Related Experiment Video
Updated: Oct 14, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.0K
GDCSeg-Net: general optic disc and cup segmentation network for multi-device fundus images
Qianlong Zhu1, Xinjian Chen1,2, Qingquan Meng1
1MIPAV Lab, School of Electronics and Information Engineering, Soochow University, Jiangsu 215006, China.
Biomedical Optics Express
|November 8, 2021
Summary
This study introduces GDCSeg-Net, a novel deep learning model for segmenting the optic disc and optic cup in retinal images. It improves generalization across diverse datasets, aiding glaucoma diagnosis.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Accurate optic disc (OD) and optic cup (OC) segmentation is vital for diagnosing retinal diseases like glaucoma.
- Existing deep learning models struggle with generalization due to domain shifts and limited training data.
Purpose of the Study:
- To develop a generalizable deep learning network for OD and OC segmentation across varied fundus image datasets.
- To address poor generalization caused by domain shift and small sample sizes in existing methods.
Main Methods:
- Proposed GDCSeg-Net, an encoder-decoder network utilizing a mixed training strategy across multiple datasets.
- Introduced a novel multi-scale weight-shared attention (MSA) module and densely connected depthwise separable convolution (DSC) module.
Main Results:
- GDCSeg-Net demonstrated competitive performance against state-of-the-art methods.
- The model was evaluated on five diverse public fundus image datasets: REFUGE, MESSIDOR, RIM-ONE-R3, Drishti-GS, and IDRiD.
Conclusions:
- The proposed GDCSeg-Net effectively overcomes generalization issues in OD and OC segmentation.
- The mixed training strategy and novel modules enhance performance on varied fundus image datasets.
