Application of an attention U-Net incorporating transfer learning for optic disc and cup segmentation
- PDF / 4,881,645 Bytes
- 9 Pages / 595.276 x 790.866 pts Page_size
- 35 Downloads / 186 Views
ORIGINAL PAPER
Application of an attention U-Net incorporating transfer learning for optic disc and cup segmentation Xiaoye Zhao1 · Shengying Wang1 · Jing Zhao3 · Haicheng Wei2
· Mingxia Xiao1 · Na Ta1
Received: 1 April 2020 / Revised: 4 October 2020 / Accepted: 4 November 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract Optic disc (OD) and optic cup (OC) regional parameters are of utmost importance in the early diagnosis of glaucoma. Improving the accuracy of OD/OC segmentation results and parameter extraction in colour fundus images plays a very important role in early glaucoma screening. To improve the accuracy and inference speed of fundus image segmentation, an algorithm for fundus image segmentation based on an attention U-Net with transfer learning is proposed in this paper. First, an attention gate was added between the encoder and decoder of U-Net to focus on the target areas, thus forming the architecture of the attention U-Net. Then, after the network had been trained on the DRIONS-DB dataset to partially obtain the weights of the encoder, it was trained on the Drishti-GS dataset to further modify the weights. Finally, the trained attention U-Net model incorporating transfer learning was used to segment fundus images. OD/OC extraction using this method shows obvious advantages in model parameter quantity and inference time compared with existing algorithms, the parameter quantity is much smaller than that of existing algorithms, and the model inference time is 0.33 s, representing a reduction of more than 50%. The proposed method can be applied to a fundus image dataset with only a small number of labels. Whilst offering fast OD/OC segmentation, it also guarantees a relatively high segmentation accuracy. Keywords Image segmentation · Optic disc and cup · U-Net · Attention · Transfer learning
1 Introduction Glaucoma is defined as a neurodegenerative disease of the optic nerve, which means that there is no regeneration capacity for the optic nerve and retina. The abnormal increase in intraocular pressure in glaucoma patients leads to compression of the peripheral optic nerve, which is a major cause of irreversible blindness [1]. Studies have previously demonstrated that morphological changes in the optic disc (OD) Electronic supplementary material The online version of this article (https://doi.org/10.1007/s11760-020-01815-z) contains supplementary material, which is available to authorized users.
B
Haicheng Wei [email protected]
1
School of Electrical and Information Engineering, North Minzu University, Yinchuan 750011, Ningxia, China
2
Basic Experimental Teaching and Engineering Training Center, North Minzu University, Yinchuan 750011, Ningxia, China
3
School of Information Engineering, Ningxia University, Yinchuan 750011, Ningxia, China
and optic cup (OC) generally precede visual field defects in glaucoma. Thus, changes in the parameters of the OD/OC could reflect damage to the optic nerve [2]. Therefore, to improve diagnostic efficiency for glaucoma, it is especia
Data Loading...
