Automated Computation of the Vascular Input Function for Dynamic Contrast-Enhanced MRI of the Brain
Butman J, Gupta S
Clinical Center, NIH
Dynamic contrast-enhanced MRI analysis using the 2-compartment General Kinetic Model can be used to characterize contrast medium kinetics based on 3 parameters (Ktrans, kep, and fPV). Solving this model requires an explicit vascular input function. Manual identification of the input function is subjective and more prone to errors. We present here a fast, fully automatic method for estimating the VIF from 3D brain DCEMRI data. The method first computes a mask from the data to emphasize vasculature, and then reliably selects pixels which exhibit enhancement characteristics of vessels. Results comparing the two methods on 15 clinical cases are presented.