This repository contains algorithms and experiments for the paper
"A cross-center smoothness prior for variational Bayesian brain tissue segmentation."
which was published in the proceedings of the International Conference on Information Processing in Medical Imaging 2019 (pdf/preprint)
Generalizing machine learning algorithms across medical centers is difficult. Data is often strongly biased towards each center, leading to different mappings from medical image X to segmented image Y.
Instead of designing an adaptive classification model that would attempt to adjust its mapping X → Y for each center, I inform an unsupervised Bayesian segmentation model with how Y is supposed to look like. Specifically, I fit a smoothness prior on segmentations produced in one medical center and incorporate that as an informative empirical prior in a variational Bayesian image segmentation model. This model will produce segmentations in a target medical center that are as smooth as the segmentations produced in the source medical center.
Comments, questions and feedback can be submitted to the issues tracker.