An improved 2D-3D medical image registration algorithm based on modified mutual information and expanded Powell method

An accurate two-dimensional-three-dimensional (2D-3D) image registration method is introduced in this paper, which is used in image-guided surgery to detect and correct patient movement during image-guided intervention treatment. In this paper, we decompose a 3D rigid transformation in the 3D patient coordinate into in-plane transformations and out-of-plane rotations in two orthogonal 2D projections, using approximate geometric relationship and the results in the two projections are then combined and converted to a 3D rigid transformation by 2D-3D geometric transformation. In order to ensure accuracy, we propose that using mutual information based on partial volume interpolation as a similarity measure to align Digitally Reconstructed Radiographs(DRRs) from original 3D CT volume and x-ray images. In the refined stage, we use expanded Powell method to search for the refined parameters. Our experiments show that it is feasible of the assessed 2D-3D registration algorithm, since it achieves sub-millimeter accuracy, when modified mutual information and expanded Powell search method are used.