June 2013
Volume 54, Issue 15
ARVO Annual Meeting Abstract  |   June 2013
Automated Detection and Enumeration of Corneal Superficial Punctate Keratitis
Author Affiliations & Notes
  • John Rodriguez
    R & D, Ora, Inc., Andover, MA
  • Patrick Johnston
    R & D, Ora, Inc., Andover, MA
  • Keith Lane
    R & D, Ora, Inc., Andover, MA
  • George Ousler
    Ora, Inc., Andover, MA
  • Footnotes
    Commercial Relationships John Rodriguez, Ora, Inc. (E); Patrick Johnston, Ora, Inc (E); Keith Lane, Ora, Inc. (E); George Ousler, Ora, Inc. (E)
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2013, Vol.54, 4341. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      John Rodriguez, Patrick Johnston, Keith Lane, George Ousler; Automated Detection and Enumeration of Corneal Superficial Punctate Keratitis. Invest. Ophthalmol. Vis. Sci. 2013;54(15):4341.

      Download citation file:

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

  • Supplements

Purpose: The evaluation of corneal superficial punctate keratitis (SPK) is an important primary endpoint for dry eye clinical trials. However, staining grading remains a challenging task, as graders must manually divide corneal regions by eye. Moreover, the required enumeration of SPK is necessarily constrained due to practical limits and can be visually difficult to resolve with sufficient sensitivity. The use of software to automate and standardize this process allows for actual enumeration of SPK to be made quickly and accurately. Added sensitivity when compared to traditional clinician based scales may be realized by implementing a continuous staining score based on the distribution of the SPK data. To test the effectiveness of this method, we investigate SPK in a dry eye population.

Methods: A sample population of 665 dry eye subjects was considered. Inferior and central corneal fluorescein images were obtained for each eye using high resolution digital photography. The images were processed using a custom software program developed using open source computer vision libraries. The software automatically selects the relevant corneal area of interest and enumerates SPK lesions and their cumulative surface area on the cornea. Finally, the original images were visually verified with images output by the program with highlights of identified SPK.

Results: The mean number of SPK detected in the inferior corneal region was 33.73 and 17.63 in the central region. SPK areas (as a percent of surface area) for the inferior and central corneal regions were 0.236% (SD:50.04) and 0.182% (SD:37.34), respectively. Visual confirmation showed close qualitative agreement with software results. Processing of all 2660 images required approximately 60 minutes. Computed log SPK number of a randomly selected subset of 54 images showed a 92% correlation with a clinical grader assessment using the Ora staining scale (0-4).

Conclusions: The results show that extensive and accurate information may be efficiently obtained with the software based approach. We note that mean SPK numbers and the stained area are higher in the inferior then the central corneal region, although the respective geometric areas are the same. The automated selection of the relevant region of interest and enumeration of up to several hundred SPK quickly and efficiently allows for precise and accurate quantification of corneal desiccation.

Keywords: 486 cornea: tears/tear film/dry eye • 479 cornea: clinical science  

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.