Several approaches to rehabilitation of the hand following a stroke have emerged over the last two decades. These treatments, including repetitive task practice, robotically assisted rehabilitation and virtual rehabilitation activities, produce improvements in hand function, but have yet to reinstate function to pre-stroke levels— which likely depends on developing the therapies to impact cortical reorganization in a manner that favors or supports recovery. Understanding cortical reorganization that underlies the above interventions is therefore critical to inform how such therapies can be utilized and improved, and is the focus of the current investigation. Specifically, we compare neural reorganization elicited in stroke patients participating in two interventions: a hybrid of robot-assisted virtual reality rehabilitation training (RAVR), and a program of repetitive task practice training (RTP).
Ten chronic stroke subjects participated in eight, three-hour sessions of RAVR therapy. Another group of 9 stroke subjects participated in eight sessions of matched RTP therapy. Functional Magnetic Resonance Imagining (fMRI) data were acquired during paretic hand movement, before and after training. We compared the difference between groups and sessions (before and after training) in terms of BOLD intensity, laterality index of activation in sensorimotor areas, and the effective connectivity between ipsilesional motor cortex (iMC), contralesional motor cortex (cMC), ipsilesional primary somatosensory cortex (iS1), ventral premotor area (iPMv), and supplementary motor area (iSMA). Last, we analyzed the relationship between changes in fMRI data and functional improvement measured by the Jebsen Taylor Hand Function Test (JTHFT), in an attempt to identify how neurophysiological changes are related to motor improvement.
Subjects in both groups demonstrated motor recovery after training, but fMRI data revealed RAVR-specific changes in neural reorganization patterns. First, BOLD signal in multiple regions of interest was reduced and re-lateralized to the ipsilesional side. Second, these changes correlated with improvement in JTHFT scores. Our findings suggest that RAVR training may lead to different neurophysiological changes when compared to traditional therapy. This effect may be attributed to the influence that augmented visual and haptic feedback during RAVR training exerts over higher-order somatosensory and visuomotor areas.
Recovery of hand function is challenging after stroke. Empirical data suggest that treatment can be beneficial if it includes many repetitions of challenging and meaningful tasks (1–3). Several approaches to delivering high volume, intense, and salient rehabilitation activities have emerged over the last two decades. These treatments, which include repetitive task practice (RTP), robotically assisted rehabilitation, and virtual rehabilitation activities, produce improvements in hand function that exceed the standard of care in the US (4, 5).
Although a strong case has been made that virtual reality (VR) and robotics can be useful technologies for delivering challenging, meaningful, and mass practice, outcome studies investigating the true benefits of VR/robotics as compared to dose-matched RTP remain mixed (6, 7). For example, we have shown significant group-level improvement in hand and arm function of chronic stroke survivors in response to RTP and robot-assisted VR (RAVR) training to be similar for both groups (8), a finding that agrees with group-level effects in other clinical studies (9, 10). However, whether the underlying neural patterns of reorganization that are induced by the different training regimes are also similar remains unknown. This becomes important to understand because it may inform researchers and clinicians whether RAVR versus RTP may preferentially facilitate distinct neural patterns of reorganization. If so, then perhaps the therapy choice can be tailored more appropriately to individuals to elicit optimal benefits.
The goal of this study was to compare the effect of RAVR- and RTP-based interventions on neural pattern reorganization. Because neural reorganization likely reflects complex processes that include the formation of new connections and/or re-weighting of existing connections, the patterns that emerge are unlikely to be reliably captured using one proxy of activation. For example, while numerous studies have shown training-induced changes in the extent of brain activity, the results of those studies conflict in terms of whether the changes reflect an increase or a decrease in brain activity (11–15). Second, there seems to be a relationship between the pattern of reorganization (increase or decrease in ipsilesional somatosensory activation) and intactness of the hand knob area of M1 and its descending motor fibers (16), and a dependence on whether the lesion is cortical or subcortical (17). Connectivity measures may be a complementary way to understand neural reorganization patterns underlying stroke recovery (18) by providing additional information about dynamic network-level changes above and beyond what can be inferred from extent and laterality of activation (19, 20).
In this study, we therefore characterize the pattern of neural reorganization using multiple measures that included the magnitude of change in brain activation, the extent of activation, the re-lateralization of brain activation in a set of homologous interhemispheric regions of interest, and interactions between multiple regions of interest based on measures of functional and effective connectivity. To our knowledge, this is the first study to characterize brain reorganization at the ROI and network interaction level with multiple functional magnetic resonance imaging (fMRI) measures before and after RAVR and RTP training. In order to delineate the relevance of brain reorganization after training, we also correlated the brain activation outcomes with clinical outcome measures.
We hypothesized that both treatments might have similar effects on the magnitude and laterality of activation in a given region of interest. However, because RAVR training provides a training environment that is enriched and augmented with visual and haptic feedback, we expected that the functional and effective connectivity between motor/premotor cortices and visuomotor areas like the superior parietal lobule may show stronger effects in the RAVR group, as compared to the RTP-based training group (21–25). We propose that identifying the neurophysiologic correlates of behavioral motor function improvement might allow strategic refinement of existing training approaches and the development of individually tailored interventions. […]