Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Automated decision tree classification of corneal shape.

Michael D Twa1, Srinivasan Parthasarathy, Cynthia Roberts

  • 1College of Optometry, The Ohio State University, Columbus, 43210, USA. twa.1@osu.edu

Optometry and Vision Science : Official Publication of the American Academy of Optometry
|December 17, 2005
PubMed
Summary

Automated decision tree classification accurately distinguishes keratoconic from normal corneas using Zernike polynomial data. This machine learning approach offers an objective, quantitative method for corneal shape analysis.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

A randomized crossover clinical trial of the Deseyne contact lens for presbyopia with extended depth of focus.

Optometry and vision science : official publication of the American Academy of Optometry·2026
Same author

Development and characterization of dibenzalacetone-loaded oleogels as a potential photoprotective agents for sunscreen formulations.

Scientific reports·2026
Same author

Wavelet time-frequency analysis enhances noise-resilient quantification of corneal wave velocity and natural frequency in optical coherence elastography.

Biomedical optics express·2026
Same author

Mental health burden associated with specific ocular conditions among Medicare Advantage beneficiaries with type 2 diabetes: examining the impact of glaucoma, cataracts, retinopathy, and dry eye on unhealthy mental health days.

Frontiers in public health·2026
Same author

Current Evidence on Options and Efficacy of Combination Therapy for Myopia Control-A Narrative Review.

Ophthalmic & physiological optics : the journal of the British College of Ophthalmic Opticians (Optometrists)·2026
Same author

Author's response.

Optometry and vision science : official publication of the American Academy of Optometry·2026

Area of Science:

  • Ophthalmology
  • Biomedical Engineering
  • Machine Learning

Background:

  • Videokeratography generates complex corneal data, challenging objective interpretation.
  • Current methods often rely on subjective pattern recognition or summary indices.
  • Objective, quantitative classification of corneal shapes is needed.

Purpose of the Study:

  • To apply decision tree induction, a machine learning method, for objective corneal shape classification.
  • To discriminate between normal and keratoconic corneal shapes using Zernike polynomial coefficients.
  • To compare the decision tree classifier's performance against established methods.

Main Methods:

  • Corneal surfaces of 132 normal eyes and 112 keratoconic eyes were modeled using seventh-order Zernike polynomials.

Related Experiment Videos

  • A C4.5 decision tree algorithm was used for classification.
  • Performance was compared with Rabinowitz-McDonnell index, Z3 index, KPI, KISA%, and Cone Location and Magnitude Index, evaluating ROC curves.

    • Main Results:

    • The decision tree classifier achieved 92% accuracy and an ROC area of 0.97.
    • It identified four key Zernike coefficients (inferior elevation, sagittal depth, oblique toricity, trefoil) for classification.
    • Performance was equal to or better than other tested classification methods.

    Conclusions:

    • Automated decision tree classification using Zernike polynomials provides accurate, interpretable, and quantitative corneal shape analysis.
    • This method can be implemented on various instrument platforms outputting raw elevation data.
    • The pattern classification approach is adaptable for other diagnostic challenges.