Reproducibility and repeatability of optical coherence tomography imaging of the optic nerve head in normal beagle eyes

• Published in: Veterinary ophthalmology Vol. 20; no. 6; pp. 480 - 487
• Main Authors: Bemis, Alexandra M., Pirie, Christopher G., LoPinto, Alexander J., Maranda, Louise
• Format: Journal Article
• Language: English
• Published: England 01.11.2017
• Subjects: canine; nerve fiber layer; optic nerve; optical coherence tomography; canine; optical coherence tomography; optic nerve; nerve fiber layer
• ISSN: 1463-5216; 1463-5224
• DOI: 10.1111/vop.12451

Objective To assess the reproducibility and repeatability of an optical coherence tomography (OCT) device for imaging the optic nerve region of normal canines. Animals studied Twelve clinically healthy beagles. Procedures All animals were anesthetized, and an OCT device was used to image the optic disk region. Total disk area and retinal nerve fiber layer (NFL) thickness in eight segments were obtained from each image. Images were collected by two operators in succession and each operator took five scans, repositioning the device between measurements. B‐scan segmentations and disk outlines were subsequently redrawn to obtain manual measurements, allowing for comparisons between automated measurements. Results Operator 1 had a mean (±SD) NFL thickness (automatic) and disk area of 73.38 ± 7.41 μm and 3.69 ± 0.52 mm², while operator 2 had 74.27 ± 7.33 μm and 3.67 ± 0.51 mm², respectively. With manual corrections, operator 1 had a NFL thickness and disk area of 86.19 ± 8.26 μm and 3.74 ± 0.68 mm², while operator 2 had 86.85 ± 6.91 μm and 3.81 ± 0.56 mm², respectively. Operators did not obtain significantly different values for any region. Intraclass correlation ranged from 0.33 to 0.97 (average 0.78). Coefficient of variation ranged from 8.8 to 36.2%. Conclusion The OCT device demonstrated high inter‐rater reliability. Intrarater reliability was strong for disk area and total NFL; however, the segmental regions demonstrated a higher degree of variability. Manual and automated measurements were significantly different for total NFL and some segmental regions; therefore, users should consistently use one method for longitudinal studies.