Abstract
Purpose: :
To develop image registration software that can be applied to a sequence of conventional fluorescein angiographic images and to analyse the early and late phases of the angiogram separately.Furthermore, to reliably represent the temporal dynamics of fluorescence change both qualitatively and quantatatively.
Methods: :
Conventional digital angiograms were exported in JPEG format. Image processing software was written in MATLAB v 6.0 to perform translational image alignment. The temporal change in pixel value was examined across the stack of images and automatically separated into two phases, early and late. The linear gradient of change of fluorescence value for the two phases was calculated. Parametric images of the two gradients were generated. Thus by separately analysing the gradients in the first and second phase of the angiogram, it was possible to produce a composite image of overlaying areas of hypo and hyperfluorescence in the second phase over the first phase gradient image. Further analysis of second phase gradients allowed the generation of 5 centile groups of minimum to maximum rates of fluorescence in the second phase. Angiograms of 15 patients with Age Related Macular Degeneration were analysed.
Results: :
For 15 angiograms of patients with age related macular degeneration, the gradients of increase in the first phase had average values ranging from 2.1 pixels/sec (standard deviation 1.81) for the lowest gradients (0–20%) to 9.35 pixels /sec (sd 6.06) for the highest areas (81–100%). For second phase gradients, the average values ranged from 0.02 pixels / sec (sd 0.05) for the lowest gradients (0–20%) to 0.7 (sd1.64) for the highest gradients. The gradient of increase in fluorescence value for the first phase of the angiogram for pathological areas of the fundus was similar to non–diseased areas in all angiograms assessed. The second phase gradient allowed for easy distinction between areas of increasing fluorescence and areas where fluorescence was steady or decreasing.
Conclusions: :
Biphasic analysis allows for a semi–quantitative assessment of a fluorescein angiogram and the quantitative identification of retinal pathology. This technique allows for the quantative assessment and identification of pathological age related macular degenerative changes including areas of early and late leakage as well as identifying specific areas with different rates of leakage.
Keywords: imaging/image analysis: clinical • retina • image processing