Interfaçage d’une prothèse bidirectionnelle chez la souris

Invasive brain-machine interfaces (BMIs) offer patients with severe motor deficits the hope of regaining adegree of autonomy by controlling a robotic arm via neuronal activity recorded in their motor cortex. Stillin a very experimental phase in humans, BMI research is mainly conducted in primates. Extending thisresearch on the mouse model, which possesses numerous genetic tools such as optogenetic, would makeit more disruptive and permissive for the questions surrounding BMI development.To this aim, we have developed a biomimetic prosthesis of the mouse upper limb. This miniature roboticarm can perform accurate movements in three dimensions thanks to its four degrees of freedom. This ismade possible by closed-loop control with an optical tracking system providing a real-time estimation ofthe prosthesis posture.To demonstrate the feasibility of using the mouse model, we set up a water-collection task: the waterdeprivedmouse had to control the prosthesis in a two-dimensional space to reach a water tank, then bringit back toward itself to lick the water carried by capillarity between the prosthesis fingers. The mice wereable to improve their ability to collect water by modulating and coordinating the activity of two neurons,each attributed to one dimension of control. The firing rate of the neuron was linked to the speed of theprosthesis along its attributed dimension. Encouraging results were also obtained with a three-dimensionalcontrol space.The current version of the prosthesis developed during this thesis has been duplicated for several projectsin the laboratory and beyond. It will allow — in parallel with the continuation of its development — toexplore several themes surrounding BMIs: motor control through mesoscale cortical activity, theimplementation of artificial sensory feedback, and prosthesis embodiment.