Three main hypotheses are currently considered to explain Neuro-PASC (8). The first is by indirect mechanism, via peripheral inflammation, also known as the “cytokine storm” that is triggered by SARS-CoV-2. This systemic inflammatory response could alter the equilibrium of the brain, similar to a septic encephalopathy (9). Cytokine storms occur when the fine balance between pro-inflammatory and anti-inflammatory cytokines is lost, and a previously localized inflammation affects the entire system. This deregulation can be non-provoked, as in auto-inflammatory illnesses, or can be precipitated by an infectious agent, as is the case for COVID-19. In the latter, studies have shown an elevation of IL-1β, IL-2, IL-6, IL-10, IFN-γ, and TNF-α, amongst others, and that the degree of elevation was proportional to the severity of the illness (10). The second theory is also one of indirect mechanism, and rests on the basis of lack of perfusion, sepsis or hyperpyrexia that can accompany the respiratory infection, each with their respective toll on the brain's homeostasis. The third hypothesis is one of direct neurotropism (11). SARS-CoV-2 infects the human body by binding to ACE2 receptors, which are primarily found in type 2 alveoli of the lungs (12) and entering the cells. Particularly, ACE2 receptors are also present in the brain's vascular endothelium and smooth muscles, as well as in skeletal muscles, thus explaining why myalgia is a widely present symptom included in post-COVID syndrome. Skeletal muscle cells express ACE2 receptors, which makes them a direct target for SARS-CoV-2 invasion (8). In support of this third theory, previous studies demonstrated that SARS-CoV-2 enters the nervous system by latching onto ACE2 receptors of the olfactory mucosa, penetrating the neural-mucosal interface, the only part of the brain not protected by the dura (3), and migrating along neuronal structures, eventually leading to the centers controlling cardiorespiratory functions (11). Whether Neuro-PASC is caused by direct attack of the virus on the brain or indirectly from systemic secondary damages, or by a combination of these (2), needs to be clarified.

Despite the immensely rapid response and development of COVID-19 management protocols, many gaps need to be filled when it comes to understanding the pathophysiology of the disease as a whole. For example, up to 55% of hospitalized COVID-19 patients have reported to have neurological manifestations 3 months after their infection (2). This is still not well investigated and studied. This article serves as a review of the radiological markers that are currently in development for the evaluation of neurological damages of COVID-19. We will discuss ¹⁸F-FDG-PET/CT, SPECT, and ¹⁸F-AMYLOID PET/CT imaging, as well as structural, functional, diffusion MRI, ASL, and susceptibility-weighted imaging, and their current contribution in the management of COVID-19 specifically from a neurological standpoint.

A record search was performed to identify neuro-imaging methods for tracing Neuro-PASC including ¹⁸F-FDG-PET/CT, SPECT, MRI and fMRI. Databases searched were PubMed and Google Scholars for papers published between March 2020 and April 2023. Searches were divided by neuro-imaging methods for tracing Neuro-PASC. Studies were included if they (1) examined neurological markers, (2) used neuro-imaging modalities, (3) studied patients with a COVID-19 history and, (4) were in English. We excluded articles that examined children, adolescents and pregnant women or consisted of post-vaccination studies.

With regards to 18 F-FDG-PET/CT imaging, the search key was ((((Neuro^*) OR (brain)) AND ((Covid^*) OR (coronavirus) OR (SARS-CoV-2))) AND ((long covid) OR (post covid))) AND ((PET) OR (positron)). For SPECT imaging, ((((Neuro^*) OR (brain)) AND ((Covid^*) OR (coronavirus) OR (SARS-CoV-2))) AND ((long covid) OR (post covid))) AND ((SPECT) OR (single photon)). For 18F-amyloid PET/CT, ((((Neuro^*) OR (brain)) AND ((Covid^*) OR (coronavirus) OR (SARS-CoV-2))) AND ((long covid) OR (post covid))) AND ((amyloid) OR (amyloid PET)). For structural MRI, (((((Neuro^*) OR (brain)) AND ((Covid^*) OR (coronavirus) OR (SARS-CoV-2))) AND ((long covid) OR (post covid))) AND ((MRI) OR (Magnetic resonance imaging)). For functional MRI, ((((Neuro^*) OR (brain)) AND ((Covid^*) OR (coronavirus) OR (SARS-CoV-2))) AND ((long covid) OR (post covid))) AND ((fMRI) OR (functional MRI)) yielded no results addressing our topic's subject. For ASL MRI, ((((Neuro^*) OR (brain)) AND ((Covid^*) OR (coronavirus) OR (SARS-CoV-2))) AND ((long covid) OR (post covid))) AND ((ASL MRI) OR (arterial spin labeling)). For diffusion MRI, ((((Neuro^*) OR (brain)) AND ((Covid^*) OR (coronavirus) OR (SARS-CoV-2))) AND ((long covid) OR (post covid))) AND ((diffusion MRI) OR (dMRI)). For susceptibility-weighted imaging, ((((Neuro^*) OR (brain)) AND ((Covid^*) OR (coronavirus) OR (SARS-CoV-2))) AND ((long covid) OR (post covid))) AND ((SWI) OR (Susceptibility-weighted imaging)). A targeted search was done on Google Scholar and PubMed for each neuro-imaging modality and post-covid (i.e.: post-covid PET neuro finding) to identify specific articles that may have been omitted.

Retained articles were checked for duplicates and were selected after undergoing a title, abstract and eventually a full text screening. The screening and selection process was done by two independent authors (OC and MM). Collectively, 92 abstracts were read and 38 were used to construct this review. Specifically, 1 for SPECT imaging, 9 for advances with ¹⁸F-FDG-PET/CT, 1 for ¹⁸F-Amyloid-PET/CT, 10 for structural MRI, 2 for functional MRI, 1 for diffusion MRI, 2 for SWI and the remainder for additional information on Neuro-PASC. A PRISMA flowchart can be found in Figure 1.

Positron emission tomography (PET) measures the emission of positrons from radiomarked tracer molecules, allowing localization of metabolically active processes. The most widely used radiotracer is currently ¹⁸F-Fluorodeoxyglucose (¹⁸F-FDG), a radiolabeled glucose molecule. This method allows localization of regions of abnormal increase or decrease glucose intake in the body to better identify hyper or hypometabolic states and is currently mainly used as part of the workup and follow-up in oncology. In the Neuro-PASC scope, the expected utility of ¹⁸F-FDG-PET/CT is to identify a specific cerebral target that explains the clinical manifestations (33). This will be useful to better determine the pathophysiology, to improve the accuracy of the diagnosis, to then better develop therapeutic options and accurate follow-ups. When the intrinsic role of the hyper/hypometabolic brain loci correlates with the clinical manifestations of Neuro-PASC, while it cannot prove causality, points toward understanding the pathophysiology of COVID-19 and helps clinicians make the diagnosis of Neuro-PASC.

Sollini et al.'s case control study in 2021 evaluated post-COVID syndrome hallmarks on ¹⁸F-FDG-PET/CT and showed brain hypometabolism in limbic, paralimbic, brainstem and cerebellum. In addition, none of the post-COVID patients exhibited regions of hypermetabolism compared to controls. Limbic and orbito-frontal hypometabolism correlated with the anosmia present during acute COVID-19 infection. Sollini's team also evaluated the whole-body ¹⁸F-FDG-PET/CTscans. Despite the hypometabolic activity in the nervous system, they found hypermetabolic activity in different organs, independent of ACE-2 receptor activity, in other words, regions of hypermetabolism were seen in target (ACE-2 presenting) as well as non-target organs, supporting the hypothesis that systemic inflammation may be in play (13).

Verger et al.'s (34) study showed that 47% of the ¹⁸F-FDG-PET/CT scans of 143 patients with suspected Neuro-PASC had hypometabolism in limbic/paralimbic and fronto-orbital olfactory regions, as well as in the brainstem and cerebellum.

Several other studies have been conducted using ¹⁸F-FDG-PET/CT imaging to reveal potential Neuro-PASC biomarkers. A study by Karimi-Galougahi et al. (14) brought forth reduced metabolic activity in the orbitofrontal cortex using ¹⁸F-FDG-PET/CT. Bilateral par hippocampal, fusiform gyri, and left insula hypometabolism were seen in an ¹⁸F-FDG-PET/CT imaging study conducted by Donegani et al. (15).

A study by Hosp et al. (9) that ran its course in parallel with the Montreal Cognitive Assessment test (MoCA), a highly sensitive tool for early detection of mild cognitive impairment (MCI), correlated cognitive decline with the evident frontoparietal hypometabolism seen on scan. This study showed that while a structural MRI could not find any sign of cerebral damage, ¹⁸F-FDG-PET/CT discovered cortical fronto-parietal hypometabolism.

A study by Guedj et al. (16), demonstrated that clusters of patients with hypometabolic patterns on ¹⁸F-FDG-PET/CT scans, flagged by whole-brain statistical analysis performed using SPM8, were very distinct in patients with Neuro-PASC and allowed to reliably differentiate their brain from one of a control subject. The findings suggested that hypometabolism was seen in bilateral rectal/orbital gyrus, including the olfactory gyrus, right temporal lobe, bilateral pons/medulla, and the cerebellum bilaterally. These patients exhibited many functional complaints. Patients with cerebellar hypometabolism experienced hyposmia, anosmia and memory impairment. Patients with frontal cortex, brainstem and cerebellum hypometabolism presented pain and insomnia.

¹⁸F-FDG-PET/CT has also proven to be useful in characterizing the timeline of damages caused by SARS-CoV-2. Post-COVID syndrome patients who had increased vascular uptake (hypermetabolism) on whole-body scan done at 1-month post-infection also showed hypometabolism in the brain, thus demonstrating the different effects of COVID-19 on different regions of the body, and the chronological sequence for brain and whole-body changes throughout the disease (13).

In contrast to these studies, Dressing et al.'s (17) study revealed that post-COVID patients reporting symptoms lasting for over 3 months after the acute infection, actually only presented mild impairment on cognitive testing (MoCA) with their distinct pathologic findings on ¹⁸F-FDG-PET/CT. In this same stretch of ideas, Ferruci et al.'s review explained that not all patients who presented cognitive decline in the post-COVID timeframe showed hypometabolism on ¹⁸F-FDG-PET/CT (20).

In Debs et al. (18) study, a time-dependent brain PET hypo- and hypermetabolism was observed in patients with a history of COVID-19. Hypometabolic activity, in the bilateral frontal, parietal, occipital, and posterior temporal lobes and cerebellum, reached its peak at 2 months post-infection, then nearly resolved at 6 months, followed by disappearance at 12 months. Additionally, 2–6 months after infection onset, hypermetabolism was observed in brainstem, cerebellum, limbic structures, frontal cortex, and periventricular white matter.

Single Photon Emission Computed Tomography (SPECT/CT) is a method that allows superposition of anatomical images with metabolic activity. While PET scans detect metabolism through emission of photons, SPECT scans detect gamma rays from tracers injected into the patient (35). Brain SPECT measures brain perfusion. A limited number of studies met the particular interest of this topic. Takao et al.'s (36) study on Neuro-PASC patients described observing hypoperfusion on SPECT imaging in various areas of the brain, particularly the frontal lobes, which correlates with the previous findings in 18 F-FDG-PET/CT scans. A study conducted on 32 patients with highly impaired functional status, determined by COOP/WONCA functional ability questionnaire (Cooperative Research Network/World Organization of National Colleges, Academies and Academic Associations of General Practitioners/Family Physicians), demonstrated cerebral perfusion changes on SPECT/CT in 29 patients (90%) (19). Fifteen patients then underwent a control SPECT-CT 3–9 months after their initial imaging taken during the acute infection. The follow-up image showed marked improvement in 8 cases and worsening in 7 cases.

Amyloid PET is an imaging modality that allows the visualization of amyloid plaques. A retrospective study conducted by Ferrucci et al. (20) found amyloid plaques in one Neuro-PASC patient 12 months after hospital discharge for his acute COVID-19 infection using an ¹⁸F-amyloid PET/CT. The plaques were found in the superior and middle frontal cortex, in the posterior cingulate cortex and in the rostral and caudal areas of the anterior cingulate cortex in a lower quantity. ¹⁸F- FDG PET/CT also detected hypometabolism in the left mesial temporal cortex of this patient (20). To our knowledge, this is the only published article that examined Amyloid PET.

The anatomy of the brain can be studied using an MRI, an imaging modality that allows the visualization of soft tissues using a magnetic field (37). Results by Hosp et al. (9) suggest that ¹⁸F-FDG-PET/CT seems to be more effective in detecting anomalies than 3T MRI. Specifically, ¹⁸F-FDG-PET/was able to detect frontoparietal hypometabolism, whereas 3T MRI found no relevant structural or vascular anomalies. The sequences used in this article were as follow: sagittal 3D-T1 rapid gradient echo (MP-RAGE) before and after contrast infusion, sagittal 3D FLAIR SPACE (sampling perfection with application-optimized contrasts using different flip angle evolutions), SWI and diffusion mesoscopic imaging (DMI). Nonetheless, 3T MRI allowed to localize a few subacute infarcts (9).

However, in a pilot study conducted by Thapaliya et al. (21), T₁-weighted 7T MRI showed an increased volume of the superior cerebella peduncle, pons, and entire brainstem in Neuro-PASC patients compared to controls. These biomarkers were also found in patients with myalgic encephalomyelitis/chronic fatigue syndrome which could explain the similarity in symptoms between these two conditions (21).

On another hand, Cecchetti et al. (22) found that COVID-19 patients showed interrelated cognitive, EEG and 3T T2-weighted structural MRI abnormalities 2 months after being discharged from the hospital. A UK Biobank longitudinal imaging study of 785 patients having been scanned by structural MRI (T1, T2 fluid attenuation inversion recovery (FLAIR) and susceptibility-weighted MRI) before, through indications for scans unrelated to COVID-19, and approximately 141 days after testing positive for COVID-19, showed statistically significant atrophy of gray matter in limbic cortical areas, directly linked to olfactory and gustatory systems 6 (23).

Another application for 3T T2-space structural MRI in post-COVID syndrome is demonstrated in Kandemirli et al.'s (24) study. Out of 23 patients with persistent COVID-19 olfactory dysfunction, nearly three-quarters had olfactory cleft opacification, a reduction in olfactory bulb volumes, change in bulb shape or signal anomalies (24). These MRI studies allowed for a better understanding of the damages inflicted on the olfactory pathway in the context of COVID-19 anosmia.

In a study conducted by Besteher et al. (25), increased gray matter volume clusters were found in post-COVID patients 8 months after their acute COVID-19 infection compared to controls using voxel-based morphometry (VBM). They used VBM with CAT12 toolbox, using T1-weighted images obtained from a 3T MRI to measure gray matter volumes. The clusters were found bilaterally in the frontal lobe, temporal lobe, insula, amygdala, hippocampus, basal ganglia, and thalamus. The volume of four of these clusters was found to decrease over time. The first cluster was in the anterior insula and some regions of the frontal lobe. The second cluster was scattered in the left temporal lobe. The third cluster was in the left post-central and precentral gyrus and finally, the fourth cluster was in parts of the temporal lobe, the right fusiform gyrus, the parahippocampal gyrus, and the hippocampus. It was hypothesized that the augmentation of the gray matter volumes was caused by compensatory processes or inflammation (25).

Hellgren et al.'s (26) observational cohort study compared the MRI imaging of post-COVID patients to the neuropsychological findings of these patients. 3T MRI (T2-FLAIR, T2-FSE, T1-FSE, T1-GRE, DWI and SWI) imaging was done on patients whose neurocognitive test results seemed concerning to researchers. The neurocognitive test performed was the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS). Thirty-five post-COVID patients who had been hospitalized for their acute COVID-19 infection and who were discharged on average 6.5 months prior underwent a brain MRI which revealed subcortical white matter lesions in 71% of these patients. The lesions were primarily found in the frontal and parietal lobes near the gray-white matter junction. Compared to the patients with normal MRI results, the patients with white matter lesions had lower scores in the RBANS visuospatial index. It is worth noting that the functionality level of the patients with white matter lesions prior to their COVID-19 infection was lower compared to controls (26).

Campabadal et al. (27) using 3T structural MRI found a reduction in gray matter volume in the caudate nucleus, putamen, olfactory cortex, parahippocampal gyrus, straight gyrus, left amygdala, and the superior and inferior orbital gyri using voxel-based morphometry. A study by Yus et al. (28) using 3T structural MRI (3D-T1, T2-FLAIR) found no statistically significant findings in brain volumes.

A functional MRI (fMRI) is able to estimate brain metabolism by using either blood-oxygen level dependant (BOLD) contrast or cerebral blood flow to measure activity in different regions of the brain. Resting-state fMRI (rsf-MRI) refers to an fMRI being performed while a patient is not performing any tasks (38). Limited studies have been published regarding functional MRI and its applicability in Neuro-PASC patients. Kiatkittikul et al. (29) showed abnormal brain connectivity on rsf-MRI which was concordant with 18 F-FDG-PET/CT findings. It was reported that one patient who had sensorineural hearing loss of the left ear as a consequence of COVID-19 showed hypometabolism of the left temporal region on 18 F-FDG PET/CT scans and anomalies in rsf-MRI in the same region (29). Another study using rsf-MRI, led by Churchill et al. (30), demonstrated that patients who are currently experiencing a larger number of Neuro-PASC symptoms tended to have altered connectivity between parietal, temporal, occipital and subcortical regions using BOLD rsf-MRI. Additionally, distinct patterns correlated to the intensity of PASC symptoms. They elude to the fact that this could be a useful tool to differentiate between Neuro-PASC and other non-COVID-related infections (30).

Arterial spin labeling (ASL) is an MRI technique that has been used to measure brain perfusion in Neuro-PASC patients. A study by Ajčević et al. (31) found that participants who still had cognitive impairment 2 to 10 months after the beginning of their acute COVID-19 infection had cerebral hypoperfusion patterns on ASL MRI imaging. The hypoperfusion was primarily found in the frontal cortex but was also present in the temporal and parietal cortex. These findings were more significant in the right hemisphere. They also found that overall, the cerebral blood flow was lower throughout the gray matter in Neuro-PASC patients compared to healthy controls (31). Yus et al. (28) found similar results in their ASL study on post-COVID patients where they found lower perfusion in the patient's orbital and medial frontal lobes.

Diffusion MRI (dMRI) is an imaging modality that utilizes the diffusion of water as contrast. Diffusion tensor imaging (DTI) is a type of dMRI that creates a 3D construction of diffusion (39). Diffusivity measurements are indicators of tissue microstructure integrity. Douaud et al. (23) found a greater increase in diffusivity, an indicator of tissue damage, in areas functionally connected to the piriform cortex, olfactory tubercle and anterior olfactory nucleus, in patients who were infected with SARS-CoV-2 compared to healthy controls (23). Additionally, a study by Campabadal et al. (27) found that Neuro-PASC patients presenting with olfactory dysfunction had higher mean and radial diffusivity in certain white matter regions on DTI compared to post-acute COVID-19 patients without olfactory dysfunction. The areas of increased mean diffusivity were found in the genu of the corpus callosum, forceps minor, orbitofrontal white matter tracts, and anterior thalamic radiations. Regions with augmented radial diffusivity were found in the genu of the corpus callosum, anterior corona radiata, and uncinate fasciculus (27). In contrast, a study done by Yus et al. (28) found no statistically significant findings with DTI—in patients with Neuro-PASC.

Susceptibility-weighted imaging was also used to try to visualize the neurological sequela of SARS-CoV-2 infection. This imaging modality is an MRI technique that uses the patient's iron, calcium, and deoxygenated blood as a source of contrast for the image. This imaging modality is useful for detecting hemorrhages, traumatic brain injury, microvasculature, and neurodegenerative diseases (40). Hellgren et al.'s (26) study using SWI found mostly subcortical abnormalities near the gray-white matter junction in frontal and parietal lobes in 8 patients out of 35 (26).

In addition, Mishra et al. (32) found regions of abnormal susceptibility bilaterally in the brain stem, the gray-white matter junction and in the white matter of the frontal lobes in patients with post-COVID condition or patients who had recovered after COVID-19 infection. The abnormalities seen in the brainstem were primarily in the midbrain and the abnormalities in the frontal lobes were located in the uncinate fasciculus tract and the inferior frontal-occipital fasciculus tracts (32). However, it is important to note that this article has not been peer-reviewed. To our knowledge, this is the only published article where the main focus was the use of susceptibility-weighted imaging to find radiological markers in post-COVID patients.