The impact of scanning density on quantitative analyses in optical coherence tomography for choroideremia

Abstract

Purpose : Manual segmentation of dense spectral domain optical coherence tomography (SD-OCT) datasets may be impractical in the setting of many clinical studies. The purpose of this analysis was to determine the impact of reducing the B-Scan density on measurements of ellipsoid zone (EZ) area, choroidal thickness, and total macular volume (TMV) in eyes with choroideremia.

Methods : SD-OCT images were captured for 27 eyes of 20 subjects with choroideremia using the Spectralis SD-OCT (Heidelberg Engineering) as part of an IRB-approved natural history study. Doheny Image Reading Center certified graders analyzed and segmented all 97 B-Scans of the volume cube (spaced at 63 μm apart over a 6.0 mm by 6.0 mm square) for each eye. To compute EZ area, the width of the EZ on each B-scan was manually drawn. In addition, the inner and outer borders of the retina (from internal limiting mebrane to Bruch’s membrane; thus including the retinal pigment epithelium) and choroid (from Bruch’s membrane to the choroid-scleral junction) were manually segmented. Mean choroidal thickness across the macula and TMV were computed. Following this calculation using all 97 B-scans (63 μm apart), the number of B-scans included in the calculation to generate the values was progressively reduced until only 1 of every 10 scans (630 μm between B scans) remained. The percent error or difference between measurements with the reduced B-scan density and the values obtained using all 97 B-scans was calculated.

Results : Table 1 shows the percent error as the B-scan density is reduced. Overall, the TMV appears to be least impacted by reducing B-scab density: using 1 of every 2 scans yielded a 0.148% error, while a B-scan density of 1 of every 10 scans resulted in a 0.345% error. In contrast, a clinically significant error in EZ area was introduced by reducing B-scan density, with the percent error of 6% even with 1 of 2 scans, rising to nearly 50% at 1 of 10 scans.

Conclusions : Whereas total macular volume can be computed reliably using a smaller sampling of B-scans in eyes with choroideremia, accurate calculation of the EZ area requires that a high-density of B-scans be used in the assessment. These findings have implications for the design of OCT image acquisition and grading protocols in choroideremia research studies.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.