Publication Date:
2013-10-24
Description:
Purpose: To compare the low-contrast detectability and image quality of computed tomography (CT) at different radiation dose levels reconstructed with iterative reconstruction (IR) and filtered back projection (FBP). Materials and Methods: A custom liver phantom with 12 simulated hypoattenuating tumors (diameters of 5, 10, 15, and 20 mm; tumor-to-liver contrast values of –10, –20, and –40 HU) was designed. The phantom was scanned with a standard abdominal CT protocol with a volume CT dose index of 21.6 mGy (equivalent 100% dose) and four low-dose protocols (20%, 40%, 60%, and 80% of the standard protocol dose). CT data sets were reconstructed with IR and FBP. Image noise was measured, and the tumors’ contrast-to-noise ratios (CNRs) were calculated. Tumor detection was independently assessed by three radiologists who were blinded to the CT technique used. A total of 840 simulated tumors were presented to the radiologists. Statistical analyses included analysis of variance. Results: IR yielded an image noise reduction of 43.9%–63.9% and a CNR increase of 74.1%–180% compared with FBP at the same dose level ( P 〈 .001). The overall sensitivity for tumor detection was 64.7%–85.3% for IR and 66.3%–85.7% for FBP at the 20%–100% doses, respectively. There was no significant difference in the sensitivity for tumor detection between IR and FBP at the same dose level ( P = .99). The sensitivity of the protocol at the 20% dose with FBP and IR was significantly lower than that of the protocol at the 100% dose with FBP and IR ( P = .019). Conclusion: As the radiation dose at CT decreases, the IR algorithm does not preserve the low-contrast detectability. © RSNA, 2013 Supplemental material: http://radiology.rsna.org/lookup/suppl/doi:10.1148/radiol.13122349/-/DC1
Keywords:
Quality Assurance/Quality Improvement, Computed Tomography
Print ISSN:
0033-8419
Electronic ISSN:
1527-1315
Topics:
Medicine
Permalink
