Artificial Intelligence for Screening of Multiple Corneal Diseases

Application of Deep Learning for Screening Multiple Corneal Diseases

This study developed a deep learning algorithm based on anterior segment images and prospectively validated its ability to identify corneal diseases.The effectiveness and accuracy of this algorithm was evaluated by sensitivity, specificity, positive predictive value, negative predictive value, and area under curve.

Inclusion Criteria:

1. The quality of slit-lamp images should clinical acceptable.
2. More than 90% of the slit-lamp image area including three main regions (sclera, pupil, and lens) are easy to read and discriminate.

Exclusion Criteria:

1）Insufficient information for diagnosis.