248. Low contrast detectability comparison in computed tomography with iterative reconstruction

Abstract

Purpose To assess low contrast detectability (LCD) of two different iterative reconstruction (IR) algorithms: the ASiR and the new ASiR-V. Methods LCD was evaluated through a statistical method [1] . A water phantom was acquired with two different devices: a CT Discovery 750HD equipped with the ASiR and a CT Revolution equipped with the ASiR-V. Acquisitions were performed at CTDI vol levels ranging from 1 to 15 mGy and images were reconstructed with FBP and 3 different levels of IR (40%, 60% and 100%, for both algorithms). A 192 × 192 pixels ROI placed over 20 images was divided into subROIs from 2 × 2 to 24 × 24 pixels. Distribution of the mean pixel value of the subROIs was taken into account: normality was assessed using Jarque-Brera test and low contrast threshold ( LC th ) was calculated according to [1] and then averaged over the 20 slices. To take into account the intrinsic differences among the two scanners, LC th was normalized to the 4.4 mm insert with the FBP. Results LC th increases passing from FBP to ASiR-V depending on insert dimension and IR level. Considering the 4.4. mm insert, LC th improves of 11% and 27% for ASiR-V 40% and ASiR-V 100%, respectively. For a fixed level of IR, improvement is more evident for smaller objects, passing from 27% for the 2.2 mm insert to 8% for the 13.2 mm insert. ASiR-V presents an improvement in LC th for all doses and all levels of IR in comparison to ASiR. Improvement in detectability increases with dose for a fixed IR level; for a fixed dose level, LC th increases with increasing IR level. Conclusions ASiR-V improves low contrast detectability in comparison to ASiR.