Isao Tanaka, Presenter: Nothing to Disclose

Haruhiko Machida MD, Abstract Co-Author: Nothing to Disclose

Rika Fukui, Abstract Co-Author: Nothing to Disclose

Kazufumi Suzuki MD, Abstract Co-Author: Nothing to Disclose

Xiao Zhu Lin BA, Abstract Co-Author: Nothing to Disclose

Eiko Ueno MD, Abstract Co-Author: Nothing to Disclose

Yun Shen PhD, Abstract Co-Author: Employee, General Electric Company Researcher, General Electric Company

To investigate in a phantom the utility of model-based iterative reconstruction (MBIR) to improve image quality and reduce radiation dose in chest CT.

Using a 64-detector CT scanner (Discovery CT750 HD, GE), we scanned a chest phantom using parameters: section thickness, 0.625 mm; rotation speed, 0.5 sec; helical pitch, 1.375; and noise index (NI), 5-30 (5 step) HU; and we used filtered back projection (FBP) and MBIR to reconstruct axial and multiplanar reformation coronal images of 0.625-mm thickness. To quantify image noise on the axial images, we placed regions of interest within the lung (Pa) and pectoral muscle (Ma) at the apex level and in the lung (Pb) and liver (Hb) at the base level. Using a 3-point scale (1: poor; 3 excellent), 3 observers independently graded subjective noise with pulmonary and mediastinal windows and sharpness of pulmonary vasculature with a pulmonary window on the coronal images. We compared those results between FBP and MBIR by Wilcoxon signed-rank test and determined the highest NI by MBIR to obtain similar or better image quality to that obtained with our routine protocol (10-NI by FBP).

MBIR significantly reduced the objective noise for Pa, Ma, Pb, and Hb (14.5 ± 1.3, 15.6 ± 1.7, 17.3 ± 1.3, 20.4 ± 1.3 at 30-NI; 11.3 ± 0.5, 12.0 ± 0.6, 12.6 ± 1.4, 12.1 ± 0.4 at 10-NI) compared to FBP (42.7 ± 1.2, 47.9 ± 3.4, 66.7 ± 3.6, 87.5 ± 4.4 at 30-NI; 17.7 ± 1.9, 19.9 ± 0.5, 22.1 ± 1.4, 28.5 ± 1.1 at 10-NI) (P < 0.05 for all). MBIR also improved the subjective noise with the pulmonary and mediastinal windows and vascular sharpness (3.0 ± 0.0, 2.0 ± 0.0, 3.0 ± 0.0 at 30-NI; 3.0 ± 0.0, 3.0 ± 0.0, 3.0 ± 0.0 at 10-NI) compared to FBP (1.0 ± 0.0, 1.0 ± 0.0, 1.0 ± 0.0 at 30-NI; 2.0 ± 0.0, 2.0 ± 0.0, 2.0 ± 0.0 at 10-NI) (P < 0.05 for all). Image quality was better by MBIR, even at 30-NI, than by FBP at 10-NI.

Compared to our routine protocol for chest CT, use of MBIR may improve image quality and still reduce radiation dose by 67%.

MBIR may dramatically improve image quality with the same radiation dose and reduce radiation dose with similar or better image quality than that of FBP in chest CT.

Tanaka, I, Machida, H, Fukui, R, Suzuki, K, Lin, X, Ueno, E, Shen, Y, Improved Image Quality and Reduced Radiation Dose in Chest CT Using Model-based Iterative Reconstruction: Phantom Experiment.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11034569.html