Brain Scans Cannot Differentiate Between Mental Health Conditions

A new study, published in the journal Human Brain Mapping, questions previous findings that specific brain regions are implicated in particular mental health conditions. Instead, according to the researchers, biased study design and the difficulty of publishing negative findings may have led to inaccurate results. While the researchers did find some differences in brain activation between people with mental health conditions and people without mental health conditions, they were not able to discriminate between specific diagnoses. The current study suggests that there are few, if any, differences in brain regions activated by specific mental health conditions. That is, there is still no brain scan that can tell whether a person has depression, social anxiety, or schizophrenia, for example.

Researchers have theorized that the different symptom clusters that form mental health diagnoses are linked to specific regions of the brain. If confirmed, such a finding would suggest that mental health diagnoses have biological components that could be targeted medically. However, the finding of the current study undermines this theory. Instead, the results indicate that while there is a general tendency for large parts of the brain (such as the amygdala and the hypothalamus) to be activated in a number of mental health conditions (as well as when humans are under stress in a number of ways), there is little difference between the varying diagnoses—even for diagnoses as seemingly different as social anxiety, depression, and schizophrenia.

The researchers were led by Emma Sprooten (Department of Psychiatry, Icahn School of Medicine at Mount Sinai, New York City). They used statistical tests to combine the results from 547 studies, which enabled them to analyze the data from 21,692 participants. The studies compared the brain scans of healthy participants with participants who were diagnosed with major depressive disorder, bipolar disorder, schizophrenia, obsessive compulsive disorder (OCD), and anxiety disorders, including social anxiety disorder, generalized anxiety disorder, panic disorder, specific phobias, and post-traumatic stress disorder (PTSD).

The studies in question used functional magnetic resonance imaging (fMRI), a common type of brain scan which creates images based on blood oxygenation levels within the brain. Higher blood oxygenation levels are assumed to indicate areas involved in more activity. Thus, an fMRI result is theorized to indicate which areas of the brain are activated or deactivated for particular tasks or states of being.

Importantly, fMRI has endured its own questions of bias. A recent article, published in the Proceedings of the National Academy of Sciences, confirmed a previous finding that up to 70% of the results in fMRI studies may actually be “false positives”—that is, finding a result when there actually is none. Nikos K. Logothetis wrote, in a 2008 article in Nature, that the fMRI “is an excellent tool for formulating intelligent, data-based hypotheses, but only in certain special cases can it be really useful for unambiguously selecting one of them, or for explaining the detailed neural mechanisms underlying the studied cognitive capacities.” That is, fMRI results can inform the questions we ask, but they can rarely answer those questions. Unfortunately, the neuropsychiatric literature is riddled with fMRI studies that purport to do just that.

Another recent study attempted to showcase just how much fMRI results rely on subjective interpretation. The researcher, Joshua Carp of the University of Michigan, examined a single fMRI event and found that there were 34,560 different results that could be reached by following different analysis procedures. He argues that the choice of analysis procedure is a subjective one, and researchers may try numerous procedures in order to achieve a positive result. He suggests that in the future, researchers must clearly specify which procedure they will use in order to reduce this extraordinary bias.

Sprooten and her colleagues framed their results as addressing the common practice of “reverse inference,” which has been challenged by other researchers as well. In reverse inference, researchers pre-select which brain regions (ROIs) they are going to study in order to maximize potential results—rather than examine the whole brain to determine which areas are activated. Put simply, if you study a particular area, then you will never see if there is activation in other brain regions during your test. You will only find activation in your pre-selected area. This result is often taken to indicate that particular disorders are associated with activation in particular regions—but this conclusion rests on the assumption that researchers would not have found other areas had they examined the whole brain.

The strength of the current study was its ability to compare ROI studies (studies that focused on only specific regions of the brain) with the results from whole-brain studies. The ROI studies tended to find differences in which brain regions were activated by different mental health conditions. However, once the whole-brain studies were factored in, these findings disappeared. When all studies were included, there were no differences between the diagnoses.

Notably, the researchers only included studies that found significant results—that is, those that purported to find differences between those with mental health diagnoses and those without. Their results would likely be even more striking if they factored in the studies with negative results—studies that did not find differences.

Sprooten writes:

“The pre-selection of ROIs, possibly in combination with the difficulty of publishing negative results, seems to bias the literature and may indirectly lead to oversimplification and over-localization of neurobiological models of behavior and symptoms.”

Choosing a brain region to examine, rather than examining the whole brain, appears to lead to biased, oversimplified results. Likewise, the conclusion that Logothetis reaches in his Nature article is that “the limitations of fMRI are not related to physics or poor engineering, and are unlikely to be resolved by increasing the sophistication and power of the scanners; they are instead due to the circuitry and functional organization of the brain, as well as to inappropriate experimental protocols that ignore this organization […]The magnitude of the fMRI signal cannot be quantified to reflect accurately differences between brain regions, or between tasks within the same region.”

The study conducted by Sprooten and her colleagues suggests that many fMRI studies misrepresent the abilities of brain scans. As Logothetis argues, using fMRI results to confirm pre-existing theories of brain region activation in mental health diagnoses is in direct contradiction of the abilities of the fMRI technology. In short, brain scan research is of limited use in explaining the complex psychological states of human beings. If a neurological answer seems clear and easy, it may be being misrepresented and oversimplified.

 

****

Sprooten, E., Rasgon, A., Goodman, M., Carlin, A., Leibu, E., Lee, W. H., & Frangou, S. (2016). Addressing reverse inference in psychiatric neuroimaging: Meta-analyses of task-related brain activation in common mental disorders. Human Brain Mapping. doi:10.1002/hbm.23486 (Abstract)