Improving Tumour ADC Estimates and Elucidating Tumour Heterogeneity Using Adaptive Bayesian Markov Random Field Monte Carlo

Simon Walker-Samuel¹, Matthew Orton¹, Jessica K R Boult¹, Simon P. Robinson¹

A method for improving ADC estimates using an adaptive Bayesian Markov random field analysis is described and evaluated using simulations and in vivo tumour models. Via the sharing of information between neighbouring pixels, the uncertainty and error in ADC estimates are significantly reduced by a factor of up to 80%. The approach also enables the segmentation of homogeneous tissue components and a novel measure of tissue (tumour) heterogeneity is described.