%0 Journal Article
%T Robust Vessel Segmentation in Fundus Images
%A A. Budai
%A R. Bock
%A A. Maier
%A J. Hornegger
%A G. Michelson
%J International Journal of Biomedical Imaging
%D 2013
%I Hindawi Publishing Corporation
%R 10.1155/2013/154860
%X One of the most common modalities to examine the human eye is the eye-fundus photograph. The evaluation of fundus photographs is carried out by medical experts during time-consuming visual inspection. Our aim is to accelerate this process using computer aided diagnosis. As a first step, it is necessary to segment structures in the images for tissue differentiation. As the eye is the only organ, where the vasculature can be imaged in an in vivo and noninterventional way without using expensive scanners, the vessel tree is one of the most interesting and important structures to analyze. The quality and resolution of fundus images are rapidly increasing. Thus, segmentation methods need to be adapted to the new challenges of high resolutions. In this paper, we present a method to reduce calculation time, achieve high accuracy, and increase sensitivity compared to the original Frangi method. This method contains approaches to avoid potential problems like specular reflexes of thick vessels. The proposed method is evaluated using the STARE and DRIVE databases and we propose a new high resolution fundus database to compare it to the state-of-the-art algorithms. The results show an average accuracy above 94% and low computational needs. This outperforms state-of-the-art methods. 1. Introduction In ophthalmology the most common way to examine the human eye is to take an eye-fundus photograph and to analyse it. During this kind of eye examinations a medical expert acquires a photo of the eye-background through the pupil with a fundus camera. The analysis of these images is commonly done by visual inspection. This process can require hours in front of a computer screen, in particular in case of medical screening. An example fundus image is shown in Figure 1. Figure 1: An example of eye-fundus image: the macula is shown in the middle, the optic disk is to the right, and the blood vessels are entering and leaving the eye through the optic disk. Our goal is to speed up the diagnosis by processing the images using computer algorithms to find and highlight the most important details. In addition we aim to automatically identify abnormalities and diseases with minimal human interaction. Due to the rapidly increasing spatial resolution of fundus images, the common image processing methods which were developed and tested using low resolution images have shown drawbacks in clinical use. For this purpose, a new generation of methods needs to be developed. These methods need to be able to operate on high resolution images with low computational complexity. In this paper, we
%U http://www.hindawi.com/journals/ijbi/2013/154860/