Abstract
Purpose: :
To investigate the impact of glaucomatous damage and optic disc size on different Heidelberg Retina Tomography (HRT) classifications for discriminating glaucoma and healthy controls.
Methods: :
Two-hundred seventy three glaucoma patients and 276 normal healthy controls were recruited from the Erlangen Glaucoma Registry for this study. Standard static white on white perimetry, 24-h IOP profile, stereographic optic disc slides and Scanning Laser Tomography (HRT II, Software Version 3.0; Heidelberg Engineering, Heidelberg, Germany) were performed in all patients and controls. The effect of glaucoma stage and disc size on the diagnostic accuracy was evaluated using receiver operating characteristic (ROC) curves. Results of the Glaucoma Probability Score (GPS) as a contour line independent procedure (relevance vector machine) was compared to other recently described HRT classifications including Moorfields regression analysis (MRA) and classifications developed by Bathija, Iester, Mikelberg and Mardin.
Results: :
Areas under the ROC curves for discrimination between glaucoma and healthy eyes of the overall classification by GPS, MRA, Bathija, Iester, Mikelberg and Mardin (95% CI) increased from 0.71 (0.66-0.77), 0.65 (0.60-0.71), 0.70 (0.64-0.75), 0.67 (0.62-0.73), 0.65 (0.59-0.71) and 0.63 (0.57-0.68) for early glaucoma(Jonas: Stage I) to 0.96 (0.94-0.99), 0.95 (0.93-0.98), 0.94 (0.88-1.01), 0.94 (0.87-1.01), 0.92 (0.86-0.98) and 0.88 (0.83-0.92) for advanced glaucomatous damage (Jonas: Stage IV).Best performance of discrimination between glaucoma and healthy eyes using area under the ROC curve were observed for medium sized discs (disc area 2.1-2.49 mm²) and worst performance for large optic discs (disc area > 3.1 mm²). GPS classification was superior to other investigated classifications in most of the investigated subgroups.
Conclusions: :
The diagnostic accuracy of all classifications depends on the optic disc size and the glaucoma stage. The diagnostic performance of the GPS was better to that of the MRA and similar to the classification of Bathija.
Clinical Trial: :
www.clinicaltrials.gov NCT00494923
Keywords: optic disc • imaging/image analysis: clinical • topography