What Is Functional Brain Imaging?

Technological developments which have been taking place over the past 15-20 years have resulted in the more recent evolution of several relatively safe methods of soft-tissue imaging. These are different than traditional x-rays, for example, which in addition to being a health risk, are more efficient at showing bone structure than soft tissues, such as muscle. Functional imaging is a type of process used to take pictures of the human body which, besides showing anatomical structure, also maps regions of varying activity. In terms of functional brain imaging, this means that the primary information received during the imaging process concerns which part (or parts) of the brain is active (or inactive) when a person performs a particular task or is exposed to a certain type of stimulus. What makes this so remarkable, is that for the first time ever it possible to determine and even map brain activity in a way which is, for the most part, non-invasive physically. Some techniques do involve the intravenous injection of radioactive chemicals, however there are few, if any, side-effects and no surgery is necessary. Thus the brain activity of, for example, a student doing a math problem or reading a sentence, can be studied directly without posing health risks or introducing additional variables into experimental results.

Better known functional imaging techniques include positron emission tomography (PET), and functional magnetic resonance imaging (fMRI) (a variation of traditional magnetic resonance imaging, or MRI). Others are variations and/or combinations of these and older methods of measuring brain activity, such as electro- and magnetoencephalography (EEG and MEG, respectively). Sometimes, the term computerized tomography is used more generally, to refer to the computer processes involved in creating images from data acquired through PET, MRI, or x-ray scans.

PET and fMRI, in particular, have both begun to play key roles in the study of mental disorders, changes in brain functioning due to aging, learning differences, and correlations between different "thought patterns" and factors such as gender and culture. For example, PET has allowed for the visual diagnoses of adult Attention Deficit Hyperactivity Disorder (ADHD), and fMRI, when interfaced with special software, can provide an inflated image of a brain which highlights localized activity during different language-related tasks. Before the development of PET and fMRI, neural functioning was studied mainly through the use of EEG and MEG. Both involve arrays of electrodes attached to specified positions on the skull, which record electrical impulses in the brain. Though useful in many ways, alone they have been inadequate for endeavors which require either a three dimensional viewpoint of the brain, or the determination of the region from which a particular signal originates. Now, use is being made of functional imaging methods which superimpose EEG and MEG maps onto PET or MRI images.

These are merely a few general examples of the exciting and revolutionary work being done in cognitive science through functional imaging. The following links provide more detailed technological, biological, and practical information on each imaging process listed below. There will be a large focus on learning-related applications of functional imaging and their potential implications for the future of both education and cognitive neuroscience. It should be pointed out in advance that while functional imaging has brought forth tremendous breakthroughs in the study of how human beings think, there is still a long way to go. While such techniques do show which regions of the brain experience heightened activity during specific functions, what exactly is behind that activity is another question. For example, similar activity in two separate brains does not necessarily signify similar thinking. Furthermore, the actual information associated with a process may be located elsewhere in the brain.


List of on-line references

Positron Emission Tomography (PET)

First Things First: What is a Positron?

A positron is a subatomic particle, like an electron in size and behavior. However, instead of having a negative electric charge a positron has a positive one. Since positrons and electrons are alike in nearly every way other than the fact that their charges are exactly opposite, when two particles, one of each species, come in contact with one another they immediately destroy one another (in other words, the positron is the antimatter counterpart of the electron). Energy is released from the collision in the form of two photons (light particles), which fly off in opposite directions (180 degrees) from one another. An interaction of this sort can be located by radiation detectors placed opposite one another on either side of the collision area. When an electron-positron annihilation occurs such that the resulting photons are emitted in line with the detectors, they receive these photons, which are of equal energy, simultaneously (or very nearly so)--the detectors are correspondingly referred to as coincidence detectors.

So, What Is PET?

In Positron Emission Tomography (PET), the most highly evolved functional brain imaging technique, the subject/patient is intravenously injected with a radioactive substance called a tracer. Different tracers are used depending on the specific goal of the PET scan (the similarity between them being that they all emit positrons). Coincidence detectors are positioned around the skull. Positrons emitted from the tracer in the blood stream quickly collide with electrons in the nearby brain tissue and annihilate (Hugdahl, 1995). Since blood rushes to the regions of the brain which are being used at a given moment, substances carried in the blood stream naturally become concentrated in those areas as well. The detectors are able to locate changes in neural activity due to the accumulation of the radioactive substance in the brain tissue for a short period of time immediately after the tracer is administered (Raichle, 1994; Hugdahl, 1995). Thus a map of the active regions of a subject's brain can be made while they, for example, perform language-related tasks, or are exposed to some other sort of visual or auditory stimuli.

In addition to determining which regions are active at a particular time, the actual brain chemistry involved in a certain type of activity can also be studied with PET by varying the type of radioactive tracer. A tracer may be radioactively labeled and still be utilized by the brain in the same manner as a similar chemical which occurs there naturally. Therefore it can be used to take measurements relating to both which areas are actively taking up the compound and to what degrees. A prime example of this is the measurement of glucose levels in the brain; this is important for many reasons since glucose is the brain's primary metabolic energy source (Groves and Rebec, 1988). More generally, PET scans can provide measurements of changes in: 1) regional blood flow (the circulation rate of blood in a region of the brain measured in milliliters per minute); 2) blood volume (blood vessels in the brain can expand and contract in order to regulate the amount of blood in a region); 3) glucose metabolism, oxygen metabolism, and the metabolisms of various neurotransmitters in given regions of the brain (Raichle, 1994; Hugdahl, 1995; Kandel and Schwartz, ed., 1981). The term metabolism refers to the levels of substances which provide energy to an area of the brain for a distinct function.

Image Construction

PET images are constructed via computers from the radiation patterns received by the coincidence detectors of different planes (or "slices") of the brain. These slices, when combined, provide a three dimensional perspective. Scans are first taken of the brain in a relatively inactive state (the control state). The subject then performs tasks or is provided with an external stimulus, and images are made of the brain in an active state.

The control images are subtracted from the stimulated-state images in order to isolate the specific regions in which the brain activity has changed due to stimulation (Raichle, 1994). The result is a computerized image of the brain, color-coded to represent areas with different levels of activity (for example, yellow, red, and white typically indicate increasing levels of higher activity while blues and purples indicate lower levels). Those areas can then be associated with the mental functions involved in processing the type of information provided through the stimulus, or with the patients outward response.

Images taken of multiple "normal" brains and averaged can be compared to images of a brain that works differently in some way. Thus the location of the atypical activity (or chemistry) related to those differences can be seen visually.

A Deeper Look at PET

The human brain consumes approximately 20% of the oxygen taken in by the body. This is pretty amazing considering that one's brain accounts for only about 2% of one's total body weight (Chien, 1981). In fact, the brain requires more energy to operate than any other part of the human body. This energy is acquired mainly from oxygen and glucose in the blood stream. The level of oxygen and/or glucose provided to a region can be at least partially regulated through a change in either the blood volume or flow rate.

In PET processes concerned with monitoring changes in the blood flow to different parts of the brain, the radioactive isotope 15O (where the superscript on the left signifies the atomic number for this particular type of oxygen), in the form of radioactively "labeled" water (H215O), is generally used as a tracer (Hugdahl, 1995; Raichle, 1994). As the blood flow increases in active areas so as to provide an increased oxygen supply, the 15O is used up just as normal oxygen would be, meaning that more of it is pumped to and used by the active areas of the brain than the inactive ones. Therefore, more emission shows up in those areas on the PET scan.

In glucose metabolism studies, the compound 18-F-flouro-2-deoxyglucose (18F-FDG) is often used (Hugdahl, 1995; Raichle, 1994). The important part here is the 2-deoxyglucose (2-DG), or radioactively labeled glucose (Pinel, 1990). As with oxygen, when the metabolism in a region increases, the 2-DG is used up more quickly by the active areas, the same as naturally occurring glucose (Pinel, 1990).

When the blood flow in an area increases in order to provide the necessary levels of oxygen and glucose, the result is that the tracer concentration, by default, increases as well, accumulating in the active neurons. Within a few seconds, the tracer molecules radiate. As mentioned above, positrons emitted from the tracer collide with electrons in the brain tissue, and annihilate with each other in order to produce two identical high-energy gamma rays (photons). In order to conserve momentum, these gamma particles are emitted exactly 180 degrees from one another, and are absorbed simultaneously by the surrounding coincidence detectors. The relative count-concentrations are determined by the PET scan. The resulting data is then mapped, using a special 3-dimensional coordinate system, to the corresponding regions of the brain.

Advantages and Disadvantages

One drawback to PET imaging is that a small cyclotron as well as chemistry facilities are necessary to run scans. This means the technology is less readily available and more costly than some other functional imaging techniques (Raichle, 1994). It is an especially useful method, however, in that it is able to monitor different metabolism levels in the brain, a task which cannot be completed using traditional Magnetic Resonance Imaging (MRI) or Computerized Tomography (CT).

Learning-Related Applications of PET

PET has been actively used to visually examine conditions related to anxiety, attention, language processing, and mental imagery and perception. In particular, it has allowed the physiological study of attention deficit hyperactivity disorder (ADHD), a condition which seems to involve a glucose deficiency in regions of the brain associated with the control of attention. PET has also been used in the study of dyslexia, a language-related disorder which most notably inhibits reading skills. Often, people who have dyslexia are also diagnosed with ADHD. Imaging studies have shown similar types of brain behavior related to both disorders. If you are interested in learning more about the use of PET and other imaging techniques in learning-related research, please visit our section on learning differences. For further reading and/or examples of PET images, please see our reference list as well as the these web-sites , which we have also used as references.

Functional Magnetic Resonance Imaging (fMRI)

What Is Functional MRI?

Traditional magnetic resonance imaging (MRI), a method which has been in use since the early seventies, is a process in which a magnetic field is used to measure variations in the density of water contained in soft-tissue. The resulting MRI image is essentially a "map" of the water density in different parts of the brain, where the boundaries of regions with particular water concentrations create a picture of separate structural features. This technology has been found by many to be superior to earlier anatomical imaging methods, mainly computerized tomography (CT) scans (an x-ray imaging method) (Conlon, et.al., 1989).

Functional MRI (fMRI) combines these advantages with the ability to observe changes in the brain as they happen along with physical or mental activity. These differences occur in blood flow and chemical metabolisms in active or inactive regions of the brain. Its applications are similar to those of positron emission tomography, although there are instances in which the use of one method as opposed to the other has its advantages.

Advantages and Disadvantages

Thanks to fMRI, changes in neural activity can now be detected within 1-2 hundred milliseconds (about 0.1-0.2 seconds) by monitoring the alignment of protons which already exist in the brain tissue via a large magnetic field. These protons are mainly in the form of hydrogen nuclei from water within the tissue, however other types of nuclei can be observed as well. This process can provide fairly accurate anatomical identifications of active regions in the brain at spatial resolutions better than 2 x 2 x 5 mm3--and improving. Image resolution can be changed by varying the magnetic field strength (Hugdahl, 1995; Raichle, 1994). This type of precision exceeds the capabilities of, for example, positron emission tomography (PET), however there are compensating drawbacks. One of these is that fMRI is a little less accurate than PET in actually measuring the changes in blood flow (Raichle, 1994). Functional MRI also differs from PET in that it generally does not require the use of a radioactive tracer (a potential health hazard). Overall, traditional MRI has been used for more than 20 years and fMRI since the early nineties, and there has been no indication of damaging biological effects with either.

How Do They Do It? (The Physics Behind fMRI)

Magnetic resonance imaging (as well as the increasingly sophisticated techniques, like fMRI, which have evolved from it) is based on a process called NMR, or nuclear magnetic resonance. This method was discovered in the mid forties and is now vital to areas of study in both physics and chemistry. It uses a magnetic field and radio waves to align molecular nuclei in a substance, in order to observe the emission given off by the particles under certain conditions, such as (for example) a temperature change. Thus NMR is actually part of the more general science of spectroscopy: the study of characteristic properties of light or particles emitted from a stimulated system.

Most NMR labs study very small samples of a substance (on the order of a milliliter) placed in a uniform magnetic field. The molecular nuclei become aligned with the magnetic field, in the sense that their axes of rotation precess (revolve) around the magnetic field lines at a characteristic frequency. The samples, which are surrounded by a metal coil, are then exposed to radio waves at this same frequency (the resonant frequency) positioned at 90 degrees to the field. This pulse of radiation causes the nuclei to rotate to an angle of 90 degrees from their prior position, at which point the radiation is turned off and the nuclei remain in equilibrium, rotating around an axis perpendicular to the magnetic field. These rotating charges induce an electric current in the surrounding coil, which then is picked up by a radio receiver on the other end and interpreted by a device (such as an oscilloscope) which is able to display the signal received from the sample. Meanwhile, the nuclei begin to relax and the signal starts to decay. Generally the radio frequencies used for such experiments are in the range of 0 to 109 Hz (or vibrations per second) (Gillies, 1994).

In MRI, the patient (in essence, the sample--albeit a rather large one) is first placed entirely inside a narrow tubular scanner (about 10 feet long and 6 feet across) (Gutin, 1996). As in NMR, a strong magnetic field is used to induce protons in the subject's neural tissue to enter an equilibrium state at 90 degrees to the magnetic field. This alignment is then disrupted by way of high energy radio pulses. A coil is positioned around all or part of the patient's head. The proton disalignment induces a current in this coil. The major difference in the case of biological imaging, however, is that a gradient is applied over the magnetic field. This means that the field strength changes over the region being imaged. This provides spatial/positional information by creating a contrast between different structures in the brain. Depending on the type of signal receiver being used, the region may be scanned point by point, line by line, or an entire "slice" of the brain may be seen all at once. Multiple slices can be put together to make a 3-dimensional image. In conventional MRI, this image is essentially a map of the hydrogen distribution in the brain tissue. Other techniques take advantage of the characteristic (i.e. spectroscopic) properties of emission from different substances in the bloodstream and other tissues. For example, the properties of hemoglobin in a magnetic field depend on the local oxygen levels in the blood.

Learning-Related Applications of fMRI

An Aside: fMRI in the study of Language

Language acquisition and linguistic processes are primary areas of interest in the study of learning. It has been thought for quite awhile that two areas of the human brain, Broca's area and Wernicke's area, are devoted to language production and comprehension respectively. Studies have shown that damage to the Wernicke's area impairs the ability to articulate meaningfully, though in some cases the patient's speech may sound fluent. Damage to the Broca's area however effects one's ability to speak in a functional manner—speech may be slow, broken, and slurred, though still meaningful in content (Gleitman, 1991; Maratsos and Matheny, 1994). Most current theories concerning the use and evolution of human language are restricted to these two areas.

Martin Sereno, a neurobiologist at the University of California at San Diego, has developed a new theory of language development, for which he is trying to acquire evidence through a combination of electrophysiological (EEG and MEG) mapping and multislice fMRI. He believes that human language abilities evolved through the slight "rewiring" of multiple pre-existing parts of the brain which are also associated with visual, auditory, and motor processes (Gutin, 1996). He does not doubt that Broca's and Wernicke's areas are strongly connected to language, but he also thinks that these locations are in league with other parts of the brain as well—especially the visual cortex. He claims about 50% of the human cortex is related in some way to visual processing, and that language centers came out of and are currently intertwined with these regions (Gutin, 1996).

Sereno's group has already shown that many visually related regions of the brain overlap with areas previously known to to be activated by written words (Sereno et al., 1995). His ideas concerning the existence of multiple laguage centers also seems to be in agreement with more recent work of others in the field as well. For example, a team of researchers led by Hanna Damasio and Antonio Damasio at the University of Iowa have found (also through functional imaging) that when individuals are called upon to name objects seen in a series of pictures, parts of the brain become active that are not only outside the traditional language centers, but which are spatially located in various different regions of the brain.

The imaging techniques use by Sereno's group is an innovative combination of new and not-quite-as-new technology. A few years ago, Sereno and colleague Anders Dale designed a computer program which maps a combination of data from electro- and magnetoencephalographs (EEG and MEG) onto 2-dimensional MRI images. A 3-D image can then be constructed from this set of 2-D "slices." This resultant image can then be expanded on the computer screen (it appears to be inflated) so that different parts of the brain can be clearly located on the image (Gutin, 1996).

Sereno and his collaborators have also been working on mapping the visual cortex using fMRI. Subjects are exposed to visual linguistic stimuli (letters, words, sentences, etc.) on a nearby screen during fMRI scans. Data from these scans have shown activity in regions of the human brain which correspond to visual centers in the brains of other sorts of primates. Studies by other neuroscientists have given evidence that damage to these same areas (simulated by electrical stimulation) impairs speech in ways similar to those caused by damage to the Wernicke's and Broca's areas (Gutin, 1996).

These discoveries, and the technology developed in order to make them, are not only interesting from the point of view of Sereno's theory on the evolution of human language. The way (or ways) in which our minds react to and use language is likely to be closely linked on a physiological level (as it is known to be on a psychological one) to how we interpret and communicate different concepts as individuals. For example, imagine a variety of subjects are visually exposed to a particular sentence which results in activity in the same region of the brain for each one. Could a detailed analysis of those images provide information concerning differences in how that sentence is processed in the brain of each individual? Given that everyone may perceive the sentence to mean something different, wouldn't it make sense that there would appear to be at least slight differences in the activity patterns on the fMRI images? And then, would it be possible to distinguish these from variations caused by functionally insignificant differences brain structure, or even systematic noise? If so, the implications for the future of functional brain imaging are enormous. If not, the use of functional imaging in cognitive neuroscience, though still potentially useful, may be somewhat limited.