AI may improve early keratoconus detection
Eyüp Özcan MD
Despite the accuracy of Placido corneal topography and tomography imaging for diagnosing keratoconus, post-LASIK keratoconus remains a challenge, occurring in about 0.33 per cent of cases. “Therefore, a better evaluation is required in order to detect keratoconus, especially in the early stages,” said Eyüp Özcan MD, at the ASCRS Virtual Annual Meeting 2020.
To that end, Dr Özcan and colleagues at the Bascom Palmer Eye Institute, University of Miami, Florida, USA, developed an artificial intelligence tool to autonomously assess corneal images for keratoconus signs.
Using techniques of machine learning and deep learning based on analysis of thousands of training images, the AI tool identifies hidden features in the data set that may be missed by unaided human analysis, Dr Özcan said. His group built on existing criteria in AS-OCT imaging for detecting keratoconus features and applied it to a technique using a single high definition OCT scan.
Data from labelled training images were assessed by a deep neural network. The resulting algorithm was then applied to assess images from 43 eyes with keratoconus and 28 normal eyes. Results were compared with diagnoses provided by board-certified corneal specialists.
“The algorithm was able to correctly differentiate all patients with keratoconus from healthy subjects and thus achieving accuracy, sensitivity and specificity of 100 per cent,” Dr Özcan reported. Similarly, it correctly diagnosed 42 out of 43 keratoconic eyes, achieving accuracy of 97.9 per cent, sensitivity of 95.3 per cent and specificity of 94.6 per cent.