Automatic Detection Of Microaneurysms In Retinopathy Fluoro-Angiogram

Publisher Website
Google Scholar
Abstract
A computerized method for detecting microaneurysms is described, based upon the concepts introduced by the Mathematical Morphology. On the grey level image of a fluoroangiogram, magnified by a microscope, a first transformation, searching the regional minima, gives a binary image where appear vessels and microaneurysms. Then the microaneurysms are extracted using a shape criterion which selects almost circular particles among stretched particles in one direction. The most complex part is to discriminate diffusing microaneurysm (leakage of fluorescein) from vessel. The algorithm is tested on 25 angiograms. We obtain a good correlation between the automatic counting and the manual counting performed by three reading-technicians.

This publication has 0 references indexed in Scilit: