OVERCOMING SEGMENTATION ERRORS IN MEASUREMENTS OF MACULAR THICKNESS MADE BY SPECTRAL-DOMAIN OPTICAL COHERENCE TOMOGRAPHY

Abstract

Purpose: To assess the frequency and severity of segmentation errors in spectral-domain optical coherence tomography, and present an appropriate scan protocol for measuring macular thickness. Methods: In this study we examined 40 eyes of healthy subjects, 45 eyes with retinal pathology and 31 eyes with subretinal pathology. Patients were prospectively imaged using 12 radial scans and 3-dimensional macular cube scans by spectral-domain optical coherence tomography (3D OCT-1000; Topcon Corp., Japan) at the same sitting. Retinal segmentation errors were noted and graded using a subjective, categoric error scale to generate an error score. We compared the macular thickness measurements with and without error correction, and between the two scan protocols. Results: In this series, 63.8% of eyes (74 of 116 eyes) gave at least 1 segmentation error. Errors were more common in eyes with pathology. There was a significant difference between measurements of macular thickness with and without correcting these errors, regardless of scan protocol (P < 0.05). After error correction of both scan protocols, the macular thickness measurements obtained with the 12 radial scan protocol were equivalent to those obtained with the 3-dimensional macular cube scan protocol. Conclusion: Segmentation errors were frequent on scans obtained by spectral-domain optical coherence tomography. We recommend the 12 radial scan protocol with error correction as the standard protocol for measuring macular thickness, particularly in clinical studies.