Extraction of Blood Vessels from Retinal Fundus Images using Maximum Principal Curvatures and Adaptive Histogram Equalization

Retinal imaging is a more popular technology among all the biomedical imaging technologies, due to its ability to acquire data that is related to various eye diseases. The exact identification of blood vessels is mandatory to help the ophthalmologist in diagnosing the diseases at a primary stage. In this paper, a novel algorithm is proposed for the segmentation of blood vessels from fundus images. This algorithm is divided into three stages. (i) Hessian-second order derivative is used to identify Maximal principal curvatures (ii) Iterative Self Organized Data Analysis (ISODATA) technique is used to separate blood vessels from the background (iii) morphological open operation is used to remove the unwanted segmented isolated regions. This proposed algorithm is tested on Digital Retinal Images for Vessel Extraction (DRIVE) database and Structured Analysis of the Retina (STARE) database. The experimental results show that the proposed algorithm achieves maximum accuracy of 97.14% and average accuracy of 94.18%