Investigative Ophthalmology & Visual Science Cover Image for Volume 57, Issue 12
September 2016
Volume 57, Issue 12
Open Access
ARVO Annual Meeting Abstract  |   September 2016
Validation of an Automated Software for Choroidal Thickness (CTh) Measurement
Author Affiliations & Notes
  • Aushim Kokroo
    NYU Langone Medical Center, New York, New York, United States
  • Alauddin Bhuiyan
    NYU Langone Medical Center, New York, New York, United States
  • Akter Hussain
    Department of Computer Science and Information Technology, University of Melbourne, Parkville, Victoria, Australia
  • Rao Kotagiri
    Department of Computer Science and Information Technology, University of Melbourne, Parkville, Victoria, Australia
  • Meleha Ahmad
    NYU Langone Medical Center, New York, New York, United States
  • Theodore Smith
    NYU Langone Medical Center, New York, New York, United States
  • Footnotes
    Commercial Relationships   Aushim Kokroo, None; Alauddin Bhuiyan, HealthScreener Inc. (E); Akter Hussain, None; Rao Kotagiri, None; Meleha Ahmad, None; Theodore Smith, None
  • Footnotes
    Support  Unrestricted funds for RPB, NH Grant EY015520
Investigative Ophthalmology & Visual Science September 2016, Vol.57, 5954. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Aushim Kokroo, Alauddin Bhuiyan, Akter Hussain, Rao Kotagiri, Meleha Ahmad, Theodore Smith; Validation of an Automated Software for Choroidal Thickness (CTh) Measurement. Invest. Ophthalmol. Vis. Sci. 2016;57(12):5954.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : The growing interest in choroidal pathology as a biomarker for systemic and retinal diseases has increased the need for automated CTh measurement. Here, we assess the accuracy of such an algorithm applied to Enhanced Depth Imaging Spectral Domain Optical Coherence Tomography (EDI SD-OCT) images.

Methods : Following B-scan cropping, the top boundary of Retinal Nerve Fiber Layer (RNFL) was detected by applying Canny edge detection, intensity profiling and Dijkstra’s shortest path algorithm; we approximately identified the Retinal Nerve Fiber Layer, Outer Nuclear Layer and Retinal Pigment Epithelium to make the search space smaller. We subsequently tracked the Bruch’s Membrane boundary (BMB) using the same approach. Using BMB as a reference, we identified the choroidal vessel boundaries through an active contour method, thereby identifying a potential choroidal region. Finally, the choroidal boundary was identified from the choroidal vessel demarcations, intensity and boundary gradient information. The software was run on foveal slice EDI SD-OCT images of 15 eyes from 14 patients, generating a continuous CTh measurement curve (Figure 1). Two masked graders also manually demarcated both choroidal boundaries using Adobe Photoshop, allowing automatic A-Scan CTh measurements across the manually graded B-scan. The accuracy of the CTh computed by our software was tested against CTh determined by expert graders using the Spearman rank correlation (ρ). We also computed the mean square error (MSE) between manually graded and automatically detected boundaries through direct pixel-by-pixel matching within the boundaries of the choroid.

Results : Spearman rank correlations (ρ) of 0.735 and 0.729, respectively, were obtained between the two expert graders’ measurements and the automatically measured thicknesses (p<0.01 for both). The MSE for boundary matching between the automatic software and first grader was 0.113 pixels2 and for the second grader 0.079 pixels2.

Conclusions : Our novel automated system for CTh quantification demonstrated strong agreement with manual CTh measurement. In the future, this method can be used to rapidly quantify CTh in various ophthalmological and systemic diseases.

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

 

Figure 1: Top, automated choroidal boundaries at foveal slice. Outer RPE (blue), outer choroid (red). Bottom, detailed choroidal thickness measurements (blue), range 245 to 285 microns

Figure 1: Top, automated choroidal boundaries at foveal slice. Outer RPE (blue), outer choroid (red). Bottom, detailed choroidal thickness measurements (blue), range 245 to 285 microns

×
×

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.

×