360° immersive virtual reality-based mirror therapy for upper extremity function and satisfaction among stroke patients: a randomized controlled trial

Stroke often results in persistent upper extremity deficits that require effective rehabilitation. Traditional mirror therapy (MT), though widely used, has limitations including trunk asymmetry from leaning to view the mirror, constrained movement space behind the mirror, a reliance on bimanual tasks with the unaffected limb, and reduced attention to the mirror image. Virtual reality-based MT (VRMT) has emerged to address these shortcomings by using head-mounted displays and immersive environments, potentially enhancing action observation and motor relearning. Prior work suggests VR-based rehabilitation can induce favorable neurophysiological changes and engage the mirror neuron system, but many VRMT implementations rely on artificial graphics that may reduce immersion or create mismatches between proprioception and visual feedback. The present study aimed to evaluate a novel 360° immersive VR-based MT (360MT) that uses recorded video of the patient’s own unaffected hand displayed via HMD, and to compare its efficacy and patient satisfaction against traditional MT and conventional physical therapy.

The paper reviews limitations of traditional MT in stroke rehabilitation (postural asymmetry, limited workspace, bimanual constraints, reduced attention). Advances in VRMT show potential benefits, including cortical reorganization favoring ipsilesional activation and enhanced frontal activity. However, many VRMT studies used artificial hand graphics, potentially diminishing immersion and creating visual-proprioceptive discrepancies. Prior video-augmented MT studies using recordings of the unaffected limb improved attention and proximal upper limb function, suggesting that displaying real limb movements may strengthen the mirror illusion and action observation effects. No prior studies had implemented a 360° camera with HMD reflecting the patient’s own hand for upper limb rehabilitation in stroke, motivating the current investigation.

Design: Prospective, active control, assessor-blinded, parallel-group randomized controlled trial; IRB-approved (SYU 2023-01-009-001) and registered (NCT05796843). Recruitment and Participants: First-ever stroke patients with hemiplegia were recruited from an M hospital in Seoul, South Korea (April 1 to May 4, 2023). Inclusion criteria: within 6 months since stroke onset, able to follow instructions, MMSE-K ≥21, mild to moderate motor impairment per FMA-UE. Exclusion criteria: mental disorder/Alzheimer’s disease, orthopedic/musculoskeletal injury, apraxia, prior MT/related investigations. Sample size: G*Power with α=0.05, power=0.8, effect size=0.55 (based on pilot FMA-UE) indicated n=36; 45 were recruited to account for dropout. Randomization and Blinding: Participants were randomly allocated 1:1:1 into 360MTG, TMTG, and CG using a computer-generated list. Two occupational therapists (≥5 years experience) performed baseline and post assessments blinded to group. Three physical therapists (≥8 years neuro PT experience) delivered treatments. Interventions: All groups received conventional physical therapy (CPT). 360MTG additionally received 360° immersive VR-based MT. TMTG additionally received traditional MT. - Conventional Physical Therapy (all groups): 60 minutes/day, 5 sessions/week, for 4 weeks. Protocol based on Sánchez-Sánchez et al.: warm-up (active-assisted scapular elevation/retraction, shoulder flexion/abduction, elbow flexion/extension, wrist circumduction, finger opposition; stretches to pectoralis major, triceps brachii, forearm flexors; ~20s each, 2 reps), circuit (elastic band strengthening for shoulder lateral rotation/flexion/abduction/serratus anterior; bimanual manipulation of small objects; pédalier tasks; elbow flexion-extension; forearm pronation-supination; wrist extension strengthening; gripping/releasing objects), cooldown (playful ring game; stretches to pectoralis major, triceps, forearm flexors; ~20s each, 2 reps). - 360MTG protocol: Before intervention, each participant performed treatment tasks using their unaffected upper limb, recorded with a 360° camera (Insta360 X3). Camera height was set below eye level and recordings were made in the intervention location. Videos were processed (Final Cut Pro) by mirroring and adjusting X-axis -10° and Y-axis +10° to minimize blind spots. During sessions, processed videos were displayed via HMD (Pico G2 VR 4K). Therapists fitted devices and guided participants to perform tasks with the affected limb while viewing the mirrored recording. MT tasks matched TMTG tasks. Schedule: 30 minutes/session, 3 sessions/week, for 4 weeks, plus CPT. - TMTG protocol: Traditional MT using an acrylic mirror (40×40×3 cm). The unaffected upper limb was mirrored; the affected limb was hidden behind the mirror. Therapists adjusted posture and instructed tasks: finger flexion-extension; wrist flexion-extension and pronation-supination; elbow flexion-extension (10 repetitions each); playable ring toy tasks (5 repetitions). Schedule: 30 minutes/session, 3 sessions/week, for 4 weeks, plus CPT. - CG protocol: CPT only (as above). Outcomes: Primary: Fugl-Meyer Assessment for Upper Extremity (FMA-UE; max 66; Korean version; reliability intra-rater r=0.99, test-retest r=0.99). Secondary: Manual Function Test (MFT; max 32; test-retest r=0.994; inter-rater r=0.993) and Box and Block Test (BBT; 2.5 cm³ blocks moved in 1 minute; test-retest r=0.97). Patient-reported experience and satisfaction: 10-item questionnaire (5-point Likert) based on Hills & Kitchen guidelines; reviewed by experienced therapists/researchers; administered to 360MTG and TMTG post-intervention. Semi-structured interviews captured perceived advantages/disadvantages and suggestions; responses summarized and analyzed. Statistical Analysis: Shapiro-Wilk for normality; Chi-square for categorical and one-way ANOVA for continuous baseline variables; paired t-tests for within-group pre-post differences; one-way ANOVA for between-group changes with LSD post hoc; significance set at P<0.05.

- Enrollment: 62 screened; 17 excluded; 45 participants completed interventions and evaluations without dropout (April 1–May 4, 2023). Randomized equally into 360MTG (n=15; 7 men, 8 women), TMTG (n=15; 8 men, 7 women), CG (n=15; 8 men, 7 women). Mean time since onset: 4.8±1.22 months. No significant baseline differences across groups. - Primary and secondary outcomes: Both 360MTG and TMTG showed significant pre- to post-intervention improvements in FMA-UE, MFT, and BBT (P<0.05). The CG did not show significant improvements. - Between-group comparisons: Change scores in 360MTG were significantly higher than TMTG and CG for FMA-UE, MFT, and BBT (P<0.001). Changes in TMTG were not significantly different from CG. - Representative pre/post means (mean±SD): FMA-UE (360MTG: 34.67±10.96 to 37.26±11.22; TMTG: 36.93±10.46 to 38.40±10.64; CG: 38.06±11.94 to 38.80±12.13); MFT (360MTG: 14.13±3.94 to 16.06±4.51; TMTG: 13.53±3.09 to 14.60±3.15; CG: 15.60±4.76 to 15.86±5.08); BBT (360MTG: 1.33±1.34 to 2.66±1.75; TMTG: 1.20±1.47 to 1.80±1.61; CG: 1.40±1.35 to 1.60±1.63). - Patient experience and satisfaction (N=30; 360MTG vs TMTG): Significant differences favoring 360MTG on items 1, 4, 5, 6, 7, 9, and 10 (P values generally ≤0.026), indicating higher engagement, comfort, and satisfaction. - Interview themes: 360MTG participants frequently reported the intervention as enjoyable, more immersive, and allowing freer movement; TMTG participants reported difficulty concentrating on the mirror, maintaining posture, and limitations due to small space. - Safety: No serious adverse events reported.

The study addressed whether a 360° immersive VR-based MT using recorded video of the patient’s own unaffected hand can improve upper extremity outcomes and patient satisfaction compared with traditional MT and conventional therapy. Both MT approaches improved motor outcomes versus baseline, but 360MT produced significantly greater gains than TMT and control, and TMT did not differ significantly from control in change scores. These findings suggest that enhanced immersion and realistic limb representation may strengthen the mirror illusion and action observation processes that underpin MT, potentially reducing mismatches between visual input and proprioception often seen with artificial graphics. The 360MT approach may also mitigate known limitations of traditional MT, such as trunk asymmetry, constrained workspace, and reduced attention due to the need to monitor the unaffected hand during bimanual tasks. The higher satisfaction and reported engagement align with literature indicating VR can enhance motivation and adherence. Improved attention to the mirrored image is theorized to facilitate mirror neuron system activation and cortical reorganization, thereby contributing to functional recovery. Collectively, the results support the relevance of immersive, unilateral-focused VRMT in sub-acute stroke rehabilitation.

The 360° immersive VR-based mirror therapy demonstrated superior improvements in upper extremity function and higher patient satisfaction compared with traditional mirror therapy and conventional physical therapy alone in stroke patients within six months of onset. By providing a highly immersive and realistic representation of the patient’s own limb, 360MT may enhance engagement and therapeutic effectiveness. Future research should include larger samples, long-term follow-up, direct neurophysiological measures of cortical activation, and comparisons with VRMT using computer-generated graphics to further validate and refine this approach.

- No long-term follow-up assessments to determine sustained effects. - No measurement of cortical activation patterns during treatment to elucidate neurophysiological mechanisms. - No direct comparison between 360MT and VRMT using computer-generated graphics. - Did not quantify patients’ concentration or attention duration during sessions. - Single-center study; broader generalizability requires multi-center trials.