CIRRUS HD-OCT User Manual 2660021169012 Rev. A 2017-12Appendix B-15Study 4: Repeatability and Reproducibility of GCA and ONH ParametersA study was conducted to determine the repeatability and reproducibility of the GanglionCell Analysis (GCA) and Optic Nerve Head (ONH) parameters. Sixty–three (63) subjectswere enrolled in this study that was performed in two phases. Phase 1 was inter–operatortesting, where four operators (A, B, C, and D) acquired measurements on one CIRRUSHD-OCT unit (C1). The order of the operators used for scanning each subject wasdetermined using a randomization table. For each subject, a series of twelve (12)RNFL/ONH scans were taken for one eye and twelve (12) macular thickness scans weretaken for the fellow eye for a total of 24 scans. Each of the four operators acquired 3 scansper eye on one CIRRUS 4000 unit. Phase 2 was inter–device testing where four CIRRUSHD-OCT units (C1, C2, C3, C4) were operated by a single operator (A). The order of devicesused for each subject was randomized.Table B-14 presents the repeatability and reproducibility standard deviation (SD) andlimits for the CIRRUS HD-OCT. Among the GCA parameters, the minimum thickness had thelargest repeatability limit and reproducibility limit (8.0165 μm and 8.1018 μm respectively)and the average thickness had the smallest repeatability limit and reproducibility limit(1.6348 μm and 2.0942 μm respectively). The repeatability of the minimum thicknessmeasurement was notably lower than the rest of the parameters; this is expected as theminimum thickness is a more variable measurement from subject to subject, dependingupon disease severity.Among the ONH parameters, the disc area measurement had the largest repeatability limit(0.1506 mm2 ) and the smallest repeatability limit was the cup volume measurement(0.0181 mm 3). The largest reproducibility limit was for the disc area (0.0942 mm2 ) and thesmallest reproducibility limit was for the cup volume measurement (0.0102 mm 3). Thereproducibility limits will affect the ability to determine when measurements have changedue to a change in pathology as opposed to random variability.The optic nerve head algorithm may show increased variability in certain anatomicalvariants. For tilted discs and discs with large clusters of blood vessels, shadowing of theunderlying RPE and Bruch's membrane may render the disc edge difficult to identify.Variability may also be increased due to ambiguity in cup marker placement for small,crowded discs with shallow cups, and discs with large clusters of blood vessels. Cups withexcavation or embryonic tissue remnants may have variable cup volume measurements.The coefficient of variation (CV) was also determined for the CIRRUS HD-OCT GCA andONH parameters. Among the GCA parameters, the minimum thickness had the largest CVof 2.5% and the average thickness had the smallest CV of 0.7%. Among the ONHparameters, the cup volume measurement had the largest CV of 7.8% and the smallest CVwas the rim area of 4.7%.
