Identification of Suitable Basis Wavelet Function for Epileptic Seizure Detection Using EEG Signals
Selection of suitable order of Daubechies (DB) wavelet for the decomposition of Electroencephalogram (EEG) signals to detect epileptic seizures is quite challenging, as experimentation is time-consuming. In existing methods, the selection of basis wavelet function for decomposition of EEG signals is carried out by considering the literature or by trial and error method. There is a very little significant literature which discusses the comparative analysis for the identification of suitable basis wavelet function (mother wavelet). However, the existing methods often fail to provide proper justification for selecting the mother wavelets. Hence, this research work addresses the fore-mentioned setback by identifying the suitable basis wavelet function based on wavelet selection methods for epileptic seizure detection. Further, the entropy-based features are extracted and classified using SVM, DT, ANN, and KNN with five complex cases (University of Bonn, Germany EEG dataset): A-B-C-D-E, AB-CD-E, C-D-E, AB-C-D, and ABCD-E. From the entropy analysis, it is evident that while extracting entropy-based features, tenth order of Daubechies wavelet (DB10) is found to be the most suitable basis wavelet function for the accurate detection of Epileptic Seizures. The performance metrics confirm the suitability of the identified basis wavelet function in terms of sensitivity, specificity and classification accuracy.
KeywordsElectroencephalogram (EEG) Discrete wavelet transform (DWT) Daubechies (DB) Basis wavelet function (BWF)
This work was supported by the IBM Shared University Research Grant and the Department of Science and Technology, India through Fund for Improvement of S&T Infrastructure (FIST) Programme (SR/FST/ETI-349/2013).
- 2.WHO Homepage: https://www.who.int/news-room/fact-sheets/detail/epilepsy. Last accessed 7 Feb 2019
- 7.Gajic, D., Djurovic, Z., Gligorijevic, J., Di Gennaro, S., Savic-Gajic, I.: Detection of epileptiform activity in EEG signals based on time-frequency and non-linear analysis. Front. Comput. Neurosci. 9(38), 1–16 (2015)Google Scholar
- 13.Zaeri, R., Ghanbarzadeh, A., Attaran, B., Moradi, S.: Artificial neural network based fault diagnostics of rolling element bearings using continuous wavelet transform. In: The 2nd International Conference on Control, Instrumentation and Automation, Shiraz, Iran, pp. 753–758 (2011)Google Scholar
- 14.Megahed, A.I., Moussa, A.M., Elrefaie, H.B., Marghany, Y.M.: Selection of a suitable mother wavelet for analyzing power system fault transients. In: IEEE Power and Energy Society General Meeting-Conversion and Delivery of Electrical Energy in the 21st Century. IEEE/Institute of Electrical and Electronics Engineers Incorporated, pp. 1–7 (2008)Google Scholar
- 15.Rodrigues, A.P., Mello, G.D., Pai, P.Srinivasa: Selection of mother wavelet for wavelet analysis of vibration signals in machining. J. Mech. Eng. Autom. 6(5A), 81–85 (2016)Google Scholar