Registration-Based Analysis of the Three-Dimensional Shape of the Retinal Nerve Fiber Layer for Detection of Glaucomatous Defects.

PURPOSE

Percentile maps of retinal nerve fiber layer (NFL) thickness identify perimetric damage in patients with glaucoma, but normal variability in NFL shape means that some healthy eyes will be scored as having arcuate NFL defects. Registration is a method from brain imaging that compensates for differences in shape across subjects. This study assessed the potential of registration as an alternative method for detecting NFL defects.

METHODS

Widefield optical coherence tomography retinal scans (55° wide, 40° high) and circumpapillary NFL thickness scans were analyzed for 29 patients with glaucoma and 112 healthy controls. Widefield scans from 3 sets of 12 control eyes each were used to compute three different templates for NFL shape. Nonrigid registration was performed for each eye with each template, generating a map showing locations where the registration failed to align the NFL with the template. Registration failures were counted for superior and inferior retinal hemifields. Perimetric data for patients were used to determine which hemifields had perimetric defects.

RESULTS

Conditions were identified for which registration for all three templates performed similarly to circumpapillary NFL thickness in identifying hemifields of patients with perimetric defects.

CONCLUSIONS

Registration-based analysis for NFL shape showed the ability to detect glaucomatous perimetric defects, and results identified areas for improvement in NFL registration methods.

TRANSLATIONAL RELEVANCE

Registration analysis has the potential to provide an alternate method to identify NFL defects in patients with glaucoma.