Experiment results 103

Table 5.6: Analysis5. In this figure, LN_L represents number of results assigned to L, N_R the number of results assigned to R, LNr the number of results assigned to resting

Dataset Classifier L or R FN FP LNr LNL LNR

G02 L:4NN,thre=0.12; L 44.44 46.27 14 25 6

R:SOM,thre=0.12; R 80.00 51.75 16 20 9

G03 L:SOM,thre=0.35; L 56.67 56.43 12 13 5

R:4NN,thre=0.35; R 50.00 54.76 8 7 15

G04 L:SOM,thre=0.30; L 66.67 48.15 11 9 7

R:SOM,thre=0.20; R 70.37 46.03 14 5 8

Chapter 6

Conclusions

In this chapter, we review the most important issues when analyzing EEGs in an asyn-chronous BCI system and then illustrate the contribution and advantages of our method.

The analysis of EEGs in an asynchronous BCI system in this thesis were:

Signal preprocessing

The aim of this step is to enhance the SNR of the EEGs. We applied artifact removal and bandpass filter to achieve the aim. When the number of the channels is large, some spatial filter may be applied to make EEGs reference-free. In our work, when analyzing asynchronous BCI, we do not apply spatial filter because we only use EEGs of three channels,C3,Cz,C4 to detect the user’s intention.

Feature extraction

Because the signals of performing the predefined mental task are usually smaller than the onging EEGs, the signals we interested are concealed in the irregular ongoing signals and noise. To overcome the problem, we will perform some approaches to extract the signals we interested. In this work, we compare four feature extraction approaches including ERD/ERS, CSP. Haar wavelet and Morlet wavelet. We finally decide to use Morlet wavelt to analyze the EEGs in an asynchronous BCI system due to the analysis result in Chapter 4.

• Feature selection

Due to the dimension of features is usually high, we perform t-statistic and FFS to select more discriminative features to represent the original data. The two approaches has a satisfied and efficient results in the analysis of our work. We finally use only 2 features to present a data and have a reasonable classification accuracy in the analysis of Chapter 5. The property of few features will have a significant influence when analyzing a online BCI because the computation time is decreased a lot.

• Classification

We involve the technique of one-class classification to settle the problem of recog-nize between the active mental tasks and resting states. Because the resting state

Conclusions 107

is composed of a lot of different ongoing states, thus is a wide-spreading distribu-tion. Besides, the resting states of training phase and testing phase is very different because the user having to pay attention to the feedback is very active at the test-ing phase. We think that if we can develop a good model of active states and output 0(zero) when the testing data is of resting states, then the problem in an asynchronous BCI may be solved as well. Thus, using the technique of one-class classification to handle the problem becomes very attractive for it can avoid to include a resting class and only focus on how to fit the data of active states. We also have discussed the two approaches of using two one-class classifiers and of using two one-class classifiers and a two-class classifier to recognize the left, right and resting states in Chapter 5.

In this thesis, the achievements can be summarized as follows.

1. We use few number of channels,3, to achieve the near same or better accuracy than Graz,2004 [51] of using 27 channels.

2. We develop a efficient feature selection process to select significatly discriminative features. For example, in our work, when analyzing EEGs of an asynchronous BCI, we use only 2 features to present the EEGs and delivered only the 2 information to the classifier yet still obtain a reasonable accuracy.

3. According to our experiments using 2005 BCI competition III datasets, the proposed methods can achieve higher stability and accuracy than Graz’s method [51] proposed in 2004.

Bibliography

[1] Eugene Tolunsky A. James Rowan. Primer of EEG With a Atlas: With a Mini-Atlas. Butterworth-Heinemann, 2003.

[2] Kotchoubey B., Haisst S., Daum I., Schugens M., and Birbaumer N. Learning and self-regulation of slow cortical potentials in older adults. Experimental Aging Re-search, 26:15–35, 2000.

[3] Jessica Bayliss and Dana Ballard. A virtual reality testbed for bracomputer in-terface research. IEEE Transactions on Rehabilitation Engineering, 8(2):118–190, 2000.

[4] N. Birbaumer, N. Ghanayim, T. Hinterberger, I. Iversen, and B. Kotchoubey. A spelling devices for the paralyzed. Nature, 398:297–298, 1999.

[5] G.L. Calhoun and G.R. McMillan. EEG-based control for human-computer interac-tion. Hics, 00(00):0–4, 1996.

[6] J.K. Chapin and G. Gaal. Robotic control from realtime transformation of multi-neuronal population vectors. Brain-Computer Interface Technology: Theory and Practice: First International Meeting Program and Papers, 1999.

[7] J.K. Chapin, K.A. Moxon, R.S. Markowitz, and M.A.L. Nicoleslis. Real-time control of a robot arm using simultaneously recorded neurons in the motor cortex. Nature Neurosci., 2(7):664–670, 1999.

[8] J. d. R. Millan, J. Mourino, M. Franze, F. Cincotti, M. Varsta, J. Heikkonen, and

F.Babiloni. A local neural classifier for the recognition of EEG patterns associated to mental tasks. IEEE Transactions on Neural Networks, 13:678–686, May 2002.

[9] John Polich Daran Ravden. Habitation of p300 from visual stimuli. International Journal of Psychophysiology, 30:359–365, 1998.

[10] R. Grave de Peralta, S. L. Gonzlez, J. del R. Milln, T. Pun, and C. M. Michel. Direct non-invasive brain computer interfaces. Proceedings of the 9th International Confer-ence on Functional Mapping of the Human Brain, June 2003.

[11] Emanuel Donchin et al. The mental prosthesis: Assessing the speed of a p300-based brain-computer interface. IEEE Transactions on Rehabilitation Engineering, 8(2):174–179, 2000.

[12] Matti Hamalainen et al. Magnetoencephalography - theory, instrumentation, and ap-plications to noninvasive studies of the working human brain. Reviews of Modern Physics, 65(2):413–497, 4 1993.

[13] Niels Birbaumer et al. The thought translation device (ttd) for completely paralyzed patients. IEEE Transactions on Rehabilitation Engineering, 8(2):190–193, 2000.

[14] Peter Jezzard et al. Functional MRI: An Introduction to the Methods. Oxford Univer-sity Press, 2001.

[15] L.A. Farwell and E. Donchin. Talking off the top of your head: toward a mental prosthesis utilizing event-related brain potentials. Electroencephalogr. Clinical Neu-rophysiology, 70:510–523, 1988.

[16] Keinosule Fukunaka. Introduction to statistical pattern recognition. ACADEMIC PRESS, INC, 1990.

[17] G.Pfurschteller and C.Neuper. Motor imagery and direct brain-computer communi-cation. Proceedings of the IEEE, 89:1123–1134, July 2001.

BIBLIOGRAPHY 111

[18] T. Hinterberger, J. M. Houtkooper, and B. Kotchoubey. Effects of feedback control on slow cortical potentials and random events. The Parapsychological Association Convention, 00:0, 2004.

[19] http://faculty.washington.edu/chudler/1020.html. 10-20 system.

[20] http://www.cortechsolutions.com/g.BSanalyze EEGtoolbox.htm. The manual of g.BSanalyze.

[21] http://www.neuro.uu.se/fysiologi/gu/nbb/lectures/EEGBas.html. Eeg basis.

[22] http://www.socialresearchmethods.net/kb/stat t.htm. t-test.

[23] Bayliss JD and Ballard DH. A virtual reality testbed for brainvcomputer interface research. IEEE Transactions on Rehabilitation Engineering, 8:188–190, 2000.

[24] Kanehisa Morimoto Jingbo Pan, Tatsuya Takeshita. P300 habituation from auditory single-stimulus and oddball paradigms. International Journal of Psychophysiology, 37:149–153, 2000.

[25] editor John G. Webster. Medical Instrumentation, chapter4. John Wiely & Sons Inc., 1998.

[26] M. Joho and K. Rahbar. Joint diagonalization of correlation matrices by using newton methods with applicaiton to blind signal separation. IEEE Sensor Array and Multi-channel Signal Processing Workshop SMA, 2000.

[27] Andrea1 et al. Kbler. Brainvcomputer communication: unlocking the locked in. Psy-chological Bulletin, 127(3):358–373, 2001.

[28] S. G. Mason and G. E. Birth. A brain-controlled switch for asynchronous control applications. IEEE Transations on Biomedical Engineering, 47(10):1297–1307, OC-TOBER 2000.

[29] Dennis J. McFarland, William A. Sarnacki, and Jonathan R. Wolpaw. Brain-computer interface (BCI) operation: optimizing information transfer rates. Biological Psychol-ogy, 63:237–251, 2003.

[30] Matthew Middendorf, Grant McMillan, Gloria Calhoun, and Keith S. Jones. Brain-computer interface based on the steady-state visual-evoked response. IEEE Transac-tions on Rehabilitation Engineering, 8(2):211–214, 2000.

[31] Matthew Middendorf, Grant McMillan, Gloria Calhoun, and Keith S. Jones. Brain-computer interfaces based on the steady-state visual-evoked response. IEEE Transa-tions on Rehabilitation Engineering, 8(2):211–214, 2000.

[32] Johannes Mller-Gerking, G.Pfurtscheller, and Henrik Flyvbjerg. Designing optimal spatial filters for single-trial EEG classification in a movement task. Clinical Neuro-physiology, 110:787–798, 1999.

[33] Birbaumer N., Elbert T., Canavan A. G. M., and Rockstro h B. Slow potentials of the cerebral cortex and behaviour. Physiological Reviews, 70(1):1–41, 1990.

[34] Ernst NiederMeyer and editors Fernando Lopes da Silva. Electroencephalography.

Lippincott Williams & Wilkins, 1999.

[35] L. Otten and E. Donchin. The relationship between p300 amplitude and subsequent recall for distinctive events: Dependence on type of distinctiveness attribute. Psy-chophysiology, 37:644–661, 2000.

[36] G. Pfurtschellera and F.H. Lopes da Silva. Event-related EEG/MEG synchronization and desynchronization: basic principles. Clinical Neurophysiology, 110:1842–1857, 1999.

[37] G. Pfurtschellera and W. Klimesch. Functional topography during a visuoverbal judgement task studied with event-related desynchronization mapping. Journal of Clinical Neurophysiology, 9:120–131, 1992.

[38] John Polich. Comparison of auditory p300 habituation from active and passive con-ditions. International Journal of Psychophysiology, 17:25–34, 1994.

[39] P.R.Kennedy and R.A.E. Bakay. Restoration of neural output from a paralyzed patient by a direct brain connection. NeuroReport, 9:1707–1711, 1998.

BIBLIOGRAPHY 113

[40] P.R.Kennedy and R.A.E. Bakay. Direct control of a computer from the human central nervous system. Brain-Computer Interface Technology: Theory and Practice: First International Meeting Program and Papers, June 1999.

[41] Brechet R and Lecasble R. Reactivity of mu-rhythm to flicker. Electroencephelin Neurophysiol, 18:721–722, 1965.

[42] Jonathan R. and Wolpaw et al. Brain-computer interface research at the wadsworth center. IEEE Transactions on Rehabilitation Engineering, 8(2):222–226, 2000.

[43] Jonathan R. and Wolpaw et al. Brain-computer interface technology: A review of the first international meeting. IEEE Transations on Rehabilitation Engineering, 8(2):164V173, 2000.

[44] D. Regan. Human Brain Electrophysiology. New York: Elsevier, 1989.

[45] A. James Rowan and Eugene Tolunsky. Primer of EEG. Butterworth Heinemann, 2003.

[46] Richard HC Seabrook. The brain-computer interface: technique for controlling ma-chines. 00.

[47] Claire Murphy Spencer Wetter, John Polich. Olfactory, auditory, and visual erps from single trials: no evidence for habitation. International Journal of Psychophysiology, 54:263–272, 2004.

[48] E.E. Sutter. The visual evoked response as a communication channel. Proc. IEEE/NSF Symp, June 1984.

[49] E.E. Sutter. The brain response interface: Communication through visually-induced electrical brain responses. Microcomput. Appl., 15:31–45, 1992.

[50] David Martinus Johannes TAX. One-class classificaiton- concept learning in the absence of counter-examples. PhD thesis, 2001.

[51] George Townsend, Bernhard Graimann, and Gert Pfurtscheller. Continuous EEG classification during motor imagery -simulation of an asynchronous BCI. IEEE Tran-sation on Neual System and Rehabilitation Engineering, 12(2):258–265, June 2004.

[52] George Townsend, Bernhard Graimann, and G. Pfurtschellera. Continuous EEG clas-sification during motor imageryxsimulation of an asynchronous BCI. IEEE Transac-tions on Neural Systems and Rehabilitation Engineering, 12:258–265, 2004.

[53] F. van der Heijden, R.P.W. Duin, D. de Ridder, and D.M.J Tax. Classification, para-meter estimation and state estimation. John Wiley & Sons, Ltd., 2004.

[54] Juha Vesanto, Johan Himberg, Esa Alhoniemi, and Juha Parhankangas. Som toolbox for matlab5. 2000.

[55] J. R. Wolpaw and D. J. McFarland. Multichannel EEG-based brain-computer com-munication. Electroencephalogr. Clinical Neurophysiology, 90:444–449, 1994.

[56] J. R. Wolpaw, D. J. McFarland, G. W. Neat, and C. A. Forneris. An EEG-based brain-computer interface for corsur control. Electroencephalogr. Clinical Neurophysiology, 78:252–258, 1991.

[57] J.R. Wolpaw, N. Birbaumer, D.J. McFarland, G. Pfurtscheller, and T.M Vaughan.

Brainvcomputer interfaces for communication and control. Clinical Neurophysiology, 113(6):767–791, 2002.