Brain-Computer Interface Visualization Training to Optimize Muscle Activation Following Orthopaedic Surgery

1. Introduction Background Patients recovering from orthopedic surgical procedures require a comprehensive physical rehabilitation process to help recover pre-operative functional mobility and strength. A limiting factor in physical rehabilitation is a patient's inability to activate the involved muscle groups postoperatively, a phenomenon termed Arthrogenic Muscle Inhibition (AMI) \[1, 2\]. AMI is a complex neurological process where the injury or surgery disrupts sensory and motor neurological pathways, resulting in decreased muscle activation and strength. AMI can be a major obstacle to a patient's return to normal mobility and muscular function \[2\]. For example, patients who have undergone anterior cruciate ligament reconstruction (ACLR), may experience ineffective quadriceps activation and persistent hamstring contracture, leading to loss of passive and active range of motion (ROM). Even with standard physical therapy rehabilitation, patients with AMI have ineffective recovery due to decreased muscular activation and movement dysfunction \[3\]. Visualization training with neurofeedback therapy (NFVT) is a non-invasive method that could be used with standard post-operative physical rehabilitation to decrease AMI and help patients recover pre-operative functional mobility and strength. The motor regions of the brain (motor cortex) play a crucial role in planning, controlling, and executing voluntary movements. The motor cortex is not only active during actual movement; mentally rehearsing motor acts without physically moving also activates the motor cortex \[4\]. For example, one could imagine themselves performing squats without squatting (visualization), this process activates the brain regions related to squatting. Such visualization training can enhance the brain's ability to plan, control and execute movement without physical load on the body \[5\]. Theoretically, this training could help restore disrupted neurological pathways, leading to reduced AMI and improved patient recovery after surgery \[6\]. When a brain region has heightened activity, passive sensors on the scalp can detect the increased electrical activity, this technique is known as electroencephalography (EEG). A computer can process the EEG signal and provide users with real time feedback on their concentration level and whether they are activating their motor cortex through mental visualization of movements (neurofeedback). This feedback process enhances the visualization training \[6\]. This study aims to investigate the effect of visualization with neurofeedback on postoperative recovery in patients undergoing physical rehabilitation from 4 orthopedic surgical procedures: anterior cruciate ligament reconstruction (ACLR), total knee arthroplasty (TKA), total hip arthroplasty (THA), and hip arthroscopy (HA) for femoroacetabular impingement syndrome (FAIS). More specifically, neurofeedback training will be implemented using a novel technology developed by i-BrainTech™. The findings of this study have the potential to revolutionize physical rehabilitation protocols for patients, offering a novel approach that integrates visualization therapy with neurofeedback to enhance standard physical rehabilitation. This could lead to faster, more complete recoveries, and potentially mitigate the long-term impacts of AMI. The successful application of this technology would also help deepen current understanding of neuroplasticity, specifically the malleability of the neuromuscular pathways and how this can improve motor control. Purpose The purpose of this study is to investigate the effect of NFVT on postoperative recovery in patients rehabilitating from orthopedic surgeries. Hypothesis It was hypothesized that through targeted NFVT using i-BrainTech™, post-surgical participants will experience improved muscle activation, which in turn will contribute to better rehabilitation outcomes, strength, such as range of motion, and functional mobility. We also hypothesize that these improvements seen throughout the recovery period will have a positive impact on short-term patient-reported outcome surveys (PROs). 2. Methods Settings \& Locations Conducted at Rush University Medical Center, specifically within: * The main campus of Midwest Orthopaedics at Rush (MOR) Sofija and Jorge O. Galante Orthopedic Building, 1611 W Harrison St, Chicago, IL 60612 * Motion Laboratory in the MOR Orthopedic Building * Physical therapy facility (Chicago location) 3. Interventions Control group: Standard post-surgical rehabilitation therapy Intervention Group: Standard post-surgical rehabilitation therapy + i-BrainTech neurofeedback training * Procedure: Patients use EEG-based neurofeedback twice a week until 8 weeks post-operatively. * Neurofeedback setup: EEG cap monitors motor cortex activation, guiding visualization exercises * Training sessions: Patients visualize movements, and EEG feedback helps optimize motor activation Physical Therapy Patients will follow a standard physical therapy protocol. The protocol will be assigned by their respective surgeon who conducted the procedure and will be specific to the procedure that the patient underwent. The standardized physical therapy protocols will be attached in supplemental materials. Treatment Group Intervention: The main study intervention for the treatment group involves NFVT using the i-BrainTech™ Platform. This is a technology that uses electroencephalography (EEG) to read the electrical activities in the brain \[7\]. Active neurons in the brain causes change in electrical activities on the scalp, detectable by electrodes placed on the scalp. The sensing electrodes are completely passive, incapable of sending electrical current to the wearer. An EEG cap will be used with sensing electrodes aligned to the frontal cortex and the motor cortex. The detected electrical activity from these locations of the scalp will transmitted to the computer, which allows for assessment of focus and motor cortex activity \[8\]. There are multiple cap sizes to ensure a comfortable fit. By concentrating and imagining themselves performing the rehabilitation movements (visualization), patients activate their own motor and pre-frontal cortices. The EEG sensors detect the increased brain electrical activity, and the i-BrainTech™ software translates the EEG signal into a virtual avatar figure performing such movements on a computer monitor, providing feedback to the patients on their visualization efforts (neurofeedback). Patients will effectively play a video game using their own brain signals. By turning this feedback process into a video game, the i-BrainTech™ platform provides an incentive for the user to intensely focus and visualize the rehabilitation exercises, and in the process activate and strengthen the neural pathways responsible for these rehabilitation movements. The repeated activation of neural pathways theoretically improves their muscle control and reduces AMI \[6\]. Instruction to Participants (how to play the "game") The i-BrainTech™ training station has a laptop and EEG caps. Participants will be seated in front of the laptop and put on the appropriately sized EEG cap. A conductive gel is injected into 2 insertion points on the cap. The column of gel touches the participant's skin on one side and the sensor on the other. The gel is water-soluble and dries up in chunks and is not sticky. The conductive gel is routinely used in the clinic and pre-operative area for ultrasound. The wet gel can be wiped off with a paper towel and the dried gel can be pulled off the participant's scalp as it does not stick to hair. The remaining fragments will be washed away when the patient showers. The session will be started and the i-BrainTech™ software will provide on screen prompts and feedback to the user. First is a 2-minute calibration period. During this time, the user is prompted to relax their mind so baseline brain activity may be detected. The brain activity above the baseline is used to control the cartoon avatar performing rehabilitation exercises. After calibration, a 20-minute NFVT session begins. Participants are prompted to imagine themselves performing various rehabilitation exercises (visualization). The selections of exercises are the rehabilitation exercises they will eventually perform at a physical therapy session, specific to their surgical procedures (Table 1). The software provides real time feedback on how concentrated the user is with the task, and how well the user is at visualizing the specific therapy exercises. The video game incentivizes participants to concentrate on the visualization therapy to maximize their score. When the participant finishes the i-BrainTech™ training session, they will remove the EEG cap and move on to their standard-of-care physical therapy session. 4. Clinic Flow and Timing of Assessments Participants in the intervention group will perform virtual rehabilitation exercises for 20 minutes 2 times per week for the first 8 weeks postoperatively. Approximately 45 minutes total is required for setup, calibration, virtual rehabilitation, and clean up. After the virtual rehab session, patients in the intervention group will move on to their standard-of-care PT session based on the surgeon's protocol specific to their operation. Patients in the control group go directly to their standard PT session. The intervention group will spend an additional 45 minutes in clinic to perform the i-BrainTech™ training session for a total study visit time of no more than 1-1.5 hours. The control group will spend 45 minutes to 1 hour of total study visit time. Participants will continue receiving their standard clinical care with their attending healthcare team throughout the study. In addition, they will attend scheduled study visits at the Motion Laboratory in the Orthopedic Building at Rush University Medical Center for motion analysis and physical testing at 2 months, 4 months, and 6 months post-surgery. At each visit, anthropometric data (age, height, and weight) will be collected first. Participants will then change into standardized clothing provided by the research team. Surface Electromyography (sEMG) sEMG data will be collected from five muscles: rectus femoris, vastus medialis oblique, vastus lateralis, semitendinosus, and biceps femoris, using a research-grade sEMG system. Electrode placement will follow the SENIAM (Surface EMG for a Non-Invasive Assessment of Muscles) protocol. Per SENIAM (www.seniam.org) guidelines, the skin will be shaved, lightly abraded with abrasion wipes, and cleaned with alcohol wipes before electrode application. Electrodes will be placed at least 2 cm apart to minimize crosstalk, and voluntary contractions will be performed to confirm correct placement. Each muscle will be assessed individually before data collection. Once all sensors are verified, simultaneous sEMG and 3D kinematic data collection will be performed using Qualisys Track Manager software or similar. Motion Capture To evaluate patient-specific movement mechanics, a markerless multi-camera motion analysis system will be used to track kinematics, while instrumented force plates will measure ground reaction forces. The markerless system allows for accurate motion tracking while significantly reducing setup time-by up to 80% since no physical markers need to be placed on the skin Participants will be evaluated while completing the following functional tasks. At 2 months postoperatively, all patients-regardless of the surgical procedure-will undergo motion analysis and strength testing. The motion analysis will include walking, bilateral squats, and a procedure-specific lunge (forward lunge for ACLR and TKA; lateral lunge for THA and hip arthroscopy). Strength assessments will focus on knee extension for ACLR and TKA, and hip extension and abduction for THA and hip arthroscopy. At 4 months postoperatively, all groups will repeat the same motion analysis tasks as at 2 months, with the addition of a single-leg vertical jump. Strength testing will now include two joint actions: knee extension and flexion for ACLR and TKA, and hip extension and abduction for THA and hip arthroscopy. At 9 months postoperatively, assessments will mirror those conducted at the 4-month follow-up for all surgical groups, including the same motion tasks and strength tests. This schedule allows for tracking recovery progression over time using both movement quality and strength performance. Strength testing Strength testing will be conducted after sEMG placement and motion analysis. Participants will keep the sEMG sensors on while performing maximum voluntary contractions (MVC) for knee flexion and extension. Isometric and isokinetic strength will be assessed using either a Biodex dynamometer or a handheld dynamometer, depending on equipment availability and participant-specific considerations. Strength data will also be used to normalize sEMG signals, with mean amplitudes of each phase expressed as a percentage of MVC. A 30% MVC normalization will be applied to allow for valid comparisons across groups. Total Testing Time The full testing session-including subject setup, EMG placement, motion analysis, and strength testing-will take approximately 30 to 45 minutes. Participants will be offered opportunities for water and seated rest breaks as needed throughout the session to ensure comfort and minimize fatigue. 5. Outcomes Primary outcomes ACLR and TKA * Outcome Measure Title o Knee extension strength * Outcome Measure Description * Maximal isokinetic knee extensor strength (Newtons/BMI). Maximal isokinetic knee extensor strength will be assessed using standardized dynamometry procedures with Biodex Isokinetic Dynamometer (Biodex System 3) at 2, 4 and 6 months after surgery. Each participant will perform three to five maximal voluntary isometric contractions of the knee extensors. The average of the peak torque values will be used for analysis. To account for individual differences in body size, values will be normalized to the participant's body mass index (Newtons/BMI). The first assessment will occur at 2 months to ensure patient safety and measurement consistency, as early postoperative conditions (e.g., pain, swelling) could compromise the reliability and validity of strength testing. * Outcome Measure Time Frame o 2, 4 and 6 months * Statistical Analysis * Linear mixed-effects models : * Linear mixed-effects models will be used to analyze changes in knee extensor strength between the intervention and control group across the 2-, 4-, and 6-month follow-up assessments. This approach accounts for repeated measures within participants and allows for the evaluation of group differences over time. Group (intervention vs. control), time (2, 4 and 6 months), and their interaction will be entered as fixed effects, with subject-level random intercepts. Significant main effects or interactions will be further examined using pairwise comparisons of estimated marginal means, with Tukey's adjustment for multiple comparisons to determine pairwise differences. All results will be reported with estimated means, p-values, and 95% confidence intervals. . Hip arthroscopy and THA * Outcome Measure Title o Hip Abduction Strength * Outcome Measure Description o Maximal isokinetic hip abductor strength (Newtons/BMI). Maximal isokinetic hip abductor strength will be assessed using standardized dynamometry procedures with Biodex Isokinetic Dynamometer (Biodex System 3) at 2, 4 and 6 months after surgery. Each participant will perform three to five maximal voluntary isometric contractions of the hip abductors. The average of the peak torque values will be used for analysis. To account for individual differences in body size, values will be normalized to the participant's body mass index (Newtons/BMI). The first assessment will occur at 2 months to ensure patient safety and measurement consistency, as early postoperative conditions (e.g., pain, swelling) could compromise the reliability and validity of strength testing.. * Outcome Measure Time Frame o 2, 4 and 6 months * Statistical Analysis * Linear mixed-effects models : * Linear mixed-effects models will be used to analyze changes in hip abductor strength between the intervention and control group across the 2-, 4-, and 6-month follow-up assessments. This approach accounts for repeated measures within participants and allows for the evaluation of group differences over time. Group (intervention vs. control), time (2, 4 and 6 months), and their interaction will be entered as fixed effects, with subject-level random intercepts. Significant main effects or interactions will be further examined using pairwise comparisons of estimated marginal means, with Tukey's adjustment for multiple comparisons to determine pairwise differences. All results will be reported with estimated means, p-values, and 95% confidence intervals. Secondary outcomes are outlined in the supplementary material attached to this protocol and include EMG activation during functional tasks and strength tests; flexibility tests, PROs, and AMI classification 6. Statistical Methods Statistical methods are summarized in the supplementary material. For the primary outcome, linear mixed-effects models will be used to assess changes in knee and hip strength following ACLR, TKA, THA, and HA procedures. Strength outcomes will be compared between the intervention and control groups at 2, 4, and 6 months postoperatively. This modeling approach accounts for repeated measures within participants and allows for the evaluation of group differences over time. Group (intervention vs. control), time (2, 4 and 6 months), and their interaction will be entered as fixed effects, with subject-level random intercepts. Significant main effects or interactions will be further examined using pairwise comparisons of estimated marginal means, with Tukey's adjustment for multiple comparisons to determine pairwise differences. All results will be reported with estimated means, p-values, and 95% confidence intervals Secondary outcomes will follow a similar analytic approach using linear mixed-effects models, accounting for repeated measures and assessing group-by-time interactions, with appropriate post hoc testing as needed. To assess the time-series data of kinematics and kinetics across different conditions and time points, Statistical Parametric Mapping (SPM) will be utilized. SPM is a robust analytical approach that allows for the statistical evaluation of entire waveforms, reducing the limitations of discrete-point analysis in biomechanical research. SPM maintains the temporal structure of the data and enables the identification of significant differences across the entire movement cycle, offering a more comprehensive understanding of biomechanical adaptations post-surgery. SPM will be applied to joint angle waveforms, ground reaction forces, and external moment profiles to compare surgical groups, timepoints, and control conditions. This approach will help detect subtle but functionally relevant alterations in movement patterns that may not be captured using traditional peak or mean value analyses. Kinematic and kinetic data will be collected bilaterally to examine compensatory strategies in the contralateral limb. EMG and strength only in the affect side. The integration of markerless motion capture, inverse dynamics, and SPM analysis will provide a detailed and objective assessment of post-surgical movement patterns. Differences in the distribution of knee AMI classification (grades 0-3) between groups will be assessed using Fisher's exact test. Statistical significance for all analyses will be set at an a priori α of 0.05. All data analyses will be completed using R version 4.2.3 (R Core Team). Loss to Follow Up Data analysis will be conducted according to the intention-to-treat (ITT) principle. Patients who are lost to follow-up for any reason will be included in the primary analysis using the last observation carried forward (LOCF) method. 7. Power Analysis Strength Separate power analyses were conducted for the knee surgery cohort (ACL and TKA) and the hip surgery cohort (HA and THA). Knee Surgery (ACL and TKA) The power analysis for the knee surgery group was based on data from a previous randomized controlled trial, which evaluated the effects of motor imagery on quadriceps strength following total knee arthroplasty (TKA). The original study included 12 patients with unilateral TKA, assessed six months postoperatively (10 females, 2 males). The primary outcome was quadriceps maximum voluntary isometric contraction (MVIC), measured using a hand-held dynamometer. This method is highly correlated with the Biodex system currently used in the present trial (R = 0.91). Statistical analysis was performed in R version 4.3.2 using the pwr package. The mean quadriceps MVIC reported was 20.58 N/BMI (SD = 1.85). Assuming a two-group parallel design (intervention vs. control) with 30 participants per group (total n = 60), the study is powered at 80% to detect a between-group difference of 1.31 N/BMI at a two-tailed α = 0.05 (adjusted α = 0.025 to account for multiple comparisons). This corresponds to a minimum detectable difference (MDD) of 6.31% between groups. Hip Surgery (HA and THA) The power analysis for the hip surgery group was based on data from a retrospective study, which evaluated hip abductor strength before and three months after hip arthroscopy for femoroacetabular impingement (FAI). The study included 29 individuals (mean age 27.4 ± 7.5 years; 76% female). The primary outcome was hip abductor MVIC, also measured with a hand-held dynamometer. The reported mean abductor MVIC was 1.97 N/kg (SD = 0.42). A two-group parallel design with 30 participants per group (n = 60 total) provides 80% power to detect a between-group difference of 0.31 N/kg at a two-tailed α = 0.05 (adjusted α = 0.025). This represents an MDD of 15.6%. To contextualize the clinical relevance of the estimated minimal detectable differences (MDDs), previously published values for minimal clinically important differences (MCIDs) were reviewed. For quadriceps MVIC, a previous study reported an MCID of 26.9% in older adults with COPD, with improvements associated with enhanced performance in the six-minute walk test. In patients undergoing ACL reconstruction, a limb strength asymmetry of 10% is considered a clinically relevant threshold and a predictor of reinjury. For hip abductor strength, although there is no universally accepted MCID after hip surgery, many studies consider a relative difference of 10-15% to be clinically meaningful. These values serve as important clinical benchmarks and support the interpretation that the proposed sample size (n = 60 per pathology group) is adequate to detect both statistically and clinically relevant between-group differences in strength outcomes. Kinematics A separate power analysis was conducted for the secondary kinematic outcomes to estimate the minimum detectable differences achievable with the current sample size and 80% statistical power. This analysis helps ensure that the study is adequately powered to detect changes that are not only statistically significant but also clinically meaningful. The power analysis was based on data from a previous investigation conducted in the same laboratory where the current study will take place, using identical equipment, camera setup, and software configuration parameters. Individuals assessed in study were 5 healthy males (mean age 26) and 5 females (mean age 28) (n=10). The primary variables of interest for the power analysis were the knee peak flexion angle (for the ACL and TKA study arms) and the hip peak extension angle (for the hip arthroscopy and THA study arms) during the gait cycle, measured using a markerless motion capture system (Theia 3D, Theia Markerless Inc., Kingston, ON). The statistical analysis was conducted using R version 4.3.2 (R Core Team, 2023) and the pwr package (Champely, 2020). For the knee flexion angle, a previous investigation reported a mean of 19° with a standard deviation (SD) of 7.68°. Using these values, a two-group parallel design with 30 participants per group (total n = 60) was powered at 80% to detect a between-group difference of at least 5.65° at a two-sided α = 0.05 (adjusted α = 0.025 for sidedness). Similarly, for hip extension angle (mean = 9.25°, SD = 5.11°), the same sample size and power would yield a minimum detectable difference of 3.76°. To contextualize the clinical relevance of the estimated minimal detectable differences (MDDs), previously published values for minimal clinically important differences (MCIDs) were reviewed. For knee flexion angle, a previous study reported an MCID of 6.81 degrees based on gait analysis of the unaffected limb in stroke patients. In a different clinical context, another study identified an MCID of 5 degrees for passive range of motion following total knee arthroplasty, with improvements in flexion associated with better knee function and higher patient satisfaction. Regarding hip extension angle, it was established an MCID of 2.86 degrees in stroke patients using motion analysis of the unaffected limb. These values serve as clinical benchmarks to interpret the statistical power and meaningfulness of the differences targeted in this study. 8. Randomization Sequence Generation Randomized list generation before trial commencement utilizing the National Institute of Health (NIH) Clinical Trial Randomization Tool. A 1:1 allocation (intervention vs. Control) per procedure will be implemented. The randomization sequence will be generated before trial commencement by an independent data manager using the Clinical Trial Randomization Tool developed by the National Cancer Institute (NCI). The tool will use the Asymptotic Maximal procedure, a restricted randomization method that limits the imbalance between trial arms to a pre-specified Maximum Tolerated Imbalance (MTI) of 3. The randomization list will be created for a total of 70 participants, with no stratification applied. Type of Randomization A simple two-arm parallel group design will be used with a 1:1 allocation ratio for each procedure (ACLR, THA, TKA, and HA), with a separate randomization list of 35 participants per group (total n=70) generated independently for each procedure. Each procedure will be treated as a distinct trial with its own allocation sequence. The final participants will be randomly assigned in accordance with the MTI threshold, allowing minor tolerable imbalances between arms. No stratification or blocking will be implemented. The increased sample size per group (n=35 instead of n=30) accounts for potential losses or allocation issues during the study. Additional allocation slots beyond the target sample size will be generated to account for unexpected exclusions, dropouts before randomization, or technical errors during allocation. Allocation Concealment Mechanism Allocation concealment will be ensured using the REDCap (Research Electronic Data Capture) Randomization Module, which provides a secure, centralized, and automated platform for treatment allocation. The randomization sequence will be generated and implemented within REDCap by an independent data manager who will not be involved in participant enrollment, intervention, and outcome assessment. The system will be configured to assign participants in real time only after eligibility has been confirmed and baseline data have been entered, thereby preventing any foreknowledge of the upcoming assignment. Investigators, study staff, and participants will remain unaware of the allocation sequence prior to assignment. REDCap's access controls and audit logs will safeguard against manipulation and preserve allocation concealment throughout the trial. Implementation All patients referred for ACLR, TKA, THA, or HA procedures in the physician's office will be referred to the research team to assess eligibility criteria and initiate the enrollment process, either in person or by phone. Eligible patients will be enrolled in the study during the pre-operative period and will complete baseline assessments. Following surgery, participants will be assigned to one of the intervention groups one day prior to the intervention date. Group allocation will be performed by an independent data manager using the REDCap Randomization Module and communicated only to the team responsible for delivering the NFVT. The individual responsible for generating the allocation sequence will be distinct from those involved in enrolling participants and assigning interventions. Blinding Investigators and physicians will be blinded to group allocation to minimize bias in clinical decision-making and post-operative care. Patients will not be blinded due to the nature of the neurofeedback intervention, which cannot be masked. Both study groups will receive standard rehabilitation, with the only difference being the addition of NFVT in the intervention arm. The neurofeedback sessions will take place within the physical therapy clinic, immediately prior to scheduled physical therapy appointments. The team delivering the neurofeedback intervention will be the only personnel aware of group allocation. Treating physicians, physical therapists, outcome assessors, and the statistical analysis team will remain blinded to participant allocation throughout the trial. Statistical analyses will be performed by the Rush Statistical Analysis team, an independent third-party ancillary resource, using a coded dataset to ensure blinding is maintained during data analysis. Unblinding Procedures Given the non-invasive nature of the EEG-based neurofeedback intervention, adverse effects are not expected. However, in rare cases where unblinding is required-such as equipment malfunction, unexpected clinical events, or participant withdrawal-a formal request must be submitted to the principal investigator (PI) or Data and Safety Monitoring Board (DSMB). All unblinding events will be logged with justification and date. For emergencies, a dedicated study coordinator will have secure access to the REDCap randomization module and group allocation. This coordinator will be available to authorized clinical staff in urgent scenarios requiring immediate unblinding. Outcome assessors and data analysts will remain blinded throughout the study. These procedures align with CONSORT guidelines and will be included in staff training to ensure adherence. 9. Recruitment The investigators will identify eligible patients in the clinic, and the research study staff will discuss the study with eligible patients. The informed consent process may occur over a period of several discussions, culminating in the signing of a consent form in the office (iPad) or sent to their verified email address. Informed consent will be obtained via the eConsent process through the secured platform Patient IQ. The patient will be prompted to reply with the appropriate passcode to access the consent form and then provide the passcode again with their signature (secured). A copy of the time stamped document will be sent to the study team through the electronic platform and a copy will be sent to the participant. All participants will be consented prior to surgery and the performance of research related testing activities.