May 2003
Volume 44, Issue 13
ARVO Annual Meeting Abstract  |   May 2003
Perceptual Enhancement of Tissue Structures in Surgery
Author Affiliations & Notes
  • D. Ebrahimi
    Dept of Surgery, University of Toronto, Toronto, ON, Canada
  • S.J. Hamstra
    Dept of Surgery, University of Toronto, Toronto, ON, Canada
  • Footnotes
    Commercial Relationships  D. Ebrahimi, None; S.J. Hamstra, None.
  • Footnotes
    Support  NSERC
Investigative Ophthalmology & Visual Science May 2003, Vol.44, 4316. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      D. Ebrahimi, S.J. Hamstra; Perceptual Enhancement of Tissue Structures in Surgery . Invest. Ophthalmol. Vis. Sci. 2003;44(13):4316.

      Download citation file:

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

  • Supplements

Abstract: : Background: Last year, we described an approach to enhance surgical images based on their statistical properties, using a carefully selected set drawn from parathyroid surgery. Such enhancements are of great value for surgeons in training or while performing delicate and visually challenging procedures. To extend this, we examined a broader range of surgical images and developed more robust image enhancement techniques. Many believe that the visual system has evolved to be most responsive to natural scenes; their amplitude spectra decrease with frequency roughly as 1/fα, with α falling in a narrow range around 1. Any major differences with surgical images may help explain trainees' initial difficulty in visual discrimination of tissues. Purpose: To characterize the statistical nature of a wide range of surgical images and employ their unique properties in selecting image enhancement techniques, and to improve surgical tissue discrimination by modifying images to better suit human visual processing. Methods: We analyzed 4 sets (n=10, n=24, n=28, n=5) of different surgical images, one set of 64 natural textures (from Brodatz), and one set of 31 outdoor scenes. These surgical images represent difficult visual problems that involve delicate differential tissue dissection. To assess the structural properties of each set, we measured the slope of log-log amplitude spectra (α), where a lower value represents more coarseness. Non-linear local image processing techniques, as well as color manipulation, take advantage of this value to visually enhance the images. Results: The mean value of α was found to be 1.15 for the set of natural textures and 1.21 for the outdoor scenes. The same measure for the 4 sets of surgical images was found to be 1.66, 1.69, 1.48, and 1.44. T-tests comparing these values to the natural texture set yielded t(72) = 19.63, t(86) = 22.83, t(90) =11.50 and t(67) =9.10, respectively (all p<<0.0001). We took advantage of this difference in developing non-linear image enhancement techniques to modify surgical images. Conclusions: Surgical images have steeper spectra than most natural scenes, as they lack sharp edges and high frequency content. Such unique properties may result in trainees battling innate vision issues in addition to the usual challenges of technical skill acquisition. Consequently, we have developed a robust algorithm to enhance surgical images, which exploits the natural response of the human visual system. CR: None Support: NSERC.

Keywords: image processing • shape, form, contour, object perception • spatial vision 

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.