Improve glaucoma diagnosis using retinal surface contour variability measured by OCT

Oregon Health & Science University, Portland OR

Investigators

Linked publications, trials & patents

Abstract

PROJECT SUMMARY Glaucoma is a leading cause of blindness. Early diagnosis of glaucoma is important because its onset is insidious and the damage is irreversible. Advanced imaging modalities such as optical coherence tomography (OCT) have been used in the past 2 decades to improve the objective evaluation of glaucoma. OCT has higher axial spatial resolution than other posterior eye imaging modalities, and it has relatively good diagnostic accuracy and reproducibility in the measurement of neural structures damaged by glaucoma. However, the primary OCT diagnostic parameters - circumpapillary nerve fiber layer (NFL) and macular ganglion cell complex (GCC) thicknesses, have limited sensitivity for detecting early glaucoma. Recently, we have developed a new structural parameter, retinal surface contour variability (RSCV), which is based on spatial frequency analysis of inner limit membrane (ILM) contour. RSCV detects focal nerve fiber bundle loss and increased relief height of larger retinal blood vessels relative to a thinned NFL. These contrasts may more reliably distinguish early glaucoma from physiologically thin NFL (normal population variation). Preliminary results show that combination of RSCV and NFL has higher diagnostic accuracy than nerve fiber layer (NFL) thickness in a study using a prototype high-speed swept-source OCT system. However, the study was too small to be definitive and the prototype is not widely available. Accordingly, the goal of the proposed project is to adapt this new approach to a standard commercial clinical spectral-domain OCT system and validate it using a large clinical database from the recently completed Advanced Imaging for Glaucoma (AIG) study, which has longitudinal data on 249 perimetric glaucoma (PG), 394 glaucoma suspect/pre-perimetric glaucoma (GS/PPG), and 145 normal participants followed for an average of 5 years. The specific aims are: 1. Optimize automatic algorithm for RSCV. (1) Refine the automated segmentation algorithm to more reliably distinguish the ILM from epiretinal membrane and vitreous face. (2) Optimize the spatial frequency band included in the RSCV parameter using the AIG baseline. 2. Improve diagnosis of early glaucoma by combining RSCV, NFL, and GCC parameters. (1) Use the AIG data to train a Structural Glaucoma Index (SGI) that combines RSCV, NFL, GCC, demographic, and ocular biometric parameters using multivariate liner regression. Establish glaucoma diagnostic criteria. (2) Evaluate the diagnostic accuracy of RSCV and SGI using the AIG data (leave-one-out validation), and a fully independent test data set (Functional and Structural OCT in Glaucoma study at a different site). 3. Predict glaucoma conversion using RSCV. Test if baseline RSCV could improve the accuracy of predicting which GS/PPG eyes in the AIG longitudinal study would convert to perimetric glaucoma (develop glaucoma visual field defects) and detect abnormality at an earlier time point relative to conversion.