Applied Krylov subspace algorithms in CT using the TIGRE toolbox #

Ander Biguri, Malena Sabate Landman, Sepideh Hatamikia, Richard Boardman, John Aston, Carola-Bibiane Schonlieb

11:10 Tuesday in 2Q48.

Part of the Advances in applied numerical linear algebra and its applications session.

Abstract #

Computed Tomography (CT) is the most widely used imaging technique in medicine, and it is starting to be a common tool for scientific discovery and industrial non destructive testing. While sufficiently sampling an object may mitigate the ill-possedness of the inverse problem that is CT image reconstruction many applications can not afford such sampling, for example medical CT or high throughput industrial CT. Algebraic iterative methods have been quite introduced into the CT field already, but fast and efficient methods that are feasible to use at the scales of current CT problems (e.g. 2000^3 images) are still a challenge.

In this talk, we introduce the application of Krylov subspace methods in the 3D CT context. These methods show fast convergence compared to more traditional iterative methods, which is a desirable property in high throughput CT imaging. The methods are presented in the context of the TIGRE toolbox, a toolbox for seamless CT reconstruction. We show their impact in both medical and industrial CT datasets.