Technology

Artificial intelligence quickly detects diseases in the eyes and lungs

A new system that can move quickly and accurately into diagnoses of diseases both in the eyes and in the lungs, using artificial intelligence and mechanical learning techniques, developed by Americans and Chinese.

The new artificial intelligence platform not only recognizes two of the most common diseases of the retina of the eye (degeneration and diabetic macular edema), but also evaluates their severity. He also analyzes children’s X-rays and can distinguish between bacterial and viral pneumonia.

Researchers, headed by Professor of Ophthalmology, Kang Zhang, of the University of California-San Diego Medical School, published the publication in “Cell” magazine. Macular degeneration and diabetic macular edema are the two most common causes of irreversible blindness, but both conditions are curable if diagnosed early, requiring the presence of a specialized ophthalmologist.

Instead, as Zhang said, “the new artificial intelligence tool can be used anywhere in the world, especially in provincial areas, which is important in places like China, India and Africa, where there are relatively few doctors.”

The new “intelligent” system, which is a “convolutive neural network” and also uses the technique of “transfer learning” (knowledge from solving a problem is then applied to solve another problem), was trained with more than 200,000 images (OCT), a widely used non-invasive eye examination of the eye. After diagnosis, artificial intelligence also goes for recommendations for further examinations or treatments.

The researchers compared computer diagnoses with those of five ophthalmologists. As Zhang said, “the machine can do just as well as a trained ophthalmologist. It is able to make a decision on whether the patient needs treatment within 30 seconds and more than 95% accuracy “. In addition, according to researchers, the new method will be cheaper for patients.

Scientists also applied the system to childhood pneumonia. Analyzing lung radiographs, artificial intelligence could distinguish the etiology of pneumonia (from bacteria or virus) to more than 90% accuracy. Rapid differential diagnosis is important, because usually more severe bacterial pneumonia (the leading cause of death for children up to five years of age) requires antibiotic therapy.

Researchers are optimistic that their system will be exploited in other applications, for example, in the future. in the distinction between cancerous and non-lesions that occur in CT and MRI.