Investigation and optimization of parameter accuracy in dynamic contrast‐enhanced MRI

Abstract

To present a modified pharmacokinetic model for improved parameter accuracy and to investigate the influence of an inaccurate arterial input function (AIF) on dynamic contrast-enhanced (DCE)-MRI parameter estimates of the transfer constant (Ktrans), blood volume (vp), and interstitial volume (ve). Tissue uptake curves were simulated over a large range of physiological values and analyzed for different AIF measurement errors and temporal resolutions. The AIF measurement was assumed to be inaccurate in the bolus amplitude (rapid sampling) or susceptible to unknown temporal offsets (slow sampling with biexponential decay fit). The modified model adequately reduces errors in parameter estimates arising from transit time effects. An error in the AIF bolus amplitude results in an inversely proportional error in Ktrans and vp; ve remains robust. More consistent error in Ktrans (approximately 20% underestimation) was obtained using a biexponential AIF, at the expense of severely underestimating vp. While an accurate, high temporal resolution AIF is essential for estimating vp, a biexponential AIF acquired at low temporal resolution (<20 seconds) provides robust estimates of ve and results in a Ktrans underestimation comparable to that from a 25% error in the initial AIF bolus amplitude.