The Role of Advanced MR Imaging in Gliomas

Abstract

Gliomas are a significant health problem with a lot of imaging challenges. The role of imaging is no longer limited to only providing anatomic details, but with the advancement of Magnetic Resonance Imaging (MRI) techniques, it now permits the assessment of the freedom of water molecule movement, the microvascular structure, the hemodynamic characteristics, and the chemical makeup of certain metabolites of lesions. These advanced imaging techniques include diffusion-weighted imaging, diffusion tensor imaging, dynamic contrast-enhanced MRI, Magnetic Resonance (MR) perfusion, MR angiography, and magnetic resonance spectroscopy. their role in the diagnosis, classification, and post-treatment follow-up of gliomas, as well as their application in radiogenomics and glioma analysis with the aid of artificial intelligence, is presented and discussed.

1. Introduction

The broad category of gliomas constitutes 24.5% of all primary brain and other central nervous system tumors and 80.9% of all malignant tumors. Glioblastoma (World Health Organization [WHO] grade 4) represents 14.3% of all tumors and 49.1% of malignant tumors [1]. The WHO classification systems of CNS tumors were published in 1979, 2000, 2007, 2016, and the last revised system was published in 2021 [2]. The grading of gliomas mainly relies on histological features, including cellularity, nuclear atypia, mitotic activity, vascularity, and necrosis, observed on light microscopy with the aid of immunohistochemistry. Recently, advancements in genetics and molecular knowledge of gliomas have shown the value of improving the correlation between diagnosis and prognosis; for example, mutations of the gene encoding for isocitrate dehydrogenase 1 (IDH1) are very common in low-grade gliomas but rare in glioblastoma. According to the latest 2021 WHO classification of gliomas, there are a total of six families of tumors: adult-type diffuse glioma, pediatric-type high-grade diffuse glioma, pediatric-type low-grade diffuse glioma, circumscribed astrocytic glioma, glioneuronal and neuronal tumors, and ependymal tumors. The types of tumors present in each tumor family have been categorized based on their molecular and genetic characteristics as well as their biological behavior. Adult-type diffuse gliomas include astrocytoma (grades 2, 3, and 4), IDH-mutant oligodendroglioma (grade 2 or 3), IDH-mutant and 1p/19q-codeleted gliomas, and IDH-wild-type gliomas (grade 4/Glioblastoma (GBM) [2].

Imaging plays an integral role in intracranial tumor diagnosis and management. Magnetic resonance imaging (MRI) is the modality of choice in the evaluation of intracranial tumors. The role of MRI in the workup of brain tumors can be broadly divided into tumor diagnosis and classification, treatment planning, and post-treatment follow-up. In recent years, apart from tumor recurrence and radiation necrosis, new post-treatment entities such as pseudoprogression and pseudoresponse have been recognized during glioma treatment [3,4].

Conventional and various advanced MR imaging techniques are currently used in clinical practice, with the latter offering more than the anatomic information provided by the conventional MRI sequences by generating microstructural tumor data and information on chemical composition. The current and more commonly used advanced techniques include diffusion-weighted imaging (DWI), along with measurement of the apparent diffusion coefficient (ADC), diffusion tensor imaging (DTI), dynamic contrast enhancement (DCE), dynamic susceptibility contrast (DSEI) perfusion, arterial spin labeling (ASL) perfusion, and magnetic resonance spectroscopy (MRS) [3,4].

For the purposes of this review, we searched the PubMed database for studies published on advanced MR imaging techniques for the diagnosis of gliomas, focusing on the role of the most commonly used advanced MRI techniques. We evaluated the utility of each technique for glioma classification, clinical decision-making, differentiation between post-treatment-related changes on follow-up, prediction of glioma molecular profile, and overall prognosis.

Literature Search Strategy

A narrative literature search was performed using the PubMed database to identify studies on advanced MRI techniques in glioma imaging. The focus was placed on diffusion, perfusion, metabolic, and emerging quantitative MRI techniques with demonstrated or potential clinical relevance in glioma diagnosis, treatment planning, and follow-up.

2. Basic Principles of Conventional and Advanced MR Imaging

2.1. Conventional MRI Techniques

The most common conventional techniques used for the assessment of intracranial tumors include T1-weighted (T1W), T2-weighted (T2W), fluid-attenuated inversion recovery (FLAIR), T2*-weighted gradient echo (T2*W), and T1W after contrast administration sequences [5]. These sequences offer valuable anatomic detail, and the intravenous contrast agent administration of gadolinium permits the detection of the tumor areas with blood–brain barrier damage. Using conventional imaging, we can detect the size and location of a lesion and evaluate the presence of edema, necrosis, hemorrhage, tumor heterogeneity, and cystic components, as well as the presence of mass effect. Nevertheless, many important limitations of brain tumor imaging persist. For example, it is difficult to differentiate glioma grades (low-grade versus high-grade tumors, or oligodendroglioma versus astrocytoma). Moreover, the appearance of treatment-related changes cannot differentiate residual or recurrent tumors and radiation necrosis [6]. Gliomas are difficult to distinguish from metastasis, abscess, and tumefactive multiple sclerosis [7]. These limitations are important to diagnosis, treatment approaches, and patient prognosis. Usually, high-grade gliomas (HGGs) show hypointense signals on T1W and hyperintense signals on T2W images, heterogeneous contrast enhancement, necrosis, hemorrhage, and edema of surrounding brain [5]. A common feature of HGGs is contrast enhancement, but it remains a nonspecific finding, because nearly a third of HGGs do not show enhancement, while almost 50% of low-grade oligodendrogliomas show some enhancement [1]. The presence of enhancement is a common feature of glioblastoma (GBM) (WHO grade 4), but a lot of other pathologies have a similar appearance, such as lymphoma, abscess, and tumefactive multiple sclerosis. Lymphoma and GBM are highly infiltrative tumors and can cross into the contralateral hemisphere via the corpus callosum [7]. According to the updated 2016 World Health Organization guidelines, oligodendrogliomas are genetically characterized by the presence of the isocitrate dehydrogenase (IDH) mutation and 1p/19q codeletion. Tumors with histologic features of oligodendroglioma and mutated IDH, but without 1p/q19 codeletion, are classified as astrocytomas [2]. Unfortunately, there is overlapping of conventional MR imaging findings between oligodendroglioma and astrocytoma. Oligodendrogliomas have been described as near the cortex and containing calcifications, but these features are rarely present, so its diagnosis is difficult [8]. Although conventional structural MR imaging lacks the capability to absolutely establish tumor grade and histologic subtypes of gliomas, in the case of IDH-mutant astrocytomas, the diagnostic T2/FLAIR mismatch sign is a validated feature that can aid in the differential diagnosis of these gliomas. This sign is observed when there is a hyperintense signal on T2-weighted imaging with complete suppression of the signal on FLAIR imaging. The hyperintense signal seen on T2-weighted imaging appears to correspond to edema, demyelination, and other neurodegenerative changes [9,10,11]. While conventional MRI is excellent in providing detailed structural evaluation of brain tumors (Figure 1), advanced MR techniques (Figure 2, Figure 3, Figure 4 and Figure 5) give important information about tumor cellularity, infiltration, neoangiogenesis, treatment response, and a better way to understand the underlying tumor biology [8]. The most commonly used advanced MR techniques include diffusion-weighted imaging (DWI), diffusion tensor imaging (DTI), perfusion-weighted imaging (PWI), and MRS [3].

2.2. Diffusion-Weighted and Diffusion Tensor Imaging

DWI and DTI are pulse sequences sensitized to the random movement of water molecules within tissues. Water molecules that remain stationary between gradient applications are refocused and contribute to signal generation, whereas moving molecules lose phase coherence, resulting in signal attenuation. Areas of restricted diffusion appear with a high signal on a DWI map and a low signal on an ADC map [12,13].

In DWI, the b-value is used to determine the sensitivity of the sequence to the diffusion of water molecules. Low b-values (0–1000 s/mm²) detect relatively free water movement, while very high b-values (2000–3000 s/mm²) highlight areas where diffusion is restricted. For the calculation of ADC values in brain tumors, b-values of 0 and 1000 s/mm² are typically used [14]. Moreover, the number of diffusion directions used for tract visualization should be greater than 6, and ideally between 15 and 32, to ensure reliable and anatomically accurate visualization of white matter tracts. Increasing the number of directions enhances the precision of anisotropic diffusion measurements, which is critical for evaluating fiber orientation, connectivity, and structural changes.

ADC measurements reflect tumor microstructure, and several tumor characteristics may influence ADC values. Cystic and necrotic areas demonstrate increased water mobility and high ADC values, whereas solid tumor regions with increased cellularity restrict extracellular diffusion and exhibit low ADC values [4]. The microstructure of brain tissue may give various directions in ADC, known as anisotropic or as isotropic diffusion [13]. The directional variations in ADC in three-dimensional space are assessed from diffusion data acquired by using a great number of multiple-gradient directions. Since diffusion is more dominant parallel to rather than transverse to myelinated nerve fibers, DTI allows for the imaging of white matter tracts, and the degree of diffusional anisotropy may be shown as color-coded, three-dimensional directional maps of nerve fibers named tractographies [5].

According to a meta-analysis involving 414 patients, DWI can be useful in distinguishing tumor progression from treatment-related abnormalities. Specifically, the mean ADC value in patients with tumor progression was calculated to be 1.13 × 10⁻³ mm²/s, while in those with treatment-related abnormalities, it was 1.38 × 10⁻³ mm²/s, and this difference was found to be statistically significant [14]. Another group of researchers reported that lower ADC values are more associated with high-grade gliomas and poorer prognosis, highlighting the potential of DWI in the differential diagnosis of gliomas [13]. Some researchers have also noted that higher b-values, such as 3000 s/mm², can help in the differentiation between HGGs and LGGs in DWI by increasing contrast and reducing the T2 shine-through effect because HGGs appear as more conspicuous hyperintense areas, while LGGs often demonstrate hypointensity in high-b-value DWI [15,16].

DTI tractography reveals the relationship between tumors and major white matter tracts. DTI is useful for detecting the invasive growth patterns of high-grade tumors such as GBM and gives information for surgical planning [17]. Other hopeful applications of diffusion imaging, such as diffusion kurtosis imaging and functional diffusion maps, are under investigation and may provide additional insights into complex brain tumor tissue environments [18,19,20,21]. Functional diffusion maps utilize DWI and ADC to compare the ADC values before and after the treatment of tumors to evaluate the early response to treatment by performing a voxel-by-voxel analysis of changes in ADC values [22]. A high volume of tissue exhibiting decreased ADC has been associated with lower survival rates and earlier disease recurrence compared to patients with a lower volume of tissue showing ADC reduction. This finding is attributed to the fact that a large decrease in ADC likely reflects an increase in tumor cell density, indicative of ineffective treatment [23].

Despite their clinical significance and the capabilities that they offer, diffusion-based techniques have certain limitations. First, ADC values vary depending on the clinical characteristics of each patient, the magnetic field, and the protocol followed for acquiring the sequences during the examination. For example, each hospital uses different protocols and may choose to calculate ADC values based on higher b-values, which could result in lower ADC values compared to those derived using the standard b-value of 1000 s/mm² [14]. Additionally, there can often be artifacts that hinder the quality and accuracy of the image, and many times, even on the same magnetic resonance imaging machine with the exact same protocol, the ACD values differ. For this reason, although DWI may play some role in the diagnosis of brain tumors, it does not seem able to be utilized for their accurate and safe differential diagnosis [13,14]. Therefore, while DWI and DTI provide valuable complementary information, they cannot be used as standalone tools for the accurate and safe differential diagnosis of gliomas, and should be interpreted in conjunction with other advanced MRI techniques and clinical data [13,14].

2.3. Perfusion Imaging

Brain tumor growth patterns are characterized by the delivery of oxygen (O₂) and nutrients and removal of waste via diffusion processes. If this method is insufficient to support cell mitotic activity and continued tumor growth, stress from hypoxia and hypoglycemia produces factors that initiate the development of neoangiogenesis [24]. Tumors with neoangiogenesis exhibit irregular arteriovenous shunts, multiple blind-ending vessels, areas of hypoperfusion, and necrosis. Advanced MR perfusion techniques enable the non-invasive assessment and quantification of these microvascular abnormalities, thereby providing insight into glioma biology.

Clinically, perfusion imaging contributes to tumor grading, treatment planning, and evaluation of therapeutic response [25]. Examples of procedures of perfusion imaging include dynamic susceptibility contrast (DSC), DCE imaging with intravenous injection of gadolinium-based contrast agents, and arterial spin labeling (ASL), which uses magnetic labeling of endogenous protons in blood to assess blood flow volume without contrast injection [26,27].

Despite their clinical utility, perfusion techniques have important limitations. Susceptibility artifacts may degrade image quality, particularly in lesions containing blood products or located near bone–air interfaces. In addition, all contrast-based techniques require intravenous contrast administration, limiting their applicability in patients with renal impairment or contrast allergy, whereas ASL, although non-contrast, remains less widely available and standardized [28,29].

2.4. Dynamic Susceptibility Contrast Perfusion

DSC perfusion MRI is the most widely used perfusion technique in the evaluation of brain tumors. A series of dynamic T2*W images are acquired during the rapid intravenous injection of a gadolinium-based contrast agent. As it passes through the brain, the contrast produces a relative decrease in signal intensity, reflecting the concentration of intravascular contrast. Analysis of these signal changes enables the generation of parametric maps of cerebral blood volume (CBV), cerebral blood flow (CBF), and mean transit time (MTT). DSC calculations are based on intravascular contrast. In brain tumors, areas of blood–brain barrier breakdown result in leaks of contrast which produce a T1 contrast effect leading to falsely low measurements of CBV. This T1 shine-through effect can be minimized by using algorithm-based leakage correction methods [25]. There is widespread use of DSC in clinical settings, along with proposals for standardized techniques in order to estimate tumor neoangiogenesis [27]. Ongoing efforts toward protocol standardization and harmonized post-processing aim to improve reproducibility and facilitate its integration into multiparametric MRI frameworks for glioma assessment.

2.5. Dynamic Contrast-Enhanced Perfusion

DCE uses dynamic T1W acquisition during contrast administration to measure changes in signal intensity from intravascular and extravascular contrast leakage from blood–brain barrier damage. This technique gives us a highly accurate determination of whether the glioma is high or low grade, provides moderate accuracy in detecting recurrence or moderate treatment-related changes, and offers a better visualization of the heterogeneity of brain lesions, making it a useful tool for diagnosis [25]. From a biological standpoint, DCE-derived parameters reflect alterations in tumor microvascular integrity and permeability, which are closely associated with glioma grade and aggressiveness. Compared with susceptibility-based techniques, DCE-MRI is less affected by magnetic susceptibility artifacts, making it particularly useful for evaluating gliomas containing intratumoral blood products or post-operative hemorrhage [26].

Clinical applications of DCE use semi-quantitative and quantitative analytic methods, usually attempting to quantify the degree of change in signal intensity relative to baseline, producing three main quantitative analytic imaging biomarkers: estimates of the vascular fraction (vp), extravascular extracellular space fraction (ve), and the transfer contrast coefficient (K^trans) [27]. These kinetic parameters can provide correlations with the microvascular environment and can be useful in identifying blood–brain barrier abnormalities, aiding in the differential diagnosis of gliomas and determining their prognosis [28]. Despite its diagnostic potential, the widespread clinical implementation of DCE-MRI remains limited by relatively long acquisition times and increased demands on post-processing, which may restrict its routine use in busy clinical settings [30].

2.6. Arterial Spin Labeling

ASL differs from other methods of MR perfusion imaging because it does not depend on contrast administration but on magnetic labeling of blood flowing into tissues of interest. This non-invasive approach allows for relatively accurate absolute measurement of cerebral blood flow (CBF), along with timing parameters that reflect the velocity of blood delivery to the tissue [31]. In a review series on advanced MRI techniques in patients with gliomas [31,32], arterial spin labeling is referred as a contrast-free technique for measuring cerebral blood flow, with attention to labeling approaches, temporal parameters, and signal-to-noise constraints, and is positioned as a useful complementary option in patients who cannot receive contrast agents. The review also explores proton MR spectroscopy, emphasizing practical aspects such as voxel selection and spectral reliability, as well as the diagnostic value of metabolite ratios, most notably choline-to-N-acetylaspartate and lipid/lactate peaks, for tumor characterization, evaluation of infiltrative disease, and distinction between recurrence and therapy-related changes. In addition, emerging techniques, including chemical exchange saturation transfer imaging and quantitative relaxometry, are presented as promising tools for probing tumor microenvironmental and molecular features. ASL can be useful in the diagnosis and classification of gliomas, as well as in the planning of surgical interventions or biopsy procedures. From a pathophysiological perspective, ASL-derived CBF values provide information about tumor vascularity and metabolic activity [33]. Clinically, ASL has shown utility in glioma diagnosis and grading, as well as in surgical and biopsy planning. This technique allows us to detect the area of the tumor with the highest blood flow, corresponding to the metabolic center, thus enabling the surgeon to obtain a more accurate histopathological diagnosis by targeting the biopsy to that point.

Additionally, the significantly increased CBF in gliomas can aid in differentiating them from metastatic lesions, non-cancerous lesions, and primary central nervous system lymphomas. Also, between HGGs and LGGs, it has generally been observed that HGGs have a high signal in ASL, while the latter have a lower signal than that of the average perfusion. Finally, it has been observed that in patients with GBM, ASL can be useful in predicting their prognosis, as it has been noted that high tumor signal intensity on ASL has been correlated with lower survival rates and shorter progression-free periods.

Although ASL is currently less widely used than contrast-based perfusion techniques such as DCE, its non-invasive nature and ability to provide quantitative perfusion metrics suggest a promising role in future multiparametric MRI protocols for glioma assessment [31,33].

2.7. MR Spectroscopy

The signals on MRS are generated from free protons, which are most plentiful in water. Furthermore, the MRI signal is also affected, but to a lesser degree, by protons bound to macromolecules, which are found in low concentrations in biologic tissues. These bound protons have specific small frequency peaks in ppm (parts per millions), which allows separation when acquiring data, each representing specific macromolecular component if sufficient suppression of the water signal can be achieved. MRS is a sensitive technique for detecting such components in low concentrations [34]. There are single and multivoxel MRS techniques for detecting metabolite concentrations. Metabolites detected by MRS include N-acetyl aspartate (NAA), choline (Cho), creatine (Cr), lactate, and lipids. The peak of NAA means neuronal integrity; choline is a marker of membrane turnover, while creatine is a marker of energy metabolism and internal cell control. Lactate is a product of non-oxidative glycolysis, seen in necrosis and hypoxia, and lipids are a marker of cellular membrane necrosis. All these metabolites give the biochemical profile of brain tumors [35].

Clinically, MRS contributes to the differentiation between neoplastic and non-neoplastic lesions, as well as between residual or recurrent tumor and radiation-induced necrosis. It can also be used to evaluate treatment response by comparing pre- and post-treatment metabolic spectra [36]. It has been shown that when the spectroscopy shows a Cho/NAA ratio > 2 and high peaks of lipids and lactate, the tumor is high grade. However, the differentiation between a primary brain lesion and a metastatic one remains unclear in the literature, and opinions are divided.

A major disadvantage of this technique is the absence of a specific spectrum for each tumor type, making it difficult to provide an accurate diagnosis, as spectra from different brain pathologies often overlap. Additionally, susceptibility artifacts may degrade spectral quality in tumors with hemorrhagic components or in lesions located near bone or at the skull base, restricting accurate evaluation [34]. The improvement of the parameters required for MRS can help in the future to optimize the analysis and reduce the artifacts, increasing the diagnostic accuracy of the technique.

2.8. MR Angiography

MR angiography (MRA) is a non-invasive technique for imaging blood vessels, and is divided into dark-blood and bright-blood techniques. The dark-blood technique is utilized for specific clinical questions, primarily for imaging stenoses or thromboses in the arterial and venous networks, respectively, as well as for imaging the heart and the course of the vessels. In contrast, the bright-blood techniques have a broader application, and among them, the most widespread is time-of-flight. (TOF). TOF can provide imaging of vascular structures without the administration of a paramagnetic substance, utilizing the signal enhancement relative to the background resulting from blood flow. Gradient echo sequences are commonly employed to minimize signal loss related to blood flow and improve vessel conspicuity [37].

The use of MRA for the differential diagnosis and histological differentiation of gliomas is not widely disseminated. Nevertheless, some researchers argue that MRA can be utilized for the identification of pathological neovascularization that is often associated with malignant processes [36]. It was found that the use of TOF-MRA can be employed for the characterization of blood vessels around the GBM area, as it can differentiate the tumor’s neo-vessels from the vessels of normal brain tissue [32,38]. The use of a 7 Tesla (7T) magnetic field in vitro and in vivo in mouse glioma models can help in the separation of pathological from normal vessels [39]. Similar results were presented by Radbruch et al., who used TOF-MRA in the same magnetic field in patients with GBM and demonstrated that it is possible to visualize and differentiate pathological arteries from those of the normal parenchyma [40]. Furthermore, some researchers suggest that using algorithms to measure the angles of vessels around the tumor also has the potential to define the tumor’s margins [32].

MRA cannot provide the information offered by conventional MRI and thus cannot currently be used for the analysis and histological distinction of gliomas. The use of higher magnetic fields such as 7T seems to enhance the capabilities of MRA and may help in the evaluation of gliomas [32,41]. Although ultra-high-field imaging appears to enhance the sensitivity of MRA for detecting tumor-related vascular changes, its high cost, limited availability, and primarily research-oriented use preclude widespread clinical implementation [32]. Consequently, at present, MRA plays a limited and largely investigational role in the imaging assessment of glioma

2.9. Chemical Exchange Saturation Transfer (CEST) MRI

Multipool CEST MRI is a novel technique that enables the quantification of metabolites present at low concentrations by exploiting exchangeable protons within those target molecules (e.g., cellular proteins). To generate the contrast, a radiofrequency saturation pulse is applied to saturate the protons of the target molecules. These saturated protons then exchange with unsaturated water protons, which leads to a reduction in the signal. The reduction in the water signal in MRI can be measured, and the difference between images acquired with and without saturation provides information on the concentration of the target metabolite or protein. Amide proton transfer (APT) is a subcategory of CEST that specifically targets amide groups and allows quantification of protein concentration in tissues [42]. A characteristic clinical application of APT in glioma imaging is the determination of IDH mutation status. The APT signal measurements can reliably distinguish IDH-wild-type from IDH-mutant gliomas and can further differentiate alpha-thalassemia/mental retardation syndrome X-linked (ATRX) loss versus retention in grade II and III IDH-mutant tumors [43]. This distinction reflects the fact that IDH-wild-type gliomas have higher APT signal than IDH-mutant gliomas [43,44]. Moreover, APT signal has been shown to correlate significantly with the Ki-67 index [43]. CEST MRI has also been investigated for the determination of MGMT promoter methylation status, due to ATP’s higher signal in gliomas with unmethylated MGMT, enabling molecular subtyping of gliomas [43]. Finally, CEST MRI can differentiate diffuse gliomas with Lys-27-Met mutations in histone 3 genes H3K27M, a mutation which is associated with poorer prognosis and shorter survival even before surgical intervention [45]. Overall, multipool CEST MRI is a highly promising technique for both grading and molecular classification of gliomas.

2.10. T1 Mapping

The quantitative T1 (qT1) mapping technique enables the creation of absolute qT1 maps, which contain the exact relaxation time values (T1) for each voxel. This allows for the accurate measurement of changes before and after the administration of paramagnetic contrast agents [46]. This technique has been employed by several research groups for evaluating blood–brain barrier (BBB) disruptions associated with tumor infiltration, as well as for predicting tumor prognosis and detecting early recurrence [46,47]. Müller et al. reported that T1 maps can reveal a “cloudy”-enhancing area located around the central tumor mass, which likely corresponds to infiltrative cancer cells extending into the surrounding tissue in patients with glioblastoma. The researchers assessed patients before and after treatment and found that a reduction of approximately 21.64% in the cloudy-enhancing region on qT1 maps was associated with a longer progression-free survival. T1 mapping also enables the distinction between necrotic tissue and tumor-associated edema, helping differentiate vasogenic edema from tumor infiltration, due to the different signal characteristics observed on T1 maps. In a preliminary study, Cao et al. demonstrated that the longitudinal relaxation time in the rotating frame (T1ρ) measured using T1 mapping can differentiate between LGGs and HGGs. HGGs showed higher T1ρ values both in the peritumoral edema and in the solid tumor area. Moreover, they found that T1ρ values in peritumoral edema and tumor location in the frontal lobe were significantly associated with the presence of IDH1 mutation [48].

All in all, T1 mapping is a quantitative MRI technique that can be effectively utilized in the imaging of gliomas, offering valuable information not available through conventional MRI sequences. However, it is not yet fully validated for all tumor types and requires both hardware and software that may not be available in all imaging centers. Moreover, it depends on patient cooperation, as the need of multiple images at different inversion or saturation times extends the duration of the examination.

2.11. Intravoxel Incoherent Motion (IVIM)

IVIM is an MRI technique that utilizes endogenous tracers to calculate perfusion by using DWI databases and multiple b-values without the need for a contrast agent. This technique allows for simultaneous assessment of diffusion and tissue microcirculation [49]. Through specialized software, it measures three parameters: the diffusion coefficient (D), the pseudodiffusion coefficient (D*), and the perfusion fraction (f), and generates corresponding maps. The values of these parameters have been utilized to evaluate patient response to treatment, predict prognosis, and for glioma classification [50,51].

In patients with HGGs, the f values were used to differentiate between residual or recurrent tumors and post-treatment changes and were compared with rCBV values from DSC images. Researchers showed a positive correlation between the two values, indicating that IVIM can be used as an alternative technique [52]. Puig et al. showed that the f and D* values in the contrast-enhancing region in newly diagnosed GBM patients could predict six-month survival. They suggested cut-off values of 9.86% for f and 21.712 × 10⁻³/s for D*, highlighting that these IVIM parameters could serve as future biomarkers for patient prognosis [50]. Gu et al. applied IVIM parameters in patients with gliomas and found that those with HGG and IDH-wild-type gliomas had higher D* values compared to patients with LGGs and IDH-mutant gliomas, respectively. They also discovered that the D and f values were significantly higher in patients with IDH-mutant gliomas [51].

IVIM is a promising technique; however, its use in routine clinical practice is not yet established, as it requires significant examination time, specialized software, and the results of the parameters vary depending on the b-values and the strength of the magnetic field. As a result, establishing specific cut-off values is not yet reliable.

3. Advanced MR Imaging of Gliomas Classification

3.1. Glioma Grade Classification

The non-invasive differentiation of LGG from HGG is of paramount importance for proper patient management, and several advanced MRI techniques have been evaluated towards this scope. Both perfusion and diffusion imaging can predict glioma grade with high accuracy. High-grade tumors exhibit increased relative cerebral blood volume (rCBV) values and a cut-off value of 0.63 has been reported to have a 100% sensitivity and a 94.4% specificity [4]. The ADC index outperforms FA values [51]. In a recent study, 72 patients with gliomas, 27 with low-grade and 45 with high-grade, were studied using synthetic magnetic resonance imaging (sy-MRI), three-dimensional pseudo-continuous arterial spin labeling (pCASL), and DWI on a 3.0 T MR scanner; the results showed that there was a significant difference between LGGs and HGGs on T1, proton density (PD), CBF, ADC, enhancement quality (EQ), and proportion enhancing (PE). Among all previous parameters, the ADC values had the highest discrimination ability. The best performance for the differentiation of low- from high-grade gliomas was the combination of T1, PD, CBF, and ADC, with a sensitivity of 95.5% and a specificity of 100% [53]. In another recent study, Li et al. evaluated the potential of diffusional variance decomposition (DIVIDE) for the assessment of glioma grades, molecular features, and microstructural characterization. The authors evaluated several parametric maps such as fractional anisotropy (FA), mean diffusivity (MD), anisotropic mean kurtosis (MKA), isotropic mean kurtosis (MKI), total mean kurtosis (MKT), MKA/MKT, and microscopic fractional anisotropy (μFA). The MKT best distinguished low- from high-grade gliomas (AUC 0.86) and FA best discriminated IDH-wild-type from mutated tumors (AUC 0.881). Apart from FA, all other diffusion metrics were significantly correlated with the Ki-67 proliferation labeling index [54].

MRS permits the quantification of tumor metabolites non-invasively by analyzing their spectra. The most frequent measured metabolites include N-acetyl aspartate (NAA), choline (Cho), creatine (Cr), myo-inositol (mIns), lactate, lipids, and certain amino acids. Nevertheless, in a meta-analysis that included thirty articles and a total of 1228 patients, MRS demonstrated moderate diagnostic performance in distinguishing LGGs from HGGs. Among the various metabolites and ratios, the Cho/NAA ratio exhibited the highest sensitivity and specificity compared to the Cho/Cr or NAA/Cr ratio [55].

3.2. Correlation with Glioma Molecular Profile

Since genetic testing requires a tissue sample for analysis, a non-invasive technique that could provide a diagnosis would be of great importance, especially for tumors for which resection is not recommended. For example, IDH mutations have been associated with longer survival compared to IDH-wild-type tumors for both GBMs and LGGS. The median survival of IDH-wild-type GMB is 15 months compared to 31 months for IDH-mutant tumors. Similarly, IDH-wild-type anaplastic astrocytomas have a median survival of 20 months, whereas the IDH-mutant gliomas have a median survival of 65 months [56].

Prediction of isocitrate dehydrogenase 1 (IDH1) genotype and 1p/19q codeletion status is important in gliomas for the assessment of individual prognosis and response to therapy. Various advanced MRI techniques have been evaluated towards this scope. In a study of 215 patients with various glioma grades, the diffusion metrics of diffusion tensor imaging, diffusion kurtosis imaging, neurite orientation dispersion, density imaging, and mean apparent propagator were evaluated, and histogram analysis was performed in both the areas of the entire tumor and peritumoral edema. The results showed that all these parameters had moderate-to-good diagnostic performance for the assessment of IDH1 and 1p/19q codeletion status [57].

The development of reproducible MRI parameters that could predict molecular subtype and risk stratification in glioma would be of paramount importance. A recent study correlated the contrast enhancement pattern, presence of necrosis, tumor margins, edema, T2/FLAIR mismatch, internal cyst formation, and CBV higher than normal cortex with five risk groups: oligodendroglioma IDH-mutant and 1p19q codeleted; diffuse astrocytoma grade—2–3 IDH-mutant; diffuse astrocytoma grade—2–3 IDH-wild-type; GBM IDH-mutant; and GBM IDH-wild-type. The highest diagnostic performance was reported for GBM IDH-wild-type with predominant enhancement, followed by diffuse astrocytoma IDH-mutant with T2/FLAIR mismatch sign and no necrosis. On the contrary, GBM IDH-mutant and diffuse astrocytoma IDH-wild-type were difficult to distinguish [58].

The O6-methylguanine-DNA methyltransferase (MGMT) gene, located on chromosome 10q26.3, encodes for a DNA repair enzyme that counteracts the lethal effects of alkylating chemotherapeutic drugs. Assessment of the MGMT status in gliomas is of therapeutic importance. The methylated MGMT promoter is correlated with better overall survival in gliomas treated with temozolomide. Furthermore, MGMT promoter methylation status has been associated with glioma pseudoprogression being more common in cases of methylated MGMT promoter [59]. In a recent meta-analysis, the diagnostic performance of radiomics using machine learning algorithms for prediction of MGMT promoter status was performed. The results extrapolated from fifteen studies showed that the pooled sensitivity and specificity of machine learning were 85% and 84% in the training cohort and 84% and 78% in the validation cohort [60].

The phosphatase and tensin homolog located on chromosome 10 (PTEN) is a common tumor-suppressor gene and plays a critical role in cell cycle regulation, angiogenesis, migration, invasion, and stem cell regulation in gliomas [61]. The mutation status is associated with disease prognosis and response to therapy. A study that included 244 patients investigated the combination of both deep convolutional neural network (CNN) and radiomic features extracted from MRI for the pretreatment prediction of the PTEN mutation status in glioma. The results showed that combining radiomics with CNN had the highest accuracy (86%). The CNN model alone had an accuracy of 81% and radiomics alone had an accuracy of 66% [62].

Diffuse midline glioma, H3K27-altered (DMG-A), is a specific entity that constitutes the most frequent of diffuse intrinsic pontine gliomas. Similar tumors can be found elsewhere in the midline, such as the thalamus and brainstem. These are aggressive tumors with unfavorable prognosis and, according to the WHO 2021 classification, are considered as grade 4 tumors regardless of histological features. In a recent study, the DSC, ADC, and conventional MRI findings were compared between DMG-A and H3K27-wild-type counterparts (DMG-W). Normalized rCBV (nrCBF) and normalized corrected rCBV (ncrCBV) values were significantly lower in DMG-A compared with DMG-W. On diffusion imaging, the normalized maximum ADC (nADC_max) was significantly higher in DMG-A compared to DMG-W. The combination of at least two parameters among nrCBF < 1.11, nADC_max ≥ 2.48, and positive T2-FLAIR mismatch sign exhibited the highest diagnostic performance [59].

4. Tumor Infiltration

Advanced MRI techniques can detect tumor infiltration in patients with gliomas due to their ability to evaluate perfusion, metabolism, and microstructural changes in both the tumor and peritumoral areas [63]. DTI can visualize white matter tracts and identify which of them are infiltrated based on reduced fractional anisotropy (FA) and elevated mean diffusivity (MD) in affected tracts [64]. However, white matter tract infiltration has not been consistently correlated with the risk of recurrence in glioma patients [63]. IVIM parameters can be used to assess peritumoral edema and identify regions of cellular infiltration, due to the elevation of f and D* values in infiltrated tissues [49]. The BBB disruption and peritumoral infiltration can also be detected using T1 mapping, which can highlight cloudy-enhancing regions corresponding to potentially infiltrated tumor cells [46,47]. MRS is also widely used to detect infiltrated areas. A reduction in NAA and/or increase in Cho in patients with HGG has been associated with higher probability of tumor infiltration and the presence of glioblastoma [65]. Croteau et al. reported a positive correlation between Cho/nCho and Cho/NAA ratios and MIB-1, a proliferation index related to tumor aggressiveness [66]. It is worth mentioning that the Cho/NAA ratio has shown the strongest association with tumor infiltration and recurrence. Perfusion sequences such as DSE, DSC, and ALS can also be used to detect infiltration. In DSC imaging, elevated rCBV indicates areas of high microvascular density and tumor invasion. Hu et al. pointed out that increased rCBV may predict the likelihood of tumor progression in patients with HGG [67]. Moreover, in HGGs, rCBV has been positively associated with the MIB-1 proliferation index. Tang et al., in their review, argued that rCBV may predict tumor infiltration with greater accuracy than other metabolic biomarkers, including MRS [63]. ALS can also be used to estimate CBV without the use of contrast agents. High CBV values measured with ALS have been associated with the presence of tumor infiltration in HGG. Furthermore, the kinetic parameters of DCE can detect BBB disruption and increased vascular permeability in the peritumoral region, which may correspond to areas of infiltrating tumor cells [63]. Finally, CEST MRI, and particularly APT imaging, can detect tumor infiltration by identifying elevated signals in regions with high concentrations of mobile tumor proteins [42,43].

5. Post-Therapeutic Surveillance

5.1. Radiation Necrosis Versus Tumor Recurrence

External-beam radiotherapy is the main adjuvant treatment after surgical resection of HGG [68]. Nevertheless, radiotherapy has several side-effects that can be divided into acute, early, and late/delayed. Radiation necrosis usually occurs 6 months to 2 years after radiation, although there have also been cases recorded after more than 5 years. In about one third of cases, radiation necrosis produced a mass effect resembling tumor recurrence. Differentiation of recurrent tumors from radiation necrosis constitutes a challenge and definite diagnosis can be made after stereotactic biopsy or repeat excision [3,4]. A recent meta-analysis of 17 high-quality studies assessed the diagnostic accuracy of DWI in differentiating glioma recurrence from post-treatment-related changes. The results exhibited a sensitivity of 82%, specificity of 83%, positive likelihood ratio of 4.9, negative likelihood ratio of 0.21, and diagnostic odds ratio of 23 [69]. Another study including 70 glioblastoma patients with suspected tumor recurrence compared the ADC, volume transfer constant (K^trans), and rCBV. The overall diagnostic accuracy was 85.8% for rCBV, 75.5% for K^trans, and 71.3% for ADC values. When combining rCBV (with a cut-off value of 2.2) with K^trans (with a cut-off value of 0.08 min⁻¹), the overall diagnostic accuracy reached 92.8% [70].

5.2. Pseudoprogression

During the radiological follow-up of patients with glioma, new or increasing areas of contrast enhancement may occur within the tumor resection bed or in its vicinity, suggesting the presence of progressive disease. However, in nearly one fourth of patients, this enhancement may represent treatment-related changes or an entity characterized as pseudoprogression. Pseudoprogression has been reported to occur in nearly two thirds of cases within the first 3 months after completing treatment [4]. This entity has been reported to occur due to tissue overresponse to treatment and involves endothelium disruption, blood–brain barrier (BBB) breakdown, and oligodendroglial injury [71]. The O6-methylguanine-DNA methyltransferase (MGMT)-methylated tumors have been predominantly associated with pseudoprogression. A recent study comprising 34 post-treatment HGG patients evaluated the added value of diffusion kurtosis imaging (DKI) in combination with DSC for the detection of pseudoprogression. The best diagnostic accuracy was 88.2%, corresponding to the combination of relative mean kurtosis (rMK) with rCBV. The diagnostic accuracy of each parameter was 82.4% for rMK, 70.6% for relative axial kurtosis (rKa), 82.4% for rCBV, and 73.5% for relative mean transit time (rMTT) [72]. Another recent meta-analysis of 24 studies with 900 glioblastoma patients compared the diagnostic performance of diffusion and perfusion imaging in differentiating pseudoprogression from true progression. The highest sensitivity and specificity were found for DWI (88% [confidence interval {CI} = 83–92%] and 85% [CI = 78–91%], respectively). Perfusion imaging had a sensitivity of 85% (CI = 81–89%) and a specificity of 79% (CI = 74–84%). Nonetheless, no significant difference was demonstrated between the imaging techniques [73].

5.3. Pseudoresponse

Bevacizumab, a humanized monoclonal antibody against VEGF, has been used for glioma treatment in several trials. By inhibiting angiogenesis, this agent minimizes the disruption of BBB, thus leading to decreased contrast enhancement on T1-weighted MRI and tumor edema on fluid-attenuated inversion recovery (FLAIR) MRI [74,75]. Although anti-VEGF agents may increase progression-free survival, the overall survival exhibits only a modest increase, with several patients developing a non-enhancing diffuse infiltrative tumor progression [74]. Ratai et al. evaluated 13 patients with recurrent GBM on anti-VEGF treatment in combination with either temozolomide or irinotecan. NAA/Cho ratio was found to be associated with treatment response and could serve as a biomarker for the distinction between response and pseudoresponse [76].

6. Predicting Prognosis

The non-invasive identification of prognostic variables in glioma patients is of paramount importance and could enhance prognostic accuracy, detect patients who are in need of closer follow-up or could benefit from aggressive treatments, and aid in selection of more homogeneous experimental populations in clinical trials [77]. By using diffusion kurtosis imaging in 33 glioblastoma patients, Li et al. found that the mean kurtosis value in the contrast-enhanced gross tumor volume before radiation therapy was significantly prognostic of overall survival. This parameter also remained significant in the multivariate analysis after adjusting for age, MGMT methylation status, and extent of resection [78].

The myo-inositol-to-total-choline (Ins/Cho) ratio on MR spectroscopy has been reported to correlate with glioblastoma, harboring the wild-type IDH, progression-free survival (PFS), and overall survival. The PFS and overall survival was significantly longer for patients with low (<0.9) Ins/Cho ratio. A total of 15 of the 16 patients with an Ins/Cho ratio higher than 0.9 experienced tumor recurrence within 12 months, whereas only 6 of the 11 patients with a low Ins/Cho ratio had recurrence within the same period [79].

7. The New Era of Advanced MR Imaging

Advanced MR imaging can be a very useful tool for the diagnosis and analysis of gliomas with the help of radiomics and artificial intelligence (AI). Radiomics is a continuously evolving field of medicine that involves the computational processing of radiological images with the aim of extracting quantified data that can be utilized for the creation of databases to improve diagnosis and prognosis prediction. For this purpose, multiple parameters from a single area of interest are analyzed and combined either automatically or with the help of AI to generate personalized insights for each patient. The use of MRI is essential for radiomics because through morphological, textural, and functional characteristics extracted from its sequences along with genomic data, we can achieve a more accurate diagnosis, predict the response to treatment, and determine the prognosis for each patient. In particular, ADC values and perfusion MR techniques provide useful data for the creation of automatic diagnostic models, which are very useful for the creation of databases [80,81].

Machine learning and deep learning are the two categories of AI that have been most widely utilized for brain tumor diagnosis and have applications in the radiomics research of gliomas [82]. These techniques utilize the data and features of radiological images obtained using MRI and process them for the purpose of creating databases. Subsequently, these databases allow for the automated identification of the tumor, comparison with genomics data to determine the type of tumor, and the combination of imaging findings with clinical data to predict prognosis and treatment response [81,82]. The combined use of advanced MR imaging techniques, radiomics, and AI seems to be able to help in the future with the better management and care of patients with gliomas. In a recent review article, Jin et al. described the growing role of machine learning and deep learning techniques in automated tumor segmentation and prediction of radiomic features. Among machine learning techniques, convolutional neural networks have been used more extensively. These ML-based algorithms and techniques allow comparison of spatial and textural patterns that are difficult to quantify visually. Despite this, the application of ML and deep learning methods may not currently be directly applicable to clinical practice due to large heterogeneity in imaging acquisition protocols and differences in the workflow. Identification of these limitations may allow for their use in clinical practice in the near future [82].

Table 1. Overview of advanced MRI techniques and AI–radiomics on gliomas and their clinical relevance.

Technique	Tumor Feature	Main Clinical Value	Limitations
DWI/ADC	Cellularity Necrosis/cystic changes	Grading classification Response OR recurrence vs. treatment-related changes	Protocol/scanner variability Overlap → not standalone
DTI/tractography	White matter tract integrity Infiltration patterns	Surgical planning Tract involvement	Sensitive to edema/crossing fibers Processing variability
DSC perfusion	Neoangiogenesis Microvascular density (rCBV/rCBF)	Grading classification Recurrence vs. necrosis/pseudoprogression Biopsy targeting	Susceptibility artifacts Requires contrast
DCE perfusion	Permeability/BBB disruption (Ktrans, vp, ve)	Heterogeneity Grading classification Recurrence	Longer acquisition Model dependence Requires contrast
ASL	Quantitative perfusion (CBF) without contrast	Useful when contrast contraindicated Grading classification	Lower SNR Less standardized/available
MRS	Metabolism (Cho/NAA, lactate/lipids)	Grading classification Recurrent Tumor vs. non-tumor	Spectral overlap Artifacts Expertise and time needed
CEST (APT)	Mobile proteins Emerging molecular surrogates	Emerging: grading and molecular profiling	Limited validation Technical complexity
Radiomics/AI	Imaging phenotype/heterogeneity linked to genomics	Molecular prediction Risk stratification (emerging)	Reproducibility and external validation required

summarizes the major advanced MRI techniques discussed, linking technical outputs to tumor biology and clinical needs, while highlighting key limitations that currently restrict broader clinical adoption.

8. Future Directions

The integration of advanced MRI techniques into routine clinical decision making for patients with gliomas is widespread but remains neither fully standardized nor universally predefined. Important unmet clinical needs persist, including reliable non-invasive tumor grading, accurate differentiation between treatment-related changes and true tumor progression, standardized assessment of tumor infiltration, and the development of robust imaging biomarkers for molecular characterization and prognostic evaluation.

A major limitation across advanced MRI techniques is the lack of standardized acquisition protocols and universally accepted quantitative thresholds, which hinders reproducibility and limits multicenter implementation. In addition, many promising techniques remain confined to research settings due to prolonged acquisition times, complex post-processing requirements, and limited external validation.

Future research should focus on the development of standardized multiparametric MRI protocols that integrate diffusion, perfusion, metabolic imaging, and emerging molecular imaging biomarkers. Within this framework, radiomics and artificial intelligence are expected to play a pivotal role, as they are increasingly incorporated into clinical practice. The extraction of large numbers of quantitative imaging features from multiparametric MRI sequences, combined with machine learning and deep learning algorithms, enables the identification of complex imaging patterns that are not discernible through conventional visual assessment. At the same time, the growing integration of artificial intelligence tools into radiology workflows is expected to enhance diagnostic accuracy, reduce interobserver variability, and facilitate the translation of advanced MRI techniques from research applications into routine clinical use.

Ultimately, prospective multicenter studies and coordinated harmonization efforts are required to transform advanced MRI techniques, together with radiomics and artificial intelligence, from research tools into clinically actionable biomarkers for modern, personalized glioma management.

9. Conclusions

Advanced MRI provides several important non-invasive insights for glioma management, including the assessment of glioma grade, treatment-induced effects, molecular tumor status, and prognosis. In particular, the integration of advanced MRI techniques such as diffusion, perfusion, and spectroscopy allows more detailed characterization of cellular density and more accurate differentiation between true tumor progression and treatment-related changes. From a neurosurgical perspective, these techniques support more optimal surgical planning for both biopsy and tumor resection and enable more accurate delineation of tumor margins, including identification of white matter tracts, eloquent regions, and tumor areas with increased mitotic activity. For diagnostic specialists, combining diffusion, perfusion, and spectroscopy findings increases diagnostic confidence, allows more reliable longitudinal follow-up, and supports clearer communication within the multidisciplinary team. As imaging methods become more standardized and artificial intelligence tools are gradually introduced into everyday practice, advanced MRI is expected to play an increasingly important role in guiding more individualized care for patients with gliomas.