Single trial discrimination of individual finger movements on one hand: A combined MEG and EEG study

Abstract

It is crucial to understand what brain signals can be decoded from single trials with different recording techniques for the development of Brain-Machine Interfaces. A specific challenge for non-invasive recording methods are activations confined to small spatial areas on the cortex such as the finger representation of one hand. Here we study the information content of single trial brain activity in non-invasive MEG and EEG recordings elicited by finger movements of one hand. We investigate the feasibility of decoding which of four fingers of one hand performed a slight button press. With MEG we demonstrate reliable discrimination of single button presses performed with the thumb, the index, the middle or the little finger (average over all subjects and fingers 57%, best subject 70%, empirical guessing level: 25.1%). EEG decoding performance was less robust (average over all subjects and fingers 43%, best subject 54%, empirical guessing level 25.1%). Spatiotemporal patterns of amplitude variations in the time series provided best information for discriminating finger movements. Non-phase-locked changes of mu and beta oscillations were less predictive. Movement related high gamma oscillations were observed in average induced oscillation amplitudes in the MEG but did not provide sufficient information about the finger's identity in single trials. Importantly, pre-movement neuronal activity provided information about the preparation of the movement of a specific finger. Our study demonstrates the potential of non-invasive MEG to provide informative features for individual finger control in a Brain-Machine Interface neuroprosthesis.

Online Covariate Shift Detection-Based Adaptive Brain–Computer Interface to Trigger Hand Exoskeleton Feedback for Neuro-Rehabilitation

“…A combination of CSD test and supervised adaptation helps to build an online adaptive BCI system that aims to trigger a hand exoskeleton device to provide a neurofeedback for healthy volunteers, and stroke patients suffering from finger impairment. We have chosen this specific area of disability as EEG pattern recognition for finger movements is relatively challenging due to low signalto-noise ratio (SNR) and finger representative areas in the somatosensory cortex are spatially over-lapping [37]. Also, there is a need for undertaking such work because BCI systems requiring quick calibration and user adaptation in real-time are in high demand, while there is a real shortage of such studies on stroke patients [38].…”

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

“…MEG is regarded as having exquisite spatial and temporal resolution [1]. In previous work [10], we demonstrated greater discrimination of adjacent motor potentials in MEG data compared with simultaneously recorded EEG data. Furthermore, a study investigating the classification of visual ERPs [11] has provided evidence that discriminable data are more focused in the MEG than in the EEG.…”

Neurological Classifier Committee Based on Artificial Neural Networks and Support Vector Machine for Single-Trial EEG Signal Decoding

“…The obtained results were higher for the most of subjects (Subj. [1][2][3][4][5]7,8) in comparison with the decoding accuracies showed by the homogeneous committee of ANNs. Thus, the introduction of new types of classifiers into the committee of classifiers did not reduce the decoding accuracy; conversely, it increased by 8 % on average and by 12-17 % in several subjects, which was significant for multiclass classification.…”

“…This approach is also used by several researchers [5,15] for classification of finger movements. As we used ERP paradigm and short trials with some synchronized endogenous events, it is important to take into consideration time localization of features.…”

“…In the study [4] achieved average decoding accuracy of 5 finger movements was of 39,7 %. Quandt et al reported about decoding of 4 fingers movements with the average accuracy 43 %, maximal accuracies for several subjects were 46 % and 56 % for 4-and 5-class classification [5].…”