All the brain imaging techniques I know fall into two categories:

  1. Tracking blood

    1. Either by looking at the magnetic (fMRI), or near-infared absorption (diffuse optical imaging, NIRS) properties of hemoglobin, or

    2. Injecting tracers into the blood stream (PET, SPECT)

  2. Tracking electric and magnetic fields in neurons (EEG, MEG).

The problem with the first approach is that it is fundamentally secondary in nature: it tracks the response of the vascular system to brain activity. Although the BOLD signal correlates well with neural activity, it is still an indirect indicator of firing rate, has time lags, and limited spatial (limited by distribution of capillaries) and temporal (limited by flow rate) resolution.

The second approach is better, but still not perfect from a pharmacology perspective. Although it gives us good information about the post-synaptic potential (EEG) or intraneuron potential (MEG), it still provides no information on specific neurotransmitters released. The approach also suffers from limitations on spatial resolution and the fact that EM-fields follow the superposition principle and thus it is mathematically impossible to reconstruct signals that interfere destructively outside the neuron but before the sensors.

The perfect approach for a pharmacologist would be to track individual types on neurotransmitters. Are there imagining techniques that use the resonance, absorption, or other properties of neurotransmitters (instead of hemoglobin)? Or is there a fundamental reason why this won't work?

I know that the same exact techniques as for hemoglobin won't work (since as far as I know most neurotransmitters don't contain nice things like iron to play with), but is there a reason other resonance or scattering techniques wouldn't work?

The closest work I know in this direction is the event-related optical signal (Gratton & Fabiani, 2001) because it tracks the light-scattering properties of something other than hemoglobin: neural tissue. Unfortunately, this is still tracking activity of the neuron and not neurotransmitters.

  • $\begingroup$ @ChuckSherrington sure, although preferable non-invasive ones and not ones where we mess with genome to insert weird trackers. However, if that is the only alternative then better than no answer! $\endgroup$ Commented Jun 19, 2012 at 3:33
  • $\begingroup$ Honey & Bullmore (2004). Human pharmacological MRI, Trends in Pharmacological Sciences, 25, 366-374. @ tinyurl.com/8mw4mtx $\endgroup$
    – Jeff
    Commented Oct 14, 2012 at 22:13

3 Answers 3


Resonance methods can be used to measure neurotransmitter levels. Magnetic Resonance Spectroscopy can measure the levels of a large number of neurotransmitters. However, this has always been viewed as a fairly static measure of neurotransmitter levels, and so has not been widely used as a measure of neural activity. However, Paul Mullins at Bangor University has been doing some work using spectroscopy to measure neurotransmitter levels in different parts of the brain. From the work I have seen him present so far, it is sensitive enough to track stimulus driven changes in neurotransmitter levels with a standard blocked design (ie 18sec blocks of stimulation followed by 18sec of rest). I'm not too sure whether it has the same functional contrast-to-noise as the BOLD response and it definitely has a lower spatial resolution than BOLD techniques. This work is currently being prepared for publication so I shouldn't say too much.


Novotny, E. J., Fulbright, R. K., Pearl, P. L., Gibson, K. M., & Rthman D. L. (2003). Magnetic resonance spectroscopy of neurotransmitters in human brain, Annals of Neurology, 54, S25-S31. PubMed.

  • $\begingroup$ I like this answer, will take a look at the paper once I am behind paywall. Is there any fundamental constrain on improving the spatial resolution, or is it low simply because they have not had the time to develop better resolution techniques for this approach? $\endgroup$ Commented Jun 19, 2012 at 14:30
  • $\begingroup$ I just went through the paper, and I was saying in chat I wasn't aware they'd gotten that far with it in the early 2000s. It does say that you can really only get glutamate, glycine, and maybe GABA, and only in coarse resolution. Has that improved with stronger magnets over time? Having a grip on glutamate concentration doesn't buy you too much, it's like saying that a crowd at a sporting event is 'a lot of people'. $\endgroup$ Commented Jun 19, 2012 at 14:56
  • $\begingroup$ Yeah, I think it has gotten somewhat better over the years though I'm not too sure - my memory of Paul's presentations is sketchy. I do know that some neurotransmitters can't be distinguished because they are too similar spectrally. I was very surprised since I hand't heard much about it before - it seems to be a very under studied area. I'll update this response once Paul publishes to include more details. $\endgroup$
    – Bronson
    Commented Jun 19, 2012 at 20:06
  • $\begingroup$ Current resolution is 5mm so it is not that far off EPI resolution anyway. I have spoken to Paul about resolution and he seemed to suggest that it could be improved. Besides it is always possible to improve resolution if you are happy having longer volume acquisition times, or reduced coverage of the cortex. $\endgroup$
    – Bronson
    Commented Jun 19, 2012 at 20:14
  • $\begingroup$ Thanks for the answer @Bronson, it would be awesome if you included information from some of the comments you made into the answer for completeness, and even better if you can convince Paul Mullins to join the site ;). $\endgroup$ Commented Jun 23, 2012 at 4:42

I found an example of a system that researchers are aiming to use in the future for determining the level of neurotransmitter activity in the brain using MRI. You were on the right track with the utility of hemoglobin. The molecule used is somewhat similar. To understand how the probes were generated requires a bit of a biological detour.

There are enzymes in the body known as P450 cytochromes, so named because they were observed as absorbing light in the 450 nm range. They are membrane bound (e.g., in the mitochondria) and one of their primary purposes is to carry out monooxygenase reactions, and liver-based P450s are responsible much of the drug metabolism that goes on in the body. So, different classes of these enzymes have a high affinity for certain molecules. The most important characteristic that the P450s possess is a heme group, which is paramagnetic, and will produce a signal in an MRI.

These P450 enzymes are also found in bacteria, which can be readily used as a test bed due to their rapid cell division. What these scientists have done is repeatedly mutated bacteria and tested the affinity of the P450s of the resulting mutants for various molecules (in this case serotonin and dopamine).

So, in retaining a culture of the bacteria, scientists can mass produce the mutated heme groups from the bacterial P450s. Presumably once the sequence of the subunit is determined, it could be created using standard techniques.

Some challenges remain. Primarily, the probes will have to be checked for safety, and measures will have to be taken to make them into robust MRI contrast agents, and ensure that they can localize and persist in the brain for the duration of testing. Presumably, the proper correlations between strength of signal and local transmitter concentration can be established with in vitro studies. It sounds like we're fairly close to non-invasive methods of measuring neurotransmitter levels.

enter image description here

Reproduced from here

Brustad, E.M., Lelyveld, V.S., et al (2012). Structure-guided directed evolution of highly selective P450-based magnetic resonance imaging sensors for dopamine and serotonin. Journal of Molecular Biology, in press, doi.


It's a bit of an art, currently. Following is one technique I witnessed in a lab that takes electrophysiological recordings of tadpole and rat neurvous sytems.

The lab that I worked with entrains the neuron, recording it's electrical activity (in a series of drug tests) and injects a marker that goes into the neuron (a GFP-like protein that binds to precursors for serotonin or dopamine or GABA and glows in pictures). Then they mark the brain tissue and the recording so they'll be associated, then after they have a bunch of recordings, they look at all the brain tissues and see which ones expressed the dopamine, GABA, or serotonin.

Now they can associate electrophysiological behavior with a particular neurotransmitter.

In addition to this, there's a general neurochemical anatomy that's known. Serotonergic neurons are only found in the brainstem, for instance and dopamine is only in the substantia nigra pars compacta, ventral tegmental area, and hypothalamus and has four different types of projections to other cell bodies.

Of course, for proof, staining (as described above) is still preferred by reviewers.

  • $\begingroup$ Staining only works at autopsy with humans ;) $\endgroup$ Commented Jun 19, 2012 at 8:27
  • $\begingroup$ @ChuckSherrington I would consider the Wada test also staining (but you don't get much detailed information out of it, I admit). $\endgroup$
    – Mien
    Commented Jun 19, 2012 at 8:33
  • $\begingroup$ @Mien Not sure how that's staining, it sounds more like directed anesthesia to me. Staining is more along the lines of permanently changing the cell, I think. $\endgroup$ Commented Jun 19, 2012 at 8:40
  • $\begingroup$ Okay, fair enough. Just to be clear: it normally has a colour, to see which hemisphere is affected. So you'd stain one hemisphere before the other. $\endgroup$
    – Mien
    Commented Jun 19, 2012 at 8:57
  • 2
    $\begingroup$ Hrm... didn't consider you'd wanted the test subject alive... :) $\endgroup$ Commented Jun 19, 2012 at 9:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.