International audienceChallenges for the next generation of Brain Computer Interfaces (BCI) are to mitigate the common sources of variability (electronic, electrical, biological) and to develop online and adaptive systems following the evolution of the subject's brain waves. Studying electroencephalographic (EEG) signals from their associated covariance matrices allows the construction of a representation which is invariant to extrinsic perturbations. As covariance matrices should be estimated, this paper first presents a thorough study of all estimators conducted on real EEG recording. Working in Euclidean space with covariance matrices is known to be error-prone, one might take advantage of algorithmic advances in Riemannian geometry and matrix manifold to implement methods for Symmetric Positive-Definite (SPD) matrices. Nonetheless, existing classification algorithms in Riemannian spaces are designed for offline analysis. We propose a novel algorithm for online and asynchronous processing of brain signals, borrowing principles from semi-unsupervised approaches and following a dynamic stopping scheme to provide a prediction as soon as possible. The assessment is conducted on real EEG recording: this is the first study on Steady-State Visually Evoked Potential (SSVEP) experimentations to exploit online classification based on Rie-mannian geometry. The proposed online algorithm is evaluated and compared with state-of-the-art SSVEP methods, which are based on Canonical Correlation Analysis (CCA). It is shown to improve both the classification accuracy and the information transfer rate in the online and asynchronous setup

Background: Neurofeedback (NF) training, as a method of self-regulation of brain activity, may be beneficial in elderly patients with mild cognitive impairment (MCI). In this pilot study, we investigated whether a sensorimotor (SMR)/theta NF training could improve cognitive performance and brain electrical activity in elderly patients with MCI. Methods: Twenty elderly patients with MCI were assigned to 20 consecutive sessions of sensorimotor (SMR)/theta NF training, during 10 weeks, on a basis of two sessions each week. Neuropsychological assessments and questionnaires, as well as electroencephalogram (EEG), were performed and compared between baseline (T0), after the last NF training session at 10 weeks (T1), and 1-month follow-up (T2). Results: Repeated measures ANOVA revealed that from baseline to post-intervention, participants showed significant improvement in the Montreal cognitive assessment (MoCa, F = 4.78; p = 0.012), the delayed recall of the Rey auditory verbal learning test (RAVLT, F = 3.675; p = 0.032), the Forward digit span (F = 13.82; p < 0.0001), the Anxiety Goldberg Scale (F = 4.54; p = 0.015), the Wechsler Adult Intelligence Score-Fourth Edition (WAIS-IV; F = 24.75; p < 0.0001), and the Mac Nair score (F = 4.47; p = 0.016). EEG theta power (F = 4.44; p = 0.016) and alpha power (F = 3.84; p = 0.027) during eyes-closed resting-state significantly increased after the NF training and showed sustained improvement at a 1-month follow-up.

Keeping a minimal number of channels is essential for designing a portable brain-computer interface system for daily usage. Most existing methods choose key channels based on spatial information without optimization of time segment for classification. This paper proposes a novel subject-specific channel selection method based on a criterion called F score to realize the parameterization of both time segment and channel positions. The F score is a novel simplified measure derived from Fisher's discriminant analysis for evaluating the discriminative power of a group of features. The experimental results on a standard dataset (BCI competition III dataset IVa) show that our method can efficiently reduce the number of channels (from 118 channels to 9 in average) without a decrease in mean classification accuracy. Compared to two state-ofthe-art methods in channel selection, our method leads to comparable or even better classification results with less selected channels.

a b s t r a c tThe essential task of a motor imagery brain-computer interface (BCI) is to extract the motor imageryrelated features from electroencephalogram (EEG) signals for classifying motor intentions. However, the optimal frequency band and time segment for extracting such features differ from subject to subject. In this work, we aim to improve the multi-class classification and to reduce the required EEG channel in motor imagery-based BCI by subject-specific time-frequency selection. Our method is based on a criterion namely Fisher discriminant analysis-type F-score to simultaneously select the optimal frequency band and time segment for multi-class classification. The proposed method uses only few Laplacian EEG channels (C3, Cz and C4) located around the sensorimotor area for classification. Applied to a standard multi-class BCI dataset (BCI competition III dataset IIIa), our method leads to better classification performance and smaller standard deviation across subjects compared to the state-of-art methods. Moreover, adding artifacts contaminated trials to the training dataset does not necessarily deteriorate our classification results, indicating that our method is tolerant to artifacts.

Background The first generation of brain-computer interfaces (BCI) classifies multi-channel electroencephalographic (EEG) signals, enhanced by optimized spatial filters. The second generation directly classifies covariance matrices estimated on EEG signals, based on straightforward algorithms such as the minimum-distance-to-Riemannian-mean (MDRM). Classification results vary greatly depending on the chosen Riemannian distance or divergence, whose definitions and reference implementations are spread across a wide mathematical literature.Methods This paper reviews all the Riemannian distances and divergences to process covariance matrices, with an implementation compatible with BCI constraints. The impact of using different metrics is assessed on a steady-state visually evoked potentials (SSVEP) dataset, evaluating centers of classes and classification accuracy.Results and Conclusions Riemannian approaches embed crucial properties to process EEG data. The Riemannian centers of classes outperform Euclidean ones both in offline and online setups. Some Riemannian