Browsing by Author "Cluff, Tyler"

Now showing 1 - 15 of 15
  • Thumbnail Image
    ItemEmbargo
    Assessing Motor Adaptation Following Adolescent Sport-Related Concussion
    (2024-07-03) Stuart, Devon Walter; Dukelow, Sean; Cluff, Tyler; Kirton, Adam; Emery, Carolyn
    Concussion is a mild traumatic brain injury that is common for adolescents participating in sports. Sport-related concussion can be acquired through biomechanically diverse mechanisms of injury, leading to diverse damage to the brain that can result in cognitive, sensorimotor, and/or vestibular impairments. Currently there are no gold standard tools for clinicians to use when diagnosing, managing, and making clearance to play decisions following sport-related concussion. When examining individuals following suspected concussion and when making clearance decisions to play, we postulated that an assessment task that engages distributed, global brain networks to capture concussion-related neurological impairments would be useful. Motor adaptation is a brain function important for sport participation that requires engagement of sensory, motor, cognitive, and association regions of the brain, leading us test this brain function. This thesis examined motor adaptation impairments in adolescents at two timepoints: within 10 days postinjury (Experiment 1), and after obtaining medical clearance to return to sports (Experiment 2). Group level differences in all tested task parameters were seen within 10 days of sport-related concussion, with approximately one-in-four participants exhibiting overall impairments in motor adaptation. When medical clearance to return to play was granted, group level differences were no longer seen compared to healthy controls. However, approximately one-in-four participants who were medically cleared exhibited overall impairments in motor adaptation. Taken together, our results suggest that motor adaptation may be a valuable brain function to assess during the clinical management of sport-related concussion, especially given the relevance of motor adaptation within sport. Further, robotic tools were shown to be capable of detecting subtle neurological deficits after sport-related concussion.
  • Thumbnail Image
    ItemEmbargo
    Characterizing the cerebello-thalamo-cortical tracts in the pathophysiology of adult-onset idiopathic dystonia
    (2023-07) Sondergaard, Rachel Elisa; Kiss, Zelma; Martino, Davide; Condliffe, Elizabeth; Cluff, Tyler
    Adult-onset idiopathic dystonia (AOID) is a movement disorder causing painful and disabling muscular contractions. The pathophysiology of AOID has evolved over time from original consideration as a purely functional neurological condition to its present understanding as a disorder of the motor network. What is presently unknown within the network model is the relative contribution of specific cerebellar outflow pathways, despite several lines of evidence in support of a role for these tracts. In this thesis, I attempt to characterize the anatomy, function, and response to intervention of the cerebello-thalamo-cortical pathway, a cerebellar outflow pathway subserving motor processes, in AOID. First, tractographic measures of the bilateral dentato-rubro-thalamic tracts were computed in patients with cervical dystonia (CD) and healthy controls. We identified bilateral reductions in diffusion tractography metrics in CD metrics relative to controls. We also computed the degree of lateralization of these tractographic measures and found that it was related to the severity of CD in subgroups of patients with similar motor phenotype. Second, I sought evidence that functions subserved by the cerebello-thalamo-cortical tracts may also be abnormal. Using a transcranial magnetic stimulation (TMS) protocol called cerebellar brain inhibition (CBI) I found that CD severity increased in association with a breakdown in normal CBI. I suspected that proprioception, another function subserved by this pathway might be abnormal and could be rescued by changing activity along this pathway using repetitive TMS. I was unable to find evidence of either in a small pilot study. I was also unable to find evidence that a single session of repetitive TMS targeting the cerebellar cortex induced changes in local field potential activity at the level of the thalamus in movement disorder patients treated with deep brain stimulators for their movement disorders. Third, I indirectly examined the response to intervention subserved by the cerebello-thalamo-cortical tracts by comparing TMS motor maps of hand representations in focal hand dystonia patients produced at peak botulinum toxin treatment effect and following washout. I examined technical factors contributing to the modest or absent changes in hand representation between conditions in these patients in an additional pilot study and structured review. Overall, I found evidence supporting abnormal anatomy and function of the cerebello-thalamo-cortical tracts in dystonia. Taken together, the findings establish structural and functional features of the cerebello-thalamo-cortical tracts in the pathophysiology of dystonia. The specific pattern of microstructural abnormalities observed and its relationship with severity also provides a plausible new target for non-invasive repetitive TMS targeting the cerebellum for therapeutic effect in AOID.
  • Thumbnail Image
    ItemOpen Access
    The Effects of Transcranial Direct-Current Stimulation on Motor Learning, Motor Maps, and Functional Networks in Children
    (2022-08) Giuffre, Adrianna; Kirton, Adam; Carlson, Helen; Kiss, Zelma; Cluff, Tyler
    Mapping the structure and function of the motor system in children informs our understanding of brain development, health, and disease. Neuronavigated robotic transcranial magnetic stimulation (TMS) is a state-of-the-art tool that can non-invasively explore primary motor cortex (M1) excitability and generate high-resolution motor maps of upper extremity muscles. However, fundamental studies are lacking in the developing brain. We propose a safe protocol, integrating methods capable of simultaneously exploring M1 modulation and TMS motor maps in typically developing children. Next, we investigated whether behavioural performance corresponds to TMS motor map outcomes and M1-excitability. We generated detailed bilateral motor maps of multiple hand muscles and observed hemispheric-specific relationships between M1-excitability, map outcomes, and motor performance. As most TMS mapping studies have reported variable results, we also determined the reliability of robotic TMS motor maps and specific outcomes across short- and long-term sessions. Our findings suggest that careful interpretation of mapping protocols and outcomes is required to interrogate M1 plasticity. M1 has become a central target to modulate plasticity for its critical role in motor control and learning. Transcranial direct current stimulation (tDCS) can non-invasively modulate M1-excitability in healthy and clinical populations. Primarily studied in adults, tDCS can enhance motor learning when paired with a motor task. However, the effects of tDCS may differ in the developing brain due to anatomical and maturational idiosyncrasies. High-definition tDCS (HD-tDCS) provides more focal targeting of cortical areas, leading to enhanced motor learning in adults, but is yet to be investigated in a pediatric population. Therefore, we aimed to determine the effects of tDCS and HD-tDCS on upper limb motor learning in typically developing children. We demonstrated that both forms of stimulation safely enhanced motor learning with long-term retention of effects. With a pressing need to determine the underlying mechanisms of such neuromodulation, we then applied our TMS mapping methods and advanced functional magnetic resonance imaging (fMRI) techniques to characterize the effects of tDCS-enhanced motor learning on motor network physiology. Alterations in motor maps and both inter-and intra-hemispheric functional motor networks were identified. We have advanced the understanding of motor system developmental and interventional plasticity in children.
  • Thumbnail Image
    ItemOpen Access
    Electrical Vestibular Stimulation for Probing the Effect of Combat Sports on Balance Control
    (2024-04-29) Banman, Christopher James; Peters, Ryan; Cluff, Tyler; Schneider, Kathryn
    Combat athletes are at a high risk for head injuries such as concussion. Recent work has uncovered that even repetitive head impacts (RHIs) may have lasting effects on cognitive, emotional, and motor function. This thesis will focus on the changes to vestibulospinal processing seen in combat athletes. Electrical vestibular stimulation (EVS) was used to evoke vestibulospinal reflexes, comparing the resulting responses between combat athletes and healthy sex/age matched controls. This is described through three separate chapters. First, I test the differences in electromyography (EMG) responses, then develop a novel EVS-induced postural sway thresholding technique, and finally investigate the EVS threshold differences between the fighters and controls. The experimental results of the first study demonstrated frequency-specific changes in vestibular sensitivity to EVS for combat athletes, via EMG responses. Combat athletes showed an increased sensitivity to low frequency EVS, and a decreased sensitivity to high frequency EVS, resulting in increased latencies on short and medium latency reflexes, scaling with increasing career RHI exposure. The second study presented shows that stochastic electric vestibular stimulation (SVS) can be used to determine EVS postural sway thresholds much lower than those that have previously been shown in the literature. The second study also shows that humans appear to be more sensitive to bipolar EVS than monopolar EVS. The third study found that combat athletes had lower EVS-induced sway thresholds using low frequency EVS than their matched controls. Taken together this thesis displays significant changes to vestibulospinal processing with increased RHI exposure in combat athletes.
  • Thumbnail Image
    ItemOpen Access
    Feedback Responses Must Disengage from Postural Control to Initiate Rapid Movements
    (2021-09) Yeung, Natalie; Cluff, Tyler; Peters, Ryan; Herzog, Walter; Dukelow, Sean
    The nervous system enables humans to respond to changes in the environment, and when necessary, update the course of an ongoing action. The process of initiating a new motor action when the task changes suddenly appears to carry a delay or time cost. Electrophysiological recordings suggest this time cost may arise from neural processing related to engaging in a new motor action. We know little about this time cost and when it arises following changes in task demands. Here we examine the time required to respond to a change in the goal of ongoing upper limb motor tasks. In 3 experiments, we applied visual perturbations while a total of 47 participants (22 females) maintained upper limb postural control (Experiment 1), performed reaching movements or maintained a fixed upper limb posture (Experiment 2), and transitioned from holding a fixed posture to initiating a reaching movement (Experiment 3). When applied in postural control, visual disturbances (‘cursor jumps’) required participants to disengage from holding a fixed limb position before initiating a rapid corrective response to drive the cursor back to the target. The same perturbations require a corrective response but do not impose a change in goal when the nervous system is already engaged in movement. Rapid muscle responses emerged as early as ~105 ms when responding to the visual perturbations during upper limb reaching movements compared to postural control. In some postural control conditions, the nervous system required ~270 ms to update muscle activity when responding to the same visual disturbances. The findings highlight slower corrective responses when required to disengage from maintaining a fixed limb position to initiate a corrective response. Taken together, our results emphasize a direct time cost when responding to changes in the goal of the task.
  • Thumbnail Image
    ItemOpen Access
    Gait Entrainment in Coupled Oscillator Systems: Clarifying the Role of Energy Optimization in Human Walking
    (2020-01-13) Schroeder, Ryan T.; Bertram, John E. A.; Croft, James L.; Sawicki, Gregory S.; von Tscharner, Vinzenz; Shrive, Nigel; Cluff, Tyler; Rubenson, Jonas
    Empirical evidence suggests that parameters of human gait (e.g. step frequency, step length) tend to minimize energy expenditure. However, it is unclear if individuals can adapt to dynamic environments in real time, i.e. continuously optimize energy expenditure, and to what extent. Two coupled oscillator systems were used to test the learned interactions of individuals within dynamic environments: (1) experienced farmworkers carrying oscillating loads on a flexible bamboo pole and (2) individuals walking on a treadmill while strapped to a mechatronics oscillator system providing periodic forces to the body. Reductionist trajectory optimization models predicted energy-minimizing gait interactions within the coupled oscillator systems and were compared to experimental data assessed with linear mixed models. On average, pole carriers significantly adjusted step frequency by 3.3% (0.067 Hz, p=0.014) to accommodate the bamboo pole – consistent with model predictions of energy savings. Novice subjects entrained (i.e. synchronized) their step frequency with machine oscillations up to ±10% of preferred step frequency and at amplitudes as low as 5% body weight (or ~33 N). Still, some subjects rarely entrained at all, and many exhibited transient entrainment, i.e. they drifted in and out of step frequencies matching the machine oscillations. Overall, subject entrainment was more robust and consistent with lower frequencies and higher amplitudes (20-30% of body weight). Although no systematic difference was found between the metabolic consumption of subjects during and not during entrainment, the net mechanical work done on subjects by the force oscillations had a strong effect on metabolic output (p<0.0001). Net work was largely determined by the alignment of oscillation forces within the gait cycle. Both the optimization model and subjects aligned force oscillations with their body velocity to increase positive power. All in all, it seems that subjects prefer entrainment with environmental oscillations under certain circumstances. However, entrainment does not appear to be motivated by energetic cost, at least not directly and not as a first priority. It is possible that individuals stabilize interactions with the environment (e.g. entrainment) as a prerequisite for effective feedforward and/or feedback gait control.
  • Thumbnail Image
    ItemOpen Access
    Grey and White Matter Correlates Underlying Proprioceptive Impairments After Stroke
    (2023-01-25) Chilvers, Matthew J; Dukelow, Sean P; Cluff, Tyler; Kirton, Adam
    Proprioception is our sense of limb position and movement and is an important sense for the accurate control of movement. Proprioception is commonly affected post-stroke, with impairments demonstrated in ~50% of individuals. Despite high prevalence rates, the understanding of the neural correlates of these impairments, at both the grey and white matter level, is incomplete. In Chapter Two, a selective lesion analysis was performed to identify other temporoparietal structures that, when lesioned, resulted in impairments on an arm position matching (APM) and kinaesthesia task, even in the absence of lesions to sensory cortex. Participants with lesions to right temporoparietal regions performed worse on both tasks than their left hemisphere counterparts. In Chapter Three, lesions to, and diffusion properties of, the Superior Longitudinal Fasciculus (SLF), Arcuate Fasciculus (AF) and Middle Longitudinal Fasciculus (MdLF) were assessed in relation to performance on an APM task post-stroke (n=26). Lesions to each tract was associated with worse APM performance, while lower fractional anisotropy of the SLF III and AF was also associated with worse performance. In Chapter Four, the association between tract damage and APM performance was assessed, in addition to differences in the disconnectome between participants with and without APM task impairments. The extent of damage to a perisylvian network of white matter was associated with APM performance. Along with a perisylvian network, disconnection of the medial lemniscus, corpus callosum, inferior fronto-occipital fasciculus, inferior longitudinal fasciculus and optic radiations was also associated with impairments on the APM task. In Chapter Five, the ability of clinical, robotic and neuroimaging data, collected in the subacute stage post-stroke (n=133), was assessed for its utility in predicting chronic proprioceptive impairments. Models which combined variables across modalities (clinical, robotic and neuroimaging) were often the highest performing and could accurately predict impairments at six-months post-stroke. Overall, this thesis advances our understanding of the grey and white matter correlates of proprioceptive impairment post-stroke, something which has been under-investigated to date. Developing a better understanding of the neuroanatomy important for proprioception, and identifying participants at risk of long-term proprioceptive impairments, is key to developing better treatments for proprioception post-stroke.
  • Thumbnail Image
    ItemOpen Access
    Hypothalamic dopamine neurons project to brainstem regions related to movement
    (2022-06-22) Grams, Joseph R; Whelan, Patrick J.; Cluff, Tyler; Wilson, Richard
    Locomotion is a complex behaviour resulting from interactions between the brain and spinal cord. Dopamine modulates locomotion via ascending projections to the basal ganglia and descending projections to the mesencephalic locomotor region (MLR). The MLR is a region in the brainstem known to initiate and control locomotion. Motor commands from supraspinal structures are relayed to the spinal cord via the medullary reticular formation (MRF). However, it remains uncertain if dopamine neurons also innervate regions of the MRF known to influence locomotor activity. Using a retrograde viral tracing approach, I identified a discrete dopaminergic pathway that extends from the A11 region of the posterior hypothalamus to locomotor regions of the MRF. Furthermore, using an anterograde viral tracing technique, I found that dopamine neurons of the A11 region project throughout the rostrocaudal extent of the MRF and terminate predominantly ipsilaterally. Contrastingly, using immunohistochemistry, I show that neurons in the MRF are devoid of D1 receptors. Lastly, using the genetically encoded dopamine sensor dLight1.2 and fibre photometry, I provide evidence of altered dopamine transients during locomotor activity. Together, these results suggest that dopaminergic innervation of the MRF is sparse and originates primarily from the A11 region of the posterior hypothalamus. Further investigation is required to determine the functional contributions to neural circuits in the MRF.
  • Thumbnail Image
    ItemOpen Access
    Modulation of Upper Limb Sensorimotor Control in Unpredictable Mechanical Environments
    (2021-12-16) Miller, Ryan; Cluff, Tyler; Aboodarda, Saied; Ryan, Peters
    Humans have the ability to adapt their movements to accommodate changes within the environment. A common strategy used to probe for adaptation is to expose people to environments where they will encounter predictable disturbances. We know comparatively little about whether people adapt to environments that are highly unpredictable, or how repeated exposure to unpredictable environments alters responses on larger time scales. This is important as many of our movements are performed in the presence of unpredictability. A basketball player with a clear lane to the hoop may alter their strategy if a defender steps in their way. The defender in this situation introduces unpredictability into the environment and the forward must adapt to achieve their task goal. In this thesis, I examine how people alter their electrophysiological, metabolic, and behavioural responses when exposed to unpredictable mechanical perturbations during upper limb posture control and reaching movements. In Chapter 3, I describe two experiments where 35 healthy participants maintained hand position within a fixed target while exposed to elbow perturbations that were unpredictable in direction. The results indicate that although the perturbations were unpredictable in direction, subjects produced stereotypical adaptation responses such as a reduction in muscle activity and energy expenditure, and an improvement in performance. Posture and reaching may involve distinct control mechanisms. In Chapter 4, 37 healthy subjects performed point-to-point reaching experiments while unpredictable perturbations were delivered during the reaching movement. The results in Chapter 4 demonstrated that humans produce adaptive responses to unpredictable perturbations during reaching, and such responses scale linearly to the amplitude of perturbation. Taken together, this work elucidates the plasticity of the nervous system and indicates that the nervous system works to minimize energy expenditure while maintaining the goals of motor control tasks.
  • Thumbnail Image
    ItemOpen Access
    Motor Learning after Stroke
    (2023-05-10) Moore, Robert Taylor; Cluff, Tyler; Dukelow, Sean; Kirton, Adam; Hill, Michael
    Motor learning is a pillar of stroke rehabilitation. Indeed, many therapeutic protocols and interventions are based on principles of motor learning. One of the assumptions made in rehabilitation is that motor learning remains intact after stroke and can be leveraged to facilitate recovery. However, a growing body of evidence has shown that motor learning can be impaired after stroke. Our understanding of how stroke influences the neural and behavioural processes that support motor learning is incomplete. This raises questions about how well our current understanding of motor learning, derived predominantly from studies in healthy adults, applies to stroke rehabilitation. The following dissertation describes four studies that examine reaching movements and a specific type of motor learning known as motor adaptation. This type of learning encompasses the processes that help to maintain accurate movements amidst changes in the body, environment, and task demands. Across three experiments in healthy adults (Chapter Two) and three experiments in participants with stroke (Chapters Three, Four, and Five), we characterized motor adaptation in health and disease. Overall, the works in this dissertation demonstrate the utility of robotics for quantifying motor adaptation. Impaired adaptation after stroke was associated with several clinical variables including: the side of the stroke affected limb (i.e., dominant versus non-dominant), time post-stroke, movement performance, proprioceptive abilities, and clinical assessments of motor impairment and functional independence. Notably these variables accounted for only a small portion of the variance in motor adaptation after stroke, suggesting that other clinical variables (e.g., lesion characteristics or other types of impairments) may be associated with adaptation after stroke. Our results reveal widespread impairments in visuomotor adaptation after stroke and generate numerous questions about the basic mechanisms underlying motor adaptation and how adaptation applies to stroke rehabilitation.
  • Thumbnail Image
    ItemEmbargo
    Muscle coactivation can prime the nervous system for rapid feedback control
    (2024-07-22) Maurus, Philipp; Cluff, Tyler; Peters, Ryan; Wong, Jeremy; Scott, Stephen; Herzog, Walter; Wolpert, Daniel
    Humans often interact with environments where they may encounter disturbances that can vary in direction and amplitude between or within movement. In turn, these disturbances can threaten success of their actions. Recent advances in Optimal Feedback Control highlight the flexible use of sensory feedback when dealing with unpredictable disturbances. In this thesis, we investigated how features of the body or task alter how the healthy nervous system processes sensory feedback. Chapter 2 examined whether right-handed individuals (N=28) express differences in stretch responses when controlling the posture of the dominant and nondominant arms. Across two experiments, we found no differences in stretch responses between the two arms. Instead, behavioral and muscle responses correlated across arms. Chapter 3 tested whether participants (N=70) alter their responsiveness to sensory feedback when reaching in the presence of random, time-varying torque disturbances. Across three experiments, we demonstrated that participants increased and tuned their responses to proprioceptive and visual feedback as well as muscle coactivation to the variability of the random torque disturbances. Chapter 4 investigated whether participants (N=90) modulate their responsiveness to sensory feedback when exposed to unpredictable visuomotor rotations that can vary in amplitude and direction between reaching movements. Across three experiments, we showed that participants increased and tuned their responses to visual and proprioceptive feedback and muscle coactivation to the variability of the unpredictable visuomotor rotations. Moreover, increases in muscle coactivation correlated with the vigor of corrective responses to visual and proprioceptive feedback. Chapter 5 examined the role of muscle coactivation in upregulating responses to sensory feedback when initiating and controlling voluntary reaching movements. Across four experiments (N=80), we found that participants spontaneously increased muscle coactivation and responses to sensory feedback when the task imposed greater temporal urgency. Moreover, instructions to coactivate upregulated responses to visual, auditory, and somatosensory feedback and expedited the initiation and control of voluntary reaching movements. Collectively, the studies included in this thesis highlight the important role of muscle coactivation in upregulating responses to sensory feedback to provide mobility. The findings raise questions about the traditional role of muscle coactivation in leveraging viscoelastic properties of skeletal muscles to provide stability.
  • Thumbnail Image
    ItemOpen Access
    Robotic and Imaging Biomarkers of Sensorimotor Dysfunction in Hemiparetic Children after Perinatal Stroke
    (2017) Kuczynski, Andrea; Kirton, Adam; Dukelow, Sean; Lebel, Catherine; Demchuk, Andrew; Boyd, Lara; Cluff, Tyler
    Perinatal ischemic stroke results from focal cerebral arterial or venous occlusion and usually causes lifelong disability. Perinatal stroke is the leading cause of hemiparetic cerebral palsy, and may result in impairment in both sensory and motor function. A major limitation in assessing sensory and motor function following stroke is a lack of objective, sensitive measurement tools which are required to advance personalized therapeutic strategies. In this study, we used a novel robotic exoskeleton (KINARM) to assess sensory and motor function in hemiparetic children with perinatal stroke. Using diffusion tensor imaging, we further sought to better understand changes in sensorimotor pathway microstructure following perinatal stroke and their relationship with behaviour. We studied 50 children with perinatal stroke (arterial or venous) and 150 typically developing children aged 6 to 19 years. Children with perinatal stroke demonstrated diverse impairments in both sensory and motor function relative to controls. Mean group differences were greater for arterial strokes compared to venous. Proprioceptive deficits included dysfunction in both robotic position-matching and kinesthesia tasks that correlated poorly with beside sensory tests. Both stroke groups demonstrated impaired motor reaching in the contralesional limb, with greater deficits in the arterial group. The ipsilesional, “unaffected” arm of arterial cases also often showed impairments. Robotic sensory and motor performance was often associated with the structural connectivity of the corresponding lesioned tract as measured by diffusion tensor imaging. Our findings contribute to a better understanding of how sensory and motor systems develop following unilateral perinatal injury and how these relate to clinical function. We add new components to emerging developmental plasticity models of perinatal stroke including detailed functional outcomes of both upper extremities and measures of structural connectivity in related major tracts. Such knowledge can inform emerging neuromodulation trials towards personalized neurorehabilitation and improved outcomes for hemiparetic children.
  • Thumbnail Image
    ItemOpen Access
    The independence of impairments in proprioception and visuomotor adaptation after stroke
    (2024-05-18) Moore, Robert T.; Piitz, Mark A.; Singh, Nishita; Dukelow, Sean P.; Cluff, Tyler
    Abstract Background Proprioceptive impairments are common after stroke and are associated with worse motor recovery and poor rehabilitation outcomes. Motor learning may also be an important factor in motor recovery, and some evidence in healthy adults suggests that reduced proprioceptive function is associated with reductions in motor learning. It is unclear how impairments in proprioception and motor learning relate after stroke. Here we used robotics and a traditional clinical assessment to examine the link between impairments in proprioception after stroke and a type of motor learning known as visuomotor adaptation. Methods We recruited participants with first-time unilateral stroke and controls matched for overall age and sex. Proprioceptive impairments in the more affected arm were assessed using robotic arm position- (APM) and movement-matching (AMM) tasks. We also assessed proprioceptive impairments using a clinical scale (Thumb Localization Test; TLT). Visuomotor adaptation was assessed using a task that systematically rotated hand cursor feedback during reaching movements (VMR). We quantified how much participants adapted to the disturbance and how many trials they took to adapt to the same levels as controls. Spearman’s rho was used to examine the relationship between proprioception, assessed using robotics and the TLT, and visuomotor adaptation. Data from healthy adults were used to identify participants with stroke who were impaired in proprioception and visuomotor adaptation. The independence of impairments in proprioception and adaptation were examined using Fisher’s exact tests. Results Impairments in proprioception (58.3%) and adaptation (52.1%) were common in participants with stroke (n = 48; 2.10% acute, 70.8% subacute, 27.1% chronic stroke). Performance on the APM task, AMM task, and TLT scores correlated weakly with measures of visuomotor adaptation. Fisher’s exact tests demonstrated that impairments in proprioception, assessed using robotics and the TLT, were independent from impairments in visuomotor adaptation in our sample. Conclusion Our results suggest impairments in proprioception may be independent from impairments in visuomotor adaptation after stroke. Further studies are needed to understand factors that influence the relationship between motor learning, proprioception and other rehabilitation outcomes throughout stroke recovery.
  • Thumbnail Image
    ItemOpen Access
    The use of machine learning and deep learning techniques to assess proprioceptive impairments of the upper limb after stroke
    (2023-01-27) Hossain, Delowar; Scott, Stephen H.; Cluff, Tyler; Dukelow, Sean P.
    Abstract Background Robots can generate rich kinematic datasets that have the potential to provide far more insight into impairments than standard clinical ordinal scales. Determining how to define the presence or absence of impairment in individuals using kinematic data, however, can be challenging. Machine learning techniques offer a potential solution to this problem. In the present manuscript we examine proprioception in stroke survivors using a robotic arm position matching task. Proprioception is impaired in 50–60% of stroke survivors and has been associated with poorer motor recovery and longer lengths of hospital stay. We present a simple cut-off score technique for individual kinematic parameters and an overall task score to determine impairment. We then compare the ability of different machine learning (ML) techniques and the above-mentioned task score to correctly classify individuals with or without stroke based on kinematic data. Methods Participants performed an Arm Position Matching (APM) task in an exoskeleton robot. The task produced 12 kinematic parameters that quantify multiple attributes of position sense. We first quantified impairment in individual parameters and an overall task score by determining if participants with stroke fell outside of the 95% cut-off score of control (normative) values. Then, we applied five machine learning algorithms (i.e., Logistic Regression, Decision Tree, Random Forest, Random Forest with Hyperparameters Tuning, and Support Vector Machine), and a deep learning algorithm (i.e., Deep Neural Network) to classify individual participants as to whether or not they had a stroke based only on kinematic parameters using a tenfold cross-validation approach. Results We recruited 429 participants with neuroimaging-confirmed stroke (< 35 days post-stroke) and 465 healthy controls. Depending on the APM parameter, we observed that 10.9–48.4% of stroke participants were impaired, while 44% were impaired based on their overall task score. The mean performance metrics of machine learning and deep learning models were: accuracy 82.4%, precision 85.6%, recall 76.5%, and F1 score 80.6%. All machine learning and deep learning models displayed similar classification accuracy; however, the Random Forest model had the highest numerical accuracy (83%). Our models showed higher sensitivity and specificity (AUC = 0.89) in classifying individual participants than the overall task score (AUC = 0.85) based on their performance in the APM task. We also found that variability was the most important feature in classifying performance in the APM task. Conclusion Our ML models displayed similar classification performance. ML models were able to integrate more kinematic information and relationships between variables into decision making and displayed better classification performance than the overall task score. ML may help to provide insight into individual kinematic features that have previously been overlooked with respect to clinical importance.
  • Thumbnail Image
    ItemEmbargo
    Understanding Stroke Rehabilitation Progression in a Robotic Rehabilitation Trial
    (2020-05-29) Keeling, Alexa Brianne; Dukelow, Sean P.; Cluff, Tyler; Hill, Michael D.
    Stroke is one of the leading causes of adult disability worldwide, leaving many individuals requiring rehabilitation to regain independence. A critical component to any rehabilitation program is progression, which is the ability of therapy program to change according to patient improvement. Currently, there is little known about therapy progression, which negatively impacts the optimization of rehabilitation programs. Therefore, the purpose of this thesis was to better understand how stroke survivor’s kinematics change throughout therapy in order to inform future rehabilitation programs. The first step in answering this question was to understand how motor learning contributes to recovery after stroke, which is explored in Chapter Two. Next, a therapy program was needed in order to study how stroke survivors progress during rehabilitation. The motor learning and stroke recovery principles discussed in Chapter Two were then used to inform the development of tasks for a robotic rehabilitation program for stroke survivors. The development, and subsequent testing, of the tasks are discussed in Chapter Three. It was found that this robotic therapy program was feasible after stroke and has the potential to improve clinical outcomes when compared only to standard of care. Using the results from this pilot study, the robotic therapy tasks were refined, as well as the study protocol, and gave rise to a Phase II Clinical Trial (RESTORE). As discussed in Chapter Four, subacute stroke patients were recruited to receive 20 days of robotic therapy for 1 or 2-hours a day, beginning either 5-9 days or 21-25 days post-stroke. Following completion of the intervention, changes in the participants’ kinematics measuring speed, accuracy, and smoothness of movements were examined. It was found that kinematics of directional error and hand path ratio (measures of accuracy), as well as smoothness, predominantly increased during the first 5 days of the intervention. Movement speed and percent time in target (a measure of accuracy), on the other hand, continued to improve throughout the intervention. These findings should be interpreted with caution due to small sample size but may be used to inform the progression of future robotic rehabilitation tasks.