Model-based iterative reconstruction versus adaptive statistical iterative reconstruction in low-dose abdominal CT for urolithiasis.

OBJECTIVE: The purpose of this study was to determine whether there is added benefit to model-based iterative reconstruction as compared with adaptive statistical iterative reconstruction on low-dose abdominal CT in the clinical context of known or suspected urolithiasis.
MATERIALS AND METHODS: Fifty-three consecutive patients (35 men, 18 women; mean [SD] age, 52.3 ± 16.6 years) underwent unenhanced low-dose abdominal CT for detection or follow-up of urinary tract stones. Ureteral definition was evaluated subjectively by two blinded readers who scored it from 1 (excellent definition) to 4 (not distinguishable) and objectively by calculating contrast-to-noise ratio (CNR) for soft tissue and fat on a standard 40% adaptive statistical iterative reconstruction and on the corresponding model-based iterative reconstruction series. The position, maximal density, and diameter on the axial plane of stones were also evaluated on both series.
RESULTS: There was an almost perfect agreement (κ = 0.872) between readers for subjective evaluation of ureteral definition. The ureteral definition was significantly better for the model-based iterative reconstruction series (mean [SD] score, 1.998 ± 0.839) compared with adaptive statistical iterative reconstruction (mean score, 2.536 ± 0.799) (p < 0.0001). Mean CNR was significantly better on the model-based iterative reconstruction (17.82 ± 4.84) compared with adaptive statistical iterative reconstruction (6.066 ± 1.985) (p < 0.0001). Sixty-nine stones were found in total. Their maximal density measured from model-based iterative reconstruction was significantly higher (754.4 ± 376.5 HU) than that measured from adaptive statistical iterative reconstruction (559.4 ± 352.4 HU) (p < 0.0001). Size of stones was overevaluated on model-based iterative reconstruction (mean diameter, 4.91 ± 2.61 mm) compared with adaptive statistical iterative reconstruction (4.52 ± 2.63 mm) (p < 0.0001).
CONCLUSION: Model-based iterative reconstruction of low-dose abdominal CT can offer significantly better ureteral definition than adaptive statistical iterative reconstruction, and its systematic use could thus be recommended. However, it has the tendency to systematically overevaluate the stones' densities and sizes.