Purchase this article with an account.
Hrvoje Bogunovic, Michael Abramoff, Elliott Sohn, Milan Sonka; Predicting the Visual Acuity of Patients with Diabetic Macular Edema from the Measured Disruption Status of External Limiting Membrane on the SD-OCT Scan. Invest. Ophthalmol. Vis. Sci. 2013;54(15):3602.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Lower visual acuity (VA) in patients suffering from Diabetic Macular Edema (DME) is known to be associated with the disruption status of the External Limiting Membrane (ELM) in the retina. Recently our group reported a method, which for the first time performs automated quantification of the ELM disruption from a spectral-domain optical coherence tomography (SD-OCT) scan. In this work, we evaluate the potential of using such measured ELM disruption status as a predictor of patient’s VA.
Macula-centered SD-OCT volumes (Heidelberg Spectralis, 512×19×496 voxels, 10.98×239.66×3.87 µm3) were acquired in 12 eyes of 12 patients diagnosed with clinically significant DME. Using the automated method, the level of disruption of patients’ ELM was estimated from SD-OCT volumes. The method detects the ELM surface and - based on the textural and morphologic properties in the vicinity of the ELM surface - estimates a probability of being locally disrupted. To obtain a measure of the disruption status, a region of interest (ROI: 30x30 voxels) corresponding to the area underneath the fovea was placed at the center of isotropically resampled ELM surface, and the mean disruption probability of voxels forming the ROI was computed. For the 12 patients, the obtained mean values were correlated with visual acuity measured at the time of the scan.
Plotting the relationship between the computed disruption status and VA revealed the generally expected trend for which more severely disrupted ELM are associated with higher VA loss. Correlation between the VA and the corresponding estimated ELM disruption status was of medium strength exhibiting Pearson’s linear correlation coefficient of 0.49 (p-value 0.11, Fig. 1).
The measured ELM disruption status has a potential to be used as a predictor of VA for DME patients, hence has a potential to serve as an imaging biomarker for patients undergoing treatment for DME.
This PDF is available to Subscribers Only