8 mm

~ 100 sec







~ 1 mm

~ 1 sec


[Oxy/deoxy Hb]





~ 1 mm

~ 1 sec

Radio waves

[Oxy/deoxy Hb]

Metal fragments

Note: Salient characteristics of structural and functional neuroimaging methods. Structural and functional methods are characterized in the upper and lower panels, respectively. Invasiveness and repeatability are rated on relative scales from low (+) to high (+++). * Autopsy, the gold standard or invasiveness is not rated. Spatial and temporal resolution refer to the smallest distance or duration that the technique is capable of sampling and resolving. Energy source is the physical phenomenon that the technique measures. Physiological measure is the property of the tissue or tracer that is measured (brackets indicate concentration; Hb = hemoglobin). Other limitations detail the primary drawbacks of the technique. N/A = not applicable.

17.2.3 Computerized Tomography without Contrast

Computerized tomography (CT) was developed in the early 1970s and is essentially a sophisticated x-ray exam. The technique makes use of mathematical algorithms to combine multiple x-ray images taken from different angles. The output is a tomographic image that represents a single slice through the brain. The procedure has characteristics similar to those of x-rays but is not as susceptible to shadows because the image is a reconstruction from multiple views.

17.2.4 Computerized Tomography with Contrast

The development of contrast media that were opaque to x-rays greatly expanded the capabilities of CT. The contrast medium is injected into a blood vessel, and a CT scan is performed in a procedure called cerebral angiography. This technique allows visualization of aneurysms, tumors, or other abnormalities that have characteristic effects on blood flow. Because the contrast medium must be injected into the patient, this method is more invasive than a CT scan performed without contrast. There are also certain risks associated with using contrast agents, such as allergic reactions to the dyes used in cerebral angiography, as well as the usual risks associated with venipuncture.

17.2.5 Magnetic Resonance Imaging

The most recently developed technique for viewing the structure of the brain is magnetic resonance imaging (MRI). It is less invasive and more repeatable than CT because it measures the response of hydrogen nuclei to radio frequency pulses in the presence of a large static magnetic field. There are no known risks of the technique itself, unlike the x-rays used in CT, so one may repeat studies with the same patient without any concerns about chronic exposure to harmful radiation. MRI's greatest advantages as an imaging modality are its submillimeter spatial resolution and its ability to resolve subtle differences in soft tissues, primarily based on their water content. The main safety concern with MRI is to avoid exposing patients to the scanner's strong magnetic field if they have any metal in their bodies, such as pins, surgical clips, or pacemakers. A typical MRI machine operates at 1.5 tesla, equivalent to 30,000 times the earth's magnetic field, and produces powerful attractive forces on metallic objects. In addition, the changing magnetic gradients that are used to generate images may induce electrical currents in metal objects that cause them to heat up.

17.3 FUNCTIONAL NEUROIMAGING TECHNIQUES 17.3.1 Electroencephalography

Electroencephalography (EEG) has a long history in human psychophysiological research as a noninvasive and repeatable technique for measuring electrical fields generated by the brain (see Brannon and Roitman, this volume; Pouthas, this volume). It was only recently, however, that application of powerful computational methods made it possible to use EEG to localize the sources of these electrical potentials and create maps of their distribution on the cortex. EEG studies use electrodes placed on various areas of the scalp to record the electrical activity generated by the brain. The number of electrodes employed by researchers using EEG has increased over the years, and recent studies have used 128 or more electrodes. Using triangulation algorithms, one can locate the electrical field sources generating the EEG signal to within a few millimeters. Because the electrical signals generated by the brain are distorted somewhat as they pass through the skull and scalp, current techniques are probably close to the practical limit for the maximal spatial resolution that can be achieved with EEG. Another fundamental limitation of this technique is that only a highly organized, parallel arrangement of neurons is capable of generating a strong enough voltage difference to be recorded from the scalp. Consequently, using surface electrodes, EEG primarily measures electrical potentials from the cortex. However, what EEG lacks in spatial resolution it amply makes up for in its temporal resolution, which is on the order of 1 msec.

17.3.2 Magnetoencephalography

Magnetoencephalography (MEG) is analogous to EEG except that it uses special superconducting detectors to measure magnetic dipoles generated by the flow of electrical current within the brain. It has characteristics, advantages, and disadvantages similar to those of EEG, with the exception that the required equipment is much more expensive and is not widely available.

17.3.3 Positron Emission Tomography

Positron emission tomography (PET) creates images of brain activity using radioactive isotopes such as 150-labeled water. When such a substance is injected into the bloodstream, a PET scanner can measure changes in either blood flow or blood volume. Using radioactively labeled glucose analogs, metabolic activity itself may be imaged. The radioactive isotopes used in PET emit positrons, which are electron antimatter. A positron may travel up to a few millimeters from where it was emitted before it collides with an electron. The interaction between a positron and an electron creates a pair of high-energy gamma rays that travel in opposite directions. Colli-mated gamma detectors surround the participant's head and determine the line along which the gamma rays were emitted. Therefore, a single source is inferred to arise from the intersection of multiple gamma rays' axes of travel. A computer reconstructs the locations of all the source events to generate a map of activity within the brain. PET's ultimate spatial resolution of about 5 mm is determined by limitations on detector sensitivity and the physical process by which an event is generated. Because the technique requires injecting radioactive substances, it is a moderately invasive imaging modality, and the need to minimize participants' exposure to radioactivity limits the repeatability of PET scans. In addition, data must be collected over periods of about a minute or so because of constraints on the concentration of radioactive isotopes that may be used. Consequently, the temporal resolution of this method is quite poor, and each imaging run results in a single snapshot of activity. Finally, the radioisotopes used in PET must be created in a particle accelerator, and the practical issues associated with this additional equipment make PET scanners somewhat uncommon.

17.3.4 Single-Photon Emission Computed Tomography

Single-photon emission computed tomography (SPECT) is a technique analogous to PET except that it uses lower-energy radioisotopes. The photons emitted by the isotopes are directly detected, unlike in PET, where the proximal signal of emitted gamma rays is an energetic by-product of the collision of a positron with an electron. As in PET, SPECT is limited by the trade-offs between detector sensitivity, spatial resolution, and radioactivity exposure to the participant. These trade-offs, combined with variability due to photon scatter, limit spatial resolution in SPECT to about 8 mm. SPECT's dependence on radioactive isotopes means that it is somewhat invasive and limited in its repeatability. Another limitation of SPECT is that it is less sensitive than PET because SPECT uses lower-energy photons. Finally, fewer radioactive ligands have been developed for SPECT than for PET.

17.3.5 Optical Recording

A relatively recent addition to the variety of functional neuroimaging tools is that of optical recording, which was developed in the 1980s and has been refined over the last few decades (Gratton and Fabiani, 2001). Optical recording arose as a noninvasive modification of an earlier technique that imaged neural changes over the course of milliseconds using voltage-sensitive dyes. It is similar to EEG in that an array of detectors is used to focus on the region of interest in the brain. Its temporal characteristics are similar to those of EEG and MEG, but it is based on a fundamentally different principle. In humans, a near-infrared laser projects a beam of light through the scalp and skull onto the cortex, where the beam is absorbed and scattered to varying degrees depending on the oxygenation state of hemoglobin in the blood. A photomultiplier then records the number of photons that are reflected back through the skull and scalp, and the transmission time gives information about the degree of scattering and absorption of the signal. Therefore, the optical signal actually measures the hemodynamic response to a neural event. The technique shows good temporal correspondence to event-related potentials recorded with EEG and good spatial correspondence with functional magnetic resonance imaging (Gratton et al., 1997). Further research has shown that the hemodynamic signal is proportional to the amplitude of the neuronal signal, providing additional support for the quantitative use of hemodynamic methods in the study of brain function (Gratton et al., 2000, 2001). In animal models, optical recording may be performed directly on the exposed cortex and may use voltage-sensitive dyes to visualize changes in neural activation dynamically. In this more invasive version of the method, optical recording is capable of measuring effects with a spatial resolution of 0.1 mm and a temporal resolution of milliseconds. Like EEG and MEG, this technique is limited to measuring functional activity in the cortex, but it provides another source of information on brain function using a different physical property of the tissue than the other imaging modalities.

17.3.6 Functional Magnetic Resonance Imaging

Functional magnetic resonance imaging (FMRI) is a technique developed within the last decade or so that uses an MRI scanner to detect changes in blood oxygenation levels while participants perform some kind of sensory, motor, or cognitive task. Like standard MRI, it is noninvasive and highly repeatable because it measures the effects of radio frequency pulses on hydrogen nuclei in a static magnetic field. The signal strength of FMRI is a function of the magnitude of the static magnetic field, but it always involves a trade-off between spatial and temporal resolution. In current incarnations, FMRI can achieve spatial resolutions of less than a millimeter (Hyde et al., 2001; Menon and Goodyear, 1999) and a temporal resolution measured in milliseconds (Menon et al., 1998), although these extremes are obtained by sacrificing resolution in the other domain. With standard systems performing whole-brain imaging, spatial resolutions of a few millimeters and temporal resolutions of a few seconds are commonly obtained. The effect on which most FMRI depends is called blood oxygenation level-dependent (BOLD) contrast (Ogawa et al., 1990). BOLD contrast relies on choosing imaging parameters that are sensitive to the relative concentrations of oxy- and deoxyhemoglobin in the blood, which produce characteristic effects in FMRI because of their different magnetic susceptibilities. When neural activity increases, local blood flow gradually increases with a rise time of 4 to 6 sec. The signal returns to baseline after another 6 to 8 sec, and this extended time course of activation is termed the hemodynamic response. Although there remains some debate about the exact physiological processes that produce a hemo-dynamic response to neural stimulation, there is general agreement that the signal that FMRI measures corresponds in reliable ways to underlying neural processing (Logothetis et al., 2001). Indeed, most early FMRI experiments used a variety of simple sensory and motor tasks in which the outcome was predicted based on data from other techniques (Bandettini et al., 1992; Belliveau et al., 1991; Binder et al., 1993; Hammeke et al., 1994; Ogawa et al., 1990, 1992; Rao et al., 1993). The great strength of FMRI as an imaging modality is that it allows one to view physiological changes throughout the brain noninvasively and repeatedly with good spatial and moderate temporal resolution. These combined characteristics make FMRI ideally suited to determine the networks of brain regions that are involved in specific aspects of cognitive function.

The cascade of events that determines the FMRI signal and generates the signal for other functional neuroimaging techniques along the way is illustrated in Figure 17.1. Sensory, motor, or cognitive activity causes a localized increase in neural activity. The graded potentials and action potentials in the activated tissue result in increased metabolic demand for glucose and oxygen. The heightened metabolic demand is satisfied by a gradual increase in regional blood flow over the course of several seconds, termed the hemodynamic response. The increased blood flow delivers a relative excess of oxyhemoglobin, which decreases the regions' magnetic


Measurement electromagnetic changes (EEG, MEG)

increased glucose uptake (PET)

increased isotope flow (PET, SPECT)

altered infrared scattering (optical recording)

increased BOLD signal (FMRI)

FIGURE 17.1 Diagram of the flow of physiological events and the associated signals measured by various functional neuroimaging techniques.

susceptibility and produces an increase in the measured BOLD signal. The BOLD signal is taken as indirect evidence of neural activity underlying the behavioral task or cognitive event of interest.

0 0

Post a comment