Prolonged disorder of consciousness (pDOC) refers to the conditions characterized by prolonged loss of awareness for more than 28 days (1). pDOC is divided into vegetative state (VS) and minimal conscious state (MCS). VS is also known as unresponsive wakefulness syndrome (UWS), which refers to the presence of basic brainstem reflexes and sleep-waking cycles, without content of consciousness (2). MCS refers to discontinuous and fluctuating signs of consciousness in patients (3). In recent years, with the progress of medical technology, it has been found that some pDOC patients show signs of brain activity, indicating that their brain has hidden consciousness. Therefore, some new diagnostic classifications have been proposed, such as “Cognitive motor dissociation,” “Recessive cortical activity,” and “MCS*” (4–6). The “mescocircurt mode “proposed in recent years believes that the connections of the thalamus, frontal lobe, parietal, occipital, and temporal sensory cortex are the basic circuits of consciousness, and the damage of changing the circuit will lead to the disorders of consciousness (7). However, questions about where consciousness comes from and what factors are related to the level of consciousness are still being explored.

Research has confirmed that the brain is not inactive when there is no stimulation. On the contrary, the brain will be active in an organized way at rest, to prepare for the next possible stimulation processing (8). Functional magnetic resonance imaging (fMRI) is one of the tools to study brain function. fMRI relies on the difference of magnetization vector between oxyhemoglobin and deoxyhemoglobin to generate the signal, which indirectly displays the activity of brain tissue with blood oxygenation level dependent (BOLD) signal (9). fMRI can generate a statistical map of the whole brain connections of a specific region or network by associating the BOLD signal in the region of interest (ROI) with all other voxels in the brain (10). Moreover, fMRI can map brain network components closely related to consciousness, such as default mode network (DMN), salience network (SN), and executive control network (ECN), and serve as neuroimaging biomarkers for pDOC prognosis (11, 12). However, the neurons are constantly active, and the time resolution of fMRI cannot meet the requirements of monitoring the complete sequence of brain activity.

Electroencephalography (EEG) is a noninvasive examination. The principle of EEG is that the image obtained by amplifying and recording the spontaneous biological potential of the brain from the scalp is the spontaneous and rhythmic electrical activity of brain cell groups recorded by electrodes. For pDOC patients, resting-state EEG has become the main modality in electrophysiological evaluation in clinic work. One of the advantages of EEG is its high temporal resolution, which is more suitable for studying the time series of resting state. EEG can be used to assess the integrity of patients’ sleep-waking cycle (13), and studies have found that the presence of sleep spindles is associated with consciousness (14). pDOC patients do not have systemic changes in the sleep spindle and slow wave oscillations between day and night (15). Event-related potential (ERP) is a kind of special evoked potential, which refers to the bioelectric response that can be detected in the system and the corresponding parts of the brain, and has a time-locked relationship with the stimulus and a specific location phase when given a specific stimulus to the brain, or the brain processes the stimulus information. The ERP is based on EEG. At present, P300 and mismatch negativity (MMN) are the most widely used. P300 is a positive wave that appears about 300 ms after stimulation triggered by the Oddball paradigm. It can be found in healthy people, MCS, and Locked-in syndrome patients, but it is rare in UWS (16). P300 reflects the brain’s ability to process information. The amplitude of P300 is correlated with the Coma Recovery Scale-Revise (CRS-R) score, and P300 latency is prolonged in pDOC patients (17–19). MMN is to superimpose and average the ERP of standard stimuli and deviant stimuli respectively, and subtract the ERP of standard stimuli from the ERP of deviant stimuli to obtain the difference wave. The negative deflection of 100-250 ms after the stimulus difference is MMN (20). MMN can distinguish healthy patients from doc patients, and the amplitude of MMN is often lower in UWS (21, 22). EEG spectral power, functional connectivity, graph theory, and nonlinear measurement are also widely used in the study of pDOC (16). Among various brain connectivity measurement methods, EEG-based brain connectivity uniquely provides resolution on the millisecond scale. Similar to fMRI, EEG can be used to construct a multiscale brain network consisting of nodes (hubs) and connections between them, in which the topology can be quantified. Some studies have proved the use of complex network analysis for pDOC by extracting the EEG parameter of network topology in the resting state and can distinguish MCS from VS (23, 24). pDOC patients have reduced local and overall efficiency of the resting state network and fewer hubs in the alpha band (25). Quantitative electroencephalogram (QEEG) is based on the classical EEG that records the activities of brain neurons. It converts the original, complex, and changeable electrophysiological curve into quantitative, and orderly data, thereby improving the sensitivity to changes in brain function (26). At present, many EEG indicators with prognostic potential have been found, such as (δ + θ)/ (α + β) Ratio (DTABR), etc., can predict the prognosis of pDOC (27).

In 1987, Lehmann et al. showed that the alpha frequency band (8-12 Hz) of resting-state EEG signals can be resolved into a limited number of distinct quasi-stable states (28). These discrete states are referred to as “microstates (MS)” and are calculated to remain stable for 80–120 ms before rapidly transitioning to other MS (28). In most normal EEG microstate studies, the four microstate maps obtained by clustering have high similarity (29). Specifically, MS 1 shows an upper left and lower right direction, MS 2 shows an upper right and lower left direction, MS 3 shows a forward-backward direction, and MS 4 shows a midline frontaltopographies, the maximum values distributed in frontal and central parts. Even if more clustering graphs are selected, these four MS always dominate in different age ranges, conditions (such as sleep and hypnosis), and pathological states. MS analysis considers signals from all electrodes on the scalp to display functional status globally, rather than individual electrode signals. The rich time series of MS provides a new quantification of EEG signals. Many studies have shown that the characteristics of MS time series of EEG vary with different behavioral states (30). MS analysis is a method to study the neural characteristics of many cognitive processes, and also a method to study brain dynamic functions and link these dynamics with cognition and disease. MS dynamics are closely related to perceptual consciousness, vision, neuropsychiatric disorders, including schizophrenia, resting-state functional network, etc. (31). Lehmann et al. believe that various MS extracted from the brain’s electric field have different levels of consciousness and psychological support, and MS are considered “atoms of thought” (32). Current research indicates that MS is associated with various mental and neurological diseases. Schizophrenia is a chronic disease with unknown etiology, usually manifested as a syndrome with varying symptoms, involving various disorders such as sensory perception, thinking, emotion, and behavior, as well as uncoordinated mental activities. In 2016, Rieger et al. conducted a meta-analysis showing the medium-sized effects of MS 3 and MS 4. The Occurrence of MS 3 is more frequent in patients with schizophrenia, while MS 4 had a shorter Duration, and MS 2 also had a shorter Duration, although not significantly (33). MS 3 and MS 4 are considered early markers of the risk of schizophrenia development, and drug treatment can normalize the patterns of these two MS. Therefore, MS can not only predict the development of certain diseases but also monitor treatment outcomes. Ricci et al. found that after 3 months of treatment with levetiracetam for temporal lobe epilepsy, the direction of MS maps and related indicators in epileptic patients changed, and proposed that MS can be used as a biomarker for the efficacy of levetiracetam in treating epilepsy (34). Similarly, MS are considered markers of diseases such as stroke, severe depression, migraine, and Parkinson’s disease (35–38). It can also study the physiological state of the brain. A study shows that MS 3 and 4 play a dominant role during Non-rapid eye movement sleep, and explains why the dreaming brain will disconnect the dreaming process and keep sleeping (39).

There are significant differences in brain function between pDOC patients and healthy individuals. Current research has focused on P300, BOLD, and other studies (40), but there is relatively little research on the MS of pDOC. EEG has the advantages of relatively low price, bedside monitoring, and easy to carry out in basic hospitals. This paper uses the MS analysis based on EEG, combined with brain connectivity analysis, to explore the differences between pDOC and normal brain, and to explore appropriate biomarkers in clinical practice.

MS analysis is the analysis of brain activity from a time series. This method first identifies several microstates at the group level and then projects them onto individual participants, thereby studying the different combinations and arrangements of MS in each participant. The transition between MS can be explained as the sequential activation of different neural networks, while the time series of MS in resting-state EEG can allow us to visually observe the fast switching between the collective activities of neurons at rest. There are three main indicators of MS, Duration refers to the average length of time a specific MS remains stable when it appears, Coverage is the proportion of the microstate to the total recording time, and Occurrence is the average number of times per second that the MS dominates during the recording period (28). This study analyzed the MS of pDOC patients and found that the morphology of MS B is unique to pDOC and can be used as a topographic map to distinguish between healthy and pDOC groups.

Studies have shown that the four MS have a spatial correlation with fMRI. MS D is mainly related to BOLD signal changes in the bilateral superior temporal gyrus and is considered to be related to language processing (43). This study shows that the Occurrence and Coverage of MS D in the DG are significantly lower than those in the HG. Therefore, pDOC patients may have obstacles in understanding instructions and are unable to communicate effectively with the outside world. From the perspective of connectivity, compared to other MS, MS D has a more significant decrease in connectivity compared to the HG. Therefore, the impairment of language comprehension function in pDOC patients may be an important reason for communication barriers with the outside world. Recent studies by Stefan et al. have also shown that the MS from the lower left and upper right is very effective in predicting coma outcomes, especially in Duration and Occurrence in the delta frequency band (0-4 Hz) and theta frequency band (4-8 Hz) (43). MS C is believed to be related to visual function (44). MS B is related to the activities of the anterior cingulate cortex, inferior frontal gyrus, claustrum, and frontal insular cortex (44). This region is closely related to the “Salience Network (SN),” which plays a switching role between the central executive network and the DMN. SN is the “mediator” of the brain, which will continuously monitor the external world and carefully determine the response of other brain networks to new information and stimuli (45). Highlighting the structural and functional integrity of the network is necessary to regulate the activity of the DMN. The MS B of pDOC group can be used as the characteristic topographic map of pDOC, and the Duration, Coverage and Occurrence of microstate in this topographic map are increased than those in the healthy group, and the Coverage is also significantly increased. Therefore, MS B of the pDOC group is of great significance both in the morphology of the topographic map and in the statistics of various micro state indicators, that is, the prominent network of pDOC is significantly damaged. MS A is associated with BOLD signals in the frontal and parietal cortex and is believed to be related to attention (44). The combination of MS and fMRI also indicates that the brain network depicted by fMRI is the sequential activation of different network components in time, rather than the simultaneous activity of all network components. The reason for this change is the low temporal resolution of MRI (46), and MS based on high temporal resolution EEG precisely compensates for this deficiency. The different combinations of various MS in time series also indicate that the brain is active every moment. Recent studies have pointed out through training a random forests classifier that MS may be the electrophysiological basis for brain dynamic functional connectivity (47).

Functional connectivity measures the statistical dependence of physiological time series recorded in different regions of the brain. Since the calculation of functional connectivity highly depends on brain activity changes in time series, high temporal resolution techniques such as EEG (<1 ms) are the best tool to reflect neural dynamics. We found that the four MS were significantly different between pDOC and healthy groups, that is, in the 1-30 Hz EEG time series, the brain connectivity of pDOC patients was significantly reduced. These results are similar to those of Lechinger et al., and the level of connectivity is related to the severity of pDOC (48). A study on neurocognitive dysfunction and pDOC showed that in the DG, EEG connections in all frequency bands were reduced, and there were also significant differences in connections from other cortical regions to frontal lobe regions (49), which seemed to indicate the role of the frontal lobe in consciousness. In this study, the connections related to the frontal lobe, especially the connections from the frontal lobe to the limbic lobe, were significantly reduced. We also found that the directional connection between the hemispheres of pDOC is reduced, and the information communication seen in the cerebral hemisphere seems to be closely related to the level of consciousness, which needs further research to prove.

The study by Xu et al. revealed that patients exhibiting microstate B, which is also present in healthy individuals, tend to have a more favorable prognosis (50). The parametric characteristics of microstate B demonstrated a strong correlation with clinical scale scores, thereby establishing it as the most accurate indicator reflecting the level of consciousness. This finding suggests a close association between microstate D and the level of consciousness (50). In a study involving pediatric pDOC patients, the results demonstrated a significant correlation between microstate-related parameters and CRS-R scores, suggesting that microstates have the potential to predict consciousness recovery (51). According to Liuzzi et al., microstates may be necessary to sustain consciousness, but cannot be considered as the only actor in consciousness presence/recovery, they also contend that the emergence of consciousness is dependent on the existence of an anterior–posterior topography (Microstate B), a map closely linked to frontoparietal activity, aligning with our study (52).

There are also some shortcomings in this study. This study included a small data volume, especially for patients in the DG, so pDOC was not further classified based on CRS-R scores or etiology. Although there was a significant age difference between the two groups, no further age-based stratification was performed in this study, given that microstate topographies and related metrics exhibit only subtle variations among adults, except in children and elderly individuals. This may have introduced some bias into the results. In the analysis, we found that although the DG had three MS topographical maps similar to the HG, the underlying neurological mechanisms may not be the same. For example, pDOC patients often have pathological changes such as cortical atrophy and subcortical fiber bundle destruction, which are inevitably accompanied by neurological dysfunction and changes in brain networks. The brain connectivity analysis of this study also showed significant differences between the two groups. Therefore, exploring the neuro-electrophysiological mechanisms of the abnormal brain is the focus of our next research.

The MS B of pDOC patients is unique compared to the HG and may be a biomarker for distinguishing pDOC from healthy individuals. At the same time, the brain connectivity of pDOC patients in the four MS decreased significantly compared to the HG, especially the connectivity between brain regions and hemispheres. Finally, microstate analysis combined with connectivity analysis showed that Significant Network damage was the key to affect consciousness.