AI diagnoses early Alzheimer’s disease from a single MRI brain scan

Researchers at Imperial College London can diagnose Alzheimer’s disease from a single magnetic resonance imaging (MRI) scan by analyzing structural features in the brain that include areas previously not associated with Alzheimer’s disease. We have developed a machine learning tool. The team states that the advantage of this technique is its simplicity and the fact that it can identify the disease in the early stages when it is very difficult to diagnose Alzheimer’s disease.

Research leader Dr. Eric Aboage, a professor of surgery and cancer at Imperial, said: Many patients with Alzheimer’s disease in memory clinics also have other neurological conditions, but even within this group, our system suffers from Alzheimer’s disease from patients who do not have Alzheimer’s disease. I was able to choose a patient.

Aboagye and colleagues Communication medicine“Predictive model using the mesoscopic architecture of the living brain to detect Alzheimer’s disease“This new data analysis technique has the potential to improve the accuracy of the diagnosis of Alzheimer’s disease,” they conclude.

The authors explained that Alzheimer’s disease is the most common cause of dementia and affects memory, thinking, and behavior. The disorder affects more than 500,000 people in the UK, primarily those over the age of 65, but young people can also develop Alzheimer’s disease. The most common symptoms of dementia are memory loss and thinking, problem-solving, and language difficulties.

There is no cure for Alzheimer’s disease, but getting an early diagnosis can help patients. It allows them access to help and support, receive treatment to manage their symptoms, and plan for the future. Accurately identifying patients in the early stages of the disease helps researchers understand the brain changes that cause the disease and supports the development and trial of new therapies. However, the researchers continued, “Diagnosis of Alzheimer’s disease can be difficult and patient care may not be optimized.”

Doctors are currently using many tests to diagnose Alzheimer’s disease, such as memory and cognitive tests, and brain scans. Scans are used to check for protein deposition in the brain and contraction of the hippocampus, a region of the brain associated with memory. All of these tests can take weeks, both in order and processing.

The new approach requires only one MRI scan and a single brain scan on a standard 1.5 Tesla machine commonly found in most hospitals. Researchers have adopted an algorithm developed for use in the classification of cancerous tumors and applied it to the brain. They divided the brain into 115 areas, assigned 660 different functions such as size, shape, and texture, and evaluated each area. We then trained algorithms to pinpoint where these functional changes could accurately predict the presence of Alzheimer’s disease, even before the apparent contraction of the brain. “For each patient, a biomarker called the’Alzheimer’s disease prediction vector'(ApV) was derived using a two-step minimum absolute contraction and selection operator (LASSO),” the researchers explained. ..

Using data from the Alzheimer’s Disease Neuroimaging Initiative, the team used brains from patients with early or late Alzheimer’s disease, healthy controls, and patients with other neurological conditions such as frontotemporal dementia and Parkinson’s disease. I tested the approach with a scan. They also tested this method using data from more than 80 patients undergoing a diagnostic test for Alzheimer’s disease at Imperial College Healthcare NHS Trust.

They found that in 98% of cases, MRI-based machine learning systems alone could accurately predict whether a patient had Alzheimer’s disease. In addition, 79% of patients were able to distinguish between early and late stages of Alzheimer’s disease with fairly high accuracy.

“This method provides biomarkers that can detect early stages of AD and has the potential to significantly improve clinical decision support systems,” the researchers say. “Our ApV is robust, reproducible throughout MRI scans, demonstrating its potential for future clinical practice.” This method has been established for both brain segmentation and radiomix analysis. The author continued that he does not need a “subject-matter expert” to use the software. “The algorithm calculates manually designed functions, allows for easy interpretation of ApV and facilitates clinical translation.”

Aboagye said: “Waiting for a diagnosis can be a horrifying experience for patients and their families. It can reduce the amount of time they have to wait, make the diagnosis a simpler process, and reduce some of the uncertainty. If possible, it would be of great help. Our new approach could also identify patients in the early stages of clinical trials of new drug treatments and lifestyle changes, but this is currently very difficult. “

The new system will change areas of the brain that were not previously associated with Alzheimer’s disease, such as the cerebellum (the part of the brain that regulates and regulates physical activity) and the ventral diencephalon (related to sensation, vision, and hearing). discovered. This opens up a potential new path for studying the link between these areas and Alzheimer’s disease.

Dr. Palais Marhotra, a consultant neurologist and researcher in the Imperial Brain Sciences Department at the Imperial College Healthcare NHS Trust, said: A scan that is invisible to experts. Algorithms that allow you to select textures and subtle structural features in the brain affected by Alzheimer’s disease can actually enhance the information available from standard imaging techniques. “

In summary, the team concluded: “… This study proposes an unsupervised approach to the development of MRI-based biomarkers for the evaluation of the biological characterization of AD. ApV is reproducible and robust. It is easy to calculate and is ready to integrate into a clinical decision support system without the need for additional sampling or patient testing. “