Maximum Likelihood Estimation of the Bias Field in MR Brain Images: Investigating Different Modelings of the Imaging Process

This article is about bias field correction in MR brain images. In the literature, most of the methods consist in modeling the imaging process before identifying its unknown parameters. After identifying two of the most widely used such models, we propose a third one and show that for these three models, it is possible to use a common estimation framework, based on the Maximum Likelihood principle. This scheme partly rests on a functional modeling of the bias field. The optimization is performed by an ECM algorithm, in which we have included a procedure of outliers rejection. In this way, we derive three algorithms and compare them on a set of simulated images. We also provide results on real MR images exhibiting a bias field with a typical “diagonal” pattern.