Purchase this article with an account.
Jennifer Hu, Cornelia Gottlieb, Daniel Barajas, Cullen Barnett, Tamera Schoenholz, Mary Durbin, Hari Iyer, Srinivas Sadda; Improved repeatability of retinal thickness measurements using LSO image-based retinal tracking. Invest. Ophthalmol. Vis. Sci. 2013;54(15):3621.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Real-time retinal tracking for optical coherence tomography (OCT) offers potentially better image registration and averaging, but many current implementations require use of cSLO with possibly greater expense and slower performance. We evaluated the performance of a Line Scan Ophthalmoscope (LSO) image-based tracking system.
17 eyes of 17 patients with various retinal diseases presenting to the USC Department of Ophthalmology Retina Clinic underwent spectral domain OCT imaging with the Cirrus HD-OCT (software ver 6.5) and the Spectralis HRA+OCT (software ver 5.3) with a pre-defined protocol using volume scan settings typically used by the imaging unit. Scans were repeated twice on each instrument, and for the Cirrus, 2 more times without tracking (6 scans total for each subject). The time to acquire each scan was recorded. Retinal thickness measurements between scans and devices were compared.
Mean age of patients was 63 years old and 59% were female. Retinal diseases included: diabetic retinopathy, age-related macular degeneration, and epiretinal membrane. Mean visual acuity was logMAR 0.5 (Snellen 20/63). Mean Cirrus foveal central subfield retinal thickness was 280 (SD 75 µ) without tracking and 279 (SD 74 µ) with tracking, compared with a Spectralis retinal thickness (with tracking) of 306 (SD 69 µ). Mean difference between repeat scans without tracking using the Cirrus was 2.4 µ with a repeatability SD of 2.6 µ and a coefficient of variability (COV) of 0.92%. This surprisingly good performance in the absence of tracking likely reflects relatively good vision (and hence good fixation) in this cohort. Repeatability improved with tracking enabled with a mean difference of 1.8 µ, repeatability SD of 1.6 µ, and a COV of 0.6%. The Spectralis demonstrated a mean difference of 5.7 µ with a repeatability SD of 7.2 µ and COV of 2.3%. Mean time to acquire the scans was 4.3, 9.1, and 12.8 seconds for the Cirrus without tracking, Cirrus with tracking, and Spectralis.
Real-time LSO image-based retinal tracking can improve the repeatability of OCT retinal thickness measurements, even in those with relatively good vision and fixation. This level of reproducibility may be particularly important in retinal sublayer analyses where layer thicknesses may be only a few microns. LSO image-based tracking also appears to have speed advantages over cSLO-based approaches.
This PDF is available to Subscribers Only