A review of classification algorithms for EEG-based brain–computer interfaces: a 10 year update

Journal of Neural Engineering, Год журнала: 2018, Номер 15(3), С. 031005 - 031005

Опубликована: Фев. 28, 2018

Most current electroencephalography (EEG)-based brain-computer interfaces (BCIs) are based on machine learning algorithms. There is a large diversity of classifier types that used in this field, as described our 2007 review paper. Now, approximately ten years after publication, many new algorithms have been developed and tested to classify EEG signals BCIs. The time therefore ripe for an updated classification BCIs.We surveyed the BCI literature from 2017 identify approaches investigated design We synthesize these studies order present such algorithms, report how they were BCIs, what outcomes, their pros cons.We found recently designed EEG-based BCIs can be divided into four main categories: adaptive classifiers, matrix tensor transfer deep learning, plus few other miscellaneous classifiers. Among these, classifiers demonstrated generally superior static ones, even with unsupervised adaptation. Transfer also prove useful although benefits remain unpredictable. Riemannian geometry-based methods reached state-of-the-art performances multiple problems deserve explored more thoroughly, along tensor-based methods. Shrinkage linear discriminant analysis random forests appear particularly small training samples settings. On hand, not yet shown convincing improvement over methods.This paper provides comprehensive overview modern presents principles guidelines when use them. It identifies number challenges further advance BCI.

Язык: Английский

---

BCI for stroke rehabilitation: motor and beyond

Journal of Neural Engineering, Год журнала: 2020, Номер 17(4), С. 041001 - 041001

Опубликована: Июль 2, 2020

Abstract Stroke is one of the leading causes long-term disability among adults and contributes to major socio-economic burden globally. frequently results in multifaceted impairments including motor, cognitive emotion deficits. In recent years, brain–computer interface (BCI)-based therapy has shown promising for post-stroke motor rehabilitation. spite success received by BCI-based interventions domain, non-motor are yet receive similar attention research clinical settings. Some preliminary encouraging rehabilitation using BCI seem suggest that it may also hold potential treating deficits such as impairments. Moreover, past studies have an intricate relationship between functions which might influence overall outcome. A number highlight inability current treatment protocols account implicit interplay functions. This indicates necessity explore all-inclusive plan targeting synergistic these standalone interventions. approach lead better recovery than individual isolation. this paper, we review advances use systems beyond particular, improving cognition stroke patients. Building on findings domains, next discuss possibility a holistic system affect synergistically promote restorative neuroplasticity. Such would provide all-encompassing platform, overarching outcomes transfer quality living. first works analyse cross-domain functional enabled

Язык: Английский

---

Brain-computer interface-based robotic end effector system for wrist and hand rehabilitation: results of a three-armed randomized controlled trial for chronic stroke

Frontiers in Neuroengineering, Год журнала: 2014, Номер 7

Опубликована: Июль 29, 2014

The objective of this study was to investigate the efficacy an Electroencephalography (EEG)-based Motor Imagery (MI) Brain-Computer Interface (BCI) coupled with a Haptic Knob (HK) robot for arm rehabilitation in stroke patients. In three-arm, single-blind, randomized controlled trial; 21 chronic hemiplegic patients (Fugl-Meyer Assessment (FMMA) score 10-50), recruited after pre-screening MI BCI ability, were randomly allocated BCI-HK, HK or Standard Arm Therapy (SAT) groups. All groups received 18 sessions intervention over 6 weeks, 3 per week, 90 min session. BCI-HK group 1 h intervention, and Both 120 trials robot-assisted hand grasping knob manipulation followed by 30 therapist-assisted mobilization. SAT 1.5 mobilization forearm pronation-supination movements incorporating wrist control grasp-release functions. all, 14 males, 7 females, mean age 54.2 years, duration 385.1 days, baseline FMMA 27.0 recruited. primary outcome measure upper extremity scores measured mid-intervention at week 3, end-intervention 6, follow-up weeks 12 24. Seven, 8 subjects underwent interventions respectively. improved all groups, but no intergroup differences found any time points. Significantly larger motor gains observed compared 12, 24, did not differ from point. conclusion, is effective, safe, may have potential enhancing recovery when combined

Язык: Английский

---

Signal Processing Approaches to Minimize or Suppress Calibration Time in Oscillatory Activity-Based Brain–Computer Interfaces

Proceedings of the IEEE, Год журнала: 2015, Номер 103(6), С. 871 - 890

Опубликована: Май 18, 2015

One of the major limitations brain-computer interfaces (BCI) is their long calibration time, which limits use in practice, both by patients and healthy users alike. Such times are due to large between-user variability thus need collect numerous training electroencephalography (EEG) trials for machine learning algorithms used BCI design. In this paper, we first survey existing approaches reduce or suppress these being notably based on regularization, user-to-user transfer, semi-supervised a priori physiological information. We then propose new tools time. particular, generate artificial EEG from few initially available, order augment set size. These obtained relevant combinations distortions original available. three different methods do so. also new, fast simple approach perform transfer BCI. Finally, study compare offline approaches, old ones, data 50 sets. This enables us identify guidelines about how time

Язык: Английский

---

HS-CNN: a CNN with hybrid convolution scale for EEG motor imagery classification

Journal of Neural Engineering, Год журнала: 2019, Номер 17(1), С. 016025 - 016025

Опубликована: Сен. 2, 2019

Objective. Electroencephalography (EEG) motor imagery classification has been widely used in healthcare applications such as mobile assistive robots and post-stroke rehabilitation. Recently, EEG methods based on convolutional neural networks (CNNs) have proposed achieved relatively high accuracy. However, these use single convolution scale the CNN, while best differs from subject to subject. This limits Another issue is that accuracy degrades when training data limited. Approach. To address issues, we a hybrid-scale CNN architecture with augmentation method for classification. Main results. Compared several state-of-the-art methods, achieves an average of 91.57% 87.6% two commonly datasets, which outperforms methods. Significance. The effectively addresses issues existing CNN-based improves

Язык: Английский

---

EEG-Based Strategies to Detect Motor Imagery for Control and Rehabilitation

IEEE Transactions on Neural Systems and Rehabilitation Engineering, Год журнала: 2016, Номер 25(4), С. 392 - 401

Опубликована: Дек. 30, 2016

Advances in brain-computer interface (BCI) technology have facilitated the detection of Motor Imagery (MI) from electroencephalography (EEG). First, we present three strategies using BCI to detect MI EEG: operant conditioning that employed a fixed model, machine learning subject-specific model computed calibration, and adaptive strategy continuously compute model. Second, review prevailing works strategies. Third, our past work on six stroke patients who underwent rehabilitation clinical trial with averaged accuracies 79.8% during calibration 69.5% across 18 online feedback sessions. Finally, perform an offline study this paper employing strategy. The results yielded significant improvements 12% (p <; 0.001) 9% all data limited preceding respectively accuracies. showed increase amount training improvements. Nevertheless, larger part improvement was due changing models did not deteriorate Hence is effective addressing non-stationarity between

Язык: Английский

---

Assessment of the Efficacy of EEG-Based MI-BCI With Visual Feedback and EEG Correlates of Mental Fatigue for Upper-Limb Stroke Rehabilitation

IEEE Transactions on Biomedical Engineering, Год журнала: 2019, Номер 67(3), С. 786 - 795

Опубликована: Июнь 5, 2019

This single-arm multisite trial investigates the efficacy of neurostyle brain exercise therapy towards enhanced recovery (nBETTER) system, an electroencephalogram (EEG)-based motor imagery brain-computer interface (MI-BCI) employing visual feedback for upper-limb stroke rehabilitation, and presence EEG correlates mental fatigue during BCI usage.A total 13 recruited patients underwent thrice-weekly nBETTER coupled with standard arm over six weeks. Upper-extremity Fugl-Meyer assessment (FMA) scores were measured at baseline (week 0), post-intervention 6), follow-ups (weeks 12 24). In total, 11/13 (mean age 55.2 years old, mean post-stroke duration 333.7 days, FMA 35.5) completed study.Significant gains relative to observed weeks 6 24. Retrospectively comparing (SAT) control group haptic knob (BCI-HK) intervention from a previous similar study, SAT had no significant gains, whereas BCI-HK 6, 12, analysis revealed positive correlations between beta power performance in frontal central regions, suggesting that may contribute poorer performance.nBETTER, EEG-based MI-BCI only feedback, helps survivors sustain short-term improvement. Analysis indicates be present.This study adds growing literature safe effective rehabilitation MI-BCI, suggests additional fatigue-monitoring role future such BCI.

Язык: Английский

---

BETA: A Large Benchmark Database Toward SSVEP-BCI Application

Frontiers in Neuroscience, Год журнала: 2020, Номер 14

Опубликована: Июнь 23, 2020

The brain-computer interface (BCI) provides an alternative means to communicate and it has sparked growing interest in the past two decades. Specifically, for Steady-State Visual Evoked Potential (SSVEP) based BCI, marked improvement been made frequency recognition method data sharing. However, number of pubic databases is still limited this field. Therefore, we present a BEnchmark database Towards BCI Application (BETA) study. BETA composed 64-channel Electroencephalogram (EEG) 70 subjects performing 40-target cued-spelling task. design acquisition are pursuit meeting demand from real-world applications can be used as test-bed these scenarios. We validate by series analyses conduct classification analysis eleven methods on BETA. recommend using metric wide-band signal-to-noise ratio (SNR) quotient characterize SSVEP at single-trial population levels, respectively. downloaded following link http://bci.med.tsinghua.edu.cn/download.html.

Язык: Английский

---

Predicting Mental Imagery-Based BCI Performance from Personality, Cognitive Profile and Neurophysiological Patterns

PLoS ONE, Год журнала: 2015, Номер 10(12), С. e0143962 - e0143962

Опубликована: Дек. 1, 2015

Mental-Imagery based Brain-Computer Interfaces (MI-BCIs) allow their users to send commands a computer using brain-activity alone (typically measured by ElectroEncephaloGraphy—EEG), which is processed while they perform specific mental tasks. While very promising, MI-BCIs remain barely used outside laboratories because of the difficulty encountered control them. Indeed, although some obtain good performances after training, substantial proportion remains unable reliably an MI-BCI. This huge variability in user-performance led community look for predictors MI-BCI ability. However, these were only explored motor-imagery BCIs, and mostly single training session per subject. In this study, 18 participants instructed learn EEG-based performing 3 MI-tasks, 2 non-motor tasks, across 6 sessions, on different days. Relationships between participants’ BCI personality, cognitive profile neurophysiological markers explored. no relevant relationships with found, strong correlations mental-rotation scores (reflecting spatial abilities) revealed. Also, predictive model performance psychometric questionnaire was proposed. A leave-one-subject-out cross validation process revealed stability reliability model: it enabled predict mean error less than points. study determined how users’ profiles impact ability thus clears way designing novel protocols, adapted each user.

Язык: Английский

---

Effects of Action Observational Training Plus Brain–Computer Interface‐Based Functional Electrical Stimulation on Paretic Arm Motor Recovery in Patient with Stroke: A Randomized Controlled Trial

Occupational Therapy International, Год журнала: 2015, Номер 23(1), С. 39 - 47

Опубликована: Авг. 24, 2015

Abstract The purpose of this study was to investigate whether action observational training (AOT) plus brain–computer interface‐based functional electrical stimulation (BCI‐FES) has a positive influence on motor recovery paretic upper extremity in patients with stroke. This hospital‐based, randomized controlled trial blinded assessor. Thirty first‐time stroke were randomly allocated one two groups: the BCI‐FES group ( n = 15) and control 15). administered AOT five times per week during 4 weeks while both groups received conventional therapy. primary outcomes Fugl‐Meyer Assessment Upper Extremity, Motor Activity Log (MAL), Modified Barthel Index range motion arm. A assessor evaluated at baseline weeks. All did not differ significantly between groups. After weeks, Extremity sub‐items (total, shoulder wrist), MAL (MAL‐Activity Use Quality Movement), wrist flexion higher p < 0.05). BCI‐based FES is effective arm rehabilitation by improving performance. improvements suggest that can be used as therapeutic tool for rehabilitation. limitations are subjects had certain limited level function, sample size comparatively small; hence, it recommended future large‐scale trials should consider stratified lager populations according function. Copyright © 2015 John Wiley & Sons, Ltd.

Язык: Английский

---