An interactive system for computer-aided retinal image analysis

This paper presents a interactive system for analyzing images of the ocular fundus to defect and quantify lesions associated with retinal diseases such as age-related maculopathy and diabetic retinopathy. Using inexpensive Windows/sup TM/ technology, the system combines low-level pattern recognition with a graphical user interface to facilitate interactive segmentation and labelling of fundus images, and calculate and report the location, size and shape of each lesion. The user always has the option to override the program's decisions in order ensure a correct analysis. The system was tested in two small pilot studies, one for analyzing drusen in age-related maculopathy patients, and the other to measure neovascularization associated with diabetic retinopathy.