Influence of arterial input function (AIF) on quantitative prostate dynamic contrast-enhanced (DCE) MRI and zonal prostate anatomy

Purpose: Reproducibility of quantitative perfusion analysis of DCE requires a standardized AIF acquisition. However, there are many different approaches for AIF assessment so that the absolute values of perfusion parameters may vary depending on the used method. This study analyzes the influence of the method of AIF determination on quantitative DCE-MRI. Methods: In this retrospective, single-center, cohort study three different methods of AIF determination in 50 consecutive patients with multiparametric MRI of the prostate were conducted. As a reference, AIF was selected manually by defining a region of interest in an artery manually (AIF(m)). The second method (AIF(a)), based on an automated algorithm and the third, population-derived AlFp where then compared. Primary endpoint were differences in the performance of the perfusion parameters K-trans, ve and k(ep) regarding the AIF acquisition methods, secondary endpoints consisted of the evaluation of differences in the peripheral and transition zone of the prostate (PZ, TZ). Results: In all three methods, K-trans, ve, and k(ep) were significantly higher in PZ than in TZ with K-trans showing least overlapping. There were no significant differences for K-trans determined with AIF(m), and AIF(a) (0.3 +/- 0.2 min(-1) for PZ for both and 0.5 +/- 0.3 min(-1) for TZ in AIF(m) and 0.4 +/- 0.3 min(-1) in AIF(a)), while there were great differences between AIF(a) and AIF(p) and AIF(m) and AIF(p) (0.1 +/- 0.03 min(-1) for TZ and PZ in AIF(p)). Spearman test demonstrated good correlation of values for K-trans and K-ep in all 3 methods (p >= 0.76). AIF(a) showed a success rate of 98% in finding the artery. Conclusion: AIF(a) is a recommendable user-independent automatical method to determine quantitative perfusion parameters allowing an objective measurement and saving interactive time for the radiologist. AIF(p) may be applied as second alternative method.