A novel immersive virtual reality environment for the motor rehabilitation of stroke patients: A feasibility study

Abstract

We designed and implemented an immersive virtual reality (VR) environment for upper limb rehabilitation, which possesses several notable features. First, by exploiting modern computer graphics its can present a variety of scenarios that make the rehabilitation routines challenging yet enjoyable for patients, thus enhancing their adherence to the therapy. Second, immersion in a virtual 3D space allows the patients to execute tasks that are closely related to everyday gestures, thus enhancing the transfer of the acquired motor skills to real-life routines. Third, in addition to the VR environment, we also developed a client app running on a PC that allows to monitor in real-time and remotely the patients' routines thus paving the way for telerehabilitation scenarios. Here, we report the results of a feasibility study in a cohort of 16 stroke patients. All our patients showed a high degree of comfort in our immersive VR system and they reported very high scores of ownership and agency in embodiment and satisfaction questionnaires. Furthermore, and notably, we found that behavioral performances in our VR tasks correlated with the patients' clinical scores (Fugl-Meyer scale) and they could thus be used to assess improvements during the rehabilitation program. While further studies are needed, our results clearly support the feasibility and effectiveness of VR-based motor rehabilitation processes.

Keywords: fugl-meyer; head-mount display; immersive virtual reality; motor rehabilitation; stroke.

FIGURE 1

Application scenario of our immersive VR environment. (A) The patients are immersed in a VR environment by means of a head-mounted display (HMD, Oculus Quest 2, Facebook Reality Labs). In this environment, they can see different objects with which they can interact. The inset shows the scene as experienced by the patient on the HMD display. (B) The program running on the HMD wirelessly communicates with a client app running on a PC that allows to monitor remotely and in real-time the patients’ behavior, set their rehabilitation routines and vocally interact with them.

FIGURE 2

Patients’ feedback on their experience in our VR-based rehabilitation system. (A) Patients’ ratings, in a scale from 1 to 5, to the question: “Did you enjoy this type of training?” (In Italian: “Ha gradito la tipologia di allenamento?”). Supplementary Figure S3 in the supplementary information shows the responses for all other questions in the satisfaction questionnaire. (B) Patients’ scores for ownership and agency as assessed by a standardized questionnaire (see Methods for further details). The 2 bars represent average across patients and the vertical lines signify variance (mean ownership = 7.4 ± 2.0; mean agency = 8.3 ± 2.0).

FIGURE 3

Distribution of completion times for three tasks and two patients. The violin plots show the distributions of the times taken to complete three of the tasks presently implemented in our system (three columns) for two patients. Patient 12 was a 78-year old male with a right-side impairment, and patient 13 was a 63-year old female with a left-side impairment. Distributions are color coded differently for the healthy and impaired limb (green and red respectively). The label on the vertical axis shows the patients’ id and their Fugl-Meyer score. See Supplementary Figure S3 in the supplementary information for similar plots for all other patients.

FIGURE 4

Correlation between behavioral results in our VR tasks and Fugl-Meyer clinical scores. The four scatterplots show the difference in completion times between the impaired and healthy limb for each patient and condition plotted against the Fugl-Meyer clinical assessment. Each panel shows results for one task and each dot represents data for one patients. The p-value of the correlation (Spearman’s rank-order correlation) between completion times and Fugl-Meyer scores is shown in the panels’ title.

FIGURE 5

Example of hand trajectories recorded during task execution. The two panels show the hand trajectories of a 77-year old male patient during the execution of the “ball in hole” (left panel) and “glasses” (right panel) tasks. This patient exhibited a left side impairment. The trajectories of the healthy and impaired hands are shown in green and red respectively.