Skip to main content

IU researcher building statistical models to identify Alzheimer’s risk

Aug 23, 2024

Amanda Mejia and student Garrett Collier review the results of fMRI scans on a computer. IU's Amanda Mejia and statistics master's student Garrett Collier review the results of an advanced functional MRI processing and quality control pipeline. Photo by James Vavrek, Indiana University

Nearly 7 million Americans are living with Alzheimer’s disease, a debilitating condition affecting memory, thinking and behavior. While there is no cure, treatment options are available that could alter its progression, so identifying at-risk patients as early as possible is vital to improving their quality of life.

As one of six global recipients of Johnson & Johnson’s prestigious Women in STEM 2 D Scholars Award, Indiana University researcher Amanda Mejia is using her experience developing and disseminating statistical models that analyze functional MRI data to tackle Alzheimer’s.

Mejia hopes to develop new, sophisticated biomarkers using information about brain function extracted from functional MRI scans, which show the activity of the brain. Functional MRI biomarkers, a biological measure that can indicate disease, could complement existing gold-standard biomarkers that are based on more invasive PET scans or spinal taps of cerebrospinal fluid.

Amanda Mejia Amanda Mejia focuses on the development of statistical methods for the analysis of brain imaging data. Photo by James Vavrek, Indiana University

To support this work, Mejia is collaborating with Alzheimer’s researchers at the IU School of Medicine and the Indiana Alzheimer’s Disease Research Center on the IU Indianapolis campus, one of the oldest Alzheimer’s research centers in the country. Mejia and her collaborators were recently awarded a prestigious $3.5 million R01 grant from the National Institute on Aging to develop functional-MRI-based biomarkers. This includes new, better statistical techniques for functional MRI and the use of machine learning and AI to construct biomarkers from complex, high-dimensional data.

“Though functional MRI is not currently a common source of biomarkers for the detection or prediction of Alzheimer’s disease, we believe that there are functional brain signatures that our advanced statistical models can help identify,” said Mejia, an associate professor in the Department of Statistics in the College of Arts and Sciences at IU Bloomington. “We hope to use our models to create a first-line screening method that will identify people who would benefit from additional testing for Alzheimer’s, helping identify at-risk individuals earlier. Although PET is the gold standard, relatively healthy patients and clinical trial participants are sometimes hesitant to undergo this more invasive scan, which is also more expensive and less widely available than MRI.”

Alzheimer’s disease is the most common form of dementia. According to the Alzheimer’s Association, it begins 20 years or more before memory loss or other symptoms begin, illustrating a significant need for identifying people most at-risk of developing it.

“Identifying Alzheimer’s risk as early as possible is really key to identifying interventions that could significantly improve quality of life and longevity for these patients, their families and communities,” Mejia said.

Functional MRI studies have been used for years to identify population trends to better understand how Alzheimer’s disease affects brain function. Unfortunately, though, functional MRI has historically been thought to lack the precision to be useful in most clinical settings. But it holds a lot of promise.

A structural MRI (left) can be visually inspected by expert radiologists, while a functional MRI (right) cannot be interpreted visually. As a result, advanced image analysis algorithms used to analyze and extract relevant information about the scan are critical to the clinical utility of fMRI. Courtesy of Amanda Mejia

Functional MRI has already proven valuable before brain surgery to isolate parts of the brain responsible for specific vital functions, such as speech or understanding. Recently, Mejia said there is also excitement around the use of functional MRI to guide brain stimulation targets to combat depression.

However, biomarkers involve more complex measures about what happens within an individual’s brain, such as how its parts are spatially organized and interact with one another. Because these complex functional brain signatures are more subtle and difficult to identify, Mejia and her team are focused on building more sophisticated and powerful statistical models that are optimized to extract these measures with a high level of precision. That data can then be used to build new biomarkers.

To disseminate these new techniques, it is important to have user-friendly software that is rigorously tested and refined. Damon Pham, an IU Bloomington graduate and former Wells Scholar, leads the software development efforts in Mejia’s lab. Pham and Mejia partner seamlessly, with Mejia developing the algorithms while Pham builds the software infrastructure to deploy them. The software is then used to analyze and validate the method before it can be applied to real data to provide scientific insights.

Mejia makes the software freely available to research laboratories around the world so they can apply the method to their own functional MRI studies. So far, Mejia and Pham’s toolboxes for fMRI analysis have been downloaded over 45,000 times, an indication of the broad impact of the lab’s work.

“Someday, we hope that patients will benefit directly from the work we do, but it’s a long game we’re playing,” Mejia said. “If we can encourage people to do the extra work and go outside their comfort zones to use more advanced statistical techniques that have been proven to work, then we can start advancing toward greater clinical utility of functional MRI in Alzheimer’s and other devastating diseases.”

Author

IU Newsroom

Kelsey Cook

Deputy director for research communication

More stories