Optical Coherence Tomography Segmentation Errors of the Retinal Nerve Fiber Layer Persist Over Time

Area of Science:

• Ophthalmology
• Medical Imaging
• Glaucoma Research

Background:

• Automated segmentation of the retinal nerve fiber layer (RNFL) in optical coherence tomography (OCT) can have errors, particularly in glaucoma suspects or those with mild glaucoma.
• These segmentation errors may impact the reliability of RNFL thickness measurements over time.

Purpose of the Study:

• To determine if optical coherence tomography (OCT) segmentation errors in RNFL thickness measurements persist longitudinally.
• To assess the repeatability and sensitivity of automated versus manually refined RNFL segmentation for glaucoma progression monitoring.

Main Methods:

• A cohort study involving 406 eyes from 213 participants using spectral domain OCT (Spectralis).
• RNFL thickness was measured in a 6-degree peripapillary circle, comparing automated segmentation results with those after manual refinement.
• Bland-Altman plots and linear regression analyzed the magnitude, location, and repeatability of segmentation errors at baseline and follow-up points up to 4 years.

Main Results:

• Segmentation errors at baseline showed a high correlation (r>0.5) with errors at follow-up time points.
• Automated segmentation exhibited smaller standard deviation of residuals (1.56 μm vs. 1.80 μm) compared to manual refinement.
• Automated segmentation demonstrated a higher ability to monitor progression, indicated by longitudinal signal-to-noise ratio analysis.

Conclusions:

• Errors in automated RNFL segmentation are relatively stable and tend to persist longitudinally in both direction and magnitude.
• Automated segmentation, without manual refinement, offers greater repeatability and may be more sensitive for detecting glaucomatous progression.
• Future advancements in segmentation algorithms can leverage these findings to enhance automated segmentation accuracy.