Wash-out of hepatocellular carcinoma: quantitative region of interest analysis on CT.

INTRODUCTION This study aims to determine if the quantitative method of region-of-interest (ROI) analysis of lesion attenuation on CT may be a useful adjunct to the conventional approach of diagnosis by visual assessment in assessing tracer wash-out in hepatocellular carcinomas. MATERIALS AND METHODS From a surgical database of 289 patients from 2 institutions, all patients with complete surgical, pathological and preoperative multiphasic CT scans available for review were selected. For each phase of scanning, HU readings of lesion obtained (Lesion(arterial), Lesion(PV) and Lesion(equilibrium)) were analysed using receiver operating curves (ROC) to determine the optimal method and cut-off value for quantitative assessment of tumour wash-out (Lesion(arterial - equilibrium), Lesion(PV - equilibrium) or Lesion(peak - equilibrium)). RESULTS Ninety-four patients with one lesion each met the inclusion criteria. The area under the curve (AUC) values for Lesion(arterial - equilibrium) (0.941) was higher than the AUC for Lesion(pv - equilibrium) (0.484) and for Lesion(peak - equilibrium) (0.667). Based on ROC analysis, a cut-off of 10HU value for Lesion(arterial - equilibrium) would yield sensitivity and specificity of 91.5% and 80.9%, respectively. ROI analysis detected 9/21 (42.9%) of lesions missed by visual analysis. Combined ROI and visual analysis yields a sensitivity of 82/94 (87.2%) compared to 73/94 (77.7%) for visual analysis alone. CONCLUSION Using a cut-off of 10 HU attenuation difference between the arterial and equilibrium phases is a simple and objective method that can be included as an adjunct to visual assessment to improve sensitivity for determining lesion wash-out on CT.