Identification of Artery/Vein in Retinal Image using Local Decision Tree Classifier
Rama Krishna V.V1, Suman.M2
1Rama Krishna V.V, Department of Electronics and Communication Engineering, Laki Reddy Bali Reddy College of Engineering, Mylavaram (Andhra Pradesh), India.
2Suman.M, Department of Electronics and Communication Engineering, KL University, Green Fields, Vaddeswaram, Gunturt (Andhra Pradesh), India.
Manuscript received on 26 March 2019 | Revised Manuscript received on 05 April 2019 | Manuscript Published on 27 April 2019 | PP: 669-673 | Volume-7 Issue-6S2 April 2019 | Retrieval Number: F11100476S219/2019©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: The importance of automatic image analysis is increasing day by day for early detection of diseases like Cancer, Diabetic Retinopathy (DR), Hypertension. Sometimes these diseases causing damage to retinal image leading to blindness. We can prevent this by careful analysis of artery/vein retinal image. From this we observe that identifying retinal vessels in to artery/vein (A/V) plays a major role in detection the vascular changes and symptoms associated with several diseases. In this work, first we extract graph from retinal image, followed by investigation of identified graph nodes (intersection point). We perform depth first graph traversal with trained decision tree applied on each graph node to classify artery/vein. The trained decision tree classifier takes node labels, vessel segment (graph links) intensity features into account for classification. The proposed framework is tested on standard dataset and is compared with trained human expert for standard datasets. The accuracy values of INSPIRE-AVR, DRIVE, and VICAVR databases are 91.4%, 92.1%, and 93.2% respectively. Experimental results show that our approach is consistent with existing A / V classification state of the art algorithms.
Keywords: Artery, Vein, Thinning Algorithm, Local Decision Tree Algorithm, Depth Graph Traversal.
Scope of the Article: Image Security