PET and fMRI Procedures

During the 1990s PET was the gold-standard of neuroimaging. In PET a small amount of a radiotracer is injected intravenously into the subject, and the concentration of tracer in brain tissue is measured by the scanner (see Fig. 2.2, top). While decaying, the radionucleotides emit positrons that, after traveling a short distance (3 to 5 mm) encounter electrons. The two types of particles annihilate each other, resulting in the emission of two gamma rays in opposite directions. The image acquisition is based on the detection in coincidence of the gamma rays in opposite directions by crystal detectors. Image reconstruction uses lines of response connecting the coincidence detectors through the brain (for more details on PET methods and a more comprehensive description of research uses of PET in psychiatry, see Chapter 6).

Once reconstructed, PET scans for (active and control conditions) are spatially normalized to stereotaxic atlas, and group averaged. Then a voxel-by-voxel parametric difference contrast is carried out, resulting in a difference image that is thresholded to a statistical cutoff, which is overlayed on the same subjects' MRI for registration and visualization (see Fig. 2.2, bottom). State-of-the-art PET scanners have a resolution of 6 mm, allowing precise localization of cortical and some subcortical structures.

Active Control Difference z-score state state (Z-score threshold)

Figure 2.2. Positron emission tomography (PET). Top: Atypical scanner. Bottom'. Images in the control task are subtracted voxel-by-voxel from images in the active task. A statistical cutoff is applied to the difference image (bottom right). See ftp site for color image.

Active Control Difference z-score state state (Z-score threshold)

Figure 2.2. Positron emission tomography (PET). Top: Atypical scanner. Bottom'. Images in the control task are subtracted voxel-by-voxel from images in the active task. A statistical cutoff is applied to the difference image (bottom right). See ftp site for color image.

However, PET has no time resolution. 15O-water, the radiotracer used for activation studies, has a half-life of 2 min. Due to the long integration time of each scan, only "block" designs are possible. Scans of active tasks and control tasks are separated by periods of no acquisition, lasting 10 min, to allow a complete return to baseline of the activations and a complete decay of the radiotracer. Typically 8 to 12 scans are acquired, with 2 to 4 repetitions of each task. The main limitation of PET is the radiation exposure, particularly for women of fertile age and children, which limits repeated testing, and the need of a cyclotron nearby, with a high cost.

In contrast to PET, fMRI is more user friendly and many more studies are available with the technology. It has rapidly replaced PET as the most popular form of neuroimaging. The MRI signal is induced with a strong magnet. When body tissues, rich in water, are placed in a strong magnetic field, all protons in the water molecules become systematically oriented. Radio-frequency pulses are then applied, producing spins in the protons. When they are no longer applied, the proton spins return to their original state releasing radio-frequency waves. Radio-frequency emissions vary with water density in different tissues and can be registered with detector coils (see Fig. 2.3, top).

The fMRI approach measures slight differences in radio frequencies produced by changes in local blood flow in activated regions during cognitive or motor tasks. An increase in oxygen accompanies increased local brain activity. Functional fMRI can measure the ratio of deoxygenated to oxygenated hemoglobin in order to obtain a measure of regional blood flow (yielding the BOLD, or blood oxygen level detection, method). For more details about the BOLD method and a more comprehensive description of research uses of fMRI in psychiatry, see Chapter 6.

Figure 2.3. Functional magnetic resonance imaging (fMRI). Top left. Typical magnet. Top right basic principles of proton density MRI, employed in fMRI: (a) protons systematically orient in a magnet; (b) radio-frequency pulses are applied, producing spins in the protons; (c) proton spins recover the original state releasing radio-frequency waves. Bottom: examples of fMRI data during cognitive tasks overlayed on three-dimensional renderings of the human head. See ftp site for color image.

Figure 2.3. Functional magnetic resonance imaging (fMRI). Top left. Typical magnet. Top right basic principles of proton density MRI, employed in fMRI: (a) protons systematically orient in a magnet; (b) radio-frequency pulses are applied, producing spins in the protons; (c) proton spins recover the original state releasing radio-frequency waves. Bottom: examples of fMRI data during cognitive tasks overlayed on three-dimensional renderings of the human head. See ftp site for color image.

Functional fMRI allows for higher spatial resolution [about 3 mm for 1.5-tesla (T) machines] and possesses better time resolution than PET. However, even if the specific imaging technique allows for superfast data acquisition (2 to 3 sec for an entire brain volume), temporal resolution is limited by the slowness of the hemodynamic response, which peaks at 5 sec from a discrete neural event, and decays slowly (in 12 to 15 sec), resulting in overlapping responses from temporally close events.

Most typically, fMRI studies utilize block designs (on-off), with alternations of tasks each typically lasting 30 to 60 sec. Since each volume is acquired in 2 to 3 sec, a large number of images can be included in task comparisons (typically 50 to 100). Voxel-by-voxel statistical comparisons are carried out with independent T tests and are then thresholded for statistical significance. Resulting differences can be overlayed on three-dimensional MRI renderings of the subject's head for impressive effects (see Fig. 2.3, bottom).

Unlike PET, fMRI is noninvasive, allowing for repeat studies in children and patients, and it can use clinical scanners with moderately priced hardware upgrades. The main limitations are claustrophobia, typically present in no more than 1 percent of the subjects, and less flexibility than PET with experimental situations because the intense magnetic fields disallow metal objects in the experimental field. Inhomogeneities in the fMRI signal are present in the ventral frontal areas and anterior temporal areas (due to nearby air cavities), making the study of emotion and memory more problematic (since these areas are implicated in those processes). One of the biggest problems is the very artificial and noisy environment in which subjects must be tested, which can easily reduce the ability of subjects to attain and sustain realistic emotional states. Thus PET may be considered better for monitoring affect (since radioactive and behavioral challenges can often be done outside the scanner, even though this is not possible for rapidly decaying tracers such as radioactive water), while fMRI, because of its temporal features, is often the better tool for monitoring cognitive processes than for affective ones.

Although functional neuroimaging methods have contributed greatly to our understanding of the neural correlates of human cognition and emotion (Posner and Raichle, 1994), a number of methodological hurdles make the study of emotion through these methodologies particularly challenging. To properly evaluate the rapidly increasing number of studies in the field, it is important to consider some of these difficulties.

0 0

Post a comment