Related Experiment Videos
Automated keratoconus detection using the EyeSys videokeratoscope.
P J Chastang1, V M Borderie, S Carvajal-Gonzalez
1Department of Ophthalmology, Hôpital Saint Antoine, Paris, France.
Journal of Cataract and Refractive Surgery
|June 1, 2000
Summary
The EyeSys System 2000 effectively detects keratoconus using a decision tree with two key indices. This method shows high sensitivity and specificity for identifying this corneal condition.
Area of Science:
- Ophthalmology
- Corneal Imaging
- Diagnostic Indices
Background:
- Keratoconus diagnosis relies on accurate corneal shape analysis.
- The EyeSys System 2000 offers advanced topographic data.
- Identifying reliable indices for keratoconus detection is crucial.
Purpose of the Study:
- To assess the efficacy of EyeSys System 2000-derived indices in identifying keratoconic corneas.
- To develop a diagnostic model for keratoconus detection.
Main Methods:
- Corneal topography data from 208 patients across 8 diagnostic groups were analyzed.
- Nine statistical indices, including Holladay Diagnosis Summary indices and a refractive power symmetry index, were evaluated.
- A decision tree model combining the most effective indices (SDSD and Asph) was developed and validated.
Main Results:
- The optimal indices for keratoconus detection were SDSD (standard deviation of the standard deviations of the radii of curvature of each ring) and Asph (coefficient of asphericity).
- The decision tree model achieved 88.5% sensitivity and 94.9% specificity in the validation set.
- False positives were primarily observed in the penetrating keratoplasty (PKP) group.
Conclusions:
- The EyeSys System 2000, utilizing a decision tree with specific indices, can reliably detect clinically apparent keratoconus.
- This approach aids in differentiating keratoconus from normal corneas and other irregular corneal conditions.
- The study highlights the potential of advanced topographic analysis for diagnosing corneal diseases.