Stochastic rank correlation - A novel merit function for dual energy 2D/3D registration in image-modulated radiation therapy

Abstract

Image-modulated radiation therapy is a field of considerable interest for both the clinical and the research community. By using on-board imaging equipment, patient pose and tumor position can be monitored during the course of therapy. 2D/3D registration is a key technology to achieve this goal. In a recent study in FluoroCT/ CT registration, we have shown that conventional cross correlation (CC), together with repeated use of conventional local optimization algorithms, provides an optimum measure for slice-to-volume registration for monoenergetic CT imaging data. If the required linear relationship between corresponding pixel pairs is offended (e. g. by using X-rays of different energy or by varying detector characteristics), CC becomes an unreliable measure of image similarity. A more general merit function like normalized mutual information (NMI) serves better in such a case but is stricken with local minima caused by sparse population of joint histograms. We present a novel merit function for 2D/3D registration named stochastic rank correlation (SRC), which is well-suited for intramodal dual-energy imaging. Here, the rank correlation coefficient is computed for pair of X-ray (or portal) images and the corresponding digitally rendered radiograph. Since computing rank correlation requires an ordering of gray values in each iteretive registration step, this approach by itself cannot be considered useful. We show that by subsampling the image using a random selection of a small number of image pixels, a valid estimate for the rank correlation coefficient can be computed. A first evaluation of SRC is given on a set of simulated and clinical image data sets. It can be seen that both in terms of accuracy, stability, numerical behaviour, and time requirements SRC is a highly competitive new paradigm for 2D/3D registration. It is independent of varying grayscale content in intramodal images and is therefore ideally suited for dual energy imaging in image-modulated radiation therapy