Bilateral sEMG Signal and Feature Fusion for Improved Parkinson's Disease Severity Assessment

Parkinson's disease (PD) is a progressive neurodegenerative disorder that requires careful monitoring of symptom severity to guide medication and related interventions. To this end, this study proposes a novel method for assessing the severity of PD using surface electromyography (sEMG) signals and feature fusion based on Fisher vector analysis. Specifically, the sEMG signals of the tibialis anterior (TA) and lateral gastrocnemius (GL) muscles were measured in 28 PD patients. To remove noise and low-frequency trends, the sEMG signals were notch filtered and detrended using a smoothness priors method. Time-domain and frequency-domain features were then extracted from the preprocessed signals using a feature extractor. The sEMG signals and features were vectorized using Fisher vector analysis and fused as input to a decision tree classifier. The results indicate that the proposed method achieved a classification accuracy of 96.00% for assessing PD severity. Furthermore, considering bilateral features was found to be more effective than unilateral features for PD severity assessment. Overall, this study highlights the potential of sEMG signals and feature fusion for improving the accuracy of PD severity assessment.