May 2003
Volume 44, Issue 13
Free
ARVO Annual Meeting Abstract  |   May 2003
Evaluation of Algorithms for the Automated Registration of Scanning Laser Tomographic and Photographic Images
Author Affiliations & Notes
  • J.E. Morgan
    Ophthalmology, University Hospital of Wales, Cardiff, United Kingdom
  • P. Rosin
    Computer Science, Cardiff University, Cardiff, United Kingdom
  • D. Marshall
    Computer Science, Cardiff University, Cardiff, United Kingdom
  • J.M. Wild
    Optometry and Vision Sciences, Cardiff University, Cardiff, United Kingdom
  • R.V. North
    Optometry and Vision Sciences, Cardiff University, Cardiff, United Kingdom
  • Footnotes
    Commercial Relationships  J.E. Morgan, None; P. Rosin, None; D. Marshall, None; J.M. Wild, None; R.V. North, None.
  • Footnotes
    Support  WORD UK
Investigative Ophthalmology & Visual Science May 2003, Vol.44, 3398. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      J.E. Morgan, P. Rosin, D. Marshall, J.M. Wild, R.V. North; Evaluation of Algorithms for the Automated Registration of Scanning Laser Tomographic and Photographic Images . Invest. Ophthalmol. Vis. Sci. 2003;44(13):3398.

      Download citation file:


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

      ×
  • Supplements
Abstract

Abstract: : Purpose:To evaluate methods for the alignment of photographic and tomographic images to facilitate the quantitative analysis of optic disc images in the detection of glaucomatous optic neuropathy. Methods:Tomographic images were obtained using the Heidelberg Retina Tomograph I (Software 2.01) with a 10 degree field of view. The extended focus image was derived from the raw HRT data and aligned to the grey scale fundus image from the Nidek 3Dx (right image digitised at high resolution). 27 images were selected from our imaging archive that provided a representative sample of optic disc appearances from normal and glaucomatous eyes. 24 alignment algorithms were evaluated, based on variations of the following: (i) mutual information (MI) method (ii) feature based registration (iii) feature based and MI combined (iv) Variance weighted MI (v) Windowed MI (in which the image was subdivided into 16 equal sized windows). Magnification characteristics of the two image sets were not included in the alignment process. Monochromatic images were graded for alignment using a 4 point scale (0 none-poor, 1 moderate, 2 good, 3 excellent) by a clinician who was expert in the analysis of the optic disc. No user input was required to control the alignment process. Results:The highest registration score was 2.39 (CI 0.29) indicating good to excellent alignment. The registrations scores (CI) for the best 5 algorithms are given in table 1. Both scaling, rotation and translation displacements of the images were performed to achieve these alignments. Conclusions: The windowed based Mutual Information approach with requantisation of the grey scale can provide clinically acceptable levels of image alignment. These alignment techniques should facilitate the fusion of tomographic and photographic data to optimise the detection of glaucomatous optic neuropathy. Algorithm alignment scores: means(CI)  

Keywords: optic disc • imaging/image analysis: clinical • imaging methods (CT, FA, ICG, MRI, OCT, RTA, S 
×
×

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.

×