Dose-wise scanning in visceral computed tomography angiography: a phantom study.

ABSTRACT

OBJECTIVES: The aim of this study was to analyze the influence of the tube current-time product in multidetector computed tomography angiography on the accuracy of stenosis quantification in a phantom model of occlusive vessel disease. MATERIALS AND METHODS: Stenosed pelvic and visceral arteries were simulated using acrylic tubes (inner diameter small, 4.0 mm; large, 6.5 mm) filled with plaque material (epoxy resin, hydroxylapatite, glass bubbles) to create different degrees of stenosis and plaque composition (calcified plaques, >1000 Hounsfield units [HU]; soft plaques, ∼50 HU; inhomogeneous plaques, 50-1000 HU). The lumen was filled with water-diluted contrast material (Iomeprol 400; Bracco Imaging, Konstanz, Germany) to increase the density to 350 HU. The vessel phantoms were inserted in an Alderson phantom and imaging was conducted on a 64-slice MDCT (Somatom Definition, Siemens, Forchheim, Germany; collimation, 0.6 mm; reconstructed slice thickness, 1 mm; 120 kVp) using 8 different image acquisition protocols (IAPs), with reference tube current-time products (IQualRef) ranging between 20 and 280 mAs (IAP20-IAP280). The signal-to-noise ratio was calculated for each IAP. The measured luminal area within a stenosis was correlated to the known value using the Kappa-Lin test (κLin). A decrease of 10% of the maximum achievable correlation was defined as substantial. The sensitivity and specificity of hemodynamically relevant stenoses (>50%) were computed. For all IAPs, the effective dose was measured with thermoluminescence dosimetry and calculated with CTEXPO 2.0 (ICRP103). RESULTS: The measured effective dose ranged from 0.8 to 10.7 mSv. The calculated effective dose was approximately 10% lower for each IAP (0.7-9.8 mSv). A total of 2592 stenosis measurements were performed. In large vessels, the correlation was almost perfect for IAP80 to IAP280 (κLin = 0.91-0.95). In comparison, overall correlation was inferior in small vessels and was substantial for IAP280 to IAP120 (κLin = 0.89-0.82). Overall, the best correlation was observed in calcified (κLin = 0.95) and soft (κLin = 0.93) plaques as compared with inhomogeneous (κLin = 0.89) plaques. A substantial decrease in the correlation was observed below IAP100 for the large vessel phantoms and IAP120 for the small vessel phantoms. The sensitivity of hemodynamically relevant stenoses was 90% to 99% for IAP20 to IAP280 and both vessel diameters, whereas the specificity decreased from 91% (IAP280) to 31% (IAP20) for the large vessel phantoms and from 81% to 25%, respectively, for the smaller vessel phantoms. CONCLUSION: In large (>6.5 mm) vessel phantoms that simulate pelvic and renal arteries, representing a high-contrast scenario, a substantial dose reduction is feasible as compared with established abdominal imaging protocols. In smaller vessel phantoms that represent intestinal arteries, the quality of luminal delineation is already limited because of the spatial resolution. Therefore, an increase in image noise can only be accepted to a smaller degree and the potential dose reduction is limited but still substantial.