Improving the Accuracy & Precision of DCE-MRI Tracer Kinetic Modelling by Imposing Inter-Variable Constraints

Leonidas Georgiou^1,2, Chris James Rose^1,2

Tracer kinetic model parameters typically have physiological interpretations, and constraints exist between them. While most fitting routines can impose constraints on parameters individually, this is not so for inter-variable constraints. Irrespective of the accuracy of the model itself, non-physiological parameter estimates will result in erroneous inferences being drawn about tumour microvascular characteristics. We study the frequency with which non-physiological fits occur in clinical data, propose a method with which those constraints can be imposed and, using simulations, compare the proposed solution to the conventional approach. We show that imposing inter-parameter constraints can result in more accurate and precise parameter estimates.