1. Introduction
Stroke is one of the main causes of long-term disability in adults [
1]. The most general stroke symptom is the limitation of upper limb function which significantly impacts on a person’s level of independence and daily living activities [
2]. Some rehabilitation interventions which may involve different exercises, type of equipment or techniques have developed to improve the upper limb function and reduce the disability [
3,
4,
5,
6]. Particularly, traditional rehabilitation techniques are effective approaches in motor function recovery. However, these methods are time-consuming, expensive, and intensive resources as well as specialized facilities are needed [
7,
8]. Therefore, post-stroke rehabilitation needs new approaches that allow individuals to enhance motor function.
Recently, mirror therapy has been considered as an effective adjunct to improve upper extremity motor function for stroke patients [
9,
10]. The principle of the mirror therapy techniques is the reflective superimposition of healthy extremity movements on the affected upper limb through a mirror [
11]. This induces a visual illusion of increased movement ability of the unhealthy extremity for the patient [
12]. Due to the simplicity of mirror therapy systems, they become difficult to provide a perfectly synchronized mirrored movement for the patient [
13].
To address limitations of the mirror therapy, avatar images or virtual objects have been constructed by image processing techniques in virtual reality (VR) systems. These systems create a challenging and motivating environment to provide enhanced feedback of movement characteristics and improved motor learning tasks for the patients [
14,
15]. However, the VR therapy systems have been widely used for the upper limb rehabilitation and have been adapted for individual patients after stroke [
16]. Furthermore, some clinical trial reports figured out the effectiveness of VR in upper limb function improvement [
17,
18,
19]. However, the systems could not provide a high sense of presence and embodiment to the patients due to using the rendered avatar images instead of real photos. Hence, the patients only obtain short-term effects from these systems. Meanwhile, augmented reality (AR) has emerged as one of the effective methods in rehabilitation. The AR combines real and virtual objects to run interactively in a common real-time environment [
20]. Despite exhibiting advantages similar to the VR in the upper limb function recovery, the AR could provide a higher degree of presence and embodiment than the VR [
21].
In order to overcome drawbacks of the VR technology, an AR system applying the mirror therapy principle has been reported to provide a higher quality of presence and embodiment for the patients [
13,
22,
23]. Moreover, the system could be feasible for long-term neuromodulation. In this study, we proposed an AR system applying mirror therapy mechanism for hemiplegic patients and evaluated its effectiveness in eliciting upper-limb motor recovery.
3. Results
A hemiplegic patient who satisfied our criteria was recruited to assess the feasibility of our AR system in a clinical trial.
Table 1 shows the detailed information of the patient.
3.1. Motor Function Improvement
To the primary outcome measure, there were small improvements in the writing (2 s of improvement) and stacking checkers (1 s of improvement) subtest of Jebsen Taylor test between pre- and post-intervention. However, no difference was found in the time total of Jebsen Taylor test in both evaluations (baseline: 65 s; post-treatment: 65 s). Noticeably, there was 4 points and 3 points improvement related to the wrist/hand and proximal arm subscale of Arm Motor Fugl-Meyer scale in the second outcome measure, respectively, after 10 days of intervention (
Table 2).
Furthermore, apparent enhancements were exhibited in the stroke impact, hand function scale, grasp force in Newtons scale, active ROM of wrist joint scale (flexion and extension), in nine-hole pegboard, and Korean version-Modified Barthel Index scale from (improvement from pre- to post-: intervention) 9 points, 5 (kg), 10° for flexion, 3° for extension, 5 s, and 3 points, respectively. Nevertheless, there were negligible decreases in both the wrist and elbow subscale of the Ashworth scale (pre-intervention: 1; post-intervention: 0) and the peak to peak Motor Evoked Potential was not evaluated after intervention due to the poor patient condition.
3.2. Validation of Virtual Reality and Agmented Reality
The authors used questionnaires consisting of six items with a scale rating of −5 (not at all) to 5 (perfect) for the validation of VR and AR (more details in
Appendix A).
Table 3 shows the validation results after 10 days of intervention. From the patient’s feedback, he did not feel the moving hand on the screen similar to his hand.
3.3. Adverse Effects
The authors utilized questionnaires including six items (dizziness, psychological anxiety, boredom, gloomy feeling, muscle twist, and motion sickness) (more detail in
Appendix B) with a scale rating of 0 (not at all), 2 (nervous), 4 (little uncomfortable), 6 (very uncomfortable), 8 (disturbed), and 10 (extremely disturbed) to assess the adverse effects after the intervention. The items dizziness, muscle twist, and motion sickness were reported after the intervention as shown in
Table 4. Therefore, the patient felt a little uncomfortable with our system after the intervention.
4. Discussion
The purpose of this study was to investigate the effects of AR system for the upper limbs of patients who experienced a stroke. Positive effects after the intervention were detected in the participant.
A few studies have chosen the Jebsen-Taylor hand function test (JTHFT) to evaluate the effect of motor function recovery because of (1) the standardized tasks relative to norms; (2) everyday activities tasks; (3) using readily available materials [
36]. In our results, there was no change in the total scores of JTHFT, but the results show the small improvements in writing and stacking checkers test of JTHFT after 10 days of intervention. Merians et al. used JTHFT to assess the fine motor dexterity of patients with brain damage. The clinical results showed the improvements in the speed and precision of fine movement in two of three patients. However, there was one patient who did not transfer the improvement to the functional activities after the intervention [
37]. These improvements of writing, stacking checker, and speed movement were consistent with the need for fine, sequential, and bimanual coordination skills of the finger and thumb.
In addition, the Arm Motor Fugl-Meyer Test is most commonly used to assess the upper limb function in the stroke patient [
38]. In this study, we used the wrist/hand and proximal arm Fugl-Meyer to assess the effect of functional recovery for the mildly impaired stroke survivor. After 10 days of intervention, the results showed the improvement in the scores of wrist/hand and proximal arm Fugl-Meyer. Particularly, the patient was able to reach the highest score of the proximal arm (34/34) and wrist/hand (23/24). Compared to the other methods, Faria et al. reported that both VR and control group showed significant improvements at the end of treatment in wrist/hand Fugl-Meyer (
p = 0.034 and
p = 0.04) [
39]. However, these improvements were significant with respect to baseline, but they were modest for both groups at end of treatment (0.8 ± 1.4 in VR versus 1.3 ± 2.3 in control). This indicated that the improvement of our system in wrist/hand Fugl-Meyer is superior to the improvement in VR. The prominence could be explained by using patients with better functions in wrist/hand Fugl-Meyer in the current study.
Furthermore, the specific task of the flexion and extension of the wrist joint promoted by our system could also explain the improvement detected in the active ROM and the nine-hole pegboard test, stroke impact scale, and grasp force in Newtons. This is also consistent with the previous report of robotic rehabilitation [
40]. Hence, our system showed to be as effective as other methods since it can promote hand dexterity, as measured by the nine-hole pegboard test and active ROM while being simpler than the robotic systems.
There were several limitations to our study. Firstly, although the results showed the improvement of the functional motor, they are not powered to detect significant changes due to the small sample size. Secondly, our system could not induce a high sense of the presence and embodiment for the patient. In order to assess the validity of presence of the system, most of VR or AR applying mirror therapy trials used a questionnaire. For instance, Koo et al. used a questionnaire including six items to check for real-time embodiment and virtual limb presence in which the mean scores were 4.0 and 4.0 (−5 to 5 rating), respectively [
13]. These scores indicated that their system could provide a high degree of presence and might induce a long-term effect. Compared to our study, the results exhibited lower scores of validation of virtual reality and augmented reality (mean 0.8 and −4.16, respectively) which is not enough to provide a high presence, indicating that our system could not induce a long-term effect for the patient in motor function recovery. Finally, some adverse events were noticed after the intervention following dizziness, muscle twist, and motion sickness. This is similar to the VR [
41,
42] and mirror therapy [
43,
44].
A future study involving a larger sample size and long-term for follow-up should be performed to enhance significant improvements in outcome measures. Besides, more exercises should be added into the system to establish a standardized program for post-stroke rehabilitation. Furthermore, we should design the system in three-dimension (3D) to provide a higher sense of presence and embodiment of the upper limb for the patient which might reduce the adverse effects and induce long-term effects for the patient.
In conclusion, we developed the AR system applying mirror therapy mechanism to supply the wrist rehabilitation exercise for hemiplegic stroke patients using real-time image processing techniques. Our findings demonstrated that the AR system was feasible in the improvement of motor function rehabilitation for a patient with hemiplegia.