Related Experiment Video
Updated: Aug 6, 2026
Full-Field Optical Coherence Microscopy for Histology-Like Analysis of Stromal Features in Corneal Grafts
Published on: October 21, 2022
Determining the Most Suitable Tomography-Based Parameters to Describe Progression in Keratoconus. The Retrospective
Marta Jiménez-García1, Elke O Kreps, Sorcha Ní Dhubhghaill
1Department of Ophthalmology (M.J.-G., S.N.D., C.K., J.J.R.), Antwerp University Hospital (UZA), Edegem, Belgium ; Department of Medicine and Health Sciences (M.J.-G., E.O.K., S.N.D., C.K., J.J.R.), University of Antwerp, Antwerp, Belgium ; Department of Ophthalmology (E.O.K.), Ghent University Hospital, Ghent, Belgium ; and Department of Medicine and Health Sciences (E.O.K.), University of Ghent, Ghent, Belgium .
Objectives:
To identify the most suitable parameters to describe keratoconus progression.
Methods:
Longitudinal retrospective cohort study. Monotonicity and consistency of over 250 parameters extracted from the Pentacam tomographies (Oculus, Germany) of 743 patients was analyzed. Repeatability was calculated for 69 patients (published elsewhere). The parameters were scored based on their performance for each desired feature and a global ranking was created.
Results:
Overall, parameters that average a higher number of corneal points performed better than single-point parameters. Zonal keratometries on areas surrounding the maximum curvature point and the steepest front keratometry performed best, followed by front best-fit sphere and mean keratometry of both surfaces. Platform-dependent indices (e.g., Belin-Ambrósio Deviation- or index height decentration-) obtained good scores, but platform-independent LOGIK performed slightly better. Finally, although minimum radius in both surfaces worked competently, minimum pachymetry (PachyMin) performed considerably poorer.
Conclusions:
We presented a list of parameters whose behavior was repeatable, monotonic and consistent, features desirable to describe change. The parameters normally used to follow keratoconus progression may not be the most adequate, as evidenced by the poor performance of PachyMin. Although calculated for a specific Scheimpflug device, most of the best-performing parameters are platform-independent variables, and results may be generalized, pending validation.
