Related Experiment Video
Updated: May 14, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
Sector-based optic cup segmentation with intensity and blood vessel priors
Fengshou Yin1, Jiang Liu, Damon W K Wong
1Institute for Infocomm Research, A*STAR, Singapore. fyin@i2r.a-star.edu.sg
Summary
A new sector-based method accurately segments the optic cup for glaucoma diagnosis. This approach refines segmentation using blood vessel features, improving automated cup-to-disk ratio measurement in retinal images.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer-Aided Diagnosis
Background:
- Accurate optic cup segmentation is essential for diagnosing glaucoma.
- Automated cup-to-disk ratio measurement aids in glaucoma diagnosis.
Purpose of the Study:
- To propose a novel sector-based method for optic cup segmentation.
- To improve automated glaucoma diagnosis through enhanced optic cup segmentation.
Main Methods:
- A two-part method: intensity-based segmentation with shape constraints and blood vessel-based refinement.
- Utilizes statistical deformable models on vessel-free images for initial estimation.
- Refines segmentation using extracted blood vessel features like bendings and boundaries.
Main Results:
- Achieved a Dice coefficient of 0.83 for optic cup segmentation.
- Outperformed existing optic cup segmentation methods.
- Demonstrated high accuracy on 650 fundus images from the ORIGA(-light) database.
Conclusions:
- The proposed sector-based method shows significant potential for automated optic cup segmentation.
- This method can enhance computer-aided diagnosis of glaucoma.
- Further development could lead to more reliable glaucoma screening tools.
