Comparison of ranked segment analysis (RSA) and cup to disc ratio in computer‐assisted optic disc evaluation

Abstract

ABSTRACT.Purpose: Ranked segment analysis is a new method for evaluation of optic nerve head topography with the Heidelberg Retina Tomograph. This analysis ranks sector measurements around the optic disc and compares these measurements with rank specific significance limits. We evaluated the diagnostic precision of the ranked segment analysis in a large clinical sample and compared with vertical cup/disc ratio measurements. Vertical cup/disc ratio measurements were not corrected for disc size. Methods: We analysed optic disc images from 153 normal and 75 glaucomatous eyes using the Heidelberg Retina Tomograph (software version 2.01). Ranked segment analyses and vertical cup/disc ratio measurements were obtained from each optic disc image, and ROC curves were plotted. Results: At the 95% specificity level, sensitivity was significantly lower for ranked segment analysis (70%) compared to vertical cup/disc ratio measurements (85%). Almost 30% of the discs in the glaucoma group were classified as normal using the ranked segment analysis. Discussion: Ranked segment analysis yielded unacceptably poor discrimination, in fact, worse than cup/disc ratio with its known clinical limitations. Glaucomatous disc damage typically occurs at the vertical poles of the disc. High rank measurements (low sector values) in normal eyes, on the other hand, are more common in the temporal disc sectors. Important spatial information is lost during the ranking procedure. This may partly explain the low sensitivity of the ranked segment analysis observed in our study. The results raise serious concerns regarding the clinical usefulness of ranked segment analysis.Key words: ranked segment analysis -glaucoma diagnosis -diagnostic precision -optic nerve head -image analysis -sensitivity and specificity.

Regression analysis of ranked segment parameters for optic nerve head classification: a pilot study

“…These data, and those from other studies (Gundersen and Åsman, 2000) demonstrate that the conventional RSA (Bartz‐Schmidt et al. , 1996; Malinovsky, 1996) yields a high number of false‐negatives (i.e.…”

“…The RSA and RARSP techniques did not appear to be effective for volumetric measures of the ONH. These data, and those from other studies (Gundersen and Å sman, 2000) demonstrate that the conventional RSA (Bartz-Schmidt et al, 1996;Malinovsky, 1996) yields a high number of false-negatives (i.e. glaucoma classified as ÔnormalÕ), which may suggest that the deviation from normal curves, which is required to register abnormality, is not set at an appropriate level, or that the technique is not sensitive enough owing to the overlap of normal and glaucomatous values.…”

“…In conventional RSA, spatial information about the ONH is lost as a result of the mathematical ranking procedure. Superior and inferior notches in the neuroretinal rim associated with glaucoma may escape detection by this method, as they may not reach significance levels (Gundersen and Åsman, 2000). The RARSP technique appears to enhance the ability of the technique to make a diagnosis of glaucoma ().…”

Ophthalmological evaluation in thyroid‐associated ophthalmopathy

“…Some of these patients may therefore represent cases of thyroid‐related optic neuropathy that eventually normalized, whereas the study by Kalman & Mourits (1998) used the cup‐to‐disc ratio as the optic disc criterion for glaucoma. The paramount physiological variability of optic disc appearance (Bengtsson 1976) hampers discrimination between normal and glaucomatous discs using the cup‐to‐disc ratio (Gundersen & Åsman 2000). There is, therefore, as yet, no firm evidence that TAO is a risk factor for glaucoma.…”