Performance Analysis of VMD for Phonocardiogram (PCG) Signal Processing

Heart murmurs are important biomedical signals that provide early indicators of cardiovascular disorders. Digital heart sound recordings, known as phonocardiograms (PCGs), enable objective analysis and automated classification of cardiac abnormalities, thereby improving diagnostic accuracy and clinical efficiency. However, PCG signals are often corrupted by noise and interference, making robust signal decomposition and feature extraction essential. Among modern signal processing techniques, VMD stands out as a powerful and principled approach for PCG analysis. By formulating signal decomposition as a constrained variational optimization problem, VMD enables the extraction of well-separated, band-limited modes with strong robustness to noise and reduced mode mixing. These properties make VMD particularly effective for preserving pathological cardiac features such as murmurs and transient abnormalities. Although classical methods such as EMD and DWT have been widely used for biomedical signals analysis, they suffer from inherent limitations, including mode mixing, sensitivity to noise, and dependence on predefined basis functions. In contrast, comparative results demonstrate that VMD consistently achieves superior reconstruction accuracy than the other techniques. Consequently, VMD represents a reliable and efficient framework for enhancing heart sound analysis and supporting early detection of cardiovascular diseases.