Functional magnetic resonance imaging is pretty technical stuff. As the Neuroskeptic aptly puts it, “when you do an fMRI scan you’re using a superconducting magnet to image human neural activity by measuring the quantum spin properties of protons. It doesn’t get much more technical.”
The technique for measuring neural activity has become a household name just 20 years after the discovery of the blood oxygen level-dependent response. The number of articles discussing “fMRI” in the abstract or title has been following this trend:
1) Biological: When neurons are more active (i.e., have a higher rate of action potentials) they require more glucose for energy quickly. Thus, task-laden neurons do more of the rapid ( ~ 2 times faster) but non-oxygen requiring process of breaking down glucose via glycolysis. One of the reasons that glycolysis is upregulated is to provide energy for the membrane pump Na, K -ATPase, which allows for synaptic glutamate reuptake by astrocytes and AMPA receptor turnover at postsynaptic densities, among other tasks.
Due to the increased need for energy following task performance, blood flow to a brain region also increases when that region is involved in a task or stimulated in some way. Using PET imaging, Fox et al (here) showed that somatosensory stimulation led to a 29% increase in cerebral blood flox in the contralateral somatosensory cortex of human participants:
Concomitantly, blood flow almost always increases more than that of local O2 demand, as indicated by measurements of the partial pressure of O2, indicating that the blood flow increase is an overcompensation. This means that the local concentration of deoxyhemoglobin in the local veins should decrease.
Deoxyhemoglobin can be measured by magnetic resonance because it has unpaired electrons, which enables researchers to track the task-induced change. The changes in aerobic glycolysis and task-induced cerebral blood flow increases probably have a similar origin.
2) Physical. Deoxyhemoglobin has 4 unpaired electrons (i.e. it is paramagnetic), whereas oxyhemoglobin does not have unpaired electrons. This means that deoxyhemoglobin will have a magnetic moment that will affect the local magnetic field and alter the ability of the MR machine to flip the spin state of protons at a given magnetic field value.
So, a change in the ratio of oxy- / deoxy- hemoglobin leads to a change in the T2* relaxation times of MR images, changing the image intensity by a few percent. This image intensity difference extends beyond just the cerebral blood volume because a local magnetic field will form across arteries / veins if one of the regions has a higher ratio of oxy- / deoxy- hemoglobin.
Bammer et al (here) explain this concept with a beautiful diagram and provide a chart describing the relationship between T2* intensity and time given different oxygenation concentrations of hemoglobin:
I refuse to go into more detail here because I didn’t do very well on the MR section of my organic chemistry class and reading more about it now is bringing up painful memories.
Statistically, these T2* differences can be extracted to give indications of the cerebral blood flow, which should correlate with the neural activity upregulation above baseline in the given region. Unfortunately, the temporal resolution isn’t great and changes in blood flow tend to occur ~ 5 seconds after changes in activity. But it is so noninvasive that the tech will likely continue to receive widespread use.
Inspired by CalTech’s Question #17 for cognitive scientists: “What is the physical and biological basis of structural and functional MRI for brain imaging?”
Bammer R, et al. 2005 Foundations of Advanced Magnetic Resonance Imaging. NeuroRX PMID: PMC1064985.
Raichle M, et al. 2010 Two views of brain function. Trends in Cognitive Sciences doi:10.1016/j.tics.2010.01.008
Fox PT, et al. 1986 Focal physiological uncoupling of cerebral blood flow and oxidative metabolism during somatosensory stimulation in human subjects. PNAS, link here.