Scheimpflug tomographers: Are they precise?
Comparative data demonstrates good agreement between devices
Corneal tomography based on optic technology, Scheimpflug and OCT, has replaced Placido topography more recently. The awareness of the importance of posterior surface and thickness topography has boosted this change.
Regarding Scheimpflug devices, two main hardware configurations compete in this arena: OnlyScheimpflug instruments like Pentacam HR (Oculus, Wetzlar, Germany) and Scheimpflug-Placido machines such as Galilei G2/G4 (Ziemer, Port, Switzerland), Sirius (CSO, Florence, Italy) and TMS-5 (Tomey, Nagoya, Japan).
The need for Placido curvature data comes from the extreme demand for precision in elevation measurements in the centre of the cornea to avoid significant curvature errors. Most Companies defend that Placido information is necessary to improve accuracy and precision. However, the Pentacam HR has maintained its only-Sheimpflug configuration over the years (first marketed in 2001) relying on its eyetracking and reconstruction algorithms. Whereas, the Galilei claims a higher precision in thickness measurement as a result of its two opposite rotating cameras, self-compensating for any decentration error.
There is no gold-standard in corneal curvature measurement to test the accuracy of any device in a certain cornea. So, accepting all of them are accurate as they can correctly measure test eyes, precision and agreement testing is the way to analyse and compare these instruments. Precision will affect the usefulness of a topographer in issues like determining keratoconus progression, IOL power calculations, calculating customized ablation patterns and so on.
We have performed a prospective randomized comparison of the Galilei G2 (software 5.2.1) and the Pentacam HR (software 1.7) by means of 3 consecutive measurements by 2 expert explorers on 35 eyes of 35 healthy subjects. Intraobserver variability defines repeatability and interobserver variability defines reproducibility of the measument. The basic statistic is the within-subject standard deviation (Sw) that allows calculation of the error Gaussian bell.
