[The Epoch Times, June 22, 2022](The Epoch Times reporter Li Shaowei compiled and reported) Scientists have invented a machine learning system that only needs to analyze the data of a patient’s brain magnetic resonance (MRI) scan. It can be determined whether he has Alzheimer’s disease. The researchers say the accuracy rate is as high as 98%.
Alzheimer’s disease (Alzheimer’s) is the most common form of dementia. Most of the patients are older than 65 years old, but now the patients are younger. Patients usually present with symptoms such as deterioration of memory, cognition, and language skills. The disease is now incurable, but if diagnosed at an early stage, patients can receive early treatment or change bad lifestyle habits, which can help control the progression of the disease.
Doctors now need to perform several tests on people to assess whether they have the disease. There will be at least a few weeks between the appointment of each test and the final report of the doctor’s assessment.
A study published June 20 in the journal Communications Medicine, part of Nature, describes a new method that allows patients to be diagnosed with a single brain MRI scan. .
The study said they borrowed an algorithm for classifying cancerous tumors and modified it to analyze brain scan data. They divided the brain into 115 regions, and checked the volume, shape, texture and other features of these regions, a total of 660 features for evaluation, and trained this machine learning algorithm to find the difference between the brain tissue of the test object and the normal brain tissue, so as to diagnose whether the patient suffers from Alzheimer’s disease.
The researchers used the system to analyze data from 400 patients with the disease, some healthy individuals and those with other types of neurological disorders, to see how effective the diagnostic tool was. The 400 patients, from the Alzheimer’s Disease Neuroimaging Initiative, included patients with early and advanced stages; patients with other types of neurological disorders included patients with frontotemporal dementia and Parkinson’s disease. In addition, they used the system to examine the data of more than 80 interviewers who were being diagnosed by the Imperial College Healthcare NHS Funding Project, and compared the results of the diagnoses.
The tool alone was able to diagnose 98 percent of those patients, and in 79 percent of them, it was able to tell whether their disease was in an early stage or an advanced stage, the study said.
Eric Aboagye, professor at Imperial College London, who led the study, said: “There is no other simple, generally feasible way to predict Alzheimer’s disease with such accuracy, so our This research is an important step forward. Many of the interviewers have other clinical neurological symptoms, and our system can tell who has Alzheimer’s and who doesn’t.”
“Waiting for a diagnosis can be a terrifying experience for both patients and their families,” Aboghi said. “It would be amazing if we could shorten this waiting period, simplify the diagnosis process, and reduce the uncertainty of the diagnosis. ”
The system found changes in multiple brain regions associated with the disease that scientists had not previously known to be involved in the disease, such as the cerebellum and the ventral diencephalon. This information provides new clues for scientists to continue studying the disease. ◇#
Responsible editor: Ye Ziwei
