Application of Virtual Reality in Neuro-Rehabilitation: An Overview

We don't need an overview, we need EXACT DEFINED STROKE PROTOCOLS. When the hell will you get there?

Maybe after you are the 1 in 4 per WHO that has a stroke? Will that finally get you to do your job properly?

Application of Virtual Reality in Neuro-Rehabilitation: An Overview

 Lucia F Lucca
1
, Antonio Candelieri
1,2
 and Loris Pignolo
1

1
S. Anna Institute and RAN - Research in Advanced Neuro-rehabilitation, Crotone,
2
Laboratory of Decision Engineering for Health Care Delivery, University of Cosenza, Italy
1. Introduction
Virtual reality (VR) collectively refers to the realistic, albeit artificial environments that are simulated by computer and are experienced by end-users via human-machine interfaces involving multiple sensory channels. In this respect, comparable technical solutions are applicable across different domains such as cyberspace, virtual environments, teleoperation, telerobotics, augmented reality,
 and synthetic environments.
This makes application possible in a variety of conditions such as (1) design, engineering, manufacturing, and marketing; (2) medicine and healthcare; (3) online monitoring of children and the elderly at home and accident prevention; (4) hazardous operations in extreme or hostile surroundings; and (5) training in military and industrial machine operation, medical teaching and surgery planning/training. An implement of VR with live direct or indirect view of a physical real-world environment whose elements are purportedly enhanced (augmented) by virtual computer-generated imagery to meet the viewer needs, Augmented Reality is extensively used in open surgery, virtual endoscopy, radiosurgery, neuropsychological assessment and medical rehabilitation. Application in psychotherapy ranked 3rd among 38 psychotherapy interventions predicted to increase in use in the next future (Gorini & Riva, 2008a; Gorini & Riva, 2008b). Application in rehabilitation is increasing and expanding; innovative technical solutions in motor and sensory-cognitive rehabilitation result in substantial developments from the available procedures and in prototypes for clinical testing. The clinical results appear promising.
2. Rationale for VR-mediated neuro-rehabilitation
The rationale for application mainly rests on the available evidence that a functional re-arrangement of the injured motor cortex can be induced with the mediation of the mirror neurons system (Eng et al, 2007; Holden, 2005; Rose et al, 2005) or through the subject’s motor imagery and learning (Gaggioli et al, 2006). Intensive training (repetition) facilitating re-arrangement of cortical function and motivation reinforced by feedback information about the ongoing improvement are necessary for motor learning to be possible after brain damage. These conditions are easily made available in VR-mediated neuro-rehabilitation paradigms. Motor impairment and recovery can be measured in real time (e.g. at the end of each trial or a series of trials) to give the user the knowledge-of-performance (about his/her movement patterns) and knowledge-of-results (about the outcome predictable at each time point during rehabilitation) that reinforce motivation and the training procedure itself. VR allows online or offline feedback, that has been extensively investigated with a general agreement that it improves learning (Bilodeau & Bilodeau, 1962; Gentile, 1972; Khan & Franks, 2000; Newell & Carlton, 1987; Winstein, 1991; Young & Schmidt, 1992; Woldag & Hummelsheim, 2002). The expectation is, that VR-mediated rehabilitation should improve the approach efficacy and the outcome by making tasks easier, less demanding and less tedious/distractive, and more informative for the subject. Interactive VR environments are flexible and customizable for different therapeutic purposes; individual treatments can be personalized in order to facilitate movement retraining, to force the user to focus on the task key elements, and to facilitate transfer of motor patters learned in VR environments to the real world.
3. Studies
3.1 VR in the upper limb motor rehabilitation
VR was first applied in the rehabilitation of the paretic upper-limb after stroke in a setting designed to promote motor (re)learning for different movements (hand, elbow and shoulder) and functional tasks or goals (Holden et al, 1999). The approach implemented a learning-by-imitation paradigm through three components: a motion tracking device to record the trajectories to be performed in the VR environment, a desktop computer display and a VR editing software specifically developed to create suitable 3D-simulated tasks at varying level of complexity. Once the scenario had been defined, the programmed motor learning tasks to be performed within the virtual environment were stored into the motion tracking device. Patients were then requested to reproduce the trajectories set by the
virtual teacher or to devise appropriate trajectories in the absence of it, while the upper limb movements were monitored by the virtual teacher, displayed in real time and recorded. The approach also assessed the degree of matching between the virtual teacher  and the patient’s trajectories and provided trainer and trainee with a measure of each trial efficency. In a pilot study (Holden et al, 1999), two chronic patients with massive stroke were trained on a reach-and-grasp task involving shoulder flexion, elbow extension and forearm supination at six increasing levels of complexity. Efficacy was assessed through a 3D kinematic reach test performed in the real world before and after VR-supported rehabilitation; the Fugl-Meyer Test of Motor Recovery for Stroke test (Fugl-Meyer et al, 1975) and the motor task section of the Structured Assessment of Independent Living Skills (SAILS) test of UE function (Mahurin et al, 1991) were used for clinical evaluation. The patients were able to export the abilities learned in VR to the real world and to similar but untrained activities, but hand orientation proved difficult to learn. In successive studies (Holden et al, 2001; Holden et al, 2002), information about the specific (as measured in real tasks designed to evaluate generalization in space, gravitoinertial force, combined spatial/gravitoinertial force, and in tasks requiring novel recombination of trained movement elements, and control tasks with untrained elements) and non-specific (as measured by variations in motor recovery tests after VR supported training activity) motor generalization was used to measure in detail the ability to transfer to the real world what learned in VR. Patients improved in three standard clinical tests of function, even if practicing in two movements only during VR training. It was suggested that VR mediated rehabilitation is an effective and efficient approach to (re)train a set of basic tasks with upgrading to a wide variety of skilled movements (Holden, 2005). Piron et al. (2005) replicated these results in a study on 50 patients with impaired upper limb motion after stroke. The VR supported rehabilitation system included a virtual environment (a PC workstation with a wall screen), a motion tracking device and the dedicated software for editing 3D-scenarios in a learning-by-imitation rehabilitation process with a
virtual teacher. Therapists set the virtual scenarios characteristics and the motor training complexity to match each patient’s motor impairment and rehabilitation protocol, set the starting position, target location and orientation, designed simple/complex tasks, added or removed non-pertinent virtual objects (distractive elements) to increase/reduce the task level of difficulty, recorded trajectory of the desired movement to be (re)learned by the subject, with the virtual teacher visible or hidden as advisable. The degree of motor impairment or recovery and the attained levels of autonomy in daily living activities were measured by the Fugel-Meyer (FM) UE score and the Functional Independence Measure scale (FIM) before and after the therapy, and by means of kinematic measures such as the movement morphology and mean duration and speed. Improvement was observed in the FM UE and FIM mean scores (with 15% and 6% increases, respectively) and in the movement mean duration (18%) and speed (23%), with better regularity of trajectories. Improvements do not appear to have been influenced by age, time since stroke or site of brain damage, as already noted in previous studies (Jeffery & Good, 1995; Johnston et al, 1992; Bagg et al, 2002; Tangeman et al, 1990; Dam et al, 1993). Instead, the severity of impairment was crucial for the outcome and a severe initial impairment was more difficult to rehabilitate. Comparison between the degrees of recovery attained after standard or VR-reinforced learning in two randomly assigned post-stroke patient groups (Turolla et al, 2007) showed significantly increased FM UE scores in all 30 patients, but improvement was greater after VR-supported therapy. The different outcome was ascribed, at least in part, to the feedback information about knowledge-of-results and knowledge-of-performance provided by the system (Todorov et al, 1997; Schmidt & Young, 1991; Winstein et al, 1996) and to the reinforced learning provided by the VR-based rehabilitation approach (Barto, 1994; Doya, 2000; Fagg & Arbib, 1992; Rummelhart et al, 1986).
3.2 VR in the lower limb motor rehabilitation
Several VR applications were designed to recover efficient walking in patients with lower limb motor impairment after stroke (Deutsch et al, 2002 and Deutsch et al, 2004). Fung, et al. (Fung, et al 2004, 2006), performed studies on gait training by using a treadmill mounted on a 6-degree-of-freedom motion platform with a motion-coupled VR environment. The system provided the unique feature of simulated turning within the environment; also provided auditory and visual cues as positive/negative feedback. Subjects were required to wear 3D stereo glasses to visualize the virtual environment. Test results from this project demonstrated improved gait speed with training. More recently, Mirelman et al. studied the effects on impaired gait kinetics of robot-assisted rehabilitation with or without VR support (Mirelman et al, 2007). Subjects in the two subgroups were trained with the same exercises; requested movements were inversion and eversion, dorsiflexion and plantar flexion, or combinations of these. A motion capture system was used to measure movements in association with a force-feedback system permitting navigation within a virtual environment displayed on a computer screen. Gait was estimated at baseline, one week after the training session, and three months after end of the therapy. Feedback information was provided directly by the system to the patients treated in the VR setting and by the therapist to those undergoing rehabilitation withouth VR support. Both groups improved, but patients treated with VR support did better, with increased ankle strenght at the end of treatment and at follow-up. Patients undergoing robotic- assisted rehabilitation without VR support reported fatigue earlier. Park and colleagues (Park et al, 2007) developed a VR system for motor rehabilitation with a PC camera and two markers of movement in a very simple virtual scenario with a crossing-stepping stone task. The success rate in 9 hemiplegic patients with stiff-knee gait after stroke was computed as the ratio of successful trials to the total gait cycles, with a ~30% improvement after treatment.
3.3 VR and telemedicine: the upper limb tele-rehabilitation
VR settings are usable in the transfer of available occupational treatments to a platform for rehabilitation at home, with remote control by therapist. Broeren and coworkers have emphasized the reduced labor, logistics and costs of the state-of art web-based video/audio systems for telemedicine and tested their protocol in a case study, with VR associated to the haptic force feedback necessary for VR object manipulation. The hand fine dexterity and grip improved after treatment (Broeren et al, 2002). More recently, Trotti and colleagues proposed VR-supported training as an integration of the conventional rehabilitation protocols (Trotti et al, 2009). They used kinematics indexes (such as movement execution time and precision) and validated clinical scales, such as Nine-Hole Peg Test (NHPT) (Mathiowetz, 1985), Frenchay Arm Test (FAT) (Heller, 1987), Medical Research Council (MRC) (Florence, 1992), Motricity Index (MI) (Bohannon, 1999), and the Motor Evaluation Scale for Upper extremity in Stroke Patients (MESUPES) (Van de Winckel, 2006) to measure the upper limb impairment in a patient with stroke before and after therapy with VR-supported upper limb rehabilitation. Kinematic analysis and most clinical scales (MRC of fingers, MESUPES and NHPT time, but not MI and FAT) showed a decrease in movement execution time and increase in precision, with improved muscle strenght and movement control (Trotti et al, 2009). VR-mediated telerehabilitation was further investigated (Piron et al, 2009) by comparing two groups (18 subjects each) of patients with stroke treated for four weeks by a VR-assisted rehabilitation program operated through Internet or by conventional therapy. Motor impairment was assessed one month before, at the beginning and end of therapy, and one month later by means of the Fugl-Meyer Upper Extremity (FM EU), Abilhand (Penta et al, 2001) and Ashworth (Bohannon & Smith, 1987) scales. The setting included a virtual teacher showing the correct trajectories as set by the therapist in association with the patient’s actual movement. The knowledge-of-performance was provided via videoconference. No differences were observed when comparing the assessments one month before and at beginning of therapy, but both groups improved after therapy and the improvement was evident also one month after the end of therapy. The FM UE showed better recovery for patients treated through VR-based telerehabilitation.
4. Systems and applications
4.1 Systems and applications for VR-supported upper limb rehabilitation
A VR system purported to measure the impairment in speed, strength, fractionation and range of fingers movements was designed to be distributed over three sites connected via Internet (for rehabilitation, data storage and data access, respectively) (Boian et al, 2002). At the rehabilitation site, the system featured a workstation and two sensing (cyber and haptic) gloves; the data storage site organized the information acquired during the VR-supported rehabilitation; open access to data was through Internet. An algorithm was implemented to increase or decrease according to the achieved performance the difficulty of the target task. The system was tested in a pilot study on 4 patients with stroke. A screen provided the patients with knowledge-of-results and performance (feedback) through a transparent hand representing the target and numerical scores about the trial execution. Trained patients achieved various degrees of improvement, with a good retention in gains and a positive evaluation of the system both by patients and therapists. A virtual tabletop environment for the upper limb rehabilitation after traumatic brain injury was developed (Wilson et al, 2007) to measure the residual function and kinematic markers like speed, precision, distance, accuracy of targeting. The system was innovative because flexible, automated, and relatively inexpensive, with components specifically designed to be user-friendly: LCD panels easy to carry and reducing the set-up time were favored; the virtual environment was displayed on the LCD panel placed horizontally on a tabletop surface, and users could interact with the system by moving sensing-objects over it; knowledge of results was provided to the patient via another LCD panel. Distractive elements appeared on the LCD to increase or decrease the task difficulty. Low-cost implements, such as commercial game controllers and marker tracking were used. Wilson and coworkers suggested that psychometric measures should be preferred in the future and predicted broad application in assessing movement impairment after stroke and ischemic or traumatic brain damage or in movement disorders (e.g. Parkinson or Huntington’s diseases). Therapy WREX (T-WREX) was designed by Reinkensmeyer and Housman (2007) for the hand and arm rehabilitation after stroke to make rehabilitation possible also in the absence of the therapist, with exercises mimiking the daily living activities in VR environment and a feedback information procedure. The system featured a passive gravity-balancing orthosis based on the Wilmington Robotic Esoskeleton (WREX) (Rahman et al, 2007), a hand grip sensor and the software needed for VR and performance evaluation, but was not a robotics/VR integration because WREX assisted patients only against gravity and by elastic bands. It focused on the re-training of function on a plane, therefore displaying the movement on the plane of interest. The system bypassed the problems of 3D complexity, but limitated the movements to be re-learned. Most patients nevertheless found T-WREX less boring than conventional therapy and their progress during rehabilitation easier to track. Reiteration of motor training by T-WREX reduced motor impairment (as measured by the Fugl-Meyer scale) in a preliminary randomized controlled study (Reinkensmaeyer & Housman, 2007).
4.2 Low-cost and open source systems for VR-supported tele-rehabilitation
Interest on tele-rehabilitation as a possible alternative to the traditional treatment of inpatients in hospital increased in recent years with the increment of costs and commitment by the private and public healthcare. Sugarman and colleagues assumed it is impossible for the therapist to monitor patients performing rehabilitation at home, emphasized the therapist’s role in motivating the patient and the need of efficient communication between the therapist and patients at any time and place, including home (Sugarman et al, 2006). Approaches combining VR and mechanical devices for rehabilitation (Fasoli et al, 2004; Coote & Sokes, 2005; Broeren et al, 2004; Reinkensmeyer et al, 2002; Jadhav & Krovi, 2004) appear encouraging, but the systems specifically developed for these purposes have high costs. In alternative, Sugarman and colleagues adopted a commercial feedback joystick in association with a specifically designed armrest and a PC with Internet connection. Their proposed VR solution could be operated in two different modes: stand-alone or cooperative. In the former, patients exercised at home without Internet connection; in the latter, the patient and therapist were online and the therapist could monitor and tutor the patient performing.