An Epileptic Seizure Detection in Electroencephalogram using Improved Entropy

Abstract

Epilepsy is a neurological disease that refers as a problem of the central nervous portrayed by the loss of awareness and spasms. Epileptic patients are dependent upon epileptic seizures brought about by irregular electrical release that lead to the development of spasms, and loss of consciousness. Roughly 50 million individuals around the globe are determined to have epilepsy. Kids and grown-ups in the age scope of 65-70 years of age are influenced the most. The fact is that the primary driver of this disease is obscure and the majority of the indications of the epilepsy seizure can be therapeutically treated. Epilepsy patients are prone to seizures, which cause natural and loss of consciousness, specific trigger and unfortunately even death in a short time. Patients with epilepsy suffer the consequences of sudden seizures, during which they are unable to hold their self and are vulnerable to asphyxia, fatality, or damage as a result of their loss of consciousness. To present, the disorder has primarily been handled with drugs and surgery; nevertheless, anticonvulsant therapy are not totally effective for all forms of epilepsy. Actually, there is an extension to improve the detection of epileptic seizures. In this study, the epileptic seizure detection is proposed to enhance the effect of the suggested approach. In this concept, the input Electroencephalogram (EEG) signals are involved in the pre-processing step, feature extraction and classification. Initially the unwanted noise is removed in pre-processing stage. After that, the features are extracted using fuzzy entropy and the last stage is classification. Here, Artificial Neural Network (ANN) is used to classify the signals. This process effectively increases the identification of epileptic seizure detection. Comparing to the classification of proposed ANN, grey wolf optimization and ANN, the experimental results of oppositional crow search algorithm for training ANN shows better performance of the proposed methods to identify the seizure detection.