Journal of neural engineering最新文献

Objective. Vagus nerve stimulation (VNS), which involves a surgical procedure to place electrodes directly on the vagus nerve (VN), is approved clinically for the treatment of epilepsy, depression, and to facilitate rehabilitation in stroke. VNS at surgically implanted electrodes is often limited by activation of motor nerve fibers near and within the VN that cause neck muscle contraction. In this study we investigated endovascular VNS that may allow activation of the VN at locations where the motor nerve fibers are not localized.Approach. We used endovascular electrodes within the nearby internal jugular vein (IJV) to electrically stimulate the VN while recording VN compound action potentials (CAPs) and neck muscle motor evoked potentials (MEPs) in an acute intraoperative swine experiment.Main Results. We show that the stimulation electrode position within the IJV is critical for efficient activation of the VN. We also demonstrate use of fluoroscopy (cone beam CT mode) and ultrasound to determine the position of the endovascular stimulation electrode with respect to the VN and IJV. At the most effective endovascular stimulation locations tested, thresholds for VN activation were several times higher than direct stimulation of the nerve using a cuff electrode; however, this work demonstrates the feasibility of VNS with endovascular electrodes and provides tools to optimize endovascular electrode positions for VNS.Significance. This work lays the foundation to develop endovascular VNS strategies to stimulate at VN locations that would be otherwise too invasive and at VN locations where structures such as motor nerve fibers do not exist.

Objectives. Recent event-based analyses of transient neural activities have characterized the oscillatory bursts as a neural signature that bridges dynamic neural states to cognition and behaviors. Following this insight, our study aimed to (1) compare the efficacy of common burst detection algorithms under varying signal-to-noise ratios and event durations using synthetic signals and (2) establish a strategic guideline for selecting the optimal algorithm for real datasets with undefined properties.Approach.We tested the robustness of burst detection algorithms using a simulation dataset comprising bursts of multiple frequencies. To systematically assess their performance, we used a metric called 'detection confidence', quantifying classification accuracy and temporal precision in a balanced manner. Given that burst properties in empirical data are often unknown in advance, we then proposed a selection rule to identify an optimal algorithm for a given dataset and validated its application on local field potentials of basolateral amygdala recorded from male mice (n=8) exposed to a natural threat.Main Results.Our simulation-based evaluation demonstrated that burst detection is contingent upon event duration, whereas accurately pinpointing burst onsets is more susceptible to noise level. For real data, the algorithm chosen based on the selection rule exhibited superior detection and temporal accuracy, although its statistical significance differed across frequency bands. Notably, the algorithm chosen by human visual screening differed from the one recommended by the rule, implying a potential misalignment between human priors and mathematical assumptions of the algorithms.Significance.Therefore, our findings underscore that the precise detection of transient bursts is fundamentally influenced by the chosen algorithm. The proposed algorithm-selection rule suggests a potentially viable solution, while also emphasizing the inherent limitations originating from algorithmic design and volatile performances across datasets. Consequently, this study cautions against relying solely on heuristic-based approaches, advocating for a careful algorithm selection in burst detection studies.

Objective.Many recent studies investigating the processing of continuous natural speech have employed electroencephalography (EEG) due to its high temporal resolution. However, most of these studies explored the response mechanism limited to the electrode space. In this study, we intend to explore the underlying neural processing in the source space, particularly the dynamic functional interactions among different regions during neural entrainment to speech.Approach.We collected 128-channel EEG data while 22 participants listened to story speech and time-reversed speech using a naturalistic paradigm. We compared three different strategies to determine the best method to estimate the neural tracking responses from the sensor space to the brain source space. After that, we used dynamic graph theory to investigate the source connectivity dynamics among regions that were involved in speech tracking.Main result.By comparing the correlations between the predicted neural response and the original common neural response under the two experimental conditions, we found that estimating the common neural response of participants in the electrode space followed by source localization of neural responses achieved the best performance. Analysis of the distribution of brain sources entrained to story speech envelopes showed that not only auditory regions but also frontoparietal cognitive regions were recruited, indicating a hierarchical processing mechanism of speech. Further analysis of inter-region interactions based on dynamic graph theory found that neural entrainment to speech operates across multiple brain regions along the hierarchical structure, among which the bilateral insula, temporal lobe, and inferior frontal gyrus are key brain regions that control information transmission. All of these information flows result in dynamic fluctuations in functional connection strength and network topology over time, reflecting both bottom-up and top-down processing while orchestrating computations toward understanding.Significance.Our findings have important implications for understanding the neural mechanisms of the brain during processing natural speech stimuli.

Objective.The steady-state visual evoked potential (SSVEP)-based brain-computer interface has received extensive attention in research due to its simple system, less training data, and high information transfer rate. There are currently two prominent methods dominating the classification of SSVEP signals. One is the knowledge-based task-related component analysis (TRCA) method, whose core idea is to find the spatial filters by maximizing the inter-trial covariance. The other is the deep learning-based approach, which directly learns a classification model from data. However, how to integrate the two methods to achieve better performance has not been studied before.Approach.In this study, we develop a novel algorithm named TRCA-Net (TRCA-Net) to enhance SSVEP signal classification, which enjoys the advantages of both the knowledge-based method and the deep model. Specifically, the proposed TRCA-Net first performs TRCA to obtain spatial filters, which extract task-related components of data. Then the TRCA-filtered features from different filters are rearranged as new multi-channel signals for a deep convolutional neural network (CNN) for classification. Introducing the TRCA filters to a deep learning-based approach improves the signal-to-noise ratio of input data, hence benefiting the deep learning model.Main results.We evaluate the performance of TRCA-Net using two publicly available large-scale benchmark datasets, and the results demonstrate the effectiveness of TRCA-Net. Additionally, offline and online experiments separately testing ten and five subjects further validate the robustness of TRCA-Net. Further, we conduct ablation studies on different CNN backbones and demonstrate that our approach can be transplanted into other CNN models to boost their performance.Significance.The proposed approach is believed to have a promising potential for SSVEP classification and promote its practical applications in communication and control. The code is available athttps://github.com/Sungden/TRCA-Net.

Objective. Brain source reconstruction through electroencephalogram is a challenging issue in brain research with possible applications in cognitive science as well as brain damage and dysfunction recognition. Its goal is to estimate the location of each source in the brain along with the signal being produced.Approach. In this paper, by assuming a small number of band limited sources, we propose a novel method for the problem by using successive multivariate variational mode decomposition (SMVMD). Our new method can be considered as a blind source estimation method, which means that it is capable of extracting the source signal without the knowledge of the location of the source or its lead field vector. In addition, the source location can be determined through comparing the mixing vector found in SMVMD and the lead filed vectors of the entire brain.Main results. The simulations verify that our method leads to performance improvement in comparison to the well-known localization and source signal estimation techniques such as MUltiple SIgnal Calssification (MUSIC), recursively applied and projected MUSIC, dipole fitting method, MV beamformer, and standardized low-resolution brain electromagnetic tomography.Significance. The proposed method enjoys low computational complexity. Moreover, our investigations on some experimental epileptic data confirm its superiority over the MUSIC method in the aspect of localization accuracy.

Neuropixels probes have become a crucial tool for high-density electrophysiological recordings. Although most research involving these probes is in acute preparations, some scientific inquiries require long-term recordings in freely moving animals. Recent reports have presented prosthesis designs for chronic recordings, but some of them do not allow for probe recovery, which is desirable given their cost. Others appear to be fragile, as these articles describe numerous broken implants.Objective.This fragility presents a challenge for recordings in rats, particularly in epilepsy models where strong mechanical stress impinges upon the prosthesis. To overcome these limitations, we sought to develop a new prosthesis for long-term electrophysiological recordings in healthy and epileptic rats.Approach.We present a new prosthesis specifically designed to protect the probes from strong shocks and enable the safe retrieval of probes after experiments.Main results.This prosthesis was successfully used to record from healthy and epileptic rats for up to three weeks almost continuously. Overall, 10 out of 11 probes could be successfully retrieved with a retrieval and reuse success rate of 91%.Significance.Our design and protocol significantly improved previously described probe recycling performances and prove usage on epileptic rats.

Objective.Peripheral nerve interfaces have the potential to restore sensory, motor, and visceral functions. In particular, intraneural interfaces allow targeting deep neural structures with high selectivity, even if their performance strongly depends upon the implantation procedure and the subject's anatomy. Currently, few alternatives exist for the determination of the target subject structural and functional anatomy, and statistical characterizations from cadaveric samples are limited because of their high cost. We propose an optimization workflow that can guide both the pre-surgical planning and the determination of maximally selective multisite stimulation protocols for implants consisting of several intraneural electrodes, and we characterize its performance in silico. We show that the availability of structural and functional information leads to very high performances and allows taking informed decisions on neuroprosthetic design.Approach.We employ hybrid models (HMs) of neuromodulation in conjunction with a machine learning-based surrogate model to determine fiber activation under electrical stimulation, and two steps of optimization through particle swarm optimization to optimize in silico implant geometry, implantation and stimulation protocols using morphological data from the human median nerve at a reduced computational cost.Main results.Our method allows establishing the optimal geometry of multi-electrode transverse intra-fascicular multichannel electrode implants, the optimal number of electrodes to implant, their optimal insertion, and a set of multipolar stimulation protocols that lead in silico to selective activation of all the muscles innervated by the human median nerve.Significance.We show how to use effectively HMs for optimizing personalized neuroprostheses for motor function restoration. We provide in-silico evidences about the potential of multipolar stimulation to increase greatly selectivity. We also show that the knowledge of structural and functional anatomies of the target subject leads to very high selectivity and motivate the development of methods for theirin vivocharacterization.

Objective.The evaluation of animals' motion behavior has played a vital role in neuromuscular biomedical research and clinical diagnostics, which reflects the changes caused by neuromodulation or neurodamage. Currently, the existing animal pose estimation methods are unreliable, unpractical, and inaccurate.Approach.Data augmentation (random scaling, random standard deviation Gaussian blur, random contrast, and random uniform color quantization) is adopted to augment image dataset. For the key points recognition, we present a novel efficient convolutional deep learning framework (PMotion), which combines modified ConvNext using multi-kernel feature fusion and self-defined stacked Hourglass block with SiLU activation function.Main results.PMotion is useful to predict the key points of dynamics of unmarked animal body joints in real time with high spatial precision. Gait quantification (step length, step height, and joint angle) was performed for the study of lateral lower limb movements with rats on a treadmill.Significance.The performance accuracy of PMotion on rat joint dataset was improved by 1.98, 1.46, and 0.55 pixels compared with deepposekit, deeplabcut, and stacked hourglass, respectively. This approach also may be applied for neurobehavioral studies of freely moving animals' behavior in challenging environments (e.g.Drosophila melanogasterand openfield-Pranav) with a high accuracy.

Objective. Electrical stimulation of visual cortex via a neuroprosthesis induces the perception of dots of light ('phosphenes'), potentially allowing recognition of simple shapes even after decades of blindness. However, restoration of functional vision requires large numbers of electrodes, and chronic, clinical implantation of intracortical electrodes in the visual cortex has only been achieved using devices of up to 96 channels. We evaluated the efficacy and stability of a 1024-channel neuroprosthesis system in non-human primates (NHPs) over more than 3 years to assess its suitability for long-term vision restoration.Approach.We implanted 16 microelectrode arrays (Utah arrays) consisting of 8 × 8 electrodes with iridium oxide tips in the primary visual cortex (V1) and visual area 4 (V4) of two sighted macaques. We monitored the animals' health and measured electrode impedances and neuronal signal quality by calculating signal-to-noise ratios of visually driven neuronal activity, peak-to-peak voltages of the waveforms of action potentials, and the number of channels with high-amplitude signals. We delivered cortical microstimulation and determined the minimum current that could be perceived, monitoring the number of channels that successfully yielded phosphenes. We also examined the influence of the implant on a visual task after 2-3 years of implantation and determined the integrity of the brain tissue with a histological analysis 3-3.5 years post-implantation.Main results. The monkeys remained healthy throughout the implantation period and the device retained its mechanical integrity and electrical conductivity. However, we observed decreasing signal quality with time, declining numbers of phosphene-evoking electrodes, decreases in electrode impedances, and impaired performance on a visual task at visual field locations corresponding to implanted cortical regions. Current thresholds increased with time in one of the two animals. The histological analysis revealed encapsulation of arrays and cortical degeneration. Scanning electron microscopy on one array revealed degradation of IrOxcoating and higher impedances for electrodes with broken tips.Significance. Long-term implantation of a high-channel-count device in NHP visual cortex was accompanied by deformation of cortical tissue and decreased stimulation efficacy and signal quality over time. We conclude that improvements in device biocompatibility and/or refinement of implantation techniques are needed before future clinical use is feasible.

Objective and Significance:This paper proposes an LSTM-enhanced multi-view dynamic emotion graph representation model, which not only integrates the relationship between electrode channels into electroencephalogram (EEG) signal processing to extract multi-dimensional spatial topology information but also retains abundant temporal information of EEG signals.Approach:Specifically, the proposed model mainly includes two branches: a dynamic learning of multiple graph representation information branch and a branch that could learn the time-series information with memory function. First, the preprocessed EEG signals are input into these two branches, and through the former branch, multiple graph representations suitable for EEG signals can be found dynamically, so that the graph feature representations under multiple views are mined. Through the latter branch, it can be determined which information needs to be remembered and which to be forgotten, so as to obtain effective sequence information. Then the features of the two branches are fused via the mean fusion operator to obtain richer and more discriminative EEG spatiotemporal features to improve the performance of signal recognition.Main results:Finally, extensive subject-independent experiments are conducted on SEED, SEED-IV, and Database for Emotion Analysis using Physiological Signals datasets to evaluate model performance. Results reveal the proposed method could better recognize EEG emotional signals compared to other state-of-the-art methods.