25 November 2020

Parkinsonism is visible in the eyes

AI helped detect Parkinson's disease in the early stages from fundus images

Maria Azarova, Naked Science

Parkinson's disease is a slowly progressive chronic disease of the central nervous system, most often affecting the elderly. Millions of people around the world are affected by this disease, but there is no way to defeat the disease. At the same time, it is not so easy to identify it: usually the diagnosis is made in the presence of symptoms such as tremor, muscle rigidity and balance disorders, but this approach has many limitations.

Scientists and doctors are working hard to create methods for determining the disease. A new approach was proposed by the staff of the University of Florida (USA). They presented their report at the annual conference of the Radiological Society of North America (RSNA), briefly described in the press release Eye exam could lead to early Parkinson's disease diagnosis on the EurekAlert portal.

"The problem is that patients show symptoms only after a long progression, when the dopamine neurons of the brain are already severely damaged," explains the study's lead author Maximilian Diaz, doctor of biomedical engineering. "Therefore, we diagnose at a later stage."

The progression of Parkinson's disease is characterized by the breakdown of nerve cells, thinning the walls of the retina – the tissue lining the inner cavity of the eyeball. In addition, the disease affects the microscopic blood vessels of the retina. Thus, the scientists concluded, the disease can be determined by the images of the fundus of patients. To do this, they used the method of support vectors (machine learning technique with a teacher), known since the late 80s. By taking pictures of the fundus of people suffering from Parkinson's disease and healthy participants from the control group, scientists were able to train AI to detect signs of the disease from these images.

Eye_exam.jpg

Comparison of vascular segmentation / © Radiological Society of North America.

As a result, AI was able to classify the signs of Parkinson's disease based on the retinal vascular network: if a person is susceptible to it, then the structure of these microcapillaries will be disrupted. "The most important conclusion of our study was that the brain disease was diagnosed based on the basic picture of the eye," said Diaz. "This is very different from traditional approaches, when you study different brain images to find neurological problems."

Portal "Eternal youth" http://vechnayamolodost.ru


Found a typo? Select it and press ctrl + enter Print version