Automated detection of drusen in the macula

• Published in: 2009 IEEE International Symposium on Biomedical Imaging: From Nano to Macro pp. 61 - 64
• Main Authors: Freund, D.E., Bressler, N., Burlina, P.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2009
• Subjects: Biomedical imaging; Blindness; Computer science; Detection of retinal abnormalities; Humans; Kernel; Laboratories; macular pathologies; Medical treatment; Physics; Retina; Shape; support vector methods
• ISBN: 1424439310; 9781424439317
• ISSN: 1945-7928
• DOI: 10.1109/ISBI.2009.5192983

Age related macular degeneration (AMD) is a condition of the retina that occurs with individuals over 50. AMD is characterized by the formation of drusen in the macula. This condition leads to a deterioration of foveal vision and eventually functional blindness. Automatically screening atrisk individuals may allow the detection of intermediate stage AMD where it is still treatable using anti-VEGH therapy. One of the difficulties in detecting and locating drusen is that their aspect (shape, texture, color, extent) varies significantly, and because of this it is often difficult to build a classifier. To address this difficulty we use a two pronged approach based on (a) multiscale analysis and (b) kernel based anomaly detection. We show experimental results on examples of fundus images taken from healthy and affected patients.