Summary: A new AI algorithm used fMRI data to identify new brain patterns associated with mental health disorders.
A source: Georgia State University
New research from Georgia State University’s TReNDS Center could lead to earlier diagnosis of devastating conditions such as Alzheimer’s disease, schizophrenia and autism — leading to better prevention and easier treatment.
In a new study published in Scientific reports A team of seven scientists from Georgia State has built a sophisticated computer program that can sift through large amounts of brain imaging data and uncover new patterns related to mental health conditions.
Brain imaging data were obtained from scans using functional magnetic resonance imaging (fMRI), which measures dynamic brain activity by detecting small changes in blood flow.
“We built artificial intelligence models to interpret large amounts of data from fMRI,” said Sergey Plis, associate professor of computer science and neuroscience at Georgia State and lead author of the study.
He likened such a dynamic image to a movie – rather than an image such as an X-ray or the more common structural MRI – and noted that “the data available is much larger, much richer, than a blood test.” continuous MRI. But that’s the challenge — it’s hard to interpret large amounts of data.”
In addition, fMRI in these specific conditions is expensive and not easy to obtain. Using an artificial intelligence model, normal fMRI data can be mined. And there are plenty of them.
“There are large datasets in people without a known clinical disorder,” explains Vince Calhoun, founding director of the TReNDS Center and one of the study’s authors. Using these large but uncorrelated available datasets improved model performance on smaller specific datasets.
“New patterns have emerged that can be clearly associated with each of the three brain disorders,” Calhoun said.
AI models were first trained on a dataset of over 10,000 individuals to learn to understand basic fMRI imaging and brain activity. The researchers then used a multi-site dataset of more than 1,200 individuals with autism spectrum disorder, schizophrenia and Alzheimer’s disease.
How it works? It can learn about you from your online behavior like Facebook, YouTube or Amazon and predict future behavior, likes and dislikes. Computer software and even brain imaging data were able to work “in the moment” associated with the mental disorder in question.
To make these findings clinically useful, they must be applied before the disorder occurs.
“If we can find and predict 40-year-old Alzheimer’s risk markers,” Calhoun said, “we can do something about it.”
Similarly, if the risk of schizophrenia can be predicted before actual changes in brain structure occur, there may be ways to offer better or more effective treatments.
“Even if we know from other tests or family history that someone is at risk for a disease like Alzheimer’s, we can’t predict when it will happen,” Calhoun said.
“When brain imaging occurs before clinical disease is evident, it can capture relevant samples and shorten that time window.”
“The vision is that we collect a large imaging data set, our AI models look at it and show us what they know about specific disorders,” Plis said. “We’re building systems to discover new knowledge that we can’t discover on our own.”
“Our goal,” said Dr. Mahfoozur Rahman, first author of the study and a doctoral student in informatics at Georgia State, “is to combine big worlds and big data sets with small worlds and disease-specific data sets and move toward relevant markers for clinical research. decisions”.
Funding: This research was supported by startup funds to SMP and in part by NIH grants R01EB006841, R01MH118695, RF1MH121885, and NSF 2112455.
This is about AI and mental health research news
Author: Noelle Reetz
A source: Georgia State University
The connection: Noelle Reetz – Georgia State University
Photo: Image is in the public domain
Original research: Open access.
“Interpreting Brain Dynamics Interpreting Models” by Sergey Plis et al. Scientific reports.
Interpreting interpretive models of brain dynamics
Brain dynamics are complex but key to understanding brain function and dysfunction.
The dynamics captured by resting-state functional magnetic resonance imaging data are noisy, high-dimensional, and not easily interpreted. The typical approach of reducing these data to low-dimensional features and focusing on the most predictable features carries strong assumptions and may miss important aspects of the underlying dynamics.
In contrast, introspection of discriminatively trained deep learning models may contain degraded elements of the signal at the level of individual time points and spatial locations. However, the difficulty of reliable training on high-dimensional low-sample datasets and the resulting imprecise relevance of predictive markers prevent the widespread use of deep learning in functional neuroimaging.
In this work, we introduce a deep learning framework for learning from high-dimensional dynamic data while maintaining stable, ecologically valid interpretations.
The results successfully demonstrate that the proposed framework enables the learning of resting-state fMRI dynamics directly from small data and provides a compact, robust interpretation of predictive features of function and dysfunction.