The minimal entropy prior for simultaneous reconstruction and segmentation of in vivo microct trabecular bone images

We present the minimal entropy prior for the iterative reconstruction of muCT projection data using the maximum a posteriori (MAP) framework. The minimal entropy prior restricts the number of intensity values in the reconstruction without requiring a priori information about the number of image tissues or tissue intensity estimates. We utilise the presented prior in the MAP framework to simultaneously reconstruct and segment trabecular bone from low quality muCT images. The MAP approach allows the straightforward incorporation of the in vivo scanner point spread function, while the entropy prior yields a segmentation that obeys to projection data.