Epithelial Thickness Maps In Patients With Epithelial Basement Membrane Dystrophy As Compared To Healthy Volunteers - A Cross-Sectional Study

Previous studies have shown that patients affected by epithelial basement membrane dystrophy (EBMD) have an increased risk of recurrent corneal erosions. Furthermore, EBMD plays an important role for preoperative keratometry measurements in cataract surgery. The purpose of this study are 1) to compare the morphological changes of the epithelium in patients affected by EBMD with healthy controls, 2) to quantify the fluctuation of these changes after 3 months and 3) to develop a machine learning algorithm for the automatic detection of EBMD.

All study-related measures as well as data collection were performed at the Vienna Institute for Research in Ocular Surgery (VIROS), Karl-Landsteiner Institute, Vienna, Austria.

In this prospective cross-sectional study, patients with EBMD and healthy volunteers are measured with two anterior segment OCT systems and the epithelial thickness maps are calculated. For this purpose, the raw data of the measurements are analysed, and the epithelial maps are calculated sectorially in different radii. In order to assess the fluctuation of the morphological constitution, patients with EBMD are measured again after 3 months. Moreover, a machine learning algorithm will be trained to test automatic detection of EBMD.

In a preliminary analysis, 48 eyes of 24 EBMD patients were compared with 27 control eyes. There were significant differences in the central and nasal-inferior and temporal-inferior epithelial sectors (p<0.05), in which the mean epithelial thickness was up to 5 µm higher in EBMD patients, as compared to healthy corneas. A preliminary random forest algorithm achieved a sensitvity of 64% and a specifity of 98% for the detection of EBMD. The complete data as well as the 3-month visit data and the final machine learning algorithm will be presented at the 42^nd Congress of the ESCRS.

The preliminary analysis showed that EBMD significantly affects the inferior epithelial region as compared to healthy corneas. A possible cause for the inferior thickening of the epithelium could be the snowplough mechanism through the eyelids as described in the literature. An automated algorithm would facilitate the detection of EBMD patients and thus patients with an increased risk of inaccurate biometry measurements.