June 2023
Volume 64, Issue 8
Open Access
ARVO Annual Meeting Abstract  |   June 2023
Automated quantification of retinal degeneration in FAF images
Author Affiliations & Notes
  • Dimitrios Pollalis
    Ophthalmology, USC Roski Eye Institute, University of Southern California, Los Angeles, California, United States
    USC Ginsburg Institute for Biomedical Therapeutics, University of Southern California, Los Angeles, California, United States
  • Timothy Lin
    University of Southern California Viterbi School of Engineering, Los Angeles, California, United States
  • Peter Chong
    University of Southern California Dana and David Dornsife College of Letters Arts and Sciences, Los Angeles, California, United States
  • Juntong Shi
    University of Southern California Viterbi School of Engineering, Los Angeles, California, United States
  • Angelamarie Ortez
    University of Southern California Viterbi School of Engineering, Los Angeles, California, United States
  • Alejandra Gonzalez Calle
    Ophthalmology, USC Roski Eye Institute, University of Southern California, Los Angeles, California, United States
    USC Ginsburg Institute for Biomedical Therapeutics, University of Southern California, Los Angeles, California, United States
  • Mark S. Humayun
    Ophthalmology, USC Roski Eye Institute, University of Southern California, Los Angeles, California, United States
    USC Ginsburg Institute for Biomedical Therapeutics, University of Southern California, Los Angeles, California, United States
  • Footnotes
    Commercial Relationships   Dimitrios Pollalis None; Timothy Lin None; Peter Chong None; Juntong Shi None; Angelamarie Ortez None; Alejandra Gonzalez Calle None; Mark Humayun None
  • Footnotes
    Support  National Science Foundation under Grant No. EMFA 1933394
Investigative Ophthalmology & Visual Science June 2023, Vol.64, 2134. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Dimitrios Pollalis, Timothy Lin, Peter Chong, Juntong Shi, Angelamarie Ortez, Alejandra Gonzalez Calle, Mark S. Humayun; Automated quantification of retinal degeneration in FAF images. Invest. Ophthalmol. Vis. Sci. 2023;64(8):2134.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : Fundus autofluorescence (FAF) imaging is a common method used to evaluate the progression of degeneration. However, traditional FAF analysis is time consuming and relies on user expertise and subjective decision-making. This project aims to 1) develop an automated software that quantifies the degenerated area in FAF images and, 2) validate this software against the conventional manual analysis.

Methods : Royal College of Surgeons (RCS) rats are commonly used for research in inherited retinal dystrophies. FAF imaging was performed using a commercially available cSLO (SPECTRALIS HRA+OCT, Heidelberg, Germany) on 8 RCS rats at two time points: 1) Baseline at postnatal day 21 (P21), which was used for excluding rats with pre-existing retina or optic nerve defects; and 2) Follow up at P49, which used for FAF analysis. For manual analysis, ImageJ (NIH, USA) was used. Briefly, the degenerated area was freehand selected, thresholded, and quantified. In designing our automated protocol, we identified an existing filter from Scikit-Image that works to isolate the bright features of an image. Then, each FAF image was split into 13 smaller areas and every region of interest (ROI) was thresholded based on the background of each individual subarea. We took advantage of different h-values for each subarea to control the sensitivity of the thresholding process to eliminate any potential confounders (e.g., shadows). The values obtained from the automated method were compared with the valued from the manual method using correlation statistical analysis.

Results : For testing the efficacy of our automated software, a small pilot study of eight RCS rats was performed. The areas of retinal degeneration in FAF images of P49 rats were successfully detected using our automated analysis protocol. The mean ± SD percentage of the ROI using ImageJ was 11% ± 8.7% and 9.3% ± 3.5% for the automated method. Correlation analysis of these two methods showed a Pearson r of 0.81.

Conclusions : This pilot study demonstrated comparable quantification of retinal degeneration in FAF images when using the automated analysis software versus manual ImageJ. The conventional manual analysis method is time-intensive and relies on user expertise and subjective decision-making. Future studies should include a larger cohort to further optimize and validate the protocol. Automated software could greatly increase efficiency and reduce user bias.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.

×
×

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.

×