| 1.1 | The association between the brain and the stomatognathic system. The traditional perspective highlights the brain as a ‘systemic factor’ associated with oral health, just like the factors related to other body systems. The functional perspective highlights that the brain and mental functions guided by the brain play an essential role in stomatognathic functions. |
| 1.2 | A general view of the neural circuitries of the brain mechanisms of orofacial functions. The circuitries between the central and peripheral sites (i.e. pathways labelled in blue and red) are investigated primarily via animal models. Notably, the circuitries within the brain (i.e. the intracortical pathways labelled in black) have not been fully elucidated. Source: Avivi-Arber and Sessle (2018).Reproduced with permission of John Wiley and Sons. |
| 1.3 | Theoretical frameworks of the association between the brain, oral functions and behaviour. (a) The oral-to-behaviour (OB) framework, (b) the oral-brain-behaviour (OBB) framework and (c) the brain–stomatognathic axis (BSA). |
| 2.1 | The general concept of the blood-oxygen-level-dependent (BOLD) mechanism. (a) Transportation of oxygenated haemoglobin during a resting condition, when neural activity is low. (b) Transportation of oxygenated haemoglobin when neural activity increases. The neurons demand more energy by consuming oxygen provided by oxygenated haemoglobin. Via a complex haemodynamic process (e.g. an increasing rate and volume of cerebral flow), the amount of oxygenated haemoglobin increases (relatively to the amount of deoxygenated haemoglobin), leading to an over-supply or compensation of the oxygen demand from neurons. |
| 2.2 | Examples of functional magnetic resonance imaging (fMRI) investigation of chewing movement. (a) The first-level analysis. In a chewing experiment, the task conditions (i.e. when subjects are chewing) are contrasted to the baseline conditions (i.e. when subjects are resting). Brain activation reflects the difference in blood-oxygen-level-dependent (BOLD) signals in the task vs. the baseline condition. The first-level analysis focuses on the pattern of brain activation at the individual subject. (b) The second-level analysis. The second-level analysis focuses on the association between brain activation and individual variability. The association can be explored by investigating the correlation between brain activation and individual performance or comparing brain activation between different clinical groups. |
| 2.3 | Methodological considerations of a functional magnetic resonance imaging study. (a) Subjects may show great inter-individual variability in their general conditions, such as sex, age and general physical/psychological conditions. (b) Subjects may show great inter-individual variability in their personal trait and performance (e.g. pain ratings) related to an experimental task. (c) Subjects differ in brain morphology. When individual brains are compared, the individual images are normalized to a template image, using linear transformation (i.e. translation, rotation, resizing and shearing) and nonlinear transformation approaches. |
| 2.4 | Statistical analysis at the individual level and the group level. (a) The analysis at the individual level focuses on the association between task progression and the blood-oxygen-level-dependent (BOLD) time series, as shown in the left panel. For each voxel, a strong association indicates that the BOLD signals of the voxel can be predicted by the task condition, in contrast to the baseline condition (e.g. Voxel A), as shown in the right panel. (b) The analysis at the group level focuses on the association between brain features (e.g. brain activation of grey matter volume) and group factors. The association may reflect the difference in brain features between patient and control groups (the left panel) or the correlation between brain features and clinical factors (the right panel). (c) A typical image result consists of the statistical values (e.g. the t-score) from multiple voxels (represented as the grid), which are visualized by a colour scale, as shown in the left panel. The result can be thresholded according to intensity (i.e. the t-score). For example, only the voxels with a t-score >6 are preserved after thresholding, as shown in the middle panel. The result can be thresholded according to the size of a cluster of voxels. For example, only the clusters with a size larger than 100 voxels will be preserved after thresholding, as shown in the right panel. |
| 2.5 | Experimental design of functional neuroimaging research. (a) Under the assumption of pure insertion, the difference of brain activation between two experimental conditions only reflects the mental function contrasted by the conditions (e.g. perception of pain intensity). However, the contrast may be associated with more than one mental function (e.g. perception of pain intensity and attention to noxious stimuli). (b) A factorial design helps to delineate the association between two mental functions. For example, the light grey area denotes the effect of increased pain on brain activation, and the dark grey area denotes the effect of increased attention. (c) A conjunction design focuses on the pattern of brain activation common to two experimental conditions (e.g. a clenching task and a chewing task). The activation may reflect the brain mechanisms of a mental function common to both task conditions. |
| 2.6 | Conceptual differences between functional specialization, functional segregation and functional integration. (a) Functional specialization highlights the association between a brain region and a specific mental function. For example, activation at the occipital lobe is considered mainly for the processing of visual perception. (b) Functional segregation highlights that a mental function is associated with multiple brain regions that are functionally connected within a module. For example, visual cognition is associated with the module consisting of the yellow regions, and motor control is associated with the module consisting of the blue regions. (c) Functional integration highlights the pattern of global communication between multiple brain regions. For example, individual variability in mental functions may be associated with the efficiency of how information is distributed in a network. |
| 2.7 | Analysis of resting-state functional connectivity. (a) The spontaneous blood-oxygen-level-dependent (BOLD) activity is acquired using resting-state functional magnetic resonance imaging. Subjects fix their eyesight on a crosshair without additional external stimuli. (b) Brain images are segmented into multiple regions according to a brain atlas. (c) To each brain region, the mean BOLD time series is extracted by averaging the time series from all the voxels within a region. (d) Association between the regional time series is quantified with correlation coefficients. (e) The correlation coefficient represents the strength of the connection between each pair of brain regions. (f) In the seed-based approach, a brain region of interest (i.e. the ‘seed’ region) is pre-selected. Functional connectivity is calculated between the seed region and all the other voxels to explore the brain regions that have a strong connection with the seed region. |
| 2.8 | Analysis of structural connectivity. (a) In deterministic tractography, each voxel is assigned with a single direction, which reflects the principal direction of diffusivity. A continuous streamline is formed by tracking the direction of voxels. (b) Probabilistic tractography assumes that there exists an uncertainty of the direction within each voxel. In the right panel, the probabilistic distribution of the directions is estimated for each voxel. A higher probability of a ‘leftward’ direction can be identified. In the right panel, the intensity of a voxel represents the frequency that a streamline passes that voxel. For example, more streamlines pass the yellow voxel (here four out of seven streamlines) compared to the red voxel (here one out of seven streamlines). (c) Structural covariance quantifies the strength of association of a brain feature between different brain regions across subjects. For example, the cortical thickness of six brain regions is assessed for eight subjects. The right panel reveals the association between regions 2 and 6, as quantified by the correlation coefficient of the cortical thickness between the regions. |
| 2.9 | Graph-based analysis of brain connectivity. (a) The pattern of functional and structural connections between brain regions can be translated from the ‘brain space’ to the ‘network space’ with applications of graph theory. In a graph, the nodes represent brain regions and the links represent the functional and structural connectivity between regions, which can be quantified by the correlation coefficient between blood-oxygen-level-dependent (BOLD) time series and the streamlines identified by tractography, respectively. (b) In the network analysis, the global metrics quantify the degree of integration of a network. For example, characteristic path length can be calculated by finding the shortest path length between a pair of nodes, such as the path A–B–D (but not A–B–E–D) between the nodes A and D. (c) The local metrics quantify the degree of segregation of a network. For example, the clustering coefficient is used to quantify the fraction of the triangular architecture in the whole network (e.g. A–B–C and E–G–H), which represents a pattern of clustered nodes. Notably, a small-world network offers a balance between the efficiency of global and local communication. A highly regular network (i.e. the middle-right panel) and a highly random network (i.e. the middle-left panel) may suffer from a lower global and local efficiency, respectively. |
| 3.1 | Methods of the assessment of oral cutting ability. (a) The sieving method quantifies the proportion of the chewed food (e.g. peanuts) with different particle sizes, using multiple sieves with different pore sizes (e.g. from the diameter of 355–3500 µm). The total weight of food particles that pass through a sieve is plotted against the pore size of the sieve. A smaller median particle size (e.g. the grey curve) represents better performance in cutting. Source: Chia-Shu Lin. (b) A test gummy jelly is customized with a standardized size and shape. The chewed fragments are collected and photographed. Colour and morphological features (e.g. the area and perimeter) of each fragment, which reflect individual cutting ability, are assessed by analyzing the image. Source: Salazar et al. (2020). Reproduced with permission of Elsevier. |
| 3.2 | Methods of the assessment of oral mixing ability. (a) In the two-colour chewing gum test, the degree of mixing food can be assessed by the colour hue of chewing gum with different colours. For example, if a piece of red and a piece of yellow gums are well mixed, the resulting bolus would in orange homogenously. The hue of the bolus can be quantified by imaging analysis. A smaller standard deviation of hue represents a greater homogeneity of colour mixing, i.e. a better mixing ability. (b) The degree of mixing is assessed according to the pattern of spatial clusters. A piece of juice chew with red and white portions was chewed by a subject for 20 strokes and collected, as shown in the left panel. The degree of clustering is assessed based on the analysis of variogram, which reflects how fine the clusters of different colours are. A pattern with finer clusters (e.g. the case in the lower-right panel) reflects better mixing ability. Source: Lo et al. (2020). Reproduced with permission of John Wiley and Sons. |
| 3.3 | Experimental design of pain/somatosensory experience. (a) Blood-oxygen-level-dependent (BOLD) signals are recorded concurrently with discrete ratings of stimulation. In this design, noxious stimuli with high and low intensities are followed by a rating phase (‘?’), which requires subjects to rate the pain intensity they perceive for the stimuli. Brain activation associated with pain can be contrasted by the phases that subjects feel strong vs. mild pain, according to their ratings (i.e. the black bars). (b) BOLD signals are recorded concurrently with continuous ratings of spontaneous pain. Patients with chronic pain continuously rate their pain, which may increase spontaneously. (c) Brain features and ratings are recorded separately. In this design, the rating of pain or somatosensory experience is conducted outside a scanner. Association between the individual ratings (e.g. pain) and their brain features (e.g. grey matter volume of the insula), which are collected separately, can be investigated by correlational analysis. |
| 4.1 | An overview of brain regions associated with motor control. The figure only displays the relative position and size of the brain regions, not depicting the anatomical details. |
| 4.2 | Sensorimotor control of the basal ganglia and the cerebellum. (a) The basal ganglia consist of a direct and an indirect pathway of motor control. In both pathways, the striatum is activated by the cortex, which forms a loop of control with the thalamus (the grey arrow). In the direct pathway (the solid black arrow), the activation of the striatum inhibits the activity of the internal segment of the globus pallidus (GPi) and the substantia nigra (SNr), which further inhibits thalamic functions. Therefore, cortical activation is associated with an increased thalamic activity via the direct pathway. In the indirect pathway (the black dashed line), the activation of the striatum inhibits the activity of the external segment of the globus pallidus (GPe), which further inhibits the activity of the subthalamic nucleus (STN). Notably, the activity of the STN further activates GPi/SNr, which decreases thalamic activity. Therefore, cortical activation is associated with a decrease in thalamic activity via the indirect pathway. (b) The cerebellum plays a key role as an internal model of motor learning. A forward model predicts the sensory outcomes when motor commands are executed. It adjusts sensorimotor processing via feedback of the predicted sensory outcomes. An inverse model calculates the motor commands that would produce the sensory outcomes from desired actions. According to Wolpert et al. (2001), both models are crucial to motor control. |
| 4.3 | Experimental design for neuroimaging of the brain mechanisms of chewing. (a) The basic concept of study design. The study consists of multiple blocks of a chewing task and a baseline (no-chewing) block. (b) Variations of the study design. Different studies may differ in the number of blocks of tasks and the definition of the baseline block (e.g. resting or clenching the teeth). The variations lead to a different interpretation of imaging results. |
| 4.4 | Brain activation associated with chewing and clenching. Source: Lin (2018). Reproduced with permission of John Wiley and Sons. |
| 4.5 | Experimental design for neuroimaging of the brain mechanisms of swallowing. (a) Study design of the swallowing tasks, including the water swallowing task and the saliva swallowing task. (b) Examples of the study design for investigating brain mechanisms of swallowing. An overt swallowing task (with either water or saliva swallowing) is associated with the execution of swallowing movement. A covert swallowing task is associated with the motor planning of swallowing. |
| 4.6 | Brain activation associated with water (the upper panel) and saliva (the lower panel) swallowing. Source: Sörös et al. (2009). Reproduced with permission of John Wiley & Sons, Inc. |
| 5.1 | Processing of oral somatosensory information. Information from individual sensory modalities is transduced via peripheral receptors at the level of somatosensation and integrated to form a holistic perceptual experience (e.g. oral stereognosis) at the level of somatoperception. Information is further integrated, with knowledge and affective–motivational experience, to form a feeling of intraoral condition at the level of somatorepresentation. |
| 5.2 | Experimental methods of investigating oral mechanoreceptors. (a) Recording signals from periodontal mechanoreceptors. Source: Trulsson (2006). Reproduced with permission of John Wiley and Sons. (b) The food splitting task. Source: Grigoriadis et al. (2017) with permission of Springer Nature under the terms of the Creative Commons CC BY 4.0 License. |
| 5.3 | Brain activation associated with gustation at the insula. A consistent pattern of brain activation is identified in the insular cortex for studies focusing on the quality, intensity and affective value of taste stimuli, respectively. Source: Yeung et al. (2018). Reproduced with permission of Elsevier. |
| 5.4 | Basic concepts of perceptual processing. (a) Top-down processing highlights the neural processing of intrinsic (personal) factors, such as one’s goal planning, on the formation of perceptual experience. The bottom-up processing highlights the neural processing of extrinsic (environmental) factors, such as the physical features of stimuli, on the formation of perceptual experience. (b) In predictive coding, the sensory inputs that we receive from the real world are compared with our prediction of the sensory outcomes. A mismatch (i.e. ‘prediction error‘) occurs when our prediction does not fit the outcome we perceive. The prediction error is associated with attentional bias and learning. For example, we may pay more attention to an unexpected event compared to an expected one. |
| 5.5 | Experimental design of manipulating threat value associated with pain. (a) The presence of noxious stimuli is associated with visual cues, which predict low-intensity stimuli (i.e. the square) constantly or predict high-or low-intensity stimuli (i.e. the circle). The latter evokes higher anxiety related to pain due to the increased uncertainty (i.e. the stimulus intensity is less predictable). Moreover, the same low-intensity stimuli would be perceived as more painful in the high-uncertainty condition (i.e. the condition predicted by a circle). (b) The threat value of pain is associated with the severity of noxious stimuli. Subjects receive different instructions regarding the severity (e.g. may cause tissue damage or not) of noxious stimuli, which are delivered at different sites (grey and black). The detection threshold of pain, i.e. the intensity that subjects feel painful and non-painful for equal times, is determined. When subjects feel a stronger severity of the stimuli (i.e. feeling more threatened), they would report higher anxiety towards the stimuli. Moreover, the same stimuli (tuned at the detection threshold) would be perceived as painful more likely in the more threatening condition (i.e. the condition with more severity) compared to the condition they regard as less threatening. |
| 6.1 | Brain activation associated with pain processing. (a) The brain regions commonly reported in functional magnetic resonance imaging (fMRI) studies of noxious stimuli. The figure only displays the relative position and size of the brain regions, not depicting the anatomical details. Note that the insular cortex (bounded by a dashed line) is covered by the frontal, parietal and temporal operculum. (b) Metaanalytical findings of the brain activation of experimental orofacial pain in healthy subjects. Increased activation is consistently found in the posterior mid-cingulate cortex (pMCC), the PPC, the insula, the thalamus, the S1 and the S2. Decreased activation is consistently found in the primary motor cortex (M1) and the S1. Source: Ayoub et al. (2018). Reproduced with permission of Elsevier. |
| 6.2 | Functional networks associated with chronic pain. Source: Davis et al. (2017) with permission of Springer Nature under the terms of the Creative Commons CC BY 4.0 License. |
| 6.3 | Mechanisms of peripheral and central sensitization. (a) In the normal status, signals induced by noxious stimuli and non-noxious (e.g. tactile) stimuli are transduced via the pathway of nociceptive and tactile processing, respectively. (b) In peripheral sensitization, neurons would show increased responsiveness to noxious stimuli. For example, the inflammation at the peripheral site (i.e. the light red area) may reduce the threshold to evoke a response. The signals for subsequent nociceptive processing are amplified. Therefore, peripheral sensitization may be associated with hyperalgesia. (c) In secondary hyperalgesia (in contrast to primary hyperalgesia (b)), noxious stimulation to the area surrounding the site of injury or inflammation (i.e. the black arrow) elicits an amplified pain. The amplification is mediated by the central neurons (i.e. the red circle), which have been sensitized by constantly receiving nociceptive inputs from the primary lesion (i.e. the circled light grey area). (d) In allodynia, non-noxious tactile stimuli are conveyed by the Aβ fibre elicit pain. Note that at the central level, both the nociceptive and tactile pathways converge on central nociceptive neurons. The central neurons (the red square) have been sensitized by constantly receiving nociceptive inputs from the primary lesion. |
| 6.4 | Brain features associated with chronic orofacial pain. (a) Brain activation associated with chronic orofacial pain. Meta-analytical findings reveal a consistent pattern of higher brain activation in the left medial and posterior thalamus and lower brain activation in the left posterior insula in patients with chronic orofacial pain (COFP) compared to healthy controls. Source: Ayoub et al. (2018). Reproduced with permission of Elsevier. (b) Functional connectivity of patients with temporomandibular disorder (TMD)-related pain. The left panel reveals that TMD patients showed enhanced functional connectivity (FC) between the medial prefrontal cortex (mPFC) and the posterior cingulate cortex (PCC)/precuneus (PCu)/retrosplenial cortex (RSC) compared to healthy controls. The right panel reveals that functional connectivity between the mPFC and medial thalamus/PAG was positively correlated with pain rumination in TMD patients. Source (insets): Kucyi et al. (2014), p.3969–3975 with permission of the Society for Neuroscience under the terms of the Creative Commons Attribution 4.0 International License. |
| 7.1 | Conceptual links between oral health and cognitive functions. (a) The framework of the brain–stomatognathic axis (BSA). The framework highlights that the brain plays a key role in behavioural adaptation in feeding and oral care behaviour, which further relates to oral health. (b) The potential role of the brain in behavioural adaptation. Poor oral health may be attributed to increasing difficulty in conducting health-related behaviour (e.g. being unable to brush teeth), which is derived from neurodegenerative disorders. (c) The potential role of oral factors in brain pathologies. The ‘oral-on-brain effect’ may be associated with multiple factors, such as reduced sensory feedback from the loss of occlusal contact or increased periodontal inflammation/infection. Notably, when the brain is affected by poor oral health, it may be followed by worse adaptation of feeding and oral care behaviour, which furthers exacerbates one’s oral health. (d) The potential role of a common factor that affects both cognitive and oral functions. For example, aging is associated with structural and functional alterations of the cerebellum, which relates to not only oral sensorimotor functions but also cognition. The arrow filled with a slash pattern denotes the potential causal links of the framework. |
| 8.1 | The concept of neuroplasticity and functional adaptation. (a) With a ‘preprogrammed’ nervous system, our behaviours responding to environmental stimuli are determined by a fixed set of stimulus–response links. For example, a great danger will elicit a stronger emotional response compared to a mild danger (the upper panel). However, our nervous system is modifiable and can be tuned according to environmental changes. Long-term experience may sculpt the brain at the structural and functional levels, leading to different behaviours responding to environmental stimuli. For example, past experience may predispose the brain to be more sensitive to danger and make a stronger emotional response (the lower panel). (b) Functional adaptation is associated with the improvement of one’s functional performance under environmental challenges. For example, individuals learn to run faster (i.e. increased performance) when they are threatened by a greater danger (the upper panel). Compensation, in contrast, highlights the restoration of performance from a worse status back to a normal status. For example, individuals with a disability can run as quickly as normal individuals when facing danger by compensating their mobility with the help from tools and rehabilitative therapy (the lower panel). |
| 8.2 | Experimental design of neuroimaging research on brain plasticity. (a) A cross-sectional study reveals a significant difference in structural brain features between subjects with and without a professional skill (e.g. driving). Individual variations in brain features further relate to the duration of skill training (the left panel). Results from the cross-sectional design only suggest, but not confirm, the causal direction of a plastic effect. The difference in brain features (as observed by neuroimaging) may be attributed to prior experience of training (the middle panel). However, it is also possible that the individual differences in the brain features predispose their performance of a skill (the right panel). (b) To better elucidate the plastic effect of training, a longitudinal design is used to assess the performance and brain features at different stages of training. Importantly, one can assess whether the plastic effect, i.e. changes in brain features during the training session, can last for a period or vanish right after the termination of training. (c) The same longitudinal design also helps to elucidate individual differences in their performance. For example, the variation in brain features may account for the difference in learning speed. |
| 9.1 | Major aims of brain–stomatognathic integrative assessment (BSIA). (a) Prediction of long-term changes in oral functions. Individual differences in brain reserve and cognitive reserve may play a key role in their susceptibility to diseases. The BSIA, which includes the assessment of cognitive functions of older people, would help to predict the future condition of individual oral health. (b) Classification of patients with different risks of oral diseases. The BSIA helps to classify the patients for their risk of oral diseases and the prognosis of treatment, based on a full-scale assessment of general physical and mental status. |
| 9.2 | Key elements for implementing the brain–stomatognathic integrative assessment. (a) The multidisciplinary investigation can be undertaken in a hospital, where different departments are in charge of different assessments. Critically, the results of an assessment from one discipline (e.g. the score of cognitive tests from neurologists) are distributed for the use of other disciplines (e.g. dentistry). For example, the cognitive performance of older patients, as assessed by neurologists, is available for prosthodontists to evaluate if patients can adapt well to their new dentures. (b) In contrast to the hospital setting, a home-based assessment can be facilitated by the use of teledentistry approaches and digital technology. For example, in long-term care institutes or at home, patients can record their own oral status by photographing (via a smartphone). The image record may include a photo about their intraoral conditions (e.g. bleeding gum) or performance of oral functions (e.g. the food bolus after chewing). The images are sent to the cloud storage service for further analysis. A preliminary assessment is performed automatically by machine-learning-based algorithms, and critical problems (e.g. a poor oral mixing ability) are screened and forwarded to dental professionals for further evaluation. |
| 9.3 | An example of neuroimaging investigation combined with animal research. Using structural MRI, Avivi-Arber et al. (2017) compared the post-mortem brain volume between the mice that received molar extraction and those who received sham operation. Tooth extraction was associated with a widespread reduction in volumes of brain regions of sensorimotor and cognitive–affective processing. Source: Avivi-Arber et al. (2017) with permission of Frontiers Media S.A. under the terms of the Creative Commons Attribution 4.0 License. |
