CT and MRI in target delineation in primary hepatocellular carcinoma.

PURPOSE: To compare the quality of computed tomography (CT) and magnetic resonance imaging (MRI) in target delineation of hepatocellular carcinoma. PATIENTS AND METHODS: Thirty-one patients with hepatocellular carcinoma received CT and MRI (including diffusion-weighted imaging [DWI]) before surgery. The maximal measurement on axial imaging and pathologic examination were extracted for statistical analysis. RESULTS: CT, MRI, DWI-based tumour size correlated well with pathologic sizes, r=0.974, 0.969, 0.964 respectively. But Wilcoxon signed-ranked test showed that differences did exist. CT overestimated the tumour size by 2.9mm compared to pathology (95% CI: -13.2 to 7.4mm). The agreement of MRI-pathology seemed to be worse than CT, with a mean difference of 3.6mm (95% CI: -14.7 to 7.7mm). The worst agreement was in DWI-pathology, with a mean difference of 5mm (95% CI: -17.9 to 7.9mm). But significant difference was found neither between CT and MRI (P=0.477) nor between MRI and DWI (P=0.079). CONCLUSIONS: CT and MRI-based tumour size correlated well with pathologic size, but differences did exist. Most of the lesions were overestimated by CT and MRI. CT and MRI were similar in the guidance of target delineation, and DWI had added little value to MRI. A margin of 10mm around the gross tumour volume to become the clinical target volume is likely not sufficient.