Description
Date depot: 1 janvier 1900
Titre: Leveraging brain connectivity networks to detect mental states in brain-computer interfaces
Directeur de thèse:
Fabrizio DE VICO FALLANI (ICM)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
A brain-computer interface (BCI) is a device that can decode brain activity alone, thus creating an alternate communication channel between a person and the external environment. BCIs are increasingly used for control and communication, and for treatment of neurological disorders.
Despite its potential societal and clinical impact, BCI performance still remain relatively low and/or unstable with several weeks needed to reach relatively high-performance (>90%) and a non negligible portion of users (15-30%) that won’t be able to achieve control (ie, BCI illiteracy phenomenon).
Current evidence shows that, even if the use of a BCI only requires the activity modulation of a few sensori/motor areas - typically captured by magneto/electroencephalographic (M/EEG) power spectrum - a larger distributed network of remote areas, including frontal and subcortical regions, appear to be involved in the acquisition of BCI control. Recent studies using noninvasive neuroimaging techniques in healthy subjects have also reported significant changes in amplitude and spatial distribution of task-related brain activity after MI-based BCI training.
Taken together, these findings suggest the use of standard signal processing methods that only capture local brain activity changes could underestimate connectivity phenomena which could instead play a crucial role in improving BCI accuracy.
Project’s goals and impact
Contrary to many approaches in BCI having mainly focused on improving the classification block through advanced machine learning algorithms (eg, common spatial filter, Riemann geometry), this project aims to primarily operate on the feature extraction block of a BCI system. In particular, the project is tightly focused on the following questions:
- Which are the brain regions and network mechanisms involved in a BCI task?
- Can we reliably estimate network-based features in real time to achieve effective online control?
- What is the potential of such features to improve BCI accuracy?
By providing fresh knowledge on the neural circuits and the dynamic mechanisms that are involved in BCI tasks this project has the potential to identify new features/biomarkers that can be used to foster the development of coadaptive frameworks and make BCIs more reliable for control and communication in healthy and diseased conditions.
Doctorant.e: Cattai Tiziana