Segmentation of Retinal Microvasculature Based on Iterative Self-Organizing Data Analysis Technique (ISODATA)

The modern ophthalmology is completely dependent on digital image processing to find out the remarkable symptoms for diagnosing the severe cardiovascular disease such as hypertensive retinopathy, and transient ischemic attack that are related to the changes of the retinal microvasculature. Employing the image segmentation techniques, the abnormalities in retinal microvasculature like vessel tortuosity, cotton wool spots, and vessel caliber can be extracted which are recognized as the salient symptoms for the abovementioned cardiovascular diseases. In this paper, an automated method for retinal image segmentation has been proposed. The proposed method was developed employing the thresholding based Iterative Self-Organizing Data Analysis Technique (ISODATA) for image segmentation combining with other existing image preprocessing techniques. The performance of the proposed method was evaluated on the healthy patient image set of (High-Resolution Fundus Image Database) HRFID. This newly developed algorithm achieved 94.3% accuracy with 97.86% specificity and 0.0054 standard deviations. The proposed algorithm can be integrated as the computer-aided clinical diagnostic tool to facilitate the ophthalmologist with further evaluation and validation.