The first human magnetic resonance imaging (Magnetic resonance imaging, a technique for viewing the stru...) scan was acquired almost 40 years ago. The scanner — hand-built by Dr. Raymond Damadian with the help of his two postdoctoral fellows — took nearly five hours to produce one snapshot of the human chest, and Dr. Damadian was eventually awarded the National Medal of Technology for his accomplishment.
We’ve come a long way since then, with MRI now giving us the ability to take high-resolution “pictures,” some on the scale of tenths of a millimeter. These remarkable technological advancements allow us to investigate the size, shape, and activity of specific parts of the brain.
Why are MRI brain scans important? Well, many researchers think that differences in the structure of the brain can provide clues as to why certain psychiatric illnesses develop. For instance, there have been clear associations between brain structure and schizophrenia, a finding that has been replicated across many different research groups. Despite the extraordinary capabilities of MRI, one recent review article by top neuroscience researchers at the Lieber Institute for Brain Development in Baltimore, MD calls into question the validity of MRI as a technique for psychiatry research.
The first thing to understand is the two major types of MRI used in neuroscience research:
Structural MRI (sMRI) is typically used to take one very high-resolution picture of the brain, which is done by measuring the magnetic properties of different brain regions, and then using that information to create an image. Researchers are able to look separately at the gray matter (which tends to contain more cell bodies) and white matter (which mostly contains axons, serving as pathways to connect different parts of the brain). On top of this, they can look at the size and shape of various brain regions that serve different functions (i.e., the size of the hippocampus, which is involved in memory, tends to be smaller in patients with Alzheimer’s disease).
Functional MRI (fMRI) is a way to measure activity in the brain, which can be inferred based on how much oxygen is in the blood near a particular brain region. When neurons in the brain are active, there is an increased demand for oxygen, which can be measured in the MRI scanner. It can then be deduced that a certain part of the brain is “in use” at a particular time. For example, it’s known that the A collection of nuclei found in the temporal lobe. The amygd... tends to be active when experiencing fear. Researchers can give study participants a specific task in the scanner (e.g., a memory test, in order to see which parts of the brain activate when storing new memories), or they can do a resting state scan, where participants lie in the scanner with no explicit task in order to see what their “natural” or “default” brain activity looks like.
Why is this useful? A popular line of research is to take patients with a psychiatric illness (e.g., schizophrenia or bipolar disorder) and compare the structure or function of their brains to that of healthy individuals. Differences identified by MRI could give insight into disease mechanism (e.g., a structural difference possibly responsible for symptoms) or identify biological markers (biomarkers) that guide patient treatment and inform prognosis.
Now that you have a primer on MRI research, let’s examine the criticisms of MRI raised by a recent paper.
“…the sounds made by an MRI scanner can be as loud as a power saw, an incredible distraction for even the healthiest of participants.”
One concern is that what we are seeing in an MRI image is actually a very zoomed-out version of the brain. One voxel (aka 3-dimensional pixel) can contain a million neurons. Drawing a conclusion about neurons within one voxel is an oversimplification, as we can assume that there’s a whole lot of variability in that one “tiny” voxel. Therefore, when studies draw specific functional conclusions from MRI studies, we need to be wary about what those conclusions mean. It’s important to note that MRI technologists are eagerly trying to develop ultra high-resolution scanners that would help to combat this problem.
Second, the authors argue that differences in the brain between a patient with psychiatric illness and a healthy individual might actually be due to confounding factors. For example, individuals with high levels of anxiety may also have hypertension, which could affect metabolism in the brain. Because of this, we cannot necessarily say that a brain difference between two patient groups is due to the psychiatric illness itself – rather it may be due to another underlying or unrelated factor. Studies that control for these confounding factors are warranted, though this is often hard to accomplish given the high rates of other conditions in psychiatric illness.Importantly, this criticism of confounding factors in psychiatric disease can be applied towards other types of biomedical research, and is not limited to MRI.
Another important consideration to make when evaluating psychiatric MRI research is head motion. Numerous studies have investigated how moving one’s head in the MRI scanner can actually change MRI results. Think about the scanner like a camera taking a picture of an object. If you try to take a picture of a moving car with a camera, the image of the car is probably going to show up blurry. This is just like when somebody moves while in the MRI scanner: if the person moves around during the scan, their final image could become distorted and blurred. In fact, because the scanner only takes one picture per second, even slow motion can be a problem! You can probably imagine that someone with a mental illness that causes irritability or hyperactivity (e.g., a child with ADHD) might be more inclined to squirm around in the scanner. If higher levels of head motion are found in certain patient groups, this may complicate matters when trying to look at brain differences between patients and controls. Many research groups are actively trying to reduce the effect of motion on on MRI results, but this is still a major obstacle in MRI research.
“While a healthy individual may be able to relax and ‘space out,’ one with a psychiatric condition may feel a sense of panic, loneliness, or – you guessed it – the urge to fidget…”
Furthermore, patients with psychiatric illnesses may have very different personal experiences while in the MRI scanner as compared to controls. For instance, in resting state fMRI studies (described above), participants are instructed to lie down and “relax” in the scanner, with no visual or auditory cues surrounding them. While a healthy individual may be able to relax and “space out,” one with a psychiatric condition may feel a sense of panic, loneliness, or – you guessed it – the urge to fidget and move his or her head around. On top of this, the sounds made by an MRI scanner can be as loud as a power saw, an incredible distraction for even the healthiest of participants. Not only might this affect head motion, but cognitive performance as well. These potential confounds make it very tricky to compare brain activity across patient groups, when the experience of being in the MRI scanner might be so fundamentally different between patient groups and controls.
Finally, because the number of voxels or data points in a scan is vast, statistically analyzing MRI data is extremely difficult. In fact, a recent study revealed that software commonly used to draw statistical conclusions from fMRI yields many false alarms or erroneous conclusions, thus drawing into question thousands of published fMRI papers.
These are just some issues that may exist in MRI brain research. They do not invalidate all research that has been done in the past, but they have some important consequences when moving forward. Experts in MRI research and methodology should consider these when designing studies and analyzing their data.
A closing thought on the future of MRI research: One of the most exciting directions in the field is in the opportunity for predictive power of MRI in psychiatry. For instance, studies that can detect brain-related biomarkers (i.e., measurable indicators of the presence of a disease state) before the onset of the disease itself have powerful clinical implications that can make a major difference in the world of psychiatry. One example is the ability to measure changes in the structure of the brain before the onset of Alzheimer’s disease. Even if these MRI-related biomarkers do not allow us to draw direct inferences on the microscopic scale of the brain, they may still be meaningful. If these biomarkers are able to successfully predict who is at high-risk for any given psychiatric disease, we can introduce preventative and early intervention techniques to potentially offset some of the effects of the impending disease.
Image by Jooyeun Lee.
Weinberger, D. R., & Radulescu, E. (2015). Finding the elusive psychiatric “lesion” with 21st-century neuroanatomy: a note of caution. American Journal of Psychiatry, 173(1), 27-33.
Shin, Lisa M., and Israel Liberzon. “The neurocircuitry of fear, stress, and anxiety disorders.” Neuropsychopharmacology 35.1 (2010): 169-191.
Schuff, N., Woerner, N., Boreta, L., Kornfield, T., Shaw, L. M., Trojanowski, J. Q., … & Disease Techniques for viewing the brain and its activity, especiall... Initiative. (2009). MRI of hippocampal volume loss in early A neurodegenerative disease characterized by neuronal loss i... in relation to ApoE genotype and biomarkers. Brain, 132(4), 1067-1077.
Buckley, P. F. (2005). Neuroimaging of schizophrenia: structural abnormalities and pathophysiological implications. Neuropsychiatr Dis Treat, 1(3), 193-204
DeLisi, L. E., Szulc, K. U., Bertisch, H. C., Majcher, M., & Brown, K. (2006). Understanding structural brain changes in schizophrenia. Dialogues in clinical neuroscience, 8(1), 71.
Haijma, S. V., Van Haren, N., Cahn, W., Koolschijn, P. C. M., Pol, H. E. H., & Kahn, R. S. (2013). Brain volumes in schizophrenia: a meta-analysis in over 18 000 subjects. Schizophrenia bulletin, 39(5), 1129-1138.
Glahn, D. C., Laird, A. R., Ellison-Wright, I., Thelen, S. M., Robinson, J. L., Lancaster, J. L., … & Fox, P. T. (2008). Meta-analysis of gray matter anomalies in schizophrenia: application of anatomic likelihood estimation and network analysis. Biological psychiatry, 64(9), 774-781.
Wachinger, C., Salat, D. H., Weiner, M., Reuter, M., & Alzheimer’s Disease Neuroimaging Initiative. (2016). Whole-brain analysis reveals increased neuroanatomical asymmetries in dementia for Structure in temporal lobe that has many functions but is es... and amygdala. Brain, 139(12), 3253-3266.
Eklund, A., Nichols, T. E., & Knutsson, H. (2016). Cluster failure: why fMRI inferences for spatial extent have inflated false-positive rates. Proceedings of the National Academy of Sciences, 201602413.