Abstract
Purpose:
To determine the agreement between clinical versus automated Disc Damage Likelihood Scale (DDLS) staging and if it was affected by disc size.
Methods:
In this prospective observational study, consecutive subjects who were diagnosed to have glaucoma or ocular hypertension were enrolled. DDLS staging was done clinically by stereo-biomicroscopy after measuring optic disc size and rim/disc ratio. Using appropriate corrective factorsfor measuring disc size, the discs werecategorized into small, average and large discs.The Kowa Nonmyd WX retinal camera (Kowa Optimed, Nagoya, Japan) was used for stereoscopic fundus imaging; a single examiner then stereoscopically observed and marked the outline of the optic disc and cup; the built-in software then calculated the DDLS staging. Agreement between the 2 methods and for evaluating effect of disc size was assessed using weighted kappa (categorical data) and reliability was measured usingintraclass correlation coefficient (ICC). We also assessed intra-observer reproducibility of the automated staging in 30 eyes.
Results:
68 eyes (36 patients) were enrolled; the mean age was 49.8 ± 16.9 years and majority were men (64%). The two methods showed good agreement (κ=0.81, 95% C.I, 0.74-0.88) and repeatability (r=0.97, 95% C.I, 0.96-0.98). Moderate sized discs showed very good agreement (r=0.85, 95% C.I, 0.77-0.93), while large discs showed moderate agreement (r=0.69, 95% C.I, 0.53-0.83). Intra-observer reproducibility was very good for the automated staging (r=0.97, 95% C.I, 0.97-0.99).
Conclusions:
Automated DDLS staging showed good agreement with clinical staging and could be used as an alternative tool in diagnosis; disc size did not appear to be a major determinant.
Keywords: 550 imaging/image analysis: clinical •
627 optic disc •
549 image processing