結合高斯混合模型與最大期望值方法於相位編碼視覺腦波人機介面之目標偵測

Detection of gazed target in an phase-tagged SSVEP-based BCI using the combination of Gaussian Mixture Model and EM method

張立昂(Li-Ang Chang)

指導教授 ： 李柏磊

國立中央大學/資訊電機學院/電機工程學系/碩士(2012年)

摘要

近年來，許多研究針對穩態視覺誘發電位(Stead-State Visual Evoked Potential, SSVEP)之大腦人機介面(Brain Computer Interface, BCI)系統進行設計，其中一種引人注目的系統，使用高頻率的相位編碼閃光技術，稱為相位編碼穩態視覺腦波人機介面。為了能改善相位編碼穩態視覺腦波人機介面在辨識方法上的速度與正確率，本研究提出使用高斯混合模型(Gaussian Mixture Model, GMM)與最大期望演算法(Expected-Maximization Algorithm, EM Algorithm)來達到目標。
本研究針對六名年齡介於22到27歲的受測者進行實驗。使用22Hz單一頻率配合四種相位閃光進行測試。使用者Oz得到的腦電波先經過濾波器濾波，接著依照觸發事件進行切割，再使用傅立葉轉換計算閃光頻率的振幅與相位並轉換為角度。演算法訓練過程利用高斯混合模型(GMM)與最大期望演算法(EM Algorithm)來建立最適合使用者的分類器；實際使用時，則利用柯爾莫諾夫-斯米爾諾夫檢定(Kolmogorov-Smirnov test, KS test) 進行有效資料區段選擇，如果通過檢定，則放入高斯混合模型中判斷使用者所注視的選項。本系統的實驗結果顯示，六位受測者使用本系統的正確率為92.45±4.36%，下達指令時間為1.17±0.42s秒/指令。

關鍵字

穩態視覺誘發電位 ； 高斯混合模型 ； 最大期望值法 ； 大腦人機介面

並列摘要

In recent year, several research groups have proposed novel techniques to improve the performance of Steady-State Visual Evoked Potential based Brain Computer Interface (SSVEP based BCI). One SSVEP-based system, using high-frequency phase-tagged flickers, denoted as phase-tagged SSVEP-based BCI, has drawn great attention. In the present study, we utilize the combination of Gaussian Mixture Model (GMM) and Expected Maximization Algorithm (EM Algorithm) to improve the speed and accuracy in phase-tagged SSVEP-based BCI system.
Six subjects, aged from 22 to 27 years, were recrited in our study. Four visual flickers, flashing at 22 Hz, with phase tagged at 0°, 90°, 180°, and 270°, were utilized as visual stimuli to induce SSVEPs. EEG signals recorded from Oz channel were filtered and segmented into epochs. The phase and amplitude information of each epoch were computed by means of Fourier method. In the system training process, we have to use GMM and EM Algorithm to establish customized classifier for each subject. In the system practice process, effective epochs were detected using Kolmogorov-Smirnov test (KS test), and only those effective epochs were subjected to the following gaze-target detections. Our results showed, the accuracies of six subjects were between the percent of 87% to 97% and the command transfer rates were between 0.68 to 1.74 second per command.

並列關鍵字

BCI ； SSVEP ； Gaussian mixture model ； EM algorithm

參考文獻

[21] 謝俊傑，多頻相位編碼之閃光視覺誘發電位驅動大腦人機介面，國立中央大學，碩士論文，2007。

[1] M. Middendorf, G. McMillan, G. Calhoun, and K. S. Jones, “Brain–Computer Interfaces Based on the Steady-State Visual-Evoked Response”, IEEE Transactions on Rehabilitation Engineering, Vol. 8, No. 2, June 2000.

[3] G. R. Müller-Putz*, Member, IEEE, and G. Pfurtscheller, Member, IEEE,” Control of an Electrical Prosthesis With an SSVEP-Based BCI”, IEEE Transactions on Biomedical Engineering, Vol. 55, No. 1, January 2008.

[4] M. A. Lopez-Gordo • A. Prieto • F. Pelayo • C. Morillas,” Use of Phase in Brain–Computer Interfaces based on Steady-State Visual Evoked Potentials”, Neural Process Lett (2010) 32:1–9 DOI 10.1007/s11063-010-9139-8.

[5] G. Pfurtscheller, C. Neuper, C. Guger, W. Harkam, H. Ramoser, A. Schlogl, B. Obermaier and M. Pregenzer, “Current Trends in Graz Brain-Computer Interface (BCI) Research,” IEEE Trans. Rehabil. Eng., Vol. 8, No. 2, June, 2000.