CT Volumetrics of Primary Liver Cancers

Abstract
A new computer algorithm is described for liver and tumor volume determinations for patients with hepatoma and primary hepatic cholangiocarcinoma. The algorithm is based on global histograms of CT numbers of the liver and primary liver cancers. The algorithm includes computer-assisted definition of the liver boundary in each CT slice. Liver and tumor volumes of 10 patients calculated by the histogram method were compared with volumes obtained from CT slices that were manually contoured by experienced observers. A correlation coefficient of 0.995 was determined for these two methods of volume computations. Mean values of the differences in volumes obtained by the two methods were 6.7 and 8.0% for the liver and tumor, respectively. The computer algorithm was tested on CT scans for an additional 46 patients by highlighting regions corresponding to normal liver and tumor tissues in each CT slice and determined to be accurate by experienced observers. The computer software is being used clinically to assess tumor response in a new treatment program for primary liver cancers that includes radiolabeled antibodies.