Study of onset time-shift and injection duration in DCE-MRI: a comparison of a reference region model with the general kinetic model.

In dynamic contrast-enhanced MR imaging (DCE-MRI), sampling of the arterial input function (AIF) is required for analysis using the general kinetic model (GKM). Alternatively, the recently proposed reference region model (RRM) may be employed to avoid the need of acquiring the AIF. This study aimed to evaluate the influence of the AIF onset-time shift and the injection duration, under various sampling intervals, on physiological parameter estimation in DCE-MRI using the GKM, and to compare the performance between GKM and RRM. Computer simulations were performed to assess the mean error (ME) and coefficient of variation (CoV) of K(trans,TOI) and v(e,TOI) from shifted and dispersed AIF with temporal resolution of 1, 5 and 10 s. With 5-s sampling, the maximal ME of K(trans,TOI) was roughly 22% for the GKM and 0.5% for the RRM. With 10-s sampling, they increased to around 28% and 0.7%, respectively. The maximal MEs of v(e,TOI) for all cases were under 5%. However, owing to the lower SNR in the reference region, the CoV obtained by the RRM were all higher than those by the GKM. The results suggested that with compromised temporal resolution, the RRM was relatively less sensitive to the AIF onset-time shift and the injection duration compared with the GKM.