Basics of iterative reconstruction methods in computed tomography: A vendor-independent overview

Abstract

Over the past two decades, technical innovations in computed tomography (CT) have constantly extended its spectrum of clinical applications and made new radiodiagnostic applications accessible. At the same time, concerns have arisen with respect to the radiation exposure to the patients caused by CT examinations. In order to address this issue, different strategies for radiation dose reduction in CT have been introduced, spanning technical approaches as well as specific examination techniques applied in clinical practice, such as reduced-dose CT. Developed technical approaches for reducing radiation dose in CT by improvements of CT scanner hardware and acquisition mechanisms, however, have not been sufficient to address the degradation of image quality caused by increasing noise and susceptibility to artifacts inherent to reduced-dose CT acquisitions. Recent advances in computing power have enabled the development of software-based methods for iterative image reconstruction (IR) in CT enabling simultaneous reduction of image noise and improvement of overall image quality. Thereby, IR allows for dose reduction by reconstruction of low-noise image data from intrinsically noisy reduced-dose CT acquisitions, thereby preserving diagnostic image quality equivalent to current clinical standards.This review provides an overview of the underlying basic principles of iterative image reconstruction methods currently available for and applied in CT imaging, independent of vendor-specific details regarding algorithms and implementations. It discusses potential strengths and weaknesses of these CT image reconstruction techniques in view of their application in clinical routine, especially in view of the potential of IR for noise and artifact reduction as well as for radiation dose reduction. Furthermore, the effect of statistical (hybrid) and model-based IR methods on image quality are exemplarily illustrated in comparison to filtered back projection (FBP) traditionally used for image reconstruction in CT.