Edited by: Zhiyong Zhao, Zhejiang University, China
Reviewed by: Zaixu Cui, Chinese Institute for Brain Research, Beijing (CIBR), China; Xin Gao, Shanghai Universal Medical Imaging Diagnostic Center, China; Li Xuejing, The Affiliated Huai’an Hospital of Xuzhou Medical University, China
†These authors have contributed equally to this work and share first authorship
This article was submitted to Brain Health and Clinical Neuroscience, a section of the journal Frontiers in Human Neuroscience
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Post-stroke depression (PSD) is a serious complication of stroke that significantly restricts rehabilitation. The use of immersive virtual reality for stroke survivors is promising. Herein, we investigated the effects of a novel immersive virtual reality training system on PSD and explored induced effective connectivity alterations in emotional networks using multivariate Granger causality analysis (GCA). Forty-four patients with PSD were equally allocated into an immersive-virtual reality group and a control group. In addition to their usual rehabilitation treatments, the participants in the immersive-virtual reality group participated in an immersive-virtual reality rehabilitation program, while the patients in the control group received 2D virtual reality rehabilitation training. The Hamilton Depression Rating Scale, modified Barthel Index (MBI), and resting-state functional magnetic resonance imaging (rsfMRI) data were collected before and after a 4-week intervention. rsfMRI data were analyzed using multivariate GCA. We found that the immersive virtual reality training was more effective in improving depression in patients with PSD but had no statistically significant improvement in MBI scores compared to the control group. The GCA showed that the following causal connectivities were strengthened after immersive virtual reality training: from the amygdala, insula, middle temporal gyrus, and caudate nucleus to the dorsolateral prefrontal cortex; from the insula to the medial prefrontal cortex; and from the thalamus to the posterior superior temporal sulcus. These causal connectivities were weakened after treatment in the control group. Our results indicated the neurotherapeutic use of immersive virtual reality rehabilitation as an effective non-pharmacological intervention for PSD; the alteration of causal connectivity in emotional networks might constitute the neural mechanisms underlying immersive-virtual reality rehabilitation in PSD.
Patients after stroke are at an increased risk of impaired mental health (
Recently, virtual reality (VR) has emerged as a valuable approach in the rehabilitation of stroke survivors (
Post-stroke depression is also associated with anomalies in brain structure and function, somewhat similar to major depressive disorder (MDD) (
In the present study, we presented a novel immersive VR training system, tested it on PSD patients, and compared the effects on depression to conventional 2D rehabilitation activities. It was hypothesized that rehabilitation combined with immersive VR training would improve depressive outcomes. Furthermore, we investigated alterations in effective connectivity in emotional networks after immersive VR rehabilitation for PSD using multivariate Granger causality analysis (GCA). GCA is a powerful technique for extracting such connectivity from data, which provide sensitive evidence for the detection of functional interactions between brain regions.
We included participants who: met the Diagnostic and Statistical Manual of Mental Disorders criteria for depressive disorder with a Hamilton depression Rating Scale (HAMD) score of 17 or higher; met the criteria for the diagnosis of cerebral infarction; were within 2 weeks to 12 months after stroke; signed informed consent; were 18–65 years old; were right-handed. There were no limits on patient sex. Experienced neuropsychologists conducted clinical interviews. The exclusion criteria included: (a) history of other severe mental disorders, epilepsy, drug abuse, or antidepressant use at stroke onset; (b) serious systemic diseases such as diabetes and uremia; (c) vision disorders and severe motor dysfunction that could interfere with the execution of the VR training task; (d) serious negative suicidal ideation; and (f) inability to cooperate and complete training and evaluation for any reason.
The research was performed in rehabilitation clinics in Yueyang Hospital, affiliated with Shanghai University of Traditional Chinese Medicine. The study protocol was approved by the local research ethics committee and registered with the Chinese Clinical Trial Registry (ChiCTR2100052132). There were 44 patients with PSD included in the study. They were allocated to two groups: the immersive-VR group (
The flow diagram of the study process. VR, virtual reality.
In addition to their usual rehabilitation treatments such as physical and occupational therapy, which mostly involved motor training, the participants in the immersive-VR group underwent an immersive-VR rehabilitation training program for 20 min/d, 5 days a week for 4 weeks, while the patients in the control group underwent 2D VR rehabilitation training.
In the immersive-VR group, 3D immersive-VR rehabilitation was conducted by the rehabilitation system developed by Shanghai University of Traditional Chinese Medicine in collaboration with the School of Mechatronic Engineering and Automation of Shanghai University. The virtual environment was developed and rendered using the Unity 3D engine and was run on a PC (CPU Intel I7 9700K). A VR headset (VIVE Pro; HTC, Taoyuan City, Taiwan) was used to present the 3D virtual training environment in the rehabilitation system. Participants could rotate their heads for a 360° view of the virtual scene and interact freely in the virtual environment with the VR headset. Assisted by a therapist, participants were asked to sit in an open room and wear a VR helmet. Through the headset, the participants could view a bright and spacious living room that could make them comfortable and stable. Within the virtual environment, the indoor furniture layout restored reality and provided the participants with an immersive training experience. A built-in positioning function in the software was conducted firstly to correspond the sofa position in the scene to the patient’s sitting position. During the training task, three different common fruits were generated in the virtual scene, and participants were asked to select the fruit prompted by sound or text within the virtual scene. After completing the interaction, encouraging vocal feedback was provided to the participants.
An evaluation score was given after all training tasks were completed. In the VR rehabilitation system, 3D upper limb reaching movements were captured and mapped onto the movements of a virtual arm. Before training, the difficulty of the training task was set according to the patient’s upper limb function to ensure that the fruit appeared within the patient’s reach. Participants were asked to select the fruits using a wireless remote control secured with straps if the affected upper limb function was sufficient to complete the harvesting task. Assistance with a manipulator was be provided if the patients could not complete the harvesting task by themselves. The immersive VR scene is presented in
The virtual reality (VR) training scenarios in the immersive-VR group and the control group.
In the control group, 2D VR rehabilitation was conducted using the Fourier M2 rehabilitation system (Fourier International Pte. Ltd., Shanghai, China). Assisted by a therapist, participants were asked to sit on a chair; the paretic hand was secured on the end-effector with straps. Through the end-effector, the upper limb movements could be captured. The user’s movements through the end-effector were mapped onto the movements of a virtual hand on the computer screen. A harvesting task was selected for the patients in the control group. Participants were asked to reach the fruits displayed on the computer screen through the virtual hand during the training task. Before training, the difficulty of the game was set according to the patient’s upper limb function. The 2D VR scene is presented in
All outcome measures were evaluated before and after a 4-week rehabilitation. A trained investigator administered the clinical measurements, and the rsfMRI data was analyzed by another trained investigator. In this study, both investigators were blinded to the group allocations.
In this study, HAMD score was considered the primary outcome measure. It is the most commonly used clinical evaluation scale for depression, which is frequently applied in patients with PSD (
The secondary outcomes concerned the effects on the activities of daily living and effective connectivity in the brain emotional networks.
The MBI is frequently used as a measure of activities of daily living. The scale comprises 10 items: feeding, bowel control, bladder control, personal hygiene, transfer, dressing, toileting, ambulation, bathing, and stair climbing. The total score ranges from 0 (completely dependent) to 100 (fully independent).
MRI measurements were conducted on a 3.0-Tesla Magnetom Verio MRI scanner (Siemens Healthcare, Erlangen, Germany) with a 32-channel phased-array head coil. All participants were asked to stay awake and still without thinking during scanning, and foam pillows were used to minimize head movement. The rsfMRI data were acquired by a single-shot gradient-recalled echo-planar imaging sequence with the following parameters: repetition time = 3,000 ms, flip angle = 90°, interleaved scanning order, slice number = 43, matrix size = 64 × 64, slice thickness = 3.0 mm (no-gap), field of view = 230 mm × 230 mm, and 200 volumes. T1-weighted data were acquired by a 3D magnetization-prepared rapid acquisition gradient-echo sequences with the following parameters: repetition time = 1,900 ms, inversion time = 900 ms, echo time = 2.93 ms, flip angle = 9°, field of view = 256 mm × 256 mm, slice thickness = 1 mm, acquisition matrix = 256 × 256, and number of averages = 1.
The rsfMRI data was pre-processed using Statistical Parametric Mapping software (version 12; Wellcome Trust Centre for Neuroimaging, London, United Kingdom)
Previous studies have shown some reliable findings that the prefrontal lobe−limbic system−cortical striatum−thalamic circuit was important in the emotional networks involved in PSD (
Map illustrating brain of the regions of interest selected in the granger causality analysis. MPFC, medial prefrontal cortex; DLPFC, dorsolateral prefrontal cortex; SFG, superior frontal gyrus; MFG, middle frontal gyrus; OrG, orbital gyrus; STG, superior temporal gyrus; MTG, middle temporal gyrus; ITG, inferior temporal gyrus; FuG, fusiform gyrus; PhG, parahippocampal gyrus; pSTS, posterior superior temporal sulcus; INS, insular gyrus; CG, cingulate gyrus; Amyg, amygdala; Hipp, hippocampus; BG, basal ganglia; CAU, caudate, Tha, Thalamus.
In this study, we used Granger causality to analyze the effective connectivity between the reference signals of the seed regions (
SPSS (Version 21.0; IBM Corp., Armonk, NY, United States) and MATLAB 2014a were used to perform the statistical analyses in this study. The baseline characteristics of the two groups were compared using the Student’s
A total of 44 patients with PSD were included in the analysis: 22 patients in the immersive-VR group and 22 patients in the control group. No significant differences were found in terms of age, gender, days after stroke, side of the lesion, and HAMD scores between the two groups at baseline. The demographic and clinical characteristics are summarized in
Demographic and clinical characteristics.
Variable | The immersive-VR group | The control group | |
Age (years) | 51.95 ± 12.41 | 50.73 ± 9.99 | 0.73 |
Gender (male/female) | 13/9 | 14/8 | 0.76 |
Duration after stroke (months) | 7.27 ± 3.79 | 6.82 ± 3.26 | 0.63 |
Side of lesion (left/right) | 16/6 | 15/7 | 0.74 |
HAMD scores | 23.09 ± 1.87 | 22.05 ± 1.94 | 0.12 |
VR, virtual reality; HAMD, Hamilton depression Rating Scale.
The Time × Group effect was significant for HAMD scores [
Outcome | The immersive-VR group |
The control group |
||||||
HAMD | 30.76 | <0.001 | 23.09 ± 1.87 | 18.27 ± 3.33 | 0.00 | 22.05 ± 1.94 | 21.09 ± 1.85 | 0.06 |
MBI | 2.02 | 0.16 | 73.64 ± 18.91 | 78.86 ± 16.54 | <0.001 | 74.55 ± 15.35 | 81.59 ± 13.75 | <0.001 |
VR, virtual reality; HAMD, Hamilton depression Rating Scale; MBI, modified Barthel Index.
At baseline, there was no significant difference in MBI between the two groups. After treatment, the two groups both showed significantly higher MBI scores compared to the baseline levels (
We used Granger causality to investigate alterations in brain effective connectivity in emotional networks.
ROI1 | ROI2 | The immersive-VR group |
The control group |
||||||
Tha_R_8_4 | pSTS_R_2_2 | 13.1626 | 0.0012 | –0.1985 | 0.0353 | 0.1310 | 0.0034 | ||
BG_R_6_1 | MFG_R_7_3 | 12.6577 | 0.0014 | –0.1451 | 0.6196 | 0.0139 | 0.2766 | –0.2037 | 0.0204 |
MTG_L_4_4 | MFG_L_7_3 | 12.6051 | 0.0014 | –0.3213 | 0.4266 | 0.0381 | 0.1422 | –0.0565 | 0.0042 |
INS_L_6_6 | MFG_L_7_3 | 12.2226 | 0.0017 | –0.2762 | 0.0367 | 0.0658 | –0.3920 | 0.0058 | |
Amyg_L_2_2 | MFG_R_7_3 | 12.1175 | 0.0017 | –0.1749 | 0.0056 | 0.1580 | –0.2898 | 0.0896 | |
Amyg_L_2_1 | MFG_L_7_3 | 11.5652 | 0.0021 | 0.0194 | 0.1183 | 0.0237 | |||
INS_L_6_3 | Tha_L_8_5 | 11.1440 | 0.0025 | –0.4270 | –0.1809 | 0.0605 | –0.3025 | 0.3370 | 0.0048 |
CG_L_7_4 | Hipp_L_2_2 | 11.0412 | 0.0026 | 0.0587 | 0.0936 | 0.3264 | − |
0.0023 | |
INS_R_6_3 | SFG_R_7_7 | 10.9122 | 0.0027 | –0.1939 | 0.0173 | 0.2837 | –0.1126 | 0.0408 | |
Tha_L_8_4 | MTG_L_4_3 | 10.2591 | 0.0035 | 0.2524 | –0.2244 | 0.1904 | 0.0348 | 0.4577 | 0.0009 |
INS_R_6_5 | MFG_L_7_3 | 10.1095 | 0.0037 | –0.3646 | 0.7435 | 0.0881 | 0.0584 | –0.2839 | 0.0048 |
Amyg_L_2_2 | MTG_R_4_2 | 9.8908 | 0.0040 | 0.3379 | 0.0658 | –0.3253 | 0.0096 | ||
Amyg_L_2_1 | SFG_R_7_7 | 9.8481 | 0.0041 | 0.0058 | 0.4969 | 0.1166 | 0.2551 |
Group mean path coefficients with predictions going from ROI1 to ROI2 are shown. Group means in bold are significantly different from zero (two-tailed tests;
Map illustrating brain of multivariate GCA signed-path coefficients alteration after treatment in the immersive-VR group
Circlize maps of signed-path coefficient alteration after treatment in the immersive-VR group
After treatment in the immersive-VR group, the causal connectivity between the following regions was strengthened significantly compared to pre-treatment: from Tha_R_8_4 to pSTS_R_2_2, from BG_R_6_1 to MFG_R_7_3, from MTG_L_4_4 to MFG_R_7_3, from INS_L_6_6 to MFG_R_7_3, from Amyg_L_2_2 to MFG_R_7_3, from Amyg_L_2_1 to MFG_L_7_3, from INS_R_6_3 to SFG_R_7_7, and from Amyg_L_2_1 to SFG_R_7_7.
After treatment in the control group, the causal connectivity in patients with PSD from INS_L_6_3 to Tha_L_8_5 and from Tha_L_8_4 to MTG_L_4_3 was strengthened significantly. The causal connectivity between the following regions was weakened significantly compared to pre-treatment: from Tha_R_8_4 to pSTS_R_2_2, from BG_R_6_1 to MFG_R_7_3, from MTG_L_4_4 to MFG_R_7_3, from INS_L_6_6 to MFG_R_7_3, from Amyg_L_2_1 to MFG_L_7_3, from INS_R_6_3 to SFG_R_7_7, from CG_L_7_4 to Hipp_L_2_2, from INS_R_6_5 to MFG_L_7_3, and from Amyg_L_2_2 to MTG_R_4_2.
In this study, we found that immersive VR training was more effective in improving depression in patients with PSD but had no statistically significant improvements activities of daily living when compared to the control group. Furthermore, we used multivariate GCA to explore the alterations in effective connectivity in the brain areas of emotional networks. Notably, the GCA results mainly showed that the causal connectivity from the amygdala, INS, middle temporal gyrus (MTG), and caudate nucleus to the dorsolateral PFC (DLPFC); from the INS to medial PFC (MPFC); and from the thalamus to the posterior superior temporal sulcus were strengthened after immersive VR training, while they were weakened after treatment in the control group.
In the past few years, many preliminary studies have shown that VR-based training combined with traditional rehabilitation was more effective for improving depressive symptoms in stroke survivors than traditional rehabilitation (
Brain remodeling is the core mechanism of stroke rehabilitation (
The basic neuroscience behind VR-based treatment has mainly been linked to functional brain remodeling (
The structures that showed altered effective connectivity with the DLPFC and MPFC were located in the caudate nucleus, insula, anterior superior temporal sulcus, and the amygdala. The affective control network, such as the thalamus, amygdala, insular, and striatum, has been shown to be involved with the emotional dysregulations in MDD (
We also found that causal connectivities from the insula to the thalamus and from the thalamus to the MTG were strengthened after treatment in the control group. Previous literature has shown that emotional networks mainly produce negative regulation signals
First, we only evaluated the patients’ depressive symptoms and did not measure recovery of motor function, although poor performance in motor activities can lead to depression in general (
Overall, this study revealed that immersive-VR rehabilitation could improve depression in patients with PSD by exposing patients to an enriched, immersive-VR environment. Furthermore, this finding was supported by increased effective connectivity in emotional networks mainly directed toward the DLPFC and MPFC, which also showed a significant decrease in the control group. Our results indicated the neurotherapeutic use of immersive-VR rehabilitation as an effective non-pharmacological intervention for PSD as well as alterations of causal connectivity in emotional networks that might contribution to symptom improvement.
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
The studies involving human participants were reviewed and approved by the Ethics Committee of Yueyang Hospital, affiliated with Shanghai University of Traditional Chinese Medicine. The patients/participants provided their written informed consent to participate in this study.
J-GX, C-LS, and X-YH defined and designed this study, and interpreted the results. M-XZ and J-JW analyzed the data and wrote the manuscript. DW, Y-LL, and XX ran the intervention. JM and X-XX collected the data. All authors revised and approved the current version of the manuscript.
This work was supported by the National Key R&D Program of China (Grant Nos. 2018YFC2001600 and 2018YFC2001604), National Natural Science Foundation of China (Grant Nos. 81802249, 81871836, 81874035, and 81902301), Shanghai Science and Technology Committee (Grant No. 22010504200), Shanghai Rising-Star Program (Grant No. 19QA1409000), Shanghai Municipal Commission of Health and Family Planning (Grant Nos. 2018YQ02 and 201840224), and Program of Shanghai Academic Research Leader (Grant No. 19XD1403600).
We would like to thank all the patients that participated in and contributed to the study.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
The Supplementary Material for this article can be found online at: