Related Experiment Video
Updated: May 22, 2026
09:17
Using Retinal Imaging to Study Dementia
Published on: November 6, 2017
Fast localization and segmentation of optic disk in retinal images using directional matched filtering and level sets
1VisionQuest Biomedical, Albuquerque, NM 87106, USA. honggang@ece.unm.edu
Summary
A new algorithm accurately locates and segments the optic disk (OD) for automatic eye disease screening. This fast, robust method is crucial for classifying retinal and optic nerve pathology.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Accurate optic disk (OD) localization and segmentation are vital for diagnosing retinal and optic nerve pathologies.
- Current methods for OD analysis often require manual intervention, limiting efficiency in large-scale screening.
Purpose of the Study:
- To develop a fast, fully automatic algorithm for optic disk (OD) localization and segmentation.
- To enhance the reliability and efficiency of automatic eye disease screening.
Main Methods:
- Template matching was used to identify initial OD location candidates, adapting to various image resolutions.
- Vessel patterns on the OD refined location detection.
- A hybrid level-set model, integrating region and gradient information, segmented the disk boundary.
- Morphological filtering removed interfering structures like blood vessels.
Main Results:
- The OD localization achieved a 99% success rate on 1200 images from the MESSIDOR database (1189/1200).
- The average segmentation boundary error was 10% of the estimated OD radius across all image sizes.
- The algorithm demonstrated efficiency, robustness, and accuracy.
Conclusions:
- The developed algorithm provides a reliable and efficient solution for automatic OD localization and segmentation.
- This technique is suitable for widespread application in automatic retinal disease screening across diverse clinical settings.
