Purchase this article with an account.
J. Einighammer, T. Oltrup, T. Bende, B. Jean; A Method for Analysing Corneal Topography Data . Invest. Ophthalmol. Vis. Sci. 2005;46(13):845.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose:A method is presented to statistically analyse measurements of corneal topography devices regarding trueness and reproducibility in approximation to ANSI Z80.23 draft. Methods: All calculations are based on difference maps of topographies. These differences are weighted to get a uniform sampling distribution. For trueness testing, differences of a series of measurements against a known reference surface are taken. Trueness is quantified in terms of offset through the mean of mean difference (MOM), the smaller the better. For reproducibility testing, difference maps are calculated from corresponding measurements. Reproducibility is expressed by standard deviation of the mean difference (SDM) and mean standard deviation of differences (MSD), the smaller the better. All values are calculated for different zones (central c=0–3mm, middle m=3–5mm, outer o=5–7mm, total t=0–7mm) and indexed (e.g. cMSD for central). Axial refractive power maps of test objects in three categories were measured on two systems (A: Keratron Scout (R) and B: Technomed C–Scan (R)): calibration spheres (42,2 & 48,2 D), test spheres (multifocal, simulated ast, simulated central keratoconus) and 4 human corneas. Results: Both systems show small offset thus good trueness when measuring the calibration spheres (A: tMOM=0.10/0.06; B: tMOM=–0.02/0.07). For calibration/test spheres the SDM and MSD are smaller for system A than for B (A: tSDM=0.06/0.17, tMSD=0.05/0.06; B: tSDM=0.26/0.28, tMSD=0.12/0.21). For human corneas the SDM of system A is higher than for B, while the MSD is slightly smaller (A: tSDM=0.40, tMSD=0.19; B: tSDM=0.23, tMSD=0.23). Means: system A’s reproducibility is better for calibration and test spheres than for human corneas. System B’s reproducibility for human corneas is better than A’s, but much worse for calibration and test spheres. On both systems the MSD decreases going to the periphery, while the SDM increases (e.g. human corneas. A: cSDM=0.31, mSDM=0.40, oSDM=0.44; B: cSDM=0.20, mSDM=0.22, oSDM=0.25). Conclusions: The proposed method allows an analysis of topography data for quantification of trueness and reproducibility. The values MOM, SDM and MSD can be useful for analysing and comparing the output of different corneal topography devices or for analysing the behaviour of a specific corneal topography device when measuring different test objects including human corneas. When doing trueness testing for known test objects other than calibration spheres, with more complex surface features, this method can likewise be used to assess a system’s resolution.
This PDF is available to Subscribers Only