A BINARIZATION APPROACH FOR CT-MR REGISTRATION USING NORMALIZED MUTUAL INFORMATION

摘要

Maximization of mutual information is one of the most popular algorithms for three-dimensional medical image registration. In practice, the multi-resolution scheme can help to improve the optimization speed and capture range. This paper presents a binarization/multi-resolution approach for Computed Tomography (CT) and Magnetic Resonance (MR) brain image registrations based on normalized mutual information. This approach gives improved accuracy with no loss of speed. We illustrate this approach by presenting the results of tests on the data of 7 patients. Registration accuracy is evaluated by J.M. Fitzpatrick, at Vanderbilt University, as part of the Retrospective Registration Evaluation Project. We also compare its performance to results available in the literature.