Abstract
Purpose :
Qualitative detection of glaucoma deterioration on optic disc photographs is challenging for clinicians. The goal of this study is to present a digital image subtraction technique of the optic nerve head to detect small amounts of structural change in glaucoma.
Methods :
Ninety-two glaucomatous eyes of 65 patients were included; 33 eyes were identified as deteriorating and 59 as stable based on review of stereoscopic disc photographs taken at baseline and last follow-up by 3 masked clinicians. The baseline and follow-up images of each eye were aligned with the i2k Align Retina software (i2k Retina, DualAlign LLC). These images were then converted to grayscale and underwent histogram equalization to enhance contrast and remove the effects of inhomogeneous illumination. The baseline image was subtracted from the follow-up and the difference was shown as a colormap superimposed on a gray-scale image (Fig. 1). Three masked clinicians reviewed the subtraction colormap images in a qualitative manner and provided their opinion of disease stability status based on observation of features of progression according to a graded scale ranging from 1 (no progression) to 5 (definite progression). Sensitivity and specificity of the classification were computed. Fleiss’ kappa statistics were utilized to summarize agreement of categorical gradings and binary classification (0=stable and 1=progressive).
Results :
The median (IQR) time difference between follow-up and baseline images was 4.8 (2.5-7.6) years. With the aid of subtraction maps, glaucoma progression was detected in 26 of the progressive eyes and 9 of stable eyes. Sensitivities of the three clinicians were 0.75, 0.76, and 0.68 and their specificities were 0.89, 0.86, and 0.87. Agreement among clinicians for detection of progression based on density subtraction images was moderate (Fleiss’ kappa=0.42; 95% CI of [.40, .44]) for progression grades and substantial (Fleiss’ kappa=0.74; 95% CI of [.71, .77]) for binary scores.
Conclusions :
Subtraction colormaps offered fair sensitivity for detecting glaucomatous progression and could also establish stability with good specificity. Integration of this method with the EMR could provide the clinician with additional information in real time for detection of glaucoma progression.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.