May 2007
Volume 48, Issue 13
ARVO Annual Meeting Abstract  |   May 2007
Digital Resolution Required for Effective Grading of Retinal Images
Author Affiliations & Notes
  • V. Silvestri
    Ophthalmology, Royal Hospital Trust, Belfast, United Kingdom
  • M. Williams
    Ophthalmology, Royal Hospital Trust, Belfast, United Kingdom
  • G. Silvestri
    Ophthalmology, Queen's University Belfast, Belfast, United Kingdom
  • Footnotes
    Commercial Relationships V. Silvestri, None; M. Williams, None; G. Silvestri, None.
  • Footnotes
    Support HPSSNI R&D
Investigative Ophthalmology & Visual Science May 2007, Vol.48, 3862. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      V. Silvestri, M. Williams, G. Silvestri; Digital Resolution Required for Effective Grading of Retinal Images. Invest. Ophthalmol. Vis. Sci. 2007;48(13):3862.

      Download citation file:

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

  • Supplements

Purpose:: Digital retinal imaging is now in common usage in clinical ophthalmology with newer cameras having increasingly large file sizes for image capture. This study seeks to assess the minimal image resolution (file size) required for digital colour fundus imaging to allow consistent and accurate grading of patients with age-related macular degeneration (AMD).

Methods:: Stereo fundus photographs were taken on a Canon CR-Dgi Non-Mydriatic fundus camera (fitted with a Canon 20D digital camera back) through a dilated pupil from 30 random patients (both eyes) with a diagnosis of AMD. Three stereo pairs of images were taken from each eye of each patient of field F2 (macula centred). Pair 1 was captured at the camera's maximum resolution: Large Fine (3504x2336 pixels), pair 2 at Canon's recommended setting of Medium Normal (2544x1696 pixels) and pair 3 at the camera's lowest resolution Small Normal (1728x1152 pixels). The 3 sets of images, large, medium and small from each patient were anonymized, saved in a coded fashion and in random order, were graded on separate days by 2 retinal specialists according to the Rotterdam AMD grading scale.

Results:: The graders were given a score of 1 for each eye when the same grade was given for all 3 capture resolutions and 0 when an eye was graded differently. The results from a subset, 10 patients (20 eyes) are as follows: Grader 1 GS scored 18 out of 20 and grader 2 MW scored 19 out of 20. The results of the pilot data show very good concordance and indicate that there is no advantage in capturing colour fundus images with the Canon CR-Dgi Non-Mydriatic fundus camera at the highest quality setting available in the Canon 20D. Transfer time of the image from the moment of capture until the file appears on screen takes 10 seconds compared to 5 seconds for the recommended setting Medium Normal and 4 seconds for the Small Normal setting. Files sizes are also much larger on average 1.8mb as apposed to 400kb for the recommended setting.

Conclusions:: The results of this study indicate that capture of retinal images using the Small Normal setting allows adequate image resolution for routine clinical retinal imaging even for lesions such as small drusen. The benefits of using the Small Normal settings were fast transfer speeds and small files sizes, approx 200kb which would ultimately save on storage space. Almost 10 images captured at the Small Normal setting could be stored in place of 1 image captured at the Large Fine setting.

Keywords: imaging/image analysis: clinical • age-related macular degeneration • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) 

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.