Artificial Intelligence for Lamellar Keratoplasty.

Abstract

The training of artificial intelligence (AI) is becoming increasingly popular. More and more studies on lamellar keratoplasty are also being published. In particular, the possibility of non-invasive and high-resolution imaging technology of optical coherence tomography predestines lamellar keratoplasty for the application of AI. Although it is technically easy to perform, there are only a few studies on the use of AI to optimise lamellar keratoplasty. The existing studies focus primarily on the prediction probability of rebubbling in DMEK and DSAEK and on their graft adherence, as well as on the formation of a big bubble in DALK. In addition, the automated recording of routine parameters such as corneal oedema, endothelial cell density or the size of the graft detachment is now possible using AI. The optimisation of lamellar keratoplasty using AI holds great potential. Nevertheless, there are limitations to the published algorithms, in that they can only be transferred between centres, surgeons and different device manufacturers to a limited extent.