Parameter Fits Using Measured Population AIF Vs. Literature AIF in DCE-MRI of Pancreatic Cancer Xenograft Model

Andrew C.H. Yung¹, Sylvia Ng^1,2, Jennifer Flexman¹, Jenny C.H. Tso¹, Donald Yapp^1,3, Piotr Kozlowski¹

We show here that perfusion parameter fits are improved when a population-averaged AIF is extracted from the DCE data in the current study, versus an assumed literature AIF (Lyng MRM 1998). DCE-MRI data were acquired in a xenograft mouse model of pancreatic cancer. Pixels were manually selected from the abdominal aorta from the contrast scans of study animals for inclusion into the population-averaged AIF. Chi-square reflects how closely the fits matched the data. In a representative tumour, using the literature AIF versus the measured population-averaged AIF produced higher chi-square in 85% of voxels (median increase in chi square was 38%).