Defining and quantifying users’ mental imagery-based BCI skills: a first step

Lotte, Fabien; Jeunet, Camille

doi:10.1088/1741-2552/aac577

Cited by 53 publications

(62 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Much more common and normally associated with higher accuracies is classification based on left vs. right hand MI or hand vs. foot MI (for a review, see [16]). The difficulty of interpreting classification accuracies has been already brought up by Lotte and Jeunet [49] emphasizing that classification accuracies should be handled carefully when quantifying user BCI/NF performance and should be complemented with other metrics.…”

Section: Implications For MI Nf Interventionsmentioning

confidence: 99%

Challenge Accepted? Individual Performance Gains for Motor Imagery Practice with Humanoid Robotic EEG Neurofeedback

Daeglau

Wallhoff

Debener

et al. 2020

Sensors

View full text Add to dashboard Cite

Optimizing neurofeedback (NF) and brain–computer interface (BCI) implementations constitutes a challenge across many fields and has so far been addressed by, among others, advancing signal processing methods or predicting the user’s control ability from neurophysiological or psychological measures. In comparison, how context factors influence NF/BCI performance is largely unexplored. We here investigate whether a competitive multi-user condition leads to better NF/BCI performance than a single-user condition. We implemented a foot motor imagery (MI) NF with mobile electroencephalography (EEG). Twenty-five healthy, young participants steered a humanoid robot in a single-user condition and in a competitive multi-user race condition using a second humanoid robot and a pseudo competitor. NF was based on 8–30 Hz relative event-related desynchronization (ERD) over sensorimotor areas. There was no significant difference between the ERD during the competitive multi-user condition and the single-user condition but considerable inter-individual differences regarding which condition yielded a stronger ERD. Notably, the stronger condition could be predicted from the participants’ MI-induced ERD obtained before the NF blocks. Our findings may contribute to enhance the performance of NF/BCI implementations and highlight the necessity of individualizing context factors.

show abstract

Section: Implications For MI Nf Interventionsmentioning

confidence: 99%

Challenge Accepted? Individual Performance Gains for Motor Imagery Practice with Humanoid Robotic EEG Neurofeedback

Daeglau

Wallhoff

Debener

et al. 2020

Sensors

View full text Add to dashboard Cite

show abstract

“…Lotte and Jeunet [19] proposed run wise cross validation as an efficient metric for offline classification algorithms. The cross-validation strategy used in this paper is similar to the metric proposed in [19] and hence, further reinforces the effectiveness of our proposed framework. The optimum regularization parameter λ is data dependent and subsequently subject dependent.…”

Section: Resultsmentioning

confidence: 99%

Formulating Divergence Framework for Multiclass Motor Imagery EEG Brain Computer Interface

Kumar

Reddy

Arora

et al. 2020

ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Similar to most of the real world data, the ubiquitous presence of non-stationarities in the EEG signals significantly perturb the feature distribution thus deteriorating the performance of Brain Computer Interface. In this letter, a novel method is proposed based on Joint Approximate Diagonalization (JAD) to optimize stationarity for multiclass motor imagery Brain Computer Interface (BCI) in an information theoretic framework. Specifically, in the proposed method, we estimate the subspace which optimizes the discriminability between the classes and simultaneously preserve stationarity within the motor imagery classes. We determine the subspace for the proposed approach through optimization using gradient descent on an orthogonal manifold. The performance of the proposed stationarity enforcing algorithm is compared to that of baseline One-Versus-Rest (OVR)-CSP and JAD on publicly available BCI competition IV dataset IIa. Results show that an improvement in average classification accuracies across the subjects over the baseline algorithms and thus essence of alleviating within session nonstationarities.

show abstract

“…The use of other types of mental tasks in BCI studies has received little attention in the literature. The papers that have studied non-motor imagery mental tasks along with the motor imagery tasks include [65][66][67][68][69][70][71][72][73][74][75][76][77][78][79][80][81][82][83][84]. Studies [85][86][87][88][89][90][91][92][93][94][95][96] have only considered non-motor imagery mental tasks.…”

Section: Introductionmentioning

confidence: 99%

“…Hand movements and word imagination are the mental tasks used in [83]. Imagination of left and right hand movements, mental rotation of a 3D geometric figure, and mental subtraction of a two-digit number from a three-digit number are considered in [84].…”

Section: Introductionmentioning

confidence: 99%

A Self-Paced Two-State Mental Task-Based Brain-Computer Interface with Few EEG Channels

Faradji¹,

Ward²,

Birch³

2020

New Frontiers in Brain - Computer Interfaces

View full text Add to dashboard Cite

A self-paced brain-computer interface (BCI) system that is activated by mental tasks is introduced. The BCI's output has two operational states, the active state and the inactive state, and is activated by designated mental tasks performed by the user. The BCI could be operated using several EEG brain electrodes (channels) or only few (i.e., five or seven channels) at a small loss in performance. The performance is evaluated on a dataset we have collected from four subjects while performing one of the four different mental tasks. The dataset contains the signals of 29 EEG electrodes distributed over the scalp. The five and seven highly discriminatory channels are selected using two different methods proposed in the paper. The signal processing structure of the interface is computationally simple. The features used are the scalar autoregressive coefficients. Classification is based on the quadratic discriminant analysis. Model selection and testing procedures are accomplished via cross-validation. The results are highly promising in terms of the rates of false and true positives. The false-positive rates reach zero, while the true-positive rates are sufficiently high, i.e., 54.60 and 59.98% for the 5-channel and 7-channel systems, respectively.

show abstract

Defining and quantifying users’ mental imagery-based BCI skills: a first step

Cited by 53 publications

References 50 publications

Challenge Accepted? Individual Performance Gains for Motor Imagery Practice with Humanoid Robotic EEG Neurofeedback

Challenge Accepted? Individual Performance Gains for Motor Imagery Practice with Humanoid Robotic EEG Neurofeedback

Formulating Divergence Framework for Multiclass Motor Imagery EEG Brain Computer Interface

A Self-Paced Two-State Mental Task-Based Brain-Computer Interface with Few EEG Channels

Contact Info

Product

Resources

About