Accelerated mutual entropy maximization for biomedical image registration

In this paper, we propose an improved acceleration scheme for the mutual entropy maximization method for biomedical image registration. Our approach is based on fast adaptive bidirectional empirical mode decomposition (FABEMD) and aims to reduce the computational complexity of the mutual entropy maximization algorithm by extracting only information essential for registration. We apply several adaptive optimization techniques in a row: image structural reduction using FABEMD, histogram sparsification, image downsampling, and multilevel parametric space search. Our experiments show that with the proposed scheme registration is performed up to 150 times faster without noticeable loss of accuracy for typical MRI images.