@article{ART003246038},
author={Inhwan Kim and Daewon Kim},
title={Corneal Ulcer Discrimination using Double Decoded U-Net},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2025},
volume={30},
number={9},
pages={31-41}
TY - JOUR
AU - Inhwan Kim
AU - Daewon Kim
TI - Corneal Ulcer Discrimination using Double Decoded U-Net
JO - Journal of The Korea Society of Computer and Information
PY - 2025
VL - 30
IS - 9
PB - The Korean Society Of Computer And Information
SP - 31
EP - 41
SN - 1598-849X
AB - Corneal ulcers are caused by damage to the cornea through infection or shock by bacteria, viruses, or fungi. Bacterial corneal ulcers are an urgent disorder that causes symptoms such as pain, foreign body sensation, redness, and light sensitivity, and can even lead to blindness in severe cases. Therefore, a quick diagnosis of early corneal ulcers is necessary, and for this, professional tests such as slit lamp examination and culture test must be performed. In addition, continuous examination by medical staff is required, and patients may find it difficult to receive prompt diagnosis and treatment in the early stages. In this study, after performing pre-processing and post-processing using various image processing techniques, we designed and experimented with a Double Decoded U-Net (DBDU-Net) model that improved the structure of the U-Net model for corneal ulcer area segmentation. The DBDU-Net model is a structure in which one more expanding path is connected to the existing U-Net model, and the contracting path and the feature map extracted from the first expanding path are successively connected to the second expanding path. After learning the location information and object shape of the corneal ulcer area using DBDU-Net, the Dice similarity as an evaluation index showed an average accuracy of 93.32% and a maximum accuracy of 98.66%.
KW - Corneal Ulcer;Semantic Segmentation;Deep-Learning;Double Decoded U-Net
DO -
UR -
ER -
Inhwan Kim and Daewon Kim. (2025). Corneal Ulcer Discrimination using Double Decoded U-Net. Journal of The Korea Society of Computer and Information, 30(9), 31-41.
Inhwan Kim and Daewon Kim. 2025, "Corneal Ulcer Discrimination using Double Decoded U-Net", Journal of The Korea Society of Computer and Information, vol.30, no.9 pp.31-41.
Inhwan Kim, Daewon Kim "Corneal Ulcer Discrimination using Double Decoded U-Net" Journal of The Korea Society of Computer and Information 30.9 pp.31-41 (2025) : 31.
Inhwan Kim, Daewon Kim. Corneal Ulcer Discrimination using Double Decoded U-Net. 2025; 30(9), 31-41.
Inhwan Kim and Daewon Kim. "Corneal Ulcer Discrimination using Double Decoded U-Net" Journal of The Korea Society of Computer and Information 30, no.9 (2025) : 31-41.
Inhwan Kim; Daewon Kim. Corneal Ulcer Discrimination using Double Decoded U-Net. Journal of The Korea Society of Computer and Information, 30(9), 31-41.
Inhwan Kim; Daewon Kim. Corneal Ulcer Discrimination using Double Decoded U-Net. Journal of The Korea Society of Computer and Information. 2025; 30(9) 31-41.
Inhwan Kim, Daewon Kim. Corneal Ulcer Discrimination using Double Decoded U-Net. 2025; 30(9), 31-41.
Inhwan Kim and Daewon Kim. "Corneal Ulcer Discrimination using Double Decoded U-Net" Journal of The Korea Society of Computer and Information 30, no.9 (2025) : 31-41.
