A Regional Smoothing Block Sparse Bayesian Learning Method With Temporal Correlation for Channel Selection in P300 Speller

Abstract

The P300-based brain-computer interfaces (BCIs) enable participants to communicate by decoding the electroencephalography (EEG) signal. Different regions of the brain correspond to various mental activities. Therefore, removing weak task-relevant and noisy channels through channel selection is necessary when decoding a specific type of activity from EEG. It can improve the recognition accuracy and reduce the training time of the subsequent models. This study proposes a novel block sparse Bayesian-based channel selection method for the P300 speller. In this method, we introduce block sparse Bayesian learning (BSBL) into the channel selection of P300 BCI for the first time and propose a regional smoothing BSBL (RSBSBL) by combining the spatial distribution properties of EEG. The RSBSBL can determine the number of channels adaptively. To ensure practicality, we design an automatic selection iteration strategy model to reduce the time cost caused by the inverse operation of the large-size matrix. We verified the proposed method on two public P300 datasets and on our collected datasets. The experimental results show that the proposed method can remove the inferior channels and work with the classifier to obtain high-classification accuracy. Hence, RSBSBL has tremendous potential for channel selection in P300 tasks.

Keywords: EEG; P300; brain-computer interface; channel selection; sparse bayesian learning; temporal correlation.

FIGURE 1

Region division. Channels belonging to a region are circled with the dotted line. The left subfigure shows the division for DS1 and DS2, while the right subfigure shows the division for DS3.

FIGURE 2

Parameter relationship graphical model in a single iteration. Parameters of the same color can be iterated simultaneously.

FIGURE 3

Diagram of the data processing framework, including pre-processing, channel selection, and classification. Using the block sparsity property of RSBSBL, we do pruning on the eligible channels by fitting the training data and labels.

FIGURE 4

The average recognition accuracy of the four methods on DS1, DS2, and DS3 when M channels are selected, where M = [4, 8, 12, 16]. The error bars are the standard deviations for DS2 and DS3.

FIGURE 5

The scalp distribution of the four methods on DS1, DS2, and DS3 when M channels are selected. The contribution value of each channel is equal to the sum of the selected numbers among all participants in the dataset. The color changes from red to blue, indicating that the channel is selected less often.

FIGURE 6

The effect of shear threshold τ in RSBSBL on the number of selected channels and accuracy. The x-axis indicates the number of selected channels, the y-axis indicates the value of τ, and the z-axis indicates the participant ID. The color of the sphere represents the normalized recognition accuracy for each participant with different thresholds.

FIGURE 7

Changes in the run-time (s) of matrix inversion when the size of the matrix increases. In the left part, the horizontal axis represents the size of the square array. The vertical axis is the value after taking the logarithm of the time. The bar chart represents the average time cost of the proposed method on the three datasets.