May 2004
Volume 45, Issue 13
Free
ARVO Annual Meeting Abstract  |   May 2004
Analysis of z–axis structural changes in the cow lamina cribrosa with changes in pressure using image matching
Author Affiliations & Notes
  • M. Balasubramanian
    Computer Science, LSU, Baton Rouge, LA
  • S.S. Iyengar
    Computer Science, LSU, Baton Rouge, LA
  • J. Reynaud
    Ophthalmology, LSU Eye Center, New Orleans, LA
  • A. Palkama
    Ophthalmology, LSU Eye Center, New Orleans, LA
  • R. Beuerman
    Ophthalmology, LSU Eye Center, New Orleans, LA
  • Footnotes
    Commercial Relationships  M. Balasubramanian, None; S. S. Iyengar, None; J. Reynaud, None; A. Palkama, None; R. Beuerman, None.
  • Footnotes
    Support  NIH Grants EY12416, RR016456, and EY02377 (departmental core grant); BOR/HEF 2000–05, RPB.
Investigative Ophthalmology & Visual Science May 2004, Vol.45, 2178. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      M. Balasubramanian, S.S. Iyengar, J. Reynaud, A. Palkama, R. Beuerman; Analysis of z–axis structural changes in the cow lamina cribrosa with changes in pressure using image matching . Invest. Ophthalmol. Vis. Sci. 2004;45(13):2178.

      Download citation file:


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

      ×
  • Supplements
Abstract

Abstract: : Purpose: To understand z–axis structural changes in the cow lamina cribrosa (LC) with the application of elevated pressure using a perfused posterior segment. Methods: The posterior segments of fresh cow eyes, perfused with oxygenated ringers, were positioned in a controlled pressure chamber to image the LC with a white light confocal microscope. Pressures tested were 12 (P1), 24 (P2) and 60mmHg (P3). With reference to P1, the z–axis displacement due to an elevated pressure (P2 and P3) is computed by matching the images at P1 with P2 and P3. We tested various area–based image matching (ABM) schemes, which compute a measure–of–match / mismatch to find the location of best match. The methods tested were: 1) Normalized inner–product method (NProd), 2) Direct correlation (DC), 3) FFT based correlation (FFTC), 4) Measure of absolute difference (MAD), and 5) Sum of squared differences (RMS). In order to validate the performance of the algorithms, a trained technician visually matched every image at P1 with the images at P2 and P3. The problem of high dimension with the ABM methods is avoided by using only the 3rd level daubechies wavelet coefficients of the images. Results: The FFTC and DC methods resulted in a high overall matching rate of 89.5% and were invariant to any illumination changes. The NProd, MAD and RMS methods resulted in a low matching rate. The images at P1 matched with the images at P2 at the same depths, indicating minimal or no laminar deformation. The image acquired at 0 micron at P1 matched with the image acquired at 110 micron at P3, indicating a 110 micron displacement. The subsequent images collected at a 20–micron interval from 0 micron at P1 and 110 microns at P3 matched respectively until a depth of 280 microns at P1 was reached. The FFTC method was the fastest matching method, but with a slightly different correlation surface than the DC method; however, the location of best match was invariant with both the methods. Conclusions: ABM was found to be a viable method for tracking any given layer of LC. The method failed at an increased depth and possibly due to maintenance of the 20–micron interval in acquiring the images.

Keywords: computational modeling • image processing • imaging/image analysis: non–clinical 
×
×

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.

×