Application of a new approach based on wavelet analysis and arithmetic programming for automatic detection and detection of epileptic seizures in EEG signals using machine learning techniques

Epilepsy, a common neurological disorder, is generally diagnosed by electroencephalogram (EEG) signals. Visual inspection and interpretation of EEGs is a slow and time consuming process that is vulnerable to errors and subjective changes. As a result, several attempts have been made to develop methods for diagnosing and classifying automatic epilepsy. The present study proposes a new computer diagnostic (CAD) method based on discrete wavelet transform (DWT) and arithmetic coding to distinguish epileptic seizure signals from normal (no seizure) signals. The proposed CAD technique consists of three steps. The first step is to separate the EEG signals using pproximations and partial coefficients using DWT, while discarding the unusual coefficients according to the threshold criteria. Therefore, limiting the number of wavelet coefficients is significant. The second step converts significant wavelet coefficients using arithmetic coding to calculate the compression ratio to bit streams. Finally, there is a set of standard compression features, according to which machine learning classifiers distinguish seizure activity from non-seizure signals. We used an extensive benchmark database from the University of Bonn to compare and validate this method with the results of previous methods. The proposed method using linear and nonlinear machine learning classifiers to classify epileptic seizure activity from EEG data achieved a complete classification function (۱۰۰% accuracy). Therefore, this CAD method can be considered with extraordinary detection capability that distinguishes epileptic seizure activity from seizure-free and normal EEG activity with simple linear classifiers. This method has the potential to be used effectively as a supplement to the clinical diagnosis of epilepsy.