An Orthonormalized Partial Least Squares Based Spatial Filter for SSVEP Extraction
In this study, a novel orthonormalized partial least squares (OPLS) spatial filter is proposed for the extraction of the steady-state visual evoked potential (SSVEP) components buried in the electroencephalogram (EEG) data. The proposed method avoids over-fitting of the EEG data to the ideal SSVEP reference signals by reducing the over-emphasis of the target (pure sine-cosine) space. The paper presents the comparison of the detection accuracy of the proposed method with other existing spatial filters and discusses the shortcomings of these algorithms. The OPLS was tested across ten healthy subjects and its classification performance was examined. Further, statistical tests were performed to show the significant improvements in obtained detection accuracies. The result shows that the OPLS provides a significant improvement in detection accuracy across subjects compared to spatial filters under comparison. Hence, OPLS would act as a reliable and efficient spatial filter for separation of SSVEP components in brain-computer interface (BCI) applications.
KeywordsSteady-state visual evoked potential (SSVEP) Electroencephalogram (EEG) Brain-computer interface (BCI) Orthonormalized partial least squares (OPLS)
- 7.Garcia-Molina, G., Zhu, D.: Optimal spatial filtering for the steady state visual evoked potential: BCI application. In: 5th International IEEE/EMBS Conference on Neural Engineering, pp. 156–160 (2011). https://doi.org/10.1109/NER.2011.5910512
- 10.Japkowicz, N., Shah, M.: Evaluating Learning Algorithms: A Classification Perspective. Cambridge University Press, New York (2011). https://doi.org/10.1017/CBO9780511921803
- 11.Lalor, E.C., et al.: Steady-state VEP-based brain-computer interface control in an immersive 3D gaming environment. EURASIP J. Adv. Signal Process. 2005, 3156–3164 (2005)Google Scholar
- 13.Nakanishi, M., Wang, Y., Chen, X., Wang, Y.T., Gao, X., Jung, T.P.: Enhancing detection of SSVEPs for a high-speed brain speller using task-related component analysis 9294, 1 (2017)Google Scholar