Journal of neural engineering最新文献

Objective.Advances in implantable brain-computer interfaces (iBCI) have rapidly accelerated in the last decade that promises to improve the quality of life of patients with communications, sensory, and motor control disabilities.Approach.In this Perspective, we quantify the extent and nature of scientific convergence across 21 research groups conducting iBCI clinical trials worldwide. Using medical subject headers and Classification of Instructional Programs taxonomies, we analyze topical and disciplinary integration within 161 publications from 1998-2023 to assess how deeply team composition aligns with research themes and translational impact.Main Results.Our findings indicate uneven patterns of convergence, with many teams combining engineering and clinical expertise yet omitting ethical, legal, and social dimensions. This represents what we term short-cut convergence.Significance.We propose an operational definition of this phenomenon and identify practical steps for researchers and funders to strengthen full convergence to accelerate iBCI translation and implementation.

Objective. Speech imagery recognition from electroencephalography (EEG) signals is an emerging challenge in brain-computer interfaces, and has important applications, such as in the interaction with locked-in patients. In this work, we use graph signal processing for developing a more effective representation of EEG signals in speech imagery recognition.Approach. We propose a dynamic graph representation that uses multiple graphs constructed based on the time-varying correlations between EEG channels. Our methodology is particularly suitable for signals that exhibit fluctuating correlations, which cannot be adequately modeled through a static (single graph) model. The resultant representation provides graph frequency features that compactly capture the spatial patterns of the underlying multidimensional EEG signal as well as the evolution of spatial relationships over time. These dynamic graph features are fed into an attention-based long short-term memory network for speech imagery recognition. A novel EEG data augmentation method is also proposed for improving training robustness.Main results. Experimental evaluation using a range of experiments shows that the proposed dynamic graph features are more effective than conventional time-frequency features for speech imagery recognition. The overall system outperforms current state-of-the-art approaches, yielding accuracy gains of up to 10%.Significance. The dynamic graph representation captures time-varying spatial relationships in EEG signals, overcoming limitations of static graph models and conventional feature extraction. Combined with data augmentation and attention-based classification, it demonstrates substantial improvements over existing methods in speech imagery recognition.

Objective. Chronically implanted microelectrode arrays (MEAs) are used for stimulating and recording neural activity in research and clinical settings. However, their reliability can be compromised by insufficient encapsulation stability. Amorphous silicon carbide (a-SiC), a chemically stable and biocompatible material, has emerged as a potential thin-film encapsulation for MEAs. We aimed to evaluate thin-film a-SiC encapsulation using electrical-accelerated aging (EAA) and to demonstrate a methodology for obtaining acceleration factors for EAA by Weibull analysis.Approach. Interdigitated electrodes (IDEs) encapsulated with a-SiC were subjected to voltage cycling and stepped-voltage protocols to measure leakage currents in buffered saline at 37 °C. EAA employed incrementally increasing voltage biases over time to induce degradation and reveal failure mechanisms.Main results. IDEs exhibited a significant change in electrical behavior on exposure to saline, with failure initiating at specific voltages and accompanied by gas evolution at defect sites. Incremental voltage biasing revealed a capacitive-to-faradaic transition in leakage current response that was used as a failure criterion.Significance. Acceleration factors for voltage-driven accelerated aging of a-SiC thin-film encapsulation can be obtained by Weibull analysis using a mechanistic failure criterion. Breakdown occurs at processing-related defects in the a-SiC. This study demonstrates the use of EAA for evaluating failure in a-SiC thin-film encapsulation used in implantable MEAs. EAA is broadly applicable to thin-film MEAs and provides a highly relevant method of predicting implanted lifetimes of bioelectronics.

Objective.Direct current (DC) electrical block of peripheral sensory axons has potential for clinical applications in pain management. The C-fiber reflex (CFR), elicited via noxious stimulation of the foot, is suitable for investigating the activation of unmyelinated C-fiber nerves, the fiber class that is responsible for lingering pain sensations.Approach.In anesthetized rats, the CFR was elicited via electrical stimulation to the plantar surface of the hindpaw, and the resulting C-fiber-evoked electromyography (EMG) signals were recorded from the ipsilateral biceps femoris muscle. A carbon separated interface nerve electrode was used to deliver DC block to arrest action potentials in the sciatic nerve. The efficacy of the block was observed as a reduction/abolition of the magnitude of the EMG in a time window corresponding to the latency of C-fibers activity.Main results. Complete cessation of nerve activity could be achieved instantaneously by applying DC at the 'block threshold (BT)'. At amplitudes below the BT, complete block could be induced over a period of seconds to minutes, with lower currents being correlated with longer induction times. When block was applied for prolonged periods of time, block was sustained following the cessation of DC delivery. This 'recovery period' was longer for longer durations of block application.Significance. The CFR is an established method to investigate pharmaceutical pain therapies, yet to date, has not been used to assess electrical block of sensory axons. Therefore, anatomical and electrophysiological methods were used to validate this method. DC nerve block shows promise for clinical pain management applications. Furthermore, the temporal properties described here could be used to reduce overall electrical current delivery and improve safety.

Objective.Effective deep brain stimulation (DBS) treatment for Parkinson's disease requires careful surgical targeting and adjustment of stimulation parameters to avoid motor side effects caused by activation of the internal capsule. Currently, patients must self-report side effects during implantation surgery and device programming-a subjective and inconsistent process that may delay optimized treatment or result in suboptimal therapy. Motor evoked potentials (mEP), the use of electromyography to record DBS-induced muscle activation, offer a promising biomarker for objective motor side effect detection.Approach.We present an automated algorithmic procedure for mEP detection and quantification.Main results.First, we design and evaluate a series of signal processing techniques to accurately detect mEP while mitigating the influence of stimulation artifacts and noise, then demonstrate a strategy for integrating multi-channel EMG responses into a single side effect biomarker (the mEP score). Next, we use data from a large patient cohort of intraoperative recordings (N= 54 subthalamic nucleus (STN) leads) to quantify several physiological features of mEP, including their response frequency, latency, amplitude, and waveform similarity properties. Last, we show that the mEP score responds to DBS amplitude and contact configuration parameters in a manner that is consistent with expected STN-capsular anatomy.Significance.The results of this study inform an end-to-end approach for side effect biomarker measurement that could aid the precision and efficiency of surgical targeting and DBS programming.

Objective. This study assesses the signal quality of state-of-the-art dry electroencephalography (EEG) under highly challenging, uncontrolled, real-world conditions and compares it to conventional wet EEG.Approach. EEG data from 530 participants recorded during a public exhibition were benchmarked against several established signal quality metrics, including spiking activity, kurtosis, auto-mutual information, spectral entropy, gamma-band power, and parameters extracted using the fitting oscillations and one-over F model. Additionally, ICLabel decomposition was applied to quantify artifact influences across EEG channels. Dry electrode results were compared with their equivalents extracted on two control datasets comprising 71 and 80 participants, respectively, recorded with wet EEG systems in laboratory, home, or clinical surroundings.Main results. The analysis revealed condition-specific susceptibility to artifacts for both EEG modalities. The dry EEG system exhibited substantial robustness in moderate-noise scenarios, with artifact profiles comparable to controlled wet EEG recordings. However, recordings obtained in highly dynamic conditions showed increased muscle artifacts and broadband activity, notably in frontal and temporal regions. Wet EEG systems, under controlled conditions, were overall less inflicted by artifacts, yet, fronto-central ocular and muscular artifacts were consistently present. ICLabel analysis further confirmed these findings, indicating similar proportions of brain-related activity across systems (approximately 31%-49.5%), but highlighted increased vulnerability to movement and environmental artifacts in dry EEG during dynamic tasks.Significance. In agreement with recent similar investigations, our findings demonstrate that dry EEG caps have significantly matured, achieving signal quality comparable to wet EEG systems even in challenging real-world conditions, provided appropriate artifact mitigation strategies are employed. These results affirm the practical readiness and broad feasibility of dry EEG technologies for diverse Brain-computer interface applications in naturalistic environments.

Objective. Electroencephalography (EEG)-based brain-computer interfaces (BCIs) can help patients with disabilities control external devices directly without peripheral pathways. Due to the limitations in EEG signal quality, the performance of EEG-based BCIs may not be satisfactory. Shared control has become an important research direction in the field of brain-controlled wheelchairs (BCWs). However, most existing studies do not achieve the flexible movement of BCW in environments with narrow spaces. This study proposes a shared controller based on the potential field method to integrate environmental information and user commands intelligently.Approach. Considering the flexibility of wheelchair movement, we incorporated EEG decoding results obtained through the motor imagery paradigm and fused them with environmental information to create a fusion field. We then used these components separately to construct the BCI and obstacle fields. Twelve subjects participated in the virtual wheelchair navigation experiment, while five subjects took part in the real-world wheelchair navigation experiment, aiming to evaluate the control performance in different scenarios under three control modes (keyboard, BCI-only, and shared control).Main results. The experimental results show that the proposed shared controller: 1) significantly enhances navigation performance in both general and narrow environments compared with BCI-only control; 2) improves the total success rate from 8.33% to 83.33% in virtual complex environments and from 23.33% to 66.67% in real-world two-way navigation; 3) achieves success rates that are statistically comparable to keyboard control (p> 0.05). Moreover, the shared control reduced the average navigation time by nearly 100 s compared with BCI-only control in real-world experiments.Significance. This new shared control method improves the ability of BCWs to move flexibly in challenging, narrow environments.

Objective.Once we learn a reliable gait, we no longer have to consciously contract individual muscles to walk, or think about the fine-grained low-level control of our joints. Instead, we mainly make decisions on where we want to end up, at what pace and through which path. Estimating this high-level (HL) intent may provide the necessary input to wearable robotic devices to adapt to their user's needs. We introduce a continuous representation of locomotion goals and investigate how it may be estimated from muscle signals and body posture.Approach.This study investigated methods to estimate a representation of HL locomotion intent, the horizontal walking path. We collected full-body motion capture and bipolar surface electromyography data from 6 subjects during non-steady-state gait. We trained temporal convolutional networks to causally predict the walking path directly or parametrically with a critically damped trajectory model, using a mixture of muscle and body posture signals.Main results.We achieved a mean trajectory estimation accuracy for a 1-second walking path corresponding tor2=0.89using a multimodal model. We simultaneously provided estimates for current and desired walking velocities as constrained by the walking path model, aiding interpretability of the estimator's output.Significance.Our approach could provide user interfacing in a subject-independent format for wearable robotic devices. Moreover, this HL intent representation is flexible and able to be synthesized in virtual environments, where it can serve as a surrogate for biosignals of simulated intent-driven robotics.

Chronic brain sensing devices, such as the Medtronic Percept™ or Neuropace RNS system, record local field potentials (LFPs) that may be vulnerable to interference and noise due to hardware limitations, environmental factors, movement, stimulation, cardiac signals, and analytical procedures. Although onboard hardware filters can attenuate some unwanted signals, additional processing is often required. Here we demonstrate that cardiac artifacts significantly alter the power spectral density (PSD) of neural activity within the theta (4-8 Hz), alpha (8-12 Hz), and beta (12-30 Hz) bands. We introduce a time-domain template subtraction method specifically designed to remove QRS complex cardiac artifacts. Separately, we describe techniques for transforming time domain data to the frequency domain and mitigating transient artifacts by estimating background neural activity-either through window rejection based on PSD characteristics or via principal component analysis. Finally, we present an approach to isolate oscillatory neural activity by subtracting the aperiodic 1/fcomponent from the power spectrum by fitting the fitting oscillations and one over F logarithmic function. While filter selection must be tailored to the specific device and participant environment to avoid over-filtering, these interference and noise mitigation strategies are crucial for ensuring the integrity of LFP recordings.

Objective.To enhance frequency recognition in steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCIs), particularly under short data acquisition and complex environmental conditions.Approach.We propose multi-stimulus discriminant fusion analysis (MSDFA), a novel method that integrates multi-stimulus strategies with discriminant modeling. MSDFA was evaluated on two public datasets (Benchmark and BETA) and compared with conventional approaches including eCCA, eTRCA, and their variants.Main results.MSDFA consistently outperformed existing methods across different data lengths and training block quantities. It achieved maximum information transfer rates of 247.17 ± 10.15 bpm on the Benchmark dataset and 192.72 ± 9.44 bpm on the BETA dataset, demonstrating superior robustness and efficiency.Significance.By combining complementary algorithmic strengths, MSDFA improves adaptability to individual variability and complex environments, advancing the practical utility and reliability of SSVEP-BCI systems.