The Effect of Image Registration on Pharmacokinetic Parameter Extraction Using 3D DCE-MRI

Andrew Melbourne¹, Matthew Orton², David Collins², Dow-Mu Koh², Martin Leach², David Hawkes¹, David Atkinson¹

Residual patient motion between images in breath-hold DCE-MRI may compromise image. This work analyses results of a Progressive Principal Component Registration (PPCR) algorithm on six 3D DCE-MRI datasets. The datasets used here are analysed using a full pharmacokinetic analysis of the liver in order to extract pharmacokinetic parameters from each pixel. Inspecting the average change in the total model fit residual for all six datasets reveals an average decrease of 1.52.7% after direct fluid-based image registration and 154% after registration by PPCR. The PPCR algorithm has been shown to allow improved model fitting by reduction in the model-fit residuals.