Published in:
01-03-2020 | Uveitis | Inflammatory Disorders
Intra and inter-rater agreement of inflammatory choroidal neovascular membrane measurements using optical coherence tomography angiography
Authors:
Inês Leal, Shi Zhuan Tan, Tariq Aslam, Laura R Steeples, Nicholas P Jones, Ramandeep Chhabra
Published in:
Graefe's Archive for Clinical and Experimental Ophthalmology
|
Issue 3/2020
Login to get access
Abstract
Purpose
Automated measurement algorithm software is not routinely available in optical coherence tomography angiography (OCTA) devices and manual measurement of choroidal neovascular membrane (CNVM) size is necessary. Our aim was to determine intra- and inter-rater agreement of inflammatory CNVM manual measurements obtained with OCTA.
Methods
OCTA (Triton® Topcon Corporation) images in patients with inflammatory CNVM were imported into ImageJ software v1.50 (NIH image). Two experienced observers performed manual area and perimeter measurements independently, and one of the observers performed the same measurements twice. Agreement was evaluated with intraclass correlation coefficients (ICC) and concordance correlation coefficients (CCC). Bland-Altman plots were plotted to graphically assess concordance. Statistical analysis was performed using STATA v13.0.
Results
Sixteen eyes of 16 subjects, with a mean age of 39.0 ± 16.6 years (range 13–71), were included. Mean CNVM area and perimeter was 124.83 ± 117.80 and 4.20 ± 2.00 mm, respectively. Intra-rater ICC for both area and perimeter measured was 0.99 (95% confidence interval (CI) 0.99–0.99). Inter-rater ICC for area and perimeter measured was 0.95 (95%CI 0.87–0.98) and 0.81 (95%CI 0.17–0.94), respectively. Intra-rater CCC for both area and perimeter measured was 0.99 (95%CI 0.99–0.99). Inter-rater CCC for both area and perimeter measured was 0.91 (95%CI 0.81–0.99) and 0.66 (95%CI 0.44–0.88), respectively.
Conclusions
Inflammatory CNVM manual measurement showed high intra-rater agreement and moderate inter-rater agreement. Repeatability and reproducibility studies are essential in manual analysis to establish thresholds that can distinguish measurements variation from true clinical change. An automatic algorithm may be helpful to accurately grade lesions and monitor disease activity and response to treatment.