iStock/Thinkstock(OAK BROOK, Ill.) — Early diagnosis of Alzheimer’s disease is challenging, and often times diagnosis occurs many years into the disease. Now researchers say earlier detection is possible with high-tech imaging.
Research published in the journal Radiology has shown that changes in glucose uptake in certain parts of the brain can be associated with Alzheimer’s, which can be detected with a type of imaging called FDG-PET scan. These scans can allow for earlier diagnosis of Alzheimer’s, which is when treatment can be the most effective.
The study from the University of California-San Francisco looked at creating and training a deep learning computer algorithm to recognize changes predictive of Alzheimer’s on FDG-PET scan. They trained the algorithm on 1,921 brain scans of patients with Alzheimer’s. They then tested the algorithm on 40 scans of new patients, seven with an ultimate diagnosis of Alzheimer’s disease, and the algorithm was able to correctly identify Alzheimer’s disease 100 percent of the time more than six years before the final diagnosis. This was compared to radiology readers, who could only identify four of the seven scans of people who were ultimately diagnosed with Alzheimer’s.
With further training of the algorithm, it could potentially be used to augment radiologist readings and improve the early prediction of Alzheimer’s disease.
Copyright © 2018, ABC Radio. All rights reserved.