Feature detection and extraction in colour fundus images of the human retina using bitplane decomposition and mathematical morphology

Abstract

Colour fundus images of the human retina are increasingly used in newlinethe diagnosis and treatment of several eye related pathologies and in ailments newlinesuch as diabetes hypertension and arteriosclerosis The effectiveness of newlinetreatment for many eye related diseases lies in the early detection through newlineregular screenings But screening a large number of patients is a significant newlineproblem faced by medical practitioners in populous developing countries like newlineIndia Moreover there are large influences of human errors and subjectivity newlineon the results of inspection by a human expert This opens up the possibility newlineof applying digital image processing techniques in fundus images to facilitate newlineand improve diagnosis in different ways Thus computer aided analysis and newlineautomatic detection of pathologies in retinal images play a major role in newlinemodern diagnostic procedures and screening systems Reliable and robust newlineextraction of retinal features like optic disc macula and vasculature is a newlineprerequisite for subsequent retinal image analysis and processing since these newlineare the predominant and most stable structures appearing in the retina newline