Diagnostic accuracy of scanning laser polarimetry with enhanced versus variable corneal compensation.

PURPOSE: To compare the diagnostic accuracy of scanning laser polarimetry (SLP) parameters between images taken with enhanced corneal compensation (ECC) and those with variable corneal compensation (VCC) and to explore the effect of atypical birefringence patterns on this accuracy. DESIGN: Cross-sectional observational study. PARTICIPANTS: Forty-one healthy subjects and 92 patients with primary open-angle glaucoma. METHODS: Variable corneal compensation and ECC images were obtained of 1 eye per subject, selected randomly if both eyes were eligible. For both ECC and VCC, the areas under the receiver operating characteristic curves (AUROCs) and the sensitivity at a specificity of > or =95% were calculated per parameter in all eyes. The analyses were reperformed separately in eyes with and without atypical birefringence patterns (ABP) images. MAIN OUTCOME MEASURES: The AUROCs and sensitivities at a specificity of > or =95% for various SLP parameters in all eyes and in eyes without ABP images. RESULTS: The diagnostic accuracy for most standard parameters (temporal-superior-nasal-inferior-temporal [TSNIT] average, superior average, inferior average, and TSNIT standard deviation) in all eyes was statistically significantly higher with ECC than with VCC, except for the nerve fiber indicator (NFI). When only eyes without ABP were used for the analysis, the diagnostic accuracy of SLP parameters with VCC improved, and the differences in diagnostic accuracy between ECC and VCC for these parameters lost their statistical significance. CONCLUSIONS: Standard SLP parameters (except for the NFI) generally had a higher diagnostic accuracy when eyes were imaged with ECC than with VCC because there were fewer ABP images with ECC than with VCC. Enhanced corneal compensation therefore may be more reliable than VCC for the detection of glaucoma. A future automated classifier, similar to the current NFI, may perform better if it is trained on data obtained with ECC. Clinically, retinal nerve fiber layer images with marked ABP, acquired with either ECC or VCC, should be viewed with caution.