early newlinestage. Hence eighty percentages of the people those who are affected by newlineepileptic are controlled and saved. The mathematical tool can extract all the newlineretrieval features of the particular disease from the brain signal. The newlineDepartment of Epileptology, Bonn University Germany EEG dataset is used in newlinethis study. The dataset consists of five groups (A to E). All the dataset from A newlineto E have 100 epochs. There are 4096 EEG samples that are recorded in each newlineepoch. The proposed work considers only two datasets such as normal set (A) newlineand seizure set (E). newline newline