Dose reduction in computed tomography using adaptive statistical iterative reconstruction (ASIR). A phantom study

Abstract

Introduction Computed Tomography (CT) has been an invaluable tool in medical diagnosis, but its increasing use has been also responsible for a significant increase in population radiation exposure. Iterative reconstruction algorithms have been recently introduced in clinical practice, in an attempt to decrease CT radiation dose. Purpose The purpose of this work was to investigate quantitatively the relationship between image quality and dose in CT and to establish the effectiveness of ASIR in reducing radiation dose, without compromising image quality. Materials and methods A CT image quality phantom was scanned in a 64-slice GE Optima CT660 scanner, using a typical head acquisition protocol with mA modulation, at 8 different Noise Index levels, corresponding to a CTDI vol range of 18.7–91.2 mGy. Scanned data were used to reconstruct 5 mm slices, using (a) only FBP and (b) FBP blended with 3 ASIR levels (20%–40%–60%). Mean CT number, image noise (SD) and CNR were measured in appropriately selected slices in all reconstructed images by three different observers. Results A reduction in image noise by 35–55% for the same radiation dose was observed in the ASIR reconstructed images compared to FBP, whereas images with similar noise levels could be produced using ASIR with 55–75% less dose compared to FBP. Similarly, ASIR with at least 40% blending, was found to increase CNR by 20–45% compared to FBP at the same radiation dose, and could produce images with comparable CNR with 20–60% less dose compared to FBP. Conclusion ASIR can be very effective in reducing the radiation dose in CT, while preserving image quality.