Trigeminal neuralgia (TN) is a paroxysmal, recurrent, electric shock‑ or knife‑like severe pain localized to the sensory distribution of the trigeminal nerve branches in the face. Often referred to as “the most painful affliction known to humankind”. First-line drugs carbamazepine and oxcarbazepine are associated with long-term side effects and diminishing efficacy, highlighting an urgent need for new therapeutic strategies. This review systematically summarizes the current progress in TN animal models (ranging from chronic constriction and inflammatory models to genetically engineered models), the multi-brain-region interactive pain transmission circuitry, molecular targets (voltage-gated sodium/calcium/potassium channels, TRP channels, chemokines, and epigenetic regulators), and drug development. Current treatments primarily rely on ion channel modulators, while emerging strategies involve Nav1.7/Cav3.2 inhibitors, CGRP antibodies, NRF2 activators, and botulinum toxin. Future efforts should focus on establishing more clinically translational precision models, dissecting the spatiotemporal dynamics of neural circuits, promoting multi-target combination therapies based on molecular subtyping, and accelerating the transition from symptomatic control to curative treatment. Not applicable.

In this letter, we propose a novel upper limb rehabilitation framework based on dual-arm robotics for therapist- like traction training. Prioritizing patient safety, an 8-DOF kinematic model of the upper limb is derived to evaluate the reachable workspace of the palm center and proximal forearm during interaction with a dual-arm robot. Leveraging the characteristics of dual-arm rehabilitation, a non-redundant inverse kinematics method is proposed to constrain joint angles, thereby establishing a safety mechanism under dual constraints. Secondly, considering the training science and compliance, a potential field control strategy is introduced to enable the robot to learn the therapist's traction characteristics from a single demonstration. Combined with the leader-follower control, it reproduces the therapist's assistance and allows for compliant interaction. Experimental results show that the proposed framework combines the strong adaptability and comfort of end-effector robots with the precise rehabilitation of exoskeleton robots. As dual-arm and humanoid robots become more widely adopted, the proposed scheme holds promise for delivering therapist- like safe, scientific, and compliant rehabilitation in clinical and home settings.

Revealing how disrupted brain dynamics lead to altered consciousness levels remains a central challenge in understanding the neural mechanisms underlying consciousness. The brain criticality framework offers a promising perspective, in which optimal neural integration and information processing occur when the brain operates near a critical point, while also reflecting fundamental neural processes such as excitation/inhibition balance. Here, we combined resting-state functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) to systematically assess brain criticality in disorder of consciousness (DoC) patients. Our results revealed that patients in an unresponsive wakefulness state (UWS) exhibited significantly greater power-law scaling exponents in co-activation clusters, higher Ising energy, and lower phase synchronization compared to those in a minimally conscious state (MCS). These findings suggest a greater deviation from critical brain dynamics in UWS, reflecting diminished neural integration and increased disorder. The extent of these deviations correlated with metabolic deficits measured by PET, highlighting the functional relevance of altered neural dynamics. Importantly, critical metrics were significantly associated with clinical scores and outperformed PET in both diagnosis and prognosis. Together, our findings advance understanding of the neural mechanisms underlying consciousness and highlight the potential of criticality-based metrics for characterizing brain states and informing prognosis in DoC.

6D pose estimation from RGB-D data constitutes a pivotal research area in computer vision, where the primary challenge resides in effectively integrating RGB and depth modalities. We propose an innovative homogeneous multimodal cross-attention fusion framework for object 6D pose through directly processing raw RGB-D data for feature extraction rather than traditional point cloud-based two-branch architectures. We employ the global-local embeddings and adaptive cross-attention fusion to exploit the inherent similarity of homogeneous multimodal information. Furthermore, we design a confidence-aware keypoint evaluation module to enhance localization accuracy and robustness. Comparative analysis experiments on three popular benchmark datasets, complemented by systematic ablation analyses, demonstrate the efficacy of our method in achieving superior performance on Occlusion-LineMOD (79.6%), YCB-Video (97.2%), and MP6D (93.60%). Finally, we verify the applicability of our method in difficult conditions.

Gesture classification based on surface electromyography (sEMG) is crucial for controlling prosthetic hands. Most current models are trained with a limited set of offline sEMG data and are only capable of recognizing pre-learned, fixed gestures. However, human gestures are diverse, and those models fail to recognize new gestures not included during the training phase. This restricts the scalability of such models, and by extension, the functionality of myoelectric prosthetic hands. Incremental gesture recognition offers a promising solution to this challenge. Among various approaches, one-class envelope-type classifiers stand out for their exceptional ability to reject unknown categories and learn new categories. This paper presented a comprehensive comparison of six types of one-class classifiers using two public NinaPro databases (DB2 and DB5). Furthermore, we proposed a two-stage recognition model that integrated one-class classifiers with a multi-class classifier to enhance the overall performance of incremental gesture recognition. Real-time control experiments with a prosthetic hand demonstrated the potential of our approach for advancing clinical myoelectric prostheses.

Background: The clinical assessment of Disorders of Consciousness (DOC) has long been constrained by the subjectivity of behavioral scales and the low-temporal resolution of neuroimaging techniques. There is an urgent need for objective, high-temporal-resolution biomarkers to improve the accuracy of DOC severity evaluation and sub-state differentiation. This study aims to develop a resting-state/task-state dual-modality EEG microstate analysis method. By integrating a multisensory stimulation paradigm with a resting-state global template, we seek to verify the validity and clinical utility of this method in quantitatively assessing the severity of DOC sub-states, namely Minimally Conscious State-positive (MCS+), Minimally Conscious State-negative (MCS-), and Vegetative State (VS). Methods: A total of 27 subjects were enrolled, including 9 healthy controls (HC), 6 MCS+ patients, 6 MCS- patients, and 6 VS patients. A multisensory stimulation paradigm (visual, olfactory, and combined visual-olfactory) was applied, and EEG microstates were extracted using a revised K-means clustering algorithm. Key microstate parameters (duration, global field power, and coverage) were quantified for systematic analysis. Results: During the resting state, the HC group exhibited a significantly posterior parietal-dominant distribution of Microstate D, while this parameter showed a gradient attenuation pattern corresponding to the severity of consciousness impairment in the DOC group (p < 0.05). During the task state, the HC group showed a significant multisensory effect under combined visual-olfactory stimulation; within the DOC group, MCS+ patients demonstrated stronger task-related responses compared to MCS- and VS patients. Conclusions: The gradient attenuation of resting-state Microstate D parameters reflects the severity of DOC, and task-specific responses to multisensory stimulation serve as a potential biomarker for distinguishing MCS+ patients. This dual-modality EEG microstate analysis method provides an objective, high-temporal-resolution basis for the precise clinical evaluation of neurological function in DOC patients.

Prosthetic hands offer significant benefits for patients with hand amputations by partially replicating the function of real hands. However, most current prosthetics lack sensory feedback on movement, leading to a gap in proprioception for users. To bridge this gap and approximate the natural experience of hand use, prosthetic hands must offer detailed motion feedback. This paper introduces a non-invasive electrical stimulation approach, which can provide motion perception feedback through modeling muscle spindles. By employing transcutaneous electrical nerve stimulation (TENS), the method generates artificial sensory signals associated with the movement of a prosthetic hand, potentially restoring a degree of proprioception for patients with hand amputations. We developed an experimental framework involving an electronic prosthetic hand, an electrical stimulator, and surface electrodes to assess our approach. Five able-body and three forearm amputees took part in our experiments. The experimental results indicated that the subjects were able to accurately discern the movement angle of the prosthetic hand, and when the sensory feedback was biomimetic, the subjects were able to identify the prosthetic hand movement state better than using a traditional encoding algorithm that only relied on the current stimulation intensity.

Cancer-associated fibroblasts (CAFs) are crucial stromal cells in the tumor microenvironment, affecting cancer growth, angiogenesis, and matrix remodeling. Developing an effective in vitro tumor model that accurately recapitulates the dynamic interplay between tumor and stromal cells remains a challenge. In this study, a 3D bioprinted fibroblast - mediated heterogeneous breast tumor model was created, with tumor cells and fibroblasts in a bionic matrix. The impact of transforming growth factor-β （TGF-β） on the dynamic transformation of normal fibroblasts into CAFs and its subsequent influence on tumor cells is further investigated. These findings reveales a profound correlation between CAFs and several critical biological processes, including epithelial-mesenchymal transition (EMT), extracellular matrix (ECM) remodeling, gene expression profiles, and tumor progression. Furthermore, tumor models incorporating CAFs exhibits reduced drug sensitivity compared to models containing tumor cells alone or models co-cultured with normal fibroblasts. These results underscore the potential of the in vitro fibroblast-mediated heterogeneous tumor model to simulate real-life physiological conditions, thereby offering a more effective drug screening platform for elucidating tumor pathogenesis and facilitating drug design prior to animal and clinical trials. This model's establishment promotes the understanding of tumor-stromal interactions and their therapeutic implications.

The search for neural correlates of consciousness has always been a hot topic in the field of consciousness. Previous studies have shown that high-density transcranial direct current stimulation (HD-tDCS) can promote the recovery of consciousness level in patients with disorders of consciousness (DOC). This study intends to explore the neuromodulatory effects of HD-tDCS at different target from the perspectives of brain spatiotemporal dynamics and directed information flow, thereby clarifying the key brain areas for the origin of consciousness. The study prospectively used a high-density transcranial direct current stimulation (HD-tDCS) protocol to perform repeated stimulation at different target points for 46 DOC patients (F3 group, n = 11; sham group, n = 19; Pz group, n = 16). The Coma Recovery Scale-Revised (CRS-R) index was calculated, and microstate analysis and symbolic transfer entropy(STE) brain network construction were performed before and 14 days after HD-tDCS stimulation. We identified seven microstates with different spatial distributions of electrode activation. There were significant differences in microstates (including spatial activation patterns and brain dynamics) for different stimulation protocols. At the same time, the F3 group strengthened the information flow from the anterior to the posterior brain and internal information flow in the anterior brain in the theta band, while the Pz group mainly increased the information flow between the left frontal and parietal lobes in the theta and alpha bands. The study suggests that consciousness is the result of the comprehensive action of the anterior and posterior brain, and there is an overlap in the mechanisms of action between the two HD-tDCS stimulation protocols, suggesting that a multi-target combined stimulation protocol may be a potentially better stimulation scheme.

Interaction force estimation using surface electromyography (sEMG) is a popular technique for applications such as powered exoskeletons, robotics, and rehabilitation. However, since the sEMG is nonstationary, it is difficult to estimate high-dimensional interaction forces accurately during physical human-robot interaction (pHRI). This work proposes an end-to-end 6-D force estimation framework that can accurately predict forces during dynamic pHRI from cartesian and joint space. First, the sEMG, pose, and velocity of the upper limb during pHRI are synchronously acquired. Subsequently, sEMG and kinematic information are deeply fused by tensor fusion and cross-attention fusion. Finally, a spatio-temporal neural network (STNN) is utilized to extract the features of the fused information and estimate the interaction force. Six subjects interact with the robot at four stiffness: 100, 150, 200, and 250 N/m. The proposed multimodal fusion scheme achieves excellent performance in different types of STNNs and is validated in different kinematic spaces. Among them, the highest $R<^>{2}$ of 0.969 is achieved using ConvLG in cartesian space. Compared to solely employing sEMG, the $R<^>{2}$ of force estimation based on multiple modalities increases by 21.4%-42.0% ( $p\lt 0.05$ ). It shows the effectiveness of the presented approach and contributes a new way to estimate high-dimensional force during dynamic pHRI.

OBJECTIVE:This meta-analysis aims to compare robotic-assisted deep brain stimulation (RA-DBS) and non-robotic-assisted deep brain stimulation (nRA-DBS) regarding accuracy, efficiency and safety. METHODS:We searched six databases to retrieve relevant studies. Two independent reviewers selected the studies and assessed the risk of bias using the Cochrane risk-of-bias tool for randomized trials version 2 and the Methodological index for nonrandomized studies score. Statistical analysis was completed by Revman 5.4. RESULTS:A total of seven trials with 341 participants entered our analysis. Our meta-analysis showed that RA-DBS demonstrated a statistically significant reduction in target point error (MD: -0.30, 95%CI: [-0.58, -0.02], I2  = 0, P  = 0.03) and deviation outliers compared to nRA-DBS. (OR: 0.15, 95%CI: [0.04, 0.51], I2  = 0, P  = 0.002). RA-DBS and nRA-DBS demonstrated comparable efficiency metrics in terms of operation room time (MD: 19.37, 95%CI: [-62.85,102.59], I2  = 99%, P  = 0.65), operating time (MD: -17.04, 95%CI: [-84.95, 50.87], I2  = 98%, P  = 0.62) and total anesthesia time (MD: 14.24, 95%CI: [-96.26, -124.73], I2  = 97%, P  = 0.80). Two groups were comparable in terms of complication rates (OR: 1.79, 95%CI: [0.79, 4.05], I2  = 5%, P  = 0.17) and intracranial hemorrhage rates (OR: 0.80, 95%CI: [0.23, 2.74], I2  = 0, P  = 0.72). CONCLUSIONS:RA-DBS exhibits efficiency and safety comparable to nRA-DBS, serving as a viable alternative to nRA-DBS. Although RA-DBS shows promise in accuracy, further high-quality studies are needed to establish its clinical superiority.

Brain and central nervous system (CNS) cancers impose a substantial and growing disease burden in China, marked by elevated mortality and disability rates. Understanding disparities in cancer control strategies between China and developed countries (the US, the UK, and Japan) may inform evidence-based policy improvements. Using Global Burden of Disease (GBD) 2021 data, we analyzed incidence, prevalence, mortality, and disability-adjusted life years (DALYs) for brain and CNS cancers (1990–2021). Trends were assessed via estimated annual percentage change (EAPC), while frontier analysis evaluated disease burden reduction capacity relative to socioeconomic development. Health inequalities were measured using the Slope Inequality Index (SII) and Concentration Index. Future trends were projected via ARIMA models. In 2021, China reported 105,541 new cases and 68,911 deaths. While China had lower age-standardized incidence (ASIR) and prevalence (ASPR) rates than comparison nations, its DALYs rate was higher. From 1990 to 2021, ASIR and ASPR increased universally, whereas age-standardized mortality (ASMR) and DALYs (ASDR) declined in China, the US, and the UK but rose in Japan. High-SDI regions exhibited greater burden mitigation capacity, with widening cross-country inequalities. Projections indicate rising ASIR in all nations by 2036. Despite declining mortality, China’s brain and CNS cancer burden remains disproportionately high. Policymakers should integrate effective strategies from developed nations while tailoring interventions to China’s unique epidemiological and healthcare context.

Background and Objective: Emotions are an integral part of our daily lives and have a significant impact on brain neural activities. However, most traditional studies on electroencephalogram (EEG) emotions have focused on identification and classification, without effectively predicting future EEG based on existing emotional EEG signals. This paper introduces a spatial-temporal graph convolutional neural network (STGCNN) to integrate temporal and spatial features of EEG, capturing more emotional information. Methods: After calculating the brain functional network operating Pearson correlation, it is utilized as the input information of the model together with the frequency domain differential entropy (DE). Information prediction in the spatiotemporal domain is mainly achieved through two cross structures of two temporal convolutions and one spatial convolution. Meanwhile, an ablation experiment is designed to illustrate the practicality of the model. Results and Conclusion: The performance of the model is evaluated on the SEED dataset for mixed and single emotions across three sessions and three frequency bands by metrics: Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Weighted Mean Absolute Percentage Error (WMAPE). All performance metrics confirm the predictability of EEG-based emotions, demonstrating the effectiveness of STGCNN for the prediction of EEG. Significance: Accurately emotional EEG prediction is crucial for brain neural rehabilitation training and forecasting.

In the EEG-based emotion recognition task, although the multi-channel acquisition method has advantages, it brings operational difficulty and increases the mental stress of the subjects, which affects the data quality. To effectively mine the emotional information in EEG and select the optimal Critical channel, we propose a novel critical channel selection framework for EEG emotion recognition: TFBOC-WOHDNN, and conduct extensive experiments on the SEED dataset. First, using the brain function network topological feature as a new independent evaluation index is proposed to filter 30 suboptimal critical channels in four brain regions. To effectively model the spatiotemporal dependence of EEG signals, a hybrid deep model WOHDNN is proposed that can automatically find the optimal hyperparameters. On this basis, an improved binary optimization channel selection method is proposed to effectively screen the optimal critical channels for a specified number of channels. We found that high-band gamma and beta contribute more to EEG emotion recognition, the temporal lobe and frontal lobe are critical brain regions. The selected 4-channel, 8-channel, and 12-channel optimal critical channel schemes achieved recognition accuracy rates of 91.59/0.97%, 97.33/0.70%, and 99.97/0.03%. Our schemes are significantly better than existing channel solutions, with stable and robust recognition performance, and excellent generalization performance in DEAP-based recognition tasks. This provides a new reference for the development of wearable and portable EEG sensor devices and promotes the practical application of emotion recognition systems.

IntroductionEEG-based emotion recognition has gradually become a new research direction, known as affective Brain-Computer Interface (aBCI), which has huge application potential in human-computer interaction and neuroscience. However, how to extract spatio-temporal fusion features from complex EEG signals and build learning method with high recognition accuracy and strong interpretability is still challenging.MethodsIn this paper, we propose a hybrid attention spatio-temporal feature fusion network for EEG-based emotion recognition. First, we designed a spatial attention feature extractor capable of merging shallow and deep features to extract spatial information and adaptively select crucial features under different emotional states. Then, the temporal feature extractor based on the multi-head attention mechanism is integrated to perform spatio-temporal feature fusion to achieve emotion recognition. Finally, we visualize the extracted spatial attention features using feature maps, further analyzing key channels corresponding to different emotions and subjects.ResultsOur method outperforms the current state-of-the-art methods on two public datasets, SEED and DEAP. The recognition accuracy are 99.12% ± 1.25% (SEED), 98.93% ± 1.45% (DEAP-arousal), and 98.57% ± 2.60% (DEAP-valence). We also conduct ablation experiments, using statistical methods to analyze the impact of each module on the final result. The spatial attention features reveal that emotion-related neural patterns indeed exist, which is consistent with conclusions in the field of neurology.DiscussionThe experimental results show that our method can effectively extract and fuse spatial and temporal information. It has excellent recognition performance, and also possesses strong robustness, performing stably across different datasets and experimental environments for emotion recognition.

Background Cholangiocarcinoma (CCA) is a biliary epithelial malignant tumor with an increasing incidence worldwide. Therefore, further understanding of the molecular mechanisms of CCA progression is required to identify new therapeutic targets. Methods The expression of RPL35A in CCA and para-carcinoma tissues was detected by immunohistochemical staining. IP-MS combined with Co-IP identified downstream proteins regulated by RPL35A. Western blot and Co-IP of CHX or MG-132 treated CCA cells were used to verify the regulation of HSPA8 protein by RPL35A. Cell experiments and subcutaneous tumorigenesis experiments in nude mice were performed to evaluate the effects of RPL35A and HSPA8 on the proliferation, apoptosis, cell cycle, migration of CCA cells and tumor growth in vivo. Results RPL35A was significantly upregulated in CCA tissues and cells. RPL35A knockdown inhibited the proliferation and migration of HCCC-9810 and HUCCT1 cells, induced apoptosis, and arrested the cell cycle in G1 phase. HSPA8 was a downstream protein of RPL35A and overexpressed in CCA. RPL35A knockdown impaired HSPA8 protein stability and increased HSPA8 protein ubiquitination levels. RPL35A overexpression promoted CCA cell proliferation and migration. HSPA8 knockdown inhibited CCA cell proliferation and migration, and reversed the promoting effect of RPL35A. Furthermore, RPL35A promoted tumor growth in vivo. In contrast, HSPA8 knockdown suppressed tumor growth, while was able to restore the effects of RPL35A overexpression. Conclusion RPL35A was upregulated in CCA tissues and promoted the progression of CCA by mediating HSPA8 ubiquitination.

Current surface electromyography (sEMG)-based gesture recognition only extracts time or frequency features from raw sEMG signals, and then puts the features together to generate sample vectors, which are further used as inputs to build fixed classification models. This way may bring out two issues. First, raw sEMG signals are often acquired from multichannel electrodes. Only extracting time or frequency features will lose the spatial topology information between different channels, and cannot reflect the movement synergy of different muscles, causing relatively low recognition accuracies. Second, fixed classifiers only recognize fixed gestures, and cannot handle dynamically increasing gestures, limiting the scalabilities of classifiers in applications. To this end, we introduce a myoelectric manifold representation based on the symmetric positive definite (SPD) matrix to express the spatial synergy of multiple muscles. Then, the growing neural gas network is extended to the SPD manifold space, and uses myoelectric matrices as inputs to realize the incremental gesture recognition, maintaining the space topology with very few prototypes. Extensive experiments were conducted on two public databases (Ninapro DB2 and DB5) and a self-collection database. Experimental results showed that our method was superior to current methods, increasing recognition accuracy by 1.63%–11.89%, and can continuously grow its recognition ability online, revealing the potential in implementing myoelectric interaction systems.

Introduction: Functional connectivity across large-scale networks is crucial for the regulation of conscious states. Nonetheless, our understanding of potential alterations in the temporal dynamics of dynamic functional connectivity (dFC) among patients with disorders of consciousness (DOC) remains limited. The present study aimed to examine different time-scale spatiotemporal dynamics of electroencephalogram oscillation amplitudes recorded in different consciousness states. Methods: Resting-state electroencephalograms were collected from a cohort of 90 patients with DOC. The sliding window approach was used to create dFC matrices, which were subsequently subjected to kmeans clustering to identify distinct states. Finally, we performed state analysis and developed a decoding model to predict consciousness. Results: There was significantly lower dFC within the forebrain network in patients with unresponsive wakefulness syndrome than in those with a minimally conscious state. Moreover, there were significant differences in temporal properties, mean dwell time, and the number of transitions in the highfrequency band at different time scales between the unresponsive wakefulness syndrome and minimally conscious state groups. Using the multi-band and multi-range temporal dynamics of dFC approach, satisfactory classification accuracy (approximately 83.3 %) was achieved. Conclusion: Loss of consciousness is accompanied by an imbalance of complex dynamics within the brain. Both transitions between states at short and medium time scales in high-frequency bands and the forebrain are important in consciousness recovery. Together, our findings contribute to a better understanding of brain network alterations in patients with DOC. (c) 2024 The Author(s). Published by Elsevier Ltd on behalf of Tsinghua University Press. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Using noninvasive technology to control prosthetics remains a real-life challenge due to the low acceptability of prosthetics stemming from their unnatural motion trajectory. The continuous angle estimation of fingers based on surface electromyography (sEMG) can effectively improve the unnatural motion trajectory. This article proposes the transformer using shifted windows and convolution networks (SCTNets) to achieve high-precision, fine-grained, and high naturalness finger continuous angle estimation. The shifted windows are used to encrease receptive field and capture multiscale sEMG signals. The transformer is used to capture the long-term dependencies within multiscale sEMG signals, facilitating the extraction of global features, while CNN is used to extract local features. We use 20 subjects of the Ninapro DB2 dataset to test the model, and compared with LS-TCN, CNN-Attention, ConvLSTM and long short-term memory network (LSTM) models. The Pearson correlation coefficient (PCC), the normalized root-mean-square error (NRMSE), and the coefficient of determination (R-2) of the SCTNet was 84.35%, 8.85%, and 69.65%. Compared with the indicator results of other models, the SCTNet has improved by 1% (PCC), 0.15% (NRMSE), and 2.13% (R-2) in each index, and achieved competitive results in real-time computing performance. The results indicate that the SCTNet can effectively estimate the natural continuous finger joint angles.