Computer Aided Diagnosis of Epileptic Seizure in Human Electroencephalogram using Discrete Wavelet Transform with an Adaptive Neuro – Fuzzy System
Ajala Funmilola A1, Oludipe Olusanmi2, Opiarighodare Donaldson Kesiena3, Olukumoro Olugbenga Sunday4
1Ajala Funmilola A., Department of Computer Science and Engineering, Faculty of Engineering and Technology, Ladoke Akintola University of Technology, Ogbomoso, Oyo State, Nigeria.
2Oludipe Olusanmi, Department of Computer Technology, School of Technology, Yaba College of Technology, Yaba, Lagos, Nigeria.
3Opiarighodare Donaldson Kesiena, Department of Computer Science and Engineering, Faculty of Engineering and Technology, Ladoke Akintola University of Technology, Ogbomoso, Oyo State, Nigeria.
4Olukumoro Olugbenga Sunday, Department of Computer Technology, School of Technology, Yaba College of Technology, Yaba, Lagos, Nigeria.
Manuscript received on 12 December 2018 | Revised Manuscript received on 23 December 2018 | Manuscript published on 30 December 2018 | PP: 27-36 | Volume-8 Issue-2, December 2018 | Retrieval Number: B2535128218/18©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: This study presents a computer aided epileptic seizure detection technique that uses discrete wavelet transform (DWT) with an adaptive neuro fuzzy system. The technique comprises electroencephalogram (EEG) signals data acquisition and synthesis of EEG signals, decomposition of the EEG signals and extraction of features, and classification of the features vectors. The technique which was implemented in Matlab (version 7.6) environment, exploits DWT strength in both time and frequency domain for the decomposition and extraction of characteristic features and an adaptive neuro-fuzzy inference system (ANFIS) for classification of transformed feature vectors that were used as input to the ANFIS. FiveEEG signals types were synthesized and decomposed. Features extracted from the decomposed signals were used to train and test the ANFIS classifier. The ANFIS blend neural network (NN) adaptive capabilities and fuzzy inference system (FIS). The performance of the ANFIS was measured in terms of sensitivity, specificity and total classification accuracy. And the results obtained showed that the hybrid model and/or technique which consist of DWT and ANFIS attained high level of accuracy in the classification of EEG signals as either epileptic or normal with minimal false detection.
Keyword: Discrete Wavelet Transform (DWT), Adaptive Neuro-Fuzzy Inference System (ANFIS), Epileptic Seizure, Electroencephalogram.
Scope of the Article: Discrete Optimization