Purchase this article with an account.
D.J. Covert, D.P. Han, J.E. Kim, W.J. Wirostko, T.B. Connor, S.J. Moon, R. Hamilton; Cognitive Variables in the Fluorescein Angiographic Interpretation of Evolving Predominantly Classic Choroidal Neovascular Membranes in Age–Related Macular Degeneration: Correlation of Digital Measurements With Physician Assessment of Lesion Size Changes and Presumed Clinical Significance . Invest. Ophthalmol. Vis. Sci. 2005;46(13):1588.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: To quantify variables in the interpretation of fluorescein angiograms (FA) of evolving predominantly classic choroidal neovascular membranes (CNVM) in age–related macular degeneration (ARMD). This quantification may be useful in setting parameters for studies involving physician recognition of CNVM growth or regression. Aims: 1. To quantify the smallest change in CNVM surface area (SA) and greatest linear dimension (GLD) that was visually detectable by retina physicians; 2. To estimate the smallest change in CNVM SA and GLD presumed to be clinically significant by physicians; 3. To describe the degree of correlation between clinicians' estimates and software estimates. Methods: Using conventional enlargement techniques, 48 FAs were used to define 29 FA pairs. Imaging software was used to measure the SA and GLD of the CNVMs. Retina specialists estimated the change in lesion SA and GLD for each pair and stated whether they believed it to be clinically significant (potentially affecting outcome). Subsequently, they viewed the same pairs with computer–assisted demarcation of the lesions' borders and GLD positioning and reanalyzed them. Results: Aim 1: For lesions that increased in size, the smallest changes consistently detected by the physicians as a group were a 5% increase in SA and a 25% increase in GLD; for lesions that decreased in size, they were a 15% decrease in SA and a 5% decrease in GLD. Aim 2: For lesions that increased in size, a 25% increase in SA and a 25% increase in GLD were thought to be clinically significant; for lesions that diminished in size, a 15% decrease in SA and a 10% decrease of GLD. Aim 3: Linear regression demonstrated a correlation between physicians' and software estimates of change in SA and GLD with r2 values of 0.38 and 0.75, respectively (p<0.001). These values improved to 0.88 and 0.93 when the lesions' boundaries and GLD were pre–drawn for the physicians (p<0.001). Conclusions: Cognitive variables influenced physician assessment of CNVM growth. Computer–assisted demarcation of lesion SA and GLD reduced variability in physicians' estimates of lesion size change and improved correlation with digital measurements. The relationship between physician presumption of clinical significance and software estimates of SA and GLD may be useful in future ARMD treatment studies.
This PDF is available to Subscribers Only