Abstract

Motivation: Accurate quantification of pharmacokinetic parameters in dynamic contrast-enhanced (DCE) MRI requires high temporal resolution, often reached through non-cartesian sampling patterns that oversample the center of k-space (e.g. radial, spiral, PROPELLOR). In pharmacokinetic models, image contrast is assumed to be formed instantly at discrete time-points. However, in acquisitions oversampling the k-space center, the signal per time-frame becomes an average over acquisition time. Goal(s): To correct for the time-averaged signals. Approach: We proposed a modification to DCE modeling and tested it in simulations and in-vivo. Results: Modern sampling patterns predominantly affect the pharmacokinetic parameter estimates for longer sampling times (>8s) per DCE frame. Impact: We verified that for short acquisitions per frame (<8s) per DCE-frame, conventional Toft's modeling is sufficient. However, for longer sampling times (>8s) per DCE frame, our time-averaged extended Toft's model is needed for accurate estimations of pharmacokinetic parameters.