Abstract
Abstract: :
Purpose: Prospective study comparing evaluation of mydriatic, non–stereo digital colour fundus photographs against clinical examination and fluorescein angiography in identifying and classifying exudative AMD. Methods: Digital colour fundus photos were obtained from patients seen between September 2001 and June 2002. Gold standard assessments for each eye were extracted from patient charts, including review of fluorescein angiograms. Colour fundus photos for each eligible patient were separated by eye, catalogued and randomly labeled before interpretation. Exact agreement, sensitivity, specificity, positive predictive value and negative predictive value of the fundus photos in diagnosing, classifying and managing cases of suspected exudative AMD were then calculated using the gold standard as the comparison. Results: A final total of 223 images were used for this study from 118 eligible patients. Exact agreement between gold standard and photo reading ranged from 89.2% (presence of PED), 87.4% (presence of CNVM) and 82.5% (evidence of RPE geographic atrophy). Against the gold standard, digital photo assessments of CNVM location had an exact agreement of 65.5%, with classic or predominantly classic membranes in higher exact agreement than eyes with occult or predominantly occult membranes (58.7% and 34.7%, respectively). The sensitivities to diagnostic parameters ranged from 89.2% (presence of CNVM) to 40.0% (presence of PED). Specificities ranged from 99.1% (extrafoveal CNVM) to 86.8% (presence of RPE geographic atrophy). Positive predictive value ranged from 88.6% (location of juxta/subfoveal CNVM) to 40.0% (presence of PED). Negative predictive value ranged from 98.2% (location of extrafoveal CNVM) to 82.4% (location of juxta/subfoveal CNVM). When comparing clinical recommendations based on digital photo assessment against the gold standard, exact agreement was 80.3%. As an overall screening tool, sensitivity, specificity, positive predictive value, and negative predictive value of digital fundus photos in screening for true and high risk wet AMD was 82.1%, 79.1%, 70.4%, and 88.0% respectively. When only treatable lesions are taken into consideration, the screening sensitivity and negative predictive value increase to 88.4% and 93.6%, respectively. Conclusions: Digital non–stereo colour fundus photographs are highly sensitive and have high negative predictive value as a screening tool. Very few treatable lesions are missed using telemedicine in age–related macular degeneration.
Keywords: age–related macular degeneration • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound)