Multimodal macula mapping by deformable image registration

This work aims to present the applied methods for the registration of different imaging modalities of the human macula. Multimodal image registration is herewith considered the process of overlaying two or more images of the same eye, taken from different viewpoints, at different times, with different sizes and fieldsof-view and using different sensors, thus imaging different retinal structures. Additionally, the registration of retinal images from in vivo patients using scanning systems has to deal with intrinsic image distortions due to saccadic eye movements. In order to correct for differences in eye fundus scanned angles, image sizes and feature expressions in different modalities, a feature vector was built for each modality embodying complementary information from the image alone.