Katsunori Miyata RT, Presenter: Nothing to Disclose

Yukihiro Nagatani MD, Abstract Co-Author: Nothing to Disclose

Masashi Takahashi MD, Abstract Co-Author: Nothing to Disclose

Satoru Matsuo, Abstract Co-Author: Nothing to Disclose

Mitsuru Ikeda MD, Abstract Co-Author: Nothing to Disclose

Norihisa Nitta MD, Abstract Co-Author: Nothing to Disclose

Hideji Otani MD, Abstract Co-Author: Nothing to Disclose

Shinichi Ota MD, Abstract Co-Author: Nothing to Disclose

Masatake Imai, Abstract Co-Author: Nothing to Disclose

Kiyoshi Murata MD, Abstract Co-Author: Nothing to Disclose

To compare detectability of simulated ground glass nodules (GGND) on chest ditigal tomosynthesis (CDT) among 12 images obtained at 6 radiation exposure levels using 2 different reconstruction algorithms and analyze the influence of nodular size and its computed tomography attenuation value (CTAV) on GGND

Seventy-four simulated GGNs (5, 8 and 10 mm in a diameter/ -630 and -800 Hounsfield of Unit (HU) in CTAV) were placed in the chest phantom with reproduced peripheral pulmonary structures in 14 different patterns of nodular distribution. In each of the 14 distribution patterns, 12 sets of reconstructed coronal images were obtained using CDT (SONIALVISION Safire, Shimadzu, Kyoto, Japan) with 6 different radiation level: 120kV-10mA/20mA/80mA/160mA, 100kV-80mA and 80kV-320mA (effective dose : 0.08/0.16/0.65/1.30, 0.39 and 0.77mSv in standard body habitus, respectively) with and without iterative dose reduction algorithm (IRA). Ten radiologists independently recorded GGN presence and their locations by continuously-distributed rating in total 168 sets of images. Receiver-operating characteristic (ROC) analysis was used to compare GGND of the 12 images in total and detection sensitivities (DS) of GGN were compared among the 12 images in subgroups classified by their nodular diameters and CTAV.

In total, GGND in images at 120kV-80mA with IRA was similar to that at 120kV-160mA with IRA, as area under ROC curve was 0.79±0.03 and 0.80±0.03, respectively, and higher than the other 6 images obtained at 120kV. (p<0.05) DS of GGN with the diameter of 8mm and -630 HU in CTAV was 73.5±6.0 % in images at 120kV-10mA without IRA and similar to those in the other 11 images. (p=0.157) DS of GGN with the diameter of 10mm and -800 HU in CTAV was 56.3±11.9 % in images at both 120kV-80mA and 120kV-160mA with IRA and higher than the other 4 images obtained at 120kV without IRA. (p<0.05)

CDT demonstrated sufficient GGND for less attenuated nodules with the diameter of 8mm or more even in the lowest radiation level (0.08mSv) and improved DS of GGN for more attenuated nodules with the diameter of 10mm at submilli-Sv with IRA.

CDT has a sufficient potential to be used for detection of pure GGN and IRA synergistically boosts its ability.

Miyata, K, Nagatani, Y, Takahashi, M, Matsuo, S, Ikeda, M, Nitta, N, Otani, H, Ota, S, Imai, M, Murata, K, A Phantom Study for Ground Glass Nodule Detection by Chest Digital Tomosynthesis with Iterative Reconstruction Algorithm: Influence of Radiation Dose and Nodular Characteristics.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14004859.html