June 2021
Volume 62, Issue 8
Open Access
ARVO Annual Meeting Abstract  |   June 2021
Image segmentation to grade Lid Wiper Epitheliopathy
Author Affiliations & Notes
  • Ayeswarya Ravikumar
    Optometry, University of Houston, Houston, Texas, United States
  • Hope M Queener
    Optometry, University of Houston, Houston, Texas, United States
  • Eric R Ritchey
    Optometry, University of Houston, Houston, Texas, United States
  • Footnotes
    Commercial Relationships   Ayeswarya Ravikumar, None; Hope Queener, None; Eric Ritchey, None
  • Footnotes
    Support  NEI P30-EY007551
Investigative Ophthalmology & Visual Science June 2021, Vol.62, 1256. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Ayeswarya Ravikumar, Hope M Queener, Eric R Ritchey; Image segmentation to grade Lid Wiper Epitheliopathy. Invest. Ophthalmol. Vis. Sci. 2021;62(8):1256.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

Purpose : The lid wiper region is responsible for tear distribution along the ocular surface during the blink. Disruption to the tear film increases the coefficient of friction under boundary lubrication conditions at blink initiation. This irritates the epithelium, resulting in staining termed lid wiper epitheliopathy (LWE). LWE is a critical clinical marker to identify and track abnormal lid/cornea interaction. We propose a semi-automated LWE grading algorithm using hue and value channels of staining images

Methods : 37 images representative of LWE staining, observed after instillation of 2% sodium fluorescein and 1% lissamine green, were analyzed. Using a custom MATLAB program, a region of interest around the lid-wiper region was manually defined. Using the hue and value channels of the image, the algorithm identified stained pixels within this region and fit a curve to these pixels. The number of stained pixels along bisector lines perpendicular to the curve were sampled at regular intervals, defining the height of staining at each bisector. LWE height (mm) was calculated as the average extent of LWE staining along all perpendicular bisectors after subtracting average Marx line height calculated from 11 images without LWE staining. LWE width in mm was determined from the length of the curve scaled by the fraction of the curve that included staining beyond the Marx line.

Results : 31 images (15 upper and 16 lower eyelids: 84%) were successfully analyzed by the algorithm, with 6 images where LWE staining was not successfully detected. Average Marx line height for eyelids without LWE was 0.06 ± 0.02mm. The mean LWE staining height was 0.12 ±0.11mm and 0.12 ± 0.07mm for the upper and lower eye lid respectively. The mean LWE staining width was 10.70 ± 3.84mm and 10.40 ± 3.84mm for the upper and lower eye lid respectively. No significant difference between upper and lower lids were observed LWE height or width (t-test, p>0.05).

Conclusions : This novel automated hue and value algorithm eliminates the potential human error in subjective grading. Automated LWE grading may allow practitioners to observe subtle changes in LWE staining between different CL materials or clinical conditions over time.

This is a 2021 ARVO Annual Meeting abstract.


This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.