Related Experiment Video
Updated: Dec 25, 2025
08:54
Author Spotlight: Understanding Age-Related Macular Degeneration Pathophysiology with QAF Workflow
Published on: May 26, 2023
2.0K
Automated Quantification of Photoreceptor alteration in macular disease using Optical Coherence Tomography and Deep
José Ignacio Orlando1, Bianca S Gerendas1, Sophie Riedl1
1Department of Ophthalmology, Medical University of Vienna, Waehringer Guertel 18-20, 1090, Vienna, Austria.
Scientific Reports
|March 30, 2020
Summary
A new deep learning model automatically quantifies photoreceptor layer changes in macular diseases like diabetic macular edema (DME) and retinal vein occlusion (RVO), matching expert accuracy.
Area of Science:
- Ophthalmology and Medical Imaging
- Artificial Intelligence in Healthcare
Background:
- Diabetic macular edema (DME) and retinal vein occlusion (RVO) are leading causes of vision loss, affecting central photoreceptors.
- Assessing photoreceptor integrity is crucial for diagnosing and managing these macular diseases, but manual quantification is laborious.
- Optical coherence tomography (OCT) is a key imaging modality for evaluating photoreceptor layer status.
Purpose of the Study:
- To develop and validate a deep learning approach for automated segmentation and characterization of photoreceptor alterations in OCT images.
- To improve the efficiency and accuracy of quantifying photoreceptor layer thickness and abnormalities in DME and RVO patients.
Main Methods:
- An ensemble of four convolutional neural networks (CNNs) was employed for segmenting the photoreceptor layer.
- En-face representations were generated to characterize layer thickness, and pixel-wise standard deviation of model outputs indicated abnormalities.
- The automated method's performance was compared against manual annotations by a human expert.
Main Results:
- The deep learning ensemble achieved performance comparable to human experts in segmenting and characterizing photoreceptor alterations.
- The automated quantification of mean photoreceptor thickness showed no statistically significant difference compared to manual measurements.
- The ensemble model outperformed individual CNN models, demonstrating the benefit of combining multiple networks.
Conclusions:
- The developed deep learning model reliably quantifies photoreceptor integrity in macular diseases, offering a significant advancement over manual methods.
- Accurate and automated photoreceptor quantification can enhance diagnostic capabilities and improve patient prognosis and management strategies for DME and RVO.
- This AI-driven approach has the potential to streamline clinical workflows and improve patient outcomes in retinal diseases.
