A contrastive learning-based segmentation network for multiple retinal lesions

Xingfang Ai

doi:10.1117/12.2658204

20 October 2022 A contrastive learning-based segmentation network for multiple retinal lesions

Xingfang Ai

Proceedings Volume 12451, 5th International Conference on Computer Information Science and Application Technology (CISAT 2022); 124515O (2022) https://doi.org/10.1117/12.2658204
Event: 5th International Conference on Computer Information Science and Application Technology (CISAT 2022), 2022, Chongqing, China

Abstract

Diabetes retinopathy is a serious eye impairment, which can lead to vision loss or blindness. An efficient marking and detection system may help early detection of DR and DME diseases. In this work, we applied patch-based CNN – MDRNet, for pixel-wise segmentation of multiple lesion associated with diabetic retinopathy. We further propose a SimCLR based MDR-Net to extract better feature representation and improve the performance of MDR-Net. The experimental results show that, MDR-Net achieves 0.768, 0.657, 0.478 and 0.607 AUC for microaneurysms (MA), hemorrhages (HE), hard exudates (EX), and soft exudates (SE), respectively. Compared to the MIB-ANet without pretraining, the application of SimCLR improves AUC of EX, HE, MA and SE segmentation by at least 2.8%, 2.4%, 1.9% and 3.5%, respectively.

Citation Download Citation

Xingfang Ai "A contrastive learning-based segmentation network for multiple retinal lesions", Proc. SPIE 12451, 5th International Conference on Computer Information Science and Application Technology (CISAT 2022), 124515O (20 October 2022); https://doi.org/10.1117/12.2658204

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available