ABSTRACT
Background Dystonia is a network level brain disorder with limited understanding of interactions between functional networks. Prior studies have demonstrated abnormalities both between and within functional networks. However, these findings were based on a hypothesis driven region of interest (ROI) approach or used resting state fMRI to extrapolate network differences in task-specific dystonia called writer’s cramp (WC). As a result, it remains unclear if these same networks would be active during the actual performance of a dystonic task. It’s also unclear from prior studies which networks play a role in the primary disease versus a compensatory region. Lastly, neuroimaging studies have used sequencing tapping task with equal motor performance to study networks changes due to concern that differences in motor performance during the dystonic writing task may confound the findings.
Objective The aim of this research study was: 1) to use a data driven approach on a task-specific focal dystonia to understand network level interactions during task-fMRI, 2) to determine if differences in observed network level could distinguish between brain regions of primary vs compensatory change in task-specific dystonia, and 3) if sequence tapping task captured similar level network changes as the dystonic writing task.
Methods We used the data-driven independent component analysis to compare functional network correlation with task time course and network connectivity during 3 motor tasks of writing, sequence tapping and finger flexion-extension.
Results Our analysis shows that in WC there is significantly decreased correlation of the basal ganglia and increased correlation of the orbitofrontal network to the writing task time course. Analysis of network connectivity showed significantly decreased connectivity of the basal ganglia to the left sensorimotor network specifically during the writing task in WC patients. We also showed significantly increased correlation of the cerebellum to the orbitofrontal and parietal networks.
Conclusions Collectively, our analysis supports a global network disorders in which the basal ganglia plays a primary role in the disease mechanism while the cerebellum plays a role as a compensatory network, partially through increased connectivity to the orbitofrontal network, a region important in motor planning. Our analysis also shows that the sequence tapping task does not capture the same network level dysfunction as the primary dystonic task. Our study is the first to apply a purely data-driven approach to understand network interactions in WC dystonia patients during task performance.
INTRODUCTION
Dystonia is a disabling brain disorder characterized by abnormal and painful postures that can manifest in the whole body (generalized dystonia) or isolated in a single limb (focal dystonia) (Albanese et al. 2013). Isolated focal dystonia is the most common type manifesting in adult patients (Warner et al. 2000; Ortiz et al. 2018). Among the focal dystonias, writer’s cramp (WC) is the most common task-specific focal dystonia with symptoms occurring specifically during the task of writing (Albanese et al. 2013; Warner et al. 2000). WC dystonia is an ideal subtype to identify the neural underpinnings of dystonia as its task-specific feature allows one to study the brain function in an “on” and “off”dystonia inducing brain state (Hallett 2006; Goldman 2015).
Neuroimaging studies of focal dystonia patients have revealed network level brain disorder with abnormal activity and within network connectivity in the parietal cortex and premotor cortex as well as subcortical regions of basal ganglia and cerebellum (Gallea et al. 2018; Burciu et al. 2017; Norris et al. 2020). These findings are based on a hypothesis driven approach called region of interest (ROI) analysis. This approach requires a-priori selection of brain regions for analysis. ROI analysis also generally assumes a canonical hemodynamic response function across all subjects, tasks and brain regions to estimate brain activity while using general linear modeling approaches.
In contrast, independent component analysis (ICA) does not require selection of a-priori brain regions and makes no assumptions about the hemodynamic response function (Calhoun et al. 2001; Ray et al. 2013; Bhaganagarapu, Jackson, and Abbott 2013; McKeown et al. 1998). ICA analysis, therefore, offers a data driven approach to understand brain dysfunction. Two prior studies of WC patients used ICA analysis to understand brain dysfunction in an “off” disease state using resting state fMRIs. These studies demonstrated abnormal connectivity within the sensorimotor and premotor-parietal networks, as well as abnormal between network connectivity of cortical to subcortical regions including the parieto-basal ganglia, and basal ganglia-cerebellar networks (Mantel et al. 2018; Battistella and Simonyan 2019). Whether a similar analysis of WC dystonia patients in an “on” disease state, while performing the dystonic task in the fMRI would capture the same functional network dysfunction remains unknown. Understanding network level dysfunction in both the on and off disease state of dystonia may offer clues to the disease mechanism and can help distinguish brain changes due to primary disease from compensatory mechanisms. Of particular interest is the difference in role between the basal ganglia and cerebellum. These two subcortical networks have long been demonstrated to play an important role in human and animal models of dystonia (Bonsi et al. 2011; Argyelan et al. 2009; Shakkottai et al. 2017; Peterson, Sejnowski, and Poizner 2010; Jinnah, Neychev, and Hess 2017). But it remains unclear if both play a role in the primary disease mechanism or if one is the primary driver and other a compensatory region.
A theoretical concern in neuroimaging studies of dystonia patients is that functional network changes observed during a dystonic task may reflect differences due to the performance of the movement rather than the disease. As a result, some fMRI studies in WC patients have used non-dystonic motor tasks with demonstrated equal motor performance between dystonia and matched healthy volunteers (HV). A common task with equal motor performance previously used to study WC patients is sequence tapping (Zeuner et al. 2015). However, a direct comparison of functional network changes during writing and sequence tapping tasks has not been performed. As a result, it remains unclear how sequence tapping can serve as a writing task substitute for studying dystonia disease mechanism. More importantly, it remains unclear if network level changes seen during the dystonic task are reflective of the primary disease or differences are due to motor performance.
The aim of this study was to use ICA analysis to understand changes in functional network connectivity in WC patients compared to HV during the dystonic task (writing), non-dystonic task of equal motor performance (sequence tapping) and a simple motor task (finger flexion-extension). To achieve these aims, we first investigated how functional networks are correlated to the time course of each motor task. We then evaluated how functional networks are correlated to each other across the 3 motor tasks.
MATERIALS AND METHODS
Participants
The study was approved by the Duke Institutional Review Board. The study was open to all isolated right hand writer’s cramp dystonia patients. All enrolled patients were more than 3 months from last botulinum toxin injections and were not on oral trihexyphenidyl, a symptomatic treatment for dystonia. All enrolled WC patients were diagnosed with isolated right-hand dystonia by a Movement Disorder Specialist through standard medical history and neurological examination. Aged matched healthy volunteers (HV) who were right hand dominant with no structural brain disorders or psychiatric diseases were also recruited for the study. All enrolled participants gave informed consent and had no contraindication to MRI. A total of 14 WC patients and 19 HV were consented and completed behavioral assessments and fMRI scans. After data collection, 1 subject’s data was excluded due to misdiagnosis of essential tremor. Additional data from 1 WC and 7 HV participants were excluded due to extensive head motion in the fMRI. All data analysis in this study is based on 12 WC and 12 HVs.
Behavioral analysis
At the initial screening visit, all study participants were asked to write a single sentence 10 times in MovAlyzer (Neuroscript) software and 4 times on OneNote (Microsoft) software on a digital tablet. Participants were then asked to perform a 20-minute writing exercise to induce writing dystonia followed by repeat of the single sentence writing in the same softwares. Writing samples performed at baseline and after 20-minute writing exercise were then analyzed for the number of dysfluent events during writing using the MovAlyzer software. Writing samples were also analyzed for legibility of sentences using writing to text conversion feature in OneNote. The number of words correctly converted from subject’s writing to text in 4 sentences were calculated as a percentage of total words. A previous study by our group showed that these writing software measures were comparable to established dystonia rating scales and show significant differences between HV and WC participants, thus allowing us to use these objective measures for clinical characterization of WC dystonia participants (Bukhari-Parlakturk et al. 2021).
fMRI research design
All participants completed a single fMRI scan during which they were presented with visual instructions to perform 3 motor tasks on an MRI compatible digital tablet using a back projection screen. The 3 motor tasks were 1) writing a single sentence 2) tapping a 4-digit sequence 3) finger flexion and extension. During the writing task, participants were shown a sentence to copy. This task was used to activate functional networks of WC during their core dystonia task. During the tapping task, participants were shown a sequence of 4 numbers every 4 seconds and asked to type it out on a 4-button keypad. The third task was the finger flexion-extension task in which the subjects were presented with alternating open or close of hand image for 0.4 and seconds respectively and they had to match their hand movements with the image. These two motor tasks were used to explore functional networks during non-dystonic motor tasks with demonstrated equal performance between the two groups for the sequence tapping task. Each task was presented for 20 seconds in a block design alternated with a 16 second rest. Twelve blocks were presented in a single run lasting 7 minutes and 26 seconds with each motor task was performed once. The task presentation timing was kept the same across tasks and subjects (Fig. 1). Performance of the writing and finger flexion-extension tasks were recorded using the digital software tablets while the sequence tapping tasks were recorded using the button box. An eye tracking device was also used to monitor subjects in the MRI.
Data acquisition and preprocessing
All participants underwent structural and functional MRI in a 3 Tesla GE scanner. Structural T1-weighted images were acquired using echo-planar imaging and the following parameters: voxel size = 1×1×1.3 mm3, TR = 6.836 s; TE = 2.976 s; FOV= 25.6mm2, bandwidth 41.67 Hz/Pixel). Functional echo-planar images were acquired while participants performed one of the three motor tasks using the following parameters: voxel size = 3.5 × 3.5 × 4.0 mm3, TR = 2 s, TE = 30 ms, flip angle = 90°, FOV = 22 cm, bandwidth = 250 Hz/Pixel, interleaved slice order with whole brain coverage). Subjects were asked to minimize their movement and any subject scans with excess movement or noise were not included in this data set as detailed above. The Cigal software was used to align each subject’s movement to a reference frame to monitor movement in real time (Voyvodic 1999). All fMRI images were preprocessed using fMRIPrep after discarding first 5 brain volumes (Esteban et al. 2019). fMRIPrep output was used to evaluate the scan quality.
Group independent component analysis
A group independent component analysis (GICA) was performed on all fMRI scans using the GIFT v3.0 software on Matlab R2018b (Calhoun et al. 2001). With GICA it is possible to identify group level functional networks and then project them to individual fMRI data to identify functional networks at the individual level. In addition, as the timing of the three different tasks were identical, we were able to apply GICA across all tasks to identify functional networks that are common across tasks. GICA was applied to decompose the group fMRI dataset into 20 and 30 functional networks. The InfoMax algorithm was used to identify spatially independent components at the group level. These spatially independent components along with their time courses are defined as functional networks. The algorithm first created group spatial maps and then used these maps and the original fMRIs at the individual subject level, the algorithm constructed the time courses of the spatial maps at the individual subject level (Sui et al. 2009). Components were then identified and labeled by comparing them against independent component network (ICN) atlas software (Kozák et al. 2017) and the Harvard Oxford MNI atlas (Desikan et al. 2006).
Components that primarily corresponded to CSF, non-brain tissue and those related to motion were removed as artifacts and were excluded from further analysis. The remaining components in this study showed a high overlap with previously identified ICN networks and anatomic regions indicating their independent validation and biological relevance. Functional networks previously shown to play a role in patients with dystonia were then selected for further analysis (Battistella and Simonyan 2019; Mantel et al. 2018; Huang et al. 2017; Baltazar et al. 2020) and they are listed in Table 1. The components denoted as the sensorimotor cortex (SM) showed bilateral activation in the motor and sensory cortical regions, however, were denoted left and right based on which side showed more activation.
Analysis of component correlation to task time course
To determine which functional networks correlated with each of the three motor task-time courses, the task presentation time course was convolved with the hemodynamic response function to construct task model time course. The model time course was then correlated with the time series of functional networks that were estimated by GICA. These correlations between the task model time course and each subject’s functional network timeseries were compared between the HV and WC groups. A t-test was performed to determine if there was a significant difference in motor task correlation to the functional networks between the two groups. Note that this is the only instance in our analyses where we apply the hemodynamic response function to validate the relevance of the functional networks with respect to the three tasks performed. By this step we show ICA is able to uncover functional networks that are closely related to the task in a purely data-driven manner.
Functional Network Connectivity
Functional networks that showed highly positive or highly negative correlations with the writing task model time course and those known to play a key role in patients with dystonia were selected for functional network connectivity analysis (Battistella and Simonyan 2019; Mantel et al. 2018; Huang et al. 2017; Baltazar et al. 2020). FNC measures was determined by computing pairwise correlations between selected functional networks in each subject (Jafri et al. 2008). Subject level FNCs were compared between the HV and WC groups using a two-sample t-test to determine group differences.
Statistical Analysis
Group level differences was tested using t-test. Gender differences between the two groups was tested by Fisher exact test. All correlational analyses were performed using Pearson’s correlation test with correlation coefficient (R-value) and significance (p-value) reported. For FNC analysis, Pearson’s correlation coefficient was transformed to a fisher z-score. A minimum p<0.05 was considered statistically significant.
RESULTS
Participants
We show results from 12 WC dystonia patients and 12 age-matched HV that met quality control standards and their demographics, motion statistics and writing behavioral measures are presented in Table 1. There were no significant HV versus WC group differences in age or gender. WC patients ranged in disease duration from 5 to 46 years with a mean of 9.6 years (14.5). There were no group differences in head motion parameters during the 3 task runs in the fMRI. Writing samples from WC patients, as expected, showed significantly higher dysfluent events and decreased legibility compared to HV.
Functional Networks
ICA analysis was used to compare 3 motor tasks in WC and age-matched HV. We ran GICA with two different numbers of components: 20 and 30. Spatial maps of the functional networks obtained from the 20-component run were mixtures of multiple brain regions and did not align with previously established networks. This led us to use functional networks from the 30-component run. Visual inspection of the functional networks from the 30-component run revealed that 11 functional networks were comparable to previously reported functional networks using resting-state fMRI (Smith et al. 2009)and task-fMRI (Laird et al. 2011; Ray et al. 2013; Kozák et al. 2017). Specifically, we report functional networks for sensorimotor (Laird ICN 8) visual (ICN 11), cerebellum (ICN 14), basal ganglia (ICN 4), left and right frontoparietal (ICN 18 and 15 respectively), insula (ICN 4), parietal (ICN 7), orbitofrontal (ICN 2), and default mode network (ICN 13). The subcortical networks of basal ganglia, cerebellum and visual networks reported in this task-fMRI analysis were comparable to the prior resting state fMRI study of WC patients (Mantel et al. 2018). However, our task-fMRI study showed different composition of cortical networks compared to Mantel and colleagues (Fig. 2 and Table 2).
Specifically, the sensorimotor networks did not divide into a dorsal and ventral network but rather by laterality of greater left vs greater right activated SM network. The parietal network did not include a premotor cluster but instead included a small cluster of precentral and postcentral gyri. The right frontoparietal and salience networks were reported in ICA analysis of resting state fMRI from blepharospasm patients, a type of focal dystonia (Huang et al. 2017). Both left and right frontoparietal networks were observed in our analysis. We also observed partial salience network with activation of bilateral insula, but not dorsal anterior cingulate cortices.
Correlation of Functional Networks to the Task
The functional networks were correlated with the time courses of the 3 motor tasks and presented by group ranked from highest to lowest correlation based on the writing task (Fig. 3). Overall, the first 8 networks with the highest correlation to the writing task showed consistently lower correlation in WC compared to HV. Similar trends were not observed during the sequence tapping or flexion-extension tasks.
Sensorimotor network
The left sensorimotor network showed the strongest correlation across all 3 tasks consistent with the nature of the 3 tasks. But there were no significant group differences in the correlation of the sensorimotor network to any of the 3 motor tasks. Furthermore, of all the functional networks reported, the SMN showed the lowest within group variabilities across all 3 tasks.
Basal Ganglia
The basal ganglia showed significant group level differences during the writing task but not tapping or extend tasks. Specifically, WC patients showed significantly decreased correlation to writing task time course compared to HV.
Cerebellum
The cerebellum showed strong correlation with all 3 task time courses. Although no significant group level differences were observed, there was a larger intragroup variability in WC during writing and tapping tasks.
Parietal
The parietal network showed weaker correlation to all 3 task time courses in HV. There was a consistent increase in the correlation between the task time course and parietal network in WC for all 3 tasks with significant difference observed during the flexion-extension task only.
Orbitofrontal
The orbitofrontal network showed weak correlation to all 3 task time courses in HV. In contrast, the WC patients showed significantly stronger correlation with orbitofrontal network during writing task.
Default Mode Network
The default mode network showed the weakest correlation to all 3 task time courses in both groups. No group differences to the task time courses were observed for all 3 tasks.
Functional Network Connectivity
We investigated functional connectivity of all 11 functional networks to each other during the 3 motor tasks. Overall, HV show a similar pattern of functional connectivity across all 3 motor tasks. In contrast, WC show a similar pattern of connectivity between writing and flexion-extension tasks and to a lesser degree, sequence tapping task (Fig. 4).
Writing task
WC patients show an overall pattern of decreased functional connectivity compared to HV.
Connectivity to Basal Ganglia
There is decreased connectivity be tween the basal ganglia and left sensorimotor cortex (HV 0.26 WC –0.11, p=0.02) that is not present in the tapping tasks (HV .15 WC .25, p=0.47) or extend tasks (HV .08 WC -.18, p=0.22). There is increased connectivity between the basal ganglia and insula during writing (HV 0.02 WC 0.36, p=0.01).
Connectivity to Cerebellum
There is increased connectivity between the cerebellum and the parietal (HV .09 WC .41, p=.04) as well as between the cerebellum and the orbitofrontal network (HV -.74 WC -.33, p = .03). There is decreased connectivity between the cerebellum and basal ganglia that approaches significance during writing (HV 0.36 WC 0.07, p=0.12).
Connectivity to Cortex
There is decreased in connectivity between the orbitofrontal and insular network (HV .26 WC - .02, p = .05). There is increased connectivity between the orbitofrontal and visual cortex (HV - .39 WC –.11, p = .05). There is increased connectivity between the visual and parietal network in the writing task (HV -.37 WC .10, p=.02).
Tapping Task
WC patients show an overall similar pattern of functional connectivity compared to HV.
Connectivity to Basal Ganglia
There is increased connectivity between the basal ganglia and parietal in the tapping task (HV - .07 WC .20, p = .04) that approaches significance in the extend task (HV .22 WC -.02 p = .10) but not in the writing task (HV -.004 WC .04, p=.72). There is no difference in connectivity between the basal ganglia and cerebellum during the tapping task (HV 0.27 WC 0.23, p=0.76).
Extend Task
WC patients show an overall pattern of decreased functional connectivity compared to HV.
Connectivity to Basal Ganglia
There is increased connectivity between the basal ganglia and insula (HV 0.05 WC 0.40, p=0.02). There is decreased connectivity between the basal ganglia and right sensorimotor cortex (HV 0.56 WC 0.15, p=0.03) as well as between the basal ganglia and cerebellum that approaches significance (HV 0.26 WC –0.09, p=0.07).
Connectivity to Cortex
There is increased connectivity between the parietal cortex and visual cortex (HV –0.34 WC 0.06, p=0.04) during flexion-extension task.
DISCUSSION
Dystonia is a complex network level brain disorder with a limited understanding of interactions between functional networks. In this study, we used a task-specific dystonia to understand the changes in functional network connectivity during the dystonic task or “on” disease state and compare it with previous findings in resting state fMRI or “off” disease state. Using independent component analysis, we found that motor tasks modulated the same networks as rest state brain in both healthy and dystonia subjects. Specifically, the sensorimotor, parietal, frontoparietal, visual, basal ganglia, and cerebellum were all modulated as previously reported (Battistella and Simonyan 2019; Mantel et al. 2018; Huang et al. 2017; Baltazar et al. 2020). Our study also found that insular, orbitofrontal and default mode networks were also modulated during motor tasks.
When correlating functional networks with each motor task, we found that the left sensorimotor network showed the strongest correlation with performance of all 3 motor tasks. This finding is consistent with the motor nature of the design tasks and use of the right hand to perform the tasks. As expected, the default mode network showed the weakest correlation with the 3 motor tasks consistent with its established association with brain’s rest state (Buckner, Andrews-Hanna, and Schacter 2008). Thus, the functional networks with the strongest and weakest task correlation serve as positive controls of our analytic approach.
At the group level, the sensorimotor network showed no group differences in task time correlation and low intragroup variability during performance of all 3 motor tasks. These findings suggest that any differences in functional networks between WC dystonia and HV could not be attributed to differences in motor performance of each task as previously theorized.
Our study also found strong correlation between the 3 motor task times and the networks of cerebellum and basal ganglia. These two networks have previously been shown in multiple human and animal model studies to play an important role in the pathophysiology of dystonia (Bonsi et al. 2011; Argyelan et al. 2009; Shakkottai et al. 2017; Peterson, Sejnowski, and Poizner 2010; Jinnah, Neychev, and Hess 2017). However, it remains unclear if one of the networks is the primary driver and the other a compensatory region or if both are directly involved in the primary disease pathophysiology. In our study, although both networks showed strong task time correlation, the only network that showed a difference in task time correlation between WC and HV was the basal ganglia. More importantly, the group difference in task time correlation with the basal ganglia was seen only during the dystonic task of writing but not the other non-dystonic motor tasks, thus suggesting that basal ganglia may be the primary network of brain dysfunction.
The orbitofrontal network also showed differences in task time correlation between WC and HV uniquely during the writing task. We attribute this group difference in orbitofrontal network to the activation of the caudate region within the network. The caudate is a subregion of the basal ganglia that plays an important role in planning the execution of motor movement. Thus, the significantly increased task time correlation of the caudate in WC patients may suggest a bigger compensatory role of motor planning in execution of the dystonic motor task in WC compared to HV. Indeed, a prior study in WC patients found increase in motor planning stage compared to HV (Delnooz et al. 2013).
We also looked at functional network connectivity differences during the 3 motor tasks in our study cohort. We found that the writing task showed the largest number of network connectivity differences between healthy and dystonia patients compared to the other two motor tasks. Of all the connectivity differences, we found decreased connectivity between the left sensorimotor and basal ganglia network in WC particularly noteworthy. This observation is consistent with prior reports (Battistella and Simonyan 2019; Gallea et al. 2016). However, unlike prior studies, we used a data driven approach to show that this difference in connectivity during the dystonic writing task but not the other motor tasks, thus identifying another primary dystonia disease mechanism.
Prior studies reported significantly decreased connectivity between the basal ganglia and cerebellum in WC compared to HV (Mantel et al. 2018). In our study, we also observed decreased connectivity between the basal ganglia and cerebellum in WC that approaches significance. However, the decrease connectivity in WC was observed during both the dystonic task (writing) and non-dystonic task (flexion-extension). The decreased connectivity in WC was not observed during the sequence tapping task.
WC patients also demonstrated differences in cerebellar network connectivity. Interestingly, these group differences in cerebellar connectivity were only observed during the dystonic writing task but not the non-dystonic motor tasks. Furthermore, the direction of the connectivity change is also noteworthy: both networks (the orbitofrontal and parietal) showed increased connectivity to the cerebellum. The lack of group differences in task time correlation to the cerebellum in conjunction with the increased connectivity between the cerebellum and other networks uniquely during writing task raises the possibility of the cerebellum playing a role as a compensatory brain network.
The cortical networks also showed differences in connectivity between WC and HV that were unique to the writing task. Specifically, the orbitofrontal network showed increased connectivity to the visual network and decreased connectivity to the insular network in WC uniquely during writing task.
The differences in network task time correlation and functional connectivities uniquely during writing task in WC patients reported in this study allows us to put forward a unifying mechanism of global network disorder in dystonia. Specifically, our study suggests that dystonia is a disease with a primary intra-network dysfunction in the basal ganglia. This dysfunction in the basal ganglia then leads to decreased connectivity of the basal ganglia with the left sensorimotor network and cerebellar network. The decrease basal ganglia connectivity to the cerebellum leads to compensatory changes of increased connectivities between the cerebellum and orbitofrontal and parietal networks. The orbitofrontal network, which plays an important role in motor planning, in turn increases connectivity to the visual network to help with planning of the motor movement while the parietal network modulates the premotor network as previously reported (Battistella and Simonyan 2019; Gallea et al. 2016).
Lastly, our study demonstrates that while all 3 motor tasks overlap in the task-time correlation to the sensorimotor network, the remaining networks show task-time correlations and functional network connectivity that are largely unique to each motor task. Therefore, the sequence tapping task does not capture all the network changes seen in dystonic writing task suggesting that it may not serve as a good substitute for studying task-specific dystonia.
Data Availability
Data available upon request from the authors
Data Code availability statement
Available upon request.
Credit authorship contribution statement
NBP was involved in all aspects of the research project from conceptualizing the study, to data collection, data analysis and manuscript writing. MF was involved in data analysis and assisted with manuscript writing. JV assisted with study design, data collection and reviewed the manuscript. AM helped conceptualize the research study, performed and critiqued data analysis, and assisted with manuscript writing.
Funding
This work was supported by grants from the National Institutes of Health (KL2TR002554), Dystonia Medical Research Foundation, Doris Duke Charitable Foundation, and Dystonia Coalition. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies.
Acknowledgments
The authors would like to thank Amber Holden, Ashley Pifer and Kelsey Ling who served as Clinical Research Coordinators for this study. The authors would also like to thank Mariusz Derezinski-Choo for his assistance in conducting the research and Dr. Simon Davis for his conceptual feedback.
Abbreviations
- WC
- Writer’s Cramp
- HV
- Healthy Volunteer
- FNC
- Functional network connectivity