Preconditioning linear systems from deformable 3D medical image registration using tensor train format

Justin Buhendwa Nyenyezi (Speaker)

Activity: Talk or presentation typesOral presentation

Description

In medical image analysis, deformable 3D image registration takes a relatively long time, especially for non-parametric transformations, for which the computing time may be quite troublesome and not even feasible for some clinical applications. Modelled as a variational problem, this registration problem needs to solve a sequence of linear systems during the optimization process. Much of the time is spent in the solution of these linear systems. Indeed, although these systems are sparse and structured, they are very large and ill conditioned. In this talk, we present and study a preconditioning technique to accelerate the solution of these linear systems using tensor train format. Appropriate preconditioners in tensor train format offer a good compromise between complexity and precision.
Period2 Feb 20173 Feb 2017
Held atORBEL: Belgian Operational Research (OR) Society, 31st conference, Belgium
Degree of RecognitionNational

Keywords

  • large linear systems, 3D medical images, registration, preconditioning, tensor train, optimization