April 2010
Volume 51, Issue 13
Free
ARVO Annual Meeting Abstract  |   April 2010
Performance of a Content-Adaptive Filtering Method for Photoreceptor Cell Counting
Author Affiliations & Notes
  • F. Mohammad
    Electrical and Computer Engineering,
    University of Illinois, Chicago, Chicago, Illinois
  • J. M. Wanek
    Ophthalmology and Visual Sciences,
    University of Illinois, Chicago, Chicago, Illinois
  • R. Ansari
    Electrical and Computer Engineering,
    University of Illinois, Chicago, Chicago, Illinois
  • M. Shahidi
    Ophthalmology and Visual Sciences,
    University of Illinois, Chicago, Chicago, Illinois
  • Footnotes
    Commercial Relationships  F. Mohammad, None; J.M. Wanek, None; R. Ansari, None; M. Shahidi, None.
  • Footnotes
    Support  NIH, Dept of VA, RPB
Investigative Ophthalmology & Visual Science April 2010, Vol.51, 2314. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      F. Mohammad, J. M. Wanek, R. Ansari, M. Shahidi; Performance of a Content-Adaptive Filtering Method for Photoreceptor Cell Counting. Invest. Ophthalmol. Vis. Sci. 2010;51(13):2314.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose: : Measurement of photoreceptor cell density may potentially become an important parameter for screening and diagnosis of diseases that affect human vision. A method was developed to automatically count the number of photoreceptor cells in adaptive optics (AO) retinal images by using content-adaptive filtering (CAF). The performance of the CAF method was evaluated using simulated images with various photoreceptor cell densities and blur levels.

Methods: : AO retinal imaging was performed in a visually normal subject. From an original image of the photoreceptor cell mosaic, images with various photoreceptor cell densities were simulated by scaling the original image by factors of 1.25, 1.5, 1.75, and 2. Blurred images were generated by applying Gaussian filters with σ = 2 and 3 pixels to the original and scaled images. The CAF method consisted of a customized bandpass filter using the McClellan transformation, a threshold-based image binarization process and cell count estimation. The error of the CAF method was determined by comparing the number of photoreceptor cells counted by the CAF method to those manually counted by 2 independent observers.

Results: : The variation in the number of photoreceptor cells manually counted on 15 images by 2 independent observers was on average 7 %. As anticipated, photoreceptor cells counted both by the automated CAF method and manually decreased as the photoreceptor cell spacing or blur level increased. Errors of the CAF method for the scaled images were 12%, 13%, 3%, 21%, and 18% for scaling factors of 1, 1.25, 1.5, 1.75, and 2, respectively. Errors of the CAF method were 12%, 7%, and 1% for original, σ = 2 blurred, and σ = 3 blurred images, respectively. For the original image scaled by a factor of 2, the errors of the CAF method were 18%, 5%, and 3% for scaled, σ = 2 blurred, and σ = 3 blurred images, respectively.

Conclusions: : Photoreceptor cell counts derived by the CAF method were in agreement with manual counting and the agreement improved with blur, which may be attributed to reduced image noise. This cell counting method may be useful for estimating changes in photoreceptor cell density due to retinal diseases that degrade image quality.

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

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.

×