Automated detection and segmentation of drusen in retinal fundus images. (October 2015)

• Record Type: Journal Article
• Authors: Mittal, Deepti; Kumari, Kajal
• Is Part Of: Computers & electrical engineering. Volume 47(2015)
• Page Start: 82
• Page End: 95
• Publication Date: 2015-10
• Subjects: Drusen -- Age-related macular degeneration -- Segmentation -- Boundary extraction -- Retinal fundus images -- Grading; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2015.08.014
• Languages: English
• ISSNs: 0045-7906

Abstract

Highlights: Designed a new drusen detection and segmentation method finding meaningful drusen boundaries. To find true edges of drusen, a gradient based segmentation procedure is described. Connected component labeling is applied to remove suspicious pixels from drusen region. Edge linking is used to connect all labeled pixels into a meaningful boundary to detect drusen. The performance of proposed method is evaluated by (i) statistical measures and (ii) quantification of drusen to grade severity of age-related macular degradation. The proposed work characterizes the detected drusen in small, intermediate, and large/soft to show its ability to grade age-related macular degradation severity level, helpful in early age-related macular degradation diagnosis. Abstract: The druse, an abnormal yellow/white deposit on retina, is a dominant characteristic of age-related macular degeneration (AMD) which is a retinal disorder associated with age. The early detection of drusen is useful for ophthalmologists to diagnose the patients that suffer from AMD. An automated method has been proposed in this work to detect and segment drusen using retinal fundus images by (i) gradient based segmentation to find true edges of drusen, (ii) connected component labeling to remove suspicious pixels from drusen region and (iii) edge linking to connect all labeled pixels into a meaningful boundary. The proposed method outperforms other existing methods in detection of drusen with an accuracy/sensitivity/specificity of 96.17/89.81/99.00 on two publicly available retinal image databases. In order to grade the severity of AMD, the detected drusen by the proposed method are further quantified into small, intermediate and large with an accuracy of 88.46, 98.55, and 88.37%, respectively. Graphical abstract: