Devin Traer Caywood MD, Presenter: Nothing to Disclose

Ben Eugene Paxton MD, Abstract Co-Author: Nothing to Disclose

Danielle Seaman MD, Abstract Co-Author: Nothing to Disclose

Lynne Michelle Hurwitz MD, Abstract Co-Author: Research Grant, Siemens AG Research Grant, General Electric Company

Rendon C. Nelson MD, Abstract Co-Author: Consultant, General Electric Company Research support, Nemoto Kyorindo Co, Ltd Research support, Bracco Group Research support, Becton, Dickinson and Company Speakers Bureau, Siemens AG Royalties, Lippincott, Williams & Wilkins

To evaluate the quality of aortic CT angiographic images reconstructed with filtered back projection (FBP), adaptive statistical iterative reconstruction (ASIR), and model-based iterative reconstruction (MBIR) scanned at progressively lower radiation doses.

A healthy 56.1 kg Yorkshire pig underwent successive thoracic arterial CT angiography on a Discovery CT 750HD MDCT scanner (GE Healthcare) at progressively lower radiation doses while under general anesthesia. The pig was scanned at 700, 400, 200, 100, and 50 mA at 120, 100, and 80 kVp, for a total of 15 scans. Each scan was reconstructed with FBP, ASIR (50% blend) and MBIR. Noise (SD) and contrast-to-noise ratios (CNR) were calculated from ROIs over the aorta and paraspinous muscle. In addition, selected axial and oblique sagittal images were scored subjectively from 1 to 5 for both aortic wall visibility (1-aortic wall margin difficult or impossible to asses; 5-superior aortic wall visualization) and for overall image quality (1-poor quality, nondiagnostic; 4-good image quality typical of expected clinical practice; 5-superior image quality with no significant noise or artifact).

Averaged across all doses, MBIR reduced noise by 73% and improved CNR by 227% compared to FBP; MBIR reduced noise by 69% and improved CNR by 107% compared to ASIR.  The effect was more pronounced in higher noise settings. At 120 kVp/400 mA, MBIR reduced noise by 50% compared to FBP and 30% compared to ASIR. At 80 kVp/50 mA MBIR reduced noise by 81% compared to FBP and 74% compared to ASIR. Subjective improvements in image quality were only noted in higher noise settings. At 120 kVp/400 mA, readers scored FBP, ASIR, and MBIR images equally. At 120 kVp/50 mA, readers scored FBP and ASIR images as nondiagnostic, while MBIR images were considered of moderate quality but remained diagnostic. At very low doses (80 KvP/50 mA), FBP, ASIR, and MBIR were nondiagnostic.

MBIR reduced image noise compared to ASIR and FBP across all dose levels. Reductions in noise were significantly greater in lower dose settings. While quantitative data suggest that large dose reductions may be possible, qualitative data is more tempered, particularly at very low doses.

Subjective data suggest that, while still beneficial, the dramatic dose reductions for MBIR implied by quantitative data may be tempered by the waxy appearance of MBIR images at low doses.

Caywood, D, Paxton, B, Seaman, D, Hurwitz, L, Nelson, R, Effects of Iterative Reconstruction on Image Noise and Contrast-to-Noise Ratios for Low Dose CT Angiography of the Thoracic Aorta in a Swine Model.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12035019.html