Statistical methods for modeling the spatial structure on the visual field in glaucoma progression research Public Deposited

Downloadable Content

Download PDF
Last Modified
  • March 22, 2019
  • Berchuck, Samuel
    • Affiliation: Gillings School of Global Public Health, Department of Biostatistics
  • Diagnosing glaucoma progression is critical for limiting irreversible vision loss. A common method for assessing glaucoma progression uses a longitudinal series of visual fields (VF) acquired at regular intervals. VF data are characterized by a complex spatiotemporal structure due to the data generating process and ocular anatomy. Thus, advanced statistical methods are needed to make clinical determinations regarding progression status. In this dissertation, we introduce new modeling techniques that produce flexible spatial dependency structures within the framework of hierarchical Bayesian spatial models. The developed methodology is applied to VF data from the Vein Pulsation Study Trial in Glaucoma and the Lions Eye Institute trial registry. In chapter 2, we work within the framework of boundary detection and introduce a spatiotemporal boundary detection model that allows the underlying anatomy of the optic disc to dictate localized spatial structure on the VF. We show that our new method provides insight into vision loss that improves diagnosis of glaucoma progression in actual glaucoma patients. Simulations are presented, showing the proposed methodology is preferred over existing spatial methods for VF data. An R package womblR is provided that implements the method. Chapter 3 aims to introduce the modeling framework from chapter 2 to the ophthalmology community. An optimal form of the metric is established and compared with standard methods for assessing glaucoma progression using a statistical diagnostic framework. In particular, we demonstrate the added value of using the novel metric in addition to established prediction models based on standard operating characteristics. Finally, we detail the procedure for implementing our novel technique in the clinical setting. In chapter 4, we present a framework that brings together vital aspects of glaucoma management, i) prediction of future VF sensitivities, ii) predicting the timing and location of future vision loss, iii) making clinical decisions regarding progression, and, iv) incorporation of anatomical information to create plausible data-generating models. We show that our method improves prediction and estimation of progression, and simulations are presented, showing the proposed methodology is preferred over existing models for VF data. An R package called spCP is provided that implements the method.
Date of publication
Resource type
  • Fine, Jason
  • Herring, Amy
  • Ibrahim, Joseph
  • Mwanza, Jean-Claude
  • Warren, Joshua
  • Doctor of Philosophy
Degree granting institution
  • University of North Carolina at Chapel Hill
Graduation year
  • 2018

This work has no parents.