Automated image analysis tools have the potential to play an important role in assisting in the diagnosis and treatment of retinal diseases.1 Problems that must be addressed in developing these tools include extraction of vascular and non-vascular features, segmentation of pathologies, unimodal and multimodal image registration, mosaic construction, and real-time systems. Research at Rensselaer Polytechnic Institute since the late 1990’s has focused on several of these problems. Most significantly, we have developed a series of registration and mosaic formation algorithms which have been validated on thousands of retinal images and have been extended beyond the retina application. While the core fundus image registration problem is essentially solved, important problems remain in many aspects of retinal image analysis.
展开▼
