May 2005
Volume 46, Issue 13
Free
ARVO Annual Meeting Abstract  |   May 2005
Extracting Subretinal Layers on Stratus OCT Images via a Structure Tensor Approach Combined With a Nonlinear Diffusion Process
Author Affiliations & Notes
  • D. Cabrera Fernández
    Bascom Palmer Eye Institute, University of Miami School of Medicine, Miami, FL
  • H.M. Salinas
    Bascom Palmer Eye Institute, University of Miami School of Medicine, Miami, FL
  • Footnotes
    Commercial Relationships  D. Cabrera Fernández, None; H.M. Salinas, None.
  • Footnotes
    Support  RO1EY008684–10S1; P30 core grant EY014801
Investigative Ophthalmology & Visual Science May 2005, Vol.46, 2575. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      D. Cabrera Fernández, H.M. Salinas; Extracting Subretinal Layers on Stratus OCT Images via a Structure Tensor Approach Combined With a Nonlinear Diffusion Process . Invest. Ophthalmol. Vis. Sci. 2005;46(13):2575.

      Download citation file:


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

      ×
  • Supplements
Abstract

Abstract: : Purpose:Segmentation of retinal layers from OCT images is a fundamental step to diagnose the progress of the disease. This study intends to show that the retinal sublayers (NFL, GCL, IPL, OPL and ONL) can be automatically and/or interactively located with good accuracy with the aid of local coherence information of the retinal structure. Methods: STRATUS OCT images from normal subjects are processed using the ideas of texture analysis by means of the structure tensor combined with the scale–space concept of anisotropic diffusion filtering. An m–dimensional scale–space for the enhancement of coherent structures is used to improve the edge detection of the global retinal layers (ILM, RPE) and the retinal sublayers (NFL, GCL, IPL, OPL and ONL). Once the images have been denoised they are segmented using a boundary detection algorithm based on local coherence information of the structure. Results: The thickness of the global and subretinal layers was measured for each set of radial scans in every normal subject, and statistical analysis was performed for each image. The average reflectance information was also extracted for each layer in the images analyzed. A significant signal to noise ratio improvement of more than 3 dB with a reduction in image sharpness of less than 9 % was observed qualitatively and quantitatively after denoising. A resulting image with the same average gray level as the original one is obtained. The boundaries for the RNFL, GCL, IPL, OPL, ONL and RPE were reliably detected. Conclusions:The currently available STRATUS OCT algorithm not only computes incorrect retinal thickness values but it is also not able to detect all the retinal sublayers. Experimental results indicate that our proposed approach has good performance in speckle noise removal and enhancement of retinal sublayers. This image enhancement not only serves as a preprocessing step for the subsequent segmentation of the retinal sublayers but also serves for facilitating the extraction of local reflectance properties and structural analysis. Thus, the methodology presented could have an important role in the future development of computer–assisted OCT quantification techniques.

Keywords: image processing • 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.

×