Purchase this article with an account.
T. Smith, M. Busuioc, N. Lee, A. F. Laine, S. Schmitz-Valckenberg, F. G. Holz; Autofluorescence and Geographic Atrophy. Invest. Ophthalmol. Vis. Sci. 2007;48(13):5114.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To determine the impact of focally increased autofluorescence (FIAF) in the development of new geographic atrophy (GA) as a function of total FIAF area and as a focal predictor.
Ten serial pairs, spaced 2 to 3 years apart, of AF images (6000 micron region) of eyes of 10 patients with GA were analyzed as follows: the later image was registered to the earlier image in Matlab so that corresponding retinal points could be precisely overlaid in Photoshop. The GA in each image was segmented by the level set automated method presented previously (Lee, Smith, et al. ARVO 2006). The FIAF in the original image was segmented by first leveling the AF image background by a mathematical model previously published and then selecting pixels whose brightness was at least 1.5 image standard deviations greater than the mean (Huang, Smith, et al. IOVS 2006). The boundary zones tested around the original area of GA were defined as 250 and 500 microns. The relative positive and negative predictive values (PPV and NPV) of focal FIAF for development of new GA in the boundary zones were calculated. The total area of FIAF in the total 6000 micron region and percent areas of FIAF in the two boundary zones were correlated with the measured rate of increase in total GA area.
The strongest focal predictive indicator of new GA was the relative PPV in the 500 micron boundary zone (1.3 ± 0.36, relative to the predictive value of chance (1.0)). The other PPV and NPV’s were close to 1.0. The strongest correlation of FIAF area with GA progression rate was the area of FIAF as a fraction of the 500 micron boundary zone. If y was the annual rate of progression of GA area as a fraction of the 6000 micron circle and x was the fraction of FIAF in the 500 micron zone, the linear regression was y = 0.44x + 0.0 (slope = 0.44 ± 0.15, p = 0.02). Other correlations did not reach statistical significance.
The results of digital image analysis suggest that as a focal predictor, FIAF has only a modest PPV for focal new GA development in the 500 micron border zone. However, the total area of FIAF in this same zone is a significant predictor of the total rate of GA progression. It appears that the level of disease activity, as reflected by area of FIAF in the border zone, is correlated with GA progression.
This PDF is available to Subscribers Only