Useful Papers - Coregistration
Maes et. al (1997), "Multimodality image registration by maximization of mutual information," IEEE Transactions on Medical Imaging 16(2), 187-198 PDF
Summary: One of the fundamental mutual information (MI) coregistration papers, this lays out the concepts of MI in a reasonably easy-to-follow manner and explains how to use MI as a cost function in coregistration. The advantages relative to other cost functions (least-squares, for example) are laid out and a set of experiments are perfomed to show the algorithm's effectiveness, which is subvoxel. A couple different interpolation schemes are also discussed.
Bottom line: An excellent paper summarizing the advantages of MI as a coregistration cost function.
Zhu & Cochoff (2002), "Influence of implementation parameters on registration of MR and SPECT brain images by maximization of mutual information," Journal of Nuclear Medicine 43(2), 160-166 PDF
Summary: The authors compare different settings of a coregistration algorithm for their effectiveness and accuracy; parameters varied included type of interpolation, number of bins to create the probability density function, and a couple different sampling optimization techniques.
Bottom line: Trilinear interpolation works best, as does adaptively changing bin numbers. Simplex optimization or multiresolution optimization were also effective ways to improve the success rate and speed the coregistration.
Ashburner & Friston (1997), "Multimodal image coregistration and partitioning - a unified framework," NeuroImage 6, 209-217 PDF
Summary: The original paper defining the old SPM (pre-SPM2) way of doing coregistration. The authors suggest defining within-modality templates that are already coregistered and using least-squares methods to coregister the experimental images to those templates. Using segmentation during the coregistration can help improve the success and accuracy of that registration.
Bottom line: A bit obsolete these days; SPM2 has moved to MI coregistration, which is simpler and shows better success rates.
Nestares & Heeger (2000), "Robust multiresolution alignment of MRI brain volumes" Magn Reson Med. May;43(5):705-15 PDF
Summary: Discusses the algorithm former Stanford prof David Heeger used in his lab to do realignment for visual cortex; they enhance the robustness of their algorithm by automatically masking out voxels whose intensity difference between images exceeds some threshold. Results from the algorithm are presented.