ERD Full-process Longitudinal Trend and Pre-post Motor Recovery Under BCI-controlled Sixth-finger Neurofeedback Intervention in Stroke Patients

ERD Full-process Longitudinal Trend and Pre-post Motor Recovery Under BCI-controlled Sixth-finger Neurofeedback Intervention in Stroke Patients | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article ERD Full-process Longitudinal Trend and Pre-post Motor Recovery Under BCI-controlled Sixth-finger Neurofeedback Intervention in Stroke Patients Wang Zhuang, Liu Yuan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6027743/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Brain-computer interface (BCI) is used in stroke rehabilitation to match brain activity with contingent feedback to establish closed-loop pathways and provide a measure of neuroplasticity changes of patients. However, most studies assessed neural function only at pre- and post-intervention, thereby longitudinal trends of neural patterns and mechanismsduring full-process of intervention remain unclear. Methods Forty stroke patients were recruited to receive a total of 8 sessions motor imagery-based (MI-based) BCI-controlled “sixth-finger” neurofeedback intervention, 4 sessions per week for 2 weeks. Electroencephalography (EEG) measure and clinical scales were assessed at three time points: baseline, post-train and 1month follow-up period, and EEG data of each intervention sessions were also tracked. ERD phenomenon induced by MI and resting-state functional connectivity were used to reflect the longitudinal trends and pre-post changes in neural activity. The upper extremity Fugl-Meyer assessment (FMA-UE) and Barthel index (BI) were used to reflect the motor improvement. Results EEG longitudinal trend shows three phases over full-process of intervention: ERD gradually increased in the first week of training, weakened and focused on the contralateral sensorimotor area in the second week, and showed a significant correlation over sessions, remaining focused and contralateral pattern in the follow-up period. And resting-state functional connectivity increased after intervention. Motor function between pre-post intervention showed significant improvement by clinical metrics, with + 7.9 in FMA-UE and + 7.1 in BI. More than half of patients (9/14) reached the minimally clinically important difference (MCID) of 6.6 points change for FMA-UE after therapy. Meanwhile the improvement was maintained until the one-month follow-up after the end of therapy. In addition, improvement of motor function is associated with the enhancement of resting-state functional connectivity. Conclusions This work reveals longitudinal trend of neural patterns over full-process of intervention and its correlation with motor recovery, so as to provide more evidence for a detailed understanding of the mechanism of neural plasticity. brain–computer interface longitudinal trend motor imagery stroke full-process Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 1. Introduction Stroke occurs due to a severe reduction or blockage of blood flow to the brain [1], potentially causing damage to motor and sensory nerve systems [2]. The key to motor rehabilitation for stroke patients lies in the brain plasticity, which supplements or reorganizes the specific function of damaged brain region by the remaining uninjured neurons and cortex [3]. Previous studies have shown that closure of sensorimotor pathways and relying on the repeated experience promote brain plasticity [4, 5]. Synergistic efforts in field of neural engineering and robotics are showcasing how neural interface techniques can be used to control devices and ultimately restore bodily functions, which open the door for establishing the closed-loop and enhancing motor relearning process. In particular, motor imagery-based brain–computer interface (MI-based BCI) has “movement property” [6], can be approximated as a mental rehearsal of actual movement execution without actual motor outputs, with a high degree of similarity in neural activation patterns to motor execution (ME) [7]. MI activated sensorimotor cortex directly and induced sensorimotor rhythm (SMR), which expresses the decrease of neural oscillatory activity of brain in mu and beta frequency bands, and is named event-related desynchronization (ERD). Meanwhile, ERD can be detected for the control of neuroprostheses and robots, e.g., functional electrical stimulation (FES) [8], exoskeleton [9]. On the one hand, it provides visual and proprioceptive feedback to the patient and closes afferent-efferent loops to restore neural function [10]; on the other hand, it can directly or indirectly stimulate the patient’s affected limb to restore motor [8, 9]. Therefore, parallel attention to the recovery of patient’s motor and neural functions as well as their correlation is necessary for evaluating the effectiveness of MI-based BCI rehabilitation systems. Despite the encouraging results achieved so far [11–14], MI-based BCI stroke rehabilitation is still a young field, with varying clinical outcomes reported across different studies [15, 16]. Furthermore, most research have conducted EEG assessments only at three or even two time points: baseline, post-train, or follow-up, missing the detailed changes in patients’ neural patterns over the intervention process. This limitation has largely resulted in a lack of evidence of the neuroplasticity mechanisms and the therapeutic effects. Brunner, I. [17] reported the changes in ERD patterns and clinical scale before and after training under BCI combined with FES therapy. The results showed that the clinical scores of only 5 patients (a total of 15 patients receiving BCI treatment) met the minimum clinical difference; the ERD pattern showed significant changes before and after training. Biasiucci, A. [18] reported a significant motor function recovery after the intervention, which remains maintained throughout the follow-up period (6–12 months after the end of therapy). EEG analysis pinpoints significant differences, mainly consisting in an increase of resting-state functional connectivity between motor areas in the affected hemisphere. Additionally, other studies [15, 16] have also shown varying degrees of changes in neural activity after training. The conclusions of these researchs rely on non-invasive BCI to collect patients’ EEG signals, which require only minimal effort to obtain relevant data. However, patients are required to perform specific BCI paradigm (e.g., MI and rest) for functional assessment, which are easily affected by their mental state on the day, such as drowsiness or emotional instability caused by the disease, thereby cannot objectively and accurately reflect the rehabilitation results. Therefore, the inclusion of multi-point EEG assessments during the intervention rather than just tracking EEG at the traditional assessment points (baseline, post-train, or follow-up), is necessary to observe the longitudinal trend of neural evolution over intervention: 1) this provides additional evidence for the mechanisms of neuroplasticity under BCI training in stroke patients and 2) makes neural assessment results more reliable. To our knowledge, few studies have focused on the EEG pattern longitudinal trends during full-process of intervention to reflect the recovery of patients’ neural functions. The main goal of this study is to achieve EEG assessment coverage full-process of intervention to investigate the longitudinal trend in neural patterns during stroke rehabilatation, as well as to assess the motor function pre-post rehabilitation to explore the potential relationship between neural and motor recovery. Our system (previous work foundation, see [19, 20]) was designed to combine BCI with supernumerary robotic finger (SRF) and sensory electrical stimulation to provide tactile and proprioceptive feedback to patients for closing the sensory pathway. The SRF fulfills the following two properties: it indirectly elicits functional movements of the affected limb through inducing the finger-to-finger or assisted grasping exercises [20, 21]; and several studies suggest that inherent or robotic supernumerary fingers could induce remodeling of the cerebral cortex [22]. Our previous research have supported this suggestion [20]. In this paper, we recruited 14 stroke patients and conducted 8 sessions of BCI-controlled “sixth-finger” neurofeedback training over 2 weeks. EEG data and clinical sacles were evaluated at three time points: baseline, post-train and follow-up period, and EEG of each intervention sessions were also tracked. Try to explore the longitudinal trend of brain activation patterns during full-process of intervention and its correlation with motor recovery of pre-post intervention, so as to better understand the neural mechanisms involved in stroke recovery. 2. Method 2.1 Design and Setting The study was designed as a self-controlled study to investigate the EEG longitudinal trends and the relationship between motor recovery and neuroplastic effects. Fourteen stroke patients (4 females, 10 males, mean age (55.5 ± 10.5) years) were recruited from Tianjin Huanhu Hospital to participate in the experiment. The experiment timeline is shown as Fig. 1 (A), assessments were performed 3 times, immediately before (baseline) and after the intervention (post-interventon), as well as 1 month after the end of the intervention (follow-up). The invertention was consisted of a total of 8 training sessions spanned over 2 weeks with 4 sessions per week, each training session lasted ~ 60 min, including ~ 15 min of BCI classifier calibration and ~ 45 min of BCI intervention therapy. The eligibility for participation of all patients was determined by medical professional and all patients signed informed consent. The experimental was consistent with the Declaration of Helsinki on the ethical treatment of human subjects, and passed the ethical approval of the Medical Ethics Committee of Tianjin Huanhu Hospital (IEC-B-003-V3.0). 2.2 Patient Characteristics Table 2 contains 14 patients demographics and basic information. The inclusion criteria for the patients were as follows: (1) between 30 and 75 years of age; (2) the first occurrence of cerebral hemorrhage, cerebral infarction; (3) disease duration is 3–12 months, this is to ensure that the period of rapid spontaneous motor recovery has ended [23]; (4) moderate-to-severe impaired with the upper extremity Fugl-Meyer assessment (FMA-UE) [24] score ≤ 47 points; (4) clearly understand and perform experiment tasks and requirements, no cognitive dysfunction, and a simple Mental State Examination score > 20 points. The exclusion criteria were: (1) motor dysfunction of the affected upper limb due to other pathologies; (2) serious psychological condition that might affect the ability to complete the experiment, epilepsy; (3) unable to collect EEG signals due to scalp wounds or other reasons. Patients who terminated or suspended the intervention for personal reasons were excluded from this study. 2.3 The BCI-controlled System The BCI-controlled system incorporated an external equipment (robotic “sixth-finger” combined with sensory electrical stimulation), a commercial EEG amplifier and a neurofeedback interface. As shown in Fig. 2 (A), the robotic “sixth-finger” was worn at the patients’ wrist on the hemiplegic side and could be controlled to bend and extend with one degree of freedom, the electrodes for sensory electrical stimulation were mounted on forearms of hemiplegic side [20]. The EEG were recorded by a NeuSen W system (Neuracle) amplifier with 64 Ag/AgCl scalp electrodes placed according to the international 10/20 System. The BCI-controlled framework is shown in Fig. 2 (B), pattern recognition was performed on the EEG signal and converted the recognition results into control commands to provide feedback for patients. The pattern recognition was performed on EEG data by sliding with 1000ms as window length and 200ms as the step distance, thus a recognition result was generated every 200ms. And the Rest state was defined as Label 1, MI state as Label 2. A smoothing filter [25] was used in order to prevent the occurrence of false-positive events. Given x t the recognition result at time t (1 or 2 respectively represent Rest or MI) and y t−1 the previous window control signal, y t the current window control signal which is definded as follows: $$\:{y}_{t}=\alpha\:·{x}_{t}+\left(1-\alpha\:\right)·{y}_{t-1}$$ where \(\:\alpha\:\) \(\:\in\:\left[0.0,\:1.0\right]\) is the smoothing factor. A neurofeedback interface based on Psychopy programming maps the control signal y t to the bending angle of the virtual “sixth-finger” in real time (if the recognition result is MI, the bending angle becomes larger, otherwise it becomes smaller). Finally, thresholding strategy (the blue outline in the interface) was used to translate the control signal y t into output command for the robotic “sixth-finger” and sensory electrical stimulation. 2.4 The BCI Intervention The experiments were conducted in a quiet and confined room. Two professionals accompanied the patient throughout the entire experiment, one should have expert EEG knowledge to be skilled in installing the EEG cap and use the EEG acquisition system, and the other should be medically qualified to deal with any discomfort of the patient. The subject sat comfortably in a chair at a distance of one-meter from the screen, the stimulation program in the computer screen is completed by Psychopy. All patients completed a calibration of individual BCI classifier to distinguish EEG activity between MI and Rest state before each training session. During the calibration block, as shown in Fig. 1 (B), each trial lasted 14s and is started by pressing space. A green cross appeared at the center of the screen for 7s with the patient remaining at rest, followed by a grey circle lasting 1s, prompting subjects to prepare for the MI tasks. Then the picture cue of bending the “sixth-finger” appeared to indicate patients to imagine and lasted 6s. The calibration block consist of 3 runs of 20 trials each. EEG data during the rest and MI state were selected to build the BCI classifier. During the therapy block, as shown in Fig. 1 (C), press the space bar when the patient is ready, followed by a grey circle lasting 1s, prompting subjects to prepare for the MI tasks. The screen then appears with the robotic “sixth-finger” bending or extending in real time, and with blue outlines representing preset thresholds. The angle of the “sixth-finger” movement is depending on the patient’s brain activity (as described in Method 2.3 and 2.6), recognised by the BCI classifier built in calibration block. If the preset threshold is reached within 6s, i.e., the bending angle of the “sixth-finger” reaches within the blue outline, a control command is output. The patients receive proprioceptive feedback in the form of robotic “sixth-finger” bending and sensory electrical stimulation. During the MI period, patients were explicitly asked to avoid any body movements, including speaking and taking their attention away from the “sixth-finger”. The “sixth-finger” MI paradigm proposed by our previous studies was adopted [19]. 2.5 Outcome Measures Three assessments (baseline, post-train, follow-up) all included clinical scales and EEG measure. The clinical outcomes were included FMA-UE and the Barthel index (BI). The primary clinical outcome of this paper is FMA-UE, from 0 to 66 for plegia to normal. And the minimally clinically important difference (MCID) was set to 6.6 (10% of the total range of the scale) [26] for the FMA-UE. BI is secondary outcome, using to evaluate individual’s ability to live independently by measuring a range of basic activities such as eating, bathing, and dressing, with a scale from 0 to 100 (best). All clinical scales were measured by professional therapists from Tianjin Huanhu Hospital. The EEG measure was used as marker of neuroplasticity recovery, including MI task-state and rest-state. The task-state paradigm is shown as Fig. 1 (B). The rest-state paradigm is acquisition of the patient’s resting EEG for 4 min. And the task-state EEG recorded at assessment and each calibration were used to investigate longitudinal trends of brain activity in response to BCI intervention. The rest-state EEG recorded at assessment were mainly used to analyzed the functional connectivity changes. 2.6 EEG data Pre-processing and Analysis For the online control in the therapy block, the forty-electrode raw EEG data from Fig. 1 (B) were processed by the band-pass filtered between 5 and 20 Hz, and 50 Hz notch filter for removing the power line interference during signals acquisition, and were processed by the common average referenced (CAR). Then, common spatial pattern (CSP) [27] was utilized to extract features of multi-channel EEG information, support vector machine (SVM) [28] was utilized for pattern recognition between MI vs. Rest state. For the offline analysis, the EEG data in calibration and assessment sessions were selected. In order to further ensure the purity of data, EEG artifacts were manually identified and rejected, abnormal channels were removed by interpolation operation, and independent component analysis (ICA) was adopted to remove eye movement artifacts. The ERD phenomenon is the main brain activation index for MI-BCI. ERSP was used to reflect the event-related power variations of the induced EEG signals in time and frequency domain [29], which defined as follows: where n is the number of trials, and \(\:{F}_{k}{(f,t)}^{2}\) is the spectral estimation of k th trial at frequency f and time t . And in order to quantify the ERD phenomenon, the baseline correction was conducted by subtracting the average ERSP value within the baseline period. The weighted phase-lag index (wPLI) was used to calculate the phase synchronization of two pairs of channels for connectivity measure. This method has been proved to be able to overcome the problem of volume conduction and is insensitive to irrelevant noise sources [30]. The non-directed coherence measures for 40*40 EEG electrode pairings wPLI [31] were calculated as follows: $$\:wPLI=\frac{\left|\right\{S\left\}\right|}{\left\{\left|S\right|\right\}}=\frac{\left|\right\{\left|\text{S}\right|sign\left(S\right)\left\}\right|}{\left\{\left|S\right|\right\}}$$ where S means the imaginary component of cross-spectrum between channel A 1 and channel A 2 , {•} means the expected value operator. The value of wPLI ranges from 0 ~ 1, with high wPLI value representing strong coupling of neural oscillatory activity. Brain functional network and their parameters such as node degree and small-world index were also analysed. And the threshold selection principle for establishing connections was 1) maximum threshold without isolated nodes and isolated parts; 2) with small-world attribute. 2.7 Statistical Analyses For continuous variables, the Lilliefors-corrected Kolmogorov-Smirnov test was first used to check the Gaussian distribution. Paired samples t-test was utilized to evaluate the statistical significance of changes in EEG data and clinical metrics before and post therapy. The EEG metrics such as wPLI values and small-world attribute and clinical metrics such as FMA-UE and BI of each patients in the baseline (T0) and post-train (T1) period were regarded as paired samples. To explore the potential mechanisms and longitudinal trends of neural patterns, correlations between variables were measured by Pearson’s correlation coefficient. Longitudinal trends in brain modulation patterns during full-process intervention were evaluated by calculating the Pearson’s correlation between mean ERSP value and the number of therapy sessions from baseline to post-train. And the Pearson’s correlation between the clinical metrics change and EEG metrics change (baseline and post-train) was conducted to reflect the relationship between motor improvement and neural plasticity before and post rehabilitation. 3. Results 3.1 ERD Longitudinal Trends In order to investigate the longitudinal trends of ERD distribution and strength over the sessions, the separate average ERD topographical maps in mu rhythm were shown in Fig. 3 (A). And as shown in Fig. 3 (B), due to the inconsistency in the hemiplegic sides of the patients (four patients with left hemiplegia, the rest with right hemiplegia), the electrodes were mirrored by setting the left side of topographical distribution as the contralateral side of hemiplegic, and the right side of topographical distribution as the ipsilateral side. The results show that bilateral ERD activation in sensorimotor cortex was induced when patients performed MI tasks, and consistently maintained global ERD activation in brain, in the first few sessions (week 1). As the session progressed, the bilateral brain modulation of patients was weakened (week 2). Until the post-train session, the ERD phenomenon focused on sensorimotor areas and contralaterally modulated was found, which more closely resembles the brain modulation patterns when healthy individuals performed unilateral limb MI. After a one-month blanking period (week 7), the obvious sensorimotor area modulation and weak contralateral dominance was still maintained. Correlation coefficients between the number of sessions (ten sessions from baseline to post-train) and the amplitude of ERD for each electrode were given in Fig. 4 and Table 1 to quantify the longitudinal trend in ERD patterns. As shown in Fig. 4 (A), degree of ERD activation in nine electrodes seems to significantly anti-correlate with the sessions, mainly distributed in the periphery of FC line, C line and CP line. The r and p for nine electrodes with correlation were given in Table 1. The contralateral and ipsilateral electrodes with correlation were averaged and calculated correlation coefficient (F5 + FC5 + FC3 + C5 + C3, r = -0.709, p = 0.012 * ; C4 + C6 + Cp4 + Cp6, r = -0.752, p = 0.021 * ), as shown in Fig. 4 (B), the intensity of ERD activation in peripheral brain regions gradually decreased across session progressed. This corroborates that after training, ERD activation is more focused on the central of sensorimotor areas and the functional compensatory effects from other brain areas are suppressed. 3.2 Alterations in EEG Connectivity In order to investigate the functional integration and close connections between different brain regions before and after rehabilitation, functional connectivity and statistical comparison based on the wPLI in mu rhythm were implemented. As shown in Fig. 5 (A), the connected edges of the total 780 electrode pairs that exceeded the threshold at baseline and post-train sessions was drawn, and the pairwise connectivity with significant differences ( p ” means the significantly increased amount of connectivity. The result shown that more synchronized neural activity between different brain regions was found after rehabilitation compared baseline. Functional connectivity values were significantly enhanced for 22 electrode pairs and significantly reduced for no electrode pairs. The average degrees across all electrodes and small-world attribute of the brain functional network before and after rehabilitation were shown in Fig. 5 (B). It could be seen that the degrees across all electrodes increased ( p = 0.0598) after rehabilitation, and the electrodes with increased degree are mainly concentrated in central of the sensorimotor area (blue dashed box) as shown in Fig. 5 (C). The small-world attribute of the brain functional network also increased ( p = 0.1144) after rehabilitation which indicated that brain network become more efficient and information is processed more efficiently. 3.3 Clinical Outcome Metrics Table 2 reports the clinical characteristics and clinical metrics (T0: baseline, T1: Post-train, T3: Follow-up) of the stroke patients. Two patients (ID 11 and 12) were unable to be evaluated in follow-up period due to personal reasons. There was a significant improvement on the primary clinical outcome FMA-UE scores of patients between post (T1) and before (T0) BCI treatment ( p < 0.01), with an average increase of 7.9 points. Meanwhile, patients still maintained improvement in the 1-month follow-up (T2) after the end of therapy. More than half of patients (9/14, ID: 02, 04, 06, 08, 10, 11, 12, 13, 14) reached the MCID of 6.6 points change for FMA-UE after therapy. The secondary clinical metrics BI also showed a significant improvement between post-before BCI treatment (with an average increase of 7.1 points, p < 0.01) and was also maintained until the follow-up period. The changes of FMA-UE and BI scores for three time points are illustrated in Fig. 6. 3.4 Neurophysiological Correlates In order to further investigate the possible role of EEG pattern changes in the improvement of motor function, the correlation between the changes in resting-state functional connectivity and improvements in FMA-UE scores before and post-train was analyzed. As shown in Fig. 7, an obvious positive correlation relationship emerged between FMA-UE score changes and the changes in the small-world attribute and degree of the resting-state brain functional network (small-world attribute: Pearson’s correlation, r = 0.502, p = 0.067, degree: Pearson’s correlation, r = 0.436, p = 0.119). 4. Discussion In this study, we aimed to explore the longitudinal trend of EEG patterns during the full-process of BCI therapy in stroke patients, providing more evidence for a detailed understanding of the mechanism of neural plasticity. This study was conducted based on our previous construction of a BCI-controlled “sixth-finger” system [20]. Our previous studies have demonstrated the advantages of the “sixth-finger” MI paradigm [19] (applied to this study) and the role of brain-controlled the “sixth-finger” training in promoting neural plasticity on healthy people [20]. MI-induced ERD phenomenon and resting-state functional connectivity revealed brain pattern changes. FMA-UE and BI were used as clinical metrics, reflecting the improvement of patients’ motor function. 4.1 Longitudinal Trends in EEG Pattern In this paper, longitudinal trends in EEG patterns were reported, which are important for a detailed understanding of the process and mechanism in neuroplasticity, but are poorly described in other literature. The results (Fig. 3) showed that the patient’s neural patterns seemed to go through three phases: (1) at baseline and the first week of training, the extent of ERD activation gradually increased in the sensorimotor cortex and even in the prefrontal cortex when the patients performed the MI tasks. On the one hand, this may be related to the novel MI paradigm that the patients were asked to perform in training. In our previous study [19], we proposed this MI paradigm based on the “sixth-finger” (imagine controlling the movement of the “sixth-finger”) and observed a larger range and stronger degree of ERD activation in healthy people, compared with the traditional MI paradigm based on the inherent hand (imagine the movement of inherent hand). Indeed, it is also widely believed that learning new movements leads to a gradual increase in brain activation in the relevant limb motor cortical areas [32]. And Pascual-Leone et al. [33] reported that learning a sequence over a few days was generally induced an increase in the size of the motor map. The stroke patients receiving a mechanical extra finger and trying to imagine controlling it, can indeed be considered to be learning a new movement or skill. On the other hand, it is possible that the compensatory mechanism of neural function plays a role. Compensation and involvement of other areas (ipsilateral sensorimotor area and prefrontal lobe) enlarged the activated brain areas and deviated from the contralateral advantage activation that is natural performance during hand MI. Bai, O.’s research also showed that neural activation may ‘spill over’ to the ipsilateral cortex when the processing load of the contralateral motor cortex increases [34]. (2) At the second week of training and post-train assessment, the ERD phenomenon gradually weakened and became more focused on the sensorimotor cortex associated with the movement of hand. As shown in Fig. 4 and Table 1, the ERD of electrodes in the periphery of the motor area and parietal lobe weakens over process of training and shows a significant correlation with the number of sessions. Some studies have shown that once the neural process is optimized, activation in cortex appears to weaken [35, 36], and our results are consistent with this conclusion. And Steele, C. [37] has reported that during initial gesture motor learning, neural activity in motor cortex increases, whereas after acquisition and consolidation of the motor task, activity decreases. In addition, the theory of “neural efficiency” [38] has also shown that an expert in a motor task displayed less pronounced and more spatially focused mu rhythm ERD. This gives us evidence to believe that the weakening and more focused ERD phenomenon over sessions is due to the improved efficiency of neural recruitment, which promotes the process of neural plasticity. The EEG analysis of resting state functional connectivity in this paper also supports this conclusion. As shown in Fig. 5, the strength of resting-state functional connection in the post-train period was significantly greater than that in the baseline period. Furthermore, the node degree and small-world attribute are higher after therapy, indicating that the overall processing efficiency of the brain functional network has been significantly improved [39], especially the sensorimotor area shown in Fig. 5 (C). (3) At follow-up assessment, the ERD phenomenon still showed natural and contralateral activation patterns, which was similar to the post-train assessment, although with some rebound. This indicates the improvement effect of neural plasticity was still maintained until the follow-up period, which was consistent with the improvement of motor function (Fig. 6). And the rebound of ERD may be due to the fact that patients became unfamiliar with the “sixth-finger” or MI tasks after a month of rest. 4.2 Clinical Metrics Outcome Improvement Although this paper focuses on exploring the longitudinal change trend of brain neural activity caused by BCI therapy, significant clinical functional improvement is still necessary for stroke rehabilitation. Overall, BCI therapy results in a significant clinical improvement, with an average increase of 7.9 points in FMA-UE and an average increase of 7.1 points in BI between post (T1) and before (T0) BCI therapy. And this improvement lasted until 1 month after the end of training. Many previous BCI clinical studies contained no follow-up assessment. The stroke patients recruited in this study were all those with onset time of 3 to 10 months and the spontaneous motor recovery almost stagnated [23]. And patients with moderate-to-severely functional impairment are less likely to recover from other methods [11], while patients with poor function (ID:10, 12) in this study have significantly improved after rehabilitation. These indicate that the improvement of the stroke patients in this study results from the BCI therapy they received. In addition, the introduction of MCID can make the improvement of clinical metrics more convincing, which is defined as the smallest change that is clinically important to patients or clinicians [40]. More than half of patients (9/14, with # marked) reached the MCID of 6.6 points change for FMA-UE after therapy. Co-analysis of clinical outcomes with EEG analysis may provide evidence for possible mechanisms behind functional improvements of stroke patients. The results (Fig. 7) showed that the enhancement in resting-state functional connectivity was correlated with the improvement in FMA-UE outcomes. Moreover, the nodes with enhanced connection strength were mainly concentrated in the sensorimotor area shown in Fig.5 (C), which is consistent with the conclusion of previous studies that the reactivation of the motor area seems to be crucial for motor recovery [18]. The global enhancement of functional connectivity and the improvement of node degree and small-world attribute indicate the improvement of brain processing efficiency, which seems to confirm the theory of neural efficiency mentioned above. Although the assertion of a rehabilitation mechanism based on the above correlation analysis is somewhat limited, the results show that the recovery of neurological function has played a role in promoting the improvement of motor function outcomes. 4.3 Limitations However, there are still some limitations that need to be overcome to make our research more comprehensive and scientific. First, the study was limited by the number of clinical cases. 14 patients may not make the results generalizable, and patients did not show consistent clinical recovery results. A larger sample size is needed in the future to confirm the preliminary findings of this study. Secondly, stroke patients differ in lesion type and onset time. Although the impact of this difference is not the main focus of this study, it reflects the actual clinical needs. Are there different characteristics of neural change trends for different types of patients? Should different clinical rehabilitation methods and doses be used? This is the ultimate goal of studying the stroke rehabilitation mechanism. To achieve this goal, further researches may be needed on different types of stroke patients in subsequent work. Finally, there was a lack of blank or traditional control groups. Although we recruited stroke patients who theoretically had little spontaneous recovery ability, we still lacked actual quantitative data to confirm the ineffectiveness of patients’ recovery without training or with traditional rehabilitation therapy. In this paper, we focused more attention to the change trends in EEG patterns and neurorehabilitation mechanisms. In subsequent work, we will provide more sufficient evidence to prove the effectiveness of BCI therapy. 5. Conclusion In this paper, the longitudinal trend in neural patterns over full-process of intervention and the potential relationship between neural and motor recovery in stroke patients were reported. We found that the longitudinal change trend of EEG showed three stages during the intervention process: ERD gradually increased in the first week of training, weakened and focused on the contralateral sensorimotor area in the second week, and showed a significant correlation over sessions, remaining focused and contralateral pattern in the follow-up period. Motor function between pre-post intervention showed significant improvement by clinical metrics, with + 7.9 in FMA-UE and + 7.1 in BI. Meanwhile the improvement was maintained until the one-month follow-up after the end of therapy. In addition, improvement of motor function is associated with the enhancement of resting-state functional connectivity. The findings of this study may be important for detailed understanding of the mechanism of neural plasticity, and subsequent studies that attempt to track the neurological patterns of patients during intervention. Declarations Acknowledgements The authors sincerely thank all patients for their voluntary participation. Author contributions Y. Liu led the design and implementation of the experiment, guiding data processing and manuscript writing; Z. Wang developed an online neural feedback interface and was responsible for conducting experiments, processing EEG data, and writing the manuscript; S. Huang was responsible for the development of the BCI system and participated in the design and implementation of experiments; H. Huang was responsible for summarizing and analyzing clinical outcomes and participates in the experiments; W. Wu, Y. Wang participated in the experiment and was responsible for data collection; X. An, D. Ming was responsible for managing experiments, interpreting data, and reviewing manuscript; J. Wu, Y. Li was responsible for clinical work. All authors reviewed the manuscript. Funding This work was supported in part by the National Key Research and Development Program of China (2023YFF1204300, 2023YFF1204305), National Natural Science Foundation of China (62273251), Research Project of State Key Laboratory of Mechanical System and Vibration (MSV202418). Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Ethics approval and consent to participate All patients signed informed consent. The experimental was consistent with the Declaration of Helsinki on the ethical treatment of human subjects, and passed the ethical approval of the Medical Ethics Committee of Tianjin Huanhu Hospital (IEC-B-003-V3.0). Consent for publication Consent for publication were given by all participants Competing interests The authors declare that they do not have any competing interests. Authors details 1 Academy of Medical Engineering and Translational Medicine, Tianjin University, Tianjin, 300072, China; 2 Department of Rehabilitation Medicine, Tianjin Huanhu Hospital, Tianjin, 300350, China; 3 Clinical College of Neurology, Neurosurgery and Neurorehabilitation, Tianjin Medical University, Tianjin, 300370, China; 4 Department of Neurology, Tianjin Huanhu Hospital, Tianjin 300350, China; 5 Tianjin Key Laboratory of Cerebral Vascular and Neurodegenerative Diseases, Tianjin Neurosurgical Institute, Tianjin 300350, China. References Stinear, C.M., et al., Advances and challenges in stroke rehabilitation. The Lancet Neurology, 2020. 19(4): p. 348-360. George, Paul M. and Gary K. Steinberg, Novel Stroke Therapeutics: Unraveling Stroke Pathophysiology and Its Impact on Clinical Treatments. Neuron, 2015. 87(2): p. 297-309. Teasell, R., N.A. Bayona, and J. Bitensky, Plasticity and reorganization of the brain post stroke. Topics in stroke rehabilitation, 2005. 12(3): p. 11-26. Hallett, M., Plasticity of the human motor cortex and recovery from stroke. Brain Research Reviews, 2001. 36(2-3): p. 169-174. Jia, J., Exploration on neurobiological mechanisms of the central-peripheral-central closed-loop rehabilitation. Frontiers in Cellular Neuroscience, 2022. 16. Sharma, N. and J.-C. Baron, Does motor imagery share neural networks with executed movement: a multivariate fMRI analysis. Frontiers in Human Neuroscience, 2013. 7. Xu, L., et al., MOTOR EXECUTION AND MOTOR IMAGERY: A COMPARISON OF FUNCTIONAL CONNECTIVITY PATTERNS BASED ON GRAPH THEORY. Neuroscience, 2014. 261: p. 184-194. Chen, L., et al., EEG-controlled functional electrical stimulation rehabilitation for chronic stroke: system design and clinical application. Frontiers of Medicine, 2021. 15(5): p. 740-749. Chowdhury, A., et al., Active Physical Practice Followed by Mental Practice Using BCI-Driven Hand Exoskeleton: A Pilot Trial for Clinical Effectiveness and Usability. IEEE J Biomed Health Inform, 2018. 22(6): p. 1786-1795. Mang, J., et al., Favoring the cognitive-motor process in the closed-loop of BCI mediated post stroke motor function recovery: challenges and approaches. Frontiers in Neurorobotics, 2023. 17. Bundy, D.T., et al., Contralesional Brain-Computer Interface Control of a Powered Exoskeleton for Motor Recovery in Chronic Stroke Survivors. Stroke, 2017. 48(7): p. 1908-1915. Tsuchimoto, S., et al., Sensorimotor Connectivity after Motor Exercise with Neurofeedback in Post-Stroke Patients with Hemiplegia. Neuroscience, 2019. 416: p. 109-125. Wang, X., et al., Differentiated Effects of Robot Hand Training With and Without Neural Guidance on Neuroplasticity Patterns in Chronic Stroke. Front Neurol, 2018. 9: p. 810. Ranzani, R., et al., Neurocognitive robot-assisted rehabilitation of hand function: a randomized control trial on motor recovery in subacute stroke. J Neuroeng Rehabil, 2020. 17(1): p. 115. Baniqued, P.D.E., et al., Brain-computer interface robotics for hand rehabilitation after stroke: a systematic review. J Neuroeng Rehabil, 2021. 18(1): p. 15. Bai, Z., et al., Immediate and long-term effects of BCI-based rehabilitation of the upper extremity after stroke: a systematic review and meta-analysis. J Neuroeng Rehabil, 2020. 17(1): p. 57. Brunner, I., et al., Brain computer interface training with motor imagery and functional electrical stimulation for patients with severe upper limb paresis after stroke: a randomized controlled pilot trial. J Neuroeng Rehabil, 2024. 21(1): p. 10. Biasiucci, A., et al., Brain-actuated functional electrical stimulation elicits lasting arm motor recovery after stroke. Nat Commun, 2018. 9(1): p. 2421. Liu, Y., et al., EEG characteristic investigation of the sixth-finger motor imagery and optimal channel selection for classification. Journal of Neural Engineering, 2022. 19(1). Liu, Y., et al., Functional Reorganization After Four-Week Brain-Computer Interface-Controlled Supernumerary Robotic Finger Training: A Pilot Study of Longitudinal Resting-State fMRI. Front Neurosci, 2021. 15: p. 766648. Hussain, I., et al., A soft supernumerary robotic finger and mobile arm support for grasping compensation and hemiparetic upper limb rehabilitation. Robotics and Autonomous Systems, 2017. 93: p. 1-12. Kieliba, P., et al., Robotic hand augmentation drives changes in neural body representation. Science Robotics, 2021. 6(54). Duncan PW, Goldstein LB, Matchar D, Divine GW, Feussner J. Measurement of motor recovery after stroke. Outcome assessment and sample size requirements. Stroke. 1992;23:1084–1089. Fugl-Meyer AR, Jääskö L, Leyman I, Olsson S, Steglind S. The post-stroke hemiplegic patient. 1. A method for evaluation of physical performance. Scand J Rehabil Med. (1975) 7:13–31. Tonin, L., F.C. Bauer, and J. del R. Millan, The Role of the Control Framework for Continuous Teleoperation of a Brain–Machine Interface-Driven Mobile Robot. IEEE Transactions on Robotics, 2020. 36(1): p. 78-91. van der Lee, J.H., et al., The responsiveness of the action research arm test and the Fugl-Meyer Assessment scale in chronic stroke patients. Journal of Rehabilitation Medicine, 2001. 33(3): p. 110-113. Ben Hamed, S., M.H. Schieber, and A. Pouget, Decoding M1 neurons during multiple finger movements. Journal of Neurophysiology, 2007. 98(1): p. 327-333. Wang, L., et al., Analysis and classification of speech imagery EEG for BCI. Biomedical Signal Processing and Control, 2013. 8(6): p. 901-908. Yi, W., et al., EEG feature comparison and classification of simple and compound limb motor imagery. Journal of Neuroengineering and Rehabilitation, 2013. 10. Lau, T.M., et al., Weighted phase lag index stability as an artifact resistant measure to detect cognitive EEG activity during locomotion. Journal of Neuroengineering and Rehabilitation, 2012. 9. Lee, M., Y.-H. Kim, and S.-W. Lee, Motor Impairment in Stroke Patients Is Associated With Network Properties During Consecutive Motor Imagery. IEEE Transactions on Biomedical Engineering, 2022. 69(8): p. 2604-2615. Floyer-Lea, A. and P.M. Matthews, Distinguishable brain activation networks for short- and long-term motor skill learning. Journal of Neurophysiology, 2005. 94(1): p. 512-518. Pascual-Leone, A., et al., Modulation of muscle responses evoked by transcranial magnetic stimulation during the acquisition of new fine motor skills. Journal of neurophysiology, 1995. 74(3): p. 1037-45. Bai, O., et al., Asymmetric spatiotemporal patterns of event-related desynchronization preceding voluntary sequential finger movements: a high-resolution EEG study. Clinical Neurophysiology, 2005. 116(5): p. 1213-1221. Haslinger, B., et al., Reduced recruitment of motor association areas during bimanual coordination in concert pianists. Human Brain Mapping, 2004. 22(3): p. 206-215. Jäncke, L., N.J. Shah, and M. Peters, Cortical activations in primary and secondary motor areas for complex bimanual movements in professional pianists. Cognitive Brain Research, 2000. 10(1-2): p. 177-183. Steele, C.J. and V.B. Penhune, Specific Increases within Global Decreases: A Functional Magnetic Resonance Imaging Investigation of Five Days of Motor Sequence Learning. Journal of Neuroscience, 2010. 30(24): p. 8332-8341. Babiloni, C., et al., "Neural efficiency" of experts' brain during judgment of actions: A high-resolution EEG study in elite and amateur karate athletes. Behavioural Brain Research, 2010. 207(2): p. 466-475. Liu, M., et al., Effects of Transcranial Direct Current Stimulation on EEG Power and Brain Functional Network in Stroke Patients. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2023. 31: p. 335-345. Jaeschke, R., J. Singer, and G.H. Guyatt, Measurement of health status. Ascertaining the minimal clinically important difference. Controlled clinical trials, 1989. 10(4): p. 407-15. Tables Tables 1 to 2 are available in the Supplementary Files section Additional Declarations The authors declare no competing interests. Supplementary Files Tables.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6027743","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":415647416,"identity":"dfe61fc2-149c-4e55-8d38-970e5e97ab86","order_by":0,"name":"Wang Zhuang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Wang","middleName":"","lastName":"Zhuang","suffix":""},{"id":415647490,"identity":"3a73235e-6600-487c-89cb-04da3c030758","order_by":1,"name":"Liu Yuan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAy0lEQVRIiWNgGAWjYLCCBwYMcgwMB4AsNmK1JBgwGPOQqIWBIbEHzCJGi/mM5GcPEgoOp+9nPGPA8KHsMAP/7Ab8WmRupJkbJBgczu1hOGPAOOPcYQaJOwfwa5GQSDCTgGlh5m07zGAgkUBIS/o3kJZ0HpCWv8RpyQHbkgDWwkiUFp43ZUAt6YY9B44VHOw5l84jcYOQFvb0bRIf/ljLs884vPHBjzJrOf4ZBLQwCMAUSBwARyYPAfVAwH8AxmggrHgUjIJRMApGJgAAtiZA0XorXBYAAAAASUVORK5CYII=","orcid":"","institution":"","correspondingAuthor":true,"prefix":"","firstName":"Liu","middleName":"","lastName":"Yuan","suffix":""}],"badges":[],"createdAt":"2025-02-14 06:17:07","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-6027743/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6027743/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":76574130,"identity":"9494bf68-6bc6-4851-80bd-bec7e510d027","added_by":"auto","created_at":"2025-02-18 14:04:00","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":191561,"visible":true,"origin":"","legend":"\u003cp\u003eExperimental procedure and paradigm. (a) Experimental \u0026nbsp;\u0026nbsp;schedule, clinical and EEG assessment were performed at three time points: baseline (T0), post-train (T1) and 1-month follow-up(T3). (b) Experimental \u0026nbsp;\u0026nbsp;paradigm of EEG assessment and calibrition of each intervention session. (c)Experimental paradigm of BCI train in intervention sessions.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6027743/v1/5c4b357332bdb76a66157045.png"},{"id":76576008,"identity":"3d451c45-b1e7-4421-9669-c62db2262f67","added_by":"auto","created_at":"2025-02-18 14:12:00","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":297234,"visible":true,"origin":"","legend":"\u003cp\u003e(a) Experimental scene. (b) Online neurofeedback framework. The patients’ MI signals was \u0026nbsp;\u0026nbsp;collected real-time pattern recognition was performed to output \u0026nbsp;\u0026nbsp;discrimination results. The results are smoothed and filtered to form control commands, which control the psychopy interface, the robotic sixth-finger, and \u0026nbsp;\u0026nbsp;sensory electrical stimulation to provide real-time feedback to patients. (c)Forty-electrode positions.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6027743/v1/292643f6346826da5ff60970.png"},{"id":76574122,"identity":"6cc62aa8-c2ae-4585-8f14-cd9564254d49","added_by":"auto","created_at":"2025-02-18 14:04:00","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":191627,"visible":true,"origin":"","legend":"\u003cp\u003e(a) Averaged EEG topographical distribution across all \u0026nbsp;\u0026nbsp;subjects in mu frequency bands over sessions. (b) Electrodes mirror position.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6027743/v1/ecf0a5929f99502cf4c15340.png"},{"id":76576007,"identity":"f0d4f984-8c60-4794-bbb8-ee9b3d825497","added_by":"auto","created_at":"2025-02-18 14:12:00","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":167938,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e(a) \u003c/strong\u003eDistribution locations of electrodes with significant correlation between ERD and sessions. (b) Line graph of ERD of specific electrodes changing with session progress. The blue line represents the contralateral (left in this study) electrodes (F5, FC5, FC3, C5, C3) with a significant correlation between the ERD and sessions, the black line represents the ipsilateral (right in this study, C4, C6, Cp4, Cp6). The green dashed line represents the sessions for correlation analysis.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6027743/v1/78342d4bc68923719826d1c4.png"},{"id":76576633,"identity":"3aa02393-92cb-411b-9da1-c2accf4898e0","added_by":"auto","created_at":"2025-02-18 14:20:00","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":183742,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of brain network functional connectivity before and after BCI therapy. (a) Electrode pairs with \u0026nbsp;\u0026nbsp;resting-state functional connectivity values above the threshold were showed \u0026nbsp;\u0026nbsp;by straight lines. And the electrode pairs with significance difference (\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.05) were drawn. “\u0026gt;” means the \u0026nbsp;\u0026nbsp;significantly increased amount of connectivity. (b) Average values in \u0026nbsp;\u0026nbsp;small-word attribute and degree of the brain network. The error bar \u0026nbsp;\u0026nbsp;represents the Std. (c) Average degree of each nodes.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-6027743/v1/b2c733e70fe6804fc31626b0.png"},{"id":76576015,"identity":"12a8fc40-88f7-4f9d-b3e2-3f8e4f308a5d","added_by":"auto","created_at":"2025-02-18 14:12:03","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":54081,"visible":true,"origin":"","legend":"\u003cp\u003eClinical scale results at baseline, post-train and follow-up assessment. Grey shading represents the therapy stage. Significant differences were found post-pre intervention in FMA-UE (\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01) and BI (\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01).\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-6027743/v1/56c03ae757467cbddb05c36a.png"},{"id":76576635,"identity":"d9269442-d3af-43fb-8c41-9ce867ef82c0","added_by":"auto","created_at":"2025-02-18 14:20:00","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":58493,"visible":true,"origin":"","legend":"\u003cp\u003eCorrelation between the change (post–pre intervention) of connectivity and the change FMA-UE. (Pearson’s correlation, small-word attribute: r = 0.502, \u003cem\u003ep\u003c/em\u003e = 0.067; degree: r = 0.436, \u003cem\u003ep\u003c/em\u003e =0.119).\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-6027743/v1/c62ad36f28d59645952f2660.png"},{"id":76578157,"identity":"f4aa91db-b75d-4d36-8e50-7065fa60ea4d","added_by":"auto","created_at":"2025-02-18 14:36:01","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1603879,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6027743/v1/a627967d-ea9b-465f-ad3a-11e7183a7fa9.pdf"},{"id":76574120,"identity":"76ded1d4-875c-4773-a01a-d9d305b33e7c","added_by":"auto","created_at":"2025-02-18 14:04:00","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":23045,"visible":true,"origin":"","legend":"","description":"","filename":"Tables.docx","url":"https://assets-eu.researchsquare.com/files/rs-6027743/v1/da7d3fa4b01d32b712ce31f0.docx"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eERD Full-process Longitudinal Trend and Pre-post Motor Recovery Under BCI-controlled Sixth-finger Neurofeedback Intervention in Stroke Patients\u003c/p\u003e","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eStroke occurs due to a severe reduction or blockage of blood flow to the brain [1], potentially causing damage to motor and sensory nerve systems [2]. The key to motor rehabilitation for stroke patients lies in the brain plasticity, which supplements or reorganizes the specific function of damaged brain region by the remaining uninjured neurons and cortex [3]. Previous studies have shown that closure of sensorimotor pathways and relying on the repeated experience promote brain plasticity [4, 5].\u003c/p\u003e\n\u003cp\u003eSynergistic efforts in field of neural engineering and robotics are showcasing how neural interface techniques can be used to control devices and ultimately restore bodily functions, which open the door for establishing the closed-loop and enhancing motor relearning process. In particular, motor imagery-based brain\u0026ndash;computer interface (MI-based BCI) has \u0026ldquo;movement property\u0026rdquo; [6], can be approximated as a mental rehearsal of actual movement execution without actual motor outputs, with a high degree of similarity in neural activation patterns to motor execution (ME) [7]. MI activated sensorimotor cortex directly and induced sensorimotor rhythm (SMR), which expresses the decrease of neural oscillatory activity of brain in mu and beta frequency bands, and is named event-related desynchronization (ERD). Meanwhile, ERD can be detected for the control of neuroprostheses and robots, e.g., functional electrical stimulation (FES) [8], exoskeleton [9]. On the one hand, it provides visual and proprioceptive feedback to the patient and closes afferent-efferent loops to restore neural function [10]; on the other hand, it can directly or indirectly stimulate the patient\u0026rsquo;s affected limb to restore motor [8, 9]. Therefore, parallel attention to the recovery of patient\u0026rsquo;s motor and neural functions as well as their correlation is necessary for evaluating the effectiveness of MI-based BCI rehabilitation systems.\u003c/p\u003e\n\u003cp\u003eDespite the encouraging results achieved so far [11\u0026ndash;14], MI-based BCI stroke rehabilitation is still a young field, with varying clinical outcomes reported across different studies [15, 16]. Furthermore, most research have conducted EEG assessments only at three or even two time points: baseline, post-train, or follow-up, missing the detailed changes in patients\u0026rsquo; neural patterns over the intervention process. This limitation has largely resulted in a lack of evidence of the neuroplasticity mechanisms and the therapeutic effects. Brunner, I. [17] reported the changes in ERD patterns and clinical scale before and after training under BCI combined with FES therapy. The results showed that the clinical scores of only 5 patients (a total of 15 patients receiving BCI treatment) met the minimum clinical difference; the ERD pattern showed significant changes before and after training. Biasiucci, A. [18] reported a significant motor function recovery after the intervention, which remains maintained throughout the follow-up period (6\u0026ndash;12 months after the end of therapy). EEG analysis pinpoints significant differences, mainly consisting in an increase of resting-state functional connectivity between motor areas in the affected hemisphere. Additionally, other studies [15, 16] have also shown varying degrees of changes in neural activity after training. The conclusions of these researchs rely on non-invasive BCI to collect patients\u0026rsquo; EEG signals, which require only minimal effort to obtain relevant data. However, patients are required to perform specific BCI paradigm (e.g., MI and rest) for functional assessment, which are easily affected by their mental state on the day, such as drowsiness or emotional instability caused by the disease, thereby cannot objectively and accurately reflect the rehabilitation results. Therefore, the inclusion of multi-point EEG assessments during the intervention rather than just tracking EEG at the traditional assessment points (baseline, post-train, or follow-up), is necessary to observe the longitudinal trend of neural evolution over intervention: 1) this provides additional evidence for the mechanisms of neuroplasticity under BCI training in stroke patients and 2) makes neural assessment results more reliable. To our knowledge, few studies have focused on the EEG pattern longitudinal trends during full-process of intervention to reflect the recovery of patients\u0026rsquo; neural functions.\u003c/p\u003e\n\u003cp\u003eThe main goal of this study is to achieve EEG assessment coverage full-process of intervention to investigate the longitudinal trend in neural patterns during stroke rehabilatation, as well as to assess the motor function pre-post rehabilitation to explore the potential relationship between neural and motor recovery. Our system (previous work foundation, see [19, 20]) was designed to combine BCI with supernumerary robotic finger (SRF) and sensory electrical stimulation to provide tactile and proprioceptive feedback to patients for closing the sensory pathway. The SRF fulfills the following two properties: it indirectly elicits functional movements of the affected limb through inducing the finger-to-finger or assisted grasping exercises [20, 21]; and several studies suggest that inherent or robotic supernumerary fingers could induce remodeling of the cerebral cortex [22]. Our previous research have supported this suggestion [20]. In this paper, we recruited 14 stroke patients and conducted 8 sessions of BCI-controlled \u0026ldquo;sixth-finger\u0026rdquo; neurofeedback training over 2 weeks. EEG data and clinical sacles were evaluated at three time points: baseline, post-train and follow-up period, and EEG of each intervention sessions were also tracked. Try to explore the longitudinal trend of brain activation patterns during full-process of intervention and its correlation with motor recovery of pre-post intervention, so as to better understand the neural mechanisms involved in stroke recovery.\u003c/p\u003e"},{"header":"2. Method","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003e2.1 Design and Setting\u003c/h2\u003e\n \u003cp\u003eThe study was designed as a self-controlled study to investigate the EEG longitudinal trends and the relationship between motor recovery and neuroplastic effects. Fourteen stroke patients (4 females, 10 males, mean age (55.5\u0026thinsp;\u0026plusmn;\u0026thinsp;10.5) years) were recruited from Tianjin Huanhu Hospital to participate in the experiment. The experiment timeline is shown as Fig.\u0026nbsp;1 (A), assessments were performed 3 times, immediately before (baseline) and after the intervention (post-interventon), as well as 1 month after the end of the intervention (follow-up). The invertention was consisted of a total of 8 training sessions spanned over 2 weeks with 4 sessions per week, each training session lasted\u0026thinsp;~\u0026thinsp;60 min, including\u0026thinsp;~\u0026thinsp;15 min of BCI classifier calibration and ~\u0026thinsp;45 min of BCI intervention therapy. The eligibility for participation of all patients was determined by medical professional and all patients signed informed consent. The experimental was consistent with the Declaration of Helsinki on the ethical treatment of human subjects, and passed the ethical approval of the Medical Ethics Committee of Tianjin Huanhu Hospital (IEC-B-003-V3.0).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\n \u003ch2\u003e2.2 Patient Characteristics\u003c/h2\u003e\n \u003cp\u003eTable \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e contains 14 patients demographics and basic information. The inclusion criteria for the patients were as follows: (1) between 30 and 75 years of age; (2) the first occurrence of cerebral hemorrhage, cerebral infarction; (3) disease duration is 3\u0026ndash;12 months, this is to ensure that the period of rapid spontaneous motor recovery has ended [23]; (4) moderate-to-severe impaired with the upper extremity Fugl-Meyer assessment (FMA-UE) [24] score\u0026thinsp;\u0026le;\u0026thinsp;47 points; (4) clearly understand and perform experiment tasks and requirements, no cognitive dysfunction, and a simple Mental State Examination score\u0026thinsp;\u0026gt;\u0026thinsp;20 points. The exclusion criteria were: (1) motor dysfunction of the affected upper limb due to other pathologies; (2) serious psychological condition that might affect the ability to complete the experiment, epilepsy; (3) unable to collect EEG signals due to scalp wounds or other reasons. Patients who terminated or suspended the intervention for personal reasons were excluded from this study.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\n \u003ch2\u003e2.3 The BCI-controlled System\u003c/h2\u003e\n \u003cp\u003eThe BCI-controlled system incorporated an external equipment (robotic \u0026ldquo;sixth-finger\u0026rdquo; combined with sensory electrical stimulation), a commercial EEG amplifier and a neurofeedback interface. As shown in Fig.\u0026nbsp;2 (A), the robotic \u0026ldquo;sixth-finger\u0026rdquo; was worn at the patients\u0026rsquo; wrist on the hemiplegic side and could be controlled to bend and extend with one degree of freedom, the electrodes for sensory electrical stimulation were mounted on forearms of hemiplegic side [20]. The EEG were recorded by a NeuSen W system (Neuracle) amplifier with 64 Ag/AgCl scalp electrodes placed according to the international 10/20 System.\u003c/p\u003e\n \u003cp\u003eThe BCI-controlled framework is shown in Fig. 2 (B), pattern recognition was performed on the EEG signal and converted the recognition results into control commands to provide feedback for patients. The pattern recognition was performed on EEG data by sliding with 1000ms as window length and 200ms as the step distance, thus a recognition result was generated every 200ms. And the Rest state was defined as Label 1, MI state as Label 2. A smoothing filter [25] was used in order to prevent the occurrence of false-positive events. Given \u003cem\u003ex\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e the recognition result at time \u003cem\u003et\u003c/em\u003e (1 or 2 respectively represent Rest or MI) and \u003cem\u003ey\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u0026minus;1\u003c/em\u003e\u003c/sub\u003e the previous window control signal, \u003cem\u003ey\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e the current window control signal which is definded as follows:\u003c/p\u003e\n \u003cdiv id=\"Equa\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e$$\\:{y}_{t}=\\alpha\\:\u0026middot;{x}_{t}+\\left(1-\\alpha\\:\\right)\u0026middot;{y}_{t-1}$$\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\alpha\\:\\)\u003c/span\u003e\u003c/span\u003e \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\in\\:\\left[0.0,\\:1.0\\right]\\)\u003c/span\u003e\u003c/span\u003e is the smoothing factor. A neurofeedback interface based on Psychopy programming maps the control signal \u003cem\u003ey\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e to the bending angle of the virtual \u0026ldquo;sixth-finger\u0026rdquo; in real time (if the recognition result is MI, the bending angle becomes larger, otherwise it becomes smaller). Finally, thresholding strategy (the blue outline in the interface) was used to translate the control signal \u003cem\u003ey\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e into output command for the robotic \u0026ldquo;sixth-finger\u0026rdquo; and sensory electrical stimulation.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\n \u003ch2\u003e2.4 The BCI Intervention\u003c/h2\u003e\n \u003cp\u003eThe experiments were conducted in a quiet and confined room. Two professionals accompanied the patient throughout the entire experiment, one should have expert EEG knowledge to be skilled in installing the EEG cap and use the EEG acquisition system, and the other should be medically qualified to deal with any discomfort of the patient. The subject sat comfortably in a chair at a distance of one-meter from the screen, the stimulation program in the computer screen is completed by Psychopy.\u003c/p\u003e\n \u003cp\u003eAll patients completed a calibration of individual BCI classifier to distinguish EEG activity between MI and Rest state before each training session. During the calibration block, as shown in Fig.\u0026nbsp;1 (B), each trial lasted 14s and is started by pressing space. A green cross appeared at the center of the screen for 7s with the patient remaining at rest, followed by a grey circle lasting 1s, prompting subjects to prepare for the MI tasks. Then the picture cue of bending the \u0026ldquo;sixth-finger\u0026rdquo; appeared to indicate patients to imagine and lasted 6s. The calibration block consist of 3 runs of 20 trials each. EEG data during the rest and MI state were selected to build the BCI classifier.\u003c/p\u003e\n \u003cp\u003eDuring the therapy block, as shown in Fig.\u0026nbsp;1 (C), press the space bar when the patient is ready, followed by a grey circle lasting 1s, prompting subjects to prepare for the MI tasks. The screen then appears with the robotic \u0026ldquo;sixth-finger\u0026rdquo; bending or extending in real time, and with blue outlines representing preset thresholds. The angle of the \u0026ldquo;sixth-finger\u0026rdquo; movement is depending on the patient\u0026rsquo;s brain activity (as described in Method 2.3 and 2.6), recognised by the BCI classifier built in calibration block. If the preset threshold is reached within 6s, i.e., the bending angle of the \u0026ldquo;sixth-finger\u0026rdquo; reaches within the blue outline, a control command is output. The patients receive proprioceptive feedback in the form of robotic \u0026ldquo;sixth-finger\u0026rdquo; bending and sensory electrical stimulation. During the MI period, patients were explicitly asked to avoid any body movements, including speaking and taking their attention away from the \u0026ldquo;sixth-finger\u0026rdquo;. The \u0026ldquo;sixth-finger\u0026rdquo; MI paradigm proposed by our previous studies was adopted [19].\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\n \u003ch2\u003e2.5 Outcome Measures\u003c/h2\u003e\n \u003cp\u003eThree assessments (baseline, post-train, follow-up) all included clinical scales and EEG measure. The clinical outcomes were included FMA-UE and the Barthel index (BI). The primary clinical outcome of this paper is FMA-UE, from 0 to 66 for plegia to normal. And the minimally clinically important difference (MCID) was set to 6.6 (10% of the total range of the scale) [26] for the FMA-UE. BI is secondary outcome, using to evaluate individual\u0026rsquo;s ability to live independently by measuring a range of basic activities such as eating, bathing, and dressing, with a scale from 0 to 100 (best). All clinical scales were measured by professional therapists from Tianjin Huanhu Hospital.\u003c/p\u003e\n \u003cp\u003eThe EEG measure was used as marker of neuroplasticity recovery, including MI task-state and rest-state. The task-state paradigm is shown as Fig.\u0026nbsp;1 (B). The rest-state paradigm is acquisition of the patient\u0026rsquo;s resting EEG for 4 min. And the task-state EEG recorded at assessment and each calibration were used to investigate longitudinal trends of brain activity in response to BCI intervention. The rest-state EEG recorded at assessment were mainly used to analyzed the functional connectivity changes.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n \u003ch2\u003e2.6 EEG data Pre-processing and Analysis\u003c/h2\u003e\n \u003cp\u003eFor the online control in the therapy block, the forty-electrode raw EEG data from Fig.\u0026nbsp;1 (B) were processed by the band-pass filtered between 5 and 20 Hz, and 50 Hz notch filter for removing the power line interference during signals acquisition, and were processed by the common average referenced (CAR). Then, common spatial pattern (CSP) [27] was utilized to extract features of multi-channel EEG information, support vector machine (SVM) [28] was utilized for pattern recognition between MI vs. Rest state.\u003c/p\u003e\n \u003cp\u003eFor the offline analysis, the EEG data in calibration and assessment sessions were selected. In order to further ensure the purity of data, EEG artifacts were manually identified and rejected, abnormal channels were removed by interpolation operation, and independent component analysis (ICA) was adopted to remove eye movement artifacts. The ERD phenomenon is the main brain activation index for MI-BCI. ERSP was used to reflect the event-related power variations of the induced EEG signals in time and frequency domain [29], which defined as follows:\u003c/p\u003e\n \u003cdiv id=\"Equb\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\u003cimg src=\"data:image/png;base64,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\"\u003e\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003ewhere n is the number of trials, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{F}_{k}{(f,t)}^{2}\\)\u003c/span\u003e\u003c/span\u003eis the spectral estimation of \u003cem\u003ek\u003c/em\u003eth trial at frequency \u003cem\u003ef\u003c/em\u003e and time \u003cem\u003et\u003c/em\u003e. And in order to quantify the ERD phenomenon, the baseline correction was conducted by subtracting the average ERSP value within the baseline period.\u003c/p\u003e\n \u003cp\u003eThe weighted phase-lag index (wPLI) was used to calculate the phase synchronization of two pairs of channels for connectivity measure. This method has been proved to be able to overcome the problem of volume conduction and is insensitive to irrelevant noise sources [30]. The non-directed coherence measures for 40*40 EEG electrode pairings wPLI [31] were calculated as follows:\u003c/p\u003e\n \u003cdiv id=\"Equc\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e$$\\:wPLI=\\frac{\\left|\\right\\{S\\left\\}\\right|}{\\left\\{\\left|S\\right|\\right\\}}=\\frac{\\left|\\right\\{\\left|\\text{S}\\right|sign\\left(S\\right)\\left\\}\\right|}{\\left\\{\\left|S\\right|\\right\\}}$$\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003ewhere S means the imaginary component of cross-spectrum between channel A\u003csub\u003e1\u003c/sub\u003e and channel A\u003csub\u003e2\u003c/sub\u003e, {\u0026bull;} means the expected value operator. The value of wPLI ranges from 0\u0026thinsp;~\u0026thinsp;1, with high wPLI value representing strong coupling of neural oscillatory activity. Brain functional network and their parameters such as node degree and small-world index were also analysed. And the threshold selection principle for establishing connections was 1) maximum threshold without isolated nodes and isolated parts; 2) with small-world attribute.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\n \u003ch2\u003e2.7 Statistical Analyses\u003c/h2\u003e\n \u003cp\u003eFor continuous variables, the Lilliefors-corrected Kolmogorov-Smirnov test was first used to check the Gaussian distribution. Paired samples t-test was utilized to evaluate the statistical significance of changes in EEG data and clinical metrics before and post therapy. The EEG metrics such as wPLI values and small-world attribute and clinical metrics such as FMA-UE and BI of each patients in the baseline (T0) and post-train (T1) period were regarded as paired samples.\u003c/p\u003e\n \u003cp\u003eTo explore the potential mechanisms and longitudinal trends of neural patterns, correlations between variables were measured by Pearson\u0026rsquo;s correlation coefficient. Longitudinal trends in brain modulation patterns during full-process intervention were evaluated by calculating the Pearson\u0026rsquo;s correlation between mean ERSP value and the number of therapy sessions from baseline to post-train. And the Pearson\u0026rsquo;s correlation between the clinical metrics change and EEG metrics change (baseline and post-train) was conducted to reflect the relationship between motor improvement and neural plasticity before and post rehabilitation.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\n \u003ch2\u003e3.1 ERD Longitudinal Trends\u003c/h2\u003e\n \u003cp\u003eIn order to investigate the longitudinal trends of ERD distribution and strength over the sessions, the separate average ERD topographical maps in mu rhythm were shown in Fig.\u0026nbsp;3 (A). And as shown in Fig.\u0026nbsp;3 (B), due to the inconsistency in the hemiplegic sides of the patients (four patients with left hemiplegia, the rest with right hemiplegia), the electrodes were mirrored by setting the left side of topographical distribution as the contralateral side of hemiplegic, and the right side of topographical distribution as the ipsilateral side. The results show that bilateral ERD activation in sensorimotor cortex was induced when patients performed MI tasks, and consistently maintained global ERD activation in brain, in the first few sessions (week 1). As the session progressed, the bilateral brain modulation of patients was weakened (week 2). Until the post-train session, the ERD phenomenon focused on sensorimotor areas and contralaterally modulated was found, which more closely resembles the brain modulation patterns when healthy individuals performed unilateral limb MI. After a one-month blanking period (week 7), the obvious sensorimotor area modulation and weak contralateral dominance was still maintained.\u003c/p\u003e\n \u003cp\u003eCorrelation coefficients between the number of sessions (ten sessions from baseline to post-train) and the amplitude of ERD for each electrode were given in Fig. 4 and Table 1 to quantify the longitudinal trend in ERD patterns. As shown in Fig. 4 (A), degree of ERD activation in nine electrodes seems to significantly anti-correlate with the sessions, mainly distributed in the periphery of FC line, C line and CP line. The r and \u003cem\u003ep\u003c/em\u003e for nine electrodes with correlation were given in Table 1. The contralateral and ipsilateral electrodes with correlation were averaged and calculated correlation coefficient (F5\u0026thinsp;+\u0026thinsp;FC5\u0026thinsp;+\u0026thinsp;FC3\u0026thinsp;+\u0026thinsp;C5\u0026thinsp;+\u0026thinsp;C3, r = -0.709, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.012\u003csup\u003e*\u003c/sup\u003e; C4\u0026thinsp;+\u0026thinsp;C6\u0026thinsp;+\u0026thinsp;Cp4\u0026thinsp;+\u0026thinsp;Cp6, r = -0.752, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.021\u003csup\u003e*\u003c/sup\u003e), as shown in Fig.\u0026nbsp;4 (B), the intensity of ERD activation in peripheral brain regions gradually decreased across session progressed. This corroborates that after training, ERD activation is more focused on the central of sensorimotor areas and the functional compensatory effects from other brain areas are suppressed.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\n \u003ch2\u003e3.2 Alterations in EEG Connectivity\u003c/h2\u003e\n \u003cp\u003eIn order to investigate the functional integration and close connections between different brain regions before and after rehabilitation, functional connectivity and statistical comparison based on the wPLI in mu rhythm were implemented. As shown in Fig.\u0026nbsp;5 (A), the connected edges of the total 780 electrode pairs that exceeded the threshold at baseline and post-train sessions was drawn, and the pairwise connectivity with significant differences (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) between baseline and post-train sessions also were drawn. \u0026ldquo;\u0026gt;\u0026rdquo; means the significantly increased amount of connectivity. The result shown that more synchronized neural activity between different brain regions was found after rehabilitation compared baseline. Functional connectivity values were significantly enhanced for 22 electrode pairs and significantly reduced for no electrode pairs. The average degrees across all electrodes and small-world attribute of the brain functional network before and after rehabilitation were shown in Fig.\u0026nbsp;5 (B). It could be seen that the degrees across all electrodes increased (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.0598) after rehabilitation, and the electrodes with increased degree are mainly concentrated in central of the sensorimotor area (blue dashed box) as shown in Fig.\u0026nbsp;5 (C). The small-world attribute of the brain functional network also increased (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.1144) after rehabilitation which indicated that brain network become more efficient and information is processed more efficiently.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\n \u003ch2\u003e3.3 Clinical Outcome Metrics\u003c/h2\u003e\n \u003cp\u003eTable 2 reports the clinical characteristics and clinical metrics (T0: baseline, T1: Post-train, T3: Follow-up) of the stroke patients. Two patients (ID 11 and 12) were unable to be evaluated in follow-up period due to personal reasons. There was a significant improvement on the primary clinical outcome FMA-UE scores of patients between post (T1) and before (T0) BCI treatment (\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01), with an average increase of 7.9 points. Meanwhile, patients still maintained improvement in the 1-month follow-up (T2) after the end of therapy. More than half of patients (9/14, ID: 02, 04, 06, 08, 10, 11, 12, 13, 14) reached the MCID of 6.6 points change for FMA-UE after therapy. The secondary clinical metrics BI also showed a significant improvement between post-before BCI treatment (with an average increase of 7.1 points, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01) and was also maintained until the follow-up period. The changes of FMA-UE and BI scores for three time points are illustrated in Fig. 6.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e3.4 Neurophysiological Correlates\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eIn order to further investigate the possible role of EEG pattern changes in the improvement of motor function, the correlation between the changes in resting-state functional connectivity and improvements in FMA-UE scores before and post-train was analyzed. As shown in Fig. 7, an obvious positive correlation relationship emerged between FMA-UE score changes and the changes in the small-world attribute and degree of the resting-state brain functional network (small-world attribute: Pearson\u0026rsquo;s correlation, r = 0.502, \u003cem\u003ep\u003c/em\u003e = 0.067, degree: Pearson\u0026rsquo;s correlation, r = 0.436, \u003cem\u003ep\u003c/em\u003e = 0.119).\u003c/p\u003e\n\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eIn this study, we aimed to explore the longitudinal trend of EEG patterns during the full-process of BCI therapy in stroke patients, providing more evidence for a detailed understanding of the mechanism of neural plasticity. This study was conducted based on our previous construction of a BCI-controlled \u0026ldquo;sixth-finger\u0026rdquo; system [20]. Our previous studies have demonstrated the advantages of the \u0026ldquo;sixth-finger\u0026rdquo; MI paradigm [19] (applied to this study) and the role of brain-controlled the \u0026ldquo;sixth-finger\u0026rdquo; training in promoting neural plasticity on healthy people [20]. MI-induced ERD phenomenon and resting-state functional connectivity revealed brain pattern changes. FMA-UE and BI were used as clinical metrics, reflecting the improvement of patients\u0026rsquo; motor function.\u003c/p\u003e\n\u003cp\u003e4.1 Longitudinal Trends in EEG Pattern\u003c/p\u003e\n\u003cp\u003eIn this paper, longitudinal trends in EEG patterns were reported, which are important for a detailed understanding of the process and mechanism in neuroplasticity, but are poorly described in other literature. The results (Fig. 3) showed that the patient\u0026rsquo;s neural patterns seemed to go through three phases:\u003c/p\u003e\n\u003cp\u003e(1) at baseline and the first week of training, the extent of ERD activation gradually increased in the sensorimotor cortex and even in the prefrontal cortex when the patients performed the MI tasks. On the one hand, this may be related to the novel MI paradigm that the patients were asked to perform in training. In our previous study [19], we proposed this MI paradigm based on the \u0026ldquo;sixth-finger\u0026rdquo; (imagine controlling the movement of the \u0026ldquo;sixth-finger\u0026rdquo;) and observed a larger range and stronger degree of ERD activation in healthy people, compared with the traditional MI paradigm based on the inherent hand (imagine the movement of inherent hand). Indeed, it is also widely believed that learning new movements leads to a gradual increase in brain activation in the relevant limb motor cortical areas [32]. And Pascual-Leone et al. [33] reported that learning a sequence over a few days was generally induced an increase in the size of the motor map. The stroke patients receiving a mechanical extra finger and trying to imagine controlling it, can indeed be considered to be learning a new movement or skill. On the other hand, it is possible that the compensatory mechanism of neural function plays a role. Compensation and involvement of other areas (ipsilateral sensorimotor area and prefrontal lobe) enlarged the activated brain areas and deviated from the contralateral advantage activation that is natural performance during hand MI. Bai, O.\u0026rsquo;s research also showed that neural activation may \u0026lsquo;spill over\u0026rsquo; to the ipsilateral cortex when the processing load of the contralateral motor cortex increases [34].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e(2) At the second week of training and post-train assessment, the ERD phenomenon gradually weakened and became more focused on the sensorimotor cortex associated with the movement of hand. As shown in Fig. 4 and Table 1, the ERD of electrodes in the periphery of the motor area and parietal lobe weakens over process of training and shows a \u0026nbsp;significant correlation with the number of sessions. Some studies have shown that once the neural process is optimized, activation in cortex appears to weaken [35, 36], and our results are consistent with this conclusion. And Steele, C. [37] has reported that during initial gesture motor learning, neural activity in motor cortex increases, whereas after acquisition and consolidation of the motor task, activity decreases. In addition, the theory of \u0026ldquo;neural efficiency\u0026rdquo; [38] has also shown that an expert in a motor task displayed less pronounced and more spatially focused mu rhythm ERD. This gives us evidence to believe that the weakening and more focused ERD phenomenon over sessions is due to the improved efficiency of neural recruitment, which promotes the process of neural plasticity. The EEG analysis of resting state functional connectivity in this paper also supports this conclusion. As shown in Fig. 5, the strength of resting-state functional connection in the post-train period was significantly greater than that in the baseline period. Furthermore, the node degree and small-world attribute are higher after therapy, indicating that the overall processing efficiency of the brain functional network has been significantly improved [39], especially the sensorimotor area shown in Fig. 5 (C).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e(3) At follow-up assessment, the ERD phenomenon still showed natural and contralateral activation patterns, which was similar to the post-train assessment, although with some rebound. This indicates the improvement effect of neural plasticity was still maintained until the follow-up period, which was consistent with the improvement of motor function\u003c/p\u003e\n\u003cp\u003e(Fig. 6). And the rebound of ERD may be due to the fact that patients became unfamiliar with the \u0026ldquo;sixth-finger\u0026rdquo; or MI tasks after a month of rest.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e4.2 Clinical Metrics Outcome Improvement\u003c/p\u003e\n\u003cp\u003eAlthough this paper focuses on exploring the longitudinal change trend of brain neural activity caused by BCI therapy, significant clinical functional improvement is still necessary for stroke rehabilitation. Overall, BCI therapy results in a significant clinical improvement, with an average increase of 7.9 points in FMA-UE and an average increase of 7.1 points in BI between post (T1) and before (T0) BCI therapy. And this improvement lasted until 1 month after the end of training. Many previous BCI clinical studies contained no follow-up assessment.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe stroke patients recruited in this study were all those with onset time of 3 to 10 months and the spontaneous motor recovery almost stagnated [23]. And patients with moderate-to-severely functional impairment are less likely to recover from other methods [11], while patients with poor function (ID:10, 12) in this study have significantly improved after rehabilitation. These indicate that the improvement of the stroke patients in this study results from the BCI therapy they received. In addition, the introduction of MCID can make the improvement of clinical metrics more convincing, which is defined as the smallest change that is clinically important to patients or clinicians [40]. More than half of patients (9/14, with \u003csup\u003e#\u003c/sup\u003e marked) reached the MCID of 6.6 points change for FMA-UE after therapy.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eCo-analysis of clinical outcomes with EEG analysis may provide evidence for possible mechanisms behind functional improvements of stroke patients. The results (Fig. 7) showed that the enhancement in resting-state functional connectivity was correlated with the improvement in FMA-UE outcomes. Moreover, the nodes with enhanced connection strength were mainly concentrated in the sensorimotor area shown in Fig.5 (C), which is consistent with the conclusion of previous studies that the reactivation of the motor area seems to be crucial for motor recovery [18]. The global enhancement of functional connectivity and the improvement of node degree and small-world attribute indicate the improvement of brain processing efficiency, which seems to confirm the theory of neural efficiency mentioned above. Although the assertion of a rehabilitation mechanism based on the above correlation analysis is somewhat limited, the results show that the recovery of neurological function has played a role in promoting the improvement of motor function outcomes.\u003c/p\u003e\n\u003cp\u003e4.3 Limitations\u003c/p\u003e\n\u003cp\u003eHowever, there are still some limitations that need to be overcome to make our research more comprehensive and scientific. First, the study was limited by the number of clinical cases. 14 patients may not make the results generalizable, and patients did not show consistent clinical recovery results. A larger sample size is needed in the future to confirm the preliminary findings of this study. Secondly, stroke patients differ in lesion type and onset time. Although the impact of this difference is not the main focus of this study, it reflects the actual clinical needs. Are there different characteristics of neural change trends for different types of patients? Should different clinical rehabilitation methods and doses be used? This is the ultimate goal of studying the stroke rehabilitation mechanism. To achieve this goal, further researches may be needed on different types of stroke patients in subsequent work. Finally, there was a lack of blank or traditional control groups. Although we recruited stroke patients who theoretically had little spontaneous recovery ability, we still lacked actual quantitative data to confirm the ineffectiveness of patients\u0026rsquo; recovery without training or with traditional rehabilitation therapy. In this paper, we focused more attention to the change trends in EEG patterns and neurorehabilitation mechanisms. In subsequent work, we will provide more sufficient evidence to prove the effectiveness of BCI therapy.\u003c/p\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eIn this paper, the longitudinal trend in neural patterns over full-process of intervention and the potential relationship between neural and motor recovery in stroke patients were reported. We found that the longitudinal change trend of EEG showed three stages during the intervention process: ERD gradually increased in the first week of training, weakened and focused on the contralateral sensorimotor area in the second week, and showed a significant correlation over sessions, remaining focused and contralateral pattern in the follow-up period. Motor function between pre-post intervention showed significant improvement by clinical metrics, with +\u0026thinsp;7.9 in FMA-UE and +\u0026thinsp;7.1 in BI. Meanwhile the improvement was maintained until the one-month follow-up after the end of therapy. In addition, improvement of motor function is associated with the enhancement of resting-state functional connectivity. The findings of this study may be important for detailed understanding of the mechanism of neural plasticity, and subsequent studies that attempt to track the neurological patterns of patients during intervention.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cem\u003eAcknowledgements\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe authors sincerely thank all patients for their voluntary participation.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAuthor contributions\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eY. Liu led the design and implementation of the experiment, guiding data processing and manuscript writing; Z. Wang developed an online neural feedback interface and was responsible for conducting experiments, processing EEG data, and writing the manuscript; S. Huang was responsible for the development of the BCI system and participated in the design and implementation of experiments; H. Huang was responsible for summarizing and analyzing clinical outcomes and participates in the experiments; W. Wu, Y. Wang participated in the experiment and was responsible for data collection; X. An, D. Ming was responsible for managing experiments, interpreting data, and reviewing manuscript; J. Wu, Y. Li was responsible for clinical work.\u0026nbsp;All authors reviewed the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eFunding\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported in part by the National Key Research and Development Program of China (2023YFF1204300, 2023YFF1204305), National Natural Science Foundation of China (62273251), Research Project of State Key Laboratory of Mechanical System and Vibration (MSV202418).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAvailability of data and materials\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eEthics approval and consent to participate\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAll patients signed informed consent. The experimental was consistent with the Declaration of Helsinki on the ethical treatment of human subjects, and passed the ethical approval of the Medical Ethics Committee of Tianjin Huanhu Hospital (IEC-B-003-V3.0).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eConsent for publication\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eConsent for publication were given by all participants\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCompeting interests\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they do not have any competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAuthors details\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003csup\u003e1\u003c/sup\u003eAcademy of Medical Engineering and Translational Medicine, Tianjin University, Tianjin, 300072, China;\u003c/p\u003e\n\u003cp\u003e\u003csup\u003e2\u003c/sup\u003eDepartment of Rehabilitation Medicine, Tianjin Huanhu Hospital, Tianjin, 300350, China;\u003c/p\u003e\n\u003cp\u003e\u003csup\u003e3\u003c/sup\u003eClinical College of Neurology, Neurosurgery and Neurorehabilitation, Tianjin Medical University, Tianjin, 300370, China;\u003c/p\u003e\n\u003cp\u003e\u003csup\u003e4\u003c/sup\u003eDepartment of Neurology, Tianjin Huanhu Hospital, Tianjin 300350, China;\u003c/p\u003e\n\u003cp\u003e\u003csup\u003e5\u003c/sup\u003eTianjin Key Laboratory of Cerebral Vascular and Neurodegenerative Diseases, Tianjin Neurosurgical Institute, Tianjin 300350, China.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eStinear, C.M., et al., Advances and challenges in stroke rehabilitation. The Lancet Neurology, 2020. 19(4): p. 348-360.\u003c/li\u003e\n \u003cli\u003eGeorge, Paul M. and Gary K. Steinberg, Novel Stroke Therapeutics: Unraveling Stroke Pathophysiology and Its Impact on Clinical Treatments. Neuron, 2015. 87(2): p. 297-309.\u003c/li\u003e\n \u003cli\u003eTeasell, R., N.A. Bayona, and J. Bitensky, Plasticity and reorganization of the brain post stroke. Topics in stroke rehabilitation, 2005. 12(3): p. 11-26.\u003c/li\u003e\n \u003cli\u003eHallett, M., Plasticity of the human motor cortex and recovery from stroke. Brain Research Reviews, 2001. 36(2-3): p. 169-174.\u003c/li\u003e\n \u003cli\u003eJia, J., Exploration on neurobiological mechanisms of the central-peripheral-central closed-loop rehabilitation. Frontiers in Cellular Neuroscience, 2022. 16.\u003c/li\u003e\n \u003cli\u003eSharma, N. and J.-C. Baron, Does motor imagery share neural networks with executed movement: a multivariate fMRI analysis. Frontiers in Human Neuroscience, 2013. 7.\u003c/li\u003e\n \u003cli\u003eXu, L., et al., MOTOR EXECUTION AND MOTOR IMAGERY: A COMPARISON OF FUNCTIONAL CONNECTIVITY PATTERNS BASED ON GRAPH THEORY. Neuroscience, 2014. 261: p. 184-194.\u003c/li\u003e\n \u003cli\u003eChen, L., et al., EEG-controlled functional electrical stimulation rehabilitation for chronic stroke: system design and clinical application. Frontiers of Medicine, 2021. 15(5): p. 740-749.\u003c/li\u003e\n \u003cli\u003eChowdhury, A., et al., Active Physical Practice Followed by Mental Practice Using BCI-Driven Hand Exoskeleton: A Pilot Trial for Clinical Effectiveness and Usability. IEEE J Biomed Health Inform, 2018. 22(6): p. 1786-1795.\u003c/li\u003e\n \u003cli\u003eMang, J., et al., Favoring the cognitive-motor process in the closed-loop of BCI mediated post stroke motor function recovery: challenges and approaches. Frontiers in Neurorobotics, 2023. 17.\u003c/li\u003e\n \u003cli\u003eBundy, D.T., et al., Contralesional Brain-Computer Interface Control of a Powered Exoskeleton for Motor Recovery in Chronic Stroke Survivors. Stroke, 2017. 48(7): p. 1908-1915.\u003c/li\u003e\n \u003cli\u003eTsuchimoto, S., et al., Sensorimotor Connectivity after Motor Exercise with Neurofeedback in Post-Stroke Patients with Hemiplegia. Neuroscience, 2019. 416: p. 109-125.\u003c/li\u003e\n \u003cli\u003eWang, X., et al., Differentiated Effects of Robot Hand Training With and Without Neural Guidance on Neuroplasticity Patterns in Chronic Stroke. Front Neurol, 2018. 9: p. 810.\u003c/li\u003e\n \u003cli\u003eRanzani, R., et al., Neurocognitive robot-assisted rehabilitation of hand function: a randomized control trial on motor recovery in subacute stroke. J Neuroeng Rehabil, 2020. 17(1): p. 115.\u003c/li\u003e\n \u003cli\u003eBaniqued, P.D.E., et al., Brain-computer interface robotics for hand rehabilitation after stroke: a systematic review. J Neuroeng Rehabil, 2021. 18(1): p. 15.\u003c/li\u003e\n \u003cli\u003eBai, Z., et al., Immediate and long-term effects of BCI-based rehabilitation of the upper extremity after stroke: a systematic review and meta-analysis. J Neuroeng Rehabil, 2020. 17(1): p. 57.\u003c/li\u003e\n \u003cli\u003eBrunner, I., et al., Brain computer interface training with motor imagery and functional electrical stimulation for patients with severe upper limb paresis after stroke: a randomized controlled pilot trial. J Neuroeng Rehabil, 2024. 21(1): p. 10.\u003c/li\u003e\n \u003cli\u003eBiasiucci, A., et al., Brain-actuated functional electrical stimulation elicits lasting arm motor recovery after stroke. Nat Commun, 2018. 9(1): p. 2421.\u003c/li\u003e\n \u003cli\u003eLiu, Y., et al., EEG characteristic investigation of the sixth-finger motor imagery and optimal channel selection for classification. Journal of Neural Engineering, 2022. 19(1).\u003c/li\u003e\n \u003cli\u003eLiu, Y., et al., Functional Reorganization After Four-Week Brain-Computer Interface-Controlled Supernumerary Robotic Finger Training: A Pilot Study of Longitudinal Resting-State fMRI. Front Neurosci, 2021. 15: p. 766648.\u003c/li\u003e\n \u003cli\u003eHussain, I., et al., A soft supernumerary robotic finger and mobile arm support for grasping compensation and hemiparetic upper limb rehabilitation. Robotics and Autonomous Systems, 2017. 93: p. 1-12.\u003c/li\u003e\n \u003cli\u003eKieliba, P., et al., Robotic hand augmentation drives changes in neural body representation. Science Robotics, 2021. 6(54).\u003c/li\u003e\n \u003cli\u003eDuncan PW, Goldstein LB, Matchar D, Divine GW, Feussner J. Measurement of motor recovery after stroke. Outcome assessment and sample size requirements. Stroke. 1992;23:1084\u0026ndash;1089.\u003c/li\u003e\n \u003cli\u003eFugl-Meyer AR, J\u0026auml;\u0026auml;sk\u0026ouml; L, Leyman I, Olsson S, Steglind S. The post-stroke hemiplegic patient. 1. A method for evaluation of physical performance. Scand J Rehabil Med. (1975) 7:13\u0026ndash;31.\u003c/li\u003e\n \u003cli\u003eTonin, L., F.C. Bauer, and J. del R. Millan, The Role of the Control Framework for Continuous Teleoperation of a Brain\u0026ndash;Machine Interface-Driven Mobile Robot. IEEE Transactions on Robotics, 2020. 36(1): p. 78-91.\u003c/li\u003e\n \u003cli\u003evan der Lee, J.H., et al., The responsiveness of the action research arm test and the Fugl-Meyer Assessment scale in chronic stroke patients. Journal of Rehabilitation Medicine, 2001. 33(3): p. 110-113.\u003c/li\u003e\n \u003cli\u003eBen Hamed, S., M.H. Schieber, and A. Pouget, Decoding M1 neurons during multiple finger movements. Journal of Neurophysiology, 2007. 98(1): p. 327-333.\u003c/li\u003e\n \u003cli\u003eWang, L., et al., Analysis and classification of speech imagery EEG for BCI. Biomedical Signal Processing and Control, 2013. 8(6): p. 901-908.\u003c/li\u003e\n \u003cli\u003eYi, W., et al., EEG feature comparison and classification of simple and compound limb motor imagery. Journal of Neuroengineering and Rehabilitation, 2013. 10.\u003c/li\u003e\n \u003cli\u003eLau, T.M., et al., Weighted phase lag index stability as an artifact resistant measure to detect cognitive EEG activity during locomotion. Journal of Neuroengineering and Rehabilitation, 2012. 9.\u003c/li\u003e\n \u003cli\u003eLee, M., Y.-H. Kim, and S.-W. Lee, Motor Impairment in Stroke Patients Is Associated With Network Properties During Consecutive Motor Imagery. IEEE Transactions on Biomedical Engineering, 2022. 69(8): p. 2604-2615.\u003c/li\u003e\n \u003cli\u003eFloyer-Lea, A. and P.M. Matthews, Distinguishable brain activation networks for short- and long-term motor skill learning. Journal of Neurophysiology, 2005. 94(1): p. 512-518.\u003c/li\u003e\n \u003cli\u003ePascual-Leone, A., et al., Modulation of muscle responses evoked by transcranial magnetic stimulation during the acquisition of new fine motor skills. Journal of neurophysiology, 1995. 74(3): p. 1037-45.\u003c/li\u003e\n \u003cli\u003eBai, O., et al., Asymmetric spatiotemporal patterns of event-related desynchronization preceding voluntary sequential finger movements: a high-resolution EEG study. Clinical Neurophysiology, 2005. 116(5): p. 1213-1221.\u003c/li\u003e\n \u003cli\u003eHaslinger, B., et al., Reduced recruitment of motor association areas during bimanual coordination in concert pianists. Human Brain Mapping, 2004. 22(3): p. 206-215.\u003c/li\u003e\n \u003cli\u003eJ\u0026auml;ncke, L., N.J. Shah, and M. Peters, Cortical activations in primary and secondary motor areas for complex bimanual movements in professional pianists. Cognitive Brain Research, 2000. 10(1-2): p. 177-183.\u003c/li\u003e\n \u003cli\u003eSteele, C.J. and V.B. Penhune, Specific Increases within Global Decreases: A Functional Magnetic Resonance Imaging Investigation of Five Days of Motor Sequence Learning. Journal of Neuroscience, 2010. 30(24): p. 8332-8341.\u003c/li\u003e\n \u003cli\u003eBabiloni, C., et al., \u0026quot;Neural efficiency\u0026quot; of experts\u0026apos; brain during judgment of actions: A high-resolution EEG study in elite and amateur karate athletes. Behavioural Brain Research, 2010. 207(2): p. 466-475.\u003c/li\u003e\n \u003cli\u003eLiu, M., et al., Effects of Transcranial Direct Current Stimulation on EEG Power and Brain Functional Network in Stroke Patients. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2023. 31: p. 335-345.\u003c/li\u003e\n \u003cli\u003eJaeschke, R., J. Singer, and G.H. Guyatt, Measurement of health status. Ascertaining the minimal clinically important difference. Controlled clinical trials, 1989. 10(4): p. 407-15.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTables 1 to 2 are available in the Supplementary Files section\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"brain–computer interface, longitudinal trend, motor imagery, stroke, full-process","lastPublishedDoi":"10.21203/rs.3.rs-6027743/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6027743/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground \u003c/strong\u003eBrain-computer interface (BCI) is used in stroke rehabilitation to match brain activity with contingent feedback to establish closed-loop pathways and provide a measure of neuroplasticity changes of patients. However, most studies assessed neural function only at pre- and post-intervention, thereby longitudinal trends of neural patterns and mechanismsduring full-process of intervention remain unclear.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e Forty stroke patients were recruited to receive a total of 8 sessions motor imagery-based (MI-based) BCI-controlled “sixth-finger” neurofeedback intervention, 4 sessions per week for 2 weeks. Electroencephalography (EEG) measure and clinical scales were assessed at three time points: baseline, post-train and 1month follow-up period, and EEG data of each intervention sessions were also tracked. ERD phenomenon induced by MI and resting-state functional connectivity were used to reflect the longitudinal trends and pre-post changes in neural activity. The upper extremity Fugl-Meyer assessment (FMA-UE) and Barthel index (BI) were used to reflect the motor improvement.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e EEG longitudinal trend shows three phases over full-process of intervention: ERD gradually increased in the first week of training, weakened and focused on the contralateral sensorimotor area in the second week, and showed a significant correlation over sessions, remaining focused and contralateral pattern in the follow-up period. And resting-state functional connectivity increased after intervention. Motor function between pre-post intervention showed significant improvement by clinical metrics, with + 7.9 in FMA-UE and + 7.1 in BI. More than half of patients (9/14) reached the minimally clinically important difference (MCID) of 6.6 points change for FMA-UE after therapy. Meanwhile the improvement was maintained until the one-month follow-up after the end of therapy. In addition, improvement of motor function is associated with the enhancement of resting-state functional connectivity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e\u003cem\u003e \u003c/em\u003eThis work reveals longitudinal trend of neural patterns over full-process of intervention and its correlation with motor recovery, so as to provide more evidence for a detailed understanding of the mechanism of neural plasticity.\u003c/p\u003e","manuscriptTitle":"ERD Full-process Longitudinal Trend and Pre-post Motor Recovery Under BCI-controlled Sixth-finger Neurofeedback Intervention in Stroke Patients","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-02-18 14:03:55","doi":"10.21203/rs.3.rs-6027743/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"8bc856bf-1c77-4339-a9de-1dc4c7a2fe67","owner":[],"postedDate":"February 18th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-02-18T14:03:55+00:00","versionOfRecord":[],"versionCreatedAt":"2025-02-18 14:03:55","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6027743","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6027743","identity":"rs-6027743","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}