Guttae can occur on the endothelial cell-bearing basement membrane of the cornea. Endothelial cells are located on the posterior surface of the cornea and are responsible for actively pumping out fluid and maintaining the transparency of the cornea. In the presence of many guttae, the endothelial cells lose their function and cannot maintain corneal transparency over time. This very often leads to irreversible visual defects, sometimes so severe that transplantation of a new cornea becomes necessary. Guttae sometimes go undetected on routine preoperative inverted specular microscopy.
The goal in KIttata (Artificial Intelligence for the Detection and Classification of Corneal Guttae in the Cornea Bank Prior to Keratoplasty) is to develop an AI classification algorithm that predicts the transplant suitability of a donor cornea. This is expected to increase the long-term survival rate of corneal grafts, reduce the need for re-transplantation, and reduce associated healthcare costs.
The likelihood of the presence of undetected guttae in a corneal graft depends on morphological criteria, such as differences in cell color, irregularities in cell shape, presence of vesicles, deformities of the cell membrane, or presence of areas without cells.
"While our PhD student Tarek Safi was establishing morphological criteria for detecting masked guttae on donor corneas in the cornea bank, he came up with the idea of using artificial intelligence for detecting alarming structural abnormalities. AI currently has many applications in many areas of medicine, including ophthalmology. By using this advanced technology as an image analysis and recognition tool, a milestone in the quality assurance of corneal donor tissue could be achieved," explains Prof. Dr. Berthold Seitz, Director of the Department of Ophthalmology at Saarland University Hospital.
Scientists from DFKI's Cognitive Assistance Systems research area are using AI methods to develop procedures that implement these criteria. At its core is a classification algorithm that uses the parameters resulting from the procedures to predict whether a given donor cornea is healthy. To do this, they use microscopically collected image data of the corneas depicting these details as input parameters for an AI-based assessment algorithm (classifier).
"Our goal is to optimize the classification of corneas and achieve more accurate results than previously used methods. For this purpose, we use a Deep Learning method, i.e. a machine learning algorithm based on complex neural networks. For this purpose, neural models are created that originate from the information in the cornea bank, i.e. images and parameters," says DFKI project manager Dr. Jan Alexandersson.
Finally, a clinical expert assesses the results. This serves as a basis for adjusting the input parameters of the models and helps to optimize the model gradually.
Funder: Dr. Rolf M. Schwiete Foundation, Mannheim, Germany
Volume: approx. 215,000 €
Duration: 1 year
Partner: German Research Center for Artificial Intelligence - DFKI, Saarbrücken
Press reports
Here you can find the article of the DFKI from 28.08.2020.
Here you can find the article in Ophthalmologische Nachrichten from 02.09.2020.
Here you can find the article of the Saarbrücker Zeitung from 08.09.2020.
Here you can find the article from EYEFOX from 09.09.2020.
Translated with www.DeepL.com/Translator (free version)