resonance imaging (sMRI) methods, in general, is to determine the size and shape of anatomical structures of the brain (Jenkinson and Chappell 2018). Various methods have been developed for investigating grey matter and white matter, two major regions of the CNS.
1.2.4.1 T 1‐Weighted Structural MRI
The most common sMRI data is acquired by T1‐weighted imaging. Imaging is acquired by weighing on the ‘T1’ value, which refers to the time constant for longitudinal relaxation, an index of the rate for protons to return to equilibrium. Critically, this value varies depending on the biochemical components of tissues: the fat‐containing tissue (e.g. neural fibres of white matter) has a shorter T1 compared to the tissue less rich in fat (e.g. grey matter). In a T1‐weighted image, the signals collected would preferably show a higher intensity (i.e. brighter) for brain tissue with a higher content of fat and lower intensity for tissue with a lower content of fat. Therefore, the spatial distribution of white matter and grey matter can be differentiated by image intensity. The greatest advantage of T1‐weighted imaging is that it can be analyzed using different methods to disclose information on brain morphology. For example, tissue‐specific segmentation is a method that separates grey matter, white matter and the space of CSF from the whole‐brain image. Voxel‐based morphometry (VBM) can be used to estimate the amount of grey matter and white matter within each voxel, which can be further used for a group‐based comparison. The method has been widely used for clinical investigation, such as assessing the reduction in hippocampal grey matter between patients with Alzheimer's disease and healthy controls. In addition, the surface‐based analysis is widely used to estimate the thickness of the cortical tissues and the volume of cortical and subcortical regions, which are critical structural features associated with clinical factors (see Section 2.3).
1.2.4.2 Diffusion MRI
While the T1‐weighted image provides a spatial feature of different brain regions, it provides less information regarding how the brain forms a connectional network. The key to understanding the connection between brain regions is to estimate the orientation of neural fibres. Diffusion magnetic resonance imaging (dMRI) is an MRI method to estimate the distribution of the ‘fibrous’ space in the brain. The method is based on the phenomenon that water molecules spread less freely in the compartment abundant of axons because the freedom to spread is limited by the axons aligned in the same direction. In contrast, the molecules spread more freely in the fluid space, such as the ventricles, where less hindrance exists to restrict the direction of spreading. Diffusion tensor imaging (DTI) is developed to quantify the directionality of diffusion. There are two major applications of dMRI. Firstly, it helps to examine the microstructural properties of the white matter (Jenkinson and Chappell 2018). For example, fractional anisotropy (FA) is a widely used index related to axonal density, the myelination of nerve fibres, and the membrane permeability (Jones et al. 2013). Secondly, dMRI is useful for exploring the structural connectivity of the brain, i.e. how the brain is wired by neural fibres. At present, it is the only tool that can probe the structural connectivity of the human brain in vivo (Jenkinson and Chappell 2018). Tractography has been used to visualize the streamlines that pass between different brain regions. The results provide further information about how brain regions are wired to form a network (see Section 2.3).
1.2.5 Functional MRI Methods
The fMRI methods have two major applications. A task‐based fMRI study investigates the brain signals associated with mental functions. Imaging is conducted when subjects are performing a behavioural task that induces the mental functions of research interest (see Section 2.2). A resting‐state fMRI study investigates the intrinsic activity of the brain, i.e. the spontaneous activity when subjects are not perturbed by external stimuli (see Section 2.4). The fMRI methods can be further categorized according to the nature of the brain signals detected, as discussed below.
1.2.5.1 Blood‐Oxygen‐Level‐Dependent fMRI
The blood‐oxygen‐level‐dependent (BOLD) fMRI detects the changes in the proportion of deoxygenated and oxygenated haemoglobin in the brain. This metabolic event (i.e. oxygenation of haemoglobin) is further associated with neural activity. Firstly, the brain region with increased neural activity is associated with more energy consumption, i.e. for synaptic activity. Secondly, oxygen consumption is associated with increased CBF and changes in cerebral vessel volume, leading to an over‐supply of the oxygenated vs. deoxygenated haemoglobin (see Section 2.2). Finally, deoxygenated haemoglobin shows a paramagnetic property that disturbs the local magnetic field and decreases the MR signal (Thulborn et al. 1982). A higher MR signal reflects the effect of an increased proportion of oxygenated vs. deoxygenated haemoglobin, i.e. the BOLD effect, coupled with increased neural activity. In a task‐based fMRI study, researchers can infer that a mental function is associated with a specific brain region by identifying changes in the BOLD signal in the brain region. Therefore, the discovery of the BOLD effect is essential for brain mapping, i.e. to map the location of brain activation associated with functions (Jenkinson and Chappell 2018).
1.2.5.2 Perfusion MRI – Arterial Spinning Labelling
A major limitation of the BOLD fMRI (also see Section 2.1) is that the BOLD signal should be interpreted in a relative sense, as the difference of brain signals between different conditions. The value from fMRI data per se cannot be directly referred to the actual neural activity. Some factors other than neural activity, e.g. CBF or vessel volume, may influence the BOLD signal. Perfusion MRI, in contrast, assesses the delivery of cerebral blood and provides a quantitative measure that can be linked to the actual state of blood perfusion by the unit ml/100 g/min for the volume of blood passing 100 g of tissue within one minute (Jenkinson and Chappell 2018). The basic concept of perfusion MRI is to label part of the blood flow and detect the labelled marker after a fixed time delay. Then, the change of the labelled content against time can be quantified. In arterial spinning labelling (ASL), water molecules are used as an intrinsic marker. In ASL‐MRI, labelling is achieved by altering the magnetic properties of the hydrogen nuclei (i.e. their spinning behaviour) using different radiofrequency. Because changes in CBF can be a critical characteristic of neurological disorders, perfusion MRI has become an important tool for diagnosing neurodegenerative disorders, tumours and migraines (Telischak et al. 2015).
1.2.6 General Considerations of the Limitations of Neuroimaging Methods
Though most of the neuroimaging methods have been developed for decades, their application has some limitations. One of the most critical considerations for all imaging methods is the spatial and temporal resolution of imaging. For both structural and functional neuroimaging methods, a poor spatial resolution renders it hard to localize the precise position of a specific brain region. The problem of low spatial resolution is significant in PET, which investigates the brain by a voxel sized between 5 and 10 mm3 (Gazzaniga et al. 2019). Therefore, it provides spatial information at the scale of gross anatomy. However, by using different radioactive neurochemical agents, PET can detect the part of the brain which specifically engages with the agents. The problem of lower spatial resolution is also significant in magnetic resonance spectroscopy (MRS). Because MRS signals are relatively weak, to increase the signals obtained in a voxel, a larger voxel will be required for an MRS scan (Gazzaniga et al. 2019). Due to the limitation of spatial resolution, neuroimaging methods provide less information about brain features at the cellular level.
