This chapter aims to provide an introduction to the neuroimaging modalities that are important to drug discovery, development and treatment. I describe the basis of each technique and compare each one with the others, offering a roadmap through the jungle of different types of neuroimaging, including in particular magnetic resonance (MR) techniques, positron emission tomography (PET), and electrophysiological measurements. I will describe the nature of each measurement or imaging signal and the functional readout from the biology, as well as what they are used for and how they are applied. I discuss the strengths and weaknesses of the different techniques, as one modality rarely has the capability of answering all of the relevant questions. I discuss principally the application of the neuroimaging modalities common to animals and humans, focusing in particular on noninvasive techniques.
I begin the chapter with an overview of the biological domains overseen by the different imaging techniques before describing each modality in turn. I then discuss the advantages and disadvantages of the modalities.
1.1 Windows to the Brain
There is no single neuroimaging modality that can offer the complete picture of brain structure or function that is needed for drug development and assessing disease or treatment effects. Each modality has its own sensitivity profile that may offer a piece of information directly linked to the action of a pharmacological agent, e.g. receptor binding, or more likely an indirect marker of pharmacological or disease action, e.g. local hemodynamic activity. It is necessary to appreciate the basis of the neuroimaging signals and their limitations in order to interpret them. This is particularly important in the process of decision making in drug discovery and development. Over- or misinterpretation of neuroimaging data could lead to the expensive failure of new compounds late in the drug development process, which could have been anticipated earlier. Underuse of valuable information contained in neuroimaging data might lead to promising compounds being abandoned unnecessarily.
The neuroimaging modalities applied in humans and animals may be categorized as offering either a structural or a functional readout. While often a useful distinction, the boundary between structure and function may be blurred, in particular at the micro scale and where function and structure are intimately linked. An elegant example lies in the characterization of the diffusion and distribution of water in the human brain. Diffusion-based MR imaging is regarded as offering structural information. However, in the context of stroke, where there is a redistribution of intra- and extracellular water, structural information and functional viability of tissue become closely linked. Furthermore, the diffusion of water in the brain has been demonstrated to change over the comparatively short timescales associated with structural plasticity during learning.1 In more general terms, the function of a neuron might be said to be defined by its connections, local and remote, to other neurons, such connections being identified noninvasively by diffusion tractography techniques in white matter.
1.1.1 Brain Structure
Alterations in brain structure with disease or treatment range from the scale of gross structure, including regional tissue volumes, through to micro and molecular structure, such as receptor distributions. Techniques sensitive to alterations in brain structure most commonly offer a long-term marker of disease progression or modification with treatment. This is because macro-scale alterations in structure, e.g. brain atrophy, normally develop over long periods, typically years in humans.
Structural magnetic resonance imaging (MRI) techniques are particularly suited to examining long-term structural alterations because of their noninvasiveness and therefore their ability to follow longitudinal changes in a research cohort. Traditional MRI structural markers, such as T1 and T2 relaxation time constants,i can reveal fat and water distribution (as well as other physicochemical differences), thus distinguishing gray matter, white matter, and cerebrospinal fluid, while developments in MR image contrast, such as magnetization transfer, can offer information on pools of liquid and macromolecular water with potential uses in examining demyelinating conditions.2 Microscopy is invasive and so only available post mortem, but is useful for characterizing altered brain microstructure. Microstructural alterations at the scale of molecular receptors can be probed with radiotracers using PET, looking for example, at alterations in spatial distributions of specific receptors. Such molecular-level alterations are often interpreted as having direct functional consequences.
1.1.2 Brain Function
Brain function is characterized by many different activities of the brain accessible to neuroimaging techniques. Functional processes are normally those occurring over short timescales from small fractions of a second to minutes. They support or are associated with information processing or the transmission and reception of signals in the brain. In animals, it is possible to place electrodes within brain tissue and to record individual cellular potentials or the electrical activity of groups of cells. In humans, this is only possible in rare circumstances when electrodes are implanted for reasons of surgical diagnosis, for example to identify seizure foci or brain stimulation. More commonly, we rely on indirect or ensemble measures of brain function, which may either be focused on specific portions of brain tissue or distributed across the brain to study the brain at the systems level.
The brain’s neuronal activity is linked or coupled to its blood supply,3,4 allowing altered hemodynamics and therefore regional blood oxygenation to be used as a marker of altered function. The growth of functional neuroimaging studies in humans in the last 20 years has exploited these phenomena, first with PET and more recently and on a larger scale with functional magnetic resonance imaging (fMRI). Optical imaging techniques, including invasive cortical infrared imaging and noninvasive near-infrared spectroscopy (NIRS) rely on changes in cerebral blood oxygenation.
A continuous energy supply to brain tissue is essential for maintaining ion concentration gradients and therefore electrical potentials. Some neuroimaging techniques have been developed that are sensitive to alterations in cerebral metabolism and, in particular, the biochemical species involved in energy supply. PET can be made sensitive to oxygen or glucose metabolism. fMRI techniques are also emerging that are able to quantify cerebral oxygen consumption. Magnetic resonance spectroscopy (MRS) can measure chemical concentrations within brain tissue and therefore monitor the energy status of tissue through species such as high-energy phosphates. With the appropriate use of tracers detectable with MRS, rate constants and chemical fluxes can be estimated to quantify cerebral metabolism.5
Specific molecules engaged in signaling or processes associated with synaptic transmission can be studied using PET and MRS. With the development of appropriate PET ligands, specific receptor activity and the distribution of receptors can now be assessed. Receptor activity has the advantage of being directly associable with the action of pharmacological agents in the brain, whereas MRS is able to measure the bulk concentration of the more common neurotransmitters and their modulation with disease and pharmacological intervention. It must be borne in mind, however, that the relationship between neurotransmitter concentration and brain function may be a complex one depending on the availability or otherwise of the neurotransmitter.
Noninvasive measures of electrophysiological activity can be made from the scalp by recording electrical potentials (i.e. electroencephalography; EEG) or the tiny magnetic fields associated with neuronal activity (i.e. magnetoencephalography; MEG). In order to be detectable at a distance of centimeters from their source, these signals necessarily arise from the coordinated activity of populations of neurons.