Magnetic resonance imaging (MRI) offers images of the body that allow doctors to diagnose injury or illness. MRI’s susceptibility tensor imaging (STI) technique measures the magnetic susceptibility of different brain tissues. This information can help to better understand, diagnose and monitor neurological diseases like multiple sclerosis (MS) and Alzheimer’s.
The team at Johns Hopkins have a new algorithm for this technique, called DeepSTI. It takes data from multiple individual scans and provides a “super-scan” with brain tissue susceptibility information. The method requires fewer images taken in fewer positions compared to traditional STI, making the process faster and more pleasant for patients, according to a post on the Johns Hopkins Website.
“Usually, STI imaging requires at least six different scans at different head orientations to achieve a good reconstruction, and that’s mainly why it’s not currently broadly used despite its potential to understand the human brain,” said senior author Jeremias Sulam, an assistant professor of biomedical engineering and an instructor in the Whiting School’s Engineering for Professionals applied biomedical engineering program. “Our AI-assisted reconstructions greatly expand the amount of useful information that can be gleaned while requiring much less data, and we hope that will help move this imaging technique from lab to clinic.”
Results from the research appeared in Medical Image Analysis, and in the proceedings of the 2023 International Workshop on Machine Learning in Clinical Neuroimaging. Zhenghan Fang, a biomedical engineering graduate student, served as the lead author of the papers. Peter van Zijl, Xu Li, and Hyeong-Geol Shin from the Kennedy Krieger Institute also contributed to the studies.
More about the STI brain scan algorithm
STI can help uncover neurodegenerative processes that affect specific structures in the brain. That includes the myelin sheath that surrounds axons and plays a central role in information transmission in the brain. The algorithm takes MRI data and creates a high-resolution 3D map of magnetic susceptibility in the brain.
Researchers say the key advance comes in the form of the ability to measure sources like myelin and iron using fewer scans than previously needed. Capturing changes in such tissues can help characterize neurological disease, type, stage or progression. The team says the algorithm generated a reconstruction to visualize myelin changes in patients with MS using data from a lone scan from one head orientation.
DeepSTI’s machine learning utilizes regularization, narrowing down the number of possible solutions to zero in on the most accurate ones. The team’s regularizers steer the model toward the most plausible solution in brain reconstructions for each new set of scans.
“The algorithm is very good at ‘learning’ what brain images should look like. When we run the algorithm on raw data, it uses those found parameters to reconstruct a clearer and more comprehensive image of the brain,” said Fang.
What’s next?
The team hopes the algorithm could make STI more affordable by reducing the time needed for scans while enhancing image quality. Researchers plan to look into how the algorithm could work in other science and engineering applications.
“Our team is excited about the mathematical framework that underpins the results here. We’ve been able to answer hard questions that we couldn’t answer before, like what are the implicit priors of data captured by these reconstruction algorithms, why this works, and why we should be using this,” said Sulam.
The team also made its data, models and code available as an open-source project for other researchers to use.
“This is a true research success story because not only could it lead to new technologies to aid decision-making in the clinic, but also is opening up new research avenues in more general restoration and reconstruction algorithms and applied mathematics,” said Sulam.