Purchase this article with an account.
M.E. Verdugo, P. Soliz, S. Wolf, B. Wyman; Towards Development of Quantitative Techniques for Analyizing Ophthalmological Images . Invest. Ophthalmol. Vis. Sci. 2005;46(13):246.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: In the field of radiological imaging moving from subjective to objective interpretation has been facilitated by image processing techniques. The purpose of this study is to investigate the potential for quantitative analysis applied to age related macular degeneration (ARMD) images. Methods: For 60 subjects in a natural progression study of ARMD the following imaging modalities were acquired: fluorescein angiography (FA), indocyanine green angiography, RGB, infrared, red–free, and auto–fluorescence. All images were registered to a common FA base. One feature, the fluorescein uptake rate, was modeled by an exponential fit of each pixel’s intensity through time. The feature calculations led to the discovery of the limitations of the equipment and acquisition techniques. Results: On the average registration of the FA data was within 2 pixels, though improvements in deformable registration to account for retina curvature would greatly enhance the ability to work with this rich data. In the FA study the 8–bit camera saturated at pixel values of 255. To compensate, the operator adjusted the gain causing uptake curve discontinuities. This will need to be solved at the instrumentation level by either moving to a camera with greater dynamic range (12 bits) or recording the gain for post processing correction. Also, during acquisition the imaging was started after the fluorescein injection resulting in no baseline for the intensity curves. Another hindrance in using multimodality analytical techniques was the lack of image standardization. Image features such as intensity could not be compared between times because the acquisition was acquired with different gain settings. Use of calibration standards, or detailed acquisition protocols should be used to ensure standardized acquisition. Conclusions: Application of image processing has great potential in the area of ophthalmology. However, a few developments are required to reach the full potential. First, modifications in the instrumentation need to be made to prevent saturation. Next, acquisition must be carefully standardized. Finally, some issues such as registration can be corrected by post–processing.
This PDF is available to Subscribers Only