Brain patterns can be difficult to predict. This is especially true when neurological disorders or traumatic brain injuries are involved.
“Traumatic brain injuries (TBI) can lead to lasting physical, emotional and cognitive impairments,” Dr. Tayo Obafemi-Ajayi, associate professor of electrical engineering at Missouri State University, said.
“For TBI, among other neurological disorders, the information provided by neuroimaging can play a crucial role in diagnosing the disorder and predicting its development.”
How does neuroimaging work?
One technique involves Magnetic Resonance Imaging (MRI) technology that documents brain activity in pictures.
The technology can serve many purposes in neuroscience. This includes providing a path to predicting TBI’s severity, as Obafemi-Ajayi and her team of researchers have found.
Their research on the subject recently received the Best Paper Award at the 2021 IEEE International Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB).
From machine-fueled models to real understanding
Obafemi-Ajayi’s team evaluated a deep learning model that draws from machine learning.
“Our model is based on neural networks, which attempt to mimic neurons in the human brain,” Obafemi-Ajayi said.
While they know the networks work, the researchers are investigating how and why they do. This ambiguity has led the neural networks’ model to be coined a “black box.”
Obafemi-Ajayi’s research team hopes to break out of the black box’s barriers. This can only become possible by learning where exactly the parameters begin and end.
“We want to understand exactly why the model produces such accurate results,” Obafemi-Ajayi said. “This would allow us to make the model explainable beyond the black box that has up to this point confined interpretations of how it works.”
More about the project
The project served as a collaboration between researchers from Missouri State and Missouri University of Science and Technology (Missouri S&T).
The universities share the cooperative engineering program.
The research team’s contributors include:
- Dacosta Yeboah, alumnus of the computer science department at MSU and leader author.
- Dr. Daniel Hier, adjunct professor of electrical and computer engineering at Missouri S&T.
- Dr. Gayla Olbricht, associate professor of statistics at Missouri S&T.
- Hung Nguyen, undergraduate computer science student at MSU and research assistant in the Computational Learning Systems (CLS) Lab.
- Dr. Tayo Obefami-Ajayi.
Their research paper is titled “A deep learning model to predict traumatic brain injury severity and outcome from MR images.”
The project’s award recognition comes with a $300 prize and certificate.
About the conference
The IEEE International Conference on CIBCB is a flagship conference of the Bioinformatics and Bioengineering Technical Committee (BBTC).
The committee serves as part of the IEEE Computational Intelligence Society (CIS).
“The conference offered exposure to some exciting developments in computational intelligence and biomedical science,” Obafemi-Ajayi said. “Participating speakers provided inspiring talks that captured the amazing pace of progress reached in automated machine learning.”