Related Experiment Video
Updated: Jun 24, 2025
07:18
Evaluation of Capillary and Other Vessel Contribution to Macular Perfusion Density Measured with Optical Coherence Tomography Angiography
Published on: February 18, 2022
1.8K
Choroidalyzer: An Open-Source, End-to-End Pipeline for Choroidal Analysis in Optical Coherence Tomography.
Justin Engelmann1,2, Jamie Burke3, Charlene Hamid4
1School of Informatics, University of Edinburgh, Edinburgh, United Kingdom.
Investigative Ophthalmology & Visual Science
|June 4, 2024
Summary
Choroidalyzer is a new open-source pipeline for segmenting the choroid and calculating its thickness, area, and vascular index. This automated tool offers objective and standardized analysis for choroidal imaging research.
Area of Science:
- Ophthalmology and Medical Imaging
- Artificial Intelligence in Healthcare
- Biomedical Image Analysis
Background:
- Accurate segmentation of the choroid and its vasculature is crucial for understanding various eye diseases.
- Manual segmentation of choroidal structures is time-consuming and prone to inter-observer variability.
- Existing automated methods often lack comprehensive analysis of choroidal thickness, area, and vascularity.
Purpose of the Study:
- To develop Choroidalyzer, an open-source, end-to-end pipeline for automated segmentation of the choroid.
- To enable precise derivation of choroidal thickness, area, and vascular index from OCT B-scans.
- To validate the performance of Choroidalyzer against manual segmentation and grading.
Main Methods:
- Utilized a dataset of 5600 OCT B-scans from 233 subjects across diverse cohorts and devices.
- Employed manual correction of state-of-the-art automatic methods for ground-truth generation.
- Trained a U-Net deep learning model for segmentation of choroidal region, vessels, and fovea.
Main Results:
- Choroidalyzer demonstrated excellent region segmentation (Dice: 0.9789 internal, 0.9749 external) and fovea localization (MAE: 3.4 external pixels).
- Achieved very good vessel segmentation performance (Dice: 0.8703 external) and high correlation for choroidal metrics (Pearson r > 0.79).
- Performance agreement with manual graders was comparable to intergrader agreement across all metrics.
Conclusions:
- Choroidalyzer provides an accurate and reliable open-source solution for choroidal analysis.
- The pipeline offers objectivity and standardization, particularly for the challenging task of choroidal vessel segmentation.
- This automated approach can significantly advance research in systemic diseases affecting the choroid.
