Process Efficient Artificial Neural Network-Based Approach for Channel Selection and Classification of Seizures

Abstract

The rate of neurological and psychiatric disorders is increasing rapidly in our day-to-day life, and epilepsy is one of the common neurological disorders in brain which is a constant, persistent brain disorder characterized by abnormal electrical activity in the brain. A single seizure cannot be predicted as epilepsy, because it is a condition of two or more arbitrary seizures. Seizures can also result in the loss of consciousness. The origin of seizures may be due to high electrical discharges from a collection of brain cells. Therefore, the development of accurate computer-aided diagnostic system for the classification of brain disorders is strongly desired. There is a significant interest in the research community for the development of reliable EEG-based automated tools. With the advancement of new signal processing techniques and mathematical algorithms in EEG analysis, supporting methods in medical decision and diagnosis can be developed to avoid tedious analysis of voluminous records and obtain clarity about the brain pathology. The importance of epileptic seizure detection is increasing due to the higher statistical occurrence of epileptic seizures in our normal day-to-day life activities. The epileptic seizure event must be appropriately detected to avoid its unexpected occurrences in the future. Hence in order to make this process as efficient as possible, this research work proposed an ANN-based intelligence method epileptic seizure classification and detection technique. The research work also includes result and discussion section to provide validation for the proposed methodology. The results showed that feedforward BPNN classifiers resulted in satisfactory classification accuracy percentages. This method affords reliable computerized methodology for appropriate EEG signal classification and better decision-making for epileptic seizure diagnosis in clinical practice.