CHESTNUT HILL, Mass. (1/5/2026) – Obtaining clearer functional MRI data about the brain and its disorders is possible by using artificial intelligence, according to Boston University researchers. They recently reported in the journal Nature Methods that they developed an AI-assisted method to remove “noise,” or image distortion, caused by movement, heartbeats, and other factors.
Functional neuroimaging, also known as fMRI, is one of the most widely used non-invasive techniques in neuroscience, with tens of thousands of studies published in 2024 alone. A major hurdle in fMRI research is that MRI data about brain responses is mixed with noise from movement and other sources.
The study's lead author, Stefano Anzerotti, an associate professor of psychology at Boston University, said that filtering out noise more effectively could pave the way for new discoveries about the brain and its disorders. A new method developed by Anzellotti and two other researchers uses generative AI to triple the performance of previous approaches.
The discovery could open new doors in brain research, Anzerotti said.
“We wanted to improve denoising from fMRI data,” says Anzellotti. “We've tried this before in other studies. What's new about our study is that thanks to the use of generative AI, we were able to improve it by more than 200 percent over previous methods.”
The method developed by the researchers, known as DeepCor, outperforms other state-of-the-art denoising approaches on a variety of simulated datasets. On real fMRI data, DeepCor outperforms another widely used method known as CompCor by 215 percent at removing noise from facial responses and by 339 percent at disambiguating realistic synthetic data generated by mimicking the characteristics of real fMRI datasets, according to Anzerotti.
The AI learns unique patterns in brain regions that contain neurons, as well as unique patterns within brain regions that do not contain neurons. It's the same as the ventricles of the heart, Anzerotti said.
“Noise typically affects both sets of regions, so when you remove the patterns common to them, the unique patterns in the regions containing neurons become more noticeable,” Anzerotti said.
The team, which included postdoctoral researcher Idas Agrinskas and then-undergraduate student Yu Zhu, studied the human brain using functional magnetic resonance imaging.
Anzerotti said he doesn't expect much room for improvement.
“We were surprised by the magnitude of the improvement,” he said. “We expected the method to be even more effective, but expected the improvement to be in the 10 to 50 percent range. The 200 percent improvement exceeded our most optimistic expectations.”
Anzerotti's research will continue to explore improvements to fMRI measurements.
“We are considering two important steps: to make this method easily accessible to many other researchers, and to use it to remove noise from large public datasets so that the field can begin to benefit from cleaner data as soon as possible,” he said.
/Open to the public. This material from the original organization/author may be of a contemporary nature and has been edited for clarity, style, and length. Mirage.News does not take any institutional position or position, and all views, positions, and conclusions expressed herein are those of the authors alone. Read the full text here.
