Phase‐based vascular input function: Improved quantitative DCE‐MRI of atherosclerotic plaques

Purpose: Quantitative pharmacokinetic modeling of dynamic contrast‐enhanced (DCE)‐MRI can be used to assess atherosclerotic plaque microvasculature, which is an important marker of plaque vulnerability. Purpose of the present study was (1) to compare magnitude‐ versus phase‐based vascular input functions (m‐VIF vs ph‐VIF) used in pharmacokinetic modeling and (2) to perform model calculations and flow phantom experiments to gain more insight into the differences between m‐VIF and ph‐VIF. Methods: Population averaged m‐VIF and ph‐VIFs were acquired from 11 patients with carotid plaques and used for pharmacokinetic analysis in another 17 patients. Simulations, using the Bloch equations and the MRI scan geometry, and flow phantom experiments were performed to determine the effect of local blood velocity on the magnitude and phase signal enhancement. Results: Simulations and flow phantom experiments revealed that flow within the lumen can lead to severe underestimation of m‐VIF, while this is not the case for the ph‐VIF. In line, the peak concentration of the m‐VIF is significantly lower than ph‐VIF (p < 0.001), in vivo. Quantitative model parameters for m‐ and ph‐VIF differed in absolute values but were moderate to strongly correlated with each other [Ktrans Spearman's ρ > 0.93 (p < 0.001) and vp Spearman's ρ > 0.58 (p < 0.05)]. Conclusions: m‐VIF is strongly influenced by local blood velocity, which leads to underestimation of the contrast medium concentration. Therefore, it is advised to use ph‐VIF for DCE‐MRI analysis of carotid plaques for accurate quantification.