A study of variance in densitometry of retinal nerve fiber layer photographs in normals and glaucoma suspects.

The main object of this research was to develop a reliable method of screening glaucoma suspects and patients for early loss of or changes in the retinal nerve fiber layer (RNFL). This study quantifies the variances due to photography, digitizing, and analysis of red-free photographs of the RNFL. The influence of pupil size, optic disc position and eye movements, film processing, digitizing, and intra- and interphotographic-session and intra- and interoperator variances were established. It was found that pupils needed to be dilated to at least 6 mm, that the optic disc had to be positioned in a standardized area in the negative, that the head of the subject had to remain still during photography, and that film processing and digitizing of the negative needed to be strictly controlled to minimize the variance in collection of densitometry data from RNFL red-free photographs. It was established that focusing of the negatives during digitization was not crucial. Criteria were defined for acceptable negatives. Interphotographic-session and intraoperator variances were not significant in most cases when negatives were digitized to these criteria. Analysis of interphotographic-session variance showed that there were still some factors in photography, film processing, and/or image digitizing that were not sufficiently controlled for long-term follow-up without normalization of the data. Densitometry data gathered using the established protocol, from negatives of 71 subjects were analyzed; best sensitivity and specificity rates of 80% and 100%, respectively, were achieved for the diagnosis of glaucoma.