Abstract
The accurate quantitation of liver tumor burden and visualization of lesions in three dimensions (3D) can assist in treatment planning and can allow monitoring of therapy. Previous attempts have used CT and standard contrast media. Because the iodinated agents rapidly diffuse into tumors, usually effacing, and at time enhancing tumor edges, they decrease accuracy and make image segmentation difficult. CT portography suffers from flow related artifacts and does not allow the distinction of tumors from hemangiomas. Blood pool contrast is ideal in this setting since it enhances liver, liver vessels and hemangiomas, but not tumors, 'physiologically' splitting the image into normal and abnormal tissues. This ongoing study assesses the feasibility of this technique to visualize tumor and presents a scheme to automatically quantitate tumor volume. It utilized a rabbit VX2 liver tumor model and CT scanning shortly after the infusion of 3 ml/kg perflubron emulsion. Cut sections of the frozen carcass served as gross pathologic correlation. Images were imported onto a Sparc workstation, 3D reformatted and tumor and liver volume calculated. Histograms of pixel intensity clearly separated tumors from liver and liver from surrounding structures allowing the easy demarcation of tumor and liver margins.