Effect of computed tomography dose on quantitative measurement and automated segmentation of airway tree

A fully automated computerized scheme that can segment an airway depicted on computed tomography (CT) images has certain advantages, particularly its simplicity and reliability. However, the CT scanning dose may affect the obtained measurement. In this study, we investigate whether the low-dose CT can be used to quantitative analysis of airway dimensions, which were measured by a computerized scheme. The study comprises twelve lung chest CT examinations acquired from three dogs with different beam currents of 30, 50, 100, and 200 mAs. At each dose, the lumen diameters, wall area, and lumen area of the matched airways (n = 82) were measured by this scheme. The bronchial segmentation ability and the measurement variability at individual dose were compared to that obtained at 200 mAs. Finally, we found that the CT dose impacts the measurement variability of the airway diameter (P <0.01) and of the percentage of wall area (WA%) (P <0.05). To obtain the same image quality, a CT dose of no less than 50 mAs should be selected since the percentage of leakage and/or obstruction bronchus at 30 mAs was significantly different from that at 200 mAs (P <0.00). In conclusion, the airway tree segmentation algorithms by the fully automated computerized scheme are feasible on low-dose CT at doses as low as 50 mAs. However, the scanning dose does affect the obtained measurements.