SSVEP-EEG Denoising Via Image Filtering Methods.
Wenqiang Yan,Chenghang Du,Yongcheng Wu,Xiaowei Zheng,Guanghua Xu
DOI: https://doi.org/10.1109/tnsre.2021.3104825
IF: 4.9
2021-01-01
IEEE Transactions on Neural Systems and Rehabilitation Engineering
Abstract:Steady-state visual evoked potential (SSVEP) is widely used in electroencephalogram (EEG) control, medical detection, cognitive neuroscience, and other fields. However, successful application requires improving the detection performance of SSVEP signal frequency characteristics. Most strategies to enhance the signal-to-noise ratio of SSVEP utilize application of a spatial filter. Here, we propose a method for image filtering denoising (IFD) of the SSVEP signal from the perspective of image analysis, as a preprocessing step for signal analysis. Arithmetic mean, geometric mean, Gaussian, and non-local means filtering methods were tested, and the experimental results show that image filtering of SSVEP cannot effectively remove the noise. Thus, we proposed a reverse solution in which the SSVEP noise signal was obtained by image filtering, and then the noise was subtracted from the original signal. Comparison of the recognition accuracy of the SSVEP signal before and after denoising was used to evaluate the denoising performance for stimuli of different duration. After IFD processing, SSVEP exhibited higher recognition accuracy, indicating the effectiveness of this proposed denoising method. Application of this method improves the detection performance of SSVEP signal frequency characteristics, combines image processing and brain signal analysis, and expands the research scope of brain signal analysis for widespread application.