Low dose multi-detector CT of the chest (iLEAD Study): Visual ranking of different simulated mAs levels

Purpose: Detailed evaluation of the lung parenchyma might be impaired by use of low dose CT as image noise increases and subsequently image quality decreases. The aim of our study was to determine the accuracy of visual perception of differences in image quality and noise at low dose chest CT. Materials and methods: Forty-four patients suffering from emphysema underwent CT (Aquilion-16, 120 kV, 150 mAs, 1 mm-collimation). Original raw data were used for simulation of 10 different mAs settings from 10 mAs to 100 mAs in 10 mAs increments. Three representative hard copy images (carina, 4 cm above, 5 cm below) were printed for evaluation of lung parenchyma (high-resolution kernel, lung window) and mediastinum (soft-kernel, soft tissue window). Ranking of expected low mAs level was performed for lung and soft tissue separately based on visual perception by three-blinded chest radiologist independently. Results were compared to the real simulated mAs. Results: The accuracy for correct ranking of the original 150 mAs scan was 89% for lung and 86% for soft tissue while it was 99% for the simulated 10 mAs for both windows. In comparison to the lowest mAs a significant error increase was found for the lung at 60-100 mAs (with error increase of 30-47%) for reader-I; 60-100 mAs for (33-64%) for reader-II and 70-100 mAs (38-57%) for reader-III. For the soft tissue: 60-150 mAs (with error increase of 28-63%) for reader-I; 50-100 mAs (35-56%) for reader-II and 50-90 mAs (35-40%) for reader-III. Conclusion: Simulated dose levels below 60 mAs (=42 mAs_eff) were clearly differentiated from higher dose levels by all readers. Therefore, imaging doses could be lowered down to 60 mAs without a diagnostically relevant increase in noise impairing image quality.