Neurotechnology for rehabilitation
Mardi 29 novembre 2016 11:00
- Duree : 1 heure
Lieu : Salle "Louis Weil" E424 - Institut Néel -bâtiment E, 3ème étage, CNRS Polygone scientifique, 25 rue des martyrs, 38000 Grenoble
Orateur : Blaise YVERT (INSERM, Clinatec, Grenoble)
Understanding the brain dynamics and finding strategies to overcome disabilities encountered following a trauma or a pathology requires devices to interface the central nervous system an d rehabilitation paradigms. Among other means, neural prosthesis and brain-computer interfaces (BCIs) aim at restoring lost functions following lesions or degeneration of the central nervous system. These neurotechnology-based approaches rely on large-scale neural interfacing using arrays of sensors to record and/or elicit activity within large neural populations. Here, I will focus on two research axes that we conduct in this context, one aiming at improving microelectrode arrays for neural sensing and microstimulation, and one aiming at using cortical implants to build a speech BCI to restore communication in locked-in aphasic patients. The presentation will thus be composed of two parts. First, I will insist on the importance of the electrode material when considering dense arrays of small-size microelectrodes, and show how in such case the noise level and stimulation capabilities of a device can be improved using novel nanoporous metal or carbon-based (diamond) electrode materia ls. Second, I will present our ongoing project aiming at using cortical neural implants to develop a BCI paradigm to restore continuous speech from real-time decoding of cortical activity. For this purpose, we have developed an articulatory speech synthesizer based on deep neural networks mapping the trajectories of the main articulators of the vocal tract (tongue, lips, jaw, velum) onto the acoustic content of speech. Different healthy subjects, in whom we recorded the movements of the main articulators and used these signals as inputs of the synthesizer, could successfully control this synthesizer in real-time. Such closed-loop paradigm will be transposed in a future clinical trial to allow patients implanted with electrode arrays to produce artificial speech from their cortical activity.
Prévenir un ami par email
Télécharger dans mon agenda