Keywords |
: |
Electrocardiogram (ECG) Signal, Feature Extraction, Interacting Multiple Modeling (IMM) Method, Heart Rate (HR), Support Vector Machine (SVM) Classification and Wavelet-based transform |
Abstract |
: |
Automatic electrocardiogram (ECG) signal classification plays a significant role in the clinical applications, to overcome the problems occur during manual annotation of the ECG recordings. The ECG beat morphologies and disease states cannot be defined easily by a single representation, since it can vary greatly for each person. This paper proposes ECG signal modeling using adaptive framework based Support Vector Machine (SVM) classification method. The main objective of the SVM classification method is to select the best cardiac parameter. Filtering of the ECG signal is performed using Lowpass Butterworth filter. Feature extraction is performed using Wavelet-based transform. The Automatic feature selection algorithm is applied for determining the best feature subset for some criterion. The heart rate (HR) is calculated based on the extracted features of the ECG signal. Detection of the cardiovascular abnormalities is performed based on the HR calculation. The proposed method achieves efficient detection of cardiovascular abnormalities, by eliminating the fault signals and reducing the error signals. |