April 2010
Volume 51, Issue 13
Free
ARVO Annual Meeting Abstract  |   April 2010
Automated Segmentation of Multiple Intraretinal Boundaries in 3D Spectral-Domain Optical Coherence Tomography
Author Affiliations & Notes
  • M. Hangai
    Ophthalmology and Visual Sciences, Kyoto Univ Graduate School of Medicine, Kyoto, Japan
  • A. Tomidokoro
    Ophthalmology, Univ of Tokyo School of Medicine, Tokyo, Japan
  • M. Araie
    Ophthalmology, Univ of Tokyo School of Medicine, Tokyo, Japan
  • N. Yoshimura
    Ophthalmology and Visual Sciences, Kyoto Univ Graduate School of Medicine, Kyoto, Japan
  • 3D OCT Study Group
    Ophthalmology and Visual Sciences, Kyoto Univ Graduate School of Medicine, Kyoto, Japan
  • Footnotes
    Commercial Relationships  M. Hangai, TOPCON, NIDEK, C; A. Tomidokoro, TOPCON, C; M. Araie, TOPCON, C; N. Yoshimura, TOPCON, NIDEK, C.
  • Footnotes
    Support  a Grant-in-Aid for Scientific Research (20592038) from the Japan Society for the Promotion of Science (JSPS).
Investigative Ophthalmology & Visual Science April 2010, Vol.51, 221. doi:https://doi.org/
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      M. Hangai, A. Tomidokoro, M. Araie, N. Yoshimura, 3D OCT Study Group; Automated Segmentation of Multiple Intraretinal Boundaries in 3D Spectral-Domain Optical Coherence Tomography. Invest. Ophthalmol. Vis. Sci. 2010;51(13):221. doi: https://doi.org/.

      Download citation file:


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

      ×
  • Supplements
Abstract
 
Purpose:
 

To automate segmentation of multiple intraretinal boundaries in 3-dimensional (3D) spectral-domain optical coherence tomography (SD-OCT) images and to assess the reproducibility of measurements of single intraretinal layers, (e.g., the ganglion cell layer (GCL)), and 2-3 combined layers, including the GCL, in normal and glaucomatous eyes.

 
Methods:
 

We used data of 3 vertical 3D scans over the macula of 43 normal eyes of 26 subjects and 59 glaucomatous eyes of 46 patients. OCT-1000 (Topcon) was used to obtain 512 A-scans over 128 frames. An automated segmentation algorithm based on a customized edge detector and edge refinery was developed. It can segment ≤9 intraretinal boundaries (Table 1). Two experts visually inspected 128 B-scans per eye to assess significant boundary-detection failures as per previously reported criteria (Invest Ophthalmol Vis Sci. 2005;46: 2012-2017). Intraclass correlation coefficients (ICC) and coefficients of variation (CV) were calculated to assess the thickness-measurement repeatability.

 
Results:
 

For each 3D scan, the mean number of B-scans deemed to have significant detection failures by either of the experts was 0.07-0.29 for the 8 boundaries but 57.9 for the external limiting membrane (ELM) in normal eyes. The ICCs and CVs were excellent in single GCL layers, combined layers including the GCL, the inner nuclear layer, and the combined outer plexiform and outer nuclear layers in both normal and glaucomatous eyes (Table 1).

 
Conclusions:
 

The automated segmentation algorithm was reliable in 8 boundaries but none in the ELM. Automated segmentation of multiple intraretinal layers enabled reliable measurement of the thickness of the GCL layer and of combined layers including GCL.  

 
Keywords: imaging/image analysis: clinical • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • retina 
×
×

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.

×