Stopping Rules for Algebraic Iterative Reconstruction Methods in Computed Tomography

TL;DR

This paper discusses four stopping rules for iterative algebraic reconstruction methods in computed tomography, aiming to optimize reconstruction quality while managing noise and computational efficiency.

Contribution

It introduces and illustrates four stopping rules specifically tailored for CT reconstruction, bridging a gap between inverse problems and practical CT applications.

Findings

Four stopping rules are described and illustrated for CT reconstruction.

The rules help balance reconstruction accuracy and noise influence.

Application to CT improves iterative reconstruction efficiency.

Abstract

Algebraic models for the reconstruction problem in X-ray computed tomography (CT) provide a flexible framework that applies to many measurement geometries. For large-scale problems we need to use iterative solvers, and we need stopping rules for these methods that terminate the iterations when we have computed a satisfactory reconstruction that balances the reconstruction error and the influence of noise from the measurements. Many such stopping rules are developed in the inverse problems communities, but they have not attained much attention in the CT world. The goal of this paper is to describe and illustrate four stopping rules that are relevant for CT reconstructions.