Syed Ayaz Ali Shah1, Tong Boon Tang2, Ibrahima Faye1, Augustinus Laude3. 1. Centre for Intelligent Signals and Imaging Research, Universiti Teknologi PETRONAS, Bandar Seri Iskandar, Perak, Malaysia. 2. Centre for Intelligent Signals and Imaging Research, Universiti Teknologi PETRONAS, Bandar Seri Iskandar, Perak, Malaysia. tongboon.tang@utp.edu.my. 3. National Healthcare Group Eye Institute, Department of Ophthalmology, Tan Tock Seng Hospital, Singapore, Singapore.
Abstract
PURPOSE: To propose a new algorithm of blood vessel segmentation based on regional and Hessian features for image analysis in retinal abnormality diagnosis. METHODS: Firstly, color fundus images from the publicly available database DRIVE were converted from RGB to grayscale. To enhance the contrast of the dark objects (blood vessels) against the background, the dot product of the grayscale image with itself was generated. To rectify the variation in contrast, we used a 5 × 5 window filter on each pixel. Based on 5 regional features, 1 intensity feature and 2 Hessian features per scale using 9 scales, we extracted a total of 24 features. A linear minimum squared error (LMSE) classifier was trained to classify each pixel into a vessel or non-vessel pixel. RESULTS: The DRIVE dataset provided 20 training and 20 test color fundus images. The proposed algorithm achieves a sensitivity of 72.05% with 94.79% accuracy. CONCLUSIONS: Our proposed algorithm achieved higher accuracy (0.9206) at the peripapillary region, where the ocular manifestations in the microvasculature due to glaucoma, central retinal vein occlusion, etc. are most obvious. This supports the proposed algorithm as a strong candidate for automated vessel segmentation.
PURPOSE: To propose a new algorithm of blood vessel segmentation based on regional and Hessian features for image analysis in retinal abnormality diagnosis. METHODS: Firstly, color fundus images from the publicly available database DRIVE were converted from RGB to grayscale. To enhance the contrast of the dark objects (blood vessels) against the background, the dot product of the grayscale image with itself was generated. To rectify the variation in contrast, we used a 5 × 5 window filter on each pixel. Based on 5 regional features, 1 intensity feature and 2 Hessian features per scale using 9 scales, we extracted a total of 24 features. A linear minimum squared error (LMSE) classifier was trained to classify each pixel into a vessel or non-vessel pixel. RESULTS: The DRIVE dataset provided 20 training and 20 test color fundus images. The proposed algorithm achieves a sensitivity of 72.05% with 94.79% accuracy. CONCLUSIONS: Our proposed algorithm achieved higher accuracy (0.9206) at the peripapillary region, where the ocular manifestations in the microvasculature due to glaucoma, central retinal vein occlusion, etc. are most obvious. This supports the proposed algorithm as a strong candidate for automated vessel segmentation.
Entities:
Keywords:
Color fundus images; Retinal image analysis; Vessel segmentation
Authors: Joes Staal; Michael D Abràmoff; Meindert Niemeijer; Max A Viergever; Bram van Ginneken Journal: IEEE Trans Med Imaging Date: 2004-04 Impact factor: 10.048
Authors: Diego Marin; Arturo Aquino; Manuel Emilio Gegundez-Arias; José Manuel Bravo Journal: IEEE Trans Med Imaging Date: 2010-08-09 Impact factor: 10.048
Authors: João V B Soares; Jorge J G Leandro; Roberto M Cesar Júnior; Herbert F Jelinek; Michael J Cree Journal: IEEE Trans Med Imaging Date: 2006-09 Impact factor: 10.048
Authors: Hai V Nguyen; Gavin Siew Wei Tan; Robyn Jennifer Tapp; Shweta Mital; Daniel Shu Wei Ting; Hon Tym Wong; Colin S Tan; Augustinus Laude; E Shyong Tai; Ngiap Chuan Tan; Eric A Finkelstein; Tien Yin Wong; Ecosse L Lamoureux Journal: Ophthalmology Date: 2016-10-07 Impact factor: 12.079