CSL researchers develop tools to aid in Alzheimer's diagnosis and prognosis

11/11/2019 Allie Arp, CSL

Written by Allie Arp, CSL

Alzheimer’s Disease has long stumped the medical community. Even now, researchers aren’t sure how it evolves or who is at risk for developing it. A team of CSL researchers, led by Ravishankar Iyer, is working to create tools that could help improve the diagnosis of the illness and produce more accurate prognoses.

 “Alzheimer’s is reaching epidemic proportions,” said Yoga Varatharajah, a CSL graduate student. “There’s no cure and a lot of unknowns in the disease pathology.”

Varatharajah and his peers in Iyer’s research group are trying to unlock some of the mysteries of Alzheimer’s. The group is working on developing a toolset that can predict the progression of the disease, and  help diagnose a patient’s brain health, an early indicator of the disease. The group recently had a paper discussing this work published in Nature’s Scientific Reports and has had a second paper accepted in IEEE Bioinformatics and Biomedicine for publication later this fall.

Both studies have used machine learning methods to analyze dozens of prior studies with thousands of data points. For the research published in Scientific Reports, the team’s prognosis work was based on a machine learning method for analyzing neuroimaging, spinal fluid, and genetic variables to predict a patient’s progression.

“This paper looked at prognosis, primarily on predicting short-term Alzheimer’s disease progression,” said Varatharajah, an electrical and computer engineering (ECE) student. “There is a continuum for the disease: normal,
Yoga Varatharajah
Yoga Varatharajah
mild cognitive impairment, and dementia. We can predict those people in the second stage who will go to the third stage in the next two years with 93% accuracy.”

They are also working on a diagnostic tool that analyzes brain health using MRI images. The IEEE paper written by Varatharajah, Krishnakant Saboo, and Chang Hu outlines a method of using an MRI image to predict a brain’s medical “age,” which isn’t its chronological age, but the age it appears to be when compared to normally aging brains. If a patient’s brain is “older” than his or her chronological age, that signals a problem in overall brain health.

At present, a diagnosis can be confirmed only after a patient has passed away.

“The gold standard diagnosis is only done using post-mortem pathology, so you have to take the brain tissue, run an analysis, and then determine this person had dementia,” said Saboo, also an ECE student. “There are ways to approximate the level of their cognition, but our research is focused on developing an objective method to come up with diagnosis and prognosis tools based on measures directly from the brain, like imaging and genetic measures.”

Data for both projects were collected from publicly available datasets, through a collaboration with the Aging and Dementia Imaging Lab at the Mayo Clinic. In partnering with neurologists and neuroscientists from Mayo, the group is developing domain-based artificial intelligence models that have a better chance of early detection of the disease. 

"Alzheimer’s is a complex disease with significant heterogeneity seen across patients. The Mayo-Illinois collaboration leverages the AI strengths of Illinois to understand this complexity and further the understanding of mechanisms of Alzheimer’s,” said Dr. Prashanthi Vemuri of Mayo Clinic. “The work published by these researchers showcases how machine learning can be applied to identify individuals at risk of developing dementia in near term, which has a significant public health impact."

Krishnakant Saboo
Krishnakant Saboo
Replication is the next step for the researchers. Varatharajah plans to replicate his work in prognosis prediction with larger groups of patients to validate the findings and generalize the model to more people. Saboo and Hu intend to expand their work to see if they can determine which clinical factors most affect cognition, in order to improve treatments that could slow cognitive decline.

“Since there is no treatment for the disease, these tools can help patients know when to undergo counseling, for example,” said Varatharajah. “Additionally, extensive research is currently being done to find new therapeutic options to treat Alzheimer’s and we believe our techniques will help identify the patients who would most likely benefit from the new therapies.”

Varatharajah and several of the other students came to the issue of Alzheimer’s while working on a DARPA project involving the cognition of epilepsy patients. They are only a few of the students in Iyer’s group working to improve the future of healthcare.

“Working with our Mayo collabortors, Yoga, Saboo, and Chang, along with other students, are working to improve the lives of patients suffering from cognitive problems,” said Iyer. “Through our efforts, and partnerships with medical providers like Mayo Clinic, there is great potential for breakthroughs in the future.”


Share this story

This story was published November 11, 2019.