Sensitivity of quantitative metrics derived from DCE MRI and a pharmacokinetic model to image quality and acquisition...

This study aims to investigate the sensitivity of quantitative metrics derived from dynamic contrast-enhanced (DCE) magnetic resonance imaging and a pharmacokinetic (PK) model to image quality and acquisition parameters.

A computer-synthesized DCE model that consisted of a large range of values of K(trans) (transfer constant of a paramagnetic contrast agent from blood to tissue), v(p) (fractional plasma volume), and k(ep) (back flux rate) was created to test the reliability of quantitative metrics derived from a standard PK model. Effects of the contrast-to-noise ratio (CNR), total acquisition time, and sampling interval on the stability and bias of the derived metrics were investigated.

The instability and bias of the estimated K(trans), v(p), and k(ep) values increased with sampling interval and decreased with increasing CNR. Total acquisition times had limited influence on the estimations of K(trans) and v(p) values, but increasing the total acquisition time improved the stability of the estimation of k(ep) values. However, for small k(ep) values, the stability was still poor even with a total acquisition time of 8 minutes. Also, the stability and bias of the estimated values of K(trans), v(p), and k(ep) are interrelated.

Our synthesized DCE model represents perfectly reproduced data except for the presence of Gaussian-distributed random noise. Our analysis suggests minimum changes that may be considered potentially significant in longitudinal therapy assessment studies. Our data are complementary to experimental data from human subjects and phantoms, and provide guidance for the design of image acquisition strategies.