Classification of Normal and Abnormal ECG signals using Support Vector Machine and Fourier Decomposition Method

Abstract

Automatic classification of abnormal electrocardiogram signals plays a vital role in developing robust unsupervised ECG analysis systems. This paper proposes a novel method for the automatic classification of normal and abnormal signals. The proposed framework utilizes the Fourier decomposition method (FDM), the statistical feature extraction, and support vector machine (SVM) based signal classification. ECG signals are decomposed using the FDM algorithm to disparate the ECG components from noise and artifacts. The statistical features such as mean, kurtosis, standard deviation, skewness, and root mean square (RMS) are computed from the extracted ECG signals. Finally, the SVM-based algorithm is applied to classifying the normal and abnormal ECG signals. The proposed method is tested rigorously using the MIT-BIH arrhythmia ECG database of different lengths. The proposed method achieves better classification of the accuracy of 99.50% compared to existing methods and accurately classifies short bursts of noises. The proposed method is also suitable for reducing false alarm rates and selecting appropriate noise-specific denoising techniques in automated ECG analysis applications.