Comparison of image quality and visibility of normal and abnormal findings at submillisievert chest CT using filtered back projection, iterative model reconstruction (IMR) and iDose4™

Objective: To compare both image quality and visibility of normal and abnormal findings at submillisievert chest CT (smSv-CT) using filtered back projection (FBP) and the two different iterative reconstruction (IR) techniques iterative model reconstruction (IMR) and iDose(4)(TM) Materials and methods: This institutional review board approved study was based on retrospective interpretation of clinically indicated acquired data. The requirement to obtain informed consent was waived. 81 patients with suspected pneumonia underwent smSv-CT (Brilliance iCT, Philips Healthcare; mean effective dose: 0.86 +/- 0.2 mSv). Data were reconstructed using FBP and two different IR techniques iDose(4)(TM) and IMR (Philips Healthcare) at various iteration levels. Objective image noise (OIN) was measured. Two experienced readers independently assessed all images for image noise, image appearance and visibility of normal anatomic and abnormal findings. A random intercept model was used for statistical analysis. Results: Compared to FBP and iDose(4)(TM) IMR reduced OIN up to 88% and 72%, respectively (p <0.001). A mild blotchy image appearance was seen in IMR images, affecting diagnostic confidence. iDose(4)(TM) images provided satisfactory to good image quality for visibility of normal and abnormal findings and were superior to FBP (p < 0.001). IMR images were significantly inferior for visibility of normal structures compared to iDose(4)(TM) while being superior for visibility of abnormal findings except for reticular pattern (p < 0.001). Conclusion: IMR results for visibility of normal and abnormal lung findings are heterogeneous, indicating that IMR may not represent a priority technique for clinical routine. iDose(4)(TM) represents a suitable method for evaluation of lung tissue at submillisievert chest CT. (C) 2016 Elsevier Ireland Ltd. All rights reserved.