Purchase this article with an account.
Xinbo Zhang, Ou Tan, Rohit Varma, Joel Schuman, David Greenfield, David Huang; Combining Nerve Fiber Layer and Ganglion Cell Complex Parameters for Glaucoma Diagnosis using Fourier-Domain Optical Coherence Tomography. Invest. Ophthalmol. Vis. Sci. 2013;54(15):2281.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To define an optical coherence tomography (OCT) based structural diagnostic index (SDI) for glaucoma diagnosis by optimally combining nerve fiber layer (NFL) and ganglion cell complex variables.
We analyzed the data from participants multi-center longitudinal Advanced Imaging for Glaucoma Study (www.AIGStudy.net). A Fourier-domain OCT system (RTVue) was used to map GCC and NFL thickness 3 times on each study visit. Five types of parameters: overall, superior and inferior average thickness, along with global loss volume (GLV) and focal loss volume (FLV) for both NFL and GCC are used. The study subjects were randomly divided into a training set and a test set. A two-stage approach was used. First, logistic regression was performed on training set in each parameter type for NFL and GCC to determine optimal weights for NFL and GCC then combine the two parameters into a single one. Next the weight-combined parameters were standardized to the reference values from the normal group. The quadratic and positively truncated form of the standardized parameters from the five types entered a logistic regression to define a single numeric value as the optimal SDI.
The analysis included 197 eyes (99 participants) from normal group; 210 eyes (141 participants) from perimetric glaucoma (PG) group. The training set contains 100 N and 100 PG eyes and the test set contains 97 N and 110 PG eyes. The optimal SDI from the logistic regression developed from the training set included superior and inferior average thicknesses, GLV and FLV, but excluded overall thickness. The area under ROC curve (AROC) for the SDI on the test set is 0.927, significantly better than the AROC for any of the component variables used alone (p=0.03). The AROC for component NFL and GCC variables ranged from 0.86 to 0.91.
Combining structural measurements of GCC and NFL from Fourier-domain OCT improved the diagnostic accuracy for glaucoma.
This PDF is available to Subscribers Only