David H. Zald and Clayton Curtis
Modern neuroimaging techniques enable researchers to noninvasively assess brain structure and function in humans. The knowledge gained from these techniques has led to a revolution in our understanding of brain-behavior relationships and has dramatically altered the psychological sciences. Several brain imaging techniques are currently in wide use, including computerized tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), single photon emission computed tomography (SPECT), magnetoen-cephalography (MEG), and near infrared optical imaging. In this chapter we focus on functional MRI (fMRI) and PET techniques because of their enormous impact on the psychological sciences. Although they are most often used in isolation, both PET and fMRI are adaptable to a multitrait-multimethod (MTMM) approach toward assessment. Indeed, it may be argued that these techniques require integration within a broad multi-method framework if they are to reach their full scientific potential. This chapter provides a brief primer on fMRI and PET imaging, followed by a discussion of the benefits of placing neuroimaging data within a MTMM approach.
Depending on the specific technique used, PET and MRI scanners can assess a number of different statewise and traitwise characteristics of the brain. These include measurement of brain structure (MRI), neurotransmitter functioning (PET and magnetic resonance spectroscopy [MRS]), glucose metabolism (PET), blood oxygenation (PET and fMRI), and blood flow (PET and fMRI). Because changes in neural activity are accompanied by changes in metabolism, blood oxygenation, and blood flow (Raichle, 1988), PET and fMRI measurements of these physiological variables allow researchers to index changes in brain functioning in relationship to specific perceptual, cognitive, and behavioral tasks. However, PET and fMRI take very different approaches to these measurements. It therefore is useful to first discuss how these measurements are made in each technique.
fMRI PHYSICS AND PHYSIOLOGY
When biological tissue is placed within a strong externally applied magnetic field, denoted Bg, the axis of individual nuclei, like hydrogen, tend to align with the field. Nuclei line up with the field because this results in the lowest energy state of the system. Outside the magnetic field, the alignment of all nuclei tends to be randomly oriented and produce no net magnetic field. However, when placed in a strong magnetic field, the nuclei align in the same direction as the field. This alignment produces a net magnetization, referred to as M, which represents the sum of all of the magnetic moments of the individual hydrogen nuclei (see Haake, Brown, Thompson, & Venkatesan, 1999, for a full review of MRI physics).
Hydrogen nuclei consist of a single positively charged particle, the proton, which spins around its axis. An individual proton not only spins around its axis, but also precesses (revolves) about the external magnetic field, much like a top both spins around its axis and precesses about the direction of gravity's magnetic field. Importantly, each type of atomic nuclei precesses at a characteristic frequency, the resonance or Larmor frequency, which is directly proportional to the strength of the applied magnetic field. This proportional dependence of the resonance frequency on the applied magnetic field forms the basis for MRI. Specifically, by spatially manipulating the field strength and measuring resonance frequencies, it becomes possible to resolve the source and location of signals from the brain.
When all the nuclei in a sample are at the resting equilibrium state, the net magnetization of the nuclei are aligned with the field, and no MR signal can be detected because each of the nuclei precesses at the same rate, but out of phase with one another. Magnetic resonance occurs when a radiofrequency (RF) pulse is transmitted to the sample at the Larmor frequency of a specific type of nuclei. For instance, hydrogen (H) precesses at a frequency of 64 MHz in a 1.5 Tesla (T) magnetic field (standard clinical scanners possess a 1.5 T field strength, whereas research dedicated scanners often use higher field strengths, such as 3 T, 4 T, or even 7 T). When an RF pulse is applied at the Larmor frequency of H, energy is selectively absorbed by H nuclei, exciting their spins from their lower resting state to an unstable higher energy state. The RF pulse also deflects the net magnetization of the nuclei away from the direction of the external magnetic field and causes each precessing nuclei to pre-cess in phase with one another (i.e., they become phase coherent). At the point in time when the RF field is extinguished, the nuclei are in an excited, high-energy state because the axes of their small magnetic fields are not oriented with that of the strong external field. This unstable state decays quickly as the nuclei begin to realign with the external field. The precessing nuclei radiate the energy that they absorbed from the RF pulse as the phase coherence exponentially decays and the net magnetization of the nuclei realign with the external magnetic field. The energy that is emitted during this brief process induces a detectable current (known as the free induction decay or FID) and is detectable by an RF coil placed around the stimulated sample (i.e., the subject's head). This is the MR signal and forms the basis of all MRI techniques.
When the application of the RF energy is terminated, the system reapproaches equilibrium, a process known as relaxation. Different types of tissue have different rates of relaxation, which is why we can obtain MR images that can distinguish between gray and white matter, bone, cerebrospinal fluid, and vasculature. For most functional MRI studies, the critical source of contrast derives from changes in the oxygen content of cerebral vasculature, typically referred to as Blood Oxygen Level Dependent (BOLD) signal (Bandettini, Wong, Hinks, Tikofsky, & Hyde, 1992; Kwong et al., 1992; Ogawa, Lee, Kay, & Tank, 1990; Ogawa etal., 1992).
The fMRl signal is a function of the metabolic demands of local neural activity. However, the coupling between the measured BOLD signal and the underlying neural activity is neither direct nor straightforward (Heeger & Ress, 2002; Logothetis, Pauls, Augath, Trinath, & Oeltermann, 2001). As neural activity increases, there are changes in both the amount of blood flow to the region and a change in the concentration of oxygenated and deoxygenated forms of hemoglobin. The oxy- and deoxyhemoglobin have different magnetic properties (diamagnetic vs. paramagnetic) and because of this behave differently within a magnetic field. The paramagnetic properties of deoxyhemoglobin lead it to have a greater interaction with the magnetic field than oxyhemoglobin such that shifts in the concentration of oxy- and deoxyhemoglobin cause changes in the MR signal. Specifically, as the concentration of oxyhemoglobin increases in response to neural metabolic demands, the BOLD signal increases. Importantly, the BOLD signal does not convey an absolute value—it is only a relative measure. Therefore, one rarely sees attempts to compare the BOLD signal between individuals. Instead, research focuses on the location and magnitude of relative changes in BOLD during different task conditions.
There are three characteristic phases of the hemodynamic response to a neural event (Figure 13.1a).
Was this article helpful?