Unifying Maximum Likelihood Approaches in Medical Image Registration

摘要

Although intensity-based similarity measures are in- creasingly used for medical image registration, they often rely on implicit assumptions regarding the imaging physics. This paper clar- ifies the assumptions on which a number of popular similarity mea- sures rely. After formalizing registration based on general image ac- qusition models, we show that the search for an optimal measure can be cast into a maximum likelihood estimation problem. We then derive similarity measures corresponding to different modeling as- sumptions and retrieve some well-known measures (correlation co- efficient, correlation ratio, mutual information). Finally, we present results of rigid registration between several image modalities to illus- trate the importance of choosing an appropriate similarity measure.