Deep feature extraction via adaptive collaborative learning for drusen segmentation from fundus images
- PDF / 5,879,992 Bytes
- 8 Pages / 595.276 x 790.866 pts Page_size
- 66 Downloads / 201 Views
ORIGINAL PAPER
Deep feature extraction via adaptive collaborative learning for drusen segmentation from fundus images Xiuxiu Ren1,2 · Xiangwei Zheng1,2 · Xiao Dong1 · Xinchun Cui3 Received: 9 February 2020 / Revised: 7 June 2020 / Accepted: 2 November 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract Drusen are an early sign of non-neovascular age-related macular degeneration which is a major factor of irreversible blindness. Drusen segmentation plays a vital role in proper diagnosis and prevention of further complications. However, most of the existing drusen segmentation approaches rely on handcrafted features which are not always guaranteed to be discriminative and therefore lead to limited performance. In this paper, we propose a deep feature extraction framework for drusen segmentation. It is formulated as a deep model which can automatically extract discriminative features. Specifically, the framework is mainly composed of three components, including feature learning, loss function and classification. The effectiveness of our method lies in the fact that the deep feature learning procedures are driven by an adaptive collaborative similarity learning technique in loss function. We evaluate the framework on STARE and DRIVE datasets, and the quantitative comparison with the state-of-the-art methods in terms of sensitivity, specificity and accuracy demonstrates the superiority of the proposed method. Keywords Drusen segmentation · Fundus images · Adaptive collaborative learning · Deep feature extraction
1 Introduction Age-related macular degeneration (AMD) is the leading cause of legal blindness among the elderly in the developed countries [11,20]. It is a kind of maculopathy that affects central vision of the visual field. The appearance of drusen is the major sign of early AMD. Drusen are considered as key clinical characteristics and significant risk factors for the development of AMD. They are focal deposits of extracellular material caused by the accumulation of rod and cone metabolism located beneath the retinal pigment epithelial (RPE) cell layer [1]. Clinically, drusen are presented as yellowish blobs in or around the macula of the retina. They are roughly divided into two morphologic groups, i.e., hard
B
Xiangwei Zheng [email protected]
1
School of Information Science and Engineering, Shandong Normal University, Jinan, China
2
Shandong Provincial Key Laboratory for Distributed Computer Software Novel Technology, Shandong Normal University, Jinan, China
3
School of Information Science and Engineering, Qufu Normal University, Rizhao, China
drusen and soft drusen. Figure 1 shows the examples of different drusen and their surface topographies. Hard drusen usually appear as small and round spots with well-defined borders. Typically, soft drusen are more irregular and larger; meanwhile, they have nonhomogeneous intensity and blurred contours. Moreover, the size and number of drusen, especially soft drusen, are essential findings that reveal the disease severity and predict t
Data Loading...