In the near future, rapid brain scans may be needed to identify the best treatments for diagnosing depression, according to a new study from Stanford Medicine.

Researchers found that brain imaging combined with machine learning could detect different types of depression and anxiety.

The study, published in Nature Medicine, identifies six biological subtypes, or “biotypes,” of depression, and describes which treatments may be most effective for three of these subtypes.

Current methods of treating depression often rely on trial and error, which can be time-consuming and frustrating.

About 30% of people with depression do not respond to multiple treatments, and up to two-thirds do not have their symptoms completely relieved. One reason for this is that there is no reliable way to know which treatment will be effective for each patient.

“The goal of our work is to get the right results the first time,” said Lynne Williams, senior author of the study and a professor at Stanford Medicine. Williams, who lost her partner to depression in 2015, focuses on precision psychiatry.

To better understand the biology of depression and anxiety, Williams and his team used functional MRI (fMRI) to scan the brains of 801 participants suffering from these conditions.

They analyzed brain activity at rest and during tasks and tested cognitive and emotional functions.

Using a machine learning technique called cluster analysis, they identified six distinct brain activity patterns.

They also tested the responses of 250 participants to three common antidepressant medications or behavioral therapy.

They found that patients with hyperactivity in cognitive brain areas responded best to the antidepressant drug venlafaxine.

People with higher relaxation activity in three areas associated with depression and problem solving benefited most from behavioral therapy.

Another group, which had lower resting activity in the attention-control circuits, was less likely to benefit from talk therapy.

The biotypes and their treatment responses align with known brain region functions, said Jun of the University of Illinois Chicago, a co-author of the study. Understanding these biotypes could lead to more precise treatments, possibly starting with medication to enhance therapy effectiveness, he explained.

Williams’ team also demonstrated that the use of fMRI improves prediction of antidepressant response.

By identifying a subtype called cognitive biotype, they predicted improvement in 63% of patients using fMRI, compared to 36% without it.

Different biotypes also correlate with differences in traits and task performance. For example, people with overactive cognitive areas had higher levels of anhedonia and performed worse on executive function tasks.