Direct estimation of pharmacokinetic parameters of DCE-MRI from raw k-space data with model-based reconstruction

Dynamic contrast enhanced (DCE) MRI is a minimally invasive technique that is able to quantitatively investigate the tumor vasculature microenvironment. Such information shows great potential for treatment stratification and response monitoring. However, DCE typically suffers from low spatial resolution, Rician noise bias, and errors due to complex perfusion modeling. Model-based reconstruction, in which DCE parameters are estimated directly from k-space, may overcome these shortcomings. In this study, we implemented model-based reconstruction for DCE-MRI data, validated it in simulations, and showed its performance in-vivo. With model-based reconstruction the estimated parameter maps exhibited less noise and preserved more anatomical details.