Detection of Error-Related Potentials in Stroke Patients from EEG Using an Artificial Neural Network.

Nayab Usama,Mads Jochumsen,Kim Dremstrup,Imran Khan Niazi

doi:10.3390/s21186274

Abstract

Error-related potentials (ErrPs) have been proposed as a means for improving brain–computer interface (BCI) performance by either correcting an incorrect action performed by the BCI or label data for continuous adaptation of the BCI to improve the performance. The latter approach could be relevant within stroke rehabilitation where BCI calibration time could be minimized by using a generalized classifier that is continuously being individualized throughout the rehabilitation session. This may be achieved if data are correctly labelled. Therefore, the aims of this study were: (1) classify single-trial ErrPs produced by individuals with stroke, (2) investigate test–retest reliability, and (3) compare different classifier calibration schemes with different classification methods (artificial neural network, ANN, and linear discriminant analysis, LDA) with waveform features as input for meaningful physiological interpretability. Twenty-five individuals with stroke operated a sham BCI on two separate days where they attempted to perform a movement after which they received feedback (error/correct) while continuous EEG was recorded. The EEG was divided into epochs: ErrPs and NonErrPs. The epochs were classified with a multi-layer perceptron ANN based on temporal features or the entire epoch. Additionally, the features were classified with shrinkage LDA. The features were waveforms of the ErrPs and NonErrPs from the sensorimotor cortex to improve the explainability and interpretation of the output of the classifiers. Three calibration schemes were tested: within-day, between-day, and across-participant. Using within-day calibration, 90% of the data were correctly classified with the entire epoch as input to the ANN; it decreased to 86% and 69% when using temporal features as input to ANN and LDA, respectively. There was poor test–retest reliability between the two days, and the other calibration schemes led to accuracies in the range of 63–72% with LDA performing the best. There was no association between the individuals’ impairment level and classification accuracies. The results show that ErrPs can be classified in individuals with stroke, but that user- and session-specific calibration is needed for optimal ErrP decoding with this approach. The use of ErrP/NonErrP waveform features makes it possible to have a physiological meaningful interpretation of the output of the classifiers. The results may have implications for labelling data continuously in BCIs for stroke rehabilitation and thus potentially improve the BCI performance.

Highlights

Licensee MDPI, Basel, Switzerland.A brain–computer interface (BCI) is a means for people with motor impairments to control external devices using only brain activity [1]
User-specific classifiers based on within-day calibration data generally perform better, but they suffer from changes in the brain activity that cause an inadequate representation with respect to the distribution of brain activity in the calibration data
A way to overcome this and optimize the performance of a BCI could be through the detection of error-related potentials (ErrPs), where the detection of this signal can be used for error correction or labelling data [9]

Summary

Introduction

Licensee MDPI, Basel, Switzerland.A brain–computer interface (BCI) is a means for people with motor impairments to control external devices using only brain activity [1]. Different approaches have been proposed in the literature to use a generalized classifier where the BCI works without the need of individualized training data [4,5,6,7]. User-specific classifiers based on within-day calibration data generally perform better, but they suffer from changes in the brain activity that cause an inadequate representation with respect to the distribution of brain activity in the calibration data. These changes may be due to shifts in attention and fatigue, which may be pronounced in people with neurological diseases such as stroke [8]. Error correction and labelling of data would work in different scenarios, e.g., error correction can be used for automatically deleting mistyped letters in a speller application or reverting the movement of a robotic arm, i.e., the application would be applicable for communication and control purposes

Objectives

Methods

Results

Discussion

Conclusion