A new research about mental health at Georgia State University’s TReNDS Center may result in early diagnosis of severe diseases such as Alzheimer’s, schizophrenia, and autism.
The findings of this investigation were reported in the Journal of Scientific Reports. A team of seven Georgia State scientists created a sophisticated computer software that could filter through massive amounts of brain imaging data and uncover surprising patterns connected to mental health conditions. The brain imaging data is produced using functional magnetic resonance imaging (fMRI) scans, which analyse dynamic brain activity by detecting minute fluctuations in blood flow.
“We created artificial intelligence models to analyse the considerable amounts of information from fMRI,” said Sergey Plis, the study’s principal author and an assistant professor of computer science and neuroscience at Georgia State University.
He compared dynamic imaging to a movie, rather than a snapshot like an x-ray or the more common structural MRI, saying that “the accessible data is so much greater, so much richer than a blood test or a regular MRI.” But that’s the issue: it’s impossible to interpret so much data.
Furthermore, fMRIs in these cases are costly and difficult to get. However, data mining on typical fMRIs using an AI model is possible. And there are a lot of them about.
“There are large datasets accessible in individuals without a known clinical condition,” says Vince Calhoun, founding director of the TReNDS Center and one of the study’s authors. Using these large but unrelated public datasets improved the model’s performance on smaller, more targeted datasets.
“New patterns emerged that we could definitely attribute to each of the three brain disorders,” according to Calhoun.
Roles of AI models
The AI models were taught the foundations of fMRI imaging and brain activity using a dataset of over 10,000 individuals as their first training ground. The researchers then examined multi-site data sets of over 1200 persons, including those with Alzheimer’s disease, schizophrenia, and autism spectrum disorder.
How does it work? It’s similar to how Facebook, YouTube, and Amazon begin to learn about you based on your online behaviour and begin to predict your future behaviour, likes, and dislikes. The computer algorithm could even pinpoint the exact “moment” when the brain imaging data most strongly showed a link to the relevant mental state.
These insights must be put to use before a condition manifests itself in order to be clinically useful.
According to Calhoun, we may be able to take action if we can identify risk factors for Alzheimer’s disease in a 40-year-old and forecast that risk using markers.
Similarly, if schizophrenia risks can be identified before actual changes in brain structure, there may be opportunities to deliver better or more efficient therapy.
“We are still unable to predict when it will develop, even if we know that someone is at risk of an illness like Alzheimer’s by past testing or family history,” Calhoun said. Brain imaging might narrow that window by detecting relevant patterns before the clinical sickness manifests itself.
Plis explains that after collecting a large imaging dataset, “our AI models will analyse it and report to us what they uncovered about specific disorders.”
We are developing methods to uncover new information that we cannot obtain on our own.”
Md. Mahfuzur Rahman, the study’s first author and a doctorate candidate in computer science at Georgia State University, claimed that the study’s goal was to “connect vast worlds and big datasets with tiny worlds and disease-specific datasets and advance towards indicators essential for therapeutic choices.”