An Evaluation of the Test-retestvariability of the HRT-II and the influence of Degree of Lenticular Opacity

Abstract

Abstract: : Purpose: to study the effect of observer-dependent image acquisition, cataract, image quality and stage of glaucoma on the repeatability of Heidelberg Retina Tomograph II (HRT II) image parameters. Methods: 108 eyes (44 glaucoma and 64 ocular hypertensive) with previous HRT imaging experience were recruited with a proportion awaiting cataract surgery. Snellen visual acuity, auto-refraction and auto-keratometry were recorded. Eyes were imaged by two experienced observers in the sequence Observer 1, Observer 2 and then Observer 1 again. Images were analysed using Eye Explorer version 1.3.5.0. with a single contour line exported from the first image to the other two. Rim area (RA) was recorded and the mean standard deviation pixel height (MSD) was used to measure image quality. Following imaging pupils were dilated and LOCS III scoring was carried out by a single observer. Scheimpflug images were acquired and analysed using the Marcher Case 2000 series, recording central nuclear dip (CND). Correlations were sought by linear regression between RA test-retest differences and LOCS III scores, CND and image MSD for intra- and inter-observer image acquisition. Bland-Altman plots were constructed to assess whether test-retest RA repeatability depended on observer or RA. Results: Mean RA was 1.19 ± 0.68mm² (± 95% confidence interval, range 0.45 to 2.21mm²). The mean inter- and intra-observer RA differences were 0.03 ± 0.19mm² and 0.01 ± 0.21 mm², respectively. There was no relation between test-retest RA differences and size of RA. Inter- and intra-observer RA differences were correlated with image MSD (r² = 0.20 and 0.42, respectively). If an image quality cut-off of MSD 30µm is applied, the 95% confidence intervals for inter- and intra-observer RA were 0.11 mm² and 0.12 mm², respectively. Inter-observer RA test-retest difference was weakly correlated with nuclear opalescence (NO) and nuclear colour (NC) (r² = 0.12 and 0.12 respectively), but not CND. Intra-observer RA test-retest difference was not correlated with NO, NC or CND. Image MSD was correlated with NO (r² = 0.34), NC (r² = 0.12) and CND (r² = 0.30). Conclusions: There is good repeatability of RA using the HRT II and this is not affected by observer image acquisition or degree of glaucomatous optic disc damage. Repeatability is affected by both image quality and subjective cataract score. Image quality is affected by the degree of nuclear sclerosis (subjective and objective measures). RA repeatability improves markedly if poorer quality images (MSD > 30µm) are excluded. These data are clinically useful when setting thresholds for progression.

Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, S • optic disc • cataract