Classification of Parkinson's Disease Using EMG Signals from Different Upper Limb Movements Based on Multiclass Support Vector Machine

Abstract

Parkinson's disease (PD) is the second most common neurodegenerative disease that affects a wide range of productive individuals worldwide. The common approach to diagnose PD is through clinical assessment of the patient, which is highly subjective and time consuming. Electromyography (EMG) can be taken as a cheap way of PD diagnosis. However, highly experienced experts are required to interpret the signals. The manual procedures are complex, time-consuming, and prone to error resulting in misdiagnosis. In this research, an automatic system for detection and classification of PD stages using EMG signals acquired from different upper limb movements is proposed. In addition, effective upper limb movement for the identification of PD has been investigated. The data required for training and testing the system was collected from flexor carpi radialis and biceps brachii muscles of 15 PD patients and 10 healthy control subjects at Jimma University Medical Center. The raw EMG signal was preprocessed and frequency and time-domain features were extracted. A multiclass support vector machine model was then trained for four-class classification (normal, early, moderate, and advanced PD levels). The performance of the system was evaluated using different performance evaluators and a promising result has been obtained. 90%, 91.7%, 95%, and 96.6% overall classification accuracies were obtained for elbow flexion by 90-degrees without load, elbow flexion by 90-degrees with load, touching the shoulder, and wrist pronation, respectively. A user-friendly interface has been also developed for ease of use of the automatic PD classification system.