Classification of two-site exchange models for DCE-MRI

PURPOSE: The Tofts models have long been regarded as a standard approach to tracer-kinetic analysis of DCE-MRI [1]. However, recent technological advances such as parallel imaging and higher field strengths have improved DCE-MRI data quality to a point where more complex models are required [2]. For tissues with a bidirectional exchange of tracer between intraand extravascular spaces, a number of alternative twosite exchange models have been proposed [3-6]. Each imposes different constraints on the tissue, but due to their complicated algebraic structure the relation between the models is poorly understood [7]. As a result it remains unclear which is more accurate in general or in particular applications, and how such a choice can be made operationally. Here we propose a generalised model which exposes the architectural differences between the models and allows transparent handling of model-comparison and -selection issues.