May 2006
Volume 47, Issue 13
ARVO Annual Meeting Abstract  |   May 2006
Image Registration and Supervised Automatic Drusen Segmentation for Use in Clinical Studies
Author Affiliations & Notes
  • M. Busuioc
    Ophthalmology, Columbia University, New, NY
  • J. Koniarek
    Ophthalmology, Columbia University, New, NY
  • J. Chan
    Ophthalmology, Columbia University, New, NY
  • N. Lee
    Ophthalmology, Columbia University, New, NY
  • S. Du
    Ophthalmology, Columbia University, New, NY
  • R.T. Smith
    Ophthalmology, Columbia University, New, NY
  • Footnotes
    Commercial Relationships  M. Busuioc, None; J. Koniarek, None; J. Chan, None; N. Lee, None; S. Du, None; R.T. Smith, None.
  • Footnotes
    Support  New York Community Trust, NEI R01 EY015520 HIGHWIRE EXLINK_ID="47:5:5707:1" VALUE="EY015520" TYPEGUESS="GEN" /HIGHWIRE –01, Research to Prevent Blindness
Investigative Ophthalmology & Visual Science May 2006, Vol.47, 5707. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      M. Busuioc, J. Koniarek, J. Chan, N. Lee, S. Du, R.T. Smith; Image Registration and Supervised Automatic Drusen Segmentation for Use in Clinical Studies . Invest. Ophthalmol. Vis. Sci. 2006;47(13):5707.

      Download citation file:

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

  • Supplements

To demonstrate two user interface driven Matlab software modules for precise, objective measurement of progression or regression of drusen in clinical trials.


Retrospective demonstration data were acquired from a pilot study: 16 digital images before and after indirect laser treatment (temporal macula) of eight eyes with macular soft drusen. We developed two interfaces, one for registration (alignment of images) and another for automatic drusen segmentation (previously validated in [1] by Smith & al., "Automated detection of macular drusen using geometric background leveling and threshold selection", Archives of Ophthalmology 123(2): 200–6). The core of the latter is a mathematical model that levels macular background reflectance for global thresholding of drusen. We aligned the image pairs with the first Matlab program. The user interface asks for manual selection of at least 4 pairs of points, normally at vascular landmarks, after which the alignment proceeds automatically. Drusen in the registered image pairs were then segmented by the algorithm in [1] driven by the second interface. The user selects three options: FewDrusenInner (Y/N) for the 3000 micron diameter center region, FewDrusenOuter (Y/N) for the 3000–6000 micron annulus, and SoftIndistinctDrusen (Y/N), after which segmentation is automatic. Color coded results are displayed as residual, resorbed and newly formed drusen to study drusen remodeling morphologically. Drusen area measurement data is automatically exported in Excel and automatic statistics are performed.


Measurements of total, residual, resorbed and new drusen that are difficult for human grader estimation were performed and displayed morphologically in easily understood format (see Fig. A–original image and Fig. B –segmented drusen, in green). For the 8 images in our test, the system solved the T test with P (T<=t) one–tail = 0.00099, very similar to human graders.


With limited human oversight these programs show promise in expediting the grading process whenever accurate drusen data is needed for AMD trials.  

Keywords: age-related macular degeneration • computational modeling 

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.