Analiza wielorozdzielcza i sieć SVM w zastosowaniu do klasyfikacji potencjałów czynnościowych

• Autorzy: Dobrowolski A. P. , Wierzbowski M. , Tomczykiewicz K.

Warianty tytułu

EN

Multiresolution analysis and Support Vector Machine for motor unit classification

• Języki publikacji: PL

Abstrakty

PL

W artykule przedstawiono nową metodę diagnozowania chorób nerwowo-mięśniowych opartą na analizie skalogramów wyznaczonych za pomocą falek Symlet 4. Z otrzymanych skalogramów wyekstrahowano 5 cech, które po analizie w sieciach SVM sprowadzono do pojedynczego parametru klasyfikującego analizowane przypadki do grupy miogennej, neurogennej i prawidłowej. Implementacja programowa metody stworzyła narzędzie diagnostyczne wspomagające badanie EMG o bardzo wysokim prawdopodobieństwie prawidłowej oceny stanu mięśnia (błąd całkowity wyniósł 0,66% - dwie błędne klasyfikacje na 300 badanych pacjentów).

EN

The paper presents a new approach to the computer aided diagnostic systems for the needs of quantitative electromyography. The approach is based on the analysis of wavelet scalograms of the motor unit action potentials calculated on the basis of 4th order Symlet wavelet. The scalograms provide the vector consisting of five features describing the state of a muscle. The vectors serve to carry out a classification of pathology by using Support Vector Machine method. The QEMG examination consists of the insertion of a needle electrode into a muscle and a registration of muscle potentials during low effort. Registered potentials are called motor unit action potentials (MUAPs). A diagnosis is usually preceded by a statistical analysis of a MUAP shape. An inconvenience of this procedure in a clinical practice is caused by high time- consumption arising, among others, from the necessity of determination of many parameters, usually between 4 and 7. Additionally, an ambiguity in determination of basic temporal parameters can cause doubts during comparison of parameters found by the physician with standard ones determined in other research centre, which mostly uses equipment of older generation. Measurement results on diagnostic method deprived of above - mentioned disadvantages are described in the paper. The aim of our work was a development of new methods for transformation of action potential signals observed in EMG records for healthy muscles and changed ones. The multiresolution decomposition method was devoted to determination of a vector of characteristic features of signals corresponding to analyzed categories. Then, this vector was used for effective recognition of these categories using linear Support Vector Machine technique. The final effect of research is development of a definition for numerical classificator directly enabling a unique diagnosis to be made. An essential advantage of the suggested classificator is a precise and algorithmically realized definition which enables an objective comparison of examination results obtained by physicians with different experience and working in different research centres. The presented diagnostic method ensures significantly better distinction between pathological and healthy cases as compared to methods using traditional parameters defined in time and frequency domains. Sensitivity of the wavelet method, for 100% specificity, amounts to 100% for myogenic and to 97% for neurogenic pathological states.

Słowa kluczowe

PL

elektromiografia ilościowa; QEMG; potencjał czynnościowy jednostki ruchowej; PJR; analiza wielorozdzielcza; aproksymacja falkowa; skalogram; sieć SVM; diagnostyka medyczna

EN

quantitative electromyography; QEMG; motor unit action potential; MUAP; multiresolution analysis; wavelet approximation; scalogram; support vector machine; SVM; medical diagnosis

• Wydawca: Wojskowa Akademia Techniczna im. Jarosława Dąbrowskiego
• Czasopismo: Biuletyn Wojskowej Akademii Technicznej
• Rocznik: 2009
• Tom: Vol. 58, nr 3
• Strony: 275--302
• Opis fizyczny: Bibliogr. 25 poz., tab., wykr.

Twórcy

autor

Dobrowolski A. P.

autor

Wierzbowski M.

autor

Tomczykiewicz K.

• Wojskowa Akademia Techniczna, Wydział Elektroniki, Instytut Systemów Elektronicznych, 00-908 Warszawa, ul. S. Kaliskiego 2

Bibliografia

• [1] E. Stalberg, S. D. Nandedkar, D. B. Sanders, B. Falck, Quantitative motor unit potential analysis, J. Clinical Neurophysiology, vol. 13, 5, 1996, 401-422.
• [2] E. Zalewska, I. Hausmanowa-Petrusewicz, Effectiveness of motor unit potentials classification using various parameters and indexes, J. Clinical Neurophysiology, vol. 111, 8, 2000, 1380-1387.
• [3] C. Bischoff, E. Stalberg, B. Falck, K. Edebol Eeg-Olofsson, Reference values of motor unit action potentials obtained with multi - MUAP analysis, Muscle & Nerve, vol. 17, 1994, 842-851.
• [4] S. Shahid, J. Walker, G. M. Lyons, C. A. Byrne, A. V. Nene, Application of higher order statistics techniques to EMG signals to characterize the motor unit action potential, IEEE Transactions on Biomedical Engineering, vol. 52, 2005, 1195-1209.
• [5] C. I. Christodoulou, C. S. Pattichis, Unsupervided pattern recognition for the classification of EMG signals, IEEE Transactions on Biomedical Engineering, vol. 46, 1999, 169-178.
• [6] C. S. Pattichis, A. G. Elia, Autoregressive and cepstral analyses of motor unit action potentials, Medical Engineering & Physics, vol. 21, 1999, 405-419.
• [7] A. Dobrowolski, K. Tomczykiewicz, P. Komur, Spectral analysis of motor unit action potentials, IEEE Transactions on Biomedical Engineering, vol. 54, 12, 2007, 2300-2302.
• [8] A. Dobrowolski, K. Tomczykiewicz, P. Komur, Fourier analysis of motor unit action potentials, Electronics and Telecommunications Quarterly, vol. 53, 2, 2007, 127-141.
• [9] A. Dobrowolski, P. Komur, K. Tomczykiewicz, Diagnosis of Muscle Condition on the Basis of MUP Spectral Analysis, IEEE Instrumentation and Measurement Technology Conference (IMTC'07), Warsaw, 2007.
• [10] K. Tomczykiewicz, A. Dobrowolski, P. Komur, Application of Fourier assessment in analysis of electromyography - an initial report, 13th European Congress of Clinical Neurophysiology (ECCN 2008), Istanbul, Turkey, 2008.
• [11] P. Komur, A. Dobrowolski, T. Dąbrowski, K. Tomczykiewicz, Automated diagnostic method supporting EMG examination, 30th Annual International IEEE EMBS Conference (EMBC'08), 1116-1119, Vancouver, Canada, 2008.
• [12] C. S. Pattichis, M. S. Pattichis, Time - scale analysis of motor unit action potentials, IEEE Transactions on Biomedical Engineering, vol. 46, 1999, 1320-1329.
• [13] A. Dobrowolski, J. Jakubowski, K. Tomczykiewicz, Linear discriminant analysis of MUAP scalograms, 30th Annual International IEEE EMBS Conference (EMBC'08), 1100-1103, Vancouver, Canada, 2008.
• [14] I. Rodriguez-Carreno, A. Malanda-Trigueros, L. Gila-Useros, J. Navallas-Irujo, J. Rodriguez-Falces, Filter design for cancellation of baseline - fluctuation in needle EMG recordings, Computer methods and programs in biomedicine, vol. 81, 2006, 79-93.
• [15] E. Zalewska, I. Hausmanowa-Petrusewicz, E. Stalberg, Modeling studies on irregular motor unit potentials, J. Clinical Neurophysiology, vol. 115, 3, 2004, 543-556.
• [16] J. Huber, Badania neurofizjologiczne, w: A. Szczeklik (red.), Choroby wewnętrzne, Podręcznik multimedialny oparty na zasadach EBM, Medycyna Praktyczna, Kraków, 2006.
• [17] I. Daubechies, Ten Lectures on Wavelets, CBMS-NSF Lecture Notes, 61, SIAM, 1994.
• [18] S. G. Mallat, A Theory for Multiresolution Signal Decomposition: the Wavelet Representation, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 11, 7, 1989, 674-693.
• [19] S. G. Mallat, A Wavelet Tour of Signal Processing, Academic Press, 1999.
• [20] V. Vapnik, Estimation of dependences based on empirical data, Springer, 2006.
• [21] S. Osowski, Sieci neuronowe do przetwarzania informacji, Oficyna Wydawnicza PW, Warszawa, 2006.
• [22] V. Vapnik, Statistical Learning Theory, John Wiley & Sons, 1998.
• [23] C. Cortes, V. Vapnik, Support vector networks, Machine Learning, 20, 1995, 273-297.
• [24] C. J. C. Burges, A tutorial on support vector machines for pattern recognition, Data mining and knowledge discovery, vol. 2, 1998, 121-167.
• [25] J. C. Platt, Fast training of support vector machines using sequential minimal optimization, in: B. Schölkopf, C. J. C. Burges and A. J. Smola, Advances in kernel methods - support vector learning, MIT Press, 1999, 185-208.