New global update mechanism of ant colony system for retinal vessel segmentation

This paper proposes two improvements of the ant clustering-based segmentation; first by applying new heuristic function of the ant colony system based on probability theory. Second by selecting any one of new two global update mechanisms of the ant colony system. These improvements evaluation is performed in the domain of automatic retinal blood vessels segmentation. The automatic segmentation of blood vessels in retinal images is vital process for early diagnosis and treatment of many diseases especially diabetic retinopathy since it´s the main cause of blindness over the world. The improved ant colony system based segmentation algorithm is tested on publicity available STARE database of retinal images in terms of the sensitivity, specificity and accuracy. The experimental results obtained, show that the proposed improvements are promising.