Researchers say that the new artificial intelligence algorithm can analyze the photosensitive layer of the eye, that is, the retina, and one day can diagnose various diseases from diabetes to autoimmune and neurodegenerative diseases on the spot.
The artificial intelligence algorithm was proposed by Dr. ursula Schmidt Elfe, director of ophthalmology at Vienna Medical University, at a scientific conference in Vienna earlier this month. The research of this algorithm was published in the journal Ophthalmology on February 8, 65438.
Schmidt Erfurth's research focuses on using artificial intelligence to detect signs of various diseases in retinal images. "Eyes" can't see: 9 eyeball injuries will make you squirm]
"From a simple color photo of retina, you can know the person's age, sex, smoking history, blood sugar level and blood pressure," Schmidt Erfort said, but we can also use the image of optical coherence tomography, which can provide us with more details.
Optical coherence tomography (OCT) is a common technique in ophthalmology. It can take three-dimensional images of the retina and allow doctors to examine in detail what happens in each layer of photosensitive tissue. However, artificial intelligence can do this more accurately and quickly, Schmidt Elforth told Life Science:
In the speech, Schmidt Erfurth showed how such an artificial intelligence algorithm can accurately find the symptoms of diabetes in the retina. Diabetic patients usually have macular edema. Macular edema is essentially the accumulation of fluid in the macula, which is a layer of retina and is responsible for the clarity of central vision. If left untreated, macular edema may lead to permanent damage and vision loss.
"The algorithm provides accurate information about how much liquid there is, but the image itself cannot provide this information," Schmidt Elforth said. She added that doctors can evaluate the effect of macular edema treatment by observing these fluid levels-over time, the decrease of fluid will show that the treatment is effective.
The same algorithm can also detect the earliest symptoms of age-related macular degeneration (ARMD) and even predict the progress of the disease. Schmidt said:
Schmidt said that arm disease is the most common cause of visual impairment in the elderly. Schmidt said that about 60% of people over the age of 50 show early symptoms. However, only 15% of these cases eventually developed into the advanced stage of the disease. Similar to diabetic macular edema, ARMD leads to blurred central vision. Schmidt Elfe said that people can no longer distinguish details, and may even find it difficult to recognize faces.
"Through this technology, we can predict a person's risk of" entering a more serious stage ". Through the algorithm training of large data sets of previous patients, patients who are more prone to diseases than other patients can be identified, and these patients will never develop advanced diseases.
High-risk individuals will receive early treatment, which may help them maintain their vision in old age.
The biggest advantage of this technology is that it doesn't need experts to explain the results.
"Artificial intelligence will enable millions of people who have not been diagnosed so far to receive treatment." She said it was so simple that you didn't even need to see an ophthalmologist.
Schmidt Erfort's early technology has been approved for commercial application in Europe and has been deployed in five hospitals in continental Europe. It uses artificial intelligence to detect signs of diabetic retinopathy from two-dimensional color photos of eyes, in which case retinal blood vessels will rupture.
Originally published in the journal Life Science.