FloatingCanvas: quantification of 3D retinal structures from spectral-domain optical coherence tomography

Area of Science:

• Ophthalmology
• Medical Imaging
• Biomedical Engineering

Background:

• Spectral-domain optical coherence tomography (SD-OCT) offers high-resolution volumetric imaging of retinal structures.
• Accurate segmentation and feature quantification are crucial for leveraging SD-OCT data.
• Existing methods may require pre-processing or lack robustness.

Purpose of the Study:

• To introduce and evaluate FloatingCanvas, a fully automated segmentation algorithm for SD-OCT.
• To assess the algorithm's ability to segment retinal tissue layers and extract features.
• To compare the performance and reproducibility of FloatingCanvas with existing methods.

Main Methods:

• FloatingCanvas performs volumetric segmentation of retinal tissue layers in 3D SD-OCT images around the optic nerve head.
• No pre-processing steps are required for the algorithm.
• Extracted features include blood vessels and retinal nerve fibre layer thickness.

Main Results:

• FloatingCanvas is computationally efficient and robust to image noise and low contrast.
• Segmentation accuracy demonstrated good agreement with manual grading.
• Retinal nerve fibre layer thickness maps are clinically realistic and highly reproducible compared to StratusOCT.

Conclusions:

• FloatingCanvas offers an effective, automated solution for SD-OCT image analysis.
• The algorithm's robustness and reproducibility make it suitable for clinical applications.
• This method enhances the utility of SD-OCT for diagnosing and monitoring retinal conditions.