Related Experiment Video
Updated: Jun 8, 2025
07:23
Retinal Vascular Reactivity as Assessed by Optical Coherence Tomography Angiography
Published on: March 26, 2020
7.1K
Quantitative characterization of retinal features in translated OCTA.
Rashadul Hasan Badhon1, Atalie Carina Thompson2, Jennifer I Lim3
1Department of Electrical and Computer Engineering, University of North Carolina at Charlotte, Charlotte, NC, United States.
Experimental Biology and Medicine (Maywood, N.J.)
|November 7, 2024
Summary
Generative machine learning translates Optical Coherence Tomography Angiography (OCTA) from OCT scans, enabling objective retinal disease diagnosis. This approach enhances accessibility to detailed vascular imaging for improved clinical detection.
Area of Science:
- Ophthalmology and Medical Imaging
- Artificial Intelligence in Healthcare
- Biomedical Engineering
Background:
- Optical Coherence Tomography Angiography (OCTA) provides detailed retinal vasculature imaging but faces adoption limitations.
- Quantitative OCTA features are crucial for characterizing retinal vascular changes and diagnosing diseases.
- Generative machine learning (ML) offers potential for image translation and feature extraction.
Purpose of the Study:
- To explore the feasibility of using generative ML to translate Optical Coherence Tomography (OCT) images into quantitative OCTA features.
- To characterize retinal vascular changes using translated OCTA (TR-OCTA) for objective disease diagnosis.
- To assess the potential of TR-OCTA for enhancing the clinical diagnostic process for retinal diseases.
Main Methods:
- Employed a generative adversarial network (GAN) framework with 2D vascular segmentation and OCTA image translation models.
- Trained models on the OCT-500 public dataset and validated with data from the University of Illinois at Chicago (UIC) retina clinic.
- Evaluated TR-OCTA quality and quantitative vascular features (tortuosity, vessel perimeter index, density) against ground truth OCTA (GT-OCTA).
Main Results:
- TR-OCTAs demonstrated high image quality (resolution, contrast) and moderate structural similarity to GT-OCTAs across different scanning ranges.
- Vascular features like tortuosity and vessel perimeter index showed consistent trends, outperforming density features affected by local distortions.
- Validation on unseen UIC data confirmed similar trends, indicating robust inference performance despite model blindness to this dataset.
Conclusions:
- Generative ML-based OCTA translation is feasible for characterizing retinal vascular changes.
- TR-OCTA enables objective disease diagnosis by providing reliable vascular features, addressing OCTA adoption limitations.
- This technology enhances accessibility to detailed vascular imaging, potentially improving retinal disease diagnostics and reducing reliance on specialized equipment.
