Improving the repeatability of Heidelberg retina tomograph and Heidelberg retina tomograph II rim area measurements.

Abstract

To devise and test strategies for improving Heidelberg retina tomograph (HRT and HRT-II) rim area (RA) repeatability and assess the benefit of the strategies in time series of HRT images. The effect of the standard and 320 microm reference planes and image quality on RA repeatability was assessed in a test-retest HRT image dataset from 74 subjects. A longitudinal HRT image dataset from 30 ocular hypertensive subjects was analysed by linear regression of RA over time, with each of the reference planes and using a manual image alignment facility. RA variability was estimated by comparing the standard deviation of residuals (RSD) generated by each linear regression. RA repeatability was better with the 320 microm reference plane (repeatability coefficient 0.17 mm(2)), improving further with only good quality images (repeatability coefficient 0.08 mm(2)). For the longitudinal data, a significant (p<0.0001) reduction in the RSD from 0.10 to 0.05 mm(2) was obtained with the 320 microm reference plane. Manual alignment led to a further significant (p<0.0001) reduction in the RSD to 0.04 mm(2). The findings support the use of a 320 microm reference plane and manual image alignment to analyse RA over time. The estimates of RA repeatability may be used to define thresholds for glaucomatous change.