September 2016
Volume 57, Issue 12
Open Access
ARVO Annual Meeting Abstract  |   September 2016
Automated Quantitative Characterization of Retinal Vascular Features in Ultra-widefield Fluorescein Angiography
Author Affiliations & Notes
  • Kevin Wang
    School of Medicine, Case Western Reserve University, Cleveland Heights, Ohio, United States
  • Sunil K Srivastava
    Cole Eye Institute, Cleveland Clinic, Cleveland, Ohio, United States
  • Justis P Ehlers
    Cole Eye Institute, Cleveland Clinic, Cleveland, Ohio, United States
  • Footnotes
    Commercial Relationships   Kevin Wang, None; Sunil Srivastava, Alcon (C), Allergan (F), Bausch and Lomb (C), Bioptigen (C), Bioptigen (P), Santen (C), Synergetics (P), Zeiss (C); Justis Ehlers, Alcon (C), Bioptigen (C), Bioptogen (P), Genentech (F), Regeneron (F), Santen (C), Synergetics (P), Thrombogenics (C), Thrombogenics (F), Zeiss (C)
  • Footnotes
    Support  NIH/NEI K23-EY022947-01A1; Ohio Department of Development TECH-13-059
Investigative Ophthalmology & Visual Science September 2016, Vol.57, 1681. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Kevin Wang, Sunil K Srivastava, Justis P Ehlers; Automated Quantitative Characterization of Retinal Vascular Features in Ultra-widefield Fluorescein Angiography. Invest. Ophthalmol. Vis. Sci. 2016;57(12):1681.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : To evaluate an automated algorithm for identification and quantification of retinal microaneurysms (MA) and leakage on ultra-widefield fluorescein angiography (UWFA) images.

Methods : An automated unsupervised algorithm was developed for multiple vascular parameters, including MA and leakage. A validation dataset consisted of UWFA obtained on the Optos 200Tx system (Optos, Scotland) from 28 patients with diabetic retinopathy. Both standard and dewarped images were analyzed. One early and one late image was selected for analysis. The early image was assessed for MAs and the late image was assessed for leakage. Two independent expert graders manually assessed all parameters in ImageJ (NIH freeware) and utilized the region of interest tool for direct comparisons between each reader and the algorithm.

Results : The algorithm successfully identified both MA and leakage on UWFA. The algorithm and 2 manual graders exhibited similar variability in both leakage and MA analysis. Dewarping images also resulted in changes to both manual and algorithm-based grading. Specifically, there was a 33% decrease in MA detection rate by both manual and algorithm grading in dewarped compared to warped images. An automated zonal measurement system was successfully implemented with concentric zones around the optic nerve. This zonal measurement system facilitated a more refined approach to peripheral artifacts, such as lashes or media opacities.

Conclusions : The automated angiography quantitative tool provided objective rapid assessment of angiogram features, in particular MA and leakage burden. The algorithm performance appeared similar to human variability. Providing quantitative longitudinal assessment of retinal vascular changes may be a critical aspect of assessing disease progression and therapeutic response. Additionally, integrative pattern analysis of angiographic patterns may provide critical biomarkers for individualizing therapeutic selection and to maximize therapeutic response.

This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.

 

Figure 1. MA grading of A) UWFA image by B) grader 1, C) grader 2, and D) automated algorithm. The number of MAs detected by grader 1, grader 2 and algorithm were 355, 295, and 335 respectively.

Figure 1. MA grading of A) UWFA image by B) grader 1, C) grader 2, and D) automated algorithm. The number of MAs detected by grader 1, grader 2 and algorithm were 355, 295, and 335 respectively.

 

Figure 2. Leakage grading of A) UWFA image by B) grader 1, C) grader 2, and D) algorithm.

Figure 2. Leakage grading of A) UWFA image by B) grader 1, C) grader 2, and D) algorithm.

×
×

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.

×