Spatial covariance improves BCI performance for late ERPs components with high temporal variability.

OBJECTIVE Event Related Potentials (ERPs) reflecting cognitive response to external stimuli, are widely used in Brain Computer Interfaces (BCI). ERPs are characterized and typically decoded through a fixed set of components with particular amplitude and latency. However, the classical methods which rely on waveform features achieve a high decoding performance only with standardized and well aligned single trials. Since the amplitude and latency are sensitive to the experimental conditions, waveform features cannot be successfully applied for challenging tasks or to generalize across various experimental protocols. Features based on spatial covariances across channels can potentially overcome the latency jitter and delays since they aggregate the information across time. APPROACH We compared the performance stability of waveform and covariance-based features as well as their combination in simulated scenarios. We investigate two cases: 1) classifier transfer between 3 experiments with Error Related Potentials and 2) the performance robustness to the added latency jitter. MAIN RESULTS The features based on spatial covariances provide a stable performance with a minor decline under jitter levels of up to ± 300 ms, whereas the decoding performance with waveform features quickly drops from 0.85 to 0.55 AUC. The classifier transfer also resulted in a significantly more stable performance with covariance-based features. SIGNIFICANCE Our findings suggest that covariance-based features can be used to: 1) classify more reliably ERPs with higher intrinsic variability in more challenging real-life applications and 2) generalize across related experimental protocols.

[1]  Andrew P. Bradley,et al.  The use of the area under the ROC curve in the evaluation of machine learning algorithms , 1997, Pattern Recognit..

[2]  David G. Stork,et al.  Pattern Classification , 1973 .

[3]  M. Berger.  A Panoramic View of Riemannian Geometry , 2003 .

[4]  G. Pfurtscheller,et al.  Optimal spatial filtering of single trial EEG during imagined hand movement. , 2000, IEEE transactions on rehabilitation engineering : a publication of the IEEE Engineering in Medicine and Biology Society.

[5]  Alexandre Barachant,et al.  A New Generation of Brain-Computer Interface Based on Riemannian Geometry , 2013, ArXiv.

[6]  Lucas C. Parra,et al.  Recipes for the linear analysis of EEG , 2005, NeuroImage.

[7]  G. McCarthy,et al.  Augmenting mental chronometry: the P300 as a measure of stimulus evaluation time. , 1977, Science.

[8]  Nicholas Ayache,et al.  Geometric Means in a Novel Vector Space Structure on Symmetric Positive-Definite Matrices , 2007, SIAM J. Matrix Anal. Appl..

[9]  R Chavarriaga,et al.  Latency correction of event-related potentials between different experimental protocols. , 2014, Journal of neural engineering.

[10]  Christian Jutten,et al.  Classification of covariance matrices using a Riemannian-based kernel for BCI applications , 2013, Neurocomputing.

[11]  E. Oja,et al.  Independent Component Analysis , 2013 .

[12]  C. C. Duncan,et al.  Event-related potentials in clinical research: Guidelines for eliciting, recording, and quantifying mismatch negativity, P300, and N400 , 2009, Clinical Neurophysiology.

[13]  Wei Wu,et al.  A Novel Algorithm for Learning Sparse Spatio-Spectral Patterns for Event-Related Potentials , 2017, IEEE Transactions on Neural Networks and Learning Systems.

[14]  Alexandre Barachant,et al.  A Plug&Play P300 BCI Using Information Geometry , 2014, ArXiv.

[15]  Christian Jutten,et al.  Multiclass Brain–Computer Interface Classification by Riemannian Geometry , 2012, IEEE Transactions on Biomedical Engineering.

[16]  Michael S. Lazar,et al.  Spatial patterns underlying population differences in the background EEG , 2005, Brain Topography.

[17]  Alfred O. Hero,et al.  Shrinkage Algorithms for MMSE Covariance Estimation , 2009, IEEE Transactions on Signal Processing.

[18]  Wei Wu,et al.  RSTFC: A Novel Algorithm for Spatio-Temporal Filtering and Classification of Single-Trial EEG , 2015, IEEE Transactions on Neural Networks and Learning Systems.

[19]  Adam Wilson,et al.  A Procedure for Measuring Latencies in Brain–Computer Interfaces , 2010, IEEE Transactions on Biomedical Engineering.

[20]  Robert Tibshirani,et al.  The Elements of Statistical Learning: Data Mining, Inference, and Prediction, 2nd Edition , 2001, Springer Series in Statistics.

[21]  Kazuyuki Aihara,et al.  Classifying matrices with a spectral regularization , 2007, ICML '07.

[22]  Wei Wu,et al.  Learning event-related potentials (ERPs) from multichannel EEG recordings: A spatio-temporal modeling framework with a fast estimation algorithm , 2011, 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[23]  Stefan Haufe,et al.  Single-trial analysis and classification of ERP components — A tutorial , 2011, NeuroImage.

[24]  Ricardo Chavarriaga,et al.  Errare machinale est: the use of error-related potentials in brain-machine interfaces , 2014, Front. Neurosci..

[25]  Christian Jutten,et al.  Common Spatial Pattern revisited by Riemannian geometry , 2010, 2010 IEEE International Workshop on Multimedia Signal Processing.

[26]  Xingyu Wang,et al.  Spatial-Temporal Discriminant Analysis for ERP-Based Brain-Computer Interface , 2013, IEEE Transactions on Neural Systems and Rehabilitation Engineering.

[27]  Christian Jutten,et al.  Transfer Learning: A Riemannian Geometry Framework With Applications to Brain–Computer Interfaces , 2018, IEEE Transactions on Biomedical Engineering.