September 2016
Volume 57, Issue 12
Open Access
ARVO Annual Meeting Abstract  |   September 2016
Automated Quantification of Morphologic Features and Vasculature of Choroid on Enhanced Depth Imaging Optical Coherence Tomography (EDI-OCT)
Author Affiliations & Notes
  • Yuanjie Zheng
    School of Information Science and Engineering, Shandong Normal University, Jinan, Shandong, China
  • Purak Parikh
    University of Pennsylvania, Philadelphia, Pennsylvania, United States
  • Brian L VanderBeek
    University of Pennsylvania, Philadelphia, Pennsylvania, United States
  • James Gee
    University of Pennsylvania, Philadelphia, Pennsylvania, United States
  • Footnotes
    Commercial Relationships   Yuanjie Zheng, None; Purak Parikh, None; Brian VanderBeek, None; James Gee, None
  • Footnotes
    Support  NIH Grant K23-EY025729; NIH Grant P30 EY001583; NFSC Grant 61572300; NSF-Shandong Grant ZR2014FM001; Taishan Scholar Program of Shandong Province
Investigative Ophthalmology & Visual Science September 2016, Vol.57, 4652. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Yuanjie Zheng, Purak Parikh, Brian L VanderBeek, James Gee; Automated Quantification of Morphologic Features and Vasculature of Choroid on Enhanced Depth Imaging Optical Coherence Tomography (EDI-OCT). Invest. Ophthalmol. Vis. Sci. 2016;57(12):4652.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : The choroid of the eye is predominately vascular structure which plays a vital role in the pathogenesis of various eye diseases. Our goal is to develop techniques for automatic segmentation and characterization of the choroid using EDI-OCT and provide a quantitative analysis of morphologic characteristics of the choroid.

Methods : A set of automated image analysis algorithms were developed to assess qualities of the choroid in normal healthy eyes, including 1) detection of choroid region by specifying the Choroid-sclera junction and the Bruch’s membrane-RPE interface via a curve fitting process accomplished by examining image gradient; 2) segmentation of vessel by exploring the geometric curvature of image intensity surface; 3) description of choroid morphologic characteristics including choroidal thickness, percentage of vessel area relative to the complete choroid region. Nine EDI-OCT images were obtained in normal healthy eyes (5 male and 4 female, average age: 34.2, all emmotropic), and from which, manual segmentation of choroidal boundaries and vasculature was conducted by a trained ophthalmologist via Adobe Photoshop. Evaluation was carried out not only by a quantitative comparison between the algorithm’s result and manual segmentation on a pixel-by-pixel basis but also via a qualitative visual assessment.

Results : For segmentation of choroid-region/choroidal-vessel, the automatic method produced AUC (area under curve) values of 0.69/0.71, accuracy of 0.68/0.70, specificity of 0.75/0.78 and sensitivity of 0.74/0.63. The average thickness of the choroid segmented by the automatic method is 291.26μm and the percentage of vessel relative to stroma was 44.23%. Subjective visual assessments by two ophthalmologists agreed that algorithm’s results are better in most cases. Vessel segmentation took 1.2 seconds on average per image by our algorithm using a CPU at 2.53GHz and ~1 hour/image by the ophthalmologist.

Conclusions : Our goal was to achieve a quantitative assessment of the vascular-structural changes of the choroid by leveraging EDI-OCT and developing automated algorithms for EDI-OCT image analysis. The proposed automatic method demonstrated a good agreement with and prominently higher speed than manual choroidal vessel segmentation. It also holds promise in detecting more subtle changes of choroidal morphological features in diseased states.

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

 

×
×

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.

×