Psychiatry and Neuroradiology

[Symmetry11]

Is it possible to do a combo like this in order to better understand the physical basis of mental illness?

---

Members don't see this ad.

 

[675015]

Is it possible to do a combo like this in order to better understand the physical basis of mental illness?

As a researcher very possible. As a clinician very difficult.

 

[hamstergang]

Is it possible to do a combo like this in order to better understand the physical basis of mental illness?

Even if entering such a combined field were possible, I don't know how much that would help you better understand mental illness. I just had a grand rounds today where the speaker presented data on fMRI studies and mental disorders. To make it grammatically simpler to write, the following is her conclusion and not necessarily facts I can vouch for:
You can find correlations between various brain regions and various psychiatric disorders. However, no region is specifically involved in just one disorder, and no disorder involves specifically one region. So we don't have any evidence of causality. And the lack of specificity means that it's not terribly helpful to differentiate the various disorders when discussing issues in these implicated brain regions.

I guess the field is still advancing and there are more imaging techniques available and being developed, but as of now it's easy to use neuroradiology to make correlations that are not really relevant.

 

[Nasrudin]

I'm not sure what knowing ever more sophisticated geography of the brain in relation to patterns of behavior, or more absurdly its meaning, is supposed to tell us. In terms of what to do. How to help people.

I'm not ruling out that maybe...it might lead somewhere. Neural network redesign with microsurgery?

It seems to me biological determinism is fundamentally misguided. As an intention or a project. I think it's more likely we'll merge with AI's before brain mapping and redesign becomes viable. I think we should be more focused on how to influence AI culture and inculcate it with the value systems of ecological maintenance and mindfulness. Or else face extinction or enslavement. The creation of intelligent self-design procedures for our brains seems like putting the cart before the horse at that point.

 

[splik]

the underlying assumption is that there is physical basis to mental illness, when historically psychiatrists have dealt with that which we cannot see, that appears not to have any physicality. Much of the imaging research is akin to the new phrenology, and none of this research, on which we have sqaundered so much money on - and usually the studies are not statistically powered to draw meaningful conclusions - has altered clinical practice. You can certainly learn to do interpret neuroimaging and become a researcher in this field but it's unlikely to change your practice unless you are snake oil salesman or work for the Amen Clinics.

That said, I do believe that psychiatrists should have a good grasp of the functional neuroanatomy of emotion and cognition, and familiarity with these brain circuits. Psychiatrists should be familiar with appropriately ordering imaging and neurophysiological investigations including MRI, SPECT, FDG-PET, PIB-PET, DAT scans, and EEGs and should have some familiarity to know what to look for and have some basic ability to look at these scans (particularly an MRI) and identify things like tumors, strokes, cerebrovascular disease and other white matter changes like MS or CADASIL, grey matter disease (like viral encephalitis), global and focal cortical atrophy, and pathology suggestive of Alzheimers, FTD, PSP ("hummingbird sign"), MSA ("hot cross bun sign"), CJD (cortical ribboning, and "hockey stick sign") etc - basically the neurological diseases that can present with psychiatric symptoms. You can get this experience through neurology rotations, doing neuroradiology electives, and ordering imaging on your patients where appropriate and going to the neuroradiology reading room and discussing the images with the neuroradiologists who are only too happy to talk with you.

 

[Jorje286]

Right now it's only useful if it's related to research. A profound understanding of radiology is certainly very helpful. But it's not worth doing a radiology residency and a psychiatry residency for that. You can do a psychiatry residency and a research fellowship in neuroimaging.

Correlations are important, even if we don't establish causative mechanisms. A lot of evidence-based medicine is simply finding consistent and clinically-relevant correlations. Check this paper out for example on recent important advances: http://www.ncbi.nlm.nih.gov/pubmed/26725338 . One of the problems in finding consistent correlations however is the heterogeneity of the disorders and there are attempts to go around that by looking at things from a symptom level first rather than DSM classifications. There are other issues as well (small sample sizes, the fact that most negative findings aren't reported to be included in metanalyses, sketchy scientific behavior...etc).

Elaborating causative mechanisms simply from imaging is a tall order. We need to bridge the gaps between multiple levels of understanding. Given that the brain is probably the most complex organ we know of, our tools are yet simply not sophisticated enough for that task (in addition to the facts that animal models of psychiatric illnesses are extremely difficult to come by and ethical challenges in research on human subjects) .

 

[DD214_DOC]

Even if entering such a combined field were possible, I don't know how much that would help you better understand mental illness. I just had a grand rounds today where the speaker presented data on fMRI studies and mental disorders. To make it grammatically simpler to write, the following is her conclusion and not necessarily facts I can vouch for:
You can find correlations between various brain regions and various psychiatric disorders. However, no region is specifically involved in just one disorder, and no disorder involves specifically one region. So we don't have any evidence of causality. And the lack of specificity means that it's not terribly helpful to differentiate the various disorders when discussing issues in these implicated brain regions.

I guess the field is still advancing and there are more imaging techniques available and being developed, but as of now it's easy to use neuroradiology to make correlations that are not really relevant.

fMRI is old hat. The speaker should have updated their talk with studies that look at both structural and functional connectivity and how comparing both types under the same testing conditions allows us to map out neural networks responsible for specific functions.

 

[clausewitz2]

fMRI is old hat. The speaker should have updated their talk with studies that look at both structural and functional connectivity and how comparing both types under the same testing conditions allows us to map out neural networks responsible for specific functions.

Yeah, comparing activation across individual regions in different conditions in a functional imaging study is very 2004.

Edit: unless it's combined fMRI/EEG, where the highly detailed time course information you get might still make plain ol' ROI analysis useful.

 

[OldPsychDoc]

Well, there's always this guy.
He did it.

 

[thoffen]

Not sure what is so uncomfortable about being a psychiatrist that we feel at least in part invalid in doing our jobs unless we have some wholly tangible marker that our approach is scientifically credible.

I am sure, though, that the more important thing when you recognize this is not to pursue that marker without either or also working on your discomfort with its (at present) non-existence. In truth, I'm highly pessimistic of its ever-existence.

 

[MacDonaldTriad]

I think functional imaging is the new phrenology because of the concept of degrees of freedom. How many voxels do you think there is in an fMRI? It would probably be easier to win the lottery than to not find any differences in a subtraction. I also like Clausewitz's "comparing activation across individual regions in different conditions in a functional imaging study is very 2004." comment. I think my first SPECT study was in 1989. I had a lot of optimism then. I can not tell you how over functional imaging I am now.

 

[OldPsychDoc]

I think functional imaging is the new phrenology because of the concept of degrees of freedom. How many voxels do you think there is in an fMRI? It would probably be easier to win the lottery than to not find any differences in a subtraction. I also like Clausewitz's "comparing activation across individual regions in different conditions in a functional imaging study is very 2004." comment. I think my first SPECT study was in 1989. I had a lot of optimism then. I can not tell you how over functional imaging I am now.

I recall an attending referring to it as "pseudocolor phrenology" back in 2002 or so.

 

[clausewitz2]

I think functional imaging is the new phrenology because of the concept of degrees of freedom. How many voxels do you think there is in an fMRI? It would probably be easier to win the lottery than to not find any differences in a subtraction. I also like Clausewitz's "comparing activation across individual regions in different conditions in a functional imaging study is very 2004." comment. I think my first SPECT study was in 1989. I had a lot of optimism then. I can not tell you how over functional imaging I am now.

I definitely agree with you in large part, but there is a responsible way to do it. From a tangentially related field, speech production, we have the DIVA model, which is sort of how to do functional neuroimaging right:

http://www.bu.edu/speechlab/research/the-diva-model/

Note that this case has several things generally lacking: a computationally explicit model of the phenomenon in question, a simulation of outputs of the system, and, perhaps most crucially, very detailed time course predictions testable via EEG/MEG. Saying "Region X and Y are involved in this process" is uninteresting; saying "Region X is active AND THEN Region Y is active" allows for a much more restricted (and thus interesting) hypothesis space.

In that framework, I think functional imaging is super useful, but you need the model with discrete functional bits to map on to what different regions might be doing. Additionally, the processing unit for the brain for many relevant things is probably the hypercolumn, but we have no way to record from those at present in living humans, so you will always be limited.

EDIT: Oh yes, also - carefully designed behavioral tasks that you can do in a scanner! The DIVA people did crazy things like putting balloons in peoples' mouths that would suddenly inflate during a speech production task, which it turns out gives you a really excellent look at the neural substrates of suddenly having to come up with a compensatory articulation and process massive speech error signals.

 

[MacDonaldTriad]

As the subject is being slid into the scanner:

Functional imaging researcher: “I need you to count backwards from 100 by 7s, but whatever you do, don’t have an erotic thought….”

I once had a patient who was surprised that we couldn’t tell what he was thinking during his scan. Even after reading and discussing the consent, he just assumed that was what we were doing. In hind sight, clearly his consent was not as informed as it could have been or at least as we thought it was.