May 2005
Volume 46, Issue 13
Free
ARVO Annual Meeting Abstract  |   May 2005
An Automatic Image Segmentation Software Can Precisely Quantitate the Severity of Corneal Disease Digitally Captured From Viral Infected Eyes
Author Affiliations & Notes
  • Y.–Y. Shieh
    Radiological Sciences,
    University of California, Irvine MC, Orange, CA
  • G. Perng
    Ophthalmology,
    University of California, Irvine MC, Orange, CA
  • Footnotes
    Commercial Relationships  Y. Shieh, None; G. Perng, None.
  • Footnotes
    Support  None.
Investigative Ophthalmology & Visual Science May 2005, Vol.46, 2731. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Y.–Y. Shieh, G. Perng; An Automatic Image Segmentation Software Can Precisely Quantitate the Severity of Corneal Disease Digitally Captured From Viral Infected Eyes . Invest. Ophthalmol. Vis. Sci. 2005;46(13):2731.

      Download citation file:


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

      ×
  • Supplements
Abstract

Abstract: : Purpose: Ocular imagines are frequently captured from HSV infected animals digitally and the severity of corneal disease is estimated by human subjective rather than objective measurement. The situation is further complicated by the variations in object orientation, the camera distance, and the zoom factors on different images. As a result, it is difficult to compare results generated by different researchers. We propose an automatic image segmentation software to reduce the human variation in estimation by the same researcher on different days as well as that among different researchers. Methods: The software is designed to scan the entire ocular surface, stained with fluorescein, including and up to the inner border of sclera. The percentage of ocular surface disease is calculated by dividing affected area over entire ocular surface. The proposed automatic image segmentation software requires the user to input: (1) the minimal rectangle that just confines the pupil region, (2) the minimal rectangle that confines the Iris region, and (3) a rectangle that contains all keratitis pixels. The first rectangle enables the program to determine the center of the eye. The Iris–sclera boundary can be determined by locating maximum contrast pixels starting from a point on the second rectangle toward the center of the eye. Finally, keratitis pixels can be separated from the normal pixels by applying simple thresholding to the green–component intensity of the pixels within the third rectangle. The algorithm has been programmed by using the Matlab matrix–oriented programming language. A set of twenty images taken from rabbit eyes infected with herpes simplex virus are used to evaluate the performance of the program. Results: Our results demonstrate the feasibility of calculating the percentage of corneal scarring by automatic image segmentation and subsequent thresholding with a small amount of initial inputs from the user. The computer processed results closely agrees with human subjective interpretation. Furthermore, it enables the user to fine–tune the discriminating threshold to satisfy the particular sensitivity criterion that he desires. Conclusions: The computer classification based upon numerical analysis provides a consistent way of separating pixels of small differences in intensity that are difficult for human eyes to discern. Furthermore, the flexibility of computer software enables the user to fine–tune the discriminating threshold to satisfy the particular sensitivity criterion that he/she desires.

Keywords: imaging/image analysis: non-clinical • image processing • cornea: epithelium 
×
×

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.

×