Professional Documents
Culture Documents
net/publication/221712001
Article in Clinical EEG and neuroscience: official journal of the EEG and Clinical Neuroscience Society (ENCS) · January 2012
DOI: 10.1177/1550059411429528 · Source: PubMed
CITATIONS READS
21 1,030
5 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Yu-Chuan Jack Li on 01 September 2014.
Published by:
http://www.sagepublications.com
On behalf of:
Additional services and information for Clinical EEG and Neuroscience can be found at:
Subscriptions: http://eeg.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
What is This?
Abstract
This study proposed a recognized system for electroencephalogram (EEG) data classification. In addition to the wavelet-based
amplitude modulation (AM) features, the fuzzy c-means (FCM) clustering is used for the discriminant of left finger lifting and
resting. The features are extracted from discrete wavelet transform (DWT) data with the AM method. The FCM is then
applied to recognize extracted features. Compared with band power features, k-means clustering, and linear discriminant anal-
ysis (LDA) classifier, the results indicate that the proposed method is satisfactory in applications of brain–computer interface
(BCI).
Keywords
amplitude modulation, brain–computer interface, discrete wavelet transform, electroencephalography, fuzzy c-means (FCM)
Figure 1. Flowchart of proposed BCI system. The system mainly consists of feature extraction and FCM clustering.
Data Acquisition
The EEG signals were recorded from 7 untrained partici-
pants (6 males and 1 female, 3 left-handed and 4 right- Figure 2. Electroencephalogram (EEG) electrode placement
handed) in a shielded room using 7 silver/silver chloride according to International 10-20 system. All electrodes are referenced
electrodes. As illustrated in Figure 2, they included 5 scalp to the position A1 at the left earlobe.
EEG channels (C3, C5, FC3, C1, and CP3. These five
channels were selected since they can be further processed for finger lifting were acquired from 2 to 2 seconds, where
by surface Laplacian filtering as discussed in the next second 0 stands for the trigger of movement by detecting the
section. One EMG channel for monitoring right muscle peak EMG signal after linear envelope processing. (Only the
activity, and 1 channel on the forehead to record possible data recorded between 2 and 2 seconds were considered to
electro-oculogram (EOG) artifacts and eye blinks during be event related.) Data segments for finger resting were
the experiment.17 All electrodes were referenced to the randomly acquired from 4-second windows within the finger
A1 lead at the left earlobe. Before being sampled at the rate resting sections in each trial.
of 256 Hz, the EEG data were filtered by an analog band-
pass filter with cutoff frequencies at 0.5 Hz and 100 Hz, and
amplified by a multiple of 10 000. During the experiments,
each participant was asked to perform 2 trials in each test,
Preprocessing
which included left finger lifting and resting. Each trial was An appropriate filtering method can increase the signal-to-
10 seconds in length, so it took 20 seconds to complete a test. noise ratio by reducing non-EEG noise. The surface
In each lifting trial, the first 4 seconds were quiet and then an Laplacian filter is a simple but effective filtering method.18
acoustic stimulus was given as a cue to signify the beginning It calculates the second derivative of the spatial voltage
of left finger lifting. At the same time, each participant was distribution for a selected electrode. It is a high-pass spatial
asked to execute finger lifting. An example of a test is shown filter that enhances localized activities and reduces back-
in Figure 3A. We recorded 60 tests for each participant, and ground noise. This filter is achieved by subtracting the
thus there were 120 trials for each participant. No trials were average potential of a set of surrounding electrodes from the
removed during the EEG data processing stage. Data segments electrode of interest,
34 Clinical EEG and Neuroscience 43(1)
Figure 3. Preprocessing with surface Laplacian filtering. A, Raw EEG signals. B, Electroencephalograph signals posterior to the preprocessing.
(The first channel in (B) is the result of surface Laplacian filtering performed from the first 5 channels in (a); the last 2 channels in (A) and (B) are
the same, and they are EMG and EOG signals, respectively.) EMG indicates electromyography; EOG, electro-oculogram.
surrounding electrodes performed the surface Laplacian filtering sub-bands and quantify event-related oscillatory activities.20
is shown as follows: Each envelope or AM waveform was calculated as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Lap 1
VC3 ¼ VC3 ðVC5 þ VFC3 þ VC1 þ VCP3 Þ ð2Þ ASJ ðkÞ ¼ SJ ðkÞ2 þ HðSJ ðkÞÞ2
4 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
The 4-second window for left finger lifting was acquired ADj ðkÞ ¼ Dj ðkÞ2 þ HðDj ðkÞÞ2 ð5Þ
from the finger-lifting trial with its center standing for the trig- where HðxÞ was in the Hilbert transform of single-trial EEG
ger of movement, while the 4-second window for finger resting signal x.
was randomly acquired from the finger-resting trial.
Table 1 Comparison of Classification Accuracy Between Band Power Table 2. Comparison of AUC Between Band Power and Wavelet-
and Wavelet-Based AM Features Under the FCM Clustering Based AM Features Under the FCM Clustering
Abbreviations: AM, amplitude modulation; FCM, fuzzy c-means. Abbreviations: AUC, area under the receiver operating characteristics curve;
AM, amplitude modulation; FCM, fuzzy c-means.
We evaluate the performance of FCM clustering. Tables 3 3. Parra L, Alvino C, Tang AC, et al. Linear spatial integration for
and 4 list the comparisons of classification accuracy and single trial detection in encephalography. Neuroimage. 2002;
AUC under the same AM features among LDA, k-means, and 7(1):223–230.
FCM. The results show that the average accuracy for LDA is 4. Ahmadlou M, Adeli H. Fuzzy synchronization likelihood with
the worst, whereas that for FCM is the best. The average application to attention-deficit/hyperactivity disorder. Clin EEG
AUC for FCM is the best, while that for k-means is the Neurosci. 2011;42(1):6–13.
worst. Classification accuracy and AUC with FCM obtains 5. Hsu WY, Sun YN. EEG-based motor imagery analysis using
the best results for all participants. Accordingly, the results weighted wavelet transform features. J Neurosci Methods. 2009;
denote that the FCM is better than the LDA and k-means 167(2):310–318.
in EEG classification. 6. Muller-Putz GR, Zimmermann D, Graimann B, Nestinger K, Kori-
In addition, 2-way ANOVA and multiple comparison tests sek G, Pfurtscheller G. Event-related beta EEG-changes during pas-
determined whether these 3 classifiers are significantly dif- siveandattemptedfootmovementsinparaplegicpatients.BrainRes.
ferent. The results reveal that there are significant differences 2007;1137(1):84–91.
in both classification accuracy and AUC: ‘‘between LDA and 7. Hsu WY. Continuous EEG signal analysis for asynchronous BCI
FCM’’ (P values .0084 and .0105) and ‘‘between k-means application. Int J Neural Sys. 2011;21(4):335–350.
and FCM’’ (P values .0066 and .0096). Hence, the FCM is the 8. Obermaier B, Neuper C, Guger C, Pfurtscheller G. Information
best of these 3 classifiers in BCI applications. transfer rate in a five-classes brain-computer interface. IEEE
Trans Neural Sys Rehabil Eng. 2001;9(3):283–288.
9. Guger C, Edlinger G, Harkam W, Niedermayer I, Pfurtscheller G.
Conclusion How many people are able to operate an EEG-based brain-
computer interface (BCI)?. IEEE Trans Neural Sys Rehabil Eng.
We have proposed an analysis system for single-trial EEG
2003;11(2):145–147.
classification. The system that contains wavelet-based AM
10. Burke DP, Kelly SP, Chazal deP, Reilly RB, Finucane C. A
features and the FCM clustering is present for the discrimi-
parametric feature extraction and classification strategy for
nant of left finger lifting and resting. The features are
brain-computer interfacing. IEEE Trans Neural Sys Rehabil
extracted by AM method from the DWT data. They are good
Eng. 2005;13(1):12–17.
features that can improve the classification of mental tasks.
11. Hsu WY. EEG-based motor imagery classification using neuro-
The FCM is then used for the recognition of EEG data.
fuzzy prediction and wavelet fractal features. J Neurosci Methods.
Results indicate that the proposed methods are better in clas-
2010;189(2):295–302.
sification accuracy and AUC in comparison with band power
12. Hsu WY. EEG-based motor imagery classification using
features, k-means, and LDA classifier. We will develop some
enhanced active segment selection and adaptive classifier. Com-
more powerful features and classifiers to further enhance the
put Biol Med. 2011;41(8):633–639.
performance.
13. Kelly S, Burke D, Chazal deP, Reilly R. Parametric models and
spectral analysis for classification in brain-computer interfaces.
Acknowledgments Proc 14th Int Digital Signal Processing. 2002;1:307–310.
The author would like to express his sincere appreciation for grant 14. Ismail MA, Selim SZ. Fuzzy c-mean: optimality of solutions and
under shared facilities supported by the Program of Top 100 Univer- effective termination of the algorithm. Pattern Recogn. 1986;
sities Advancement, Ministry of Education, Taiwan. 19(6):481–485.
15. Hsu WY. Analytic differential approach for robust registration of
Declaration of Conflicting Interests rat brain histological images. Microscopy Res Tech. 2011;74(6):
The authors declared no potential conflicts of interest with respect to 523–530.
the research, authorship, and/or publication of this article. 16. Hsu WY, Kuo WF. Unsupervised fuzzy c-means clustering for
motor imagery EEG recognition. Int J Innov Comput Inf Control.
2011;7(8):4965–4976.
Funding
17. Jasper H. The ten-twenty electrode system of the international
The authors received no financial support for the research, authorship,
federation. Electroenceph Clin Neurophysiol. 1958;10(2):371–375.
and/or publication of this article.
18. McFarland DJ, McCane LM, David SV, Wolpaw JR. Spatial fil-
ter selection for EEG-based communication. Electroencepha-
References logr Clin Neurophysiol. 1997;103(3):386–394.
1. Nicolaou N, Georgiou J. The use of permutation entropy to 19. Daubechies I. Ten Lectures on Wavelets. CBMS-NSF Lecture
characterize sleep electroencephalograms. Clin EEG Neurosci. Notes nr. 1992; 61 SIAM.
2011;42(1):24–28. 20. Clochon P, Fontbonne JM, Lebrun N, Etevenon P. A new method for
2. Xu J, Zhao SZ, Zhang HS, Zheng CX. Decreased delta event- quantifyingEEGevent-relateddesynchronization:amplitudeenvel-
related synchronization in patients with early vascular dementia. ope analysis. Electroencephalogr Clin Neurophysiol. 1996;98(2):
Clin EEG Neurosci. 2011;42(1):53–58. 126–129.
38 Clinical EEG and Neuroscience 43(1)
21. Dunn JC. A fuzzy relative of the ISODATA process and its use in 24. Hsu WY, Lin CC, Ju MS, Sun YN. Wavelet-based fractal fea-
detecting compact well-separated clusters. J Cybern. 1974;3(3): tures with active segment selection: application to single-trial
32–57. EEG data. J Neurosci Methods. 2007;163(1):145–160.
22. Bezdek JC. Fuzzy Mathematics in Pattern Classification. Ph.D. 25. Hsu WY, Poon WFP, Sun YN. Automatic seamless mosaicing of
Dissertation, Applied Mathematics. New York, NY: Cornell microscopic images: enhancing appearance with color degrada-
University, Ithaca; 1973. tion compensation and wavelet-based blending. J Microscopy-
23. Demiralp T, Yordanova J, Kolev V, Ademoglu A, Devrim M, Oxford. 2008;231(3):408–418.
Samar VJ. Time-frequency analysis of single-sweep event- 26. Hanley JA, McNeil BJ. The meaning and use of the area under
related potentials by means of fast wavelet transform. Brain Lang. a receiver operating characteristic (ROC) curve. Radiology.
1999;66(1):129–145. 1982;143(1):29–36.