Advances in the application of neuroinflammatory molecular imaging in brain malignancies

The prevalence of brain cancer has been increasing in recent decades, posing significant healthcare challenges. The introduction of immunotherapies has brought forth notable diagnostic imaging challenges for brain tumors. The tumor microenvironment undergoes substantial changes in induced immunosuppression and immune responses following the development of primary brain tumor and brain metastasis, affecting the progression and metastasis of brain tumors. Consequently, effective and accurate neuroimaging techniques are necessary for clinical practice and monitoring. However, patients with brain tumors might experience radiation-induced necrosis or other neuroinflammation. Currently, positron emission tomography and various magnetic resonance imaging techniques play a crucial role in diagnosing and evaluating brain tumors. Nevertheless, differentiating between brain tumors and necrotic lesions or inflamed tissues remains a significant challenge in the clinical diagnosis of the advancements in immunotherapeutics and precision oncology have underscored the importance of clinically applicable imaging measures for diagnosing and monitoring neuroinflammation. This review summarizes recent advances in neuroimaging methods aimed at enhancing the specificity of brain tumor diagnosis and evaluating inflamed lesions.


Introduction
The most prevalent brain tumors comprise meningiomas, gliomas-particularly glioblastomas (GBMs)-and intracranial metastases originating from various cancers (1). The treatment approach for intracranial metastases involves a combination of chemoradiotherapy and neurosurgery (2). Meningiomas are predominantly benign and are typically managed by surgical resection (3). GBM represents the primary malignancy in the central nervous system (CNS) and is known for its aggressive nature, with limited benefits derived from advanced therapeutic strategies (4). Therefore, early diagnosis and accurate monitoring through neuroimaging are crucial for managing brain tumors. Neuroimaging is an indispensable aspect of clinical practice, offering valuable insights into brain tumors.
The advancement of computed tomography (CT) and magnetic resonance imaging (MRI) has enabled more precise diagnoses, improved clinical monitoring, and enhanced accuracy in prognostic prediction (1). In particular, neuroimaging has played a crucial role in diagnosing and treating brain tumors. By combining molecular pathology and histopathology, significant progress has been made in classifying various types of brain tumors. Precision neuro-oncology integrates tumor-specific neuroinflammation and distinct protein alterations (5). However, the translation of the precision oncology paradigm to neurooncology faces several significant challenges, including the criteria for assessing therapeutic effects and the need for accurate monitoring approaches. Advanced neuroimaging techniques currently offer potential solutions to overcome these challenges. The interactions between brain tumors and immune responses are critical focal points (6,7). Consequently, current immunotherapeutic strategies have been developed to target specific immune cells or inflammatory mediators within the tumor microenvironment (TME) (8). Neuroinflammation plays a crucial role in the progression of brain tumors. GBM, in particular, is characterized by tissue necrosis accompanied by heightened neuroinflammation. Additionally, immunosuppressive neuroinflammation and induced necrosis also contribute to treatment resistance and poor prognosis (5,9). However, immunotherapies also affect imaging phenotypes in clinical practice. Therefore, it is essential to gain a better understanding of the inflammatory factors and their associated neuroimaging characteristics.
In this review, the key aspects of neuroimaging in diagnosing and distinguishing brain tumors from inflammatory-associated lesions were emphasized. This information is immensely valuable for gaining a better understanding of imaging characteristics and common patterns that aid in diagnosis. Furthermore, a comprehensive understanding of the inflammatory mechanisms within the TME and their corresponding imaging features facilitates the development of noninvasive prognostic and predictive imaging strategies in clinical practice.

Current limitations in neuroimaging in the context of brain malignancy
The clinical diagnosis, evaluation, and monitoring of therapeutic effects in patients with brain malignancies heavily rely on neuroimaging techniques. There is a wide range of neuroimaging options available for clinical use. Structural MRI is a useful choice for identifying and classifying tumors and guiding surgical strategies. Additionally, PET imaging is predominantly used to assess tissue metabolism, providing valuable information about cancer cell proliferation and early-stage tumor detection (10). PET is a type of molecular imaging technique that is suitable for detecting specific molecules, such as choline, fluorodeoxyglucose, methionine, and phenylalanine (11,12). Thus, the critical mechanism of PET-based neuroimaging involves defining and tracing the targeting molecules that are specific or sensitive to immune cells, including microglia and macrophages.
The efficiency of current neuroimaging techniques is widely acknowledged; however, there are still several limitations requiring further improvement. For instance, distinguishing relapsed brain tumors, brain metastasis, or inflammatory and necrotic lesions poses a significant challenge in clinical treatment, as both tumors and metastatic sites exhibit similar intratumoral textures on MRI (13,14). The application of PET techniques also faces various challenges. The low permeability of biotracers across the bloodbrain barrier (BBB) and systemic plasma binding affect the imaging results. Additionally, inherent characteristics of the neuroimaging technique itself impose certain limitations. The most commonly used PET probe for neuroimaging, 2-deoxy-2-18F-fluoro-Dglucose, is based on glucose metabolism (15). Glucose transporters are highly expressed not only in tumor cells but also the inflamed areas (16). Importantly, glucose metabolism plays a crucial role in inflammatory responses, leading to false-positive results in tumor diagnosis. Therefore, in this review, the recent advances in neuroimaging approaches aimed at improving the evaluation of brain tumors and brain metastases combined with neuroinflammatory response information have been discussed. The first radiotracer used for neuroinflammation is based on an 18-kDa protein named translator protein (TSPO), a peripheral benzodiazepine receptor (17). TSPO is a mitochondrial transmembrane protein predominantly located in macrophages and microglia (18). Accumulative evidence indicated that TSPO up-regulation can be observed in pro-inflammatory and immunosuppressive conditions. Previous studies have shown that TSPO expression can be up-regulated in response to neuroinflammation and tumor malignancy (19). Additionally, TSPO can also be found in neoplastic glioma cells. Consequently, over 13 unique TSPO radiotracers have been developed for use in malignant brain pathologies, enhancing tumor-to-background brain signals (20). Furthermore, TSPO-based PET imaging shows promise in delineating clinically important neuroinflammatory tumor features (21). Therefore, the establishment of novel neuroimage techniques is urgently required for patients with brain tumors.

Brain tumors as neuroinflammatory diseases: tumor immune microenvironment
Neuroinflammation plays a crucial role in brain TME, resulting in carcinogenesis and tumor progression. In particular, primary brain malignancies, such as GBMs, are characterized by significant immunosuppressive neuroinflammation and necrosis, resulting in resistance to chemotherapy and poor clinical outcomes (22). Tumor cells produce and secrete various immunomodulatory substances, including interleukin (IL)-10, IL-1b, galectin-1, and transforming growth factor-b, which regulate the behavior of infiltrating immune cells and establish a pro-tumoral microenvironment (23,24). Among the components of the pro-tumoral microenvironment, tumor-associated macrophages (TAMs) are particularly prominent (6,25). Neuroinflammation induces immunosuppressive changes and further enhances the proliferation, migration, and therapeutic resistance of tumor cells (26).

Immune component of the TME in CNS malignancies
Under normal conditions, the CNS is protected by the BBB, which renders it immune-privileged and shields it from the influence of systemic inflammation. However, immune surveillance and inflammatory regulation in the CNS are mediated by resident immune cells and infiltrating peripheral cells. The microenvironment within brain tumors comprises a diverse array of cellular and immune components, which significantly affect the growth, drug resistance, and recurrence of primary and metastatic brain tumors (27,28).
First and foremost, it has been reported that astrocytes can be immediately infiltrated around brain tumors, exhibiting varying morphology and cellular functions (29). Active astrocytes, notably, have the ability to promote the invasive capacity and drug resistance of tumor cells, thereby enhancing tumor growth and survival, while suppressing the therapeutic efficiency in clinical treatment (30). Conversely, several studies have also reported the anti-tumor function of astrocytes. It has been demonstrated that specific astrocyte-derived exosomes comprising micro-ribonucleic acid (RNA) can significantly inhibit tumor growth (31). Additionally, microglia, macrophages, endothelial cells, and pericytes play key roles in tumor growth and neuroinflammation, potentially serving as clinical imaging features (32). Accumulative evidence illustrates the critical roles of microglia and macrophages in tumor malignancy, accounting for more than 30% of the cell populations (33). Microglia derived from the yolk sac precursor cells during embryonic development are resident phagocytic cells, while macrophages are derived from the bone marrowderived monocytes (BMDMs) and migrate into the brain parenchyma. Lineage-tracing experiments in brain metastases have demonstrated the infiltration of microglia and BMDMs in such malignancies. Collectively, microglia and macrophages are widely considered to be associated with tumor malignancy, referred to as TAMs (34).
The tumor vasculature primarily comprises endothelial cells, playing a critical role in tumor growth and facilitating immune reactions within the TME through the release of various neuroimmune substances, including IL-1b, IL-10, and granulocytemacrophage colony-stimulating factor (GM-CSF) (35,36). Additionally, pericytes are localized around the vascular vessels and are responsible for tumor vascularization and maintaining BBB integrity (37). The tumor vasculature comprises established vasculature and neo-vasculature, with the latter being significantly enhanced under the regulation of local hypoxia, metabolic demands, and vascular endothelial growth factor stimulation (38). Furthermore, the tumor vasculature exhibits abnormalities in integrity and function. Previous studies have shown a higher prevalence of dysfunctional vasculature within the tumor region, while the peritumoral vasculature resembles normal blood vessels. Following the intravenous administration of a gadolinium-based contrast agent (GBCA), the GBCA enters the extracellular space in the brain parenchyma through this abnormal vasculature, resulting in hyperintensity on T1-weighted magnetic resonance (MR) images (39). However, tumor cells might still be present in the nonenhancing regions around the areas of enhancement. Therefore, the presence or absence of tumor cells is not directly associated with the compromised BBB, increasing the difficulty of clinical diagnosis.

The interactions between the immune responses and brain tumors
The CNS possesses a unique immune monitoring system that involves meningeal or brain parenchymal lymphatic vessels and immune surveillance. TAMs can be classified into proinflammatory (M1) or anti-inflammatory (M2) types based on their cellular function and morphological features, which are associated with various processes such as the inflammatory cascade, immune activation, angiogenesis, tissue remodeling, and tumor survival. This classification framework allows for a better understanding of the multifaceted roles of TAMs (40). M1 TAMs are responsible for pathogen recognition and tumor killing. On the contrary, M2 TAMs primarily regulate inflammatory reactions and establish a favorable TME through the release of anti-inflammatory cytokines, thereby promoting tumor survival and growth (41).
Furthermore, the dysregulated adaptive immune reactions within the TME also contribute to tumor growth. The function and prognostic predictive value of adaptive immune cells in brain tumors have gained considerable attention. A previous clinical study has shown a correlation between prolonged survival in patients with GBM and the infiltration of the cluster of differentiation CD 8 + T lymphocytes, whereas regulatory T cells are associated with worse prognosis in patients (42,43). The adaptive immune responses are profoundly influenced by immune checkpoints, which regulate the immune response to self-antigens and contribute to tumor control. Accumulating evidence indicates that various tumors, including CNS malignancies and metastases, evade recognition and surveillance by the immune system through immune checkpoint regulation (44).
Evidence has revealed that therapeutic mobilization of innate and adaptive immune responses can be employed to induce immune-mediated tumor cell death. As a result, a wide array of immunotherapeutic strategies has been established, targeting the interaction between tumor cells and immune cells. These strategies include immune checkpoint inhibition, engineered chimeric antigen receptor T cells, oncolytic viral therapies, and vaccines (45,46). The underlying mechanisms of immune checkpoint inhibition and vaccines involve the activation of T lymphocytes by eliminating inhibitory signals. Additionally, tumor antigens are presented to antigen-presenting cells, such as dendritic cells (47). These cells are then administered to patients with GBMs after being stimulated with tumor antigens, triggering adaptive immune activation. Based on this approach, vaccine-based immunotherapies use tumor-associated antigens to enhance cytotoxic effects through the enhancement of immune reactions. Due to the critical role of immune components within the TME, neuro-inflammation-based imaging techniques may be a potentially effective approach in the diagnosis of patients with brain tumors.

Novel neuroimaging approaches for evaluating brain tumor recurrence
Despite the current efficiency of existing neuroimaging techniques in brain tumor detection, there is still a huge challenge in distinguishing recurrent brain tumors, solitary brain metastases and inflammatory and necrotic lesions (48). It has been revealed that conventional neuroradiological techniques struggled to distinguish recurrent tumors and inflammatory alterations, due to their similarly-showing imaging in MRI. Therefore, a better understanding of novel approaches to diagnose recurrent brain tumors may present great clinical potential in the follow-up of patients after anti-tumor therapies. With the rapid advances in neuroimaging, the direct labelling of critical immune cells has emerged as a potential option for neuroinflammation imaging with clinical relevance. Given that TAMs are considered the predominant immune cells within the brain TME, significant efforts have been made to label TAMs based on accumulating evidence. Furthermore, MRI plays an essential role in clinical assessment and monitoring, making novel approaches using MRI for neuroinflammation imaging likely to be applicable in clinical practice. In conventional MRI, GBCAs are commonly used to examine contrast dynamics and identify CNS malignancies. However, it is noteworthy that ultrasmall superparamagnetic iron oxide (USPIO) nanoparticles possess a unique capability to detect neuroinflammation by being taken up by immune cells (49). One such USPIO nanoparticle is ferumoxytol, which has been extensively studied in patients with brain tumors (49). Previous clinical studies have indicated that ferumoxytol exhibits a preferential accumulation in inflamed sites (50). Similarly, an experimental study revealed that ferumoxytol is predominantly observed in activated astrocytes, microglia, and macrophages but not in tumor cells. This underscores the potential of ferumoxytolenhanced MRI for neuroinflammation imaging in patients with brain tumors (51). Currently, ferumoxytol has received Food and Drug Administration approval for clinical use. In cases where the administration of GBCA is restricted due to clinical conditions such as GBCA allergy or renal failure, ferumoxytol is considered an alternative contrast agent (52). Additionally, ferumoxytol has demonstrated potential advantages over GBCA in the assessment of neuroinflammation. Its plasma half-life is approximately 14-21 h, allowing delayed localization within TAMs or the neovascular space (53). This prolonged intravascular signal persistence contributes to the observed parenchymal signal during delayed imaging (54). A prospective pilot study involving patients with gliomas revealed that delayed ferumoxytol imaging captures TAM signals within the TME. Furthermore, there is a positive correlation between susceptibility and relaxation and the number of macrophages (54).
The interpretation of T1-weighted ferumoxytol contrastenhanced MRI depends on clinical conditions (55). During the diagnostic phase, similar imaging characteristics can be observed in primary or metastatic brain tumors using either GBCA or ferumoxytol-enhanced MRI. However, after chemoradiotherapies, these enhancing images might differ. Specific immunotherapies can induce neuroinflammation, leading to gadolinium enhancement and T2 hyperintensity on MRI. Although the enhanced signals resemble those of solid tumors, they subsequently undergo spontaneous regression, known as pseudoprogression. Psuedoprogression can be observed in approximately 30% of patients with GBMs after chemotherapy or radiotherapy. Given the underlying mechanisms of immunotherapies, neuroinflammation induced by immunotherapies is more likely to be detected than that induced by chemoradiotherapies. Therefore, distinguishing pseudoprogression from actual tumor growth becomes crucial. Additionally, iron nanoparticle imaging requires specific attention. The prolonged clearance of iron nanoparticles should be given careful consideration in clinical practice (56). A previous study has revealed an increased level of methionine in metastatic tumor tissues (57,58). Furthermore, elevated levels of the neuroinflammatory marker PBR-TSPO can be observed in necrotic sites and quantified using a specific PET biotracer. Therefore, PET biotracers based on 11 Carbon (C)-methionine have been used in patients with metastatic brain tumors to accurately diagnose tumor recurrence (59,60). In one clinical trial, 11 C-methionine-based PET biotracers correctly diagnosed tumor recurrence, as confirmed by pathological examination, in seven patients. In comparison, 11 C-PBR28-based PET biotracers only identified three lesions, highlighting the reliability of 11 C-methionine-based biotracers for tumor recurrence detection. Thus, 11 C-methionine is considered a reliable marker for tumor recurrence compared with 11 C-PBR28 PET (61). PET imaging provides significant benefits in advanced photon techniques, particularly in treatment planning and the sparing of normal tissue for skull base meningioma using advanced photons and protons. Additionally, incorporating 68 Gallium-DOTATOC-PET information has a substantial impact on target volumes (62). Moreover, the combination of Multiparametric 18 F-FET PET/MRI improves the therapeutic effectiveness by distinguishing between tumor progression and therapy-related alterations (63). According to another clinical study, combined dynamic and static 18 F-FET PET/CT parameters can be used in differentiating radiation necrosis from recurrent tumor after cyberknife robotic radiosurgery (64). Effective imaging techniques are still required to be explored in future studies.

Neuroimaging approaches for evaluating brain tumors and radiationinduced necrosis
Currently, MRI (Table 1), PET, and single-photon emission CT (SPECT) are the predominant imaging techniques for the noninvasive assessment of neuroinflammation (87). Among them, SPECT can detect gamma rays emitted by radioactive isotopes used in clinical imaging. By employing various radiotracers, SPECT enables precise evaluation of metabolic and molecular processes, such as glucose utilization, nucleoside and amino acid transporter expression, and protein and deoxyribonucleic acid (DNA) synthesis (88). SPECT has been extensively used to investigate the molecular neurodegeneration mechanisms in drug addiction and to enhance therapeutic strategies with minimal adverse effects. Furthermore, SPECT plays a valuable role in neuroscience research, particularly in neurodegeneration and neuro-oncology (89). However, conventional neuroradiological techniques have several limitations. The increased contrast enhancement observed in MRI could lead to difficulty in distinguishing between radiation necrosis and tumor recurrence, as the lesions appear similar. Currently, radiotherapy and stereotactic radiosurgery are commonly used for treating brain metastasis. However, the incidence of radiation-induced necrosis poses a significant challenge in clinically differentiating between metastasis and radiation-induced necrosis. A previous study has demonstrated that conventional MRI has a specificity of 75% and a sensitivity of only 44% in distinguishing between tumors and inflamed necrotic tissue (90,91).
Neuroinflammation and necrosis are prominent side effects of radiotherapies for treating brain malignancies. A previous retrospective study reported that the rate of radiation-induced necrosis in patients with GBM ranged from 2.5% to 30% (92,93). This necrotic tissue disrupts the vasculature and perivascular parenchyma, resulting in an abundance of neuroinflammation. The compromised BBB facilitates peripheral immune cell infiltration and brain edema, thereby amplifying the inflammatory response initiated by activated and infiltrated immune cells. Currently, proton MR spectroscopic imaging is employed to assess cellular metabolism by detecting the distribution of proton metabolites, including creatine, lactate, lipid, and especially choline, which is particularly elevated in cell populations exhibiting enhanced cell proliferation. This advanced MRI technique allows for a detailed evaluation of tissue biochemical composition and blood perfusion, providing valuable information for distinguishing solid tumor tissues and necrotic regions. In a previous clinical trial involving 29 consecutive patients, the feasibility and utility of two-dimensional (2D) chemical shift imaging MR spectroscopy were investigated, demonstrating that increased ratios of choline content can differentiate brain tumors from necrotic tissues with an accuracy of up to 97% (66). The choline/N-acetylaspartate and choline/phosphocreatine ratios are particularly effective discriminators in necrotic lesions. However, it is important to note that a transient increase in choline levels can be observed in patients with radiation-induced necrosis, resulting in a false-positive diagnosis of tumor recurrence when using proton MR spectroscopy (68). Another study focusing on patients undergoing high-dose radiotherapy revealed that elevated lactate/creatine ratios and decreased metabolites are more commonly observed in patients with radiation necrosis than in those with recurrent GBM (65).
A retrospective study revealed significant differences in the mean, maximum and minimum relative peak height, and relative cerebral blood volume between patients with GBM and those, as detected by T2-weighted dynamic susceptibility-weighted contrast materialenhanced MRI (67). Furthermore, lower recovery values were observed in recurrent GBM compared with radiation necrosis. Additionally, a meta-analysis of 13 studies demonstrated that the detection of choline/N-acetylaspartate and choline/phosphocreatine ratios, along with relative cerebral blood volume (rCBV) using MR spectroscopy and MR perfusion, significantly improves accuracy in diagnosing primary or metastatic brain tumors (69).
PET scan, as a widely applied noninvasive neuroimaging approach, is useful for imaging neuroinflammation (Table 2). According to a previous study, PET scans exhibit a specificity of 69% and a sensitivity of 92% in differentiating between tumor recurrence and radiation necrosis, surpassing nuclear MR spectroscopy for choline/N-acetylaspartate and choline/creatine ratios at various thresholds (98). Another experimental study conducted on orthotopic GBM rat models demonstrated the excellent ability of PET with 18 Fluorine (F)-fluorodeoxyglucose and 18 F-fluoroethyltyrosine to distinguish primary GBMs from necrosis (96). Furthermore, a meta-analysis of six studies suggested that 11 C-choline PET is the most accurate diagnostic approach for distinguishing tumor relapse from necrosis in patients with glioma (97). Additionally, a retrospective study with long-term follow-up revealed that 11 C-choline PET/CT outperform MRI and 18 F-fluorodeoxyglucose in evaluating tumor recurrence or radiation-induced necrosis (94). Furthermore, a previous clinical study involving 50 patients demonstrated that 11 C-methionine PET outperforms 11 C-choline and 18 F-fluorodeoxyglucose PET in distinguishing primary GBMs from necrosis (95). Additionally, the L-type amino acid transporter 1 tumor-specific PET tracer, 18 F-2-fluoroethyl-L-phenylalanine, is also a superior option compared with 1 8 F-fluorodeoxyglucose PET in clinical differentiation, as it exhibits low sensitivity to neuroinflammation. However, the high rate of false-positive and false-negative results in PET scans remains a significant limitation in clinical practice. Therefore, there is an urgent need for improved neuroimaging techniques based on the aforementioned methods to achieve accurate diagnosis and clinical evaluation. Currently, advanced approaches offer substantial benefits in evaluating neuroinflammation through immune substance labelling, assessing BBB integrity via contrast agent leakage, and identifying inflammatory consequences combined with phenotypic imaging patterns and imaging genomics.
6 Neuroinflammatory molecularbased imaging strategies for brain metastases Brain metastasis is more prevalent than primary brain tumors, mainly due to the limited therapeutic efficacy against various primary cancer, such as breast, lung, and colorectal cancers. The prognosis for patients with brain metastasis is significantly compromised, underscoring the importance of early detection and accurate diagnosis. The TME plays a crucial role in the development of brain metastasis. Notably, the cancer stem cells (CSCs) are the predominant population involved in mediating metastasis, while neuroinflammation also plays a decisive role. Inflamed regions within the brain parenchyma facilitate the adhesion of peripheral tumor cells to activated endothelial cells, initiating invasion and metastasis. Tissue lesions caused by brain metastasis contribute to the establishment of neuroinflammation, characterized by persistent activation of astrocytes and microglia, increased production and release of pro-inflammatory substances, compromised BBB permeability, and infiltration of immune cells. Consequently, a diverse array of immune cells and inflammatory factors promote the progression of brain metastasis, exhibiting high heterogeneity depending on the origin of the primary malignancy and the specific brain site involved. Similar to GBMs, hypoxiainduced stimulation triggers pro-inflammatory substance expression in CSCs, further promoting a pro-inflammatory phenotype in GBM. These inflammatory factors contribute to tumor growth and metastasis in GBM (99,100).
Distinguishing between primary brain tumors and brain metastasis using neuroimaging remains a significant challenge in clinical diagnosis. Both primary and metastatic brain malignancies exhibit similar peritumoral hyperintensities and intratumoral texture on MRI. Previous studies have shown limited differences in apparent diffusion coefficient (ADC) measurements between primary brain tumors and brain metastasis. However, several studies have revealed that GBMs exhibit higher homogeneity and inverse difference moment than brain metastasis (79). Regarding peritumoral edema, MRI demonstrates greater heterogeneity of  (64) peritumoral edema in GBMs compared with that in metastasis, with high sensitivity and specificity of 80% and 90%, respectively (76). Additionally, the use of 2D texture features extracted from MRI images enables fast and noninvasive discrimination between GBM and brain metastases (82). Machine learning algorithms have gained significant attention in neuroimaging applications to improve the accuracy of clinical diagnosis. Quantitative analysis of MRI using machine learning and deep learning-based models facilitates the differentiation between primary and metastatic malignancies, emphasizing the significance of texture feature analysis (81,85). Furthermore, texture features extracted from post-contrast three-dimensional T1-weighted MR images, optimized by machine learning classifiers, have demonstrated high diagnostic performance and generalizability in differentiating solitary brain metastasis from GBM with high diagnosis performance and generalizability (83). In pediatric medulloblastoma (MB), radiogenomics combined with MRI-based machine learning offers an opportunity for MB risk stratification.
Studies have reported the beneficial use of MRI-based machine learning in identifying four clinically relevant molecular pediatric MB subgroups (86).
A previous study demonstrated that extracting texture features from post-contrast diffusion tensor imaging (DTI) MRI contributes to distinguishing between primary and metastatic brain tumors, offering high performance and generalizability in clinical diagnosis. Moreover, combining arterial spin labelling perfusion and DTI has shown significant clinical value (77). However, no differentiation in ADC values and ratios, as well as standard deviation values and ratios, was observed between GBMs and brain metastasis (84). In computational-aided quantitative analysis of MRI images (T2weighted/susceptibility-weighted/contrast-enhanced T1-weighted MRI), high accuracy was achieved in differentiating GBMs from metastases, emphasizing the significance of texture features rather than fractal-based features in clinical practice (75). Trained multiclass multilayer perceptron models using non-enhancing T2 hyperintense regions can differentiate glioblastoma, brain metastasis, and CNS lymphoma with modest diagnostic accuracy, resulting in an approximately 19% increase in diagnostic yield (78).
The integration of DTI, neurite orientation dispersion, intracellular or extracellular volume fraction, and metabolite analysis with neuroimaging techniques have promoted accurate clinical discrimination. Furthermore, the rCBV in the peritumoral brain zone (PBZ) distinguishes GBMs from metastases. Moreover, the CBV gradient or color map obtained from phase difference enhanced imaging in the PBZ also serves as an effective approach for distinguishing GBMs from metastasis. Significantly higher rCBV ratios and rCBV gradient were observed in the PBZ of GBMs compared with brain metastasis (80). Intratumoral proton MR spectroscopy reveals high levels of creatine in primary brain malignancies, particularly GBMs, whereas the absence of intratumoral creatine suggests the presence of metastatic brain malignancies (73). Definite lipid signals indicate tumor tissue necrosis, while the absence of lipid signals might rule out metastasis. Currently, the evaluation of peritumoral areas using color phase difference enhanced imaging, a novel phase-related MRI technique, also aids in the differentiation between GBM tumors and metastases (74). The prognosis for patients with brain metastasis is compromised, therefore, early detection and accurate diagnosis of brain metastasis are critical in clinical management.

Neuroinflammatory molecular imaging to distinguish CNS malignancies from intracranial infections
In neuroimaging, brain abscess manifests as expansile, rimenhancing masses surrounded by edema, which can resemble necrotic malignant tumors, particularly GBMs (101). Consequently, lesions caused by brain infections, particularly brain abscesses, are often misdiagnosed as brain tumors due to their similar MRI appearance and characteristics. However, brain abscesses and GBMs could cause nonspecific headaches in the absence of fever, focal neurologic deficits, and epileptic seizures (102). Therefore, the development of rapid and accurate diagnostic techniques is necessary to distinguish between brain abscesses and malignancies. Pathological examination reveals that the enhancing rim of GBMs comprises infiltrated tumor cells, whereas the enhancing rim of pyogenic abscesses comprises inflammatory components such as neutrophils, macrophages, and lymphocytes (103, 104). Therefore, the choline/creatine ratio of the rimenhancing lesion in abscesses is expected to be lower than that in GBMs. During MRI, the ADC and diffusion-weighted imaging provide valuable information for clinical diagnosis. Previous studies have reported that abscesses exhibit hyperintense signal changes on diffusion-weighted imaging, while GBMs demonstrate varying degrees of hyperintense to hypointense signal conversion. Significant differences have been observed in the choline/creatine, choline/N-acetylaspartate, and choline/choline-n ratios within the contrast-enhancing rim, allowing for differentiation between abscess and GBMs (71). Furthermore, combining dynamic susceptibility contrast perfusion-weighted imaging and DTI has shown improved efficacy in distinguishing inflammatory lesions compared with using a single neuroimaging technique. Research has indicated that choline levels in the ring-enhancing portion of abscesses are significantly lower compared with that of brain tumors (70). Moreover, a subsequent clinical study has demonstrated the significant role of MR spectroscopic imaging in discriminating abscesses from brain tumors. Metabolite ratios and maximum choline/choline-n, choline/creatine, and choline/Nacetylaspartate ratios within the contrast-enhancing rim could effectively differentiate abscesses from brain tumors (72).

Conclusion
Imaging genomics has emerged as a technique for evaluating neuroinflammation, involving the use of novel imaging biomarkers that capture DNA and RNA patterns associated with the biology or immune states of cancers. The imaging features are closely associated with gene expression patterns, mutations, and protein modifications (105). In the context of GBMs, imaging genomics has been employed, and several biomarkers have been established, such as the isocitrate dehydrogenase 1 mutation status and immunoreactivity. A radiogenomic profiling of 60 patients with GBM demonstrated positive correlations between CD68, CSF1 receptor, CD33, CD4, and CBV (106). Thus, imaging genomics holds the potential for bridging the gap between neuroimaging and tumor diagnosis in clinical practice.
In the era of immunotherapy and precision oncology, focusing solely on isolated imaging of brain tumor is not enough to establish predictive biomarkers and define neuroinflammation. Therefore, novel MRI and PET scans based on the tumor-associated neuroinflammation have attracted great attention. Secondary surgery towards recurrent brain tumors always accompanies elevating surgical risk and high therapeutic costs, therefore, it is important for accurate discrimination between tumor recurrence and radiation-induced necrosis. Multi-parametric MRI presents versatile imaging information and is considered an effective and useful imaging approach in clinical diagnosis. Accumulative evidence emphasized the clinical value of the apparent diffusion coefficient, volume transfer constant, and relative cerebral blood volume in the distinction of tumor cancer, radiation-induced necrosis, and other brain diseases in daily neuro-oncological practice. PET scans present a unique function in determining tumor microenvironment, assessing drug delivery, and evaluating therapeutic effects. Although the application of PET scans presents great advantages in clinical use, high economic costs and restricted devices limit its generalization. The development of neuroimaging and the combination of novel MRI contrast agents and PET radiotracers and imaging genomic techniques enable the evaluation of neuroinflammatory components and the improvement of accurate diagnosis and clinical discrimination. Of note, there are still several limitations requiring further improvement in neuroinflammation-based neuroimaging to minimize the false-positive diagnosis of tumor recurrence and misdiagnosis with necrosis or intracranial infection. In conclusion, the efforts on such noninvasive neuroinflammation imaging towards accurate diagnosis and personalized therapeutic efficacy monitoring will help the establishment of precision oncology strategies for patients with brain tumors or other malignancy.

Author contributions
WD and ZQ designed the study; WD wrote the draft; ZQ and NW revised the manuscript. All authors contributed to the article and approved the submitted version.

Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.