Assessment of the severity of lower extremity peripheral arterial disease based on gastrocnemius surface electromyography and mechanomyography signals.

To compensate for the shortcomings of existing diagnostic methods in assessing the severity of peripheral arterial disease (PAD), this study proposed a novel non-invasive protocol for both diagnosing PAD and stratifying its severity. Surface electromyography (sEMG) and mechanomyography (MMG) signals from the gastrocnemius muscle were collected from subjects. The continuous wavelet transform was used to generate the Morlet scalogram of the processed sEMG. Subsequently, conventional feature extraction techniques were employed to procure time domain, frequency domain, and nonlinear features from the sEMG and MMG. A multimodal cascade fusion network was developed to integrate information from scalograms and multidomain features, enabling the detection and severity assessment of lower extremity PAD. The proposed method was subjected to a fourfold cross-validation experiment in 40 lower extremities. The mean sensitivity, specificity, accuracy, and F1 score were 92.36%, 96.23%, 95.00%, and 94.17%, respectively. The clinical evaluation of the method was performed on 18 lower extremities, with 17 lower extremities accurately classified. The achieved performance demonstrates the feasibility of assessing lower extremity PAD severity using gastrocnemius sEMG and MMG signals, providing a promising tool for routine PAD screening and aiding clinicians in informed decision-making.