On breast imaging from joint microwave and acoustic data within a Bayesian framework

Breast cancer is most common, so early diagnosis of tumors is wished for. Microwave (MW) and ultrasound (US) are non-invasive, non-ionizing, low-cost, and can be run without registration for free pending breasts. MW is to yield high-contrast images of low resolution, the converse with US, with the benefit of the common breast structure. That is, fusion of MW and US data should produce images with both high contrast and high resolution. Here a Bayesian formalism is chosen to that effect (Variational Bayesian Approximation or VBA), edges as hidden variables, a number of hyperparameters involved as expected. Once the mathematics sketched, one insists on imaging of a MRI- derived breast model, a tumor added into it. Comparison with a joint edge-preserving contrast source inversion (JCSI-EP) in a deterministic framework will illustrate pros and cons of VBA.