Press Release

New AI-driven initiative could optimize brain stimulation for treatment resistant depression

Uniting experts in artificial intelligence, psychiatry, and neuroimaging, the ENIGMA Consortium will apply AI to brain scans of people with depression in the first global analysis for new ways to predict which patients will respond best to a promising new treatment.

Sidney Taiko Sheehan February 14, 2024
Illustration of a machine learning network used in neuroimaging.

Illustration of a machine learning network used in neuroimaging. The Stevens INI is using artificial intelligence methods for improved prediction of treatment response. Image: USC Stevens INI

About a third of people worldwide will be clinically depressed during their lifetime, making major depression the most prevalent mental health condition. Major depressive disorder (MDD) is the leading cause of disability worldwide, and around half of MDD patients have treatment-resistant depression. As the co-founder and leader of the ENIGMA Consortium, which coordinates the world’s largest neuroimaging studies in psychiatry and neurology, Paul M. Thompson, PhD, recognizes an urgent need to find better treatments for people suffering from depression.

“The ENIGMA Consortium has studied depression and over 30 other brain diseases since 2009. By analyzing brain scans from all over the world, we found structural and functional brain differences in people with depression. Recently, one of our collaborators, Dr. Roberto Goya-Maldonado, reached out with the idea of applying artificial intelligence to these brain scans to identify patients who may benefit the most from a promising new treatment using repetitive transcranial magnetic stimulation (rTMS),” explains Dr. Thompson, who is also the associate director of the USC Mark and Mary Stevens Neuroimaging and Informatics Institute (Stevens INI) at the Keck School of Medicine of USC.


Illustration of brain stimulation
The Stevens INI uses brain scans to predict who will respond best to brain stimulation and to find new ways to optimize the stimulation treatment. Image: USC Stevens INI


Repetitive transcranial magnetic stimulation (rTMS) is a noninvasive, FDA-approved procedure that stimulates the brain’s nerve cells with magnetic fields to alleviate symptoms of MDD. rTMS is generally used when other treatments have failed. During rTMS sessions, an electromagnetic coil is placed against the scalp. This coil delivers magnetic pulses that stimulate nerve cells in brain regions involved in mood control and depression.

Brain stimulation with rTMS has gained popularity in recent years for the treatment of depression, with proven clinical benefits. Multiple studies show it reduces suicide and suicidal thoughts and behaviors. However, only 40-50% of patients respond to rTMS, so there is great interest in predicting who responds best and which brain features predict whether a patient will respond.

“rTMS is a novel treatment that offers hope when other treatments have failed. If someone is clinically depressed, a psychiatrist may prescribe medication, but for many people, medication does not work. Our study uses brain scans to determine if a treatment like rTMS will likely help,” says Dr. Thompson.

Dr. Roberto Goya-Maldonado, a psychiatrist and neuroimaging specialist at the University Medical Center Göttingen in Germany, emphasized the importance of the study. “Identifying how depressed a person is and whether or not this person can benefit from rTMS is a task achievable by powerful artificial intelligence methods,” said Dr. Goya-Maldonado. “With data from up to 60,000 participants, our research groups will investigate how clever combinations of different brain scan protocols could predict treatment response most simply and accurately.”

Now, with new funding from the National Institute of Mental Health, Drs. Thompson and Goya-Maldonado are forging a new path forward with their study. They are applying artificial intelligence methods to extract features from brain images to predict clinical status and treatment response. Using ENIGMA’s immense data repository, these artificial intelligence methods will be tested on diverse test data from globally representative populations.

“There are so many thrilling opportunities with this study. The first practical step is to determine if rTMS may help an individual based on their brain imaging, but the study will also optimize how rTMS is used—how often, for how long, and on which regions of the brain, for example. Additionally, by learning how rTMS helps with depression, we can see how it could help in people with mild cognitive impairment or dementia, which are key areas of focus at the Stevens INI. This is the first step towards a larger international study. We aim to provide treatment and relief for people with depression and work out how targeted brain stimulation may help for other disabling conditions,” says Dr. Thompson.

The Stevens INI maintains the world’s most extensive collection of clinical brain scans to train machine learning methods, comprising imaging, genetics, clinical, cognitive, and electrophysiology data. Their researchers create software and computing resources to streamline disease diagnosis and treatment selection based on brain scans. Their algorithms are applied to worldwide medical data to discover factors influencing brain development, aging, and disease.

“The ability to begin a treatment plan for depression equipped with data that says you are choosing the option most likely to help is incredibly powerful. The trial and error of seeking treatment, particularly for mental illness, can be daunting and complicated by socioeconomic issues and barriers to health care,” says Stevens INI Director Arthur W. Toga, PhD. “Dr. Thompson’s study exemplifies our institute’s strengths of using neuroimaging and informatics in tandem, using brain scans and AI to show the potential benefits of a specific clinical treatment. This is translational medicine at its best.”

This research is supported by the National Institute of Mental Health (R01MH131806) and will also support neuroscientists at Cornell University and the University of Illinois, Chicago.