Meeting Banner
Abstract #1758

Can the Two-Compartment Model for DCE-MRI Fit Contrast Concentration Curves Accurately Over a Heterogeneous Region of Interest?

Xiaobing Fan1, Aytekin Oto1, Gregory S. Karczmar1

1University of Chicago, Chicago, IL, United States

The nonlinearity of two-compartment model (TCM) of dynamic contrast-enhanced MRI (DCE-MRI) was studied using computer simulations and an example of clinical prostate DCE-MRI. Even when each individual contrast media concentration v.s. time curve (C(t)) within a region of interest (ROI) satisfies the TCM, the averaged C(t) over the ROI may not satisfy the TCM. Our computer simulation results showed that the Ktrans derived from fitting the averaged C(t) was 20% to 30% smaller than average of all of the Ktrans values for each pixel in the ROI. This is substantial error, and the error for many individual patients will be even higher, leading to errors in diagnosis.

Keywords

accuracy accurately agent approximate arterial audience becomes benefit best bolus cancer cancers characterize clinical closely commonly compartment computer concentration constant contrast curve curves decreased derived detection diagnosis diagnostic distribution dominant drawn dynamic enhanced equation error errors evaluate even extract finally fits fitted fitting function generate generated generator gets goal good goodness heterogeneity heterogeneous important importantly individual injection input larger leads likely limits linearity many media model nonlinearity normally parker people physiological pixel pixels plays poor population practice preferred produce prostate radiology random randomly rather redistribution reduced relatively resolution role sampled satisfies satisfy satisfying selected sensitivity significantly simple simulated simulations smaller solutions source specificity step steps still suggesting systematic target temporal thereafter tissue took trans transfer tumor underestimates underestimation uptake visual volume washout