Dynamic contrast‐enhanced MRI is used to estimate microvascular parameters by tracer kinetics analysis. The time for the contrast agent to travel from the artery to the tissue of interest (bolus arrival time (BAT)) is an important parameter that must be measured in such studies because inaccurate estimates or neglect of BAT contribute to inaccuracy in model fitting. Furthermore, although the precision with which these parameters are estimated is very important, it is rarely reported. To address these issues, two investigations were undertaken. First, simulated data were used to validate an independent method for estimation of BAT. Second, the adiabatic approximation to the tissue homogeneity model was fitted to experimental data acquired in prostate and muscle tissue of 22 patients with prostate cancer. A bootstrap error analysis was performed to estimate the precision of parameter estimates. The independent method of estimating BAT was found to be more accurate and precise than a model‐fitting approach. Estimated precisions for parameters measured in the prostate gland were 14% for extraction fraction (median coefficient of variation), 19% for blood flow, 28% for permeability‐surface area product, 35% for volume of the extravascular‐extracellular space, and 36% for blood volume. Techniques to further reduce uncertainty are discussed. Magn Reson Med, 2006.

Dynamic contrast‐enhanced MRI has been used in conjunction with tracer kinetics modeling in a wide range of tissues for treatment monitoring, oncology drug development, and investigation of disease processes. Accurate measurement of model parameters relies on acquiring data with high temporal resolution and low noise, particularly for models with large numbers of free parameters, such as the adiabatic approximation to the tissue homogeneity model for separate measurements of blood flow and vessel permeability. In this simulation study, accuracy of the adiabatic approximation to the tissue homogeneity model was investigated, examining the effects of temporal resolution, noise levels, and error in the measured arterial input function. A temporal resolution of 1.5 s and high SNR (noise sd = 0.05) were found to ensure minimal bias (10%). A 10% error in the measured height of the arterial input function first pass peak resulted in an error of at most 10% in each model parameter. Magn Reson Med, 2010.

The purpose of this study was to assess the effects of cellular‐interstitial water exchange on estimates of tracer kinetics parameters obtained using rapid dynamic contrast‐enhanced (DCE) MRI. Data from the internal obturator muscle of six patients were examined using three models of water exchange: no exchange (NX), fast exchange limit (FXL), and intermediate rate (shutter‐speed [SS]). In combination with additional multiple flip angle (FA) data, a full two‐pool exchange model was also used. The results obtained using the NX model (transfer constant, K^trans = 0.049 ± 0.027 min^–1, apparent interstitial volume, v_e = 0.14 ± 0.04) were marginally higher than those obtained using the FXL model (K^trans = 0.045 ± 0.025 min^–1, v_e = 0.13 ± 0.04), but the error bars overlapped in two‐thirds of these parameter estimate pairs. Estimates of K^trans and v_e obtained using the SS model exceeded those obtained using the NX model in half the patients, and many estimates, including all those of intracellular residence time of water, t_i, were imprecise. Results obtained using the full two‐pool model fell between those obtained using FXL and NX models, and estimates of t_i were also imprecise. The results suggest that data obtained using clinically relevant DCE‐MRI are exchange‐insensitive and unsuitable for the assessment of cellular‐interstitial water exchange. Magn Reson Med 60:1011–1019, 2008.