For this scoping review, we use the methodological framework developed by Arksey and O'Malley (26). Arksey and O'Malley suggest that there are five stages to a scoping review: (1) identifying the research question; (2) identifying relevant studies; (3) selecting studies; (4) charting the data; (5) collating, summarizing, and reporting the results.

In this review, we will concentrate on the alteration of brain networks measured by fMRI after rTMS over M1 in stroke patients with movement disorders, and understand the brain mechanisms of rTMS. The scoping review may provide guidance for rTMS use as a therapeutic tool in movement disorders after stroke.

PubMed, Embase, Web of Science, Embase, WanFang Chinese database, and ZhiWang Chinese database were accessed and searched from inception until December 2022. In addition, reference lists of relevant articles were screened to identify key articles that had been missed. The following search terms were entered using the Boolean operators AND/OR: “stroke”; “cerebrovascular accident”; “hemorrhagic stroke”; “ischemic stroke”; “cerebral stroke”; “brain vascular accident”; “transcranial magnetic stimulation”; “transcranial magnetic stimulations”; “repetitive transcranial magnetic stimulations”; “repetitive transcranial magnetic stimulation”; “transcranial magnetic stimulation, repetitive”; “rTMS”; “TMS”; “functional magnetic resonance imaging”; “fMRI”; “Functional MRI”; “fMRI”; “MRI”; “hemiplegia”; “motor deficit”; “dyskinesia”; “movement disorder”; and “motor dysfunction”; “motor deficits”; “movement disorders”; “motor dysfunction.” The search method was modified for each database following the example of PubMed (Table 1).

Databases searches identified 711 articles. After screening titles, abstracts, and full texts for eligibility, nine articles were included. Four studies were conducted in China, two in Japan, two in Germany, and one in Turkey. Five studies were RCTs, three were pre-post-test trials, and one was non-randomized controlled trials. A summary of the characteristics of the included papers is summarized in Table 2. Figure 3 shows the findings regarding the effects of rTMS on fMRI and motor performance in stroke survivors.

Table 3 displayed the outcomes of the quality evaluation of each study. Five studies were rated as having good quality and four as fair according to the Downs and Black criteria. The experimental hypothesis, the primary clinical features of the patients, the intervention techniques, and the key findings were all well reported in the papers under review. However, the studies met the requirements for the reporting section, but none provided the principal confounders in the groups and reported adverse events of intervention. Six studies failed to adhere to the requirement for blinded outcome assessment and participants blinded to treatment.

By measuring the effects of the HF-rTMS (10 Hz) applied in the ipsilesional M1 on BOLD signals, Du et al. (32) found that HF-rTMS increased BOLD signal in the ipsilesional motor areas. Motor performance was observed in conjunction with fMRI changes. The effects of HF-rTMS (10 Hz) and LF-rTMS (1 Hz) were contrasted by Juan et al. (31). When applied the 10 Hz in the ipsilesional M1 they found an increase in resting-state functional connectivity (FC) between the bilateral M1, which has a positive relationship with motor performance. Guo et al. (33) assessed the effect of functional reorganization following rTMS in stroke survivors as well as the differences between HF-rTMS and LF-rTMS. In the HF-rTMS group, they found higher FC between ipsilesional M1 and contralesional premotor area (PMA), which suggests that HF-rTMS can increase the FC of the ipsilesional motor network and enhance the motor functions. Significant functional connectivity changes after rTMS are summarized in Figure 4. Significant activations of the brain areas after rTMS are summarized in Figure 5.

Du et al. (32) show LF-rTMS reduced brain excitability and fMRI activation in contralesional motor region. These changes were accompanied by improved motor activity. Juan et al. (31) reported 1 Hz rTMS applied to contralesional M1 resulted in enhanced FC between contralesional M1 and ipsilesional SMA. Motor improvement evaluated using the FMA and MRC scale was significantly enhanced in the real rTMS group compared with the sham group. Gottlieb et al. (37) found that connectivity to the left angular gyrus increased after LF-rTMS over M1. The modified Ashworth scale (MAS)score was reduced and the FMA score improved in the LF-rTMS group, suggesting motor improvement. Significant functional connectivity changes after rTMS are summarized in Figure 4. Significant activations of the brain areas after rTMS are summarized in Figure 5.

Chen et al. (34) found that inhibitory-facilitatory rTMS treatment induced greater increases in FC between multiple brain regions in comparison to the other groups using single-course or sham rTMS, resulting in great improvements in motor function. Significant functional connectivity changes after rTMS are summarized in Figure 4. Significant activations of the brain areas after rTMS are summarized in Figure 5.

Ueda et al. (30) found significant FC changes in bilateral cerebral hemispheres after LF-rTMS during task-fMRI. According to Grefkes et al. (36), 1-Hz rTMS has suppressive effects on ipsilesional M1 and facilitates more efficient motor processing in the contralesional hemisphere, as shown by the improved coupling of SMA and M1. Wanni et al. (35) show significant activations were seen in the ipsilesional PMA, M1, and thalamic-cortical regions with the paretic hand movements after rTMS. However, significant activations in the contralesional primary somatosensory cortex (S1), superior parietal cortex, and bilateral cerebellum with unaffected hand movements after the intervention were observed. There was a considerable improvement in FMA and WMFT from pre- to post-rTMS. Tosun et al. (38) reported that greater activation of the affected M1 was observed during the movements of the paretic hand in most patients of the TMS group and TMS + NMES group. Significant functional connectivity changes after rTMS are summarized in Figure 4. Significant activations of the brain areas after rTMS are summarized in Figure 5.

There is growing evidence to suggest an association of rTMS with the induction of neural plasticity to promote stroke recovery (47). “Neural plasticity” refers to the ability of modification of the nervous system in response to suffering and to adapt following trauma by remodeling its structure, functions, or connections (48). The processes of neural plasticity include altered excitability of neuronal circuits, reorganization by using redundant connections, and formation of new functional connections, which may partly compensate for the lost function. Based on neuroplasticity research, motor function recovery after stroke is related with spontaneous neuroplasticity changes and rTMS-induced plasticity (49). During the post-stroke recovery stage, the spontaneous remodeling neural networks were accompanied by functional recovery (50, 51). It is possible that these spontaneous changes in neuroplasticity are associated with pathophysiological mechanisms, such as salvage ischemic penumbra by revascularization, the release of neurotropic factors and neurotransmitters (52). However, these spontaneous neuroplasticity changes have a limited effect on stroke recovery (53). After stroke, rehabilitation can promote dynamic processes due to increase effective neuronal information input, promote neural repair and functional compensation (54). rTMS was identified to be a successful rehabilitation method for improving stroke recovery by promoting neuroplasticity. The application of fMRI provides a good understanding of the mechanisms involved. Recently, plasticity-induced rTMS was combined with fMRI to map plastic aftereffects (55). The neuroplastic mechanisms of motor function recovery after stroke can see the Figure 6.

Neural plasticity takes place at a very early stage after stroke, lasts for some time, and involves brain regions remote from the lesioned site. Schulz et al. (56) proved that the corticospinal tract fibers not only originating from the M1 and the ipsilesional PMA and SMA contributed to motor function after stroke. Pathologically, damage to the brain related to motor function, such as M1, PMA, SMA, and S1 (57), contribute to hemiplegia following stroke. Activation in M1 of the affected hemisphere is reduced and the activation is relocated toward the PMA and SMA during movements of the affected hand after stroke (58, 59). By comparing a shift in sensorimotor cortical excitability in both hemispheres, it was found the activation of the contralesional sensorimotor cortex increases accompanying movements of the paretic hand in most post-stroke (60). After a stroke, there are alterations in neural activity in bilateral cerebral hemispheres, which can be beneficial, but can also contribute for maladaptive recovery (61). Conventionally, LF-rTMS or HF-rTMS is effective in improving motor functions by rebalancing hemispheres' the excitability (32).

There have been many studies on local and global functional connections analysis. The most consistent conclusion is that the interhemispheric connection between regions involved in motor function has changed significantly, and the degree of connectivity is related to motor functions. In the early post-stroke period, motor network resting-state connections progressively decrease (62). fMRI studies demonstrated that the patients with reduced the functional connections between bilateral M1 are more severe in motor performance and with increased functional connection between the ipsilesional M1 and the contralesional thalamus, SMA and middle frontal gyrus at the stage of acute stroke was beneficial to the motor recovery (63). The decreased interhemispheric functional connections between homologous motor areas were associated with the degree of motor function in the acute phase (64), While the enhanced functional connections had a positive correlation with the spontaneous recovery of motor function in the weeks to months after stroke (62). The patients with good motor function recovery showed functional connections between homologous motor regions came back to normal levels in the stable phase after stroke. However, the patients with poor motor function recovery found that functional connections remained low (65, 66). Therefore, it is necessary to seek alternative approaches that strengthen the natural plasticity of the sensorimotor system, which is particularly effective in enhancing motor improvement. In rTMS studies, functional connectivity (FC) evaluation is of value because it enables the evaluation of rTMS effects.

The interhemispheric competition, assuming the presence of mutual, balanced inhibition between the hemispheres in the healthy people, is found to be altered after stroke (12). Damage to one hemisphere in stroke results not only from neuronal loss within the affected hemisphere but also from the downregulation of remaining neurons within the affected hemisphere resulting in increased inhibition of the affected hemisphere by the unaffected hemisphere (12). That is to say that both the stroke itself and the excessively high interhemispheric inhibition from the contralesional hemisphere result in a double impairment of the ipsilesional cortex. Based on TMS therapy (30), ipsilesional motor areas play an active and vital role in stroke recovery. Therefore, the downregulation of contralesional cortical excitability may be helpful for enhancing motor recovery. Researchers and clinicians have used rTMS to restore the interhemispheric balance to gain motor improvement by either enhancing the activtion of the lesioned hemisphere (38) or inhibiting the healthy hemisphere.

However, it has been found that HF-rTMS given over the M1 of the unaffected side is more effective than LF-rTMS when patients with significant harm to the affected side (67). The significance of the contralesional hemisphere in motor recovery following stroke has been thoroughly investigated in the past few years. According to Ueda et al. (30), the cortical activity of the ipsilesional and the contralesional motor areas changes synchronously during movements of the paralyzed hand after rTMS in stroke survivors with moderate disability. fMRI research found the contralesional PMA was significantly activated during the movement of the paralyzed hand in stroke survivors (68). A meta-analysis (69) shows that consistently activated regions involved the contralesional M1, the bilateral PMA, and SMA in stroke patients compared to healthy controls. The extent of this activation is likely to be influenced by the size and location of the injured region, being greatest in stroke survivors with the greatest impairment. From what has been discussed above, the enhanced activation of the PMA, SMA, and the contralesional hemisphere make up for the lost function in stroke survivors with severe motor impairment. Thus, the vicariation model (12) was presented, which assumes activity in residual networks helps stroke people recover function lost by damaged areas. Similarly, rTMS can support residual motor function following stroke by inducing positive compensatory effects of contralesional mirror motor regions (67).

The vicariation and interhemispheric competition are the two models of functional recovery, which hold opposite views and represent different neuromodulatory treatments. According to interhemispheric competition, stroke recovery would be facilitated by the downregulation of the contralesional hemisphere since it would free the injured hemisphere from aberrant inhibition by the ipsilesional hemisphere. The vicariation model, however, contends that such a tactic would impede compensating activity in the contralesional hemisphere. However, neither the interhemispheric competition model nor the vicariation model could account for all stroke recovery. Therefore, a novel theory called the bimodal balance-recovery model (12) was put out to account for the various contributions made by the contralesional hemisphere to post-stroke motor recovery. The biphasic recovery model of the transcallosal suppression model and the compensatory model may be the neurophysiological basis of functional recovery after stroke. Due to the high level of structural preservation, the interhemispheric competition model is more helpful for stroke survivors with moderate motor impairment. In contrast, patients with little structural reserve may find the vicariation model more helpful in predicting a recovery. Excessive activation in contralesional M1 and nonprimary motor areas can be seen in the early phase after stroke, suggesting recruiting of these brain regions after the cerebral vascular accident (70). This is in accordance with the result that shows overactivity in the ipsilesional and contralesional PMC, SMA, parietal cortex, and contralesional M1 during movements of the hemiplegic hand (71). In a word, bilateral cerebral hemispheres may both participate in the restoration of motor functions brought about by rTMS intervention. It is necessary to conduct more researches to examine whether rTMS therapy can benefit more stroke patients using individualized stimulation techniques.

It has been shown that the long-lasting after-effects of plasticity-inducing rTMS can easily impact human behavior. The combination of rTMS with fMRI provides a unique opportunity to elucidate the mechanistic basis for such behavioral effects. Thus, this combination allows us to gain insight into the local and distant effects of different interventions and provides a means of addressing changes in functional connectivity that underlie potential behavioral effects. The currently accepted strategy to promote the recovery of motor function after stroke is to either apply HF-rTMS (31–33) to the M1 region of the ipsilesional hemisphere or apply LF-rTMS (30–33, 35–38) to the M1 region of the contralesional hemisphere. In clinical practice, the combination of HF-rTMS and LF-rTMS is relatively rare, but some studies have shown that the combination of the two is feasible and safe, and the therapeutic effect is better than the single application (90). In comparison to sham stimulation or single-course rTMS, coupled inhibitory-facilitatory rTMS significantly improved motor function.

rTMS is a noninvasive neuromodulation technique whose variability in the stimulation intensity, frequency, duration, and target location influence modulatory effects (91). The HF-rTMS stimulation and the LF-rTMS stimulation both can improve motor function by rebalancing motor cortex excitability and regulating connectivity in patients after stroke (32). Du et al. (25) found that rTMS could dramatically enhance motor function, and this enhancement was strongly connected with changes in the motor cortex excitability, and LF-rTMS may have more profound effects than HF-rTMS. The HF-rTMS group offers better advantages for the functional connection recombination of the motor network on the ipsilesional brain, bringing greater benefits for the recovery of motor disorders (33). A meta-analysis revealed that HF-rTMS is marginally more efficient than LF-rTMS (92). Another meta-analysis, however, revealed a different outcome (93). Chen et al. (34) also found that the combined application of low-frequency and high-frequency rTMS had a synergistic effect on improving motor function and cortical excitability of patients. At present, the optimal effective frequency of rTMS is still uncertain. The 2014 European Guidelines for the treatment of rTMS indicate that both LF-rTMS or HF-rTMS can be used to restore motor function after stroke (94). The application of LF-rTMS shows level A evidence for motor stroke in the post-acute phase, as well as level C evidence in the chronic phase. While the application of HF-rTMS shows level A evidence in the post-acute phase (15).

The summary of current evidence suggests that rTMS may change neural plasticity, which was associated with movement improvement after stroke. This scoping review revealed some important findings. First, in the process of searching the literature, we discovered most rTMS studies only focused on the functional connectivity changes in motor network, and few studies have involved deep brain areas. However, many movement diseases, including Parkinson's disease and stroke, are associated with the damage of the thalamus, putamen, cerebellum, and other subcortical regions (95, 96). As the included studies in the review seldom focused on whole-brain connectivity, but rather specific regions of interest, more researches should pay attention to the whole-brain connectivity. In the future, the regulation mechanism of rTMS on deep brain regions can be explored by analyzing the changes in functional connectivity in cortical and subcortical regions, which will provide a reference for the treatment of these diseases. Second, it should be highlighted that the number of studies applied fMRI to investigate the aftereffects of rTMS on the functional brain network following stroke is relatively limited, and the effects of rTMS, in particular, have seldom been mapped with task-based fMRI. Future research should concentrate on mapping the effects of rTMS on task-related activity and connectivity (97). Third, the last three decades have seen the progress of neuroimaging technology, which allow people to examine neuroplasticity noninvasively. However, each of them has its merits and demerits. Multimodal neuroimaging technology can merge information and overcome some of the limitations of the stand-alone neuroimaging method. Multimodal fusion technology permits more granular inspection of the underlying mechanisms of plastic after-effects of rTMS protocols. However, studies using multimodal fusion technology to explore neural mechanisms underlying rTMS intervention are relatively rare. Therefore, incorporating multimodal neuroimaging technology into clinical trials will help expand our knowledge of neural mechanisms underlying rTMS intervention and ultimately tailor the therapeutic use of rTMS. Fourth, fMRI studies have confirmed the role of bilateral M1 areas after a stroke, indicating functional connectivity between the areas and functional connectivity between M1 and other brain areas is associated with motor recovery (63). Although M1, as an important brain area responsible for planning and performing motor functions, is the ideal stimulated brain area, stimulating rather the PMA has many benefits (98). When brain damage is severe, PMA of the contralesional hemispheres exerts an excitatory effect on the M1 affected hemisphere. When brain lesions are small, PMA of the contralesional hemispheres exerts an inhibitory effect (99). SMA is related to complicated motions and the programming of these complicated motions, which is a crucial issue to address in motor skill recovery. In addition, for patients with motor dysfunction accompanied by cognitive impairment or depression, the left dorsolateral frontal lobe can be selected as the stimulation target (100), and the recovery of motor function can be promoted by improving the cognitive and depression symptom. To date, studies have focused mostly on the motor cortex, whereas other parts of the brain are still underrepresented. Future researches may deepen the topic to concomitantly evaluate different targets areas. Lastly, none of the studies perform fMRI during rTMS, and a fMRI scan was usually performed at baseline and after rTMS that lasted several weeks. This approach is limited in its sensitivity to capture post-stimulation effects, and is likely to fail to map the immediate consequences of the stimulation. For solving this problem, simultaneous rTMS-fMRI, in which rTMS is applied during fMRI scan, permits research into the immediate effects of rTMS and the underlying processes of rTMS-mediated neuronal regulation.