Detection of Blood Vessels From Retinal Images Using Watershed Transformation

Abstract

Automated analysis and interpretation of retinal images has become an incontournable diagnostic step in ophthalmology. Retinal blood vessels morphology can be an important indicator for diseases such as diabetic retinopathy; and their detection also serves for image registration. This paper presents a method based on mathematical morphology for extraction of vascular tree in color retinal image with low contrast. It consists in contrast enhancement and application of watershed transformation in order to segment blood vessels in digital fundus images.

Improvement of Automatic Hemorrhages Detection Methods Using Shapes Recognition

“…It is very important for ophthalmologists to determine these diseases in early stages because some of these diseases, if not detected early, can make people blind (A. Bessaid et al, 2009). …”

Fundus characterization for automatic disease screening through retinal image processing

“…In those cases, the presented algorithm works properly and reduces over-segmentation problems which can be found in methods based on a previous segmentation (Martinez-Perez et al, 2007;. With regard to other methods that obtain the skeleton in a direct way and use the watershed transformation (Bessaid et al, 2009; instead of the stochastic watershed, the proposed work achieves a more robust detection and decreases the number of spurs. Despite good results, it must be mentioned that the main disadvantage of the method is that some vessels can lose their continuity if some part of them is not detected and it should be corrected in a post-processing stage.…”

“…In the same way as explained above, the homotopic skeleton was performed after the segmentation process in both cases. On the other hand, as for the methods that obtain directly the retinal vessel centerline, the analysis was focused on other two approaches based also on the watershed transformation Bessaid et al, 2009). In Figures 3.48 and 3.49, the strengths and weaknesses of the proposed method can be observed in two examples of both databases.…”

“…Jaccard's 96 Chapter 3. Retinal structure characterization (Bessaid et al, 2009) and (f ) Walter and Klein method .…”

“…Although most of the state-ofthe-art methods look for detecting all vessel pixels of the vascular tree, there are also some attempts based on finding the vessel skeleton or, in other words, the vessel centerline. The work of Chen et al is based on shortest path connection , Sofka and Stewart on the use of matched filters (Sofka and Stewart, 2006), Wu and Derwent on ridge descriptors (Wu et al, 2009) and Walter and Klein and Bessaid et al on the application of watershed transformation Bessaid et al, 2009). …”