ABSTRACT:

The electroencephalogram is the gold standard for diagnosing epilepsy (EEG). A massive amount of EEG information can be found in long-term recordings from a patient with epilepsy. Because of this, spotting epileptic activity is a challenging task that requires a thorough analysis of the entire EEG data set by a trained specialist. This article describes an automated classification of EEG signals for the detection of seizures using wavelet transform and statistical pattern recognition. Wavelet transform-based feature extraction, scatter matrix-based feature space dimension reduction, and quadratic classifier-based classification make up the three main phases of decision-making. EEG data sets from three groups of subjects (a) healthy volunteers, (b) epileptics during a seizure-free period, and (c) epileptics during a seizure were analysed using this methodology. Overall, we achieved a classification accuracy of 98%. Results indicated that the suggested method could successfully categorise EEG signals and identify epileptic seizures. This may aid in the enhancement of epilepsy diagnosis.

Keywords: Diagnosis of epilepsy; Detection of convulsions; Scatter matrices; Dimension reduction; Quadratic classifiers