96th DOG Annual Meeting, 1998



C. Bräuer-Burchardt, W. Vilser, T. Riemer

The RVA is a new automatic measuring system for recording and analysing vessel diameter in dependence on time and position along the vessels. It is a powerful tool to study retinal branch vessel behaviour in online measurements. The aim of this study was the determination of coefficients of variation (CV) as a measure for the reproducibility in dependence on different measuring conditions and kind of examination.

Methods: 136 measuring sequences were made by the RVA at 43 retinal branch vessels of 19 healthy volunteers with a mean age of 30.3 ± 7.1 years. The measurements were made at the same position of a 1 to 2 mm long vessel section under same measuring conditions (measuring length and time). The coefficient of variation was determined within the online measuring sequences (CVM) and between measuring sequences of different sessions (CVS). The time between the sessions which were independent concerning the adjustment of device and eye was about 45 minutes. The measuring time of the sequences was 3 minutes to exclude known biological influences like vasomotions and Meyer-waves.

Results: The mean CVM was 6.8 % (n=20). Thus the random error of the vessel diameter of one measuring sequence becomes less than 1 % in the case of 25 measuring values per second and measuring time longer than 10 seconds. The mean CVS value representing the difference between the two sessions was 1.8% (n=26).

Conclusions: The results show the big influence of the biologically determined local variation of the vessel diameter on the variation of the measuring values. Regarding a single local measuring position of about 0.2 mm length over the vessel section a mean CVM value of 2.7 % was obtained (n=16). That means that with the RVA also local differences of the vessel diameter values smaller than 1% can be detected in sequences with a measuring time longer than 10 seconds. The detection limit of alterations between different sessions is determined by the CVS. The whole random error is negligible in long measuring times. The interpretation of alterations of the vessel diameter from session to session can be made only if the systematic error represented by the CVS is taken into account.

Supported by TMWFK: B303-96144

Technical University Ilmenau, BMTI, PF 100565, D-98684 Ilmenau