Abstract #3490

Walker-Samuel S, Collins D, Orton M, Leach M

When applying pharmacokinetic (PK) models to DCE-MRI data it is necessary to provide on estimate of the arterial input function (AIF). We propose a novel Bayesian method that allows the AIF to be directly estimated from the tissues of interest. This leads to accurate estimates of the PK parameters, and allows for the incorporation of prior information that can assist the inference. This approach was tested on data taken from liver metastases. In this example the posterior distribution for the AIF parameters was well localised, and the maps of the Bayesian derived PK parameters agreed well with existing methods.