Abstract
Purpose: :
To establish criteria for optic disc progression that best discriminate between healthy and glaucomatous optic discs followed over time.
Methods: :
Heidelberg Retina Tomograph (HRT) images of the optic disc were obtained from 60 healthy controls (60 eyes, median follow–up 8.2 yrs, 16 exams) and 91 glaucoma patients (172 eyes, median follow–up 9.3 yrs, 18 exams). The data were analyzed with the Topographical Change Analysis of the Heidelberg Eye Explorer software (v. 3.0.2.0, incorporating a new alignment algorithm). The size of the largest cluster of superpixels within the optic disc area with statistically significant change from baseline was derived for height changes of ≥0, 20, 50, 100, and 200 µm, and Kaplan–Meier analyses were performed for cluster sizes of ≥1, 2, 5, 7.5 and 10% of disc area. The log–rank test was used as a rate–independent measure of the separation between paired event rates. We compared event rates between controls and glaucoma patients with red (negative height change) and green clusters (positive height change), as well as between red and green clusters within each of the two groups of subjects.
Results: :
The largest separation between healthy and glaucomatous subjects (Χ2 between 8.4 and 21) was observed with red clusters of small size (between 1% and 2% of disc area) and moderate height changes (20 to 50µm). Larger cluster sizes reduced both the event rates as well as the separation between healthy and glaucomatous subjects. Positive height changes (green clusters) also occurred more frequently in the glaucoma patients but the differences between the subject groups were smaller (Χ2 between 0 to 4.1). Greater incidence of red compared to green clusters occurred across all criteria in the glaucoma patients (Χ2 between 3.1 and 35) but was observed only with the less conservative criteria in the controls (Χ2 between 0 and 4.6).
Conclusions: :
Glaucoma progression, physiological ageing, disc remodeling, and noise all contribute to topographical height changes (both positive and negative) observed in the optic discs of patients followed over time. The best statistical discrimination between healthy and glaucomatous subjects occurred with small cluster sizes (1–2%). Height change criteria (between 20 to 50 µm) decreased event rates but did not reduce the separation between healthy and glaucomatous subjects. Criteria incorporating both size and height change improve the detection of glaucomatous optic disc progression with the HRT.
Keywords: optic disc • imaging/image analysis: clinical • clinical (human) or epidemiologic studies: systems/equipment/techniques