Predicting the likelihood of need for future keratoplasty intervention using artificial intelligence
Siamak Yousefi1, Hidenori Takahashi2, Takahiko Hayashi3
1Department of Ophthalmology, University of Tennessee Health Science Center, Memphis, USA; Department of Genetics, Genomics, And Informatics, University of Tennessee Health Science Center, Memphis, TN, USA.
The Ocular Surface
|March 3, 2020
Summary
Artificial intelligence (AI) can now identify corneal conditions and predict keratoplasty needs using optical coherence tomography (OCT) scans. This AI tool helps surgeons identify high-risk patients for future corneal transplant surgery.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Corneal diseases often necessitate keratoplasty (corneal transplant surgery).
- Accurate prediction of future keratoplasty needs is crucial for patient management.
- Optical coherence tomography (OCT) provides detailed corneal parameter data.
Purpose of the Study:
- To develop and validate an AI system for automated corneal condition identification.
- To predict the likelihood of future keratoplasty intervention based on OCT parameters.
- To assist surgeons in identifying patients at higher risk for corneal transplantation.
Main Methods:
- Collected 12,242 corneal OCT images from 3,162 subjects.
- Developed a pipeline using data transformations and unsupervised machine learning.
- Identified five distinct clusters of eyes based on corneal parameters.
Main Results:
- The AI system identified five clusters with varying likelihoods of needing keratoplasty.
- Normalized likelihoods for keratoplasty ranged from 1.0% to 33.1% across clusters.
- The system utilizes corneal shape, thickness, and elevation parameters.
Conclusions:
- The AI system aids in identifying patients at higher risk for future keratoplasty.
- The identified clusters can guide clinical decision-making for corneal transplant evaluation.
- Further validation with independent datasets is recommended.
