Enhancing Detection of SSVEPs through Spatial Filtering: An Inter-Trial Distance Minimization Perspective

Abstract

The brain-computer interface (BCI) technology has a great potential in providing more intelligent robots control for future disaster management systems. However, before brain-controlled robots can finally be brought to reality, many practical problems need to be solved. One of them is to further improve the detection performance of existing BCI systems. In steady-state visual-evoked potential (SSVEP) based BCIs, the selection of spatial filters poses a direct impact on the detection accuracy and the information transfer rate that can be achieved. To derive the optimum spatial filters, a popular approach is to maximize the inter-trial correlation over the training set. The relevant algorithms include task-related component analysis (TRCA), correlated component analysis (CORCA), sum of squared correlation analysis (SSCOR), etc. In this paper, a new perspective for calculating the spatial filters is proposed and it is named inter-trial distance minimization analysis (ITDMA). Literally, different from the conventional methods, the proposed ITDMA algorithm derives the spatial filters through minimizing the inter-trial distance over the training set. The detection performance of ITDMA is tested on a public benchmark SSVEP dataset and results show that the proposed ITDMA algorithm outperforms the three benchmark algorithms. The validity of ITDMA is verified.