Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Publication
Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
All human actions involve motor control. Even the simplest movement requires the coordinated recruitment of many muscles, orchestrated by neuronal circuits in the brain and the spinal cord. As a consequence, lesions affecting the central nervous system, such as stroke, can lead to a wide range of motor impairments. While a certain degree of recovery can often be achieved by harnessing the plasticity of the motor hierarchy, patients typically struggle to regain full motor control. In this context, technology-assisted interventions offer the prospect of intense, controllable and quantifiable motor training. Yet, clinical outcomes remain comparable to conventional approaches, suggesting the need for a paradigm shift towards customized knowledge-driven treatments to fully exploit their potential. In this thesis, we argue that a detailed understanding of healthy and impaired motor pathways can foster the development of therapies optimally engaging plasticity. To this end, we develop and apply multimodal methodologies to investigate the central and peripheral mechanisms underlying motor control and recovery. In the first part of this work, we concentrate on the transition from one-suits-all approaches to patient-tailored protocols, in the context of robot-assisted rehabilitation. We start addressing this question from a technical viewpoint and propose methods to assess individual dynamics of recovery in stroke patients. First, we demonstrate the applicability of a model-based approach to continuously personalize training based on kinematic motor improvement. Then, we show how complementary knowledge can be gleaned from kinematics, muscular and neural signals, and we introduce a versatile framework to distill this multimodal information into a set of clinically relevant variables. These results highlight the pivotal importance of multimodality, stressing the need for a comprehensive view of the human motor hierarchy. To this end, the second part of this work focuses on the spinal cord, whose functional properties remain largely unexplored in humans. As this gap of knowledge primarily pertains to the dearth of non-invasive methods to assess its function in vivo, we first propose a pipeline for spinal cord functional magnetic resonance imaging (fMRI) and demonstrate its ability to capture cervical activation patterns during upper limb movements. We then present a dynamic functional connectivity framework to dissect spinal spontaneous fluctuations into fine-grained components mirroring neuroanatomical and physiological principles. Next, we extend the use of this approach to fMRI data acquired in the entire neural axis during motor sequence learning, hence shedding light on specific cortical, subcortical and spinal correlates of skill acquisition and consolidation. Finally, we glimpse into the implementation of these methodologies in the scope of translational applications, providing evidence of their potential to explore spinal plasticity following stroke. These findings are a valuable contribution towards an extensive characterization of human motor control. This system-level view deepens our understanding of motor pathways, fully acknowledging the active and plastic nature of the spinal cord and emphasizing its key role in sensorimotor functions. We envision that the synergy between technology and knowledge will open promising avenues for strategies leveraging each patientâs residual function to optimize clinical outcome.
, , , ,
,