1Department of Radiology and Nuclear Medicine, St. Olav's University Hospital, Trondheim, Norway, 2Department of Physics, Norwegian University of Science and Technology (NTNU), Trondheim, Norway
Intravoxel incoherent motion (IVIM) modelling has the potential to provide pixel-wise maps of pseudo-diffusion parameters that offer insight into tissue microvasculature. However, standard approaches using least-squares fitting yield parameter maps that are typically heavily corrupted by noise. Bayesian modelling has been shown recently to be a promising alternative. In this work we test the robustness of one such Bayesian approach by applying it to simulated noisy data, and obtain clearer parameter maps with much lower estimation uncertainty
than least-squares fitting. However, certain features are found to disappear completely, indicating that a level of caution is required when implementing such techniques.