Abstract:
To compare the performance of scanning laser polarimetry with variable corneal compensation (GDx–VCC), confocal scanning laser ophthalmoscopy (HRT II), and optical coherence tomography (Stratus OCT) for detection of early perimetric glaucoma.
One hundred–nine eyes from 109 patients (49 normal controls and 60 patients with open angle glaucoma) were included. The area under ROC curves (AUC) and sensitivity and specificity values were used to compare performance. Likelihood ratios were calculated based on the normative database of each device. Pairwise agreement between the methods was assessed with the kappa statistic.
The average (±SD) visual field MD was –4.1±2.5 dB and 0.08±1.3 dB in the glaucoma and normal groups, respectively. The parameters with greatest AUC (Figure) for GDx–VCC, HRTII, and Stratus OCT were nerve fiber layer index (NFI) (0.938), Mikelberg discriminant function (0.899), and NFL average thickness (0.965). Comparison of AUCs for the best parameter of each device did not show a significant difference (p > 0.05 for all). The greatest sensitivities parameters at 95% specificity were: OCT average thickness (90%), GDx–NFI (82%), and HRT infratemporal cup/disc and rim/disc area ratios (78%). Abnormal test results were associated with very high positive likelihood ratios (>34) for each of the three methods. The best agreement among the techniques was found between NFI (GDx–VCC) and NFL average thickness (OCT),(kappa = 0.71; 95% CI: 0.57–0.85).
ecent versions of quantitative imaging devices perform equally well for the detection of early perimetric glaucoma. However, their sensitivities remain fair at high levels of specificity.
Keywords: imaging/image analysis: clinical