The weight load inconsistency effect on voluntary movement recognition of essential tremor patient

Abstract

Essential Tremor (ET) refers to involuntary movements of a part of the body. ET patients have serious difficulties in performing their daily living activities. Our ultimate goal is to develop a system that can enable ET patients to perform their daily living activities. We are in the process of developing an exoskeletal robot for ET patients. This robot is controlled by estimation of voluntary movement using surface electromyogram (EMG) signal input and a Neural Network (NN) learning algorithm. However, the EMG signal of ET patients contains not only signals from voluntary movements but also noise from involuntary tremors. We have therefore developed a signal processing method to suppress tremor noise present in the surface EMG signal. The proposed filter is based on the hypothesis that tremor noise can be approximated to powered sine wave. It have been confirmed that the proposed filter increases the accuracy of recognition. In this paper, we have focused on the effect of inconsistency of weight load between instruction signal and input signal. When the instruction signal comprised unloaded motion, our voluntary movement estimation method worked stably with the loaded motion's EMG input.