Related Experiment Video
Updated: Jun 18, 2025
08:58
Artificial Intelligence Approaches to Assessing Primary Cilia
Published on: May 1, 2021
3.4K
A novel artificial intelligence model for diagnosing Acanthamoeba keratitis through confocal microscopy
Omar Shareef1, Mohammad Soleimani2, Elmer Tu3
1School of Engineering and Applied Sciences, Harvard College, Cambridge, MA, 02138, USA; Department of Ophthalmology, Massachusetts Eye and Ear, Harvard Medical School, Boston, MA, 02114, USA.
The Ocular Surface
|July 31, 2024
Summary
An artificial intelligence (AI) model was developed to diagnose Acanthamoeba keratitis (AK) using Heidelberg Retinal Tomograph 3 (HRT 3) in vivo confocal microscopy (IVCM) images. The AI model achieved 76% accuracy, demonstrating promise for early AK diagnosis.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Acanthamoeba keratitis (AK) is a serious eye infection requiring prompt diagnosis.
- In vivo confocal microscopy (IVCM) is a valuable tool for diagnosing AK.
- Standardized AI-driven diagnostic tools are needed to improve AK detection.
Purpose of the Study:
- To develop and evaluate an artificial intelligence (AI) model for diagnosing Acanthamoeba keratitis (AK).
- The AI model utilizes images from the Heidelberg Retinal Tomograph 3 (HRT 3) in vivo confocal microscopy (IVCM).
Main Methods:
- A retrospective cohort study used HRT 3 IVCM images from culture-confirmed AK patients.
- Cornea specialists independently labeled images as AK or nonspecific finding (NSF).
- A convolutional neural network (CNN) was developed using Python and TensorFlow for image classification.
Main Results:
- The study analyzed 3312 IVCM images from 17 AK patients.
- After expert consensus, 2782 images were used for model training and validation.
- The AI model achieved 76% accuracy, sensitivity, and specificity, with 78% precision.
Conclusions:
- An AI model was successfully developed for diagnosing AK using HRT-based IVCM images.
- The model demonstrated good diagnostic performance, indicating potential for clinical use.
- This AI tool shows promise in enhancing early detection and management of Acanthamoeba keratitis.
