Evaluating similarity metrics in an image matching tool for image guided spine surgery

This paper presents a comparison study of the effect of three similarity measures (mutual information, normalized mutual information, and mean squared error on the edges of the input image) in an image matching tool that can be used in image guided spine surgery. Using 3D rotational x-rays and magnetic resonance images of a spine phantom, it was found that the similarity measures had similar effect. Therefore, experiments with other datasets are needed before making conclusions about the suitability of these similarity metrics for image guided surgery.