RESUMO
INTRODUCTION: The purpose of this work was to determine the exposure-optimised slice thickness for hepatic lesion detection with CT. METHODS: A phantom containing spheres (diameter 9.5, 4.8 and 2.4 mm) with CT density 10 HU below the background (50 HU) was scanned at 125, 100, 75 and 50 mAs. Data were reconstructed at 5-, 3- and 1-mm slice thicknesses. Noise, contrast-to-noise ratio (CNR), area under the curve (AUC) as calculated using receiver operating characteristic analysis and sensitivity representing lesion detection were calculated and compared. RESULTS: Compared with the 125 mAs/5 mm slice thickness setting, significant reductions in AUC were found for 75 mAs (P < 0.01) and 50 mAs (P < 0.05) at 1- and 3-mm thicknesses, respectively; sensitivity for the 9.5-mm sphere was significantly reduced for 75 (P < 0.05) and 50 mAs (P < 0.01) at 1-mm thickness; sensitivity for the 4.8-mm sphere was significantly lower for 100, 75 and 50 mAs at all three slice thicknesses (P < 0.05). The 2.4-mm sphere was rarely detected. At each slice thickness, noise at 100, 75 and 50 mAs exposures was approximately 10, 30 and 50% higher, respectively, than that at 125 mAs exposure. CNRs decreased in an irregular manner with reductions in exposure and slice thickness. CONCLUSION: This study demonstrated no advantage to using slices below 5 mm thickness, and consequently thinner slices are not necessarily better.
