ECG AND EEG BIOSIGNALS CLASSIFICATION BY HYBRID DWT-ICA-SVM

Malika D. Kedir-Talha and Saliha Ould Slimane

Keywords

Electrocardiogram, electroencephalogram, DWT, ICA, SVM

Abstract

This work shows the possibility of achieving a diagnostic assistance of the biomedical signals: ECG and EEG, using the same processing and classification methods. The processing of these two non-stationary signals requires the application of a time-frequency transform, such as the wavelet. The frequency characteristics of the different waves (δ, θ, α, β) for the EEG signal and (P, QRS, T) for ECG, justifies our choice for the discrete wavelet transform (DWT). Our work proves that only one statistical feature, the energy variance of detail coefficients for the first five decomposition levels, is sufficient to represent these two types of signals. We applied the independent component analysis (ICA) to reduce the data space and keep the most relevant ones. To detect cardiac abnormality or prevent epileptic seizures, classification by support vector machines (SVM)has made it possible for us to achieve recognition rates of 100% using a wise choice of DWT.

Important Links:

Go Back