May 2008
Volume 49, Issue 13
ARVO Annual Meeting Abstract  |   May 2008
Image Averaging in Fundus Autofluorescence Utilizing Fundus Photography
Author Affiliations & Notes
  • M. D. Ober
    Department of Ophthalmology, Henry Ford Health System, West Bloomfield, Michigan
  • J. E. Kim
    Department of Ophthalmology, Medical College of Wisconsin, Milwaukee, Wisconsin
  • T. Trozak
    Department of Ophthalmology, Henry Ford Health System, West Bloomfield, Michigan
  • H. Demirci
    Department of Ophthalmology, Henry Ford Health System, West Bloomfield, Michigan
  • Footnotes
    Commercial Relationships  M.D. Ober, None; J.E. Kim, None; T. Trozak, None; H. Demirci, None.
  • Footnotes
    Support  Supported in part by The CONNECT Network
Investigative Ophthalmology & Visual Science May 2008, Vol.49, 1842. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      M. D. Ober, J. E. Kim, T. Trozak, H. Demirci; Image Averaging in Fundus Autofluorescence Utilizing Fundus Photography. Invest. Ophthalmol. Vis. Sci. 2008;49(13):1842. doi:

      Download citation file:

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

  • Supplements

Purpose: : To determine the effect of image averaging on fundus autofluorescence images taken with a fundus camera.

Methods: : Fundus autofluorescence images from 38 eyes of 23 patients were captured into WinStationTM (Ophthalmic Imaging Systems, Sacramento, CA) using a Topcon EX fundus camera equipped with an excitation filter of 580 nm (bandwidth 500-610 nm) and a barrier filter of 695 nm (bandwidth 675-715 nm). Each eye had multiple images taken of the macula which were then processed using an image averaging algorithm designed by Ophthalmic Imaging Systems to yield a single composite image. The best single original image was chosen by an unmasked retinal specialist (MDO) and altered only to match the contrast and brightness of the composite image for more equivalent comparison. A retinal specialist (JEK) who was masked to the image processing history then compared the 38 sets of 2 images, without knowing which was the composite image and which was the single original image. The grading process included an indication of whether the 2 images in each set were "equivalent" or one was "superior," in terms of image quality. When there was a preferred image, it was rated as "improved" or "greatly improved" compared to the non-preferred image. Subjective grading criteria included resolution, sharpness, and pathology determination to yield an overall impression.

Results: : All composite images (100%) were rated superior or equivalent to the corresponding single images by the masked grader. In 35 of the 38 sets of images (92%), the composite image was judged to be "superior" to a corresponding single image. The remaining 3 sets (8%) were judged to be "equivalent". Thirteen composite images (37%) were rated "greatly improved" while 22 (63%) were rated "improved". A second identical reading performed several hours later by the same masked observer again selected composite images to be "superior" or "equivalent" to single image in all 38 sets. In 36 of 38 sets (95%), there was agreement between first and second reading. The number of images used to create the composite images did not correlate with the degree of improvement; the "greatly improved" category included composite images created from 2 up to 7 original images. A series of images with close registration yielded superior composite images.

Conclusions: : Image averaging can significantly improve the quality of fundus autofluorescence imaging acquired by a fundus camera even when only two images are combined.

Keywords: image processing • imaging/image analysis: clinical 

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.