ارزیابی شیوه های کامپیوتری جهت قطعه بندی در تصاویر شبکیه ای و افزایش رگ خونی

ABSTRACT A Comparison and Evaluation of Computerized Methods for Blood Vessel Enhancement and Segmentation in Retinal Images

Diabetic retinopathy is a disease which damages the retina and it is caused by the sophistications of diabetes, which leads to blindness. All patients who have diabetes mellitus have the risk of diabetic retinopathy. The risk factor increases with the increase in the span of diabetes. There are no early indications of Diabetic retinopathy. Even sometimes macular edema does not show any indication rather it creates vision loss rapidly. In retinopathy new blood vessels can be formed at the back of eye and due to their leakage blurred vision is caused. In most cases some blood spots appear on the surface of eye which may disappear after some hours.

A tool which can be used to assist in the diagnosis of diabetic retinopathy should automatically detect all retinal image features such as optic disk, fovea and blood vessel and all abnormalities in retinal image such as microaneurysms hard and soft exudates, hemorrhages and edema.

Computer assisted diagnosis for various diseases is very common now a days and medical imaging is playing a very important part in such computer assisted diagnosis. So, image processing techniques can help in detecting all above mentioned abnormalities from fundus image as they appear with obvious properties (color size and shape).

Most of the vessel segmentation methods are based on watershed algorithm [1], probabilistic filters [2], Hessian based filters [3], improved matched filters [4] and Gabor filter [5]. Methods based on vessel tracking to obtain the vasculature structure, along with vessel diameters and branching points have been proposed by [6] and [7]. Papers [8]-[9] used deformable models for vessels segmentation.

In this paper, we compare the retinal image enhancement techniques that enhance the blood vessels using Gabor wavelet and LoG filtering. We also compare the retinal image vessel segmentation techniques that segment the vascular pattern using global thresholding and adaptive thresholding.

The paper is organized in four sections. In section II, comparison and evaluation of the two techniques for automated vessel enhancement and segmentation is done in detail. Experimental results on the retinal images of the DRIVE and STARE databases and their analysis are given in Section III followed by conclusion in Section IV.