April 2014
Volume 55, Issue 13
Free
ARVO Annual Meeting Abstract  |   April 2014
Automated Detection of Fluorescein Leakage in Diabetic Macular Edema
Author Affiliations & Notes
  • Amani Al-Tarouti
    Ophthalmology, University of Michigan Kellogg Eye Center, Ann Arbor, MI
  • Grant Michael Comer
    Ophthalmology, University of Michigan Kellogg Eye Center, Ann Arbor, MI
  • Pavan S Angadi
    Ophthalmology, University of Michigan Kellogg Eye Center, Ann Arbor, MI
  • Christopher Ranella
    Ophthalmology, University of Michigan Kellogg Eye Center, Ann Arbor, MI
  • Nathan Patel
    Ophthalmology, University of Michigan Kellogg Eye Center, Ann Arbor, MI
  • Daniel Albertus
    Ophthalmology, University of Michigan Kellogg Eye Center, Ann Arbor, MI
  • Maxwell Stem
    Ophthalmology, University of Michigan Kellogg Eye Center, Ann Arbor, MI
  • Matthew Johnson-Roberson
    Ophthalmology, University of Michigan Kellogg Eye Center, Ann Arbor, MI
  • Thiran Jayasundera
    Ophthalmology, University of Michigan Kellogg Eye Center, Ann Arbor, MI
  • Footnotes
    Commercial Relationships Amani Al-Tarouti, None; Grant Comer, None; Pavan Angadi, None; Christopher Ranella, None; Nathan Patel, None; Daniel Albertus, None; Maxwell Stem, None; Matthew Johnson-Roberson, None; Thiran Jayasundera, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science April 2014, Vol.55, 4832. doi:https://doi.org/
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Amani Al-Tarouti, Grant Michael Comer, Pavan S Angadi, Christopher Ranella, Nathan Patel, Daniel Albertus, Maxwell Stem, Matthew Johnson-Roberson, Thiran Jayasundera; Automated Detection of Fluorescein Leakage in Diabetic Macular Edema. Invest. Ophthalmol. Vis. Sci. 2014;55(13):4832. doi: https://doi.org/.

      Download citation file:


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

      ×
  • Supplements
Abstract
 
Purpose
 

To determine the accuracy of a novel method of quantitative image analysis for fluorescein angiography (FA) that demonstrates clinical utility in identifying and following treatment efficacy of fluorescein leakage extravasation in diabetic macular edema (DME).

 
Methods
 

This proof of concept study obtained a small FA image database (88) of diabetic macular edema eyes, before and 6 months after treatment with intravitreal angiogenesis inhibitors. Eight images were chosen. Two of these images had no leakage. The baseline and 6 months post-treatment images were analyzed for quantification of area change based on pixel area difference between early (25-30 seconds) and late (after 2-3 minutes) FA images. For each image, areas of fluorescein leakage extravasation were outlined and graded according to the ETDRS grading scheme by two human graders (retina specialists) utilizing an interactive pen display thus generating quantifiable values across the macula. A comparison of the generated signatures was performed by calculating histogram information of the pixels in the outlined regions, producing a single quantitative value of the area change ( pixel difference from early to late phase ); the Index of Retinal Leakage (IRL). Inter-grader variability was calculated between the indices outlined by the human graders. The IRL was compared to a (texture based feature vector classification) algorithm generated value and calculated as an IRL as well, thus quantifying leakage area change.

 
Results
 

The area change generated by the vector algorithm were similar to that observed by the specialists’ interactive grading. The algorithm was able to demonstrate lack of leakage as well as the human graders by producing zero area change. There was less variability demonstrated by the algorithm compared to the human grader variability.

 
Conclusions
 

The IRL generated by the new algorithm exhibited acceptable inter-grader variability for effective quantification of changes in FA images due to fluorescein leakage inDME extravasation and may serve as a new outcome measure tool to in quantify disease quantifying disease progression and monitoring response to treatment. Further investigation of the algorithm will be done to demonstrate consistency and reproducibility of results to establish an objective leakage assessment process in DME.

 
 
First human grading outline.
 
First human grading outline.
 
 
Second human grader outline.
 
Second human grader outline.
 
Keywords: 549 image processing • 550 imaging/image analysis: clinical • 499 diabetic retinopathy  
×
×

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.

×