<div> <div> <div> <p>Patients with Spinal Cord Injury (SCI) have limited time with supervised therapy in rehabilitation hospitals. This makes it imperative for them to continue regular therapy at home so they can maximize motor recovery especially for performing Activities of Daily Living (ADL). However, physical therapy can be tedious and frustrating leading to a lack of motivation. A novel upper extremity movement measurement tool was developed using a commercial VR system to rapidly and objectively measure an individual’s range of motion, velocity of movement on an individual gesture basis, and frequency of movements in a three-dimensional space. Further, an exergame with varied and customizable gameplay parameters was developed. Through the analysis of participant interaction with the exergame, we identified gameplay parameters that can be adjusted to affect the player’s perceived and physiological effort. We observed that VR has a significant motivational effect on range of motion of upper limbs in individuals with tetraplegia. The motion data and kernel density estimation is used to determine areas of comfort. Moreover, the system allowed calculation of joint torques through inverse kinematics and dynamics to serve as an analysis tool to gauge muscular effort. The system can provide an improved rehabilitation experience for persons with tetraplegia in home settings while allowing oversight by clinical therapists through analysis of mixed reality videos or it could be used as a supplement or alternative to conventional therapy. </p> </div> </div> </div>

La réalisation d’une action motrice implique l’activation simultanée des muscles agonistes et antagonistes contrôlés par le système nerveux central. Un traumatisme médullaire détériore la moelle épinière, entrainant une déficience motrice et des modifications du contrôle des activations musculaires. Ce travail étudie la réorganisation des activations musculaires, des activations corticales et des interactions corticomusculaires (ICM) d’un groupe traumatisé médullaire cervical (SCI) et d’un groupe contrôle (AB) lors de flexions et d'extensions isométriques autour de l’articulation du coude. En extension, nos résultats ont mis en évidence une altération des capacités de force maximale chez les SCI, associée à une augmentation des activations musculaires, une activation corticale identique aux AB et une diminution de l’implication du M1 dans le contrôle des activations musculaires. En flexion, la force développée, les activations corticales et les ICM étaient similaires chez les SCI et AB, mais les activations antagonistes et la difficulté à inhiber la contraction étaient plus importantes chez les SCI. Pour l’ensemble des participants, les ICM en flexion étaient différentes selon la fonction des groupes musculaires. Ces résultats suggèrent une altération du contrôle cortical des mécanismes inhibiteurs spinaux de la contraction musculaire après un traumatisme médullaire mais indiquent que le cortex moteur reste fonctionnel pour contrôler un acte moteur malgré l’atrophie des muscles extenseurs. Ces résultats pourraient trouver des applications cliniques pour l’élaboration de neuroprothèses nécessitant un contrôle simultané de différents groupes musculaires. / The realization of a motor action involves simultaneous activation of both agonist and antagonist muscles controlled by the central nervous system. Following spinal cord injury, damage to the spinal cord causes both a loss of motor efficiency and changes in the control of muscle activations. In the present work, we studied the reorganization of muscle activations, cortical activations and corticomuscular interactions (ICM) in spinal cord injured (SCI) and able-bodied (AB) participants during voluntary isometric contractions in flexion and extension around the elbow joint. In extension, our results showed altered capacity of maximum force production in SCI participants, associated with increased muscle activations, similar cortical activation and decreased involvement of M1 in the control of muscle activations when compared to AB participants. In flexion, the force capacities, cortical activations and ICM were similar between SCI and AB participants, but the activation of antagonistic muscles and the difficulty to inhibit the contraction were greater in SCI participants. For all participants in flexion, ICM were different depending on the function of the muscle groups. Taken together, these results suggest an alteration of the cortical control of spinal inhibitory mechanisms following a spinal cord injury, but suggest that the motor cortex remain functional to control a motor act despite the atrophy of the extensor muscles. These results could find clinical applications for the development of neuroprotheses involving simultaneous control of different muscle groups.

Robotic-powered exoskeletons are increasingly used to assist patients with movement disorders in daily life and rehabilitation. Accurately estimating joint torque, especially for dynamic movement conditions using EMG, is crucial for effective assistance. Machine learning and deep learning have been employed for EMG-based force/torque estimation, but their precision and robustness have been limited, particularly for dynamic movements. This thesis aims at comparing and analyze the results using MLP, CNN, and CNN-LSTM methods to estimate ankle joint torque in dynamic movements based on HD-EMG. Meanwhile, this thesis designed and tested different data augmentation to enhance the performance using HD-EMG data augmentation techniques. The CNN-LSTM model demonstrated superior performance among the machine learning models. Additionally, the combination of spatial and signal augmentation methods showed notable improvements in the inter-subject case performance of the prediction. No augmentation methods have shown notable improvements in the intra-subject case or inter-session case. / Robotdrivna exoskelett används i allt större utsträckning för att hjälpa patienter med rörelsestörningar i det dagliga livet och rehabiliteringen. Att noggrant uppskatta ledmomentet, särskilt för dynamiska rörelseförhållanden med EMG, är avgörande för effektiv assistans. Maskininlärning och djupinlärning har använts för EMG-baserad kraft/vridmomentuppskattning, men deras precision och robusthet har varit begränsad, särskilt för dynamiska rörelser. Denna avhandling syftar till att jämföra och analysera resultaten med MLP-, CNN- och CNN-LSTM-metoder för att uppskatta fotledsvridmoment i dynamiska rörelser med hjälp av HD-EMG datadrivna modeller. Samtidigt designade och testade denna avhandling olika dataförstärkningar för att förbättra prestandan med hjälp av HD-EMG dataförstärkningstekniker. CNN-LSTM-modellen visade överlägsen prestanda bland maskininlärningsmodellerna. Dessutom visade kombinationen av rumsliga och signalförstärkningsmetoder anmärkningsvärda förbättringar i prediktionens inter-subject case performance. Inga förstärkningsmetoder har visat märkbara förbättringar i fallet inom ämnet eller fallet mellan sessioner.