Enhancing ECG Signal Alignment Accuracy with Smoothed-Quasi-Periodic Dynamic Time Warping

An accurate alignment of electrocardiogram (ECG) signals is essential for effective diagnosis and classification in healthcare by minimizing the variability due to noise interference, motion artifacts, and individual anatomical or physiological differences. The traditional dynamic time warping (DTW) methods, while commonly used, often struggle with noise and morphological variability in ECG data. This study proposes smoothed-quasi-periodic dynamic time warping (SQP-DTW), a novel approach incorporating a smoothing regularization feature into the quasi-periodic DTW (QP-DTW) framework. Using the MIT-BIH Arrhythmia Dataset, we compare SQP-DTW against standard DTW, Shape-DTW, and QP-DTW methods. Our results demonstrate significant enhancements in alignment accuracy in the case of Baseline Wander, Muscle Artifact, and Electrode Motion Noise, evaluated through correlation and percent root difference (PRD). The correlation evaluation shows that QP-DTW and SQP-DTW consistently achieved higher alignment accuracy than standard DTW and ShapeDTW, where SQP-DTW reached a correlation as high as ${0. 9 8}$. Meanwhile, SQP-DTW, particularly under electrode motion noise, performed best, with a percent root difference evaluation of only 20.82 percent. Prospective studies will explore the application of SQP-DTW in real-time ECG monitoring wearable systems and its adaptation to other biomedical signal processing tasks.