An Efficient Decoder for the Recognition of Event-Related Potentials in High-Density MEG Recordings

Christoph Reichert,Stefan Dürschmid,Hermann Hinrichs,Rudolf Kruse

doi:10.3390/computers5020005

Abstract

Brain–computer interfacing (BCI) is a promising technique for regaining communication and control in severely paralyzed people. Many BCI implementations are based on the recognition of task-specific event-related potentials (ERP) such as P300 responses. However, because of the high signal-to-noise ratio in noninvasive brain recordings, reliable detection of single trial ERPs is challenging. Furthermore, the relevant signal is often heterogeneously distributed over several channels. In this paper, we introduce a new approach for recognizing a sequence of attended events from multi-channel brain recordings. The framework utilizes spatial filtering to reduce both noise and signal space considerably. We introduce different models that can be used to construct the spatial filter and evaluate the approach using magnetoencephalography (MEG) data involving P300 responses, recorded during a BCI experiment. Compared to the accuracy achieved in the BCI experiment performed without spatial filtering, the recognition rate increased significantly to up to 95.3% on average (SD: 5.3%). In combination with the data-driven spatial filter construction we introduce here, our framework represents a powerful method to reliably recognize a sequence of brain potentials from high-density electrophysiological data, which could greatly improve the control of BCIs.

Highlights

People who have lost the capability to communicate with their environment due to severe paralysis could greatly benefit from a brain–computer interface (BCI) [1]
An appropriate combination of these channels enhances the signal strength and reduces the noise. Such channel weightings are known as spatial filter coefficients which were applied to analyze P300 potentials, taking varying approaches to estimating the coefficients, e.g., independent component analysis (ICA) [3], common spatial patterns (CSP) [4], spatial whitening [5], canonical correlation analysis (CCA) [6], and others [7,8]
We reduce the number of components, and the number of filters to be applied on test data by selecting only the first q filters that provide a significance of pk < 0.05 where at the same time the canonical correlation obtained from training data have to exceed a value of ρk > 0.1

Summary

Introduction

People who have lost the capability to communicate with their environment due to severe paralysis could greatly benefit from a brain–computer interface (BCI) [1] In such systems, communication can be realized by translating voluntarily modulated brain activity of a user into commands. An appropriate combination of these channels enhances the signal strength and reduces the noise Such channel weightings are known as spatial filter coefficients which were applied to analyze P300 potentials, taking varying approaches to estimating the coefficients, e.g., independent component analysis (ICA) [3], common spatial patterns (CSP) [4], spatial whitening [5], canonical correlation analysis (CCA) [6], and others [7,8]. When a high number of channels is available and no hypothetical selection of channels can be made, optimal spatial filtering is relevant

Methods

Results

Discussion

Conclusion