Parametric Modeling of the Brain Vascular System and its Application in Dynamic Contrast-Enhanced Imaging Studies.

Abstract

Dynamic Contrast-Enhanced Imaging (DCE) is one of the main tools for in vivo measurement of vascular properties of pathologies such as brain tumors. In DCE imaging, one of the key components for estimation of vascular perfusion and permeability parameters using Pharmacokinetic models is the Arterial Input Function (AIF). To measure these parameters more accurately, there have been approaches for estimating the AIF profile at the capillary level; however, a practical and realistic estimate is still missing. As a solution, we have developed a model of the brain vascular system, based on laws of fluid dynamics and vascular morphology, to address dispersion and delay of the contrast agent (CA) concentration profile at different levels of the brain vascular tree. Using this model, we introduced a transfer function that can describe changes of the AIF profile along a vascular pathway, from a major artery to the capillary bed. Our simulations and also testing this model on DCE Imaging data of the human brain, all showed that our model can give a realistic estimation of the CA concentration profile, at all levels of the vascular tree in the brain. In the next step, we extended our model to address vascular leakage as well. Using this extended vascular (EV) model, we are able to decompose the tissue response signal in DCE images to its intravascular and extravascular components. This feature has provided us with an excellent tool that can lead to relatively unbiased measurements of perfusion and permeability parameters, especially in areas with vascular leakage. We tested this on DCE-CT and DCE-MR images and compared the performance of our model to conventional methods. Also, using a simulation study, we measured the levels of overestimation and underestimation of the permeability parameters using conventional processing methods and demonstrated the superior performance of the EV model for more accurate estimation of these parameters. Overall, the results show that the EV model can provide a platform for better understanding of the role of the AIF in DCE studies as well as estimation of AIF for more accurate measurement of perfusion and permeability parameters in clinical studies.