A hybrid approach for machine learning based beat classification of ECG using different digital differentiators and DTCWT.

This research paper presents a systematic approach to ECG beat classification using advanced machine learning techniques. The study classifies ECG beats into six distinct classes based on annotations from the MIT-BIH Arrhythmia Database. The methodology incorporates manual feature extraction using the Dual-Tree Complex Wavelet Transform (DTCWT) to capture critical information from ECG signals. Four novel digital filters are employed for ECG signal differentiation to further enhance the discriminative power of the extracted features, effectively isolating key components such as R peaks and QRS complexes. The Pan-Tompkins algorithm is enhanced with these digital differentiators, improving its effectiveness in QRS detection. A comprehensive feature set is constructed, combining morphological features derived from DTCWT with four statistical features. This feature set is then used to train different machine learning classifiers, facilitating the precise classification of ECG beats. The performance of these classifiers is evaluated using key metrics, including accuracy, precision, recall, and F1 score. The experimental analysis, conducted on the complete MIT-BIH Arrhythmia Database comprising 48 records, ensures the robustness and generalizability of the proposed approach. The results demonstrate the efficacy of this methodology in accurately classifying ECG beats, making significant advancements in automated ECG analysis and cardiac signal processing.