Date of Award

January 2012

Document Type


Degree Name

Medical Doctor (MD)



First Advisor

M B. Shields

Subject Area(s)




Joshua K. Au, Haben Kefella, Nancy Hite, Marc Bodman, M. Bruce Shields, Department of Ophthalmology, Yale University School of Medicine, New Haven CT

PURPOSE: The objective of this study was to design a practical analytical model of interpreting Cirrus Optical Coherence Tomography (OCT) retinal nerve fiber layer (RNFL) data and to use this model to evaluate the ability of Cirrus OCT to diagnose glaucoma severity as compared to the gold standard ophthalmologic exam. While Spectral Domain (SD)-OCT has been studied as an effective tool in glaucoma diagnosis, to the author's knowledge, this is the first study investigating the correlation of SD-OCT RNFL thickness to clinical glaucoma severity. Additionally, as early glaucoma manifests as a focal optic nerve defect affecting the superiotemporal or inferiotemportal RNFL peaks, we hypothesize that increasing difference in peak RNFL thickness may correlate with increasing glaucoma severity. Finally, we sought to further characterize the horizontal deviation according to glaucoma severity with the hypothesis that an increase in the degree of RNFL peak deviation correlates with increasing glaucoma severity.

METHODS: A retrospective cross sectional study of 148 patients undergoing Cirrus SD-OCT imaging and clinical ophthalmologic exam was performed. Eyes were excluded for scans with signal strength less than 6. In total, 263 eyes were included. Each eye was graded in a masked fashion by a glaucoma specialist on a scale of 1 to 4 of increasing glaucoma severity - normal, glaucoma suspect, moderate glaucoma, severe glaucoma, respectively. An independent Cirrus OCT grading scheme was developed to generate a corresponding glaucoma severity grading scale from 1 to 4. Peak RNFL difference and horizontal deviation were measured and calculated. Clinically significant peak difference was considered > 50 μm and clinically significant horizontal deviation was considered >20 degrees. Analysis of variance and Pearson correlation studies were used to assess the correlation between study parameters and glaucoma severity.

RESULTS: The Cirrus OCT glaucoma severity grading model correlated significantly with clinical glaucoma severity (r = 0.57, p-value<0.001). The sensitivity and specificity of the Cirrus OCT grade was 65.10% and 81.67%, respectively. RNFL peak horizontal deviation at the superior and inferior poles also significantly correlated with clinical glaucoma severity (p-value <0.001). Horizontal deviation increased in frequency and degree with increasing glaucoma severity.

CONCLUSIONS: We have developed a novel analytical model for utilizing the Cirrus OCT in diagnosing glaucoma severity. We have reported that it is a robust model with significant correlation to clinical glaucoma severity grading. The main limitation of this Cirrus OCT glaucoma severity diagnostic model is in identifying glaucoma suspects and moderate glaucoma. It correlates strongly with clinical diagnoses of the extreme, normal and severely glaucomatous patients. Furthermore, we have further elucidated the recently published phenomenon of RNFL peak horizontal deviation. While, anatomical deviations of RNFL peak in normal, healthy eyes may exist, this study suggests that increasing glaucoma severity may also influence horizontal deviation, which increases both in frequency and degree with increasing glaucoma severity. As glaucoma damages the optic nerve, the structural result is not only neural atrophy and thinning of the RNFL but also a regional restructuring of the RNFL resulting in significant deviation of the peak.


This thesis is restricted to Yale network users only. This thesis is permanently embargoed from public release.