Automated optic nerve head image fusion of nonhuman primate eyes using heuristic optimization algorithm

Multi-sensor biomedical image registration and fusion usually require intensive computational effort. This article presented a novel automated approach of the multi-sensor retinal optic nerve head image registration and fusion using heuristic optimization algorithm. The reference and the to-be-registered images are from two different modalities, i.e. angiogram grayscale images and fundus color images. The optic nerve head vasculature is extracted using Canny edge detector. Control points are detected at the vessel bifurcations using adaptive exploratory algorithm. mutual-pixel-count (MPC) maximization based heuristic optimization adjusts the control points at the sub-pixel level. The iteration stops either when MPC reaches the maximum value, or when the maximum allowable loop count is reached. A refinement of the parameter set is obtained at the end of each loop, and finally an optimal fused image is generated at the end of the iteration. Comparative evaluation is performed with the genetic algorithm. The results show the advantages of the presented method in terms of novelty, efficiency and accuracy.