Related Experiment Video
Updated: May 24, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
Depth discontinuity-based cup segmentation from multiview color retinal images.
Gopal Datt Joshi1, Jayanthi Sivaswamy, S R Krishnadas
1Centre for Visual Information Technology, International Institute of Information Technology Hyderabad, Hyderabad 500032, India. gopal@research.iiit.ac.in
IEEE Transactions on Bio-Medical Engineering
|February 16, 2012
Summary
This study introduces a novel depth discontinuity method for segmenting the optic cup boundary in retinal images, improving glaucoma assessment. The approach significantly reduces errors in cup-to-disk ratio measurements, especially with multiple views.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Accurate optic cup segmentation is crucial for glaucoma diagnosis.
- Existing methods often focus on the cup region, not its boundary.
- Glaucoma assessment relies on metrics like cup-to-disk ratio (CDR) and cup-to-disk area ratio (CAR).
Purpose of the Study:
- To propose a novel method for estimating the optic cup boundary using depth discontinuity.
- To shift the segmentation focus from the optic cup region to its boundary.
- To improve the accuracy of glaucoma-related measurements derived from retinal images.
Main Methods:
- A depth discontinuity-based approach is used to estimate the cup boundary.
- Utilizes motion boundary and partial occlusion cues from sequential images.
- Approximates the cup boundary with best-fitting circles and a confidence measure.
- Evaluated on synthetic and real retinal image datasets across various multiview scenarios.
Main Results:
- The proposed method achieved a 16% error reduction in CDR and 13% in CAR compared to a monocular segmentation method.
- Incorporating a third view further improved performance, reducing CDR error by 33% and CAR error by 18%.
- Demonstrates superior accuracy in cup boundary segmentation and subsequent ratio estimations.
Conclusions:
- The depth discontinuity-based method offers a promising approach for accurate optic cup segmentation.
- Multiview imaging significantly enhances the performance of the proposed segmentation technique.
- This method has the potential to improve the reliability of glaucoma assessment tools.
