April 2011
Volume 52, Issue 14
Free
ARVO Annual Meeting Abstract  |   April 2011
Comparison of Spectral Domain Optical Coherence Tomography Segmentation Algorithms in the Analysis of Pigment Epithelial Detachments
Author Affiliations & Notes
  • Nishi Gulati
    Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, Palm Beach Gardens, Florida
  • Fernando M. Penha
    Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, Miami, Florida
  • Philip J. Rosenfeld
    Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, Miami, Florida
  • Giovanni Gregori
    Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, Miami, Florida
  • Zohar Yehoshua
    Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, Miami, Florida
  • Footnotes
    Commercial Relationships  Nishi Gulati, None; Fernando M. Penha, Carl Zeiss Meditec (F); Philip J. Rosenfeld, Carl Zeiss Meditec (F, C, R); Giovanni Gregori, Carl Zeiss Meditec (F, P, R); Zohar Yehoshua, Carl Zeiss Meditec (F)
  • Footnotes
    Support  Research to Prevent Blindness; NEI grant P30 EY014801; Carl Zeiss Meditec; Foundations: Jerome A. Yavitz; Emma Clyde Hodge; Florman Family; Gemcon Family
Investigative Ophthalmology & Visual Science April 2011, Vol.52, 4043. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Nishi Gulati, Fernando M. Penha, Philip J. Rosenfeld, Giovanni Gregori, Zohar Yehoshua; Comparison of Spectral Domain Optical Coherence Tomography Segmentation Algorithms in the Analysis of Pigment Epithelial Detachments. Invest. Ophthalmol. Vis. Sci. 2011;52(14):4043.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose: : To compare the performance of segmentation algorithms in the analysis of pigment epithelium detachments (PEDs) using two different spectral domain optical coherence tomography instruments (SDOCT).

Methods: : Twenty seven eyes from 25 patients with different types of PEDs were enrolled. All eyes were scanned using both the Cirrus SDOCT (Carl Zeiss Meditec, Dublin, CA) and the Spectralis SDOCT (Heidelberg Engineering, Inc., Heidelberg, Germany) instruments during a single visit. All the scans were centered on the fovea. Two different Cirrus scan patterns were used; the 512 x 128 and 200 x 200 A-scan raster pattern covering a 6X6 mm (20x20 degree) area. The Spectralis SDOCT scan pattern consisted of 7 B-scans, each averaged 51 times, covering a 30X5 degree area . The main outcome measure was to compare the segmentation performance of the B-scan which included the foveal center.

Results: : The Spectralis algorithm failed to properly segment the central B-scan in 25 eyes (92.6%); the algorithm failed to follow the RPE. The Cirrus ialgorithm successfully segmented the RPE layer in every central B-scan. The mean center point thickness recorded by the Spectralis was 362.29 µm ± 156.37 (176-919 µm). The mean center point thickness using the Cirrus was 220.18 µm ± 69.30 (119 - 393 µm) and 223.33 µm ± 70.04 (119-399 µm), using the 200x200 or 512x128 scan patterns, respectively. There was no difference between the two Cirrus scan patterns. There was a statistically significant difference between the Cirrus and Spectralis central thickness measurements (F=15.61; p < 0.001).

Conclusions: : The Cirrus SDOCT algorithm is superior to the Spectralis algorithm for identifying appropriate segmentation boundaries in the presence PEDs. The Spectralis SDOCT algorithm is unreliable when segmenting PEDS. This leads to erroneous retinal thickness measurements. The Spectralis instrument should only useful for the qualitative assessment of PEDs unless manual adjustments to the boundary layers are performed.

Keywords: imaging/image analysis: clinical • age-related macular degeneration • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) 
×
×

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.

×