Introduction: RTA imaging covers an area of 20* 20 degree of the central macula where 50% of the total number of the retinal ganglion cells are located. The ganglion cell layer and the nerve fiber layer in this region constitute 33% of the retinal thickness. A previous study has shown that the RTA is capable to detect 13% thinning of the retina in this region with a confidence of 97.5%. Therefore, the posterior pole is an applicable area for identifying glaucomatous changes with the RTA.
Purpose: I. To present two novel parameters for identifying glaucomatous changes in the posterior pole utilizing the results of the RTA – the PPGI and neural network. II. To assess the specificity and sensitivity of both methods to detect glaucomatous changes.
Results: 16 posterior pole RTA parameters are incorporated to create the PPGI. The results of each patient create a point within 16 dimensional space and the distance between this point and the center of the space is determined as the PPGI. The specificity and sensitivity of the PPGI in discriminating between normal subjects and glaucoma patients is 90% and 75%, accordingly. Utilizing neural network, the specificity is 90% sensitivity 89% and the area under ROC curve is 0.92.
Conclusions: The PPGI was found to have good specificity and sensitivity to differentiate between normal subjects and glaucoma patients based on posterior pole measurements obtained by the RTA. Further improvement was found utilizing neural network.