Abstract
Purpose: :
1) To describe an automated edge-driven image segmentation algorithm to measure the area of atrophic lesions (AL) shown on AF imagery, and 2) to correlate panretinal function with AL area in STGD.
Methods: :
AF images were obtained with a confocal scanning laser ophthalmoscope. Image processing was accomplished automatically using Matlab, ImageJ, ImageMagick, and Edge-flow Image Segmentation linked together with LONI Pipeline. ALs were defined as areas with low intensities based on a two class Gaussian mixture model, and whose perimeter edge strengths were strong. Serial images were obtained from 24 STGD subjects (mean age = 49.12 ± 14.04 years) over a period of up to three years. Structure-function relationships were evaluated with electroretinography (ERG).
Results: :
The mean variation of AL area measurements across a series of AF images collected on the same day with different acquisition settings was 1.1% ± 0.3% (range = 0.6% - 2.8%), suggesting a highly reliable methodology. The average size of the AL(s) in STGD patients at baseline was 22.4 (± 28.7) mm2. AL area was not correlated with age or the underlying ABCA variant. The median growth of the AL over 1.0, 2.0, and 3.0 years was 7.7%, 16.2% and 20.9%, respectively. The rate of change of the AL was uncorrelated with age but was correlated with lesion size at baseline; AL 20 mm2). Rod- and cone- mediated ERG amplitudes were correlated with the total area of the AL; the larger the AL, the poorer the ERG response, with a greater impact of lesion size on the cone responses.
Conclusions: :
A highly reliable, sensitive and automated algorithm for measuring the size of ALs is described. Progression of AL area greater than 2.0% can be considered to represent meaningful change not attributable to normal variation in image processing parameters. This methodology offers a highly sensitive means of monitoring progressive change in retinal structure in patients with well-defined ALs that will be useful in clinical trials to quantify phenotypic changes in response to therapy.
Keywords: image processing • degenerations/dystrophies • anterior segment