IEEE Transactions on Neural Systems and Rehabilitation Engineering最新文献

{"title":"Study on prosthetic hand proprioception feedback based on hybrid vibro-electrotactile stimulation.","authors":"Guangfei Wu, Wenqing Gu, Yi Luo, Xin Zhang, Lei Li, Jingming Hou, Haoyue Deng, Wensheng Hou, Lin Chen, Xing Wang","doi":"10.1109/TNSRE.2025.3593354","DOIUrl":"https://doi.org/10.1109/TNSRE.2025.3593354","url":null,"abstract":"<p><strong>Objective: </strong>Sensory substitution technologies represent a significant advancement in modern prosthetics, with hybrid tactile feedback approaches gaining increasing attention. However, limited studies have examined the role of hybrid tactile feedback in closed-loop prosthetic control, particularly in relation to varying visual conditions.</p><p><strong>Methods: </strong>This study employed a 2-factor mixed design involving 10 able-bodied participants and 3 transradial amputees. Three visual conditions--Optimal Vision, Limited Vision, and Blocked Vision were tested. Participants performed prosthetic wrist and hand position-matching tasks utilizing vibrotactile, electrotactile, and hybrid vibro-electrotactile (HyVE) feedback modes provided by a custom-developed prosthetic position feedback control system. Outcome measures included categorical analysis of task completion outcomes, control precision error (CPE), completion time (CT), and feedback preference.</p><p><strong>Results: </strong>The HyVE feedback mode elevated success rates across all visual conditions, while effectively reducing the incidence of errors and confusion trials. Compared to electrotactile mode, HyVE yielded significantly lower CPE, and compared to vibrotactile mode, it enabled significantly shorter CT. Furthermore, 7 out of 13 participants (>50%) selected HyVE as their preferred feedback mode.</p><p><strong>Conclusion: </strong>The proposed HyVE tactile feedback method effectively combines the advantages of electrotactile and vibrotactile stimulation, mitigating their respective limitations. It enables accurate perception of prosthetic motion without the need for continuous visual monitoring, showing significant potential for enhancing everyday prosthetic functionality.</p>","PeriodicalId":13419,"journal":{"name":"IEEE Transactions on Neural Systems and Rehabilitation Engineering","volume":"PP ","pages":""},"PeriodicalIF":5.2,"publicationDate":"2025-07-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144730098","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adaptive Neurofeedback Training Using a Virtual Reality Game Enhances Motor Imagery Performance in Brain-Computer Interfaces.","authors":"Kun Wang, Yuwei Liu, Feifan Tian, Weibo Yi, Yang Zhang, Tzyy-Ping Jung, Minpeng Xu, Dong Ming","doi":"10.1109/TNSRE.2025.3592988","DOIUrl":"https://doi.org/10.1109/TNSRE.2025.3592988","url":null,"abstract":"<p><p>Neurofeedback training (NFT) has been widely used in motor rehabilitation. However, NFT combined with motor imagery-based brain-computer interface (MI-BCI) faces challenges such as mental fatigue and non-personalized training strategies. Therefore, we proposed an adaptive NFT based on a VR game that simulates real-life motor tasks to improve training efficiency. We conducted a detailed comparative analysis of the efficiency of the VR-based NFT and traditional Graz-based NFT. Forty-eight healthy subjects were randomly assigned to five groups and underwent various NFT protocols. Among them, the subjects in the four experimental groups were required to perform the NFT three times over five days, including virtual or real scenarios, as well as unilateral or bilateral hands training. We evaluated training effects by analyzing EEG features and classification performance, while online recognition duration served as the primary measure for assessing the adaptive NFT strategy. EEG analysis showed that VR-based NFT significantly enhanced the Event-related desynchronization (ERD) activations in the sensorimotor cortices over five days. The VR-based NFT group achieved a classification accuracy of 81.85%, representing a 10.14% improvement from baseline, which exceeded the 6.43% increase observed in the Graz-based NFT group. Furthermore, implementing the adaptive NFT strategy reduced the mean task duration by over 30% compared to the fixed-time training protocol. The results demonstrated that the adaptive MI-BCI-based NFT in a VR game achieves superior training outcomes while reducing training duration. These findings suggest the promising potential for applying MI-BCI NFT with VR games in motor rehabilitation following a stroke.</p>","PeriodicalId":13419,"journal":{"name":"IEEE Transactions on Neural Systems and Rehabilitation Engineering","volume":"PP ","pages":""},"PeriodicalIF":5.2,"publicationDate":"2025-07-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144730097","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"TBI-Related EMG Characterization of Neuromuscular Responses to Anterior Perturbations While Standing.","authors":"Kiran K Karunakaran, Easter S Suviseshamuthu, Prasad Tendolkar, Guang H Yue, Rakesh Pilkar","doi":"10.1109/TNSRE.2025.3592477","DOIUrl":"https://doi.org/10.1109/TNSRE.2025.3592477","url":null,"abstract":"<p><p>Traumatic brain injury (TBI) causes deficits in sensory systems, sensorimotor integration, and/or neuromuscular response, thus impairing essential postural response mechanisms such as compensatory postural adjustments. This, in turn, results in balance deficits and increases the risk of falls, affecting the activities of daily living and quality of life. Therefore, the goal of this study is to quantify the differences in neuromuscular responses based on electromyography (EMG) between people with TBI (pwTBI) and age-matched healthy controls (HCs). We investigated the differences between eight HCs and nine pwTBI in the following EMG characteristics: muscle activity (EMG) onset, EMG burst area, and median frequency, in response to anterior (forward) platform perturbations at four different amplitudes during standing. The results showed delayed muscle activation onset, larger EMG bursts, and decreased EMG median frequency in pwTBI compared to HCs, suggesting an altered neuromuscular response to platform perturbations in pwTBI.</p>","PeriodicalId":13419,"journal":{"name":"IEEE Transactions on Neural Systems and Rehabilitation Engineering","volume":"PP ","pages":""},"PeriodicalIF":4.8,"publicationDate":"2025-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144707388","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Imagined Speech Detection Using Multi-Receptive CNN for Asynchronous BCI Communication and Neurorehabilitation","authors":"Byung-Kwan Ko;Seo-Hyun Lee;Seong-Whan Lee","doi":"10.1109/TNSRE.2025.3592312","DOIUrl":"10.1109/TNSRE.2025.3592312","url":null,"abstract":"Imagined speech-based brain-computer interface (BCI) facilitates brain signal-driven intuitive communication which holds great promise as an effective speech rehabilitation tool, enabling real-time, hands-free interaction for individuals with speech and motor impairments. While speech-based assistant systems rely on wake-word detection (e.g., “Hey Siri”), BCI-based communication system must capture imagined onset from EEG signals to turn on the ‘brain switch’ to further convey user’s imagined command. Nevertheless, the absence of reliable ground truth for the endogenous paradigm adds to the complexity to train the model to capture exact onset from continuous EEG. To address these issues, we introduce a multi-receptive field convolutional neural network, designed to capture speech and idle states based on behaviorally-aligned EEG features. We propose a voice-based ground truth alignment method with voting strategy that aims to synchronize imagined speech with overt speech onset and offset, providing a structured approach for capturing speech events in asynchronous BCI systems. Furthermore, spectral and phonological analyses revealed that beta and alpha bands, as well as syllable count, appear to influence speech state discriminability. Evaluations on imagined and overt speech tasks, including pseudo-online experiments, demonstrate the potential to enhance asynchronous BCI systems, supporting real-time communication for both healthy and impaired individuals.","PeriodicalId":13419,"journal":{"name":"IEEE Transactions on Neural Systems and Rehabilitation Engineering","volume":"33 ","pages":"2904-2914"},"PeriodicalIF":5.2,"publicationDate":"2025-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11095808","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144707387","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Brain Connectivity During Walking and Obstacle Avoidance in Persons with Multiple Sclerosis and Healthy Controls: A Pilot EEG Study.","authors":"Fares Al-Shargie, Michael Glassen, John DeLuca, Soha Saleh","doi":"10.1109/TNSRE.2025.3592492","DOIUrl":"https://doi.org/10.1109/TNSRE.2025.3592492","url":null,"abstract":"<p><p>This study investigated effective connectivity and hemispheric asymmetry in persons with multiple sclerosis (pwMS) compared to healthy controls (HC) during two walking conditions: walking alone and walking while avoiding unpredictable obstacles. Cognitive-motor interference (CMI) was analyzed using electroencephalography (EEG) across beta, alpha, and theta frequency bands. Directed functional connectivity was estimated using partial directed coherence (PDC) to assess differences in connectivity patterns between conditions and groups. In healthy controls, obstacle avoidance increased connectivity in motor and cognitive regions including left central (LC), left temporal (LT), and right frontal (RF) regions, p<0.0014. In contrast, pwMS demonstrated weaker and more localized connectivity, primarily in the left central regions (sensorimotor cortices) p<0.0013, suggesting reduced efficiency in brain networks and compensatory mechanisms to maintain task performance. Further, pwMS showed left laterality toward the central region during both walking conditions compared to HC, p<0.05. Correlational analysis revealed that connectivity during obstacle avoidance in HC positively correlated with comfortable walking speed (r = 0.57), indicating efficient neural pathways. In pwMS, connectivity showed a negative correlation with walking speed (r = -0.65), indicating compensatory but inefficient neural engagement. These findings highlight disruptions in brain connectivity during motor-cognitive tasks in pwMS, with potential implications for designing targeted rehabilitation strategies to improve gait and neural efficiency.</p>","PeriodicalId":13419,"journal":{"name":"IEEE Transactions on Neural Systems and Rehabilitation Engineering","volume":"PP ","pages":""},"PeriodicalIF":4.8,"publicationDate":"2025-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144707386","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Age, Motor Function, and Cognitive Function Influence Preferences for Telerehabilitation Mediated by a Social Robot Augmented with Telepresence.","authors":"Michael J Sobrepera, Anh T Nguyen, Ajay Anand, Laura A Prosser, Sally H Evans, Michelle J Johnson","doi":"10.1109/TNSRE.2025.3592020","DOIUrl":"https://doi.org/10.1109/TNSRE.2025.3592020","url":null,"abstract":"<p><p>Social robot augmented telepresence (SRAT) is a potential approach to provide rehabilitative care to remote patients, while overcoming barriers to physical clinician-patient interaction. This study evaluated the preference of the subjects, stratified by age, motor impairment level and cognitive impairment level, for three modes of rehabilitation care delivery: face-to-face (FTF), classical telepresence (CT), and via social robot-augmented classical telepresence (SRAT). Forty-two participants completed the experiment that included assessments of upper-limb motor function and cognitive function followed by simulated rehabilitation interaction sessions, where the FTF interaction was the first, followed by CT and SRAT interactions in randomized order. Participants completed surveys on their impression and experience receiving simulated care in each mode. Survey responses were analyzed using descriptive statistics and regression methods. Although in-person interaction (FTF) was the preferred option, 71% of subjects enjoyed and preferred SRAT over CT and this preference was mediated by age and severity of motor and cognitive impairment. Our analysis suggests that young children will rank SRAT above CT except for when they have severe cognitive impairment, adults will prefer SRAT less as their upper-limb impairment becomes more severe, and adults over 70 years old will prefer SRAT less if they have moderate to no upper-limb motor impairment and no cognitive impairment.</p>","PeriodicalId":13419,"journal":{"name":"IEEE Transactions on Neural Systems and Rehabilitation Engineering","volume":"PP ","pages":""},"PeriodicalIF":4.8,"publicationDate":"2025-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144698446","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multi-Class Decoding of Attended Speaker Direction Using Electroencephalogram and Audio Spatial Spectrum","authors":"Yuanming Zhang;Jing Lu;Fei Chen;Haoliang Du;Xia Gao;Zhibin Lin","doi":"10.1109/TNSRE.2025.3591819","DOIUrl":"10.1109/TNSRE.2025.3591819","url":null,"abstract":"Prior research on directional focus decoding, a.k.a. selective Auditory Attention Decoding (sAAD), has primarily focused on binary “left-right” tasks. However, decoding of the attended speaker’s precise direction is desired. Existing approaches often underutilize spatial audio information, resulting in suboptimal performance. In this paper, we address this limitation by leveraging a recent dataset containing two concurrent speakers at two of 14 possible directions. We demonstrate that models relying solely on EEG yield limited decoding accuracy in leave-one-out settings. To enhance performance, we propose to integrate spatial spectra as an additional input. We evaluate three model architectures, namely CNN, LSM-CNN, and Deformer, under two strategies for utilizing spatial information: all-in-one (end-to-end) and pairwise (two-stage) decoding. While all-in-one decoders directly take dual-modal inputs and output the attended direction, pairwise decoders first leverage spatial spectra to decode the competing pairs, and then a specific model is used to decode the attended direction. Our proposed all-in-one Sp-EEG-Deformer model achieves 14-class decoding accuracies of 55.35% and 57.19% in leave-one-subject-out and leave-one-trial-out scenarios, respectively, using 1-second decision windows (chance level: 50%, indicating random guessing). Meanwhile, the pairwise Sp-EEG-Deformer decoder achieves a 14-class decoding accuracy of 63.62% (10 s). Our experiments reveal that spatial spectra are particularly effective at reducing the 14-class problem into a binary one. On the other hand, EEG features are more discriminative and play a crucial role in precisely identifying the final attended direction within this reduced 2-class set. These results highlight the effectiveness of our proposed dual-modal directional decoding strategies.","PeriodicalId":13419,"journal":{"name":"IEEE Transactions on Neural Systems and Rehabilitation Engineering","volume":"33 ","pages":"2892-2903"},"PeriodicalIF":5.2,"publicationDate":"2025-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11091336","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144698359","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Differential Cortical Responses of Functional and Sensory Electrical Stimulation in Closed-Loop Tremor Suppression for Parkinson’s Disease","authors":"Xiaoqi Zhao;Tinglan Huang;Mengyue Jin;Hongbo Zhao;Yu Shi;Yanlin Wang;Xiao Shen;Zhen Li;Qingqing Shi;Xiaodong Zhu;Lin Meng","doi":"10.1109/TNSRE.2025.3591134","DOIUrl":"10.1109/TNSRE.2025.3591134","url":null,"abstract":"Functional electrical stimulation (FES) and sensory electrical stimulation (SES) are widely used in tremor suppression for Parkinson’s disease (PD), however, their therapeutic efficacy varies significantly across individuals. This study investigated the differential cortical effects of FES and SES during closed-loop tremor suppression in PD patient, aiming to identify neurophysiological biomarkers for guiding personalized neuro modulation strategies. We developed an inertial based closed-loop tremor suppression system that delivers out-of-phase FES and continuous SES based on real-time tremor detection. Fifteen PD patients were recruited in tremor suppression trials while surface electroencephalography (EEG) and inertial-based movements of hand and forearm were measured. Both FES and SES significantly reduced tremor amplitude, with FES showing overall greater suppression (hand suppression rate: 60.72% vs. 48.31%, p >0.05; forearm suppression rate: 62.25% vs. 54.41%, p >0.05) where substantial inter-individual variability was observed. EEG analysis revealed that FES induced contralateral beta-band event-related desynchronization (<inline-formula> <tex-math>$beta $ </tex-math></inline-formula>-ERD), whereas SES elicited beta-band event-related synchronization (<inline-formula> <tex-math>$beta $ </tex-math></inline-formula>-ERS). These distinct cortical response patterns were significantly correlated with tremor suppression performance (FES <inline-formula> <tex-math>$beta $ </tex-math></inline-formula>-ERD: r = -0.629, p = 0.012; SES <inline-formula> <tex-math>$beta $ </tex-math></inline-formula>-ERS: r = 0.679, p = 0.005). Resting-state spectral analysis further revealed modality-specific changes in alpha power across sensorimotor regions. These findings revealed functional neurodynamic signatures associated with individual responsiveness to stimulation. The observed <inline-formula> <tex-math>$beta $ </tex-math></inline-formula>-band oscillatory responses may serve as candidate biomarkers for predicting individual treatment outcomes, offering a potentially biomarker-guided approach for personalized neuromodulation for PD tremor.","PeriodicalId":13419,"journal":{"name":"IEEE Transactions on Neural Systems and Rehabilitation Engineering","volume":"33 ","pages":"2814-2822"},"PeriodicalIF":5.2,"publicationDate":"2025-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11087652","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144690086","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Novel Hybrid Brain–Computer Interface Integrating Motor Imagery and Multiple Visual Stimuli","authors":"Chao Zhang;Guojing Li;Xiaopei Wu;Xiangping Gao","doi":"10.1109/TNSRE.2025.3591616","DOIUrl":"10.1109/TNSRE.2025.3591616","url":null,"abstract":"Brain-Computer Interface (BCI) that integrate Motor Imagery (MI) with Steady-State Visual Evoked Potentials (SSVEP) or Overt Spatial Attention (OSA) have demonstrated superior performance compared to MI only BCI. Nonetheless, the exploration of BCI that combine MI with visual tasks remains limited, and the synchronization between MI and visual tasks is often weak. To address this gap, our study introduces a novel BCI paradigm that combines MI with two visual tasks: SSVEP and OSA. In this paradigm, dynamic images depicting left and right arm movements flash at distinct frequencies, serving as visual stimuli positioned on both sides of the screen. Four classification methods are used for testing. The MI+SSVEP+OSA paradigm achieves higher average accuracy than the MI, MI+SSVEP, and MI+OSA paradigms. This validates the effectiveness of our novel paradigm and confirms the feasibility of simultaneously integrating MI with two visual stimuli. Moreover, we observe that the integration of SSVEP offers significant improvements, especially for participants who exhibit limited performance in the MI only paradigm. Additionally, our results indicate comparable performance between the MI+SSVEP and MI+OSA paradigms. Overall, this study offers valuable insights that can guide future research in hybrid BCI development, paving the way for more efficient and user-friendly BCI.","PeriodicalId":13419,"journal":{"name":"IEEE Transactions on Neural Systems and Rehabilitation Engineering","volume":"33 ","pages":"2847-2857"},"PeriodicalIF":5.2,"publicationDate":"2025-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11090001","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144690085","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhanced Online Continuous Brain-Control by Deep Learning-Based EEG Decoding","authors":"Jiaheng Wang;Lin Yao;Yueming Wang","doi":"10.1109/TNSRE.2025.3591254","DOIUrl":"10.1109/TNSRE.2025.3591254","url":null,"abstract":"Objective: A growing amount of deep learning models for motor imagery (MI) decoding from electroencephalogram (EEG) have demonstrated their superiority over traditional machine learning approaches in offline dataset analysis. However, current online MI-based brain-computer interfaces (BCIs) still predominantly adopt machine learning decoders while falling short of high BCI performance. Yet, the generalization and advantages of deep learning-based EEG decoding in realistic BCI systems remain far unclear. Methods: We conduct a randomized and cross-session online MI-BCI study on 2D center-out tasks in 15 BCI-naive subjects. A newly proposed deep learning model named interactive frequency convolutional neural network (IFNet) is leveraged and rigorously compared with the prevailing benchmark namely filter-bank common spatial pattern (FBCSP) for online MI decoding. Results: Through extensive online analysis, the deep learning decoder consistently outperforms the classical counterpart across various performance metrics. In particular, IFNet significantly improves the average online task accuracy by 20% and 27% in two sessions compared with FBCSP, respectively. Moreover, a significant cross-session training effect is observed by the IFNet model (<inline-formula> <tex-math>${P}={0}.{017}$ </tex-math></inline-formula>) while not for the controlled method (<inline-formula> <tex-math>${P}={0}.{337}$ </tex-math></inline-formula>). Further offline evaluations also demonstrate the superior performance of IFNet over state-of-the-art deep learning models. Moreover, we present unique behavioral and neurophysiological insights underlying online brain-machine interaction. Conclusion: We present one of the first studies about online MI-BCIs using deep learning, achieving substantially enhanced online performance for continuous BCI control. Significance: This study suggests the good utility of deep learning in MI-BCIs and has implications for clinical applications such as stroke rehabilitation.","PeriodicalId":13419,"journal":{"name":"IEEE Transactions on Neural Systems and Rehabilitation Engineering","volume":"33 ","pages":"2834-2846"},"PeriodicalIF":5.2,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11087643","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144682561","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}