Purchase this article with an account.
Donald Hood, C Gustavo De Moraes, Lola Grillo, Paula Alhadeff, Ravivarn Jarukasetphon, Rithambara Ramachandran, Diane Wang, Dana Blumberg, Jeffrey M Liebmann, Robert Ritch; Improving inter-individual diagnostic agreement in early glaucoma through better use of optical coherence tomography (OCT) scans.. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):2060.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To test if individuals trained in a method integrating key visual field (VF) and optical coherence tomography (OCT) information without stereo disc photographs (SDP) would show better inter-individual agreement in diagnosing early glaucoma than glaucoma specialists using traditional commercial reports and SDP.
Two glaucoma fellows and two students were trained with a method  for interpreting spectral domain OCT scans and for comparing them to 24-2 VFs. The training involved learning to use a single-page report  containing inner retinal layer thickness information from OCT macular and disc cube scans, along with superimposed 24-2 VF (total deviation) and OCT probability plots (Fig. 1). They were then given reports for 50 eyes (50 patients) with 24-2 MDs better than -6 dB and suspicious discs based upon SDP. For each eye, they judged whether it had glaucomatous damage. The fellows were then given the commercial 24-2 VF reports and SDP, and could change their judgment. Inter-individual agreement (kappa statistic, k) was compared to that of 3 glaucoma specialists who judged these eyes based upon SDP and commercially available 24-2 VF and OCT reports. A “reference standard” based upon the majority opinion of 3 glaucoma experts plus the referring specialist, consisted of 32 glaucomatous and 16 non-glaucomatous eyes (there was insufficient agreement for 2 eyes).
With the report alone, the agreement among the 4 graders (mean paired k of 0.93, range of 0.87 to 0.96) was substantially better than that among the 3 specialists (mean paired k of 0.53, range of 0.36 to 0.69) who used traditional diagnostic methodologies. The estimated “sensitivity” and “specificity” were also good. Of the 32 eyes judged as abnormal by the specialists, only one was judged as normal by the 4 graders, and of the 16 eyes judged as normal by the specialists, between 0 and 2 were judged as abnormal by the graders. The students and fellows did equally well. Adding the full VFs and stereo-photographs only improved the agreement by one eye for one grader.
With training to read a report that integrates OCT and VF information, inter-individual agreement and diagnostic performance can be improved without the need for fundus photographs. 1. Hood, Raza (2014) BJO. 2. Hood, Raza, De Moraes et al. (2014) TVST.
This PDF is available to Subscribers Only