Automatic change detection of retinal images

Abstract

The aim of the presented study is the development of an automatic method for change detection in multitemporal digital images of the human retina. The images are acquired from the same patient at different times by a color fundus camera. The method proposed here is based on the preliminary automatic registration of multitemporal images, and the detection of the changes that can occur in the retina during time, by comparing the registered images. In order to achieve the temporal registration of the retinal images, an automatic approach based on global optimization techniques is proposed here. In particular, in order to estimate the optimum transformation between the input and the base image, a genetic algorithm is used to optimize the match between previously extracted maps of curvilinear structures in the images to be registered (such structures being represented by the vessels in the human retina). The proposed approach for the detection of temporal changes within the registered images is based on the application of an unsupervised algorithm, in order to cope with the lack of training information about the statistic of the changed areas in fundus images. The algorithm is tested on color fundus images with small and large changes. The comparison between the registered images using the implemented method and a manual one points out that the proposed algorithm provides an accurate registration. The image registration is not possible only when dealing with images taken from very different view-points. The analysis of the change-detection performances by a human expert suggests that the method is able to provide accurate change maps, when registration is successful. © 2009 Springer-Verlag.