Virtual Reality Therapy for Arm Recovery (VVITA) Stroke

BACKGROUND. Cerebral stroke is one of the main causes of disability in adults and involves, in the most common case, a hemiparesis of the contralateral upper limb. This motor deficit, which occurs for 80% of patients in the acute phase and 40% in the chronic phase, can be improved through the involvement of patients in appropriate rehabilitation with increased intensity and task-specificity. Technological advances in recent years have provided new methodologies to support and promote the rehabilitation process by increasing its repeatability and intensity. In particular, rehabilitation systems integrated into virtual reality environments (VR, Virtual Reality) can provide more complex work scenarios than rehabilitation systems and, thanks to the immersiveness of the systems with a visor integrated into a helmet wearable, it is possible to simulate activities of daily life, thus effectively increasing the active participation and motivation of the patient in carrying out and completing the rehabilitation path. Further recent methodological advances have allowed the development of approaches to personalized rehabilitation programs, aimed at restoring the specific components of the control motor deficits. In particular, it has been noted that deficits in motor function that accompany patients in stroke outcomes are characterized by specific patterns of muscle activation or pathological muscle synergies. In this regard, recently quantitative approaches have been developed for the identification of motor coordination in healthy subjects that relies on the decomposition of patterns of muscle activation as a combination of physiological muscle synergies through factorization algorithms. These approaches of decomposition have highlighted the changes induced in the coordination of muscle following stroke. Hence, the characterization of the alterations of activation patterns post-stroke muscle by decomposition in combination of muscle synergies can provide an indicator of the pathological state, with prognostic value on recovery induced by a therapeutic intervention and more accurate than current clinical scales and quantitative measurements based on purely kinematic measurements. In addition to the applications of muscle synergies to rehabilitation. In recent years, the use of myoelectric interfaces for the control of actuators such as exoskeletons or prostheses has become increasingly widespread. Myoelectric interfaces can decode the patient's intention through the residual myoelectric activity of the paretic limb allowing the execution of voluntary movements through their normal cortico-spinal pathway, thus providing them with feedback (e.g. visual) and effectively establishing a closed-loop system that promotes learning and fostering active participation, increasing coordination muscle strength and reducing spasticity. Myoelectric interfaces for rehabilitation also aim to activate neuroplasticity mechanisms that reshape neuromuscular activity and lead to motor learning and in some cases, to the restoration of motor function.

AIM AND DESIGN OF THE STUDY. The aim of the present pilot study is therefore to evaluate the effectiveness of a protocol for the upper limb post-stroke rehabilitation through a Virtual Reality system embedded with myoelectric control. All the devices used in this study such as for the visualization of environments virtual (HTC-Vive) and for recording electromyographic signals (Myo Armband) are available on the consumer electronics market and have been previously used in other clinical study involving humans. Study Design is a pilot Randomized Controlled Trial.

PARTICIPANTS. We enrolled patients who had experienced stroke more than a year before the time of the first evaluation were enrolled for this study.

Criteria for enrolment were: 1) upper limb deficits with sufficient level of muscle power, so that movement is possible with gravity eliminated (F=2), or muscle movement is possible against gravity (F=3), as assessed with Medical Research Council (MRC) scale grades not below 1; 2) at least 18 but not more than 54 out of 66, respectively 27% and 80% of the upper limb functionality, as assessed with the Upper Limb Extremity section of the Fugl-Meyer Assessment (FMA) scale; 3) absence of severe linguistic impairments which may limit the understanding of the instructions; 4) absence of cognitive impairments, assessed with the Addenbrooke's Cognitive Examination - Revised version for Italian population, visual deficit or other neurologic disease in comorbidity, which may affect the patient's ability to interact with the VR environment.

Since age-related and severity-related differences were found in motor outcomes following VR upper limb treatments, we created a blocked randomization list with two levels of stratification (age and severity), which guaranteed a balanced within-group split throughout the data collection process. Therefore, participants aged between 18 and 55 years old were considered as young participants according to previous studies, while those aged between 56 and 80 years were considered as older. The level of impairment was computed based on two cut-offs obtained through the administration of the FMA-UE scale. We considered patients with severe impairment, whose scores fell between 18 and 36, namely, those with 27-54% motor proficiency. On the other hand, scores ranging from 37 to 54 (56-80%) were considered moderate impairment. We used the sealed envelope tool to create the randomization list of participants as a function of the stratification number of randomization blocks (age, level of impairment), which revealed the need to enrol at least 12 patients. This sample size was also used in previous work about the effectiveness of the VR system on the rehabilitation of the upper limb in stroke patients.

INNOVATION AND DEVELOPMENTS. The development of such an approach is the result of the collaboration between the Santa Lucia Foundation and the Institute of Robotics and Mechatronics of the Deutsches Zentrum für Luft- und Raumfahrt e.V. (DLR) in Munich, Germany, which is the leader partner of a project (VVITA) funded by the Helmholtz Association e.V. (Germany). The system will allow the practice of exercises simulating activities of daily living in immersive virtual reality environments, providing real-time visual feedback on the movement of the virtual limb.

Briefly, the system consists of an HTC Vive VR System including a head-mounted display (HMD), two motion trackers, two controllers, two base stations, which emit infrared light and synchronization signals to enable accurate spatial tracking of the head-mounted display (HMD), controllers, and trackers, and two Myo armbands (Thalmic Labs). Each tracker of HTC and each Myo armband were positioned respectively on the hand dorsum and on the forearm of each participant. The tracker system provides the position and orientation of the patient's hand. The rehabilitation software has been built with the Unity game engine, a platform commonly employed for the creation of interactive 3D applications and virtual reality environments. It consists of three different virtual environments in which a table in from of the patients matches the position and dimensions of a rehabilitation table. For each patient, an anonymous form is created and filled with the participant ID, date of birth, impaired side, and dominant hand. Before starting the training session, the program requires to register the positions of the table corners with one of the two controllers in use for the VR session for calibration, so that the virtual model of the table is spatially aligned with the corresponding physical table where the patient is performing the exercise. This was necessary to provide correct visuo-haptic feedback during the VR exposure. The system was developed in the theoretical framework of Mirror Therapy and according to the personalized medicine approaches. The impaired virtual limb is designed to move in the virtual environment according to the actual movement and electromyographic activity recorded with the trackers positioned on the hands and the wearable bracelets respectively. This implies that all exercises are trustily based on the ability of patients, taking into account the possible range of movement and ability to reproduce gestures as assessed before the beginning of the interventional session.

PROCEDURE. VR neurorehabilitation therapy will be administered in addition to normal therapy, and the effectiveness of such therapy will be evaluated through comparison with a control group that will carry out regular therapy plus a neuromotor therapy dedicated to the upper limb of equal time and dose of investigational therapy. Motor proficiency will be monitored before and after the entire rehabilitation process through clinical evaluations and kinematic and electromyographic systems. After the clinical evaluation, each patient was enrolled for the experimental training with VR mirror therapy. Before starting, a researcher explained the task to facilitate the patient's interaction with the VR system. For all the patients, this was the first experience with a VR system in which myoelectric control must be exploited to achieve a task. The patient performed VR treatment while sitting on a chair with the possibility of resting the hands on a table. At the start of each session, the researcher carefully positioned the two Myo armbands over the brachioradialis muscles on each arm and attached the trackers to the back of the patient's hands, aligning the z-axis between the index and middle fingers.

For each session, to personalize the training exercises by calibrating the reachable workspace the patient was asked to perform : 1) to assume a resting pose, with a comfortable position of the hand on the table; 2) to extend the unimpaired arm to the maximum range in the antero-posterior and medio-lateral directions; 3) to extend the unimpaired arm in the same directions with the impaired limb, as much as they could. For each arm, the software evaluates the reachable area. A secondary assessment was dedicated to calibrating the visualization of the opening and closing of a virtual hand (gesture) according to the EMG signals recorded by the Myo armband. The patient is asked to: (1) assume a resting pose with a comfortable position of the hand on the table and minimal muscle activity; (2) reproduce the gesture of holding a ball with the fingers extended; (3) reproduce the gesture of holding the handles of an accordion with finger flexed.

TRAINING PROTOCOL. The training consisted of 24 sessions, each lasting approximately 20 minutes. In the initial session, the researcher guided the patient to explore the virtual environment to help them become familiar with the VR setup and minimize distractions. Following this, the patient received verbal instructions for the first tasks. The training involved reaching for two different virtual objects, a ball and an accordion, which were shown one after the other to encourage various hand movements. To interact with the accordion, the patient needed to make a fist, while reaching for the ball required an open hand. Both objects had handles on opposite sides to assist with proper hand placement, and a movement was deemed successful if both virtual hands were within 5 cm of the handles. The VR system provided immediate feedback on hand positioning and gesture accuracy, when the hands were correctly placed, the handles changed from orange to green. An icon on the screen also indicated the required hand position, disappearing once the correct gesture was completed. After meeting the positioning and gesture criteria, the target object gradually turned green, signalling the time needed to hold it before finishing the task. Auditory cues were also included, with a positive sound for successful task completion and a negative sound if the trial exceeded 20 seconds. The targets varied in height and distance from the starting hand position, which was recorded at the beginning of each session. The maximum distance was customized based on the patient's arm reach, while the minimum matched their resting position, resulting in 12 possible target locations. The order of the targets was randomized to reduce fatigue from repetitive movements. The researcher instructed the patient to place their hands in two spheres next to the target object, keeping their hands open for the ball and clenched for the accordion.

The system uses advanced algorithms to support rehabilitation for both hands, utilizing VR to provide mirrored assistance for arm movements and hand gestures. The arm movements are monitored using motion trackers, while the hand gestures are detected through EMG signals from a Myo armband. There are two adjustable settings, alpha and beta, which therapists can modify to customize the level of assistance based on the patient's abilities. There are ten preset levels of assistance, ranging from complete help to total independence. The alpha setting determines how much the less affected limb influences the movement of the more affected limb. If alpha is set to 1, the virtual limb moves exactly like the real affected limb, providing no assistance. If set to 0, it mimics the less affected limb, similar to traditional mirror therapy. For values in between, the system combines input from both limbs to provide a gradual transition between assistance and independence. The beta setting allows therapists to adjust the difficulty of the tasks. In terms of arm movements, a beta value of 0 places targets within reach of the more affected limb, while a value of 1 sets them at the maximum reach of the less affected limb. For gestures, beta affects how much muscle effort is needed for recognition, with lower values requiring less effort. The relationship between these settings and task difficulty is designed so that higher alpha and beta values mean less assistance and greater challenges, encouraging active participation. Lower values increase assistance, which is crucial for patients with more significant motor difficulties.

The first training session focuses on determining the appropriate settings for each patient's ability level, establishing a baseline that also serves as a familiarization phase. This involves adjusting the alpha and beta settings to find the right balance between independence and dependence. The goal is to set a challenging yet manageable level of difficulty, ensuring that the exercises are sustainable throughout the training session without causing frustration. After each block of 12 reaching tasks, the therapist adjusts the assistance settings based on the patient's performance score. This score, which reflects the success rate during the block, helps to fine-tune the difficulty for the next block. If the patient achieves a success rate above 90%, indicating at least 11 successful attempts, the difficulty level is increased to enhance engagement. If the success rate is between 70% and 90%, the difficulty remains the same to keep the patient at an optimal challenge level. Conversely, if the success rate drops below 70%, the difficulty is reduced to ensure the tasks remain achievable and to prevent frustration. This adaptive approach aims to maintain motivation and maximize motor improvement by continuously adjusting the difficulty of the tasks to match the patient's performance, striving to keep the success rate within the desired range of 70% to 90%. The specific adjustments to be made are determined by the therapist based on their expertise and real-time observations of the patient's performance, fatigue, and overall condition.

USABIITY, ACCEPTABILITY AND FEASIBILITY EVALUATION.

For the evaluation of the usability and feasibility of the VITA program, the following questionnaires were administered:

• The User Satisfaction Evaluation Questionnaire (USEQ), for evaluation of the evaluation of the VR system usability.
• Visual Analogical Scale (VAS) with range from 0 to 10 with respect to subjective motivation and satisfaction related to exercise.
• Italian version of Pittsburgh Participation to rehabilitation Scale (PPRS) compiled by the researcher/therapist to report the patient's participation levels in the exercise on a Likert scale ranging from 1 to 6.
• The NASA Task Load Index for the multidimensional subjective assessment that rates perceived workload to assess a task, system, or team's effectiveness or other aspects of performance.