Automated segmentation of optic nerve head structures with optical coherence tomography

Area of Science:

• Ophthalmology
• Medical Imaging
• Glaucoma Research

Background:

• Accurate segmentation of optic nerve head structures is crucial for glaucoma diagnosis.
• Manual segmentation can be time-consuming and subjective.
• Automated methods offer potential for faster and more consistent analysis.

Purpose of the Study:

• To compare manual and automated segmentation of internal limiting membrane (ILM) and Bruch's membrane opening (BMO) using spectral-domain optical coherence tomography (SD-OCT).
• To quantify differences in BMO-minimum rim width (BMO-MRW) calculations between the two methods.
• To assess the impact of image quality on segmentation accuracy.

Main Methods:

• 107 glaucoma patients and 48 healthy controls underwent SD-OCT imaging.
• Manual and automated segmentation of ILM and BMO were performed.
• BMO-MRW was calculated, and differences (ΔILM, ΔBMO, ΔBMO-MRW) were analyzed.
• Correlation with image quality scores was explored.

Main Results:

• Median ΔILM was 8.9 μm in patients and 7.3 μm in controls.
• Median ΔBMO was 11.5 μm in patients and 12.4 μm in controls.
• Subject-averaged ΔBMO-MRW was not statistically different between patients (13.4 μm) and controls (12.1 μm).
• Image quality did not correlate with segmentation differences.

Conclusions:

• Automated segmentation of ILM and BMO shows small differences compared to manual methods.
• No significant differences were found in BMO-MRW calculations between automated and manual segmentation.
• Automated SD-OCT segmentation is a reliable tool for optic nerve head analysis in glaucoma.