Can combining corneal OCT and air puff tonometry help better diagnose the earliest stages of keratoconus? That was the question ELZA researcher, Dr. Nanji Lu, was trying to answer (1). To do so, he performed a study that examined 622 eyes from 622 patients with already-known diagnoses. Some had normal corneas, and the rest had either advanced keratoconus (AKC), early keratoconus (EKC), or, the very earliest stage of keratoconus development, and the hardest to diagnose: sub-clinical or “forme-fruste” keratoconus (FFKC).
Nanji and his collaborators used corneal tomographic (shape and thickness) maps obtained from OCT imaging, combined with corneal biomechanical strength data from an air puff tonometer (which emits a puff of air to the eye, and a high-speed camera records and measures how the cornea changes shape in response to the air deflection). As the researchers already knew the which type of keratoconus these eyes had, they were able to feed the OCT and air puff tonometer data into an artificial intelligence (AI)/ machine learning program which then progressively refined a diagnostic model, to the point where it was able to detect FFKC with greater accuracy than existing corneal tomography-based methods.
The importance of detecting keratoconus early cannot be overstated: the disease tends to be progressive, and keratoconus progression is associated with greater and greater loss of vision. A surgical procedure, corneal cross-linking (CXL) can stop keratoconus progression, but it does not tend to recover the vision that was lost up until that point, meaning that the earlier progressive keratoconus is detected, the more progression can be avoided through the use of earlier CXL.
The Medical Director of the ELZA Institute, Prof. Farhad Hafezi said “We are proud to see our postdoc Dr. Nanji Lu’s work featured on the cover page of the Journal of Refractive Surgery.”
Click here to read the paper.
