IEEE Transactions on Neural Systems and Rehabilitation Engineering最新文献

Objective: Transcranial magnetic stimulation (TMS), as a non-invasive means of neuromodulation, plays a crucial role in rehabilitation. Recent studies highlight that modeling the TMS-induced electric field (E-field) is essential to maximize the personalized treatment efficacy. Despite advancements in various E-field calculation methods, classic numerical calculation pipelines remain time-consuming and rely on whole head segmentation, and deep learning-based pipelines suffer from limited interpretability and stability.

Methods: We develop a comprehensive pipeline that supports both numerical methods and deep learning methods for TMS targeting and optimization based on local E-field, called PLED. This pipeline mainly consists of local image patch extraction, tissue segmentation, local E-field estimation, and coil placement optimization. Notably, prior information about tissue conductivity and primary E-field from the coil is embedded into the deep learning model.

Results: We have conducted extensive experiments on four datasets involving millions of local image patches in total. It is examined that our pipeline runs over 40 times faster on CPU and 100 times faster with GPU acceleration than classic numerical calculation pipelines for coil placement optimization. Meanwhile, compared to other deep learning-based pipelines, our pipeline achieves higher accuracy at most potential stimulation sites across the entire brain.

Conclusion: Our proposed pipeline enables rapid, accurate, and robust local E-field estimation and coil placement optimization.

Significance: Our pipeline would enhance stimulation efficacy and reduce data processing time in the precise personalized TMS treatment and rehabilitation.

Objective: The purpose of this paper is to present a chronic, preclinical, pilot study of a multielectrode auditory nerve implant. To date, the most successful auditory neuroprosthesis is the cochlear implant, but these devices have known limitations. Intraneural stimulation of the auditory nerve is a promising alternative, but preclinical investigations of multi-electrode devices have been limited to acute or semi-chronic studies.

Methods: Six cats were chronically implanted in the right auditory nerve with a 15-channel Utah Slant Array^TM. The array's position in the internal auditory canal was confirmed using computed tomography. Electrode impedance and evoked responses were recorded regularly during the indwelling period to evaluate performance.

Results: Two of the six implants met endpoint criteria at 6 months, with stable electrode impedances and robust evoked responses on at least 50% of their channels. Another two implants exhibited stable electrode impedances, but activation thresholds increased over three months. Two implants failed within the first two weeks due to device migration caused by connector-related issues.

Conclusion: This study reports the first successful chronic deployment of an multi-electrode auditory nerve implant in a preclinical model. Two implanted cats exhibited stable chronic performance, marked by a consistent set of electrodes capable of eliciting evoked responses. The failure modes of the remaining four cats are discussed and recommendations are made to stabilize implant placement. The study lays a solid foundation for future chronic safety and efficacy investigations as part of an effort to translate this new neuroprosthetic technology into the clinic.

With the global population aging, understanding the mechanisms of human quiet stance is crucial for preventing falls and improving quality of life. Although human quiet stance has been studied for decades, there is still no consensus on the underlying neural control mechanisms. Existing studies have proposed a variety of biomechanical simplifications and neural control models, but their fragmented development has led to persistent debates and different interpretations of experimental data. This review integrates these diverse approaches, deepens the understanding of human quiet stance, and provides practical insights for rehabilitation, prosthetic design, and humanoid robotics. We first survey biomechanical models, from the simple inverted pendulum to detailed musculoskeletal representations. Then we examine neural control models, including stiffness, continuous, intermittent, optimal control, and multisensory integration, which explain how stability is maintained. By directly comparing these models, the review clarifies the causes of existing challenges in the field and emphasizes the interconnections among different control models. Beyond theoretical significance, the review also discusses practical applications and identifies future research directions to guide the development of integrated neuromechanical models that combine biomechanical complexity with realistic neural control schemes.

Brain functional networks (BFNs) derived from multi-site fMRI data have been widely explored using transformer-based models to extract discriminative connectivity features for the diagnosis of psychiatric disorders, such as schizophrenia (SZ). However, existing transformer-based methods ignore physiological priors like the scale-free property, causing the attention to fail to capture the intrinsic topology of BFNs. Additionally, multisite heterogeneity makes source-free domain adaptation (DA) essential in clinical practice where data sharing is restricted. However, existing methods mainly rely on heuristic data augmentations or pseudo-labeling, without leveraging the intrinsic inter-regional dependencies, resulting in poor robustness to cross-site variability. To overcome these challenges, we proposed a source-free DA framework based on a scale-free transformer encoder for SZ classification. The transformer encoder was pre-trained on labeled source domains to capture discriminative connectivity patterns while integrating a scale-free prior to bias the attention toward hub nodes. The pretrained encoder and classifier were used to initialize the target model, where the encoder learned latent representations derived from inter-regional interactions for causal graph construction. To enhance robustness, the causal structure was perturbed via random permutation and counterfactual interventions, while entropy minimization jointly optimized the encoder and predictor to learn domain-invariant representations. Results showed that our method outperformed the other 4 transformer, 7 DA, 6 multi-site and 6 state-of-the-art methods across two SZ datasets (87.18%±0.91% and 88.39%±0.13%). Ablation results highlighted the contributions of the causal, permutation, counterfactual, and entropy minimization constraints to the performance improvement. Furthermore, the identified discriminative temporal regions provided insights into the dysfunctional neural-mechanisms in SZ.

Objective: Effective cross-subject decoding is essential for reducing calibration time and enhancing the practical usability of brain-computer interfaces (BCIs). However, large inter-subject variability in EEG features poses a major challenge, particularly for motor imagery (MI) paradigms. Recent studies have shown that rhythmic MI can induce steady-state movement-related rhythms (SSMRR), which provide more structured electrophysiological features than conventional sensorimotor rhythms (SMR) and may offer a promising basis for efficient cross-subject decoding.

Methods: In this study, we comprehensively explored ways to achieve high-performance cross-subject decoding based on the rhythmic MI paradigm from both model and data perspectives.

Results: We achieved an encouraging cross-subject four-class decoding accuracy of 72.94%±13.80% using a streamlined multilayer perceptron (MLP)-based network on a self-collected dataset comprising 100 BCI-naïve participants. From a model perspective, networks composed of simple MLP-based functional modules can achieve results comparable to, or even superior to, those of several state-of-the-art (SOTA) models. From a data perspective, increasing the training set size substantially improves cross-subject decoding performance (from 61.78% to 72.94%). Moreover, we revealed a strong positive correlation between EEG feature consistency and cross-subject decoding accuracy, providing a physiological explanation for why enlarging the training data scale enhances cross-subject generalization. Finally, we explored strategies for selecting high-quality training data. We found that feature-consistency-based selection serves as a more reliable criterion than within-subject decoding accuracy.

Significance: Overall, our study provides novel insights into cross-subject EEG decoding from the perspectives of model design, data scale and quality. The code is available in https://github.com/SJTUwyxuan/RhythmicMI-CrossSubject.

Phase-amplitude coupling (PAC), reflecting the modulation of high-frequency amplitude by the phase of lower-frequency oscillations, is increasingly recognized as a key mechanism underlying neural information processing. While PAC is typically associated with higher-order perceptual and cognitive processes, some studies explored PAC in relation to auditory steady-state responses (ASSR), a paradigm commonly used to assess gamma-band synchronization. However, findings from these studies remain inconclusive due to methodological variability and challenges in PAC analysis. In this study, we systematically investigated PAC in the EEG signal recorded during 40 Hz auditory steady-state stimulation using a rigorous analysis pipeline with three established PAC estimation methods: Mean Vector Length, Kullback-Leibler Modulation Index, and Phase-Locking Value. Our approach was validated on simulated EEG-like signals and applied to scalp EEG data from 12 participants (26.7±3.6 years, 5 females) in 40 Hz ASSR and resting-state (rsEEG) conditions. We found no significant differences in PAC between ASSR and rsEEG conditions in the ASSR-associated frontocentral region, regardless of PAC estimation methods. Furthermore, individual-specific peak PAC values and their associated frequencies showed no consistent patterns across conditions. These results suggest that PAC is not reliably elicited by auditory steady-state stimulation in EEG, challenging the utility of the ASSR paradigm for assessing PAC.

Psychological factors such as ownership, agency, and trust are critical to the acceptance and effective use of prosthetic devices, yet their relationship to control reliability remains underexplored. We investigated how induced delays and artificial malfunctions influence these factors during prosthesis simulator use in a fully immersive virtual reality environment. A Pasta Box Task was implemented in Unity with MuJoCo physics simulation, using surface electromyography myocontrol and integrated eye tracking to measure subjective and visuomotor responses. Thirty non-disabled participants completed six within-participant conditions crossing two control delay and three artificial malfunction levels. Validated questionnaires assessed ownership, agency, and trust, while gaze metrics quantified fixation percent, target-locking strategy, and eye arrival and leaving latencies. Both delay and malfunction significantly reduced psychometric scores, with artificial malfunctions exerting the largest overall effect, while delay particularly diminished agency. Artificial malfunctions also increased fixations on the prosthesis and altered gaze strategies, suggesting compensatory behavior. Delay primarily affected eye-arrival latency and the number of fixations, whereas artificial malfunctions influenced target-locking strategy and eye-leaving latency, indicating distinct visuomotor adaptations to each reliability factor. Weak but significant correlations emerged between gaze behavior and psychometric measures. The results of the experiment highlight the value of immersive, physics-accurate virtual reality as an early-stage platform for the controlled evaluation of myocontrol and prosthesis behavior and for capturing relevant psychometric and visuomotor indicators relevant to user-centered design.

In the contemporary society, mental fatigue is a major challenge to the efficiency of production and safety of daily life, highlighting the necessity of exploring effective intervention strategies. Here, 73 healthy adults (male/female = 36/37, age = 22.6 ± 2.2 yrs) were recruited to participate in a randomized, single-blind, sham-controlled study. Specifically, the high-frequency repetitive transcranial magnetic stimulation (rTMS) was applied to the left dorsolateral prefrontal cortex (DLPFC) in-between a sustained attention task to explore, for the first time, the feasibility and efficacy of promoting fatigue recovery. EEG data obtained during task were then projected to cortical space through source imaging, followed by functional brain network construction and quantitative graph theoretical analysis. Compared to the sham group, the real rTMS group exhibited preserved reaction times and significantly alleviated subjective worry. These behavioral benefits were accompanied by an intervention-related reorganization of brain networks. Specifically, rTMS-related modulation of long-range fronto-temporal connectivity and shifts in nodal centrality within the right fronto-parietal network and the posterior cingulate cortex (PCC) was also revealed. Further inspection of global graphical properties showed that brain network developed toward a higher segregation state after the intervention, as indicated by the increased local efficiency and clustering coefficient. Our findings indicate that high-frequency rTMS mitigates mental fatigue not through passive rest, but by promoting a functional transition of large-scale brain networks toward a more efficient topological state. This study confirmed the feasibility of rTMS as an effective intervention for fatigue recovery, and contributed to the development of precise, neurobiologicallybased strategies for fatigue intervention.

Neurogenic lower urinary tract dysfunction (NLUTD) represents a common and severe complication following spinal cord injury (SCI), profoundly impacting the quality of life and contributing to significant morbidity through recurrent infections and renal complications. Current management strategies impose significant compromises, with clean intermittent catheterization (CIC) being burdensome and showing poor long-term adherence, while sacral anterior root stimulation (SARS) typically restores emptying but lacks sensory feedback. In this study, we explore whether targeted electrical microstimulation of ascending bladder afferent pathways within the lumbar dorsolateral funiculus (DLF) can recreate afferent signals of bladder fullness. Using custom-fabricated intraspinal microelectrode arrays, we characterized the spatiotemporal patterns of bladder-responsive neural activity in the L2-L4 spinal segments of anesthetized rats during controlled bladder filling cycles. Mapping experiments revealed highly localized neural responses within DLF that exhibited robust firing patterns associated with bladder filling. Furthermore, patterned electrical microstimulation delivered to DLF coordinates corresponding to filling-responsive zones successfully triggered coordinated voiding in 91.7% of trials, characterized by appropriate intravesical pressure increases and rhythmic external urethral sphincter (EUS) activity. The evoked responses demonstrated remarkable spatial specificity without concurrent hindlimb motor activation, as contrasted with spinothalamic tract (STT) stimulation. These findings identify DLF as a promising anatomical substrate for targeted electrical microstimulation and establish proof-of-concept for the sensory component of a future closed-loop neuroprosthetic system aimed at restoring bladder function following SCI.

Computerized posturography has been employed to quantify an individual's intrinsic balance control under varying stances, thereby presenting the potential to support autonomous and ambulatory fall risk assessment when integrated with machine learning (ML) techniques. However, the superiority of posturography-based approaches over conventional methods such as questionnaires or physical performance tests remain insufficiently documented. In this study, we compared the predictive performance of various combinations of input data and introduced a novel ML approach that incorporates a Large Language Model (LLM) to enhance prediction while enabling feature-based, summarized explanations to improve the transparency of the predictions. We followed 206 community-dwelling older adults over a 6-month period to monitor fall events. At baseline, all participants completed a survey capturing demographic information, self-reported questionnaires, various physical performance tests, and four standing tasks assessed via tracker-based posturography. The predictive validity of these data in distinguishing fallers from non-fallers was evaluated using traditional ML models, and an LLM enhanced with Quantized Low-Rank Adaptation (QLoRA). The 6-month fall incidence was 16.9%. Traditional ML models achieved an area under the curve (AUC) ranging from 0.54 to 0.71 using different combinations of questionnaire responses, physical performance data, and posturographic parameters. Notably, a higher AUC (0.88) and accuracy (0.86) were achieved by applying the LLM with QLoRA to posturographic parameters alone. In conclusion, this study contributes to a deeper understanding of the relationship between postural control and fall risk, and demonstrates the potential of LLMs to improve predictive accuracy while minimizing the need for labor-intensive expert annotation.