Physical characterization of a new CT iterative reconstruction method operating in sinogram space

Caterina Ghetti,Giulio Serreli,Francesca Palleri,Livia Ruffini,Ornella Ortenzia

doi:10.1120/jacmp.v14i4.4347

Abstract

Recently a new iterative reconstruction algorithm named Sinogram Affirmed Iterative Reconstruction (SAFIRE) has been released by Siemens. This algorithm works in the raw data domain with noise reduction as main purpose, providing five different strengths. In this study, the effect of SAFIRE on image quality has been investigated using selected phantoms and a comparison with standard filtered back projection (FBP) has been carried out. The following quantitative parameters have been evaluated: image noise, impact of different reconstruction kernels on noise reduction, noise power spectrum (NPS), contrast‐to‐noise ratio (CNR), spatial resolution, and linearity and accuracy of CT numbers. The influence of strengths on image quality parameters has also been examined. Results show that image noise reduction is independent of reconstruction kernel and strongly related to the strength of SAFIRE applied. The peak of NPS curve for SAFIRE reconstructions is shifted towards low frequencies; this effect is more marked at higher levels of strength. Contrast‐to‐noise ratio is always improved in SAFIRE reconstruction and increases with higher strength. At different dose levels SAFIRE preserves CT number accuracy, linearity, and spatial resolution, both in transversal and coronal planes. These results confirm that SAFIRE allows for image noise reduction with preserved image quality. First clinical data to validate this phantom analysis and confirm that commercially available iterative algorithms can play an effective role in dose containment.PACS number: 87.57.Q

Highlights

The use of iterative reconstruction algorithms in computed tomography (CT) has become a crucial issue for dose reduction in CT examinations
The main advantage of iterative algorithms as opposed to filtered back projection (FBP) is the incorporation of physical models, which allow for CT studies at reduced doses with preserved image quality and low levels of image noise.[1,2,3,4,5,6,7,8,9] The most important iterative reconstruction methods and the solutions introduced by CT manufacturers have been recently reviewed.[10]
Standard deviation reduces as the inverse square root of CTDIvol as expected for FBP, which means that the Poisson distribution of noise is conserved in this iterative reconstruction method

Summary

Introduction

The use of iterative reconstruction algorithms in computed tomography (CT) has become a crucial issue for dose reduction in CT examinations. Previous clinical studies exploring SAFIRE reconstruction have measured parameters such as contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR), and provided a subjective assessment of image quality. Most of these studies report an image noise reduction without loss of diagnostic information, and consistent dose reduction.[12,13,14,15,16,17,18,19,20,21,22]. The aim of this study is to evaluate the SAFIRE algorithm using image quality parameters measured on phantoms in order to describe the effect of iterative reconstruction with objective metrics. Noise power spectrum (NPS), CNR, kernel impact on noise reduction, linearity and accuracy of CT numbers, and both transverse and coronal spatial resolution have been investigated using dedicated phantoms, and results have been compared to traditional FBP

Objectives

Methods

Results

Conclusion