Tremor is the most common of all involuntary movement. A lot of tremor patients in upper limb have serious difficulties performing daily living activities. We have developed the exoskeleton robot for tremor patient. In this paper, we focused on to develop a signal processing method to extract the voluntary movement from the electromyogram (EMG) signal in which the voluntary movement and tremor were mixed. We have researched about following two methods to recognize the voluntary movement: one is Low pass filter and neural network (NN), the other is Short Time Fourier Transform and NN. The low pass filter and neural network (NN) were effective for recognition of healthy subject's movement. However, these methods were not applied to the tremor patient's movement due to the characteristic oscillation of the EMG signal in the tremor patient. The proposed algorithm, which was composed of the Short Time Fourier Transform and NN, dramatically improved the recognition rate of tremor patient's movement. It was confirmed that the signal processing using STFT and NN is suitable for the recognition of the tremor patients In future, we will develop more accurate algorithm based on this study, and finally conduct the clinical test to show effectiveness of our system.
|ホスト出版物のタイトル||2009 4th International IEEE/EMBS Conference on Neural Engineering, NER '09|
|出版ステータス||Published - 2009|
|イベント||2009 4th International IEEE/EMBS Conference on Neural Engineering, NER '09 - Antalya|
継続期間: 2009 4月 29 → 2009 5月 2
|Other||2009 4th International IEEE/EMBS Conference on Neural Engineering, NER '09|
|Period||09/4/29 → 09/5/2|
ASJC Scopus subject areas