Researchers at Imperial College London have developed a machine learning tool that can diagnose Alzheimer’s disease from a single magnetic resonance imaging (MRI) scan, by analyzing structural features within the brain, including in regions not previously associated with Alzheimer’s. The team says advantages of the technique are its simplicity and the fact that it can identify the disease at an early stage, when Alzheimer’s can otherwise be very difficult to diagnose.

Research lead Eric Aboagye, PhD, professor, Imperial’s department of surgery and cancer, said, “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. Many patients who present with Alzheimer’s at memory clinics do also have other neurological conditions, but even within this group our system could pick out those patients who had Alzheimer’s from those who did not.”

Aboagye and colleagues reported on their work in Communications Medicine, in a paper titled, “A predictive model using the mesoscopic architecture of the living brain to detect Alzheimer’s disease,” in which they concluded, “This new data analytic method demonstrates potential for increasing accuracy of Alzheimer diagnosis.”

Alzheimer’s disease is the most common cause of dementia, impacting memory, thinking, and behavior, the authors explained. The disorder affects over half a million people in the U.K., mostly those over the age of 65 years, although younger people can also develop Alzheimer’s. The most frequent symptoms of dementia are memory loss and difficulties with thinking, problem solving, and language.

Although there is no cure for Alzheimer’s disease, getting a diagnosis quickly at an early stage helps patients. It allows them to access help and support, get treatment to manage their symptoms, and plan for the future. Being able to accurately identify patients at an early stage of the disease will also help researchers to understand the brain changes that trigger the disease, and support development and trials of new treatments. However, the researchers continued, “It can be challenging to diagnose Alzheimer’s disease, which can lead to suboptimal patient care.”

Doctors currently use a raft of tests to diagnose Alzheimer’s disease, including memory and cognitive tests and brain scans. The scans are used to check for protein deposits in the brain and shrinkage of the hippocampus, the area of the brain linked to memory. All of these tests can take several weeks, both to arrange and to process.

The new approach requires just one MRI scan brain scan taken on a standard 1.5 Tesla machine, which is commonly found in most hospitals. The researchers adapted an algorithm developed for use in classifying cancer tumors and applied it to the brain. They divided the brain into 115 regions and allocated 660 different features, such as size, shape, and texture, to assess each region. They then trained the algorithm to identify where changes to these features could accurately predict the existence of Alzheimer’s disease, even before obvious shrinkage of the brain occurs. “For each patient, a biomarker called “Alzheimer’s Predictive Vector” (ApV) was derived using a two-stage least absolute shrinkage and selection operator (LASSO),” the investigators explained.

Using data from the Alzheimer’s Disease Neuroimaging Initiative, the team tested their approach on brain scans from over 400 patients with either early- or late-stage Alzheimer’s, healthy controls, and patients with other neurological conditions, including frontotemporal dementia and Parkinson’s disease. They also tested the method with data from over 80 patients undergoing diagnostic tests for Alzheimer’s at Imperial College Healthcare NHS Trust.

They found that in 98% of cases, the MRI-based machine learning system alone could accurately predict whether the patient had Alzheimer’s disease or not. It was also able to distinguish between early- and late-stage Alzheimer’s with fairly high accuracy, in 79% of patients.

“This method provides a biomarker able to detect an early stage of AD with a significant potential improvement of the clinical decision support system,” the investigators stated. “Our ApV is robust and repeatable across MRI scans, demonstrating its potential for applicability in clinical practice in the future.” The method also doesn’t require a “subject matter expert,” as it uses established software for both brain segmentation and radiomics analysis, the authors continued. “The algorithm computes manually engineered features allowing an easy interpretation of the ApV and facilitating clinical translation.”

Aboagye noted, “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. 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.”

The new system spotted changes in areas of the brain not previously associated with Alzheimer’s disease, including the cerebellum (the part of the brain that coordinates and regulates physical activity) and the ventral diencephalon (linked to the senses, sight and hearing). This opens up potential new avenues for research into these areas and their links to Alzheimer’s disease.

Co-author Paresh Malhotra, PhD, who is a consultant neurologist at Imperial College Healthcare NHS Trust and a researcher in Imperial’s department of brain sciences, said: “Although neuroradiologists already interpret MRI scans to help diagnose Alzheimer’s, there are likely to be features of the scans that aren’t visible, even to specialists. Using an algorithm able to select texture and subtle structural features in the brain that are affected by Alzheimer’s could really enhance the information we can gain from standard imaging techniques.”

In summary, the team concluded, “… this study proposes an unsupervised approach for the development of an MRI-based biomarker for the biological characterization of AD. The ApV is reproducible and robust. It can be easily computed with the calculation of manually engineered features and is ready to be integrated into the clinical decision support system without the need for additional sampling or patient testing.”

Previous articleGive A Dog A Bone: Touchlight’s Cell-free DNA Vector Synthesis Ignites Genetic Medicine Manufacturing
Next articleCRISPR-Cas12a Editing Rates Improve with Better Directions to the Nucleus