Brain Scan Can Aid Early Detection of Alzheimer’s Disease

Machine learning technology can read MRI scans to identify Alzheimer’s disease in the earliest stages when it’s easier to treat, a new study suggests.

Fact-Checked
Female doctor looking at MRI scanner monitor brain scans
Adapting a computer algorithm used to identify cancer tumors, researchers examined features like shape, size, and texture across different regions within the brain to see where changes might predict Alzheimer’s disease on MRIs.Getty Images

One of the most vexing things about diagnosing Alzheimer’s disease is that many patients don’t know they have this common form of dementia until it’s more advanced and challenging to treat. Now, a new study suggests that it may be possible to catch Alzheimer’s early with a single brain scan.

Even though Alzheimer’s disease can’t be cured, early detection can help patients get treatment to help manage and sometimes slow the progression of symptoms, researchers noted in a paper published June 20 in Communications Medicine. Many techniques currently used to diagnose Alzheimer’s disease — including cognitive assessments, memory tests, and brain scans to look for telltale protein deposits in the brain and tissue shrinkage in the hippocampus region of the brain — can take weeks to schedule and then get the results.

To speed up this process, scientists wanted to see if they could use what’s known as machine learning technology to teach computers to more quickly spot signs of Alzheimer’s disease during magnetic resonance imaging (MRI) scans of the brain. Adapting a computer algorithm used to identify cancer tumors, they examined 660 distinct features like shape, size, and texture across 115 different regions within the brain to see where changes to these features might predict Alzheimer’s disease on MRIs.

Scientists tested out this new algorithm on brain scans several hundred patients with various stages of Alzheimer’s disease as well as people with other neurological disorders and a control group of healthy individuals without any cognitive issues.

With this algorithm, machines correctly identified whether patients had Alzheimer’s disease or not 98 percent of the time. And, among those who did have Alzheimer’s disease, the computer correctly determined whether cases were early or advanced in 79 percent of patients.

“Currently no other simple and widely available methods can predict Alzheimer’s disease with this level of accuracy, so our research is an important step forward,” senior study author Eric Aboagye, PhD, of the Comprehensive Cancer Imaging Center at Imperial College London in England, said in a statement.

The ability to detect Alzheimer’s from a single brain scan, without the aid of additional cognitive tests or memory assessments, may also make it easier for patients and families to get answers quickly, Dr. Aboagye said.

“If we could cut down the amount of time they have to wait, make diagnosis a simpler process, and reduce some of the uncertainty, that would help a great deal,” Aboagye said. “Our new approach could also identify early-stage patients for clinical trials of new drug treatments or lifestyle changes, which is currently very hard to do.”

It’s also noteworthy that the computer could correctly distinguish between Alzheimer’s disease and other neurological disorders, because many people seen at memory clinics don’t actually have Alzheimer’s, Aboagye added.

Beyond this, using machine learning to interpret MRIs helped pinpoint some areas of the brain involved in Alzheimer’s disease that haven’t been previously associated with the condition, the study team notes.

The hippocampus has longed been linked to memory and, changes in this part of the brain have traditionally been used as one indicator of dementia or Alzheimer’s disease. However, the study also found some changes in other areas of the brain associated with Alzheimer’s disease including the cerebellum, which plays a role in movement and coordination, and ventral diencephalon, which is involved in sensory perception, sight, and hearing.