Discrimination of Parkinsonian Tremor from Essential Tremor Using the Nonlinear Analysis of Wrist Muscle EMG Signals

摘要

Discrimination of Parkinsonian tremor from Essential tremor is a controversial diagnostic category. Similar clinical traits, usually results in incorrect diagnosis, and degrade the performance of clinical diagnostic methods. In this paper, a new methodology is presented to discriminate the patients suffering from Parkinsonian tremor and the patients suffering from Essential tremor through the analysis of wrist muscle surface EMG signals recorded in rest and a postural position. Twenty patients with Parkinson disease and twenty patients with Essential tremor participated in the experimental studies. Some frequency and time features were extracted from EMG signals and were fed to nonlinear classifiers. The MLP neural network and SVM were used as the classifiers. According to the results, using the EMG signals recorded in rest position, the SVM classifier reached the performance of 89% accuracy and MLP classifier reached the performance of 92.25% accuracy. In addition, using the EMG signals recorded in a postural position, the SVM classifier reached the performance of 90.25% accuracy and MLP classifier reached the performance of 95.75% accuracy.