AI recognizes Alzheimer's for years before diagnosis


Shoot the human brain using PET

Thousands of PET images of Alzheimer's patients at an early stage are used by researchers to train their AI. (Photo: North American Radiology Society)

BerlinIn the fight against Alzheimer's disease, early detection is very important. If dementia that can still be cured is detected early, it can at least slow down their journey with drugs.

"If we diagnose Alzheimer's only when obvious symptoms appear, the loss of brain volume is so large that it's usually too late for effective intervention," explained Jae Ho Sohn.

Together with his team from the University of California in San Francisco, doctors have developed a new tool for early detection of Alzheimer's disease: a reliable adaptive algorithm that predicts dementia years before a doctor's diagnosis.

The researchers focused their development on subtle metabolic changes in the brain caused by the onset of disease. These changes can be visualized using imaging techniques known as positron emission tomography (PET).

Seven factors that support Alzheimer's

However, the trace in the early stages of the disease is so weak that it can hardly be recognized even for experienced doctors. "It's easier for humans to find biomarkers of certain diseases," explained Sohn. "But metabolic change is a more subtle process."

The researchers trained their artificial intelligence using data from Alzheimer's & Disease Neuroimaging Initiative (ADNI). Among other things, this data collection contains thousands of PET images of Alzheimer's patients in the early stages of the disease. 90 percent of these recordings, researchers used to train algorithms, the remaining 10 percent to control success.

For the last test, the AI ​​finally had to analyze 40 images that had not been submitted to him until then. The results describe the child as follows: "The algorithm can detect any case, which then comes to the onset of Alzheimer's disease."

In addition to the 100 percent hit rate, the doctors were impressed by all the very early identification of cases. On average, the system recognizes symptoms more than six years before the diagnosis of the actual disease. "We are pleased with this result," Son said. However, doctors also know that the test series is still relatively small and further tests must confirm the results.

However, he looked in his algorithm for an important important tool in Alzheimer's treatment: "If we can detect this disease earlier, it will give researchers the opportunity to find better ways to slow down or even stop the disease process."


Source link