摘要
對於因中風或癱瘓導致部分肢體無法自主運動之患者，臨床上常用電刺激或輔具，使患者恢復部分運動能力。這類的輔助方式通常需要患者以正常側的動作做為控制源，所以只對偏癱的患者有效。至於全身癱瘓的患者，上述輔助方法完全無效。因此希望在患者大腦未受損的情況下，以其大腦皮質m波之變化做為輔具或電刺激之控制源。如果要將m波之變化當成控制源，必須先能夠準確的鑑別出m波的變化。本研究之目的在於建構一人腦與電腦的介面系統，偵測受測者之m波變化以判斷其是否進行姆指之自主動作。
目前應用於頻率分析的訊號處理方法除了傳統傅利葉分析外，以小波分析的效率最好，對未知訊號的頻率分佈在時間軸上可以得到很好的解析度，適合應用於腦波的分析處理。另外，類神經網路的學習與感知能力也是合應用於m波之判別。因次本研究結合這兩種方法，以小波分析分解出m波，再以類神經網路進行辨識，稱為小波神經網路。
實驗中以具有即時訊號處理功能的介面卡作為主要的數據分析工具。除了即時判別以外，也作離線的數據分析，並且與短時間傅利葉轉換(STFT)的分析結果互相比較。人體實驗結果證實以小波神經網路確實可以提高m波的判別結果。

Abstract
In rehabilitation of the paralyzed patients, the functional electrical stimulation(FES) or prosthesis are often adopted in clinical practice. But these assistive devices are not suitable for fully paralyzed patients, because they have no intact side to generate control command. If their brain cortex is not fully damaged, they may be able to control the prosthesis with their own EEG of m wave. That means if the m wave is taken as a control source, the variations of m wave have to be detected precisely via a brain-computer interface. The objective of this research is to construct a real-time system to detect whether the subject had the voluntary movements of thumb via functional variation of m wave.
Fourier analysis is a traditional method in frequency analysis of EEG, but recently, the wavelet analysis has receive more attention. The wavelet analysis is not only accurate at frequency analysis, but also has good resolution on the time axe, so it is feasible to decompose the brain wave in time frequency domain. In another away, neural network has the ability for supervised learning and pattern recognition, it could be suitable for classifying m wave. In this research, we combine the above mentioned methods into a wavelet neural network and use it to decompose the m wave and to detect the suppression of m wave in real time.
A real-time signal processing system based on DSP chip was developed in this study. Beside real-time classification, off-line analyses are also performed, and the results are compared with these of a traditional short-time Fourier transform. Human test results indicated that wavelet neural network is feasible in increasing the success rate of m wave classification.

參考文獻
1. C. Neuper, A. Schlögl, G. Pfurtscheller, “Enhancement of left-right sensorimotor EEG differences during feedback-regulated motor imagery,” Clin. Neurophysio., Vol. 16, No. 4, pp.373 – 382, 1999.
2. G. Pfurtscheller and A. Berghold, “Patterns of cortical activation during planning of voluntary movement,” Electroenceph. Clin. Neurophysiol., Vol. 72, pp.250–258, 1989.
3.王政道，”以m波作為控制源的可行性之研究”，1999，國立成功大學機械研究所碩士論文。
4. 黃志旭，”拇指運動對m波的影響”，2001，國立成功大學機械研究所碩士論文。
5. R.Wolpaw, N. Birbaumer, “Brain–Computer Interface Technology: A Review of the First International Meeting,” IEEE Transaction On Rehabilitation Engineering, Vol. 8, No.2, pp.164-173, 2000.
6. A. Stancak, G. Pfurtscheller, “The Effects of Handeness And Type of Movement On The Contralateral Perponderance of mu-Rhythm Desynchronisation,” EEG and Clinical Neurophysiology, pp.174-182, 1996.
7. G. Edlinger, Paul Wach, and G. Pfurtscheller, “On the Realization of an Analytic High-Resolution EEG,” IEEE Transaction On Biomedical Engineering, Vol. 45, No.6, pp.736 – 745, 1998.
8. E. Haselsteiner, G. Pfurtscheller, “Using Time-Dependent Neural Networks for EEG Classification,” IEEE Transaction On Rehabilitation Engineering, Vol. 8, No. 4, pp.457-464, 2000.
9. W. Klimesh, H. Russegger, M. Doppelmayer and T.Pachinger, “A method for the calculation of induced band power: implications for the significance brain oscillations,” Electroencephalogr. Clin. Neurophysiol, Vol.108, pp.123-130, 1998.
10. C. Andrew, G. Pfurtscheller, “On The Existence of Different Alpha Band in The Hand Area of Man,” Neuroscience Letters, pp.203-206, 1999.
11. R.Lacourse, “Performance Monitoring From The EEG Power Spectum With a Radial Basis Function Neural Network,” IEEE, pp.19-21, 1997.
12. J. McFarland, J .McCane, J. Wolpaw, “EEG-Based Communication and Control: Short-Term of Feedback,”
IEEE Transaction on Rehabilitation Engineering , Vol. 6, No. 1, pp.7-12, 1998.
13. A. Kubler, B. Kotchoubey, T. Hinterberger, N. Ghanayim, “The thought translation device: a neurophysiological approach to communication in total motor paralysis,” Exp. Brain Res. Vol.124, pp.223 – 232, 1999.
14. C. Ming, G. Shangkai, “An EEG-Based Cursor Control System,” IEEE Transaction on Rehabilitation Engineering 1999, pp.669 – 672.
15. C. W. Ko, B. Y. Kim, “Automatic sleep stage scoring system using genetic algorithms and neural network,” Engineering in Medicine and Biology Society, Vol.2, pp.849 -850, 2000.
16. J. Carballido, A. Kostov, J. Martenez, “Enhancement of EEG control signals in the development of Brain Computer Interface,” Proceedings of the First Joint MBES Conference.
17. N. Karayanis, A. Venfisanopoulos, “On the training and performance of high order neural network,” Mathematical Biosciences, pp.129-143, 1995.
18. M. Pregenzer, G. Pfurtscheller, “Distinction Sensitive Learning Vector Quantization-application as a classifier based feature selection method for the BCI ,” Exp. Brain Res. , Vol. 6, No. 1, pp.7-12, 1999.
19. L. Wang, K. K. Teo, “Prediction time series with wavelet package neural network,” IEEE Transaction on Rehabilitation Engineering, pp.1593-1599, 2001.
20. C. T. Lin, C. S. George, Neural fuzzy ystems, Wiley Inter-Science, New York, 1995.
21. C. Goswami, A. K. Chan, Fundamental of wavelets: Theory, algorithm and application, Wiley Inter-Science, New York, 1999.
22. Q. Chan, A. Benvenist, “Wavelet neural network,” IEEE Transactions on neural network, Vol.3, No.6, pp.889-898, 1992.
23. P. Dopente, L. Angresani, “Wavelet-Network based detection and classification for transients,” IEEE Transactions on Instrumentation and Measurement, Vol. 50, No.5, pp.186~194, 2001.