A new optic disc segmentation method using a modified Dolph-Chebyshev matched filter

Abstract

In this paper, a novel algorithm for automatic optic disc segmentation on fundus images is presented. In the proposed algorithm, a modified Dolph-Chebyshev type I function matched filter is utilized to detect optic disc boundary candidates. Subsequently, the Circular Hough Transform and B-spline approximation are employed to obtain the final optic disc boundary. The matched filter, Circular Hough Transform and B-spline approximation work on a localized region of interest that is obtained based on three independent methods: template-based, vessels density map-based, and maximum entropy-based methods. The performance of the proposed algorithm is evaluated using fundus images from the MESSIDOR and DRIVE datasets. From the experiments, it can be observed that the proposed algorithm successfully locates optic discs with an average success rate of above 99%. Also, the proposed algorithm can achieve best scores in terms of Area Overlap, Dices coefficient, Accuracy, and False Positive Fraction, which makes the algorithm suitable for practical applications such as an automatic detection tool for retinal diseases.