THE EFFECT OF CONTOUR-LINE DRAWING CRITERIA ON OPTIC DISC PARAMETERS AS MEASURED WITH THE HEIDELBERG RETINA TOMOGRAPH*

F.A. ENNIS, C. BLACHLY, G.A. CIOFFI, S.L. MANSBERGER and C.A. JOHNSON

Discoveries in Sight, Devers Eye Institute, Portland, OR, USA

Abstract

Purpose

Numerous studies have examined the reliability and reproducibility of measurements using the Heidelberg Retina Tomograph (HRT). However, the effect of criteria used in defining the optic disc margin and placement of the contour line, has not been thoroughly assessed. This is clinically relevant, since measurements from the HRT that are contour-line dependent are used in the final analysis in discriminating glaucomatous from non-glaucomatous subjects. Therefore, the purpose of this study was to examine the effect of contour-line placement upon HRT measurements.

Methods

Contour lines were drawn by an experienced technician on 105 HRT (Version 1) mean topography images of 105 glaucomatous eyes. Optic disc parameters were then measured via the HRT software. The contour lines were erased and then later redrawn using a second set of optic disc margin criteria. HRT disc parameters were then re-measured and compared with the previous results.

Results

Correlations between the two contour-line criteria were statistically significant for 13 of the 14 HRT parameters assessed, although some correlation coefficients were as low as 0.31. However, the third moment showed no significant correlation (r = 0.032; p = 0.75) for the two sets of criteria. Furthermore, comparisons of the ranked position of the height variation contour, third moment, and rim volume above reference plane, showed poor agreement. As these three parameters are utilized in the Mikelberg equation, which the HRT uses for classifying results as normal or abnormal, these classifications also showed poor agreement between the sets of optic disc margin criteria.

Discussion

The relatively low correlation coefficients for many of the HRT parameters and the poor agreement for the Mikelberg classifications implies that caution must be exercised in interpreting HRT data if the specific

*This study will appear in full elsewhere.

Address for correspondence: F.A. Ennis, MD, Discoveries in Sight, 1225 NE Second Avenue, PO Box 3950, Portland, OR 97208-3950, USA

Perimetry Update 2002/2003, pp. 329–330

Proceedings of the XVth International Perimetric Society Meeting, Stratford-upon-Avon, England, June 26–29, 2002

edited by David B. Henson and Michael Wall

© 2004 Kugler Publications, The Hague, The Netherlands

contour-line drawing criteria are not known. Furthermore, diagnostic criteria that use the third moment in particular, should not be used on HRT images where the contour line has been drawn using criteria other than those used in the development of the diagnostic criteria in question. A consensus on methods for drawing contour lines is necessary, and a detailed description of the method used should be included in all reports that utilize HRT data.

EVALUATION OF EFFECTIVENESS OF NEW GDx

PARAMETERS

KUNIO YAMADA,1 MASAHIRO OSAKO,1 KAZUYA TACHIBANA,1 TADASHI OKANO1 and MASAHIKO USUI2

1Department of Ophthalmology, Kasumigaura Hospital, Tokyo Medical University;

2Department of Ophthalmology, Tokyo Medical University; Tokyo, Japan

Abstract

Purpose

The authors evaluated the effectiveness of new GDx parameters for the diagnosis of glaucoma by comparing them with the results of conventional parameters.

Subjects and methods

Thirty-three eyes of 33 glaucoma patients (primary open-angle, 12 eyes; normal tension, 21 eyes) and 33 eyes of 33 normal persons were studied (mean age, 59.4 years). Three consecutive measurements of the peripapillary retinal nerve fiber layer (RNFL) were performed by a single operator using the nerve fiber analyzer (GDx), and the mean values of three images were calculated. The new GDx parameters used for evaluation were ellipse standard deviation (ESD), normalized superior ratio (NSR), normalized inferior ratio (NIR), discriminant analysis (LDF), and sector analysis (SA). For comparison, fundus photography and the 24-2 program of the Humphrey Field Analyzer (HFA) were performed on all subjects.

Results

All the new GDx parameters had better sensitivity and specificity compared to the conventional parameters except for the Number. The sensitivity of LDF was 97% and was the most sensitive of all the parameters studied, but the specificity of LDF was the lowest of all the parameters. For SA, if glaucoma was defined as having more than one abnormal sector, then the sensitivity and specificity were 55 and 82%, respectively. The agreement between SA and RNFL defect on fundus photography was higher in the inferior part of the disc than in the superior part, and the correlation was significant in the inferior part (p < 0.05). Significant correlations were observed between new GDx parameters and visual field indices (MD, CPSD) of the HFA. In the comparison between early (MD > -5 dB) and moderate or advanced (MD < -5 dB) glaucoma cases, LDF and the Number showed less differences in sensitivity than the other parameters.

Conclusions

Compared to conventional parameters, the new GDx parameters were effective for the diagnosis of glaucoma. We need to review the standard value for LDF by examining more subjects.

Address for correspondence: Masahiro Osako, MD, Department of Ophthalmology, Kasumigaura Hospital, Tokyo Medical University, 3-20-1, Amimachi-chuo, Inashiki-gun, Ibaraki 300-0332, Japan. Email: cd9m- osk@asahi-net.or.jp

Perimetry Update 2002/2003, p. 331

Proceedings of the XVth International Perimetric Society Meeting, Stratford-upon-Avon, England, June 26–29, 2002

edited by David B. Henson and Michael Wall

© 2004 Kugler Publications, The Hague, The Netherlands