Purpose: To compare different quantification tools based on confocal scanning laser ophthalmoscopy for assessment of retinal pigment epithelium (RPE) tear area size.
Methods: Confocal scanning laser ophthalmoscopy fundus autofluorescence (FAF) and near-infrared reflectance (IR) images were retrospectively evaluated in 23 patients with RPE tear after intravitreal injection for pigment epithelium detachment due to exudative age-related macular degeneration at baseline and additionally in 11 patients after 5.1 ± 1.8 months of follow-up. Retinal pigment epithelium tear area was measured by three independent readers using three methods: manually on confocal scanning laser ophthalmoscopy FAF images, manually on confocal scanning laser ophthalmoscopy IR images, and using an FAF-based semiautomated software.
Results: Confidence intervals were 0.08 and 0.12 for FAF, 0.11 and 0.09 for FAF-based semiautomated software, and 0.25 and 0.27 for IR for intraobserver (Reader 1) and interobserver agreements (Readers 1 and 2), respectively. The average values of the square errors of the quantification methods were 0.040 ± 0.033 mm (FAF), 0.035 ± 0.060 mm (software), and 0.187 ± 0.219 mm (IR). Mean area of RPE tears at baseline given as the average measurement of all 3 readers using FAF-based semiautomated software was 5.77 ± 4.62 mm (range, 0.13-14.74 mm). Follow-up measurements of unilobular RPE tears (8 patients) showed no change in lesion area size (0.14 ± 0.33 mm); in contrast, multilobular RPE tears (3 patients) showed a progression in lesion area size of 1.80 ± 0.74 mm.
Conclusion: Manual FAF-based and semiautomated FAF-based quantifications of RPE tear area are accurate and reproducible and superior to manual IR-based measurement. Retinal pigment epithelium tear area quantification is clinically relevant regarding further intravitreal treatment, particularly in multilobular RPE tears.