Équipe IMAGeS : Images, Modélisation, Apprentissage, Géométrie et Statistique

Séminaire du 27/06/2014

De Équipe IMAGeS : Images, Modélisation, Apprentissage, Géométrie et Statistique
Aller à la navigation Aller à la recherche

vendredi 27 juin 2014, 10h30

Fourth Order Tensor Statistical Models for DW-MRI: Application to population comparison

Conférencier : Theodosios Gkamas

Résumé : Early diagnosis, prognosis and patient's monitoring for neurodegenerative diseases are crucial tasks in neuroscience. Detecting areas of the brain where the difference between two groups of interest (normal, abnormal) is statistically significant is a solution to these tasks. Knowing the T2 model's limitations to describe crossing fibers, we select a T4 model and we propose a specific statistical group comparison. Moreover, we choose to eliminate the curse of dimensionality, by calculating robust statistics in a reduced sub-manifold of the initial tensor's space.