Long-Term BCI Training of a Tetraplegic User: Adaptive Riemannian Classifiers and User Training

Several inaccurate or erroneous conceptions and misleading propaganda about brain-computer interfaces

Brain-computer interface (BCI) is a revolutionizing human-computer interaction, which has potential applications for specific individuals or groups in specific scenarios. Extensive research has been conducted on the principles and implementation methods of BCI, and efforts are currently being made to bridge the gap from research to real-world applications. However, there are inaccurate or erroneous conceptions about BCI among some members of the public, and certain media outlets, as well as some BCI researchers, developers, manufacturers, and regulators, propagate misleading or overhyped claims about BCI technology. Therefore, this article summarizes the several misconceptions and misleading propaganda about BCI, including BCI being capable of "mind-controlled," "controlling brain," "mind reading," and the ability to "download" or "upload" information from or to the brain using BCI, among others. Finally, the limitations (shortcomings) and limits (boundaries) of BCI, as well as the necessity of conducting research aimed at countering BCI systems are discussed, and several suggestions are offered to reduce misconceptions and misleading claims about BCI.

Transfer learning promotes acquisition of individual BCI skills

Subject training is crucial for acquiring brain-computer interface (BCI) control. Typically, this requires collecting user-specific calibration data due to high inter-subject neural variability that limits the usability of generic decoders. However, calibration is cumbersome and may produce inadequate data for building decoders, especially with naïve subjects. Here, we show that a decoder trained on the data of a single expert is readily transferrable to inexperienced users via domain adaptation techniques allowing calibration-free BCI training. We introduce two real-time frameworks, (i) Generic Recentering (GR) through unsupervised adaptation and (ii) Personally Assisted Recentering (PAR) that extends GR by employing supervised recalibration of the decoder parameters. We evaluated our frameworks on 18 healthy naïve subjects over five online sessions, who operated a customary synchronous bar task with continuous feedback and a more challenging car racing game with asynchronous control and discrete feedback. We show that along with improved task-oriented BCI performance in both tasks, our frameworks promoted subjects' ability to acquire individual BCI skills, as the initial neurophysiological control features of an expert subject evolved and became subject specific. Furthermore, those features were task-specific and were learned in parallel as participants practiced the two tasks in every session. Contrary to previous findings implying that supervised methods lead to improved online BCI control, we observed that longitudinal training coupled with unsupervised domain matching (GR) achieved similar performance to supervised recalibration (PAR). Therefore, our presented frameworks facilitate calibration-free BCIs and have immediate implications for broader populations-such as patients with neurological pathologies-who might struggle to provide suitable initial calibration data.

Beta bursts question the ruling power for brain-computer interfaces

Objective: Current efforts to build reliable brain-computer interfaces (BCI) span multiple axes from hardware, to software, to more sophisticated experimental protocols, and personalized approaches. However, despite these abundant efforts, there is still room for significant improvement. We argue that a rather overlooked direction lies in linking BCI protocols with recent advances in fundamental neuroscience.Approach: In light of these advances, and particularly the characterization of the burst-like nature of beta frequency band activity and the diversity of beta bursts, we revisit the role of beta activity in 'left vs. right hand' motor imagery (MI) tasks. Current decoding approaches for such tasks take advantage of the fact that MI generates time-locked changes in induced power in the sensorimotor cortex and rely on band-passed power changes in single or multiple channels. Although little is known about the dynamics of beta burst activity during MI, we hypothesized that beta bursts should be modulated in a way analogous to their activity during performance of real upper limb movements.Main results and Significance: We show that classification features based on patterns of beta burst modulations yield decoding results that are equivalent to or better than typically used beta power across multiple open electroencephalography datasets, thus providing insights into the specificity of these bio-markers.

Towards user-centric BCI design: Markov chain-based user assessment for mental imagery EEG-BCIs

Objective.While electroencephalography (EEG)-based brain-computer interfaces (BCIs) have many potential clinical applications, their use is impeded by poor performance for many users. To improve BCI performance, either via enhanced signal processing or user training, it is critical to understand and describe each user's ability to perform mental control tasks and produce discernible EEG patterns. While classification accuracy has predominantly been used to assess user performance, limitations and criticisms of this approach have emerged, thus prompting the need to develop novel user assessment approaches with greater descriptive capability. Here, we propose a combination of unsupervised clustering and Markov chain models to assess and describe user skill.Approach.Using unsupervisedK-means clustering, we segmented the EEG signal space into regions representing pattern states that users could produce. A user's movement through these pattern states while performing different tasks was modeled using Markov chains. Finally, using the steady-state distributions and entropy rates of the Markov chains, we proposed two metricstaskDistinctandrelativeTaskInconsistencyto assess, respectively, a user's ability to (i) produce distinct task-specific patterns for each mental task and (ii) maintain consistent patterns during individual tasks.Main results.Analysis of data from 14 adolescents using a three-class BCI revealed significant correlations between thetaskDistinctandrelativeTaskInconsistencymetrics and classification F1 score. Moreover, analysis of the pattern states and Markov chain models yielded descriptive information regarding user performance not immediately apparent from classification accuracy.Significance.Our proposed user assessment method can be used in concert with classifier-based analysis to further understand the extent to which users produce task-specific, time-evolving EEG patterns. In turn, this information could be used to enhance user training or classifier design.

Transfer Learning in Trajectory Decoding: Sensor or Source Space?

In this study, across-participant and across-session transfer learning was investigated to minimize the calibration time of the brain-computer interface (BCI) system in the context of continuous hand trajectory decoding. We reanalyzed data from a study with 10 able-bodied participants across three sessions. A leave-one-participant-out (LOPO) model was utilized as a starting model. Recursive exponentially weighted partial least squares regression (REW-PLS) was employed to overcome the memory limitation due to the large pool of training data. We considered four scenarios: generalized with no update (Gen), generalized with cumulative update (GenC), and individual models with cumulative (IndC) and non-cumulative (Ind) updates, with each one trained with sensor-space features or source-space features. The decoding performance in generalized models (Gen and GenC) was lower than the chance level. In individual models, the cumulative update (IndC) showed no significant improvement over the non-cumulative model (Ind). The performance showed the decoder's incapability to generalize across participants and sessions in this task. The results suggested that the best correlation could be achieved with the sensor-space individual model, despite additional anatomical information in the source-space features. The decoding pattern showed a more localized pattern around the precuneus over three sessions in Ind models.

Impact of dataset size and long-term ECoG-based BCI usage on deep learning decoders performance

Introduction

In brain-computer interfaces (BCI) research, recording data is time-consuming and expensive, which limits access to big datasets. This may influence the BCI system performance as machine learning methods depend strongly on the training dataset size. Important questions arise: taking into account neuronal signal characteristics (e.g., non-stationarity), can we achieve higher decoding performance with more data to train decoders? What is the perspective for further improvement with time in the case of long-term BCI studies? In this study, we investigated the impact of long-term recordings on motor imagery decoding from two main perspectives: model requirements regarding dataset size and potential for patient adaptation.

Methods

We evaluated the multilinear model and two deep learning (DL) models on a long-term BCI & Tetraplegia (ClinicalTrials.gov identifier: NCT02550522) clinical trial dataset containing 43 sessions of ECoG recordings performed with a tetraplegic patient. In the experiment, a participant executed 3D virtual hand translation using motor imagery patterns. We designed multiple computational experiments in which training datasets were increased or translated to investigate the relationship between models' performance and different factors influencing recordings.

Results

Our results showed that DL decoders showed similar requirements regarding the dataset size compared to the multilinear model while demonstrating higher decoding performance. Moreover, high decoding performance was obtained with relatively small datasets recorded later in the experiment, suggesting motor imagery patterns improvement and patient adaptation during the long-term experiment. Finally, we proposed UMAP embeddings and local intrinsic dimensionality as a way to visualize the data and potentially evaluate data quality.

Discussion

DL-based decoding is a prospective approach in BCI which may be efficiently applied with real-life dataset size. Patient-decoder co-adaptation is an important factor to consider in long-term clinical BCI.

Riemannian geometry-based metrics to measure and reinforce user performance changes during brain-computer interface user training

Despite growing interest and research into brain-computer interfaces (BCI), their usage remains limited outside of research laboratories. One reason for this is BCI inefficiency, the phenomenon where a significant number of potential users are unable to produce machine-discernible brain signal patterns to control the devices. To reduce the prevalence of BCI inefficiency, some have advocated for novel user-training protocols that enable users to more effectively modulate their neural activity. Important considerations for the design of these protocols are the assessment measures that are used for evaluating user performance and for providing feedback that guides skill acquisition. Herein, we present three trial-wise adaptations (running, sliding window and weighted average) of Riemannian geometry-based user-performance metrics (classDistinct reflecting the degree of class separability and classStability reflecting the level of within-class consistency) to enable feedback to the user following each individual trial. We evaluated these metrics, along with conventional classifier feedback, using simulated and previously recorded sensorimotor rhythm-BCI data to assess their correlation with and discrimination of broader trends in user performance. Analysis revealed that the sliding window and weighted average variants of our proposed trial-wise Riemannian geometry-based metrics more accurately reflected performance changes during BCI sessions compared to conventional classifier output. The results indicate the metrics are a viable method for evaluating and tracking user performance changes during BCI-user training and, therefore, further investigation into how these metrics may be presented to users during training is warranted.

Open multi-session and multi-task EEG cognitive Dataset for passive brain-computer Interface Applications

Brain-Computer Interfaces and especially passive Brain-Computer interfaces (pBCI), with their ability to estimate and monitor user mental states, are receiving increasing attention from both the fundamental research and the applied research and development communities. Testing new pipelines and benchmarking classifiers and feature extraction algorithms is central to further research within this domain. Unfortunately, data sharing in pBCI research is still scarce. The COG-BCI database encompasses the recordings of 29 participants over 3 separate sessions with 4 different tasks (MATB, N-Back, PVT, Flanker) designed to elicit different mental states, for a total of over 100 hours of open EEG data. This dataset was validated on a subjective, behavioral and physiological level, to ensure its usefulness to the pBCI community. Furthermore, a proof of concept is given with an example of mental workload estimation pipeline and results, to ensure that the data can be used for the design and evaluation of pBCI pipelines. This body of work presents a large effort to promote the use of pBCIs in an open science framework.

Modeling the acceptability of BCIs for motor rehabilitation after stroke: A large scale study on the general public

Introduction

Strokes leave around 40% of survivors dependent in their activities of daily living, notably due to severe motor disabilities. Brain-computer interfaces (BCIs) have been shown to be efficiency for improving motor recovery after stroke, but this efficiency is still far from the level required to achieve the clinical breakthrough expected by both clinicians and patients. While technical levers of improvement have been identified (e.g., sensors and signal processing), fully optimized BCIs are pointless if patients and clinicians cannot or do not want to use them. We hypothesize that improving BCI acceptability will reduce patients' anxiety levels, while increasing their motivation and engagement in the procedure, thereby favoring learning, ultimately, and motor recovery. In other terms, acceptability could be used as a lever to improve BCI efficiency. Yet, studies on BCI based on acceptability/acceptance literature are missing. Thus, our goal was to model BCI acceptability in the context of motor rehabilitation after stroke, and to identify its determinants.

Methods

The main outcomes of this paper are the following: i) we designed the first model of acceptability of BCIs for motor rehabilitation after stroke, ii) we created a questionnaire to assess acceptability based on that model and distributed it on a sample representative of the general public in France (N = 753, this high response rate strengthens the reliability of our results), iii) we validated the structure of this model and iv) quantified the impact of the different factors on this population.

Results

Results show that BCIs are associated with high levels of acceptability in the context of motor rehabilitation after stroke and that the intention to use them in that context is mainly driven by the perceived usefulness of the system. In addition, providing people with clear information regarding BCI functioning and scientific relevance had a positive influence on acceptability factors and behavioral intention.

Discussion

With this paper we propose a basis (model) and a methodology that could be adapted in the future in order to study and compare the results obtained with: i) different stakeholders, i.e., patients and caregivers; ii) different populations of different cultures around the world; and iii) different targets, i.e., other clinical and non-clinical BCI applications.

Application of rapid invisible frequency tagging for brain computer interfaces

BACKGROUND

Brain-computer interfaces (BCI) based on steady-state visual evoked potentials (SSVEPs/SSVEFs) are among the most commonly used BCI systems. They require participants to covertly attend to visual objects flickering at specified frequencies. The attended location is decoded online by analysing the power of neuronal responses at the flicker frequency.

NEW METHOD

We implemented a novel rapid invisible frequency-tagging technique, utilizing a state-of-the-art projector with refresh rates of up to 1440 Hz. We flickered the luminance of visual objects at 56 and 60 Hz, which was invisible to participants but produced strong neuronal responses measurable with magnetoencephalography (MEG). The direction of covert attention, decoded from frequency-tagging responses, was used to control an online BCI PONG game.

RESULTS

Our results show that seven out of eight participants were able to play the pong game controlled by the frequency-tagging signal, with average accuracies exceeding 60 %. Importantly, participants were able to modulate the power of the frequency-tagging response within a 1-second interval, while only seven occipital sensors were required to reliably decode the neuronal response.

COMPARISON WITH EXISTING METHODS

In contrast to existing SSVEP-based BCI systems, rapid frequency-tagging does not produce a visible flicker. This extends the time-period participants can use it without fatigue, by avoiding distracting visual input. Furthermore, higher frequencies increase the temporal resolution of decoding, resulting in higher communication rates.

CONCLUSION

Using rapid invisible frequency-tagging opens new avenues for fundamental research and practical applications. In combination with novel optically pumped magnetometers (OPMs), it could facilitate the development of high-speed and mobile next-generation BCI systems.

A pediatric near-infrared spectroscopy brain-computer interface based on the detection of emotional valence

Brain-computer interfaces (BCIs) are being investigated as an access pathway to communication for individuals with physical disabilities, as the technology obviates the need for voluntary motor control. However, to date, minimal research has investigated the use of BCIs for children. Traditional BCI communication paradigms may be suboptimal given that children with physical disabilities may face delays in cognitive development and acquisition of literacy skills. Instead, in this study we explored emotional state as an alternative access pathway to communication. We developed a pediatric BCI to identify positive and negative emotional states from changes in hemodynamic activity of the prefrontal cortex (PFC). To train and test the BCI, 10 neurotypical children aged 8–14 underwent a series of emotion-induction trials over four experimental sessions (one offline, three online) while their brain activity was measured with functional near-infrared spectroscopy (fNIRS). Visual neurofeedback was used to assist participants in regulating their emotional states and modulating their hemodynamic activity in response to the affective stimuli. Child-specific linear discriminant classifiers were trained on cumulatively available data from previous sessions and adaptively updated throughout each session. Average online valence classification exceeded chance across participants by the last two online sessions (with 7 and 8 of the 10 participants performing better than chance, respectively, in Sessions 3 and 4). There was a small significant positive correlation with online BCI performance and age, suggesting older participants were more successful at regulating their emotional state and/or brain activity. Variability was seen across participants in regards to BCI performance, hemodynamic response, and discriminatory features and channels. Retrospective offline analyses yielded accuracies comparable to those reported in adult affective BCI studies using fNIRS. Affective fNIRS-BCIs appear to be feasible for school-aged children, but to further gauge the practical potential of this type of BCI, replication with more training sessions, larger sample sizes, and end-users with disabilities is necessary.

Competing at the Cybathlon championship for people with disabilities: long-term motor imagery brain-computer interface training of a cybathlete who has tetraplegia

BACKGROUND

The brain-computer interface (BCI) race at the Cybathlon championship, for people with disabilities, challenges teams (BCI researchers, developers and pilots with spinal cord injury) to control an avatar on a virtual racetrack without movement. Here we describe the training regime and results of the Ulster University BCI Team pilot who has tetraplegia and was trained to use an electroencephalography (EEG)-based BCI intermittently over 10 years, to compete in three Cybathlon events.

METHODS

A multi-class, multiple binary classifier framework was used to decode three kinesthetically imagined movements (motor imagery of left arm, right arm, and feet), and relaxed state. Three game paradigms were used for training i.e., NeuroSensi, Triad, and Cybathlon Race: BrainDriver. An evaluation of the pilot's performance is presented for two Cybathlon competition training periods-spanning 20 sessions over 5 weeks prior to the 2019 competition, and 25 sessions over 5 weeks in the run up to the 2020 competition.

RESULTS

Having participated in BCI training in 2009 and competed in Cybathlon 2016, the experienced pilot achieved high two-class accuracy on all class pairs when training began in 2019 (decoding accuracy > 90%, resulting in efficient NeuroSensi and Triad game control). The BrainDriver performance (i.e., Cybathlon race completion time) improved significantly during the training period, leading up to the competition day, ranging from 274-156 s (255 ± 24 s to 191 ± 14 s mean ± std), over 17 days (10 sessions) in 2019, and from 230-168 s (214 ± 14 s to 181 ± 4 s), over 18 days (13 sessions) in 2020. However, on both competition occasions, towards the race date, the performance deteriorated significantly.

CONCLUSIONS

The training regime and framework applied were highly effective in achieving competitive race completion times. The BCI framework did not cope with significant deviation in electroencephalography (EEG) observed in the sessions occurring shortly before and during the race day. Changes in cognitive state as a result of stress, arousal level, and fatigue, associated with the competition challenge and performance pressure, were likely contributing factors to the non-stationary effects that resulted in the BCI and pilot achieving suboptimal performance on race day. Trial registration not registered.

EEG-fNIRS-based hybrid image construction and classification using CNN-LSTM

The constantly evolving human–machine interaction and advancement in sociotechnical systems have made it essential to analyze vital human factors such as mental workload, vigilance, fatigue, and stress by monitoring brain states for optimum performance and human safety. Similarly, brain signals have become paramount for rehabilitation and assistive purposes in fields such as brain–computer interface (BCI) and closed-loop neuromodulation for neurological disorders and motor disabilities. The complexity, non-stationary nature, and low signal-to-noise ratio of brain signals pose significant challenges for researchers to design robust and reliable BCI systems to accurately detect meaningful changes in brain states outside the laboratory environment. Different neuroimaging modalities are used in hybrid settings to enhance accuracy, increase control commands, and decrease the time required for brain activity detection. Functional near-infrared spectroscopy (fNIRS) and electroencephalography (EEG) measure the hemodynamic and electrical activity of the brain with a good spatial and temporal resolution, respectively. However, in hybrid settings, where both modalities enhance the output performance of BCI, their data compatibility due to the huge discrepancy between their sampling rate and the number of channels remains a challenge for real-time BCI applications. Traditional methods, such as downsampling and channel selection, result in important information loss while making both modalities compatible. In this study, we present a novel recurrence plot (RP)-based time-distributed convolutional neural network and long short-term memory (CNN-LSTM) algorithm for the integrated classification of fNIRS EEG for hybrid BCI applications. The acquired brain signals are first projected into a non-linear dimension with RPs and fed into the CNN to extract essential features without performing any downsampling. Then, LSTM is used to learn the chronological features and time-dependence relation to detect brain activity. The average accuracies achieved with the proposed model were 78.44% for fNIRS, 86.24% for EEG, and 88.41% for hybrid EEG-fNIRS BCI. Moreover, the maximum accuracies achieved were 85.9, 88.1, and 92.4%, respectively. The results confirm the viability of the RP-based deep-learning algorithm for successful BCI systems.

2020 International brain-computer interface competition: A review

The brain-computer interface (BCI) has been investigated as a form of communication tool between the brain and external devices. BCIs have been extended beyond communication and control over the years. The 2020 international BCI competition aimed to provide high-quality neuroscientific data for open access that could be used to evaluate the current degree of technical advances in BCI. Although there are a variety of remaining challenges for future BCI advances, we discuss some of more recent application directions: (i) few-shot EEG learning, (ii) micro-sleep detection (iii) imagined speech decoding, (iv) cross-session classification, and (v) EEG(+ear-EEG) detection in an ambulatory environment. Not only did scientists from the BCI field compete, but scholars with a broad variety of backgrounds and nationalities participated in the competition to address these challenges. Each dataset was prepared and separated into three data that were released to the competitors in the form of training and validation sets followed by a test set. Remarkable BCI advances were identified through the 2020 competition and indicated some trends of interest to BCI researchers.

Neural correlates of user learning during long-term BCI training for the Cybathlon competition

Background

Brain-computer interfaces (BCIs) are systems capable of translating human brain patterns, measured through electroencephalography (EEG), into commands for an external device. Despite the great advances in machine learning solutions to enhance the performance of BCI decoders, the translational impact of this technology remains elusive. The reliability of BCIs is often unsatisfactory for end-users, limiting their application outside a laboratory environment.

Methods

We present the analysis on the data acquired from an end-user during the preparation for two Cybathlon competitions, where our pilot won the gold medal twice in a row. These data are of particular interest given the mutual learning approach adopted during the longitudinal training phase (8 months), the long training break in between the two events (1 year) and the demanding evaluation scenario. A multifaceted perspective on long-term user learning is proposed: we enriched the information gathered through conventional metrics (e.g., accuracy, application performances) by investigating novel neural correlates of learning in different neural domains.

Results

First, we showed that by focusing the training on user learning, the pilot was capable of significantly improving his performance over time even with infrequent decoder re-calibrations. Second, we revealed that the analysis of the within-class modifications of the pilot’s neural patterns in the Riemannian domain is more effective in tracking the acquisition and the stabilization of BCI skills, especially after the 1-year break. These results further confirmed the key role of mutual learning in the acquisition of BCI skills, and particularly highlighted the importance of user learning as a key to enhance BCI reliability.

Conclusion

We firmly believe that our work may open new perspectives and fuel discussions in the BCI field to shift the focus of future research: not only to the machine learning of the decoder, but also in investigating novel training procedures to boost the user learning and the stability of the BCI skills in the long-term. To this end, the analyses and the metrics proposed could be used to monitor the user learning during training and provide a marker guiding the decoder re-calibration to maximize the mutual adaptation of the user to the BCI system.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12984-022-01047-x.

A Transfer Learning Algorithm to Reduce Brain-Computer Interface Calibration Time for Long-Term Users

Current motor imagery-based brain-computer interface (BCI) systems require a long calibration time at the beginning of each session before they can be used with adequate levels of classification accuracy. In particular, this issue can be a significant burden for long term BCI users. This article proposes a novel transfer learning algorithm, called r-KLwDSA, to reduce the BCI calibration time for long-term users. The proposed r-KLwDSA algorithm aligns the user's EEG data collected in previous sessions to the few EEG trials collected in the current session, using a novel linear alignment method. Thereafter, the aligned EEG trials from the previous sessions and the few EEG trials from the current sessions are fused through a weighting mechanism before they are used for calibrating the BCI model. To validate the proposed algorithm, a large dataset containing the EEG data from 11 stroke patients, each performing 18 BCI sessions, was used. The proposed framework demonstrated a significant improvement in the classification accuracy, of over 4% compared to the session-specific algorithm, when there were as few as two trials per class available from the current session. The proposed algorithm was particularly successful in improving the BCI accuracy of the sessions that had initial session-specific accuracy below 60%, with an average improvement of around 10% in the accuracy, leading to more stroke patients having meaningful BCI rehabilitation.

Riemannian channel selection for BCI with between-session non-stationarity reduction capabilities

OBJECTIVE

Between-session non-stationarity is a major challenge of current Brain-Computer Interfaces (BCIs) that affects system performance. In this paper, we investigate the use of channel selection for reducing between-session non-stationarity with Riemannian BCI classifiers. We use the Riemannian geometry framework of covariance matrices due to its robustness and promising performances. Current Riemannian channel selection methods do not consider between-session non-stationarity and are usually tested on a single session. Here, we propose a new channel selection approach that specifically considers non-stationarity effects and is assessed on multi-session BCI data sets.

METHODS

We remove the least significant channels using a sequential floating backward selection search strategy. Our contributions include: 1) quantifying the non-stationarity effects on brain activity in multi-class problems by different criteria in a Riemannian framework and 2) a method to predict whether BCI performance can improve using channel selection.

RESULTS

We evaluate the proposed approaches on three multi-session and multi-class mental tasks (MT)-based BCI datasets. They could lead to significant improvements in performance as compared to using all channels for datasets affected by between-session non-stationarity and to significant superiority to the state-of-the-art Riemannian channel selection methods over all datasets, notably when selecting small channel set sizes.

CONCLUSION

Reducing non-stationarity by channel selection could significantly improve Riemannian BCI classification accuracy.

SIGNIFICANCE

Our proposed channel selection approach contributes to make Riemannian BCI classifiers more robust to between-session non-stationarities.