domeet webmaster

Some studies have also shown that extending TOT will cause the energy to change from low frequency to high frequency(beta band: 13~30 Hz), and The activity in the β band of EEG is closely related to the cognitive process, and this energy transfer may reflect the compensatory function of the brain in order to maintain certain executive ability when vigilance is reduced. In ERP, when a person is in a state of mental fatigue, the magnitude of ERP components related to error monitoring and suppression is also significantly reduced, reflectingIn the fatigue state, the ability of task misrecognition is reduced.

In a recent article, Borghini et al. reviewed relevant studies on the changes in neurophysiological signals of pilots and car drivers during mental load and mental fatigue, and found that more and more relevant studies revealed that EEG The increase in power of δ, θ and α frequency bands is used as a representative feature of the transition from mental load to mental fatigue; this review paper The result then reflects from a new perspective the characteristics of neuro-electrophysiological activities when workers show obvious TOT effects in real life.

(2) Functional Magnetic Resonance Imaging

Magnetic resonance imaging technology can non-invasively measure brain activity by measuring the changes in hemodynamics caused by neuronal activity. Compared with EEG imaging technology, it has higher spatial resolution, and has obvious advantages in locating brain regions affected by mental fatigue. Lim et al. used arterial spin labeling (ASL) perfusion functional magnetic resonance imaging during the 20-minute psychomotor alert task.(fMRI) detected time-related changes in brain function and found that after the task was completed, the cerebral blood flow in the frontal lobe, cingulate gyrus and parietal lobe regions was (CBF) is significantly reduced. In addition, the changes in the CBF of the prefrontal network before and after the test are linearly related to the decline in executive ability.

Based on the results of this study, Gui et al. used a similar experimental design to study the blood oxygen level dependence during the 20-minute psychomotor alert task(BOLD) ‘s low frequency fluctuation (ALFF) amplitude, it turns out that it is the same as the pre-task (Alert state) measured value compared to the participants after the task (Under fatigue) ALFF in the default mode network is significantly reduced, and ALFF in the thalamus is significantly increased. More interestingly, the study also found that in the resting state, the ALFF of the posterior cingulate cortex and the medial prefrontal cortex before the task can be used to predict the degree of the subject’s subsequent decline in executive ability, that is, these two The higher the initial ALFF of the brain area, the more stable the performance that can be expected in the entire 20-min task.

Recently, Nakagawa et al. used a series of visual and auditory attention distraction tasks with different attention loads to study the effect of mental load on the adjustment of TOT effect. The results show that, similar to previous fMRI studies, the (including frontal, temporal, occipital, and parietal cortex) view of the cerebral cortex is similar to the previous fMRI study. Span> A decrease in brain activity caused by fatigue was observed. At the same time, the cerebellum and midbrain also showed a significant decrease in activity related to fatigue. In addition, significant interaction (that is, the activity of the midbrain area decreases more under high load conditions) was found in the midbrain region, which may reflect In addition to the inhibition of the negative feedback system, the system usually triggers restorative rest to maintain homeostasis.

(3) Functional near-infrared spectroscopy technology

Functional near-infrared spectroscopy technology(fNIRS)Using the main components of blood to pair 600~900 nm near-infrared light has good scattering properties, so as to obtain the changes of oxyhemoglobin and deoxyhemoglobin during brain activity. Similar to functional magnetic resonance imaging technology, it is also a non-invasive brain functional imaging technology that reflects neuronal activity by measuring changes in hemodynamics.

In a recent study, Jiao et al. used functional near-infrared spectroscopy (fNIRS) to study the relationship between the hemodynamic response of the prefrontal cortex and 4 hn-back(such as n = 2) the relationship between the degree of mental fatigue under the task of working memory. The results show that the information entropy of hemodynamic response is related to task performance and subjectiveThe self-assessment index is significantly correlated, indicating that hemodynamic response can be used as a neurobiological marker for fatigue classification.

De Joux et al. used fNIRS technology to find that the oxygenation of the right hemisphere of the brain was enhanced with the increase of TOT. At the same time, the oxygenation of the left hemisphere of the brain was significantly increased in the local task, which indicated that it was accompanied by the TOT effect under local rather than overall conditions. Increased utilization of bilateral brain resources.

Derosière et al. studied the neuroadaptation of task-related motor brain structure to TOT effect. They used single-pulse transcranial magnetic stimulation to measure the time course of exercise-related brain activity changes and corticospinal excitability. In addition, in a simple continuous attention response time task, they measured hemodynamic changes in exercise-related brain regions through fNIRS . It was found that after the TOT effect appeared, the oxygenation of the prefrontal and right parietal regions increased significantly, which is consistent with the findings in the literature. In addition, the study also mentioned the significant enhancement of corticospinal excitability and primary motor area activity, showing a way to adapt to TOT-related attention deficit in the form of changes in motor activity.

Recently, Chuang et al. combined EEG and fNIRS to study the hemodynamic characteristics in the process of driving fatigue. They found that the EEG α band inhibition in the occipital cortex increased, and the frontal lobe and primary movement Oxygenation in the area, parieto-occipital lobe and auxiliary exercise area is enhanced. These findings are roughly the same as the previous observations of fNIRS.

(4) Positron emission computed tomography technology

Positron emission computed tomography technology (PET) is an imaging technology that reflects molecular metabolism. The main principle of action in brain imaging is to mark short-lived radioisotopes with substances (such as glucose, etc.) which are necessary for the metabolism of brain neurons , Inject it into the human body, and realize the detection of brain activity by non-invasive imaging of the accumulation of the substance in the brain during the metabolism.

Paus et al. used PET to monitor the CBF area during the continuous 60-minute auditory alert task. As a result, it was found that CBF, which is a function of TOT, was significantly decreased in most areas of the cortex, including the thalamus, frontal lobe, parietal lobe, and temporal lobe of the right hemisphere.

Coull et al. used different experimental paradigms

In 1736, Swiss mathematician Euler proposed the famous “Seven Bridges of Königsberg” problem, introducing graph theory into mathematics for the first time, making it a new branch of the ancient science of mathematics. Graph theory is a mathematical analysis method used to quantitatively evaluate the network topology. In 1998, Watts and Strogatz used graph theory analysis to study the neural network structure of Caenorhabditis elegans, and found that the network structure is organized in a specific form to maintain a balance between local and global information transmission efficiency. They define this type of network architecture as a “small world” network. Once this influential article was published, it made the research on the structure and function of various complex systems a hot topic in research, including but not limited to neuroscience, social science, physics, biology, and computer science. Although there are huge differences in the microscopic details of each system component or their interaction mechanisms, the macroscopic behaviors of many complex systems are still very similar.

(1) Establishment of brain network

Brain network consists of a set of nodes and the connecting edges between them. The nodes of the brain network at the macro scale usually represent brain regions or sensors, such as the region of interest in fMRI (ROI), EEG or magnetoencephalogram< Span class="text-remarks" label="note">(MEG) electrodes or brain atlas solutions based on physiological structure. Using sulcus/gyri/cranial nucleus information in physiological structure, advanced brain segmentation framework based on connectivity, functional connection clustering and machine learning classification model based on multimodal images, many research teams have proposed Different brain atlas (node ​​definition) schemes. However, in a recent study, Arslan et al. systematically compared brain connection studies based on different brain atlases and found that there is still a lack of an optimal universal brain atlas scheme for brain network research. In view of the fact that different node definition methods may result in different properties of the brain network, it is recommended that different brain mapsThe reproducibility of the results of the spectrum study is evaluated in order to make a more comprehensive disclosure of the researched problems.

Compared with relatively simple node definitions, the nature and construction methods of edges or connections are more complicated. For example, edges can come from different but related connection forms: structural connection, functional connection or causal connection. The structural connections correspond to the white matter fiber bundles between different areas of the brain, which are usually observed by diffusion tensor imaging (DTI). Functional connection corresponds to the time dependence of distributed activities in the brain regions that are far apart in space, or manifests as the synchronization between distributed activities, which is usually estimated based on fMRI data. According to the different methods used in the analysis, functional connectivity can reflect the linear or nonlinear interactions between brain areas, as well as the interactions on different time scales. Causal connection reflects the direct or indirect influence of one area on another area, and has been widely used in EEG/MEG signal analysis. The relevant network connection construction process is shown in Figure 1.

Figure 1 The process of constructing brain structural connections, functional connections, and causal connections at a macro scale. SOG: upper occiput; SFGmed: upper frontal gyrus, medial part; L and R represent left and right respectively

In addition to the above three connection types, edges or connections can also be based on weight (weighting and binarization) and directionality (directed and undirected) to classify. The weight of the edge in the structural network can represent the size or density of white matter fiber bundles, and the weight of the edge in the functional and causal network can be expressed as the correlation between the activities of different brain regions. The binary network only contains edges that indicate whether the connection exists or not(0 or 1), usually can be realized by using different thresholds on the weighted network to achieve binarization, such as greater than the set threshold, it means that there are connected edges , And vice versa means that the connecting edge does not exist. At the same time, connecting edges can also be distinguished by the presence or absence of directional information. Although biologically-based structural connections can be represented by directed networks, current neuroimaging methods cannot directly detect the causal directionality of connections. On the other hand, based on the EEG/MEG record with high time resolution, the causal network can be constructed more simply by using different measurement methods such as Granger causality.

The constructed network is represented by its connection matrix (or adjacency matrix), where the rows and columns represent nodes, and the matrix items represent connected edges. The connectivity matrix is ​​used as input for quantitative analysis of graph theory.

(2) Graph theory analysis

Graph theory analysis aims to provide various quantitative measurement methods for evaluating the topology of the network(such as the spatial organization structure of nodes and edges). This section briefly introduces some network analysis parameters (table 1) often used in fatigue connection research. Regarding graph theory parameters and more detailed mathematical formulas, you can refer to the relevant review, which will not be repeated in this article. In addition, if you are interested in the practice of graph analysis algorithms, you can apply the relevant open source software toolbox. It should be noted that in the specific operation process, network analysis parameters should take into account the characteristics of the network(weighted network/binarized network, causal network/non-causal network ), using corresponding mathematical formulas based on specific network characteristics.

Table 1 Network parameter definition and measurement information in graph theory analysis

Four. Research on Connectivity of Mental Fatigue

In April 2018, we searched the Web-of-Science database with keywords “mental fatigue/TOT” and “brain connection” and found 99 relevant research papers. Although it is generally accepted that chronic fatigue syndrome changes the structural connections of the brain, it is still unclear whether the mental fatigue caused by TOT in the short-term cognitive process will cause changes in the brain structure. Therefore, this article mainly focuses on the functional connectivity changes related to mental fatigue in previous studies. By further excluding research papers that used samples of unhealthy people or used sleep deprivation experiments as a fatigue-induced paradigm, we finally got 29 research results, which constitute the main content of this section. Table 2 lists representative articles about mental fatigue and brain function connection.

(1) EEG connectivity research results

Ten Caat et al. proposed a data-driven functional unit (FU) method based on high-density EEG. The signal EEG electrode can better reflect its spatial connection group. In this paper, a visual-alertness experimental design (such as long-term switching tasks) is used to induce mental fatigue. The FU method has found that the fatigue effect mainly affects the low-frequency δ segment of the EEG signal. text-remarks” label=”Remarks”>(1~3 Hz). In addition, the results showed that the largest FU was located in the front and back of the brain in the non-fatigue group, but only in the back of the fatigue group.

However, in the next study, they found that the power and coherence of EEG increased significantly in multiple frequency bands, indicating that mental fatigue has a wide-ranging effect on the power and coherence of EEG. The significant difference in the results of the two studies may be caused by the different sample sizes. The literature has a smaller sample size(5), compared to The 26 samples in the literature may have large individual differences.

Clayton and others recently published anA review of the cortical oscillations in Yizhong, which emphasizes the importance of the θ, α and γ frequency bands in the EEG signal in the continuous attention task Sex. Recent network research shows that with the extension of TOT, long-distance functional connection plays a vital role in maintaining the best performance level. For example, we use the experimental design in the literature (20 min psychomotor alert task) to measure TOT-related changes with brain function connections. The results showed that the characteristic path length of the low α band EEG functional network significantly increased, and the increase was significantly correlated with the degree of decline in behavioral performance.

More interestingly, we found that the brain function connection (Figure 2) in the left and right hemispheres increases with the increase in fatigue. Presenting an asymmetrical change, this finding further validates previous neuroimaging observations. Specifically, the significant drop in the connectivity of the left frontal and parietal lobe is related to sustained attention. In addition, fatigue studies using different visual attention tasks and simulated driving paradigms also reported similar de-integrated network topologies.

Figure 2 The alert state (a) and fatigue state (b) corresponding to the first 5 minutes and the last 5 minutes of the 20-minute PVT experiment EEG Causal Function Connection Image Example (Reproduced from Ref. with permission of Elsevier Inc., © 2017)

In experiments in which cognitive tasks induce mental fatigue, subjects usually need to perform single or combined tasks for a long time. Common types of tasks include alertness or continuous attention tasks, visual attention tasks, and working memory tasks. Compared with this type of experiment,For tasks or simulated operation tasks, such as driving a vehicle, the process is bored by engaging in simple tasks with low load for a long time, which leads to a decrease in the subject’s perception ability and mental alertness.

In order to further study the neural mechanism of mental fatigue, we systematically compared by psychomotor alert tasks(PVT)Similarities and differences between paradigm and simulated driving task-induced mental fatigue in behavioral and neuro-electrophysiology. In terms of behavioral performance, we have found obvious psychological fatigue effects under the two fatigue-inducing paradigms, and then verify the effectiveness of these two commonly used paradigms in inducing mental fatigue. In terms of neuro-electrophysiology, we found that there are different reorganization phenomena in the EEG functional network between PVT and simulated driving: the characteristic path length and clustering coefficient of the network are increased under simulated driving conditions; while in the PVT paradigm, only the characteristic path length is increased. This discovery further reveals the complex neural mechanism of mental fatigue, and at the same time points out the regulation effect of mental load as a key factor on mental fatigue.

According to the resource theory of mental fatigue, mental load can be defined as the level of mental ability actually required to perform cognitive tasks in limited cognitive resources. Cognitive psychology research has shown that there is an “inverted U-shaped” relationship between mental load and task performance (that is, optimal task performance requires an appropriate mental load. High or too low will cause a decline in task performance), this phenomenon is in line with the underload theory and resource theory of mental fatigue. According to a recent review by Borghini et al., there is an obvious and measurable correlation between mental load and mental fatigue, which can be assessed by detecting continuous and consistent changes in a series of physiological signals. The detection of mental load has extremely important significance in practical applications. For example, it can maintain the optimal task performance by adjusting the mental load, while effectively avoiding the generation of mental fatigue.

More interestingly, several recent studies have shown that the mental load effect not only affects personal status, but is also related to mental fatigue in cognitive tasks of teamwork. Based on the EEG connectivity of mental load, we introduce an analysis framework that uses cross-frequency phase synchronization to evaluate cognitive mental load. Recently, we have proved that cross-task mental load can be achieved by extracting important EEG connection feature subsets in a single task. The feasibility of classification, the research results laid a certain foundation for the application of psychological load assessment in actual production and life.

(4) Connectivity results of fatigue classification

In addition to the above-mentioned graph theory research on mental fatigue, researchers also tried to use functional connection as a feature for fatigue detection and classification, and obtained satisfactory classification accuracy. In previous studies, we used a multivariate model analysis combining function connection (MVPA) and support vector machine (SVM) realizes the automatic classification of mental fatigue. The research divides the 20-minute PVT task into 4 stages to construct a functional connection matrix, where the first 5 minutes represents the alert state, and the last 5 minutes represents the fatigue state, and a good classification result is finally obtained.(81.5% accuracy, 77.8% sensitivity and 85.2% specificity). In addition, we found that most of the functional connections with high discrimination are significantly reduced under fatigue, which further verifies the de-integration of the network structure in previous EEG studies.

Recently, we have improved the classification model based on functional connection, using the most discriminative feature subset of the sequence floating forward selection method, and the input uses the radial basis function(RBF) SVM classifier and sequence minimum optimization learning method for analysis. The results show that (30 min PVT and 60 min simulated driving) have achieved a high classification accuracy rate under two different fatigue states(30 min PVT and 60 min simulated driving) span class=”text-remarks” label=”remarks”>(>90%). In addition, we also found significantly different functional connection characteristics, indicating that these two conditions may have different fatigue-related neural mechanisms. (one of the experiments included Intermission) uses the graph theory analysis method of functional connection (Figure 4). It was found that the experimental group without intermediate rest had lower efficiency of the resting brain network after the end of the task, which is consistent with the previous functional connection results on fatigue. Specific to the brain regions affected by fatigue, we found that the subcortical brain regions are more susceptible to mental fatigue. In addition, compared with the experimental group that did not rest, the experimental group that took a break did not observe a significant improvement in task performance. We have partially explained this finding from the perspective of rest regulation, that is, regulatory factors such as the length of rest, the nature of rest, and the introduction time of rest may all affect the degree of fatigue recovery.

For example, Lim and Kwok conducted an interesting study on the effect of rest time on TOT, and found that there is a clear link between rest time and improved performance. In addition, Ross et al. tested the effect of 1 min of rest introduction on fatigue recovery, and found that the recovery of TOT is only effective when rest is in the early stage of the task. Therefore, we can study more effective ways of fatigue recovery from different rest control perspectives. Although no significant behavioral improvement was found in the experiment of introducing short breaks in the middle of the task, the local efficiency of the brain network showed obvious interaction. This was mainly due to the significant decrease in the local efficiency after the task was completed under the condition of no rest. The partial efficiency of rest is maintained. In addition, the node efficiency of the left frontal gyrus and the right parietal area is also improved due to the short break introduced midway. The results of this analysis are roughly the same as those in the literature. The better recovery effect after a long period of rest is related to the increase in the activities of the putamen and the left middle frontal gyrus. In view of the lack of research on the neural mechanism of fatigue recovery, we can use a comprehensive experimental design and consider various rest regulation factors to further explore to verify our research results.

Figure 4 In this study, an in-group design was used to study the effect of introducing a short rest state during the execution of continuous tasks on mental fatigue recovery. The figure shows the results of the analysis of the efficiency of the brain area nodes in the resting brain function network before and after the task by repeated measurement multivariate analysis (block effect comparison before and after the task, phase effect comparison whether short breaks are introduced). (A) Significant block-effect brain area; (b) Significant stage-effect brain area; (c) Examples of functional brain network local characteristics of interaction effects; posterior analysis with significant interaction areas is shown in (d). INS: Insula; SFGdor: superior frontal gyrus, dorsal part; PUT: putamen; THA: thalamus; CAU: caudate nucleus; HIP: hippocampus; PAL: globus pallidus; IOG: inferior occipital gyrus; ORBinf: subfrontal Gyrus, orbital part; FFG: fusiform gyrus; PHG: parahippocampal gyrus; SFGmed: superior frontal gyrus, medial part; PCL: central parietal lobe; TPOmid: temporal pole, middle part (Reproduced from Ref. with permission of Elsevier Inc. , © 2018)

V. Discussion and Prospects

Through the summary of the research of brain function connection, it is found that under fatigue, the global integration generally shows a downward trend(Higher path length and lower global efficiency), Local specificity shows an upward trend(Clustering coefficient and local efficiency) (Table 2). The global neuron workspace theory believes that a globally integrated network is needed to support higher task requirements. According to the view of resource theory, the subject’s ability to maintain concentration in alert tasks depends on the amount of available psychological resources, and the repeated use and consumption of limited cognitive resources may in turn lead to a decrease in the integration and specificity of the network . We have observed in a recent fatigue recovery study that inserting a short period of time in the continuous alert taskDuring rest, the brain network reorganization has a recovery effect, which further supports the resource theory, indicating that the rest phase in continuous tasks may release the continuous demand for cognitive and neural resources, thereby alleviating or even reversing the fatigue observed The neural effect of .

Table 2 Main findings of research related to TOT affecting brain function connection

AAL: automated anatomical labeling atlas; HOA: Harvard–Oxford atlas; Craddock: Craddock’s functional atlas. PVT: psychomotor vigilance test; PDC: partial directed coherence; GPDC: generalized PDC; PCC: posterior cingulate cortex; MePFC: medial prefrontal cortex.

a Abbreviations of network metrics: FC: functional connectivity; D: physical distance; Str: connectivity strength; γ: normalized clustering coefficient; λ: normalized characteristic path length. The superscript arrow indicates the development trend of the network metric due to mental fatigue, while, NS indicates non-significant.

Multiple evidences show that the default mode network, the prominent network, and the thalamus-cortical circuit are related to mental fatigue. According to the literature, the mind is automatically restricted by the default mode network and the forehead attention network, while the prominent network is responsible for modulating the default mode network and the attention network. Because the default mode network is dominant in the resting state, it is usually considered to be a negative task network; the network with increased brain activity in the corresponding task state is usually considered to be a task positive network. Therefore, people associate the enhancement of neural activity of the default mode network with the decrease of attention to the external environment (“Thinking roaming” state).

Gui et al. proposed that when performing high-demand tasks, people with a highly active resting state default mode network may have more flexibility and more brain resources to redistribute the network from the negative to the positive task. Therefore, it is more resistant to the TOT effect. According to the resource-control model of continuous attention, the increased TOT will consume the execution resources of the highlight network, which is used to inhibit the activation of the default mode network and promote the task forward network to effectively perform the first task. In addition, the TOT effect not only affects the activation within the network, but also affects the connections between the networks. Specifically, there is a greater negative correlation between the default mode network and the attention network due to the TOT effect. In a recent study, we found that the brain activity affected by the TOT effect is mainly distributed in the subcortical area. This finding further supports the idea that the striatum-thalamic-cortex circuit may be the source of central fatigue. This phenomenon occurs in more extreme cases (such as sleep deprivation) and various diseases (such as chronic fatigue syndrome, multiple sclerosis) is especially obvious.

Most graph theory analysis indicators depend on the number of network nodes and connection edges. Therefore, network comparisons are meaningful only if the number of nodes matches the number of connections. Table 2 lists a variety of node definition methods from the signal channel of EEG to the brain atlas in fMRI research. However, the universal optimal brain atlas for brain connection group research has not yet been discovered, and different node definition methods may lead to different brain network properties. Therefore, it is necessary to further study a variety of neuroimaging technologies and brain networks defined by nodes to obtain a more comprehensive understanding. In addition, you can also explore new methods (such as meta-connectom analysis). Through comprehensive statistics of many studies, you may get more accurate and objective result. With the increasing number of fatigue-related neuroimaging studies, our understanding of the neural mechanisms of mental fatigue will surely make greater progress in the near future.

VI. Conclusion

This article briefly introduces the latest achievements of the brain connection group, and at the same time shows that it is of great value in revealing the neural mechanism of mental fatigue. It is hoped that this article can provide help to researchers who are interested in brain connectome technology in fatigue research. Although it is still difficult to fully reveal the mechanism of mental fatigue, it will bring great benefits to the prevention of the adverse consequences of fatigue. In the rapidly developing field of neuro-ergonomics, constantlyThe emergence of new methods can help us better reveal the neural mechanism of mental fatigue. We believe that the brain connection group can not only clarify the underlying neural mechanism of fatigue, but also provide quantitative analysis indicators to realize the automatic detection of mental fatigue.

Note: The content of this article has been slightly adjusted, you can view it if necessaryOriginal .

Adapted from the original text:

Peng Qi, Hua Ru, Lingyun Gao, Xiaobing Zhang, Tianshu Zhou, Yu Tian, ​​Nitish Thakor, Anastasios Bezerianos, Jinsong Li, Yu Sun.Neural Mechanisms of Mental Fatigue Revisited: New Insights from the Brain Connectome[J].Engineering,2019,5(2):276-286.

This article is from WeChat official account:Journal of Chinese Academy of Engineering (ID: CAE-Engineering)< / span> , author: “Engineering”