Certain investigations on detection of glaucoma using image processing and machine learning approaches

Abstract

Identification of physiological changes inside the human body is a challenging task in bio-medical engineering. To be specific, identifying the abnormalities in the human eye is very difficult and tedious, as it includes several complexities related with the procedure. Hence retinal image analysis gained more attention among the researchers, due to the necessity of disease detection systems at an early stage aiding in the screening and management of the disease. In retinal image analysis, many investigations are carried-out for automated glaucoma diagnosis. Glaucoma disease leads to blindness in human beings worldwide and stands third in causing blindness in India. Thus, early detection of glaucoma disease is essential for preventing eye diseases from getting severe. Glaucoma is said to increase the Cup to Disc ratio in the eye, affecting peripheral vision. Variety of image processing techniques that are used in diagnosing glaucoma based on Cup to Disc ratio evaluation of the pre-processed retinal fundus images are addressed in the research work. These computational algorithms are evaluated on the public fundus image datasets and the performance results are compared. Most of the existing glaucoma detection systems depend on human intervention, the assessments being much ambiguous, time-consuming and warrant skilled professionals. In most of the circumstances, manual investigation of the retinal image leads to misdiagnosis due to human error, mostly occurring by visual fatigue. To enhance the diagnostic accuracy of screening retinal images, computer-aided design systems are developed serving as a baseline for screening and diagnosis. Presently, numerous automated systems are developed on retinal image analysis for various newline