2007-04

BCI Monthly Roundup — April 2007

1 April 2007 – 30 April 2007

Introduction

April 2007 was dominated by the push from lab prototypes toward practical, user-ready BCIs. Adaptive algorithms took center stage: online adaptive LDA and P300 classifiers addressed EEG non-stationarity and session-to-session variability, while regularized discriminant analysis and wavelet/CSP-style features improved robustness with limited data. Landmark demos—Graz motor imagery driving navigation in a virtual apartment and self-paced control in Google Earth—showed that non-invasive BCIs could support continuous, goal-directed interaction. At the same time, clinical applications were spelled out clearly: communication restoration in ALS and locked-in patients, closed-loop neurofeedback for stroke, and rehabilitation and targeting considerations in spinal cord injury.

Invasive and semi-invasive work advanced on several fronts. ECoG proved suitable for real-time cursor control and continuous hand trajectory decoding, with high-gamma power across multiple regions supporting multimodal control; language cortex ECoG and decoding during natural speech pointed toward communication BCIs beyond motor cortex. Intracortical BMIs emphasized better decoding (ensemble and probabilistic population decoders, velocity Kalman filters) and high-performance 3D control; cortical microstimulation for artificial tactile feedback was framed as a first closed-loop sensory–motor prosthesis. IEEE NER 2007 and references to SfN 2007 and BrainGate2 design kept the community focused on shared platforms, next-generation trials, and deployment challenges.

Suggested Titles

• From lab to living room: adaptive BCIs and the Graz virtual apartment
• ECoG and intracortical BMIs: real-time control, trajectories, and sensory feedback
• Closing the loop: adaptive classifiers and clinical BCIs in April 2007
• Multimodal and hybrid BCIs: P300, motor imagery, eye tracking, and passive monitoring
• BrainGate2, NER 2007, and the practical BCI agenda

Papers and Prototypes

Decoding and control. Large-scale ensemble and probabilistic population decoding improved BMI trajectory and cursor accuracy; probabilistic decoding helped especially in low firing-rate regimes. A velocity Kalman filter gave smooth closed-loop cursor control in human BrainGate participants. High-performance 3D cursor control from motor cortex reached new levels of speed and accuracy. Population dynamics in motor cortex predicted trial-by-trial BMI performance, and neural correlates of BMI skill learning mirrored natural motor skill consolidation. Goal-directed movement representations in motor cortex were shown to carry target information for BMI. LFP-based decoding enabled 3D robotic arm control comparable to single-unit decoders.

• Simultaneous firing of neurons in motor cortex during BMI — J Neuroscience · Large-scale cortical ensembles improve BMI trajectory decoding.
• Probabilistic neural population decoding for BMI — J Neural Eng · Probabilistic decoding improves accuracy in low firing-rate conditions.
• Closed-loop BMI with velocity Kalman filter — J Neural Eng · Smooth closed-loop cursor control in human BrainGate participants.
• High-performance BMI for 3D cursor control — J Neural Eng · High-performance 3D cursor control from motor cortex.
• Local field potential BCI for arm control — J Neural Eng · LFP signals for 3D robotic arm control comparable to spike decoders.
• Real-time neural state estimation for BCI — J Neural Eng · Real-time state-space neural state estimation for smooth cursor control.

ECoG. High-gamma ECoG across multiple regions supported multimodal BCI control. Real-time 1D and 2D ECoG cursor control was achieved in presurgical patients across diverse electrode layouts. Continuous hand trajectory decoding from ECoG approached single-unit-level accuracy. ECoG during natural speech was decoded with high accuracy, and language-cortex ECoG was proposed for communication BCIs.

• ECoG factors underlying multimodal BCI control — J Neuroscience · High-gamma ECoG power and multimodal BCI control.
• ECoG-based real-time cursor control — J Neural Eng · Real-time 1D/2D ECoG cursor control in presurgical patients.
• Decoding hand movement trajectories from ECoG — J Neurosci Methods · Continuous hand trajectory from ECoG.
• Decoding ECoG signals during natural speech — J Neuroscience · ECoG speech decoding; relevance for speech BCIs.
• Language cortex BCI for communication — J Neural Eng · Language cortex ECoG for communication BCIs.

EEG: adaptive and practical. Online adaptive LDA handled EEG non-stationarity without full retraining; online adaptive P300 reduced session-to-session variability. Regularized LDA improved classification with small training sets. Dry EEG electrodes enabled gel-free, wearable BCI use. High-resolution EEG with source reconstruction improved motor imagery spatial specificity; adaptive spatial filtering tailored topography to the user.

• Study of online adaptive discriminant analysis for EEG-BCI — IEEE Trans Biomed Eng · Online adaptive LDA for non-stationary EEG.
• Online adaptation of P300 classifier — J Neural Eng · Online adaptive P300 for stable performance across sessions.
• Regularized discriminant analysis for BCI — J Neural Eng · Regularized LDA with small training datasets.
• Practical dry electrode EEG for BCI — IEEE Trans Biomed Eng · Dry electrodes for wearable BCI.
• High-resolution EEG techniques for BCI applications — J Neural Eng · High-resolution EEG and source reconstruction for motor imagery.
• Adaptive spatial filtering for EEG BCI — IEEE TNSRE · User-specific adaptive spatial filters.

EEG: features and paradigms. Wavelet packet and sub-band CSP gave multi-resolution and frequency-specific features for motor imagery; combined temporal (IRA) and spatial (CSP) features improved single-trial classification. Power spectral density was proposed as a stable long-term EEG feature. SSVEP with canonical correlation analysis gave high accuracy with short segments. Selective attention (SSVEP/P300) supported multi-target control. TMS before a BCI session enhanced cortical excitability and motor imagery quality.

• Wavelet packet features for EEG motor imagery — J Neural Eng · Multi-resolution wavelet packet features.
• Sub-band Common Spatial Pattern for BCI — IEEE NER 2007 · Sub-band CSP for motor imagery.
• Temporal and spatial features of single-trial EEG for BCI — Comput Intell Neuroscience · Temporal (IRA) and spatial (CSP) features for single-trial EEG.
• Power spectral density features for EEG BCI — IEEE EMBS 2007 · PSD as stable long-term feature.
• SSVEP-BCI using canonical correlation analysis — IEEE Trans Neural Syst · CCA-based SSVEP with short signal length.
• Attention-based BCI paradigms — J Neural Eng · SSVEP and P300 for multi-target control.
• Transcranial magnetic stimulation for BCI enhancement — Exp Brain Res · TMS to boost motor imagery signal quality.

Hybrid and passive BCIs. Hybrid P300 + motor imagery improved ITR; EEG motor imagery combined with eye tracking improved accuracy. fNIRS captured cognitive/hemodynamic signals for passive monitoring; passive BCI was used for workload monitoring without explicit control.

• Simultaneous P300 and motor imagery BCI — IEEE Trans Biomed Eng · Hybrid P300 and motor imagery for higher ITR.
• Hybrid BCI combining EEG and eye tracking — J Neural Eng · EEG motor imagery plus eye gaze for accuracy.
• fNIRS cognitive activity detection for BCI — IEEE NER 2007 · fNIRS for passive cognitive/hemodynamic monitoring.
• Passive BCI for workload monitoring — J Neural Eng · Workload monitoring without user intent.

Usability and applications. Motor imagery BCI enabled self-paced navigation in a virtual apartment; the self-paced Graz BCI was used for virtual navigation and Google Earth. Auditory ERP BCI allowed communication without vision. Prefrontal signals were proposed for cognitive-control BCIs. Sparse Bayesian learning was used for automatic EEG feature selection; online ICA separated neural signals from artifacts in real time.

• Brain–computer communication: virtual apartment control — IEEE TNSRE · Motor imagery navigation in a virtual apartment.
• Self-paced Graz BCI: methods and applications — Comput Intell Neuroscience · Self-paced Graz BCI for virtual navigation and Google Earth.
• Auditory ERP BCI for patients without vision — J Neural Eng · Auditory ERP BCI for non-visual communication.
• Prefrontal cortex BCI for cognitive control — J Neural Eng · Prefrontal signals for executive-function BCIs.
• Sparse Bayesian learning for BCI — J Neural Eng · Sparse Bayesian learning for EEG feature selection.
• Online independent component analysis for BCI — J Neural Eng · Online ICA for artifact separation in real time.

Sensory feedback and closed-loop. Cortical microstimulation encoded artificial tactile feedback in BMI, described as a first closed-loop sensory–motor prosthesis. Perceptual thresholds for brain stimulation were quantified for sensory BCI design. Parietal reach region population coding of intended reach targets was highlighted as a cognitive BCI signal source.

• Sensory feedback restoration via cortical microstimulation — PLoS ONE · Closed-loop tactile feedback via cortical microstimulation.
• Brain stimulation and perception for sensory BCIs — J Neuroscience · Perceptual thresholds for stimulation-based sensory BCIs.
• Population coding of intended movement from parietal cortex — Neuron · Parietal reach region as cognitive BCI signal.

Hardware, biocompatibility, and platforms. Tissue response to implanted electrodes was reviewed as a determinant of chronic recording longevity. Multi-electrode array fabrication for neural interfaces was advanced. Motor cortex reorganization after spinal cord injury was discussed for BCI targeting. BCI for restoring neurological function (including ECoG) was reviewed; practical issues in clinical BCI systems and a special issue on practical implementations framed deployment challenges. Computational neuroscience frameworks for BCI and DBS/neuromodulation advances were also covered.

• Biocompatibility of implanted neural electrodes — Biomaterials · Tissue response and chronic recording longevity.
• Multi-electrode array fabrication for neural interfaces — J Neural Eng · MEA fabrication and materials.
• Motor cortex reorganization after spinal cord injury — J Neurophysiology · Reorganization after SCI and BCI targeting.
• Brain-machine interface: restoring neurological function — Physical Medicine & Rehab · BCI/ECoG for restoration.
• Practical issues in clinical BCI systems — J Neural Eng · Reliability and setup in clinical BCI.
• Brain-computer interfaces: towards practical implementations — Comput Intell Neuroscience · Practical implementation across EEG, ECoG, intracortical.
• Computational neuroscience of BCI — Comput Neurosci · Computational frameworks for BCI.
• DBS and neuromodulation advances in 2007 — Movement Disorders · DBS and adaptive closed-loop prototypes.

Prosthetics and related. EMG control of an anthropomorphic prosthetic hand with individual finger control, implantable wireless EMG for prosthetics, and decoding individual finger movements from motor cortex for dexterous prosthetic control were reported. EEG correlates of motor adaptation during BCI use and training-induced changes in cortical oscillations were also highlighted.

• EMG control of anthropomorphic prosthetic hand — IEEE EMBS 2007 · EMG-controlled hand with finger-level control.
• Implantable wireless EMG sensor for prosthetics — IEEE Trans Biomed Eng · Wireless implantable EMG for prosthetic control.
• Towards a BCI for dexterous control of multi-fingered prosthetic hand — IEEE NER 2007 · Decoding finger movements for prosthetic control.
• EEG correlates of motor adaptation for BCI — Clinical Neurophysiology · EEG correlates of adaptation during BCI use.
• Cortical oscillations modulate during BCI training — J Neural Eng · Oscillation changes during BCI learning.

Clinical and Regulatory

ALS and locked-in. BCI was positioned for communication restoration in ALS across disease stages; decoding of residual motor intentions from EEG in patients with moderate impairment and BCI use in locked-in patients demonstrated communication when voluntary movement is lost.

• BCI for restoration of communication in ALS — Amyotrophic Lateral Sclerosis · Communication restoration in ALS at various stages.
• Decoding motor intentions from EEG in ALS — Exp Brain Res · Decoding residual motor intentions in ALS.
• EEG-based BCI for locked-in patients — Clinical Neurophysiology · BCI communication in locked-in syndrome.

Rehabilitation and injury. Closed-loop neurofeedback BCI for stroke was linked to neuroplasticity benefits. Spinal cord injury rehabilitation and functional recovery with BCI were reviewed; motor cortex reorganization after SCI was discussed for BCI targeting.

• Closed-loop neurofeedback BCI for stroke rehabilitation — Arch Phys Med Rehab · Neurofeedback BCI and neuroplasticity in stroke.
• Spinal cord injury and BCI rehabilitation — Topics Spinal Cord Inj Rehab · BCI in SCI rehabilitation and recovery.

Next-generation trials and meetings. BrainGate2 design and preparation for the next-generation clinical trial were outlined. SfN 2007 and IEEE NER 2007 highlighted BrainGate, Schwartz robotic arm, and cortical prosthetics, plus BCI, prosthetics, and neural recording in the NER proceedings.

• Donoghue BrainGate 2: preparation and design — IEEE EMBS 2007 · BrainGate2 clinical trial design.
• Neuromotor prosthesis development — Overview of BrainGate system and 2007 goals.
• SfN 2007: BCI and neural interface highlights — Society for Neuroscience 2007 · BrainGate, Schwartz arm, cortical prosthetics.
• IEEE NER 2007 proceedings: neural engineering highlights — IEEE NER 2007 · BCI, prosthetics, neural recording.

Companies and Funding

No dedicated company or funding announcements appeared in the April 2007 briefs. Conference and program activity (IEEE NER 2007, SfN 2007, BrainGate2 design) is covered under Clinical and Regulatory and Papers and Prototypes.

Emerging Themes

Practical and adaptive BCIs. The month emphasized deployment-oriented work: online adaptive LDA and P300, regularized methods for small datasets, dry electrodes, and reviews of practical issues in clinical BCI and toward practical implementations. Usability landmarks included the Graz virtual apartment and self-paced Google Earth control.

Hybrid and multimodal control. Hybrid designs combined P300 with motor imagery and EEG with eye tracking; ECoG leveraged high-gamma across multiple regions for multimodal control. Passive BCIs (fNIRS, workload monitoring) extended the scope beyond direct motor or P300 control.

Closed-loop sensation and cognition. Cortical microstimulation for tactile feedback was framed as a first closed-loop sensory–motor prosthesis; perceptual thresholds for stimulation were quantified. Parietal and language-cortex signals were proposed for cognitive and communication BCIs; decoding during natural speech and prefrontal cognitive control were highlighted.

Clinical populations and next trials. ALS and locked-in communication, stroke neurofeedback, and SCI rehabilitation and targeting were recurring themes. BrainGate2 preparation and design and conference coverage (SfN, NER) kept the focus on translation and shared engineering standards.

previous << monthly >> next