Related Experiment Videos
Placido-based indices of corneal irregularity
Darío Ramos-López1, Andrei Martínez-Finkelshtein, Gracia M Castro-Luna
1Department of Statistics and Applied Mathematics, University of Almería, Almería, Spain.
Summary
New corneal irregularity indices directly analyze Placido disk images, bypassing reconstruction. These sensitive and specific metrics accurately detect keratoconus, offering a faster, simpler diagnostic tool for clinical practice.
Area of Science:
- Ophthalmology
- Biomedical Engineering
- Medical Imaging
Background:
- Corneal topography is essential for diagnosing conditions like keratoconus.
- Current Placido topographers rely on complex surface reconstruction.
- There is a need for direct, efficient methods to assess corneal irregularity.
Purpose of the Study:
- To develop novel indices for measuring anterior corneal surface irregularity directly from Placido disk images.
- To bypass the traditional surface or curvature reconstruction step in corneal topography analysis.
- To achieve high sensitivity and specificity in detecting corneal abnormalities.
Main Methods:
- Proposed several basic indices analyzing geometric and mathematical properties of Placido mires.
- Developed primary indices directly from digitized ring displacements.
- Introduced compound metrics (e.g., generalized linear model, classification trees) combining primary indices for improved efficiency.
- Tested metrics on the CSO topography system, adaptable to other Placido topographers.
Main Results:
- Primary indices demonstrated excellent accuracy in discriminating between normal and keratoconic corneas.
- Receiver operating characteristic (ROC) curve analysis confirmed high sensitivity and specificity.
- Combined indices achieved optimal efficiency, showing perfect classification within the test group.
Conclusions:
- The proposed primary indices effectively differentiate normal from irregular corneas.
- Combined indices offer optimal accuracy for clinical use as disease markers.
- All developed indices are computationally fast and easily implementable in existing corneal topography systems.