A VIRTUAL REALITY-BASED ENVIRONMENT FOR ERROR-AUGMENTED TREATMENT IN PATIENTS WITH STROKE

Stroke is a leading cause of long-term sensori-motor deficits in upper limb function. Yet, current upper limb interventions have limited effectiveness. We developed a method and system for an innovative treatment of upper limb motor deficits following stroke, based on joint level error augmentation. The method, based on referent control theory, aims at dynamic remapping of muscle-level control mechanisms. We hypothesize that addition of error at the joint angle level will lead to an increase in the joint’s active control zone and thus to an increase in patient’s ability to perform isolated voluntary movement. The system has three components, a passive armrest supporting the arm against gravity, a Kinect motion tracking system, and a virtual-reality (VR) training environment. In the current presentation, we focus on the VR environment.

The proposed method requires accurate visualization of the entire arm that can include an error in the presented joint angle. The visualization should invoke a high degree of presence, so that the patient accepts the visualized arm position as representing his/her actual arm location, despite conflicting input from his/her proprioception. The game should encourage a functional gross arm motion task suitable for patients who have not yet re-gained fine motor skills. The elbow training range should be adapted to the zone in which each patient can control his or her arm motion. Motion is limited to a horizontal plane facilitated by the supporting manipulator. The game should be motivating but the visual scene should not be overwhelming, since the patient’s perception capabilities may be reduced.

According to these specifications, a stationary reaching task was designed, in which targets are presented within a horizontal plane. The virtual camera is situated behind the subject at a tilt angle of 45°, to emulate first-person view. A full arm model is presented on the screen, based on the tracked motion with (or without) an inserted error of 10 degrees in elbow angle. The reaching task is time-limited, and the game environment facilitates adaptation of the time for reaching a target. A motivational icon is presented above the patient’s hand upon each successful reach, and every 30 reaches an encouraging message is presented on the screen. Prior to the training, a calibration procedure is performed to ensure that the training zone is suitable for the patient’s capabilities. The patient is presented with an area-covering task, and 30 targets are later uniformly distributed throughout this space. The system records the average success rate and time to reach the targets.

A pilot clinical experiment is currently underway. One system is installed in the rehabilitation department at the Soroka University Hospital, Beer-Sheva and an additional system will be installed in Canada. 24 sub-acute patients with stroke (six weeks to three months post stroke) will undergo three thirty-minute training sessions on consecutive days. Half of the patients will receive visual feedback with error and half will receive visual feedback without error. The experiment in Israel has been approved by the Helsinki committee at the Soroka University Hospital.