Magnetic Resonance Imaging (MRI) is a structural brain-imaging procedure in which high-resolution images are constructed from the measurement of radio-frequency waves that hydrogen atoms emit as they align with a powerful magnetic field. MRI provides clearer images of the brain than does ct. A two-dimensional MRI scan of the midsagittal plane of the brain. In addition to providing relatively high spatial resolution (the ability to detect and represent differences in spatial location), MRI can produce images in three dimensions.
In magnetic resonance imaging (MRI), a large magnet (M) and a specific radiofrequency pulse (R) generate a brain signal that produces an image (I). MRI is used to study both brain anatomy and neural function noninvasively, and because it does not make use of ionizing radiation, it is safe enough to use repeatedly on volunteers and patients, adults and children alike. The technique is based on the principle that a hydrogen atom’s nucleus, which consists of a single proton, behaves like a spinning bar magnet. Each proton has a dipole: one end is a north pole and the other end a south pole. Each spinning proton produces an electrical current. Ordinarily, protons are oriented at random, so a given piece of tissue (all soft tissue contains water, which contains hydrogen) has no net dipole and consequently generates no net electrical current.
When hydrogen atoms are placed in a magnetic field, the spinning protons orient themselves with respect to the field’s lines of force. In other words, the protons behave like a compass needle that aligns itself north and south with Earth’s magnetic field. When aligned, the protons’ summed electrical current is large enough to be measured. Because proton density varies in different brain tissue (cerebrospinal fluid, myelin, neurons), largely in proportion to its water content, the electrical currents produced by the aligned protons are different, higher for some tissues and lower for others. Measures of the electrical current are used to create the MRI image. Another way to make an MRI image is to perturb the protons when they are aligned and record the changes occurring in the electrical field as the protons realign after the perturbation. A brief radiofrequency pulse is applied to a brain, horizontal to the magnetic field. The pulse forms a second magnetic field that pushes the aligned protons over onto their sides. The tipped protons now have two motions: they spin about their own axes and they spin about their longitudinal (north-south) orientation. The protons wobble like a slowly spinning top, a motion called precession. When the horizontal magnetic field is turned off, the synchronously spinning protons begin to relax: they begin to “stand up” again and fall out of synchrony with one another. Both relaxation processes are measured, using a current detector, by two-time constants, T1 and T2:
• For T1, a current detector oriented horizontally to the vertical axis, that is, to the protons’ initial alignment, measures how long it takes after the magnetic pulse is turned off for the protons to “right” themselves from their tipped positions and realign with the original magnetic field.
• For T2, a second detector oriented perpendicular to the first measures the rate at which protons lose synchrony about the horizontal axis after the horizontal pulse is turned off. Protons in differing tissue types have different relaxation rates and corresponding T1 and T2 time constants. For example, the relaxation rates for cerebrospinal fluid are slower than those for brain tissue. Therefore, at a set time—for example, at the midpoint of relaxation—differences in electrical current related to and indicating the composition of tissue can be measured. T1 and T2 can be translated into brain-image gradients that correspond to its different tissues, with darker gradients indicating low-density tissue and lighter indicating high-density tissue. Either T1 or T2 constants are used, though one may be more suitable than the other in a given situation. For example, T2 imaging is more sensitive than T1 to differences between damaged tissue and intact tissue and so is useful for detecting lesions.
In the MRI procedure, the subject lies prone with his or her head centered within the magnetic coils and must remain as still as possible. (Corrections are made for the slight head and brain movement produced by pulsations of cerebral blood flow.) Density differences in the imaged slice through the head are portrayed as colors, in this case producing a horizontal cross-section of the head and of the brain. Although the MRI procedure is safe, the noisy, enclosed magnetic coils produce claustrophobia in some people. People with surgical implants that contain metal should not undergo MRI because of the magnets’ strength.
MRI image resolution is derived from the strength of the magnetic field, measured in teslas. A 1.5-Tesla magnet is referred to as a 1.5T magnet; magnets for medical use range from 0.5T to 3.0T. The resolution of a large magnet is 1-cubic millimeter voxels, twice that of PET. Despite this high image resolution, like PET, each MRI voxel indirectly infers the activity of thousands of neurons.
Functional Magnetic Resonance Imaging (fMRI) refers to a series of in-vivo MRI techniques employed most frequently to provide a link between brain anatomy and neural processes such as cognition, perception, and sensation. The blood oxygenation level-dependent (BOLD) response is the basis of the most commonly used fMRI technique employed to image neural processes. BOLD fMRI is based on the assumption that there is a disproportionate increase in the supply of oxygenated blood to the site of increased neuronal activity (Huettel et al., 2004). Hemoglobin is diamagnetic when bound to oxygen, but deoxygenated hemoglobin is paramagnetic. Thus, alterations in hemoglobin oxygenation result in local distortions to the magnetic field that the MR system has applied to the brain (Jezzard, Matthews, & Smith, 2001). Essentially, BOLD fMRI produces image contrasts based on the ratio of oxyhemoglobin to deoxyhemoglobin that accompanies neuronal responses (Ogawa, Menon, Kim, & Ugurbil, 1998). Ogawa and colleagues (Ogawa, Lee, Kay, & Tank, 1990) were able to demonstrate that in-vivo changes in blood oxygenation could be detected with MRI, and several demonstrations of BOLD signal changes in normal humans during stimulus presentations and task performance quickly followed (Bandettini, Wong, Hinks, Tikofsky, & Hyde, 1992; Kwong et al., 1992; Ogawa et al., 1992). It should be noted, however, that researchers continue to debate scientifically with regards to what BOLD fMRI truly measures (Logothetis, 2002). Because the BOLD response is based on the oxygen content of brain vasculature, the peak of the hemodynamic response (increased oxygenated blood flow to the site of neuronal response) typically occurs within 6 to 9 seconds after the actual firing of the neurons. Similarly, this lag results in the BOLD effect lasting longer than neural activity within a brain region (McRobbie et al., 2003). Delay of the hemodynamic response results in the BOLD signal typically rising and falling within 12 to 20 seconds of the assumed neuronal firing in response to the presentation of a stimulus or event, which must be taken into consideration when analyzing fMRI data.
A fundamental limitation to experimental design with fMRI is that the measured signal changes are small (Jezzard et al., 2001). The blocked design has been the primary type of experimental design utilized in fMRI experiments and is statistically powerful. Blocked designs consist of multiple, discrete epochs (time periods) that alternate the presentation of two or more experimental conditions. Each epoch is many seconds in duration, and epochs of like conditions are averaged together to obtain neural responses that are contrasted with responses for other conditions. For example, in a study of response inhibition, one may have condition A where both a respond cue and an inhibit cue are presented on multiple occasions in a random manner. The control condition may involve only the presentation of the respond cues, with the contrast between conditions A and B interpreted as reflecting neural processes related to inhibitory control. Such blocked designs are disadvantageous in part because of the limited information regarding the time course of neural response to a given stimulus type that may be obtained, as well as the increased risk of confounding influences from fluctuations of attention or arousal over time that may differentially affect the experimental conditions given their extended durations.
In contrast to blocked designs, event-related fMRI designs allow one to investigate brain activation in response to individual event types because these are interspersed among each other. Using our example from the blocked design described previously, an event-related design would involve the random or pseudo-random presentation of the respond and inhibit cues within the same epochs, with subsequent off-line averaging of brain activation for all stimuli of a given type presented in the experiment. Event-related designs thus have several advantages such as reducing the potential confound of having separate conditions of interest in different epochs, allowing for increased flexibility in the use of stimuli, exploring changes in neural response over time, and facilitating post-hoc binning of stimuli or responses based on experimenter interest (e.g., correct versus incorrect responses to a stimulus). A significant disadvantage of event-related designs is a reduced signal-to-noise ratio (SNR) and loss of statistical power, which may be addressed in part by increasing the number of stimulus trials.
The resulting functional MR images of the brain are typically presented as colored blobs superimposed on a grey-scale two-dimensional anatomical background image, or as a color overlay on a 3D-surface rendered image of the cortical surface (Jezzard et al., 2001). It is important to consider that BOLD fMRI does not measure neuronal activity directly, but is believed to be an indirect measure of changes in synaptic activity in gray matter (Jezzard et al., 2001). Therefore, the colored blobs may not represent neuronal activity per se, but areas of statistically different MR signal, with the intensity of the color usually representing the degree of statistical confidence in the results of the contrast of interest.
As noted in the introduction, fMRI has been employed increasingly since its inception to investigate the neural correlates of various functions of the brain in healthy and ill individuals. For example, fMRI studies have demonstrated brain abnormality during the performance of a variety of cognitive tasks in patients with schizophrenia (Ford et al., 2004; Tan, Choo, Fones, & Chee, 2005), bipolar disorder (Altshuler et al., 2005; Roth et al., 2006), multiple sclerosis (Wishart et al., 2004), and traumatic brain injury (McAllister et al., 2001). fMRI has also been employed to study neural correlates of developmental changes in cognitive processes (Booth et al., 2003; Konrad et al., 2005), and recently to investigate the effects of medications on brain activation in various clinical populations (e.g., Saykin et al., 2004).
|Magnetic Resonance Imaging||Functional Magnetic Resonance Imaging|
|Views anatomical structure||Views metabolic function|
|High spatial resolution||Long-distance resolution|
|Utilized for experimental purpose||For diagnostic purpose|
|Studies water molecule’s hydrogen nuclei||Calculates oxygen level|
|Less expensive||More expensive|
|Views tissue with respect to space||Views tissues differences with respect to time|
|Widely used||Not much in use|