Science and artificial intelligence combined at the Medical University of South Carolina in a study that could lead to personalized repetitive transcranial magnetic stimulation, or rTMS, for smokers who want to quit.
"We want to improve the effectiveness and specificity of rTMS and reduce side effects," said study leader Xingbao Li, M.D. He's an associate professor in the Department of Psychiatry and Behavioral Sciences who has done extensive research on TMS.
His team published its findings in the journal Brain Connectivity.
TMS uses electromagnetic pulses to affect brain activity and may be best known for its role in treating depression and obsessive-compulsive disorder. Side effects include discomfort at the stimulation site and headaches.
TMS has also been approved by the Food and Drug Administration for smoking. MUSC Health was the first place in South Carolina to offer TMS to smokers. Research has shown that multiple sessions of rTMS, specifically over the left dorsolateral prefrontal cortex of the brain, can cut cravings and cigarette consumption.
The new MUSC study gets even more targeted than that, using a form of AI called machine learning to analyze images from the brain's neural networks to see if it's possible to predict which smokers are likely to benefit from multiple sessions of rTMS, also known as repetitive TMS.
To do that, the researchers used functional magnetic resonance imaging, or fMRI, to detect changes in blood flow to measure brain activity. They looked at neural networks when participants were in a resting state, relaxed with their eyes closed, and when they were exposed to smoking photos.
That analysis found that one neural network stood out: the salience network. It filters information to determine what's salient, or important, to focus on. In the study, connectivity in the salience network was the best predictor of rTMS effectiveness.
"The study gives us a roadmap to extend personalized rTMS and build an fMRI and multimodal biomarker pipeline. The methods can be used for other substance use disorders as well," Li said.
"Historic studies focus on the reward network in cigarette smokers," he continued, referring to the parts of the brain involved in motivation and pleasure.
"We were surprised to find that the salience network plays such a crucial role in smoking behavior. This makes the salience network a mechanistic bridge between rTMS neuromodulation and successful smoking cessation."
They found that bridge with the help of machine learning. In machine learning, computers analyze and learn from data without being programmed to do so. They use algorithms that can spot statistical patterns and adapt to them. That, in turn, lets researchers automate that part of their work and improves accuracy.
In this case, machine learning analyzed data collected during an earlier MUSC study on TMS in smokers.
Here's how that earlier study was set up. The researchers recruited 42 people who wanted to quit smoking. They were split into two groups. One group got real TMS. The other got sham TMS that felt real. They all spent a minute-and-a-half before each TMS session interacting with things like cigarettes and ashtrays, then during the TMS, real or sham, watched videos of people smoking. There were 10 sessions per person over a two-week period.
In the end, the researchers found participants who got the real TMS "smoked significantly fewer cigarettes per day during the two-week treatment," were more likely to quit by their target date and had lower cravings for tobacco.
Li said thanks to the fMRI scans that were also part of that work, the newer study was able to build on its findings. "Using machine learning to identify an individual's dysfunctional brain network and then applying rTMS to the dysfunctional network, we can select who prefers to use rTMS or who prefers medicine to help them stop smoking."
The study was supported by a grant from the National Institutes of Health. The authors didn't report any conflicts of interest. The research team includes Kevin Caulfield, Ph.D.; Andrew Chen, Ph.D.; Christopher McMahan, Ph.D.; Karen Hartwell, M.D.; Kathleen Brady, M.D., Ph.D.; and Mark George, M.D.
Li said their relatively small study lays the groundwork for larger studies to further explore targeted TMS for smokers. "This demonstrates that MUSC researchers can use novel and high-impact technology to move beyond fixed-target stimulation into precision neuromodulation."
