Abstract
Purpose:
To compare segmentation and decentration errors of SDOCT volume scans in eyes with neovascular age-related macula degeneration (NVAMD) and to analyze the impact of visual acuity, CNV lesion type and SDOCT parameters.
Methods:
SDOCTs (Spectralis, Heidelberg Engineering) of 64 eyes of 62 patients with NVAMD were retrospectively collected. Foveal central subfield (FCS) and foveal center point (FCP) thickness measurements (ILM to Bruch's membrane) were calculated before and after manual correction of segmentation errors and positioning of the ETDRS grid using computer-assisted manual grading software. Statistical analysis was performed using the Wilcoxon signed-rank test.
Results:
The mean absolute difference between FCP (FCS) thickness values before and after manual correction of segmentation errors was 63±97µm (58±87µm), which increased to 73±110µm (65±100µm) after additional manual centration of the ETDRS grid on the fovea. Absolute FCP (FCS) difference >50µm was seen in 36% (31%) of cases after correction of segmentation errors, and in 42% (39%) after additional correction of the grid position. Eyes with greater retinal thickness showed larger differences after manual segmentation, eyes with cystoid spaces or absence of SRF showed larger differences after manual centration. There was no impact of CNV lesion type, presence of PED, or visual acuity.
Conclusions:
Manual grading of SDOCT volume scans improves the quality of thickness measurements not only by correcting segmentation errors but also by adjustment of the grid position. Segmentation error was higher in eyes with greater retinal thickness values. Decentration error was higher in eyes with cystoid spaces or absence of SRF.
Keywords: 552 imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) •
412 age-related macular degeneration