Yücel KOÇYİĞİT, Mehmet KORÜREK

EMG işaretlerini dalgacık dönüşümü ve bulanık mantık sınıflayıcı kullanarak sınıflama

Deri yüzeyinde algılanan elektromiyografik (EMG) işaretleri, kas liflerinin kasılması sonucu oluşan çok sayıda aksiyon potansiyellerin birleşimidir. Şimdiye kadar biyomedikal mühendisliğinde çeşitli uygulama alanları bulmuştur. Bu uygulamalardan biri de protez kontrolüdür.Bu çalışmanın hedefi, öznitelik çıkartma yöntemi olarak zaman-frekans domeni analiz yöntemlerini kullanarak protez koluna ait dört farklı hareket için EMG işaretlerini daha iyi sınıflamayı gerçekleştirmektir. Bunun için boyut azaltma ve bulanık sınıflama yöntemleri de incelenmiştir. Sınıflama problemi öznitelik çıkartma, boyut azaltma ve örüntü sınıflama aşamalarına ayrılır. Dalgacık dönüşümü öznitelik çıkartma yöntemi olarak büyük üstünlük sağlar. Özniteliklerin çıkartma aşamasında yüksek boyuta sahip olmalarından dolayı sınıflama başarısı, Ana Bileşenler Analizi (ABA) ve Bağımsız Bileşenler Analizi (BBA) gibi uygun boyut azaltma yöntemleriyle gerçekleştirilebilir.

EMG signal classification using wavelet transform and fuzzy logic classifier

The electromyographic (EMG) signal observed at the surface of the skin is the sum of thousands of small potentials generated in the muscle fibers. After this signal are processed it can be used as a control source of artificial limbs. The objective of this work is to achieve better classification for four different movement of a prosthetic limb making an analysis of time-frequency domain methods as a feature extraction tools in the problem of the EMG signal while investigating the related dimensionality reduction and fuzzy classification methods. The classification problem may be divided into the stages of feature extraction, dimensionality reduction, and pattern classification. It is shown that wavelet transform (WT) provide a powerful framework for feature extraction. Because of high dimension of features at the extraction stage, the success of classification can be achieved by employing suitable dimensionality reduction methods which are Principal Component Analysis and Independent Component Analysis outperform WT features. The other stage is the pattern classification in which fuzzy clustering methods and artificial neural networks (ANN) are used. The clustering methods are used to obtain membership values of the EMG signals for each class or cluster. The values are necessary during the classification stage. As classifier, Fuzzy K-Nearest Neighbor classifier is used. ANN are used to compare these methods as classifier.

PDF

___

Asres, A., Dou, H., Zhou, Z., Zhang, Y. ve Zhu, S. (1996). A combination of AR and neural network technique for EMG pattern identification, 18th Annual International Conference of the IEEE Engineering in Medicine And Biology Society, 1464-1465, Amsterdam.
Chan, F.H.Y., Yang, Y.S., Lam, F.K., Zhang, Y.T. and Parker, P.A., (2000). Fuzzy EMG classification for prosthesis control, IEEE Transactions on Rehabilitation Engineering, 8, 305-311.
Del Boca A. Park, D.C., (1994). Myoelectric signal recognition using fuzzy clustering and artificial neural networks in real time, IEEE World Congress on Computational Intelligence, 5, 3098-3103.
Doerschuk, P.C., Gustafson, D.E. and Willsky, A.S., (1983). Upper extremity limb function discrimination using EMG signal analysis, IEEE Transactions on Biomedical Engineering, 30, 1, 18-29.
Englehart, K., (1998). Signal representation for classification of the transient myoelectric signal, Ph. D. Dissertation, University of New Brunswick, Fredericton, N.B., Canada.
Englehart, K., Hudgins, B., Parker, P.A., (2001). A Wavelet-based continuous classification scheme for multifunction myoelectric control, IEEE Transactions on Biomedical Engineering, 48, 302–311.
Geva, A. B., (1997). Dynamic unsupervised fuzzy clustering in forecasting events from biomedical signals, Ministry of Science, International Conference on Fuzzy Logic and Applications, Zichron Yaakov, Israel.
Graupe, D., Salahi, J. ve Zhang, D., (1985). Stochastic analysis of myoelectric temporal signatures for multifunction single-site activation of prostheses and orthoses, Journal of Biomedical Engineering, 7, 1, 18-29.
Hudgins, B., Parker, P.A. ve Scott, R.N., (1993). A new strategy for multifunction myoelectric control, IEEE Transactions on Biomedical Engineering, 40, 1, 82-94.
Hyvärinen, A. ve Oja, E., (2000). Independent Component Analysis: Algorithms and Applications, Neural Networks, 13, 411-430.
Kang, W., Shiu, J., Cheng, C., Lai, L., Tsao, H. ve Kuo, T., (1995). The application of cepstral coefficients and maximum likelihood method in EMG pattern recognition [movements classification], IEEE Transactions on Biomedical Engineering, 42, 777-785.
Karlık, B., Pastacı, H. ve Korürek, M., (1994). Myoelectric neural networks signal analysis, Proc. 7th Mediterranean Electrotechnical Conference, 1, 262-264, Antalya, Turkey.
Kelly, M., Parker, P.A. ve Scott, R.N., (1990). The application of neural networks to myoelectric signal analysis: A preliminary study, IEEE Transactions on Biomedical Engineering, 37, 3, 221-227.
Şeker, H., (1995). Elektromiyografik işaretlerin bulanık sınıflayıcılarla sınıflandırması, Yüksek Lisans Tezi, İstanbul Teknik Üniversitesi, İstanbul.
The FastICA MATLAB package.
http://www.cis.hut.fi/projects/ica/fastica