June 2015
Volume 56, Issue 7
Free
ARVO Annual Meeting Abstract  |   June 2015
Inter- and Intra-observer agreement of Newly Developed Software for Retinal Arteriolar Focal Narrowing Quantification
Author Affiliations & Notes
  • Alauddin Bhuiyan
    Australian E-Health Research Centre, Commonwealth Scientific &Industry Rsch Org, Floreat, WA, Australia
  • Pallab Roy
    The University of Melbourne, Melbourne, VIC, Australia
  • Akter Hussain
    The University of Melbourne, Melbourne, VIC, Australia
  • Kim Lee
    The University of Melbourne, Melbourne, VIC, Australia
  • Ecosse Luc Lamoureux
    Singapore Eye Research Institure, Singapore, Singapore
    DUKE-NUS, Office of Clinical Sciences, Singapore, Singapore, Singapore
  • Tien Yin Wong
    Singapore Eye Research Institure, Singapore, Singapore
    DUKE-NUS, Office of Clinical Sciences, Singapore, Singapore, Singapore
  • Elsdon Storey
    Monash University, Melbourne, VIC, Australia
  • Yogesan Kanagasingam
    Australian E-Health Research Centre, Commonwealth Scientific &Industry Rsch Org, Floreat, WA, Australia
  • Rao Kotagiri
    The University of Melbourne, Melbourne, VIC, Australia
  • Footnotes
    Commercial Relationships Alauddin Bhuiyan, None; Pallab Roy, None; Akter Hussain, None; Kim Lee, None; Ecosse Lamoureux, None; Tien Wong, None; Elsdon Storey, None; Yogesan Kanagasingam, None; Rao Kotagiri, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2015, Vol.56, 5247. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Alauddin Bhuiyan, Pallab Roy, Akter Hussain, Kim Lee, Ecosse Luc Lamoureux, Tien Yin Wong, Elsdon Storey, Yogesan Kanagasingam, Rao Kotagiri; Inter- and Intra-observer agreement of Newly Developed Software for Retinal Arteriolar Focal Narrowing Quantification . Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):5247.

      Download citation file:


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

      ×
  • Supplements
Abstract
 
Purpose
 

To assess the accuracy and repeatability of novel computer based semi-automated software for quantifying the severity of retinal arteriolar focal narrowing (FN), a hypertensive retinopathy sign, from colour retinal photographs.

 
Methods
 

23 retinal photographs were randomly selected from the Envision Study, a population-based study of persons aged 70 years or over. Total 300 images were considered. The rest images were excluded for not having FN point or being poor quality. A novel computer assisted software was developed to quantify retinal FN by selecting vessel region semi-automatically where FN exists. The software then crops the FN region, calculates the vessel edges, computes the widths and finds the FN points by finding the width's sudden changes. The software provides quantified FN as continuous numerical values and the severity level as none, mild, moderate and severe. The accuracy of the FN computation method was tested against the manually measured FN severity level by expert grader through the Spearman rank correlation. Inter-grader reliability were assessed by measurements from two individual graders of the same eye and intra-session repeatability was assessed by grading the same image at two different times, using Intra-class Correlation Coefficients (ICC).

 
Results
 

Accuracy measurements indicated a Spearman rank correlation (ρ) of 0.88 (p<0.001) between software measured FN values and expert graders’ ranking on FN severity. Between-graders and intra-session ICC values for FN were 0.94 (95% Confidence Interval [CI]: 0.85-0.97) and 0.93 (95% CI: 0.83-0.97) respectively, which indicates high repeatability of the proposed software for FN severity measurement.

 
Conclusions
 

A novel semi-automatic measurement system for retinal arteriolar focal narrowing quantification is demonstrated for accuracy and grading repeatability for the first time. A good agreement with standard grading and excellent reproducibility, indicating that quantitative assessment of retinal FN is possible.  

 
Steps of of FN computation method (a) Region of interest<br /> selection from original image (b) spatial normalization<br /> (c) noise removal and vessel enhancement (d) detected edge<br /> after applying canny edge detector (e) mapped vessel edge using<br /> Dijkstra’s shortest path algorithm (f) mapped edge overlapped<br /> on the ROI (g) computed vessel caliber or width for finding FN.
 
Steps of of FN computation method (a) Region of interest<br /> selection from original image (b) spatial normalization<br /> (c) noise removal and vessel enhancement (d) detected edge<br /> after applying canny edge detector (e) mapped vessel edge using<br /> Dijkstra’s shortest path algorithm (f) mapped edge overlapped<br /> on the ROI (g) computed vessel caliber or width for finding FN.

 
×
×

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.

×