No abstract

AimsTo quantify the change in drusen volume over time and identify its prognostic value for individual risk assessment.MethodsA prospective observational study over a minimum of 3 years and maximum of 5 years and follow-up examination every 3 months was conducted at the ophthalmology department of the Medical University of Vienna. 109 patients presenting early and intermediate age-related macular degeneration (AMD) were included, of which 30 patients concluded a regular follow-up for at least 3 years. 50 eyes of 30 patients were imaged every 3 months using spectral-domain and polarisation-sensitive optical coherence tomography (OCT). Drusen volume was measured using an automated algorithm. Data of a 6-month follow-up were segmented manually by expert graders.ResultsGradings from 24 000 individual B-scans showed solid correlation between manual and automated segmentation with an initial mean drusen volume of 0.17 mm3. The increase in drusen volume was shown to be comparable among all eyes, and a model for long-term drusen volume development could be fitted as a cubic polynomial function and an R2=0.955. Spontaneous drusen regression was observed in 22 of 50 eyes. In this group, four eyes developed choroidal neovascularisation and three geographic atrophy.ConclusionsDrusen volume increase over time can be described by a cubic function. Spontaneous regression appears to precede conversion to advanced AMD. OCT might be a promising tool for predicting the individual risk of progression of AMD.

The predictive model proposed in this study represents a promising step toward image-guided prediction of AMD progression. Machine learning is expected to accelerate and contribute to the development of new therapeutics that delay the progression of AMD.

Modern optical coherence tomography (OCT) devices used in ophthalmology acquire steadily increasing amounts of imaging data. Thus, reliable automated quantitative analysis of OCT images is considered to be of utmost importance. Current automated retinal OCT layer segmentation methods work reliably on healthy or mildly diseased retinas, but struggle with the complex interaction of the layers with fluid accumulations in macular edema. In this work, we present a fully automated 3D method which is able to segment all the retinal layers and fluid-filled regions simultaneously, exploiting their mutual interaction to improve the overall segmentation results. The machine learning based method combines unsupervised feature representation and heterogeneous spatial context with a graph-theoretic surface segmentation. The method was extensively evaluated on manual annotations of 20,000 OCT B-scans from 100 scans of patients and on a publicly available data set consisting of 110 annotated B-scans from 10 patients, all with severe macular edema, yielding an overall mean Dice coefficient of 0.76 and 0.78, respectively.

Mapping retinal sensitivity to distinct retinal pathologies revealed outer retinal layers, in addition to the RPE, as significant for sensitivity loss. Therefore in GA the RPE loss and the alteration of outer retinal layers should be analysed, which could also provide insight into lesion progression.