Corneal Edema Visualization With Optical Coherence Tomography Using Deep Learning: Proof of Concept.

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Optical coherence tomography (OCT) is crucial for diagnosing and monitoring corneal edema.
• Assessing minimal or localized corneal edema with OCT can be challenging.

Purpose of the Study:

• To develop and validate a novel automated tool for detecting and visualizing corneal edema using OCT.
• To improve the accuracy and efficiency of corneal edema assessment.

Main Methods:

• A convolutional neural network was trained to classify pixels in OCT images as normal or edema.
• The model generated colored heat maps to visualize edema distribution.
• Performance was evaluated using metrics like accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUROC).

Main Results:

• The automated tool achieved high diagnostic performance with an AUROC of 0.994 for edema fraction (EF).
• An optimal threshold of 6.8% EF distinguished normal from edematous corneas with 98.7% accuracy, 96.4% sensitivity, and 100% specificity.
• Significant differences in mean EF were observed between normal and edematous corneas (P < 0.0001).

Conclusions:

• The developed model accurately detects corneal edema from OCT scans.
• The tool effectively differentiates between normal and edematous corneas.
• Colored heat maps provide valuable visualization of edema presence.