April 2014
Volume 55, Issue 13
Free
ARVO Annual Meeting Abstract  |   April 2014
Quantifying Retinal Vascular Differences in SDOCT Images from Diabetics vs. Controls
Author Affiliations & Notes
  • Joel A Papay
    School of Optometry, Indiana University, Bloomington, IN
  • Ann E Elsner
    School of Optometry, Indiana University, Bloomington, IN
  • Andrea Walker
    School of Optometry, Indiana University, Bloomington, IN
  • Footnotes
    Commercial Relationships Joel Papay, None; Ann Elsner, None; Andrea Walker, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science April 2014, Vol.55, 211. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Joel A Papay, Ann E Elsner, Andrea Walker; Quantifying Retinal Vascular Differences in SDOCT Images from Diabetics vs. Controls. Invest. Ophthalmol. Vis. Sci. 2014;55(13):211.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose: Vascular changes in the retina are crucial to diabetic patients. Retinal vessels can become dilated and hyper reflective, quantifiable in spectral domain optical coherence tomography (SDOCT) via a frequency analysis.

Methods: One foveal centered SDOCT scan (Spectralis, Heidelberg Engineering) was taken for each of 35 diabetic subjects (20 females, 15 males) and age/gender-matched controls (age range 39-73). These images were processed using custom routines in Matlab (Mathworks). The images had the nerve fiber layer (NFL) and the region from the exterior limiting membrane (ELM) outward manually removed. The mean gray level of what remained filled in the reminder of the image. The images were made larger and square by adding rows of the mean gray level to the bottom. Five columns from each side of the image were set to the mean gray level, and 25 columns ramped to the average grayscale. Images were flattened with respect to ELM. These images were then filtered with 3x3, 5x5, 7x7, and 9x9 pixel Wiener filters and saved individually as tiff files. 2D FTs of the baseline and filtered images were computed and the difference between them was taken. The artifact caused by flattened ELM boundary and the surrounding average gray image were removed. The average values for rings around the DC, 3-4 pixels wide were computed.

Results: Processing with a Wiener filter decreased high frequency noise and increased visibility of retinal blood vessels. Consistent differences were seen in the 2D FTs as a function of eccentricity from the DC and also with filter size. Significant differences were found between unfiltered and filtered images at the frequencies that are associated with blood vessel sizes at p<0.05. The filter size shifted the appearance of the filtered blood vessel images and also the frequencies at which the peak difference was found between diabetic and normal subjects, with the smallest filter emphasizing slightly lower spatial frequency differences in the FT.

Conclusions: Retinal vessels become more predominant in Wiener filtered SDOCT cross-sections, especially for diabetics. Our data are consistent with an enhancement of the power in a specific range of spatial frequencies rather than more noisy images in general, because the high frequencies overlapped for diabetics and controls.

Keywords: 550 imaging/image analysis: clinical • 498 diabetes • 688 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.

×