Across-subject offline decoding of motor imagery from MEG and EEG

Hanna-Leena Halme,Lauri Parkkonen

doi:10.1038/s41598-018-28295-z

Abstract

Long calibration time hinders the feasibility of brain-computer interfaces (BCI). If other subjects’ data were used for training the classifier, BCI-based neurofeedback practice could start without the initial calibration. Here, we compare methods for inter-subject decoding of left- vs. right-hand motor imagery (MI) from MEG and EEG. Six methods were tested on data involving MEG and EEG measurements of healthy participants. Inter-subject decoders were trained on subjects showing good within-subject accuracy, and tested on all subjects, including poor performers. Three methods were based on Common Spatial Patterns (CSP), and three others on logistic regression with l1 - or l2,1 -norm regularization. The decoding accuracy was evaluated using (1) MI and (2) passive movements (PM) for training, separately for MEG and EEG. With MI training, the best accuracies across subjects (mean 70.6% for MEG, 67.7% for EEG) were obtained using multi-task learning (MTL) with logistic regression and l2,1-norm regularization. MEG yielded slightly better average accuracies than EEG. With PM training, none of the inter-subject methods yielded above chance level (58.7%) accuracy. In conclusion, MTL and training with other subject’s MI is efficient for inter-subject decoding of MI. Passive movements of other subjects are likely suboptimal for training the MI classifiers.

Highlights

Long calibration time hinders the feasibility of brain-computer interfaces (BCI)
As the online decoding was done with passive movements (PM) training, we assumed these results probably did not reflect the robustness of motor imagery (MI) or the ability to perform MI
We investigated methods for inter-subject classification of motor imagery from MEG and EEG signals

Summary

Introduction

Long calibration time hinders the feasibility of brain-computer interfaces (BCI). If other subjects’ data were used for training the classifier, BCI-based neurofeedback practice could start without the initial calibration. Hemiparetic patients can learn to control an orthosis attached to the paretic hand using motor imagery (MI) and concurrent feedback (see a review by Teo and co-workers[1]) This so-called neurofeedback training may enhance recovery of sensorimotor function compared to traditional physiotherapy, but the clinical evidence of BCI therapy still remains vague[2,3]. Due to large inter-subject variability of these signals, the user-specific brain responses are typically collected in the beginning of each session for calibrating the BCI and the total calibration time can be significant. This process can be exhausting for patients suffering from neurological disorders. A new BCI user can begin the neurofeedback practice immediately without the initial calibration session, which saves both time and effort of the patients

Methods

Results

Conclusion