Abstract

The accuracy and reproducibility of the optic nerve head analyser (ONHA) and the Heidelberg retina tomograph (HRT) were compared and the performance of the HRT in measuring fundus elevations was evaluated. The coefficient of variation of three repeated measurements in a model eye and in volunteers and the relative error in a model eye was calculated. With ONHA measurements the pooled coefficient of variation in volunteers was 9.3% in measuring cup areas and 8.4% in measuring the cup volume. In a model eye the pooled coefficient of variation was 7.6% for the parameter 'cup area' and 9.9% for the parameter 'cup volume'. The pooled relative error in the model eye was 6.6% for the parameter 'cup area' and 5.1% for the parameter 'cup volume'. With HRT measurements in volunteers the pooled coefficient of variation of both the parameters 'volume below contour' and 'volume below surface' was 6.9%. In the model eye the pooled coefficient of variation was 2.4% for the 'volume below contour' and 4.1% for the parameter 'volume below surface'. The pooled relative error in the model eye was 11.3% for the 'volume below contour' and 11% for the 'volume below surface'. The pooled relative error in measuring retinal elevations in the model eye was 3.8%. The coefficient of variation was 3.5%. The accuracies of the HRT and ONHA were similar. However, as the ONHA 'cup volume' is unreliable in patients because of the design of the ONHA whereas the HRT volume parameters are reliable it seems reasonable to assume that the HRT is superior to the ONHA. Only the HRT is capable of quantifying retinal elevations.