Sector-based analysis with the Heidelberg retinal tomograph 3 across disc sizes and glaucoma stages : a multicenter study

PURPOSE: To investigate the ability of sectorial analysis using the Heidelberg Retinal Tomograph 3 (HRT3) to discriminate between healthy and glaucomatous eyes and to determine whether this is affected by disc size and glaucoma severity. DESIGN: Multicenter, cross-sectional evaluation of diagnostic tests. PARTICIPANTS: Two hundred thirty-three eyes from 137 normal subjects and 96 glaucoma patients classified by the presence of a repeatable visual field defect. TESTING: Participants underwent imaging with the HRT3, and the diagnostic accuracy of stereometric parameters, Moorfields regression analysis (MRA), and glaucoma probability score (GPS) were analyzed sectorially by glaucoma stage and optic disc size. MAIN OUTCOME MEASURES: Sensitivity, specificity, positive and negative predictive values, and the area under the receiver operating characteristics curve (AUC). RESULTS: Of stereometric parameters, the cup-to-disc area ratio of the inferotemporal sector had the largest AUC (AUC, 0.74). Both MRA and GPS algorithms showed the best diagnostic accuracy in the inferotemporal sector, with a sensitivity and specificity of 63% and 88% for MRA and 80% and 62% for GPS, respectively. In small discs, sectorial MRA analysis had higher diagnostic accuracy than the global optic nerve head (ONH) analysis (sensitivity of 70% and specificity of 82% in the inferonasal sector), and these findings were confirmed in very large discs (sensitivity of 74% and specificity of 85% in the inferotemporal sector). Similarly, stereometric parameters discriminated better sectorially rather than globally, with different parameters giving the best results in different optic disc size subgroups. MRA sensitivity was weak in the early-glaucoma stage, with slightly higher figures if considered sectorially rather than globally. GPS diagnostic accuracy was very consistent across ONH sectors in each disc size and glaucoma stage subgroup, with no single sector demonstrating better diagnostic accuracy than the global analysis. CONCLUSIONS: Heidelberg Retinal Tomograph 3 sectorial analysis showed moderate diagnostic performance and may offer potential advantages over global analysis in the clinical diagnostic process. Small discs are classified more accurately by examining the inferonasal sector, whereas larger discs are classified more accurately by examining the inferotemporal sector. Neither HRT parameters nor classification algorithms seem to be good at the earlier stage of the disease