Optimization of deformable registration using numerical tensor methods for 3D medical images

Justin Buhendwa Nyenyezi (Orateur)

Activité: Types de discours ou de présentationPrésentation orale

Description

We investigate and analyse numerical algorithms for 3D medical image registration problem. These algorithms include different versions of the Demons algorithms and the package of Flexible Algorithms for Image Registration (FAIR). Inspired by new tools from numerical tensor decomposition and low-rank approximation, we address a new formulation of the problem. This allows us to approximate data, operators and preconditionners in a low-rank tensor representation that may lead to a significant gain in memory storage and running time.
Période15 nov. 2016
Conservé àNamur Institute for Complex Systems
Niveau de reconnaissanceLocal