Effects of Mind–Body Exercise on Brain Structure and Function: A Systematic Review on MRI Studies

Mind–body exercise has been proposed to confer both physical and mental health benefits. However, there is no clear consensus on the neural mechanisms underlying the improvements in health. Herein, we conducted a systematic review to reveal which brain region or network is regulated by mind–body exercise. PubMed, Web of Science, PsycINFO, SPORTDiscus, and China National Knowledge Infrastructure databases were systematically searched to identify cross-sectional and intervention studies using magnetic resonance imaging (MRI) to explore the effect of mind–body exercise on brain structure and function, from their inception to June 2020. The risk of bias for cross-sectional studies was assessed using the Joanna Briggs Institute (JBI) checklist, whereas that of interventional studies was analyzed using the Physiotherapy Evidence Database (PEDro) scale. A total of 15 studies met the inclusion criteria. Our analysis revealed that mind–body exercise modulated brain structure, brain neural activity, and functional connectivity, mainly in the prefrontal cortex, hippocampus/medial temporal lobe, lateral temporal lobe, insula, and the cingulate cortex, as well as the cognitive control and default mode networks, which might underlie the beneficial effects of such exercises on health. However, due to the heterogeneity of included studies, more randomized controlled trials with rigorous designs, similar measured outcomes, and whole-brain analyses are warranted.


Introduction
Mind-body exercise is a form of multicomponent exercise that combines movement sequences, breathing control, and attention regulation, which is different from traditional physical exercise [1]. It is also referred to as movement-based contemplative practice [2] or mindful movement [3], which emphasizes moving mindfully, commonly including Tai Chi Chuan (TCC), Qigong, and yoga. TCC is a form of mind-body exercise incorporating physical, cognitive, social, and meditative components [4]. Qigong involves a set of relatively slow exercises through coordinated physical movements, breathing, and meditative state to cultivate one's internal energy called "Qi" to achieve body healing, and Baduanjin (BDJ) is one of the most common forms of Qigong [5]. Yoga is an ancient mind-body exercise which focuses on the present moment, consisting of physical postures (asanas), control of breath (pranayama), and the use of meditation (dyana), and the most common form is Hatha yoga [6]. Compared with aerobic or resistance exercise, mind-body exercises are relatively low in intensity and slow in pace, particularly suitable for the elderly and groups with chronic diseases [7].
In recent years, increasing research evidence has shown that mind-body exercise could improve and promote physical health [8][9][10] as well as benefit mental health, including improving general cognition, executive function, learning, memory, and verbal fluency [11][12][13]. Moreover, it aids in relieving stress [14,15], anxiety, depression, and other impairment or suffering from organic diseases such as diabetes, fibromyalgia, knee osteoarthritis, or tinnitus to avoid interference of the results by these factors. (ii) Type of exercise: the experimental group engaged in mind-body exercises, including Tai Chi Chuan (TCC), Qigong, and yoga. Studies with multimodal interventions comprising mind-body exercises were also included to increase the number of enrolled articles. However, we excluded studies that examined the sole effects of mindfulness or meditation since the aim of this study was to examine the effect of holistic mindbody exercises which involve structured movements, breath control, and attention modulation on brain health. (iii) Study design: to gain an overall understanding of mind-body exercise-related changes in the brain plasticity of healthy adults, there were no restrictions on the study design. Therefore, randomized and non-randomized controlled interventions and within-subjects intervention studies as well as cross-sectional studies comparing experts to novices were all included. With reference to previous studies [25,26], for the intervention studies, the experimental group must have exercised for at least 4 weeks with more than one session per week, and for the cross-sectional studies, the regular exercise duration must have been not less than 3 years to provide sufficient time for changes in the brain structure and function to occur. (iv) Outcomes: the imaging technique was restricted to MRI, including sMRI, task fMRI, and resting-state fMRI. The outcome measures included changes in structure (i.e., gray matter volume, density, and cortical thickness) and task or resting-state (de)activation, functional connectivity for pre-to post-mind-body intervention, or mind-body expertnovice comparison. (v) Literature type: peer-reviewed articles published either in English or Chinese language.

Data Extraction
Two independent researchers (ZX and ZW) performed data extraction from the eligible studies. Any disagreements were resolved via discussions with a third researcher (ZB). For the cross-sectional studies, we extracted data on the sample size, group age, group description, main outcome measures, primary MRI results, and the association with behavior results. At the same time, the extracted data for intervention studies included participants and study design, sample size, group age, intervention frequency and duration, main outcome measures, primary MRI results, and the association with behavior results.

Quality Assessment
We assessed the methodological quality of the included cross-sectional studies using the Joanna Briggs Institute (JBI) checklist for analytical cross-sectional studies [26,27]. The checklist comprised 8 items and possible answers were "yes", "no", "unclear", or "not applicable". According to relevant studies [28,29], studies were characterized according to the following: (i) low risk of bias if studies reached more than 70% score "yes"; (ii) moderate risk of bias if "yes" scores were between 50% and 69%; and (iii) high risk of bias if "yes" scores were below 49%. For the intervention studies, we used the Physiotherapy Evidence Database (PEDro) scale developed by the Delphi list [30]. The scale consists of 11 items which were scored as either 1 (the answer was "yes") or 0 (the answers were "no", "unclear", or "not applicable"). The studies were classified, using the total rating score (item 1 not scored), as having excellent (9-10), good (6)(7)(8), fair (4)(5), or poor (<4) quality [13]. In addition, to reduce the risk of bias in the assessment, two researchers (ZX and ZW) scored the quality of the included articles independently. Any conflicting scores between the two researchers were resolved via a discussion with a third researcher (ZB). Figure 1 shows a flow chart summarizing the study selection process recommended by the PRISMA (preferred reporting items for systematic reviews and meta-analyses) Brain Sci. 2021, 11, 205 4 of 19 guidelines. Overall, 29 studies focused on evaluating mind-body exercise and brain plasticity using MRI met the inclusion criteria. It is worth noting that in 14 cases, authors published multiple articles on one actually performed study; therefore, we merged the related articles into one study, resulting in 15 actual studies. The characteristics of the included cross-sectional and intervention studies are shown in Tables 1 and 2, respectively, and studies using the same dataset were merged into one column and we reported the overall results. All the included studies involved healthy adult participants, with 10 studies involving elderly participants. There were nine cross-sectional studies which compared mind-body experts to controls. Of these, five studies focused on yoga experts and four focused on TCC experts who had regularly practiced for at least three or more years. The remaining six were intervention studies with different durations of the interventions (range 6-24 weeks) and frequency (range 1-5 sessions per week). Of these, four were randomized controlled trials (RCTs), one was a controlled trial, and one was a "before-and-after" study with no control group. In addition, three studies applied TCC, two studies applied yoga, one applied BDJ, and one applied a multimodal intervention comprising cognitive training, TCC exercise, and group counseling. These intervention studies examined the brain health outcomes at baseline and the end of the intervention. Figure 1 shows a flow chart summarizing the study selection process recommen the PRISMA (preferred reporting items for systematic reviews and meta-analyses) gui Overall, 29 studies focused on evaluating mind-body exercise and brain plasticity usi met the inclusion criteria. It is worth noting that in 14 cases, authors published multiple on one actually performed study; therefore, we merged the related articles into one resulting in 15 actual studies. The characteristics of the included cross-sectional and in tion studies are shown in Tables 1 and 2, respectively, and studies using the same datas merged into one column and we reported the overall results. All the included studies in healthy adult participants, with 10 studies involving elderly participants. There we cross-sectional studies which compared mind-body experts to controls. Of these, five focused on yoga experts and four focused on TCC experts who had regularly practice least three or more years. The remaining six were intervention studies with different du of the interventions (range 6-24 weeks) and frequency (range 1-5 sessions per week). O four were randomized controlled trials (RCTs), one was a controlled trial, and one wa fore-and-after" study with no control group. In addition, three studies applied TCC, tw ies applied yoga, one applied BDJ, and one applied a multimodal intervention com cognitive training, TCC exercise, and group counseling. These intervention studies ex the brain health outcomes at baseline and the end of the intervention.   Years of yoga experience correlated with GMV in the insula, frontal operculum and OFC in the left hemisphere, and the number of hours of weekly practice correlated with GMV in SPL, PCC/precuneus, and hippocampus. Combination of postures and meditation contributed the most to the size of the hippocampus, precuneus/PCC, and S1/SPL, and combination of meditation and breathing exercises contributed the most to V1 volume. The insular GMV positively correlated with yoga experience and uniquely correlated with pain tolerance. Compared to WA group, the TCC group had less brain activation in the right lateral occipital cortex during the Stroop Word Color Task and presented less brain activation in the right frontal pole and SFG during the N-back task. The TCC group practiced Yang-style TCC 4 ± 1 times weekly for about 1.5 h each time for more than 6 years. WG mainly exercised by walking, not less than 5 times a week, with each time being no less than 1.5 h for more than 6 years. Two groups matched for age, handedness, and years of education.  The ALFF changes in the right MFG were positively correlated with changes in TMT and SWLS, and the baseline ALFF in the right MFG was negatively correlated with changes in TMT and SWLS. For IG, the ReHo changes in the left STG were positively correlated with changes in CFT, and changes in the right MTG were negatively correlated with changes in total PALT. For IG, changes for the mPFC-MTL FC were positively correlated with changes in CFT. putamen, and no significant differences were observed between the two groups. TCC increased fALFF in the dlPFC and BDJ increased fALFF in the mPFC in the slow-5 and low-frequency bands; both TCC and BDJ (at a lower threshold) significantly increased the rsFC between the bilateral hippocampus (ROI) and mPFC (ROI) and no significant difference between the two groups was observed; TCC showed a significant decrease in rsFC between the dlPFC (ROI) and the left SFG, ACC, and insula/putamen (at a lower threshold); the BDJ showed a significant decrease in rsFC between the dlPFC(ROI) and the left putamen and insula and ACC (at a lower threshold), and there was no significant difference between the two groups. TCC increased rsFC between the PCC (ROI) and the right putamen/caudate, while BDJ decreased rsFC between the mPFC (ROI) and right orbital prefrontal gyrus/putamen and left ACC; compared to BDJ, TCC increased rsFC between the mPFC (ROI) and right putamen/caudate.

Study Search and Characteristics
MQ and visual reproduction subscores were both associated with GMV increases in the putamen and hippocampus. Increased fALFF in both TCC and BDJ groups was positively associated with memory function improvement. The increases in rsFC between the bilateral hippocampus and mPFC were significantly associated with memory function improvement across all subjects. Mental control improvement was negatively associated with rsFC dlPFC-putamen changes across all subjects. Baseline visual reproduction subscore was negatively correlated with rsFC between the mPFC and right orbital prefrontal gyrus. TCC group with greater PFC activation increases in the switch condition presented greater reductions in task-switching errors from pre-to post-intervention.  (10 min

Quality Assessment
The quality assessment results for all the cross-sectional and intervention studies are shown in Tables 3 and 4, respectively, and studies using the same dataset were merged into one column and the overall results are reported. Most of the included cross-sectional studies were characterized as having low risk of bias (n = 8), with only one being categorized as having moderate risk of bias. Among these studies, the lack of a clear description of the measurement approach of the exercise conditions was a common problem (n = 7). The quality of the intervention studies most often ranged between good and excellent (n = 5), with only one being of fair quality. The most frequent missing point was the description and application of instructor or assessor blinding (n = 6).

Prefrontal Cortex
Structural and functional differences in sub-regions in the prefrontal cortex (PFC) were reported by various studies. Most prominently, the dorsolateral PFC (dlPFC) and the medial PFC (mPFC) were affected by mind-body exercises in the included studies. Structural changes in the PFC were reported in four cross-sectional studies. Compared with novices, TCC experts [36] had a greater cortical thickness (CT) in the right middle frontal sulcus (part of the dlPFC) and yoga experts [41] showed greater CT in the left prefrontal lobe cluster that included part of the middle and superior frontal gyri (MFG and SFG). Greater gray matter volume (GMV) was reported in the orbitofrontal cortex (OFC), right MFG, and the mPFC in yoga experts compared with controls, which were positively correlated with fewer cognitive failures [32,38].
Changes in task-induced brain activation were reported in three cross-sectional studies and one RCT study. For the cross-sectional studies, less activation was reported in the dlPFC during the Sternberg working memory task [43] and in the left SFG during the Affective Stroop task [33] in yoga experts relative to controls. Besides, less dlPFC activation was reported regarding the N-back task in the TCC experts compared with the water aerobics practitioners [45]. However, in the RCT study, there was an increased trend of left SFG activation following the TCC intervention during the modified task-switching fMRI paradigm. Intriguingly, the TCC group with greater PFC activation in the switch condition reported better cognitive function [57].
Two cross-sectional and two RCT studies reported changes in spontaneous brain neural activity. In the cross-sectional studies, compared with novices, the TCC experts had less regional homogeneity (ReHo) in the dlPFC and lower voxel-mirrored homotopic connectivity (VMHC) as well as greater fractional amplitude of low-frequency fluctuations (ALFF or fALFF) in the MFG [34,37,46]. Moreover, two RCT studies reported increased ALFF in the dlPFC after a 6-week multimodal intervention including TCC and after a 12-week TCC intervention [50,53]. Tao et al. also reported increased fALFF in the mPFC after a 12-week BDJ intervention [53].

Hippocampus/Medial Temporal Lobe
Two cross-sectional and three intervention studies reported structure changes in the hippocampus and medial temporal lobe (MTL) following mind-body exercise. Compared with novices, greater hippocampal GMV was observed in experienced yoga experts [32,43]. Both the 12-week TCC and BDJ interventions increased the GMV in the hippocampus/MTL, and the increased GMV was positively associated with improved memory abilities [55]. Increased hippocampal GMV (compared with controls) and GMD (compared with active sport and passive groups) were reported in two yoga intervention groups, respectively [48,59].

Lateral Temporal Lobe
In addition to the hippocampus/MTL, structural changes in the lateral temporal lobe, majorly constituting the superior temporal gyrus (STG) and the middle temporal gyrus (MTG), were reported in three cross-sectional and two RCT studies. Here, TCC experts displayed greater CT in the left STG [36] and yoga experts exhibited greater GMV in the left STG [32]. In addition, 8-week TCC intervention increased GMV in the left STG and the right MTG relative to controls and the aerobic exercise group, respectively [58]. Moreover, significantly changed ReHo in the left STG and MTG was reported following 6week multimodal intervention including TCC, which both correlated with better cognitive function [51].

Insula
One RCT and three cross-sectional studies reported on structural changes of the insula, including great CT and GMV. Increased CT in insula was observed in TCC experts [36] and yoga experts [38] compared with controls. Increased GMV in insula was also reported in yoga experts [32] and TCC and BDJ intervention groups [55]. In addition, the reported structural changes in the insula uniquely correlated with pain tolerance in yoga experts [38].

Cingulate Cortex
Only one RCT and two cross-sectional studies reported structural or functional changes in the cingulate cortex. Compared with controls, yoga experts [38] exhibited both greater CT and GMV in the cingulate cortex, and TCC experts [34] had less ReHo in the left anterior cingulate cortex (ACC). Meanwhile, 8-week TCC intervention increased GMV in the left precuneus/posterior cingulate cortex (PCC) [58].

Changes in Brain Functional Connectivity and Network
Two cross-sectional and three RCT studies reported differences or changes in brain functional connectivity. Compared to controls, a higher resting-state functional connectivity (rsFC) between the mPFC and right angular gyrus was reported in the yoga experts [42], whereas a weaker rsFC between the dlPFC and MFG was observed in the TCC experts [44]. For the RCT studies, significantly greater rsFC was observed between the left MFG and superior parietal lobule (SPL) after 8-week TCC intervention [58], as well as between the mPFC and MTL/hippocampus after 6-week multimodal intervention including TCC [49]. Another RCT study with different a priori regions of interest (ROIs) (dlPFC, hippocampus, and mPFC/PCC) using a seed-based rsFC analysis reported decreased rsFC between the dlPFC and left SFG and ACC after a 12-week TCC intervention, as well as decreased rsFC between the dlPFC and left putamen and insula compared to control groups [52]. At the same time, both the TCC and BDJ interventions increased the rsFC between the hippocampus and mPFC, and there was no significant difference between the two groups [54]. Furthermore, the BDJ intervention decreased the rsFC between the mPFC and OFC/putamen. Whereas, the TCC intervention increased the rsFC between the PCC and right putamen/caudate, and the baseline rsFC between the mPFC and OFC was negatively correlated with memory function [56].
Besides the aforementioned FC results, a few brain networks were reported. TCC experts had less fALFF in the bilateral frontoparietal network (main part of the cognitive control network, CCN) and the default mode network (DMN) [35]. Moreover, one crosssectional study used a group independent component analysis (gICA) to examine the different correlations of intrinsic connectivity network and found that there were significant differences in the DMN and sensory-motor network (SMN) of rsFC between the TCC and walking groups [47]. Froeliger et al. also reported that yoga experts exhibited greater rsFC within the dorsal attention network (DAN) [31].

Discussion
Herein, we systematically reviewed evidence of the effect of mind-body exercise on the structure and function of the brain to further understand the possible neural mechanisms underlying the health benefits. We found that both long-term and relatively short-term mind-body exercises induced structural or functional changes, mainly in the PFC, hippocampus/MTL, lateral temporal lobe, insula, and the cingulate cortex and within the CCN and DMN.

Prefrontal Cortex
The PFC is one of the later-mature brain regions and plays a critical role in a series of advanced cognitive activities [60]. There were inconsistent results regarding the structure of and task-related activation in the PFC between the included cross-sectional and intervention studies. First, four included cross-sectional studies reported structural changes in the PFC, including greater GMV and CT [32,36,38,41]. This implies that long-term mind-body exercise (lasting for at least 3 years of regular practice) can cause positive plasticity in the PFC structure, hence improving the cognitive performance of the individual. However, there were no changes in the PFC structure in the included intervention studies. We speculated that this could be because the relatively short mind-body intervention duration was insufficient for significant changes in the PFC region. Besides, three included cross-sectional studies reported less task-related activation in the PFC, whereas one RCT study reported more [33,43,45,57]. This might be attributed to two different theoretical frameworks that explain the neural plasticity induced by exercise, i.e., reduced neural activity could reflect improved neural processing efficiency, and increased neural activity could indicate more specialized and enhanced neural processing ability [57]. Based on this, we speculate that the relatively short-term mind-body intervention (e.g., the 12-week TCC) could have enhanced the neural processing ability of the PFC, resulting in the greater task-related activation. However, with the continuous increase in regular exercise (e.g., more than 3 years), the neural processing efficiency of the PFC could become more and more efficient, resulting in a subsequent decreased task-related activation in the PFC. Taken together, RCT studies of further prolonged duration should be conducted to clarify the effect of intervention duration on the structure and functional activation of the PFC region.
The findings of the spontaneous neural activity studies indicated that mind-body exercise could decrease the regional homogeneity and increase the functional specialization and intrinsic activity intensity of the PFC, improving the cognitive function [34,37,46,50,53]. Nevertheless, the sub-regions in the PFC affected by different types of mind-body exercises might not be the same. Particularly, TCC affected the spontaneous neural activity of the dlPFC, whereas BDJ affected the mPFC. This difference could be attributed to the different characteristics related to TCC and BDJ. Compared with TCC, which is much more complex and requires moving the trunk and four limbs by spatial navigation toward oneself [36], BDJ is much simpler and only involves eight simple fixed movements of arms with almost no movement of legs [61].

Hippocampus/Medial Temporal Lobe
Both cross-sectional and intervention studies reported the effect of mind-body exercise on structural changes in the hippocampus/MTL [32,43,48,55,59]. Notably, the positive changes in the hippocampus/MTL structure were related to the improvement in the cognitive function, particularly the memory function [32,55]. The hippocampus plays a crucial role in learning and memory processing [62], and the MTL, consisting of the hippocampus, adjacent parahippocampal, and entorhinal cortex, is also essential for memory processing [63], which shows that different types of mind-body exercises can improve the memory of an individual by changing the structure of the hippocampus/MTL. The effects of mindbody exercises on the hippocampus are similar to the findings of physical activity [64] and mindfulness meditation [65], suggesting that physical activity alone or meditation alone, as well as the combination of both, improves the brain structure related to memory. In addition, the intervention study [59] conducted by Garner et al. indicated that yoga experts have a significantly increased GMD in the hippocampus compared with a sport group (stretching and strengthening training), implying that mind-body exercises are more effective in improving the hippocampal structure than physical exercises alone. However, because of the methodological defects of the study (participants were not randomly allocated), the baseline hippocampal GMD in the yoga group was significantly lower than in the sport group. Therefore, the efficacy of mind-body exercise on hippocampal structure should be further explored using methodologically sound RCT studies.

Lateral Temporal Lobe
Several included studies reported structural and functional changes in the STG and MTG in the mind-body exercise groups compared to controls or individuals performing aerobic exercises [32,36,51,58]. The STG is thought to be sensitive to emotional information and plays an important role in a goal-directed behavior [66]. The MTG has extensive connectivity with the frontal parietal regions and is associated with semantic and memory processing [67]. A recent meta-analysis of RCT studies showed that physical activity significantly increases the structure of the lateral temporal lobe [68]. Our results revealed that the mind-body exercises equally or, in some cases, more effectively induced brain plasticity in the lateral temporal lobe, leading to more functional improvements than other forms of exercise, such as aerobics.

Insula and Cingulate Cortex
Mind-body exercises mainly induced structural changes in the insula and cingulate cortex [32,36,38,42,55]. The insula is believed to be related to interoceptive awareness [69] and the integration of sensory information to produce emotional experiences [70], as well as pain processing [71]. Meanwhile, the extensive connection between the cingulate cortex (particularly the ACC) and the PFC and insula is linked to emotion regulation and attention control [72,73]. Young et al. showed that changes in ACC activity reflect the mental processing of non-judgmental acceptance, one of the core components of meditation training [60]. Besides, the ACC is also correlated with pain procession [74]. Villemure et al. indicated that the insula GMV is associated with pain tolerance, with yogis using mental strategies based on relaxation, focusing on the pain, and non-judgmental acceptance to tolerate pain [38]. Combined with the authors' interpretations, we believe that those changes could be attributed to the inclusion of meditation training during the mind-body exercise, including present moment awareness and non-judgmental acceptance. Moreover, the repeated and prolonged use of these strategies causes positive plasticity in the related brain regions such as the insula and ACC. This further enhances the level of meditation of the individual and corresponding tolerance to noxious stimuli such as pain. Even so, relatively few included studies reported changes in these two regions; hence, more research is required to elucidate this issue.

Brain Networks
Mind-body exercise not only affects a specific brain region, but also the interconnection among different brain regions and various brain networks. This allows more meticulous access to the underlying neural mechanisms. The rsFC results of the included studies mainly reported changes between the PFC (two major sub-regions: dlPFC and mPFC) and other related regions. There were four a priori ROIs in the included rsFC analysis studies: the dlPFC, mPFC, PCC/precuneus, and the hippocampus/MTL. The dlPFC is a vital region of the CCN which plays an essential role in cognitive control processes [75]. The mPFC, PCC/precuneus, and MTL/hippocampus constitute the main nodes of the DMN which are functionally relevant to internal mental explorations and memory function [76,77].

Cognitive Control Network
We found weaker rsFC within the CCN of older TCC experts [44], as well as older TCC and BDJ intervention groups [52]. Similarly, a weaker fALFF in the CCN was reported in older TCC experts [35], and the decreased rsFC and fALFF within the CCN were correlated with better cognitive control and emotion regulation [35,44,52]. In contrast, there was an increased FC within the CCN after TCC intervention in young college students [58]. The conflicting results may be attributed to the different ages of the participants, to some extent. In the first three studies, all of the participants were elderly. Research shows that old people frequently over-recruit frontal neural resources to compensate for the disruption of the CCN and to overcome cognitive control deficits [78]. However, this hyperactivation usually represents a dysfunctional condition and cannot prevent a decline in cognitive function [79]. The decreased rsFC or fALFF in the CCN of the old mind-body exercise experts might suggest an increased efficiency of the cognitive control system and eliminate the need for compensatory hyperactivation of the network affected by mind-body exercise. However, Cui et al. assessed young college students whose executive function is still undergoing development, and the increased FC within the CCN after TCC exercise reflects the specialized and enhanced cognitive control. Furthermore, some studies showed that the beneficial effect of physical exercise on the CCN and cognitive function has been widely reported [80]. Similarly, mindfulness meditation enhances self-regulation through three components, i.e., attention control, emotion regulation, and self-awareness [81]. It is reasonable to imply that mind-body exercise, with both the physical activity and meditation components, induces more effective functional changes in the CCN to improve self-regulation in both young and older adults.

Default Mode Network
Consistently increased rsFC within the DMN was reported in yoga experts [42], a 6-week multimodal intervention including TCC [49], and 12-week TCC and BDJ interventions [54]. Besides, the increased rsFC within the DMN was associated with better general cognition and memory function [49,54]. These findings suggest that mind-body exercises, including TCC and BDJ, enhance cognitive function through the same neural mechanisms by increasing intrinsic connectivity between the mPFC and hippocampus/MTL within the DMN. On the contrary, the study of Liu et al. [56] selected different ROIs (mPFC and PCC) and different methods for multiple comparison correction based on the same dataset, and reported different results, including changes in the direction and related regions. This implies that both TCC and BDJ could improve memory function through other different neural circuits, e.g., decreased rsFC between the mPFC and OFC. With the different results taken into account, we speculate that the DMN is a complex brain network encompassing different vital brain regions (e.g., mPFC and PCC), and mind-body exercise could modulate various features of the DMN, through its multifaceted nature that combines movement, breathing, and attention. Besides, the characteristics among different types of mind-body exercise, such as TCC and BDJ, are not exactly the same, which could affect the DMN through different neural mechanisms. Furthermore, as mentioned above, the two articles [54,56] used different methods for multiple comparison correction based on the same dataset. The former applied the family-wise error (FWE) correction method, whereas the latter utilized the false discovery rate (FDR), which might have different influences on the results.

Limitations
Our study faced several limitations. Firstly, because of the wide variety of outcome measures (structure, neural activity, and functional brain connectivity), different study designs, and some studies reporting ROI analyses, there was no quantitative meta-analysis such as activation likelihood estimation (ALE) applied. Secondly, given the relatively few related studies, our systematic review included cross-sectional studies. Nevertheless, due to the inherent defects, the cross-sectional studies could not certainly attribute the group differences in brain structure and function to mind-body exercise. Although the age, gender, and years of education between groups were matched in the included studies, other confounders could still influence the results. For instance, people with a specific brain activation pattern may have the tendency to engage in mind-body exercises. Besides, there is no long-term follow-up study yet, which would be very meaningful to clarify the long-term health effects affected by mind-body exercise. Thirdly, there was a selection and reporting bias due to the focus on a priori defined ROIs rather than whole-brain analyses, particularly in the rsFC analysis studies. Therefore, it is impossible to fully uncover the impact of mind-body exercises on brain plasticity. Fourthly, we still cannot ascertain the role of the physical and mental components of mind-body exercise on brain structural and functional changes due to the lack of direct comparisons between these components. Finally, because of the few relevant studies, we did not include studies of clinical populations with cognitive issues, mood disorders, etc. As an important nonpharmacological therapy, mind-body exercises have shown promising effects on older adults with cognitive impairments [82] and persons with depression [83]. Therefore, to clarify the influence of mind-body exercise on the brain plasticity of clinical populations is an important direction for future studies.

Conclusions
In the present study, 15 studies which employed MRI to investigate the effects of mind-body exercise on brain plasticity were included. Our synthesis of results revealed that mind-body exercises induced changes in the structure, neural activity, or functional connectivity in various regions of the brain, primarily the PFC, hippocampus/MTL, lateral temporal lobe, insula, and the cingulate cortex, as well as brain networks, including the CCN and the DMN. These changes were associated with health benefits for healthy adults. However, due to the heterogeneity in the study designs, varied age of participants, exercise types, outcome measures, and a priori ROIs, there were some inconsistent results among the included studies. Therefore, findings of this study should be interpreted with caution. RCTs with rigorous designs and similar measured outcomes, as well as wholebrain analyses, should be conducted to unravel the precise underlying neural mechanisms of mind-body exercise.