Researchers have developed an “experimental artificial intelligence (AI) system that can speed the scan review process” for non-invasive eye scans “and help doctors spot subtle signs of eye disease sooner.”
The technology is called OCTCube-M and includes a family of “three AI models that are designed to read and interpret 3D images of the eye’s retina as well as other types of eye scans.” The new AI system was more accurately able to identify eight different retinal diseases, including age-related macular degeneration. It was also more accurate in predicting how fast a severe form of this condition, called geographic atrophy, would progress.
The study also showed that the model “could infer health risks beyond the eye, predicting outcomes such as heart attack, stroke, and kidney failure based solely on retinal imaging.”
The researchers sought to determine if adding 3D tomography to the AI model could further improve disease diagnosis and prognosis, as they had already found that training the model on 2D tomography images was effective. OCTCube-M “more accurately identified six of the eight retinal diseases by about four to six percentage points,” finding “43 to 60 additional cases out of every 1,000 individuals with eye diseases” with the 3D images included in the model’s training. The researchers now plan to train the model with “larger datasets encompassing more patients, more diseases and even more types of imaging data to continue improving upon it.”
