Harmonized low-dose computed tomographic protocols for quantitative lung imaging using dose modulation and advanced reconstructions.

BACKGROUND: Quantitative computed tomography (QCT) lung imaging is employed in many multi-center studies. Standardized protocols have used fixed volumetric CT dose index (CTDIvol) adjusted for body mass index to minimize dose while accounting for participant size. Dose modulation and iterative/deep-learning reconstruction (IR/DLR) offer new opportunities for QCT standardization for a multi-center protocol. PURPOSE: To develop harmonized reduced dose lung QCT protocols implementing dose modulation and IR/DLR in the context of the Precision Intervention for Severe Asthma (PrecISE) multi-center study. METHODS: A low-dose protocol was first developed on one state-of-the-art scanner having similar quantitative characteristics to a widely used standard-dose protocol as a reference for establishing harmonized protocols for a range of CT systems across four major manufacturers and ten sites. An anthropomorphic chest phantom with outer chest plates (LUNGMAN Chest Phantom, Kyoto; 43.5 cm left-right, 22.9 cm anterior-posterior) and custom inserts containing differently attenuating materials was imaged using varying dose modulation and IR/DLR settings. Hounsfield Unit (HU), standard deviations (SD), and coefficient of variation (CoV) in multiple density standards, including standardized foams, lung tissue, air, and water were compared for measures of accuracy, noise, and precision. The in-plane and z-direction modulation transfer functions (MTF) were also derived from a cubic insert. Purpose-built segmentation software (Pulmonary Analysis Software Suite, PASS) assured sampling of similar regions of interest. Final protocols included dose modulation-IR/DLR combinations yielding target low-dose CTDIvol, which minimized HU mean differences and SD, and maximized MTF compared to the reference-standard. RESULTS: The low-dose protocols achieved a mean CTDIvol reduction of 54% ± 7% (range 42%-70%) compared with the current standard-dose (SPIROMICS and MESALung). Compared to the reference-standard, mean HU difference was 12.0 ± 9.2 HU (range 0.4-28.5 HU) for air and 1.9 ± 1.3 HU (range 0.0-4.5 HU) for water inserts across the other nine low-dose protocols, and HU SD was lower in nine of ten low-dose protocols compared to standard-dose. HU CoV for all 10 low-dose protocols were near 0 for air and ranged 2.3-33.4 for water. MTF measurements were 2.71-4.22 and 4.02-7.13 cycles/cm for 50% and 20% cutoffs, respectively, compared with 3.17-3.50 and 4.68-5.54 cycles/cm for standard-dose. CONCLUSION: We provide harmonized low-dose QCT protocols using manufacturers' current dose modulation and IR/DLR techniques to reduce radiation dose by up to 70% and broadly maintain measurement accuracy and precision suitable for multi-center studies.