Comparison of different kinetic models for dynamic 18F-FDG PET/CT imaging of hepatocellular carcinoma with various, also dual-blood input function.

A kinetic modeling analysis was performed on hepatocellular carcinoma (HCC) as well as healthy liver tissue regions from dynamic FDG positron emission tomographys/computer tomography (PET/CT) cans. On basis of image derived input function from hepatic artery and portal vein, various kinetic models were compared among each other in order to check whether HCC can be classified and differ from healthy tissue within the kinetic parameters. 14 HCC and 10 healthy liver regions from FDG PET/CT scans of ten patients were analyzed from their time activity curves (TACs) were extracted from the PET dynamic images. Also the hepatic artery and the portal vein were delineated in the fused PET/CT images, which were used as input functions. Four kinetic models were applied to the TACs, using both or only one input function. Results were analyzed with several information criteria according to Akaike and Schwartz as well as be the F-Test. The paired student's t-test was used to determine the differences between HCC and healthy regions. All applied models revealed significant differences of p   <  0.01 of k 3 between HCC and healthy liver and three out of four models produced almost identical values for k 3  =  0.03 min-1 and K i   =  0.03 min-1. According to the information criteria tests, a simple two-tissue model with only a venous input function is clearly preferred in case of HCC TACs. After dividing all HCC regions into two groups having a low and a high HCC-to-healthy ratio, respectively, this model also showed significant differences of k 3 between these two groups. In conclusion, results indicate that the portal vein is sufficient to describe kinetic FDG processes in HCC and healthy liver regions.