Functional magnetic resonance imaging (fMRI) is an emerging medical diagnostics technique that is sensitive to the concentration of oxygen in biological tissue and therefore to the level of functional activity that the tissue exhibits. The most exciting application of this technique is the study of activity patterns in the human brain – there are well-founded concerns that fMRI could become sufficiently powerful to be able to read human emotion and eventually thought.
The fundamental principle of fMRI is the same as that of standard magnetic resonance imaging – the precession frequency of nuclear spins in an applied magnetic field can be made position-dependent by applying field gradients (Orrison Jr et al., 2015):
Where y is the magnetogyric ratio of the nucleus in question and are the amplitudes of the magnetic field gradients. The nuclei absorb and radiate back this position-dependent frequency. When the resulting radiofrequency intensity distribution across frequencies is detected, it is equivalent to the distribution of proton concentrations across space. The most common proton-containing substance in the body being water, the result is a water concentration image across the body – bones are dark, fatty tissue grey and blood vessels very bright. An example is given in Figure 1 (top left).
Figure 1. MRI images of human brain with four different contrast settings: PD (proton density), T2 (transverse relaxation time), T1 (longitudinal relaxation time) and MRA (magnetic resonance angiography). Source: Radiology Department, UCLA.
Other sources of contrast are also available in MRI – for example, the time it takes for the nuclei to get magnetised along the applied magnetic field (known as T1 time) and demagnetised in the perpendicular direction (T2 time) is different for different types of tissues (Verwilst et al., 2015). Images recorded with T1 and T2 contrast are also shown in Figure 1. The most exciting part, however, is that both of these parameters depend on the local oxygen concentration (Ma et al., 2015). Oxygen is consumed at a greater rate by the parts of the brain that are loaded with some function and it is therefore possible to tell which parts of the brain are active at any given moment. This technique is known as functional magnetic resonance imaging (Orrison Jr et al., 2015).
Figure 2. Difference in the brain activation areas during decision-making activity in the brains of healthy (a) and schizophrenic (b) individuals. Reproduced from (Koch et al., 2011).
The primary reason for the existence of blood-oxygen-level-dependent (BOLD) contrast in tissues is the fact that molecular oxygen is paramagnetic – unlike most stable molecules at room temperature its ground electron state is not a singlet, but a triplet (Zhu et al., 2016), meaning that oxygen can engage in magnetic dipolar interactions with the nuclei of the surrounding molecules (Li et al., 2015). Because the motion of molecules in a liquid is chaotic, this generates noise in the magnetic interactions between oxygen molecules and the nuclei and this noise would typically have a component at the nuclear spin transition frequency, meaning that it would cause the nuclear state to return to the thermal equilibrium.
This is called paramagnetic relaxation (Schlagnitweit et al., 2015) and is the primary reason for the emergence of oxygen concentration dependent contrast in MRI images. It is important to note that the haemodynamic response that alters the local oxygen concentration lags the neuronal events by between one and two seconds – that is how long it takes the circulation system to respond to the increased local workload (Donahue et al., 2015).
Spatial and temporal resolution
The spatial resolution of fMRI is limited by the need to suppress the signal from major blood vessels and by the intrinsically low signal-to-noise ratio of the technique that has to rely on relatively small changes in the local relaxation times. This is typically between millimetres and centimetres, meaning that different Brodmann areas (histologically distinct region of the cortex) and subcortical nuclei (identifiably distinct groups of neurons) may be identified as active (Yacoub et al., 2015).
The time resolution of fMRI is determined by the acquisition time of the MRI instrument, the haemodynamic response function and the need to record a ‘background’ image that is free of the activation pattern in question (Safi-Harb et al., 2015). Modern multiplexed echo-planar imaging pulse sequences (Boyacioğlu et al., 2016) are generally able to obtain a high-quality 2D slice in less than 100 milliseconds and a high-quality volumetric image in less than 30 seconds.
Haemodynamic response to neuronal activity takes a few seconds and the time it takes for the brain to return to its resting state after an experiment strongly depends on the type of activity and may be anywhere between minutes and hours (Narsude et al., 2015). Assuming that the oxygen concentration differential is sufficiently strong to see the necessary changes in one scan, a three-dimensional fMRI image may thus be recorded in about a minute. An example is given in Figure 3.
Figure 3. Resting state network (activation areas pertaining to a brain at rest) fMRI images recorded in a fraction of a second using multiplexed echo-planar imaging (Feinberg et al., 2010).
It is beyond reasonable doubt that the activation maps returned by fMRI are accurate – multiple studies involving implanted electrodes (Chang et al., 2016), electroencephalograms (Usami et al., 2016) and magnetoencephalograms (Natsukawa and Kobayashi, 2015), as well as cerebral blood flow data from PET scans (Fan et al., 2016) have confirmed that the areas reported by fMRI are indeed activated. It is clear, however, that there is a time lag relative to EEG and MEG data and that fMRI images are broad averages over the volumes of several cubic millimetres even at the highest available resolution (Safi-Harb et al., 2015).
Also, fMRI cannot differentiate between inhibitory and excitatory activities because both types consume oxygen and glucose. Current applications Although fMRI is currently used in the clinic for pre-operative assessment, neurosurgical planning and to diagnose pathology in the functioning of the brain (an example is given in Figure 2), its primary application is in the neurophysiological research, where fMRI is used to:
1. Determine function localisation and functional connectivity of the brain by detecting simultaneously activated areas. A recent example is the demonstration that recognition takes place in the same areas of the brain regardless of the object being recognised (Kaiser et al., 2016).
2. Examine differences in brain structure and function between bearers of different genetic polymorphisms in the absence of clear phenotypic, metabolic or behavioural differences. The ability to tie genetic polymorphisms to low-level cognitive function is important because post-natal adaptations can in many cases compensate neurophysiological pathology and make it difficult to detect (Poline et al., 2015).
3. Determine the changes in the functioning of the brain associated with pharmacological treatment. Recent studies include the effects of analgesic drugs (Wager and Woo, 2015), side effects of certain antibiotics and detailed studies of brain function pattern alteration by hallucinogenic drugs (Spain et al., 2015) and narcotics (Ianno, 2016).
4. Document individual differences in cognitive neurophysiology as well as diagnosing psychiatric disorders, such as panic disorder, anxiety disorder, autism, bipolar disorder and ADHD. Methods based on fMRI look for known patterns of brain activation in response to external triggers. For example, patients suffering from social anxiety are presented with images and videos exhibiting triggering behaviour (Poletti et al., 2015).
5. Study commonalities in brain activation. It was recently demonstrated that social exclusion discomfort shares many activation areas with physical pain (Wudarczyk et al., 2015) and analgesic response to placebo has a surprising number of areas in common with the response to the actual analgesic (Colagiuri et al., 2015).
Many more current application areas exist – see recent reviews (Cendes, 2015; Ernst et al., 2015; Fan et al., 2016; Smirnakis, 2016) for further information. Future applications Future application areas for fMRI can often resemble science fiction. To take a famous example, because the acts of retrieving a memory and of constructing a consistent lie involve very different brain areas, it is actually very easy to tell, using fMRI, that a person is lying. This has generated both a significant industry of companies offering fMRI lie detection and a significant area of research in ethics and philosophy (Miklin and Fiore, 2015). It is quite clear that fMRI lie detection works (Choi et al., 2015), is difficult to circumvent and may eventually become a standard method in the toolkit of various organisations that, for one reason or another, need to know with certainty whether someone is telling the truth.
A more conceptually and technically challenging proposition that is unlikely to ever deliver fully on its promises is ‘thought reading’ using fMRI. Due to the inevitable latencies and resolution issues discussed above, it is not realistic to literally ‘read thoughts’ using MRI. It is, however, perfectly feasible to read emotions and there are many studies that indicate disgust, fear, anxiety, hunger, happiness and other distinct human emotions can be detected (Kleinhans et al., 2016). Research in this direction currently continues apace and a time may eventually come when a tinfoil hat (conductive materials reflect radio waves) would no longer be deemed an eccentricity.
BOYACIOĞLU, R., SCHULZ, J. & NORRIS, D. G. 2016. Multiband echo‐shifted echo planar imaging. Magnetic resonance in medicine.
CENDES, F. 2015. Epilepsy: Can fMRI predict memory decline after temporal lobe resection? Nature Reviews Neurology, 11, 372-374.
CHANG, C., LEOPOLD, D. A., SCHÖLVINCK, M. L., MANDELKOW, H., PICCHIONI, D., LIU, X., FRANK, Q. Y., TURCHI, J. N. & DUYN, J. H. 2016. Tracking brain arousal fluctuations with fMRI. Proceedings of the National Academy of Sciences, 201520613.
CHOI, Y., KIM, S., DO, H., SHIN, K. S. & KIM, J. E. 2015. Detecting Deception Using Neuroscience: A Review on Lie Detection Using Functional Magnetic Resonance Imaging. Korean Journal of Biological Psychiatry, 22, 109-112.
COLAGIURI, B., SCHENK, L., KESSLER, M., DORSEY, S. & COLLOCA, L. 2015. The placebo effect: from concepts to genes. Neuroscience, 307, 171-190.
DONAHUE, M. J., STROTHER, M. K., LINDSEY, K. P., HOCKE, L. M., TONG, Y. & DEB FREDERICK, B. 2015. Time delay processing of hypercapnic fMRI allows quantitative parameterization of cerebrovascular reactivity and blood flow delays. Journal of Cerebral Blood Flow & Metabolism, 0271678X15608643.
ERNST, M., TORRISI, S., BALDERSTON, N., GRILLON, C. & HALE, E. A. 2015. fMRI Functional Connectivity Applied to Adolescent Neurodevelopment*. Annual review of clinical psychology, 11, 361-377.
FAN, A. P., JAHANIAN, H., HOLDSWORTH, S. J. & ZAHARCHUK, G. 2016. Comparison of cerebral blood flow measurement with [15O]-water positron emission tomography and arterial spin labeling magnetic resonance imaging: A systematic review. Journal of Cerebral Blood Flow & Metabolism, 36, 842-861.
FEINBERG, D. A., MOELLER, S., SMITH, S. M., AUERBACH, E., RAMANNA, S., GLASSER, M. F., MILLER, K. L., UGURBIL, K. & YACOUB, E. 2010. Multiplexed Echo Planar Imaging for Sub-Second Whole Brain FMRI and Fast Diffusion Imaging. PLoS ONE, 5, e15710. IANNO, D. 2016. Broadening perspectives on the management of chronic pain.
KAISER, D., AZZALINI, D. C. & PEELEN, M. V. 2016. Shape-independent object category responses revealed by MEG and fMRI decoding. Journal of neurophysiology, 115, 2246-2250.
KLEINHANS, N. M., RICHARDS, T., GREENSON, J., DAWSON, G. & AYLWARD, E. 2016. Altered Dynamics of the fMRI Response to Faces in Individuals with Autism. Journal of autism and developmental disorders, 46, 232-241.
KOCH, K., WAGNER, G., SCHACHTZABEL, C., SCHULTZ, C. C., GÜLLMAR, D., REICHENBACH, J. R., SAUER, H. & SCHLÖSSER, R. G. M. 2011. Neural activation and radial diffusivity in schizophrenia: combined fMRI and diffusion tensor imaging study. The British Journal of Psychiatry, 198, 223-229.
LI, W., GRGAC, K., HUANG, A., YADAV, N., QIN, Q. & VAN ZIJL, P. 2015. Quantitative theory for the longitudinal relaxation time of blood water. Magnetic resonance in medicine.
MA, Y., BERMAN, A. J. & PIKE, G. B. 2015. The effect of dissolved oxygen on the relaxation rates of blood plasma: Implications for hyperoxia calibrated BOLD. Magnetic resonance in medicine.
MIKLIN, D. J. & FIORE, R. N. 2015. fMRI Imaging and Decision Making in Vegetative Patients: Ethics, Technology, and Welfare. AJOB Neuroscience, 6, 49-51.
NARSUDE, M., GALLICHAN, D., VAN DER ZWAAG, W., GRUETTER, R. & MARQUES, J. P. 2015. Three‐dimensional echo planar imaging with controlled aliasing: A sequence for high temporal resolution functional MRI. Magnetic resonance in medicine.
NATSUKAWA, H. & KOBAYASHI, T. 2015. Cortical activation associated with determination of depth order during transparent motion perception: A normalized integrative fMRI–MEG study. Human brain mapping, 36, 3922-3934.
ORRISON JR, W. W., LEWINE, J., SANDERS, J. & HARTSHORNE, M. F. 2015. Functional brain imaging, Elsevier Health Sciences.
POLETTI, S., RADAELLI, D., CUCCHI, M., RICCI, L., VAI, B., SMERALDI, E. & BENEDETTI, F. 2015. Neural correlates of anxiety sensitivity in panic disorder: a functional magnetic resonance imaging study. Psychiatry Research: Neuroimaging, 233, 95-101.
POLINE, J.-B., BREEZE, J. & FROUIN, V. 2015. Imaging Genetics with fMRI. fMRI: From Nuclear Spins to Brain Functions. Springer.
SAFI-HARB, M., PROULX, S., VON ELLENRIEDER, N. & GOTMAN, J. 2015. Advantages and disadvantages of a fast fMRI sequence in the context of EEG–fMRI investigation of epilepsy patients: a realistic simulation study. Neuroimage, 119, 20-32.
SCHLAGNITWEIT, J., TANG, M., BAIAS, M., RICHARDSON, S., SCHANTZ, S. & EMSLEY, L. 2015. Nanostructure of Materials Determined by Relayed Paramagnetic Relaxation Enhancement. Journal of the American Chemical Society, 137, 12482-12485.
SMIRNAKIS, S. M. 2016. Probing Human Visual Deficits with fMRI. Annual Review of Vision Science, 2. SPAIN, A., HOWARTH, C., KHRAPITCHEV, A. A., SHARP, T., SIBSON, N. R. & MARTIN, C. 2015. Neurovascular and neuroimaging effects of the hallucinogenic serotonin receptor agonist psilocin in the rat brain.
Neuropharmacology, 99, 210-220. USAMI, K., MATSUMOTO, R., SAWAMOTO, N., MURAKAMI, H., INOUCHI, M., FUMURO, T., SHIMOTAKE, A., KATO, T., MIMA, T. & SHIROZU, H. 2016.
Epileptic network of hypothalamic hamartoma: An EEG-fMRI study. Epilepsy Research, 125, 1-9. VERWILST, P., PARK, S., YOON, B. & KIM, J. S. 2015. Recent advances in Gd-chelate based bimodal optical/MRI contrast agents. Chemical Society Reviews, 44, 1791-1806. WAGER, T. D. & WOO, C.-W. 2015. fMRI in analgesic drug discovery. Science translational medicine, 7, 274fs6-274fs6.
WUDARCZYK, O. A., KOHN, N., BERGS, R., GUR, R. E., TURETSKY, B., SCHNEIDER, F. & HABEL, U. 2015. Chemosensory anxiety cues moderate the experience of social exclusion–an fMRI investigation with Cyberball.
Frontiers in psychology, 6. YACOUB, E., HAREL, N. & SHMUEL, A. 2015. High-resolution fMRI. fMRI: From Nuclear Spins to Brain Functions. Springer. ZHU, H., TRAORE, K., SANTO, A., TRUSH, M. A. & LI, Y. R. 2016. Oxygen and Oxygen Toxicity: The Birth of Concepts. Reactive Oxygen Species, 1, 1-8.