Related Experiment Video
Updated: May 29, 2026
09:17
Using Retinal Imaging to Study Dementia
Published on: November 6, 2017
Interactive segmentation for geographic atrophy in retinal fundus images
Noah Lee1, R Theodore Smith, Andrew F Laine
1Biomedical Engineering Department, Columbia University, New York, NY 10027 USA (phone: 212-854-5996; fax: 212-854-5995).
Conference Record. Asilomar Conference on Signals, Systems & Computers
|August 26, 2011
Summary
This study introduces a new method for automatically segmenting geographic atrophy (GA) in fundus auto-fluorescence (FAF) images. The novel approach significantly improves the accuracy of GA quantification, aiding in age-related macular degeneration (AMD) diagnosis.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Age-related macular degeneration (AMD) is a leading cause of blindness.
- Geographic atrophy (GA), an advanced form of AMD, causes severe visual loss.
- Accurate quantification of GA is crucial for disease progression monitoring and clinical diagnosis.
Purpose of the Study:
- To develop and evaluate an intuitive and simple approach for automatic geographic atrophy segmentation.
- To compare the proposed method with the state-of-the-art random walker algorithm for interactive segmentation.
Main Methods:
- Leveraging watershed transform and generalized non-linear gradient operators for interactive segmentation.
- Quantitative evaluation using Receiver Operating Characteristic (ROC) statistics on 100 FAF images.
Main Results:
- The proposed approach achieved a mean sensitivity of 98.3% and specificity of 97.7%.
- The random walker algorithm achieved a mean sensitivity of 88.2% and specificity of 96.6%.
Conclusions:
- The developed method offers a highly accurate and effective solution for geographic atrophy segmentation.
- This technique has the potential to significantly improve the clinical diagnosis and management of AMD.
