Design of an effective method to accurately register retinal fluorescence images for future quantitative analysis in patients with nAMD

Neovascular age-related macular degeneration (nAMD) is characterized by an increased choroidal vascularization underneath the macula through the upregulation of vascular endothelial growth factors (VEGF). Therefore, current therapies for this disease are focused on preventing VEGF from binding to its corresponding receptors using antibodies such as bevacizumab. Since the effect of these antibodies is still uncertain, the Optical Molecular Imaging Groningen (OMIG) research group has set up a study in which they combine bevacizumab with the near-infrared dye IRDye800CW. This antibody tracer is then used to image the retinal vascularization using fluorescence angiography alongside the tracer fluorescein and the baseline retinal autofluorescence. The research group obtained baseline images, bevacizumab-800CW images and fluorescein angiographic images from eight patients with nAMD using Scanning Laser Ophthalmoscopy (SLO). Two registration methods were developed for the alignment and overlay of these images. The first method was developed using the DS4H Image Alignment plugin in Fiji. This method relies on the manual selection of at least three points on distinctive positions in each image, hence being a semi-automatic method for image registration. After this, a fully automatic method was written in Matlab using the SURF algorithm. The script for this method detects specific features in the SLO images, after which the script matches these points between images and registers the fluorescence image to the baseline image.