April 2011
Volume 52, Issue 14
ARVO Annual Meeting Abstract  |   April 2011
Computer-aided Screening of Cardiovascular Disease Based on Retinal Imaging: Results using Vasculature Features
Author Affiliations & Notes
  • Honggang Yu
    VisionQuest Biomedical LLC, Albuquerque, New Mexico
  • Carla Agurto Rios
    VisionQuest Biomedical LLC, Albuquerque, New Mexico
    Electrical & Computer Engineering, University of New Mexico, Albuquerque, New Mexico
  • Eduardo S. Barriga
    VisionQuest Biomedical LLC, Albuquerque, New Mexico
  • Wendall C. Bauman, Jr.
    Ophthalmology, Retina Institute of South Texas, San Antonio, Texas
  • Peter Soliz
    VisionQuest Biomedical LLC, Albuquerque, New Mexico
  • Gilberto Zamora
    VisionQuest Biomedical LLC, Albuquerque, New Mexico
  • Footnotes
    Commercial Relationships  Honggang Yu, VisionQuest Biomedical (E); Carla Agurto Rios, University of New Mexico (F), VisionQuest Biomedical (E); Eduardo S. Barriga, VisionQuest Biomedical (E); Wendall C. Bauman, Jr., Retina Institute of South Texas (I); Peter Soliz, VisionQuest Biomedical (I); Gilberto Zamora, VisionQuest Biomedical (E)
  • Footnotes
    Support  None
Investigative Ophthalmology & Visual Science April 2011, Vol.52, 1344. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Honggang Yu, Carla Agurto Rios, Eduardo S. Barriga, Wendall C. Bauman, Jr., Peter Soliz, Gilberto Zamora; Computer-aided Screening of Cardiovascular Disease Based on Retinal Imaging: Results using Vasculature Features. Invest. Ophthalmol. Vis. Sci. 2011;52(14):1344.

      Download citation file:

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

  • Supplements

To develop and test a system for computer-aided screening of systemic cardiovascular disease based on the topographical characteristics of the retinal vasculature.


Data consisted of a set of 55 retrospective cases (2224×1888 pixels, 45-degree field of view, mydriatic) with ground truth on the presence of systemic cardiovascular diseases (CVD). Twenty cases were normal, 25 had hypertension, 15 had coronary heart disease, and 12 had stroke (several cases presented more than one CVD). Image analysis consists of vessel segmentation, optic disc (OD) detection, and vascular network analysis. Vessels are segmented using multiscale Hessian eigensystems and local entropy thresholding. The OD is located using template matching and vessel branch detection inside OD candidates. Vasculature analysis includes arterio-venous ratio (AVR), tortuosity index (TI), and fractal dimension. AVR is calculated using the Parr-Hubbard method within an annulus surrounding the OD of radius 1 to 1.5 disc diameters. TI is calculated using the normalized arc to chord length ratio and integrated curvature. Fractal dimension is calculated using the box-counting method.


AVR for the normal group was statistically different from the CVD group (0.75 and 0.49, respectively, p<0.05). Similar results held for fractal dimension (1.42 and 1.29, p<0.05) and tortuosity index (7.35 and 19.91, p<0.01). The correlation between the computer-generated TI and expert visual tortuosity grading was 0.96 (R2=0.8). As a screening tool, and using leave-one-out for validation, the system had a sensitivity of 0.85 and a specificity of 0.8. Area under the ROC curve was 0.85.


This work presents a feasible methodology to conduct computer-aided screening of CVD by non-cardiologists or ophthalmologists that leverages the known association between retinal vasculature and CVD, automatic assessment of retinal features, and the availability of high resolution retinal images.

Keywords: image processing • retina • imaging/image analysis: non-clinical 

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.