Intraimage reproducibility of measurements in the macular area using a computerized system.

An accurate analysis of the morphological changes which take place during pathological processes of the posterior pole is important for a correct diagnosis and therapeutic approach. The purpose of the study was to determine the intraobserver and interobserver reproducibility of the Image-net system 100 (Topcon, Japan) to take measurements on the retina. The program 'Linear/Areal Measurement functions' of Image-net system 100 which is an image digitalization technique, was tested. Twelve patients were consecutively selected from the patients of the Retina Center of the Department of Ophthalmology, University of Genoa. Three images of each eye were taken from each subject and only the best image was used in this study. The intraobserver and interobserver reproducibility of both the distance between two pre-set points (linear measurement), and the perimeter and area of preselected retinal zones were calculated. The repeatability (or intraobserver reproducibility) of the linear sizes was measured by the coefficient of variation and ranged from 0.32% to 7.38%, while the interobserver reproducibility ranged from 0.46% to 5.22%. The repeatability and reproducibility of the perimeters ranged from 0.72% to 9.63% and from 0.6% to 5.7%, respectively, while the repeatability and reproducibility of the areas ranged from 0.72% to 9.63% and from 0.6% to 5.7%, respectively. Although the results were quite good, the quality of the image of the fundus and the number of observers influenced the coefficient of variation; furthermore, the anatomy of the areas to be measured and the computer 'mouse' could increase the value of the coefficient of variation.