An Enhanced Automated Epileptic Seizure Detection Using ANFIS, FFA and EPSO Algorithms

Abstract

Objectives: Electroencephalogram (EEG) signal gives a viable perception about the neurological action of the human brain that aids the detection of epilepsy. The objective of this study is to build an accurate automated hybrid model for epileptic seizure detection. Methods: This work develops a computer-aided diagnosis (CAD) machine learning model which can spontaneously classify pre-ictal and ictal EEG signals. In the proposed method two most effective nature inspired algorithms, Firefly algorithm (FFA) and Efficient Particle Swarm Optimization (EPSO) are used to determine the optimum parameters of Adaptive Neuro Fuzzy Inference System (ANFIS) network. Results: Compared to the FFA and EPSO algorithm separately, the composite (ANFIS+FFA+EPSO) optimization algorithm outperforms in all respects. The proposed technique achieved accuracy, specificity, and sensitivity of 99.87%, 98.71% and 100% respectively. Conclusion: The ANFIS-FFA-EPSO method is able to enhance the seizure detection outcomes for demand forecast in hospital.