Partial volume-related errors in parameters derived from dynamic contrast-enhanced MRI – implications for clinical trials

Introduction Tracer kinetic model parameters are frequently used in clinical trials to measure the efficacy of cancer therapeutics. Typically, the parameters are estimated for each voxel within a volume of interest (VOI) that is drawn around the tumour. The mean or median value is then calculated for the VOI and compared at different stages of the treatment. A simultaneous reduction in the tumour volume may also be observed. Voxels on the edge of the VOI are likely to include non-tumour tissue, and in this work we aim to investigate the influence of partial volume effects (PVEs) on the mean and median values as the tumour shrinks. This is of particular interest in the liver where the dual blood supply may lead to contrast agent concentration time courses that differ significantly from that of an arterial tumour rim. The liver is a common site of metastasis from colorectal cancer and there is also an increased incidence of hepatocellular carcinoma when cirrhosis is present. We use a software phantom generator to synthesise tumour and liver data sets to provide controlled test data with which to evaluate the influence of partial volume effects.