A researcher or medical professional looks at the results from a brain scan or MRI on a lab computer near a medical exam roomShare on Pinterest
  • Alzheimer’s disease is the most common form of dementia, affecting around 70% of people with dementia, but Alzheimer’s disease can be challenging to diagnose.
  • Doctors currently use multiple cognitive tests and scans to diagnose Alzheimer’s, which can take a long time.
  • Researchers have developed an algorithm to be used with a single brain MRI scan to rapidly detect early signs of Alzheimer’s.
  • In their trial, the system detected 98% of cases of Alzheimer’s disease.

Dementia is, according to the World Health Organization, the seventh leading cause of death worldwide. The most common form, affecting up to 70% of those with a dementia diagnosis, is Alzheimer’s disease.

People with suspected Alzheimer’s usually undergo multiple tests to diagnose the condition. During the assessment, the person will:

  • Give their medical history, both physical and mental.
  • Undergo a medical examination.
  • Undergo a neurological examination to test reflexes, speech and coordination.
  • Take several cognitive tests to assess memory, thinking and simple problem-solving.
  • Have a magnetic resonance imaging (MRI) or CT scan to look for any changes in the brain, such as atrophy or shrinkage of the hippocampus.
  • Undergo cerebrospinal fluid (CSF) or blood tests to measure the levels of beta-amyloid, a protein that accumulates in the brains of people with Alzheimer’s disease.

However, these diagnostic tests may not be accurate, have limited availability, or can take a lot of time, during which the disease could progress without treatment.

Now, a team from Imperial College London has developed an MRI-based machine-learning system to quickly and accurately diagnose Alzheimer’s disease. In their study, published in Communications Medicine, the method could detect both early and more advanced Alzheimer’s disease.

The researchers developed an algorithm based on those used for classifying cancer tumors. Having divided the brain into 115 regions, they allocated 660 features, such as shape, size and texture, to each region. They trained the algorithm to predict Alzheimer’s by identifying changes in these features from a single, standard MRI scan.

They tested their method on brain scans from more than 400 patients in the Alzheimer’s Disease Neuroimaging Initiative. These patients had either early or late-stage Alzheimer’s and were compared with healthy controls and patients with other neurological conditions.

They then tested it using data from 80 patients undergoing diagnostic tests for Alzheimer’s at Imperial College Healthcare NHS Trust.

“This new research approach is using machine learning and MRI scans in an attempt to identify biological brain changes early in the Alzheimer’s disease continuum. That being said, this research is in its early days and it is not ready to be used as a stand alone diagnostic tool.”

Dr. Rebecca Edelmayer, Ph.D., Senior Director of Scientific Engagement, Alzheimer’s Association

The researchers found that the MRI-based machine-learning system accurately predicted Alzheimer’s in 98% of cases in their initial study. It could also distinguish between early and late-stage Alzheimer’s 79% of the time.

When tested on an external data set, the algorithm still detected 86% of Alzheimer’s cases, a higher figure than previously published studies.

Dr. Anton Porsteinsson, professor and director of the Alzheimer’s Disease Care, Research and Education Program (AD-CARE) at the University of Rochester Medical Center, welcomed their results:

“Their method appears to be highly predictive in this population and adds to the number of imaging techniques and fluid biomarkers that make the diagnosis of dementia more accurate.”

The algorithm also showed higher accuracy than the measures currently in use — hippocampal atrophy and cerebrospinal fluid (CSF) beta-amyloid measure — which show 26% and 62% accuracy, respectively.

The researchers suggest that their scan and algorithm method could be an alternative to invasive CSF measurements.

However, Dr. Porsteinsson told Medical News Today: “There is an intense exploration going on right now to find the most convenient and yet highly accurate biomarkers for the diagnosis, prognosis, and possible treatment outcomes in Alzheimer’s and related dementias. This study suggests that the authors’ technique can find a role here, but the competition is formidable, especially the fluid biomarkers.”

Because the new method can detect the early changes in Alzheimer’s disease, it could lead to earlier diagnosis, allowing treatments to begin before symptoms become life changing.

Lead researcher, Prof. Eric Aboagye, from Imperial College’s Department of Surgery and Cancer, called their research “an important step forward.”

“Waiting for a diagnosis can be a horrible experience for patients and their families. 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,” said Prof Aboagye.

Commenting for MNT, Dr. Edelmayer agreed: “[T]his research is addressing an important issue in Alzheimer’s disease: early detection. With FDA accelerated approval of the first anti-amyloid disease-modifying Alzheimer’s treatment and more coming down the pipeline, it is vital that individuals with Alzheimer’s be diagnosed early in the disease process when treatment may be most beneficial.”

Biogen’s Aduhelm (aducanumab) is the treatment Dr. Edelmayer is referring to, for which the FDA granted accelerated approval in 2021.