June 2022
Volume 63, Issue 7
Open Access
ARVO Annual Meeting Abstract  |   June 2022
Automatic detection of cone photoreceptors in the fovea using confocal AOSLO imaging
Author Affiliations & Notes
  • Mikhail Tsaritsyn
    Hopital ophtalmique Jules-Gonin, Lausanne, Vaud, Switzerland
  • Alain Jacot-Guillarmod
    Hopital ophtalmique Jules-Gonin, Lausanne, Vaud, Switzerland
  • Jelena Potic
    Hopital ophtalmique Jules-Gonin, Lausanne, Vaud, Switzerland
  • Thomas Wolfensberger
    Hopital ophtalmique Jules-Gonin, Lausanne, Vaud, Switzerland
  • Ciara Bergin
    Hopital ophtalmique Jules-Gonin, Lausanne, Vaud, Switzerland
  • Mattia Tomasoni
    Hopital ophtalmique Jules-Gonin, Lausanne, Vaud, Switzerland
  • Adam M Dubis
    Moorfields Eye Hospital NHS Foundation Trust, London, London, United Kingdom
    University College London Institute of Ophthalmology, London, London, United Kingdom
  • Footnotes
    Commercial Relationships   Mikhail Tsaritsyn None; Alain Jacot-Guillarmod None; Jelena Potic None; Thomas Wolfensberger None; Ciara Bergin None; Mattia Tomasoni None; Adam Dubis None
  • Footnotes
    Support  None
Investigative Ophthalmology & Visual Science June 2022, Vol.63, 4554 – F0468. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Mikhail Tsaritsyn, Alain Jacot-Guillarmod, Jelena Potic, Thomas Wolfensberger, Ciara Bergin, Mattia Tomasoni, Adam M Dubis; Automatic detection of cone photoreceptors in the fovea using confocal AOSLO imaging. Invest. Ophthalmol. Vis. Sci. 2022;63(7):4554 – F0468.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : The Adaptive Optics Scanning Laser Ophthalmoscope (AOSLO) can be used to image photoreceptors, in vivo. Recently, several algorithms for cone characterisation have exploited the specificity of split dector to robustly image cone inner segements. The high packing density of cones within the fovea, impedes split dector efficacity[1]. We present a novel algorithm for automated cone detector in confocal images: the Automatic Adaptive Thresholding and Maxima Search (ATMS).
[1] (Morgan, J. I. W., Vergilio, G. K., Hsu, J., Dubra, A., & Cooper, R. F. (2018). The reliability of cone density measurements in the presence of rods. Translational Vision Science and Technology, 7(3), 1–11. https://doi.org/10.1167/tvst.7.3.21).

Methods : ATMS first detects darker areas by adaptive local filtering. Those correspond to blood vessels where cones cannot be reliably counted, so they are excluded from further analysis. The image is then split into Regions Of Interest (ROIs) and contrast enhancement is performed separately in each ROI, normalizing by brightness and subtracting its the second derivative. This enhances the bright spots corresponding to cones. Based on the local distribution of luminosity in each ROI, the image was then segmented using adaprive thresholding, and the cones are identified via a local maxima search. Finally, the output was refined by adjusting the threshold probability based on Veronoi properties of the cell mosaic. From this output, cell densities were calculated.

Results : Results were validated on 53 images from 10 subjects. Each image was annotated by 2 graders: the overall mean Dice’s coefficient was 0.96 (Precision = 0.97, Recall = 0.95).

Conclusions : This paper presented a novel, fully automated labelling method which is suitable for analysis of central foveal regions (0° to 3°) in confocal images. Our method was integrated with existing tools that operate in more peripherical regions (4° to 10°), thus building a cell density estimation pipeline which works across the full range of retinal eccentricities.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.

×
×

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.

×