Texas A&M researchers receive grant funding to develop A.I. that can detect signs of Alzheimer’s Disease
COLLEGE STATION, Texas (KBTX) - Researchers at Texas A&M recently received grant funding to develop a way for artificial intelligence to find signs of Alzheimer’s Disease.
The team is looking to find a way to detect signs of the disease by analyzing MRI or PET scan images. They hope this technology will also help them find the link between the observable characteristics of Alzheimer’s in the brain, called the phenotype, with the genetic mutations that cause the disease, known as the genotype.
”That’s like us trying to compare apples with oranges. We cannot compare them directly. We have to convert each of them into some unified format in order to compare it,” Texas A&M Associate Professor of Computer Science & Engineering Shuiwang Ji said. “The genome is like a library. The library itself is a great trove of information, but it’s not functional on its own. From genome to phenotype, it’s basically like reading different parts of the library. We gain different knowledge, and that knowledge can be translated into something that will impact society. That’s exactly how the genome and the phenotype are related.”
Ji says the complexity of the brain is one reason why a majority of neurological diseases have no real cure. He says when it comes to other organs in the human body, scientists have a much better grasp on how they work, and therefore why a dysfunction exists when something is wrong.
“The brain by itself is the most complex piece of organized matter in the universe,” Ji said. “We have no idea how it works. There are some treatments for different neurological diseases like Alzheimer’s Disease or Parkinson’s, but all of these are based on some sort of trial and error and empirical experiences.”
The goal of Ji’s team is to create an instrument to capture what the brain looks like and correlate that with the information that can be taken from the genotype.
“We have the neuroimaging representing the phenotype and the genome information representing the genotype, but these are very different types of information,” Ji said. “How can you correlate one set of information that’s presented through images while the other is presented through sequencing? Basically, we use deep learning to extract the information through the neuroimages into some kind of integer representation. That’s why we need to use multiple layers of a network to create a linear representation. Then we combine them together.”
Ji’s team is receiving $1.2 million from a grant provided by the National Institutes of Health to pursue this project.
“I think our ultimate goal is by correlating the genotype with the phenotype, we are hoping to identify a set of genes that may lead to the change in phenotype that eventually leads to Alzheimer’s Disease,” Ji said. “If we are able to identify those sets of genes, potentially we can develop a method like a gene therapy to really cure the disease from the root.”
Copyright 2021 KBTX. All rights reserved.