Introduction

The coronavirus disease 2019 (COVID-19) pandemic has had profound global public health effects after its initial outbreak in late 2019 [1]. In addition to the direct health impacts caused by the infection, associated preventive measures disrupted daily routines and hindered social processes, leaving people in a state of uncertainty, shock, fear and helplessness [2,3,4]. The fear of the virus itself, coupled with sudden changes in the social environment and economic pressures, has made a subset of individuals prone to mental health problems, including negative emotions such as anxiety and depression symptoms, decreased cognitive performance, poor sleep quality, and post-traumatic stress symptoms [5, 6]. Consequently, a number of studies have focused on the psychosocial implications of the pandemic [7, 8], and individual differences in psychological reactivity to prolonged exposure to the stress imposed by the pandemic [9].

To identify individuals susceptible to mental health issues arising from stressors such as the COVID-19 pandemic, some researchers are applying neuroimaging tools to investigate and validate biomarkers that characterize the brain’s stress response [10]. Indeed, neuroimaging has been widely used in clinical psychology and psychiatry research as a non-invasive technique for exploring brain mechanisms and determining biomarkers of mental health [11,12,13]. Researchers have used imaging methods such as structural magnetic resonance imaging (sMRI), functional MRI (fMRI), and electroencephalography (EEG) to explore the neural mechanisms at the level of brain regions, connectivity and large-scale networks in the context of mental health problems arising during the COVID-19 pandemic [14,15,16]. However, findings are heterogeneous and reliable neural signatures underlying pandemic-induced mental health problems remain elusive.

Herein, we conducted a systematic review to synthesize neuroimaging-based findings linked to various types of COVID-19-related mental health problems into a coherent framework, which may assist in identifying vulnerable individuals and implementing targeted measures pre-emptively. In particular, in this study, we focused on a large group of general population without pre-existing neurological or psychiatric disorders to ensure that the observed neural correlates were specifically driven by the psychosocial impact of the pandemic in the realm of subclinics. This enables isolation of pandemic-related neural changes from those attributable to other neurobiological abnormalities, thereby strengthening the validity of our findings regarding pandemic-specific neural mechanisms. We also hope to shed light on how to handle other major threatening life events similar to the COVID-19 pandemic in the future.

Methods

Search strategy and study selection

This systematic review was conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [17]. A systematic strategy was used to search for relevant studies published in PubMed, Web of Science and Medline up to March 19, 2025, with the following search terms: (1) COVID; coronavirus; pandemic; or SARS and (2) brain; magnetic resonance imaging; gray matter; white matter; computer tomography; electroencephalography; diffusion tensor imaging; single-photon emission computed tomography; positron emission tomography; functional near-infrared spectroscopy; MRI; CT; EEG; DTI; SPECT; PET; fNIRS; or neuroimaging and (3) mental health; psychological health; anxiety; depression; stress; PTSD; negative affect; cognition; fear; worry; distress; fatigue; burnout; or emotions. In addition, we conducted manual searches in the reference lists of the obtained articles to identify additional studies that need to be included. Details of the literature search and eligibility assessment for our analysis are shown in Fig. 1.

Fig. 1
figure 1

Flowchart of literature search and selection criteria for systematic review.

Full size image

The inclusion criteria were as follows: (1) using self-reported or other measures for the psychosocial effects of COVID-19; (2) using neuroimaging measures and reporting findings on the brain-mental health association. Studies were excluded if they were: (1) non-English publications; (2) non-human studies; (3) literature review, meta-analysis, conference abstract, case report or letter; (4) exploring biological effects of COVID-19; (5) exclusively enrolled participants with clinically diagnosed neurological or psychiatric disorders. Two authors (Y. G. and N. P.) independently completed the screening and assessment procedures for each article to ensure their appropriateness for inclusion. Any disagreement will be mediated and a consensus was reached after discussion with a third researcher (S. W.).

To provide a detailed description of each study in the systematic review, we collected information including sample size, gender ratio, age range, neuroimaging measures, psychometric measures, and primary findings (see Tables 1, 2 and 3).

Table 1 Summary of sMRI studies.
Full size table
Table 2 (A) Summary of task-based fMRI Studies. (B) Summary of rs-fMRI.
Full size table
Table 3 Summary of other studies.
Full size table

Results

Included studies and sample characteristics

A total of 6228 candidate articles were retrieved after excluding 8058 duplicates. After reviewing titles and abstracts, 4942 unrelated articles were excluded. Next, we assessed the eligibility of the remaining 1286 full-text articles. The following studies were excluded: studies without neuroimaging measures (n = 433), studies without psychometric measures (n = 45), and studies with unrelated themes (n = 762). Thus, 46 articles that met the inclusion criteria were identified in the systematic review (Fig. 1). 9 studies used sMRI to analyze the relation between COVID-19-related mental health problems and brain structure (for demographic characteristics and research details, see Table 1), 30 fMRI reports analyzed their relation to brain functional connectivity and activation (Table 2), 7 investigations used EEG, and 1 used functional near-infrared spectroscopy (fNIRS) (Table 3).

Regional-level neural substrates

Refer to Fig. 2 for the weighted contributions of brain regional mechanisms to COVID-19-related mental health issues.

Fig. 2: Weighted Contributions of Brain Regional Mechanisms to COVID-19-Related Mental Health Issues.
figure 2

Brain regions with a longer and more pronounced radar indicate a higher number of reported studies in that particular region. CG cingulate gyrus, INS insula, OG occipital gyrus, PCL paracentral lobule, PFC prefrontal cortex, SUB subcortical regions, TPC, temporoparietal cortex.

Full size image

Prefrontal cortex (N = 12)

Four articles revealed the role of the lateral prefrontal cortex (PFC) in COVID-19-related mental health issues. First, in a sMRI study using the prospective cohort design, higher right lateral PFC thickness prior to the pandemic was found to be associated with worse mental health outcomes, mainly manifested as a higher risk of anxiety and depression symptoms [18]. Second, another study revealed that COVID-related vicarious traumatization exhibited a positive correlation with gray matter volume (GMV) in the right dorsolateral PFC, while a negative association with resting-state functional connectivity between the right dorsolateral PFC and right precuneus [19]. Furthermore, the right dorsolateral PFC volume and dorsolateral PFC-precuneus connectivity mediated the effect of childhood cumulative trauma on vicarious traumatization [19]. Third, a task-based fMRI research reported that the fear for infection during the pandemic was positively correlated with the right dorsolateral PFC activation patterns prior to COVID-19 during a Theory of Mind task [20]. Forth, in another task-based fMRI research, investigators found that neural response to sad faces in the right ventrolateral PFC was positively correlated with negative memory biases [21].

Several studies have also implicated the medial PFC. For example, an sMRI study found that occupational burnout of medical professionals during the COVID-19 period was negatively correlated with GMV in the ventromedial PFC [22]. Lan et al. found a positive association between post-traumatic stress symptoms during the COVID-19 period and GMV in the left medial PFC, and a negative association between post-traumatic growth and GMV in the left dorsolateral PFC [23]. Specifically, participants with lower GMV in the ventromedial PFC showed higher levels of emotional exhaustion and depersonalization, two subcomponents of burnout [22]. In a resting-state fMRI (rs-fMRI) study, compared to local students, overseas students showed significantly lower regional homogeneity (ReHo) values in left superior and medial frontal gyri, pre-central gyrus, and paracentral lobule during the COVID-19 pandemic. Importantly, these ReHo reductions showed a significant positive correlation with depressive symptoms [24]. Another rs-fMRI study showed that pre-pandemic degree centrality of the PFC (mainly the posterior orbital gyrus) was positively associated with social anxiety during the pandemic [25]. In four studies using brain functional connectivity, weaker connectivity in the PFC was associated with increased anxiety [26], depression [27], post-traumatic stress symptoms [10], and perceived stress [15].

Cingulate cortex (N = 9)

Diffusion-weighted imaging studies on cingulum fiber density and cross-section (FDC) reported different neuroimaging patterns, and after accounting for known sex differences. For example, cingulum FDC was negatively associated with depressive symptom severity during the pandemic in just females, while higher cingulum FDC predicted higher resilience and lower stress in both sexes [28]. Indeed, a positive correlation has been reported between rostral anterior cingulate cortex (ACC) thickness and “tension” symptoms in individuals with low maladaptive coping [14]. Similarly, two sMRI studies found a positive association between GMV in the dorsal ACC and post-traumatic stress symptoms during the pandemic and a negative association between GMV of the dorsal ACC extending to the dorsal medial PFC and the optimism score [23, 29]. Further seed-based structural covariance network (SCN) analysis found an optimism-linked SCN covarying with the combined dorsal ACC and dorsal medial PFC. Additionally, optimism mediated the impact of dorsal ACC-dorsal medial PFC volume and its SCN on COVID-19-specific posttraumatic stress symptoms [29]. In task-based fMRI using sad faces, neural responses in the bilateral dorsal ACC have been positively associated with negative memory biases, and negatively associated with future depressive symptoms during the COVID-19 pandemic [21]. Three functional connectivity studies have shown that there are nodes in the ACC that play a significant role in COVID-19-related anxiety [26, 30] and post-traumatic stress symptoms [10] respectively, while another one suggested that node strength in the posterior cingulate cortex (PCC) is associated with COVID-19-related depression symptoms [27].

Insula (N = 6)

Two studies have revealed that anxiety induced by COVID-19 is related to pre-pandemic reduced insula thickness [14] and weaker connectivity from insula [26]. Two other functional studies focused on COVID-19-related depressive symptoms: one rs-fMRI study exhibited that the nodes strength located in the right insular can predict depression during the pandemic [27]; a task-based fMRI study found that adolescent girls with high shy/fearful temperaments showed negative associations between insula activation to social reward and COVID-19 depressive symptoms, whereas girls with lower shy/fearful temperament showed positive associations [31]. In addition, in a task-based fMRI study, greater anterior insula activation to unpredictable shock (U-threat) was associated with greater COVID-related negative affect [32]. Finally, levels of emotional exhaustion in medical professionals were negatively correlated with GMV in the left insula [22].

Temporoparietal regions (N = 7)

One sMRI study found that the GMV in the right supramarginal gyrus might be positively associated with social anxiety symptoms during the pandemic [33]. An rs-fMRI revealed that enhanced pre-COVID neural activity (measured by fractional amplitude of low-frequency fluctuation, fALFF) in the right fusiform gyrus could predict severe COVID-related post-traumatic stress symptoms and social anxiety during the pandemic, and the COVID-related post-traumatic stress symptoms may serve as mediators [34]. One investigation reported that the mother’s subjective burden of care increased during the pandemic, and the higher right temporoparietal junction activation during Theory of Mind task preceded a higher subjective burden [20]. A prospective longitudinal study found that pre-pandemic measures of functional connectivity could predict post-traumatic stress symptoms, and its neural connectome was mainly anchored on the temporoparietal junction [10]. Further, Mao et al. found that weaker connectivity stemming from the right inferior temporal gyrus was associated with depression during the COVID-19 pandemic [27]. In a study aimed to identify the functional connectome that encodes individual variations of pandemic-related vicarious traumatization, only the negative network model stably predicted individuals’ vicarious traumatization scores, with the contributing connected nodes primarily distributed in the bilateral angular gyrus and inferior temporal gyrus, right temporal pole, fusiform gyrus [35]. Similarly, the nodes that play a key role in predicting the individual perceived stress level during the pandemic were located in the bilateral precuneus, right middle temporal gyrus, and left inferior temporal gyrus [15].

Subcortical regions (N = 11)

Three resting-state fMRI studies have respectively proven that hub nodes related to anxiety are located in the thalamus, hippocampus, and parahippocampal gyrus [26], nodes related to post-traumatic stress symptoms are located in the hippocampus and amygdala [10], and nodes that play a key role in predicting individual perceived stress levels are located in the parahippocampal gyrus [15]. Task-based fMRI paradigms showed that higher activation of the left amygdala during fearful faces before the pandemic was related to greater internalizing problems during the pandemic [36]. The hippocampus has been considered the most critical hub region in predicting the emergence of distress after COVID-19 [37], while the node that exerts a critical role in predicting depression during the COVID-19 pandemic has been located in the thalamus [27]. Girls with high levels of shy/fearful temperament showed negative associations between caudate and putamen activation to social reward and COVID-19 depressive symptoms, whereas girls lower in shy/fearful temperament showed positive associations [31]. In a passive avoidance task, lower expected value signaling in the right nucleus accumbens predicted higher withdrawn symptoms throughout the pandemic [38]. Hardi et al. identified a distinct neural network variation associated with greater anxiety symptoms during the COVID-19 pandemic, characterized by more connections involving the left amygdala and ventral striatum [30]. Morphologically, higher levels of depersonalization (a dimension of occupational burnout) were related to the lower GMV in the left thalamus [22]. Another sMRI study revealed that there was a negative correlation between amygdala volume and anhedonia symptoms in individuals with high maladaptive coping [14].

Network-level neural substrates

Refer to Fig. 3 for the weighted contributions of large-scale brain network mechanisms to COVID-19-related mental health issues.

Fig. 3: Weighted Contributions of Brain Network Mechanisms to COVID-19-Related Mental Health Issues.
figure 3

Large-scale brain networks with a longer and more pronounced radar indicate a higher number of reported studies in that particular network. AFN affective cortical network, CEN central executive network, DMN default mode network, SMN sensorimotor network, SN salience network, VN visual network.

Full size image

Affective cortical network (AFN, N = 10)

Perica et al found that greater connectivity within the AFN (mainly between anterior ventromedial PFC and posterior hippocampus) predicted greater COVID-19-related stress in youth under 18 years old [39]. Perceived stress is also related to weaker connectivity between AFN and two other networks [DMN and visual network (VN)] [15]. Additionally, Pan et al. identified an anxiety-related mode characterized by high loadings in connectivity between the AFN and VN when exploring the multivariate patterns of brain functional connectome predicting COVID-19-related negative affect symptoms [40]. Higher AFN system segregation (SyS, a graph-theoretical parameter to quantify the balance between within- and between-network integration and represent efficient functioning of the network, which is formally calculated as the difference between the mean magnitudes of between-system correlations from the within-system correlations as a proportion of mean within-system correlation [41]) could increase the effect of coping strategies that regulate COVID-19-related perceived stress, anxiety and depression symptoms [42]. Using rs-fMRI analyses, one study found that fear symptoms and their severity during the COVID-19 pandemic were associated with stronger connectivity within the AFN and between AFN and salience network (SN) [43]. COVID-19-related depressive symptoms have been associated with weaker connectivity within the AFN [27] and between the AFN and VN, especially between the left amygdala and the bilateral lingual gyrus [44]. In another study, stronger AFN-SN connectivity (mainly the basolateral amygdala–subgenual ACC connectivity) was associated with heightened depressive symptoms in adolescents, both before and during the pandemic [45]. Some investigators have reported that pre-pandemic measures of functional connectivity between AFN and SN can predict COVID-19-related post-traumatic stress symptoms [10]. However, an fMRI region of interest (ROI) analysis failed to find any significant association between depression or anxiety symptoms during the pandemic and AFN functional connectivity [46].

Central executive network (CEN, N = 4)

One study reported that higher CEN-SyS levels enhance the positive association between perceived stress and anxiety and depression symptoms during the COVID-19 pandemic [42], and another showed that lower pre-pandemic CEN coherence predicted more severe internalizing symptoms during the pandemic in early-maturing youths [47]. An investigation reported an association between COVID-19 vicarious traumatization and lower connectivity within the CEN pre-pandemic [35]. Researchers also found weaker local connections within the CEN in participants with post-traumatic stress symptoms compared to those without such symptoms during the COVID-19 pandemic [10].

Default mode network (DMN, N = 6)

Higher levels of DMN-SyS were found to attenuate the positive association between perceived stress and anxiety and depression symptoms during the COVID-19 pandemic [42]. Further, a connectome-based predictive modeling analysis showed that the negative DMN model stably predicted individuals’ COVID-19 vicarious traumatization. Mediation analysis revealed that this vicarious traumatization mediated the influence of DMN on general distress [35]. Another rs-fMRI study also found that lower connectivity between inferior temporal gyrus and the DMN predicted worse vicarious traumatization during the pandemic [48]. Pan et al. found that the weaker connections within DMN and between network connections of AFN-DMN and dorsal attention network-DMN predicted more severe distress symptoms during the pandemic [37]. An fMRI study investigating the multivariate patterns of brain functional connectome in predicting COVID-19-related negative affect symptoms revealed that connectivity of the DMN with dorsal attention network were remarkably prominent in mode stress [40]. In addition, it was reported that functional connectivity between the DMN and SMN/attention networks was decreased during the pandemic. And the absolute values of the DMN-SMN connectivity changes showed a negative correlation with the improvement in stress, anxiety, depression, and negative affect symptoms (positive correlation with improvement in positive affect symptoms), while the absolute values of the DMN-attention network connectivity changes showed a negative correlation with the improvement in mental status measurements in stress, anxiety, and depression [49].

Salience network (SN, N = 4)

Lower connectivity within the SN may predict COVID-19-related vicarious traumatization [35]. Further, depression during the COVID-19 pandemic is associated with weaker intrinsic connectivity within the SN [27]. Higher connectivity within the SN and between the SN and two other networks (DMN and CEN) are associated with less negative emotions during and after COVID-19. Further, the buffering effect of these functional networks was stronger in the context of higher levels of social support [50]. Similarly, Hu et al. found that higher resilience of the SN predicted better mental health during the COVID-19. Further, lower connectivity of left SN, reward, limbic, and PFC and its thalamic, striatal, amygdala connections, predicted higher stress and sadness. And lower bilateral robustness (higher fragility) and/or connectivity of these networks predicted higher sadness [51].

Sensorimotor network (SMN, N = 1)

COVID-19 vicarious traumatization was associated with weaker connectivity of the motor network, and mediation analysis indicated that this vicarious traumatization mediated the influence of the motor network on general distress [35].

Other studies

EEG (N = 7)

Compared to pre-pandemic controls, who completed two test sessions of EEG recording and cognitive tasks before the pandemic, participants during the pandemic who completed a second testing session of EEG recording and cognitive tasks during lockdown, were characterized by higher extraversion (tested by self-reported measures before the pandemic), improved cognitive performance and stronger EEG brain global connectivity. Importantly, stronger EEG connectivity and higher extroversion was inferred as a potential defense mechanism against stress-related deterioration of cognitive functions [52]. Among COVID-19 frontline healthcare professionals, reduced cognitive performance and poorer sleep quality was related to higher theta relative power, lower peak alpha frequency, and higher interhemispheric coherence of both alpha and theta rhythms [53]. In one event-related potential (ERP) study, greater COVID-19-related depressive symptoms were associated with reduced late positive potentials (LPPs) while viewing pleasant interpersonal images. In another study, enhanced LPPs while viewing threatening interpersonal images predicted increases in traumatic intrusions after the outbreak of COVID-19 [16]. Interactions between anxiety and LPPs recorded during passively viewing task of affective pictures could predict pandemic-related post-traumatic stress symptoms: greater anxiety symptoms predicted post-traumatic stress symptoms specifically in individuals with greater LPPs to unpleasant stimuli and with reduced LPPs to pleasant stimuli [54]. In two other publications using ERP analyses, larger pre-pandemic error-related negativity (ERN) and correct-response negativity (CRN) amplitudes were associated with increased perceived internalizing symptoms [55, 56]. Specifically, enhanced delta-ERN predicted greater anxiety among adolescents with an increased reliance on reactive control [55]. Further, pre-pandemic ERN and CRN correlated with increased perceived risk regarding a COVID-19 infection and increased stress during the pandemic, which mediated indirect effects of ERN and CRN on internalizing psychopathology, including anxiety, depression, and obsessive-compulsive symptoms [56]. Finally, EEG response to transcranial magnetic stimulation (TMS) could predict an individual’s capacity to resist COVID-19-related psychological stress, where larger late EEG responses locally post-left dorsolateral PFC stimulation predicted increased mental distress [57].

Functional near-infrared spectroscopy (fNIRS, N = 1)

One publication using fNIRS reported that the reduced hemodynamic response of the PFC is associated with negative emotions, especially anxiety, during the COVID-19 lockdown, but immediate music stimulation alleviated those negative emotions by altering connectivity patterns in the PFC [58].

Discussion

To our knowledge, this is the first systematic review examining neural substrates of COVID-related psychosocial impact. Despite some discrepancies in results between studies, our study shows that COVID-related mental health problems are mainly associated with brain structural and functional patterns in the PFC, insula, cingulate, hippocampus, and amygdala, as well as AFN at the brain-network level. These findings provide insights about the neural basis of adverse psychological issues triggered by COVID-19, which may provide inform the development of non-invasive clinical interventions. Below, we will discuss our results briefly, integrating the brain regional findings into a framework of large-scale network systems that may be relevant to pandemic-related mental health impact.

The AFN mainly includes the amygdala and ventromedial PFC, which widely participates in engendering and regulating vigilance and arousal responses to biologically salient stimuli [59, 60]. The structural and functional changes within the AFN may be associated with symptoms such as reactive aggression, anger, and irritability [61, 62] and manifest in mood and personality disorders [63,64,65]. The amygdala is widely known to generate negative and unpleasant emotions, especially fear, and in associating environmental stimuli with emotionally charged and aversive sensory inputs [66]. Long term chronic stress partially mediates changes in amygdala neuron excitability through potassium channel function, which may lead to overactivity of circuits related to fear and anxiety, and reduce the ability of other regions involved in fear inhibition, such as the hippocampus and medial PFC, to dampen amygdala output [67]. The unprecedented uncertainty of the COVID-19 infection and enforced quarantine [7] likely increased the risk of stress activation processing. During the COVID-19 pandemic, weaker AFN connectivity was associated with the negative emotions (perceived stress, anxiety and depression) [27, 40, 42]. However, Perica et al. found the opposite: the higher connectivity within AFN predicted greater COVID-19-related stress [39]. In addition, the inter-connections between AFN and other networks were also involved in regulating negative emotions. Specifically, weaker AFN-DMN and AFN-VN connectivity respectively predicted more severe perceived stress and depression symptoms during the COVID-19 pandemic [15, 44], while the AFN-SN and AFN-VN connectivity was positively correlated with depressive and anxiety symptoms, respectively [40, 45]. Meanwhile, dysregulated interaction between AFN and SN was also associated with COVID-19-related fear and PTSD symptoms [10, 43]. For example, higher bottom-up arousal signaling from amygdala, insula, ACC, and thalamus could increase the susceptibility of negative emotions to COVID-19 [68,69,70], whereas the stronger top-down control by the PFC of these subcortical regions could inverse this process [71].

The CEN, mainly consisting of the dorsolateral PFC and posterior parietal cortex [72], is crucial for the cognitive regulation of emotion, behavior, and thought, especially activating in efforts to exert self-control, reassess threatening stimuli, and suppress invasive and unpleasant thoughts [73,74,75]. The deficits in these processes are often characteristics of multiple adverse mental health outcomes [76,77,78]. The dorsolateral PFC is a well-recognized cognitive control region that subserves inhibitory control, cognitive flexibility, and working memory [79], especially the right dorsolateral PFC is involved in the downregulation of negative emotional conditions [80]. COVID-related vicarious traumatization is associated with dorsolateral PFC GMV and connectivity, which is broadly consistent with a previous study showing that trauma-exposed individuals had greater dorsolateral PFC thickness relative to controls over a year after trauma, and greater dorsolateral PFC thickness was associated with greater post-traumatic stress symptoms reductions and better recovery [81]. Therefore, it can be inferred that a larger dorsolateral PFC GMV may facilitate better management and suppression of distressing memories and emotions, which are more frequently arised from witnessing others’ traumatic experiences in these individuals. One task-based emotion regulation fMRI study showed that lateral PFC activity was associated with greater stress reduction in participants with elevated trait anxiety scores, suggesting that lateral PFC dysfunction may be associated with psychiatric symptoms that manifest dysfunctional down-regulation of negative emotion and excessive fluctuation of emotions [82]. Similarly, research from the COVID-19 pandemic also confirmed that abnormalities in lateral PFC contributed to robust predictive markers of during-pandemic negative mental health symptoms [10, 18, 20, 21]. Deficits in the CEN, including weaker intrinsic connectivity within its nodes, aberrant cross-network connectivity patterns, or impaired access to salient task-relevant stimuli, have all been reported in psychiatric disorders such as depression, schizophrenia and autism [83,84,85]. Similarly, lower CEN coherence has also been associated with severe mental health problems during the COVID-19 pandemic period [10, 35, 47].

The DMN is a large-scale brain network identified in this review to be associated with severe mental symptoms during the COVID-19 pandemic. As this review summarizes, DMN plays a crucial role in attenuating the positive association between perceived stress and anxiety/depression symptoms, predicting vicarious traumatization, and influencing general distress during the COVID-19 pandemic [30, 32, 34, 40]. Additionally, the functional connectivity between the DMN and other networks (e.g. the SMN and attention network) is also involved in the impact of COVID-19-related negative mental status from the prolonged isolation, reduced physical and social activities [49]. This is in line with prior research on neural correlates of broad psychological impairments [86, 87]. DMN dysfunction has been recognized as a critical biomarker in various psychiatric disorders involving disrupted self-referential processing, such as PTSD [88], depression [83], anxiety [89], and schizophrenia [90]. Concerning DMN’s components, the medial PFC subsystem plays an important role in regulating emotion and stress responses [85, 86, 91]. Medial PFC impairment may be associated with anxiety and PTSD through impaired fear memory regulation [92], which could be exacerbated by pandemic-related stressors. The dorsomedial PFC is instrumental in mentalizing and metacognitive processing [93], and its hyperactivity coincides with rumination, a core symptom of major depressive disorder [94]. The hippocampus, another key region of the DMN, is crucial for cognitive functioning and the regulation of stress and emotion [95, 96]. Besides being generally regarded as a major target of stress mediators [97], the hippocampus is also involved in the extinction of fear memories, and lesions in the hippocampus may be associated with extinction deficits in PTSD [97]. Reduced activation of brain regions of DMN (including medial PFC, PCC, posterior inferior parietal lobule, and the parahippocampal gyrus), as well as decreased functional connectivity within DMN nodes (including the PCC, posterior hippocampus and ventromedial PFC), have been reported in PTSD patients [98, 99]. A decrease in DMN coherence is also believed to be the basis for impairments in self-referential processes, autobiographical memory and altered sense of self in PTSD patients [88]. Altogether, individuals with weaker connections of DMN are more likely to experience mental distress, especially during the period of the COVID-19 pandemic with a high risk of stress [35, 37, 42].

The SN, anchored in the ACC and insular cortex, is responsible for detecting, integrating, and filtering relevant interoceptive, autonomic and emotional information [72]. Reduced volumetric patterns of ACC might represent a biomarker of depression and predict long-term prognosis [100, 101]. Existing evidence has demonstrated that the dorsal ACC strongly communicates with dorsomedial PFC in the process of emotion regulation, especially in regulating negative emotions, such as fear and anxiety [102, 103]. Dysfunctional increasing neural activity in the dorsal ACC and dorsomedial PFC potentially leads to severe anxiety symptoms and even major depressive disorders [102, 104]. The indirect connection from ACC and the PFC to the amygdala through a top-down inhibitory pathway contributes to the assessment, acquisition, and cognitive regulation of fear [105]. An increase in the strength of connectivity between the amygdala-ACC and dorsolateral PFC may be related to increased fear processing function and PTSD symptoms [106]. As for the insular section of SN, disturbances including those related to several sensory and multimodal perceptions, as well as body awareness, the emotion of disgust, mood and willed action, may all appear following insular damage [107]. Insular activation patterns might also serve as a focal point in addressing the mental health challenges arising from the pandemic given their involvement in processing aversive interoceptive stimuli and altered prediction of an aversive body state that may trigger an increase in anxious and worrisome thoughts in individuals prone to anxiety symptoms [108]. The anterior insula plays a key role in the anticipation and emotional experience of aversive stimuli, and participates in the allocation of attention and initiation of appropriate action through the ACC [109]. Patients with generalized anxiety disorder have shown reduced insula activation after taking anti-anxiety drug citalopram to reduce worry symptoms [110]. The SN plays a dynamic switching role between self and inner world attention mediated by the DMN and task-related directed attention of outside stimuli maintained by the CEN [111]. This may explain the findings that individuals with higher SN-DMN and SN-CEN connectivity exhibited less negative emotions during and after the COVID-19 pandemic [50]. Overall, strength and robustness of the SN prior to the pandemic may have been a protective factor for mental health, whereas its fragility and weaker connections were risk factors that increased vulnerability to the pandemic’s adverse effects.

Quantitative parameter analysis of EEG signals could evaluate altered rhythms that are related to mental health [112, 113]. Higher theta relative power and lower peak alpha frequency during COVID-19 in frontline healthcare workers were associated with experiences of increased work stress, reduced cognitive performance, and poor sleep quality [53]. Theta waves are commonly present in EEG images during trance or hypnotic states and excessive theta activity may indicate depression, while low alpha frequency may indicate greater stress or anxiety [114, 115]. EEG also reflects brain functional connectivity and quantifies the interactions between different neuronal networks [116]. Overall, stronger EEG connectivity during the pandemic may have served as a protective mechanism to prevent stress-related cognitive deterioration [52]. The LPPs, as a neurophysiological measure of emotional reactivity, have sensitivity to discriminate emotional relative to neutral cues [117]. Specifically, reduced LPPs to positive stimuli were found in individuals at risk for depression [118], while elevated LPPs, generally to negative or threatening stimuli, were recorded in individuals at risk for anxiety [119, 120]. Therefore, LPPs may be a potential signal to reflect mental health during COVID-19 period. TMS interferes with ongoing brain activity by suddenly injecting a certain amount of current into the neural circuit, resulting in phase reset and being recorded by EEG [121]. Therefore, EEG quantification of TMS response could be used to track an individual’s capacity to resist COVID-19-related psychological stress [57]. Finally, fNIRS, with the advantages of relatively high spatial and temporal resolution and better comfort, may indirectly monitor brain activity by measuring changes in oxyhemoglobin and deoxyhemoglobin concentrations in cerebral blood vessels [122]. A study using fNIRS analysis revealed the relationship between the reduced hemodynamic response of the PFC and negative emotions during the pandemic [58], illustrating its potential utility as a tool to track neuropsychiatric symptoms.

There are several limitations of our systematic review. First, the imaging modalities used in the literature are mostly fMRI and EEG. It is well known that fMRI exhibits notable spatial resolution, but poor temporal resolution, while EEG does the opposite [123]; simultaneous EEG and fMRI, to bridge the high temporal resolution of EEG and the high spatial resolution of fMRI, could be a promising research approach in the future. In addition, other imaging technologies, such as positron emission tomography (PET), which can quantify brain metabolic information [124], has not been applied to relevant research. Second, many psychometric measures in research rely on participants self-administered questionnaires, introducing a notable degree of subjectivity into the obtained results; future studies that combine self-reports with biological and validated clinical and neuropsychological tests could improve objectivity. Third, the sample of participants across studies varied widely in terms of sociodemographic characteristics, and the primary psychological outcomes measured, the imaging modalities, the assessment tools and analysis protocols were highly heterogeneous across studies. This heterogeneity hindered the comparability of findings and synthesis of results. With more studies using more standard measures, quantitative meta-analysis could be conducted. Forth, this review primarily focuses on elucidating neural mechanisms underlying COVID-19-related mental health issues. While these findings provide important mechanistic insights, we recognize the current limitations in direct clinical application, highlighting the need for future translational studies to bridge this gap. Finally, it is recommended that future studies should incorporate large and diverse samples of well-characterized participants (i.e., frontline healthcare workers and other vulnerable samples), contributing to more reliability and validity of the subject matter.

Conclusion

The COVID-19 pandemic caused tremendous harm to human health, including an unprecedented threat to mental health in various populations. To our knowledge, this is the first systematic review of neuroimaging applications aimed to characterize neural features associated with adverse mental symptoms related to the COVID-19 pandemic. Our study identified how COVID-19 related-mental health problems are linked to specific brain regions and networks, principally including the PFC, insula, cingulate, hippocampus, amygdala and AFN. Our findings may serve as a foundation to effectively address the acute stress-related effects of the pandemic, and may also provide insights for proactively preventing adverse mental symptoms in individuals during future large-scale stressful events.