Abstract
Purpose :
Assessment of retinal nerve fiber layer thickness (RNFLT) is important for managing juvenile glaucoma. However, some children have poor fixation (latent nystagmus, poor attention, etc) that prevent reliable measurement of RNFLT by either a 2-dimensional volume scan or a single circular peripapillary scan. We examined whether RNFL can be reliably measured from repeated misaligned scans. The analysis is possible since multiple OCT scans acquired during nystagmus or unsteady fixation will result with matching locations to a reference scan.
Methods :
Data were collected from retrospective review. Imaging was obtained by combined spectral domain OCT and scanning laser ophthalmoscope (Spectralis OCT-SLO; Heidelberg Engineering). Images were acquired in high speed mode (40,000 A-scans/second 768 x 768 pixels). RNFLT measurements were obtained from circumpapillary (~ 3.5 mm diameter, 768x497 pixels) or single line b-scans. Software was developed that performed semi-automated segmentation of RNFLT and then piecemeal-matching of misaligned scans to a reference circumpapillary scan centered in the optic disc. RNFLT segmentation errors could be manually corrected if needed and the measurements were corrected for rotation of the fovea with respect to the optic disc.
Results :
Twenty-five subjects (mean 9.0 years age; s.d. = 4.9) with either congenital (n=9), or juvenile (n=4) glaucoma, or concerns of secondary (n=9) glaucoma (aphakia, anterior segment dysgenesis, sturge-weber, aniridia), or glaucoma suspect (n = 3). Across all subjects, 5476 sample points obtained within a single visit (mean 219 per subject). The difference in RNFLT from the reference scan averaged 0.84 microns (s.d. 12.5; range -27 to 33). However, errors on individual matches could be > 60 microns. The variance in RNFLT was poorly correlated with visual acuity, cup-to-disc ratio, refractive error, reference global RNFLT, number of matching points, HEYEX image quality score, or presence / absence of automated real-time tracking (ART). There were only minor increases in variability near the vessel arcades.
Conclusions :
This study demonstrates the variability in RNFLT in children with poor fixation and eye movement artifacts. The data have important implications for longitudinal measurements of RNFLT when alignment to a baseline b-scan must be made. We will present future directions in assessing RNFLT when scans are misaligned.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.