Cerebral Gliomas and Glial Tumours — MRI Interpretation

Cerebral Gliomas and Glial Tumours — MRI Interpretation Deep Dive

MRIninja Knowledge Base | Focus Deep Dive Page

Related master pages: MRI Brain Generic Standard Protocol · DSC Perfusion · RANO 2.0 · SWI · DWI/ADC

Version 1.0 — May 2026

1. Scope and Clinical Context

This deep dive covers the multiparametric MRI interpretation of cerebral gliomas and related glial tumours across the full WHO CNS Classification 2021 spectrum [1]. It assumes familiarity with the general brain MRI protocol (see MRIninja Brain Generic Standard Protocol) and builds directly on the perfusion (DSC/DCE), spectroscopy, SWI, and DWI sequence pages.

The WHO CNS 2021 classification [1] fundamentally reorganised glioma diagnosis by mandating integrated molecular-pathological diagnosis. Imaging alone cannot assign a WHO grade; MRI provides the spatial and functional phenotype that — in combination with IDH mutation status, 1p/19q codeletion, TERT promoter mutation, EGFR amplification, CDKN2A/B homozygous deletion, and MGMT methylation status — constitutes the complete diagnostic characterisation. This integration of imaging and molecular data is the operational context in which all MRI interpretation of gliomas must occur.

The seven tumour entities addressed on this page:

1. Astrocytoma, IDH-mutant (WHO grade 2/3/4)

2. Oligodendroglioma, IDH-mutant, 1p/19q-codeleted (WHO grade 2/3)

3. Glioblastoma, IDH-wildtype (WHO grade 4)

4. Diffuse midline glioma, H3 K27-altered (WHO grade 4)

5. Diffuse hemispheric glioma, H3 G34-mutant (WHO grade 4)

6. Paediatric-type low-grade gliomas (pilocytic astrocytoma, BRAF-altered)

7. Diffuse low-grade glioma, MAPK pathway-altered

2. WHO CNS 2021 Classification — Imaging Implications

The 2021 WHO classification [1] introduced several changes with direct MRI implications that differ from prior editions:

Astrocytoma, IDH-mutant: now a single entity covering WHO grades 2, 3, and 4. The grade 4 astrocytoma IDH-mutant (previously GBM with IDH mutation) is distinguished from GBM IDH-wildtype by the presence of CDKN2A/B homozygous deletion. On imaging, IDH-mutant grade 4 astrocytoma shares ring enhancement with GBM but tends to have a more compact, better-circumscribed lesion without the extensive satellite spread of GBM IDH-wildtype. This distinction is clinically important for prognosis but cannot be reliably made on imaging alone.

Glioblastoma, IDH-wildtype: now requires IDH-wildtype status AND at least one of: TERT promoter mutation, EGFR amplification, or chromosome 7 gain/10 loss. Many lesions previously classified as grade 2 or 3 gliomas are now re-classified as GBM based on molecular features even without the typical imaging of GBM. This means a "low-grade looking" MRI can represent GBM IDH-wildtype when molecular features are present — a critical radiological awareness point.

Diffuse midline glioma, H3 K27-altered: located in the midline structures (pons, thalamus, spinal cord, other midline). Previously called DIPG (diffuse intrinsic pontine glioma) in the paediatric literature; the 2021 classification incorporates all H3 K27-mutant diffuse midline tumours regardless of age.

No more "secondary GBM": the concept has been eliminated. The new nomenclature distinguishes astrocytoma IDH-mutant grade 4 from GBM IDH-wildtype, which are biologically distinct entities despite both being grade 4.

3. Standard MRI Protocol for Glioma Assessment

The minimum multiparametric protocol for glioma assessment follows the BTIP (Brain Tumour Imaging Protocol) and RANO 2.0 framework [2, 3]. It extends the standard brain protocol with mandatory functional sequences.

RANO 2.0 mandate for 3D isotropic post-contrast T1: RANO 2.0 [2] requires 3D isotropic T1 post-contrast (MPRAGE/BRAVO/TFE at ≤ 1 mm isotropic, TE/TI standard) for all GBM baseline and follow-up imaging to enable volumetric enhancement measurement and reliable subtraction. 2D post-contrast T1 is no longer acceptable for GBM treatment monitoring at trial-participating centres.

4. Sequence-by-Sequence Glioma Analysis

4.1 T1-Weighted Pre-Contrast (3D IR-GRE)

Normal brain: white matter is T1-bright, grey matter is T1-intermediate, CSF is T1-dark. Any deviation from this pattern in the tumour region is diagnostically informative.

T1-hypointense signal in the tumour region (most gliomas): reflects increased free water from oedema, cellular infiltration replacing normal white matter, or necrosis. Low-grade gliomas are typically T1-hypointense without mass effect-proportional bright signal.

T1-hyperintense signal within a glioma (without contrast): strongly suggests haemorrhage (methaemoglobin in subacute haemorrhage), calcification (mild T1 shortening in some calcified tumours), or proteinaceous material within cysts. This pre-contrast T1 hyperintensity must be documented before contrast injection because it will simulate or mask enhancement on post-contrast images. Subtraction imaging (post minus pre T1) is mandatory when pre-contrast T1 signal is elevated.

T1-hypointense cavities: necrosis within high-grade gliomas; central necrosis of GBM. Distinguished from cystic components by irregular wall, internal debris signal, and enhancement pattern on post-contrast.

Haemorrhagic signal stratification by MRI stage (see also SWI section 4.5):

• Hyperacute (< 6h): T1-isointense, T2-bright (oxyHb)
• Acute (6h–3d): T1-isointense to hypointense, T2-dark (deoxyHb, intracellular)
• Early subacute (3d–7d): T1-bright (extracellular metHb developing peripherally)
• Late subacute (7d–30d): T1-bright throughout, T2-bright (extracellular metHb)
• Chronic: T1-dark rim, T2-dark rim (haemosiderin)

4.2 T2-Weighted TSE

T2 is the primary sequence for tumour and infiltration extent assessment. All glial tumours produce T2 hyperintensity — the degree and character of this hyperintensity varies diagnostically:

T2 bright, well-circumscribed, without significant mass effect: characteristic of low-grade glioma (astrocytoma IDH-mutant grade 2, oligodendroglioma). The "T2-FLAIR mismatch sign" (described below in FLAIR section) is the key low-grade astrocytoma imaging marker.

T2 bright, poorly defined, infiltrative: the radiological definition of diffuse glioma. The tumour cells spread along white matter tracts, producing T2 signal increase that extends beyond visible mass. This infiltrative margin on T2 is the surgical boundary challenge — the visible T2 edge consistently underestimates true tumour cell infiltration [4].

T2 bright with central dark signal: necrosis — the centre of GBM. On T2, necrosis appears heterogeneously bright with dark debris or blood products. The characteristic GBM appearance is a thick, irregular enhancing ring surrounding a T2-bright necrotic core.

T2-hypointense signal within a glioma:

• Haemorrhage (deoxyHb, haemosiderin): dark on T2
• Calcification: can appear T2-hypointense (blooming on GRE/SWI)
• Dense cellularity in some high-grade components
• Melanin or protein-rich material (rare in primary gliomas)

Peritumoral T2 hyperintensity: the T2 bright region surrounding the enhancing tumour. In GBM, this zone combines vasogenic oedema AND tumour cell infiltration — these cannot be reliably distinguished on T2 alone. In lower-grade tumours, the T2 zone is predominantly the tumour itself. In metastases, the T2 zone is predominantly vasogenic oedema with minimal tumour infiltration — this difference in infiltration pattern is a useful differential diagnosis criterion (see Section 7).

Cortical T2 signal: diffuse cortical T2 hyperintensity suggests cortical involvement by infiltrating tumour, encephalitis (differential), or cerebrovascular disease. In glioma context, cortical T2 signal extending into the subarachnoid space suggests leptomeningeal spread.

4.3 FLAIR (3D)

FLAIR suppresses CSF signal, making periventricular, juxtacortical, and cortical tumour infiltration more conspicuous than on T2. For glioma assessment, FLAIR provides three critical additional pieces of information:

Extent mapping: FLAIR is more sensitive than T2 for detecting the peripheral infiltrative margin of diffuse gliomas because it eliminates CSF signal competition at the cortical surface and in the sulci. In high-grade gliomas, the FLAIR extent frequently underestimates true tumour infiltration but provides the surgical planning boundary in most centres.

The T2-FLAIR mismatch sign [5]: a T2-hyperintense lesion that appears relatively FLAIR-hypointense (i.e., the lesion is bright on T2 but the FLAIR signal within the lesion is substantially lower than the surrounding T2 hyperintensity) is a relatively specific sign for IDH-mutant astrocytoma. The pathophysiological basis is the homogeneous myxoid matrix of these tumours, which produces very long T1 (approaching the FLAIR null point) and very long T2. Reported sensitivity 27–74%, specificity 89–96% for IDH-mutant non-codeleted astrocytoma vs. other gliomas [5, 6]. The sign is most reliable for pure astrocytoma IDH-mutant and is less reliable for oligodendroglioma (which typically shows full FLAIR positivity due to calcium and heterogeneity).

Subependymal / periventricular extension: FLAIR demonstrates spread along the ependymal surface and subependymal layer, which T2 alone may miss due to adjacent CSF brightness.

Post-contrast FLAIR timing: post-contrast 3D FLAIR acquired at 3–5 minutes after injection (before significant CSF gadolinium accumulation) demonstrates leptomeningeal and dural enhancement with high sensitivity. For glioma staging, post-contrast FLAIR complements post-contrast T1 for: (a) leptomeningeal spread detection; (b) cortical spreading; (c) small enhancement foci near the sulcal surface. See the FLAIR sequence page for the post-contrast timing artefact that must be avoided.

4.4 DWI and ADC Map

DWI reflects tissue cellularity (restricted diffusion = high cellularity) and is the primary sequence for necrosis characterisation and treatment response monitoring.

ADC in glioma grading:

Lower ADC values correlate with higher cellularity and higher tumour grade. Published mean ADC values (at b=1000 s/mm²):

• GBM core: 0.9–1.1 × 10⁻³ mm²/s
• High-grade glioma (grade 3): 1.0–1.3 × 10⁻³ mm²/s
• Low-grade glioma (grade 2): 1.3–1.7 × 10⁻³ mm²/s
• Necrosis: often high ADC (> 1.5 × 10⁻³ mm²/s) due to absence of intact cells
• Normal WM: 0.7–0.8 × 10⁻³ mm²/s

These ranges overlap substantially between grades — ADC alone is not a reliable grading tool. The ADC value must be interpreted in context of the multiparametric profile.

DWI for treatment response monitoring:

After treatment, ADC changes precede T2/FLAIR and contrast enhancement changes. The functional diffusion map (fDM) — comparing ADC voxelwise between baseline and follow-up — identifies regions of increasing restriction (recurrence) vs. increasing ADC (treatment effect/necrosis) weeks before conventional imaging changes.

ADC cutoff 1.36 × 10⁻³ mm²/s for tumour progression vs. pseudoprogression: per the RANO 2.0 ancillary criteria [2] (supported by Cho et al. [7]), an ADC below this threshold in the T2-hyperintense peritumoral region suggests tumour infiltration rather than pure treatment effect. This cutoff is not a standalone diagnostic criterion but contributes to the integrated assessment.

DWI for abscess vs. necrotic tumour: this is the most important practical DWI application in the glioma differential. Pyogenic brain abscess produces intense restricted diffusion (very low ADC, ~0.3–0.5 × 10⁻³ mm²/s) due to the high viscosity of pus. Necrotic glioma has elevated ADC. This distinction is clinically urgent. Caveats: (1) treated abscesses and some non-pyogenic infections have higher ADC; (2) some GBM may show focal restriction in necrotic areas due to debris.

Diffuse midline glioma DWI: H3 K27-altered midline gliomas frequently demonstrate restricted diffusion corresponding to densely cellular regions. In the pons, DWI restriction in a DIPG-equivalent lesion indicates high-grade (WHO grade 4 behaviour) and correlates with poor prognosis.

4.5 SWI and Phase Image

SWI is mandatory in the glioma protocol for four specific diagnostic contributions:

Intratumoral haemorrhage detection: SWI is the most sensitive sequence for haemosiderin and deoxyhaemoglobin within gliomas. Dark foci on SWI within a T2-bright glioma indicate prior haemorrhage and suggest higher grade (GBM haemorrhages more frequently than low-grade gliomas). SWI dark foci in the tumour periphery (not in the core) suggest haemorrhagic satellites — a sign of GBM's aggressive spread.

Tumour-associated susceptibility signal (TVASN): the network of small dark dots within or at the periphery of the enhancing tumour on SWI corresponds to: (a) deoxyhaemoglobin in densely packed tumour neovessels; (b) haemosiderin from prior microhaemorrhages; (c) mineralisation. TVASN correlates with the SWI-visible microvasculature density and has been proposed as an indirect marker of angiogenesis and higher grade [8].

Calcification detection: oligodendrogliomas calcify in approximately 70–90% of cases. On SWI, calcifications appear dark on the magnitude image and bright on the phase image (diamagnetic susceptibility). This phase brightness of calcification (opposite sign to haemosiderin) allows imaging-based distinction between calcium and iron — a capability not available on T2 or CT MRI. For oligodendroglioma characterisation, the combination of (a) T2 bright/FLAIR positive; (b) cortical/subcortical location; (c) SWI phase-bright (diamond = calcium) is diagnostically important. Always verify phase sign convention on your scanner (see SWI sequence page).

Central vein sign (CVS) in differential diagnosis: the CVS — a dark vein traversing the centre of a T2/FLAIR hyperintense lesion on SWI — is a specific marker for demyelinating lesions (MS) rather than glioma. A lesion with CVS is unlikely to be a glioma. When glioma is the differential diagnosis in a patient with multiple brain lesions, CVS presence in ≥ 3 lesions strongly supports MS over GBM or lymphoma. (See SWI sequence page and MS child protocol for CVS threshold criteria.)

4.6 Post-Contrast T1 (3D IR-GRE/MPRAGE)

Post-contrast imaging is the primary tool for tumour grading and treatment response monitoring. Enhancement indicates blood-brain barrier (BBB) disruption — not inherent to tumour grade but strongly correlated with it.

Enhancement patterns by entity:

Glioblastoma ring enhancement morphology: the RANO-defining enhancing component is measured on the post-contrast T1 (axial plane, two perpendicular diameters). The ring is characteristically:

• Thick (> 5–8 mm wall thickness)
• Irregular inner margin
• Variable outer margin (indistinct on T2/FLAIR)
• Incomplete (the "open ring" sign is not pathognomonic for MS — GBM can also show incomplete ring)

Enhancement subtraction: when pre-contrast T1 signal is elevated (methaemoglobin in haemorrhagic GBM, treatment-related protein deposition, gadolinium deposition from prior injections), the apparent enhancement on post-contrast T1 is ambiguous. Subtraction (post minus pre T1) reveals true enhancement by cancelling pre-existing T1-bright signal. This is mandatory in the follow-up GBM protocol. The RANO 2.0 framework [2] explicitly requires subtraction imaging when baseline T1 signal is elevated.

No-enhancement does not mean low-grade: in the WHO 2021 framework, a non-enhancing tumour with IDH-wildtype molecular features is GBM, not a low-grade glioma. The absence of enhancement at diagnosis in an IDH-wildtype tumour (which occurs in approximately 20% of newly diagnosed GBMs) does not preclude aggressive biology and carries the same poor prognosis as enhancing GBM.

Enhancement beyond the enhancing ring: satellite enhancement foci, ependymal enhancement, and leptomeningeal enhancement beyond the primary mass are staging-relevant findings:

• Subependymal/ventricular enhancement: indicates CSF pathway spread
• Contralateral hemisphere enhancement via corpus callosum: "butterfly glioma" — characteristic of GBM
• Leptomeningeal enhancement: systemic dissemination; rare at initial diagnosis but increases at progression

4.7 DSC Perfusion (rCBV)

DSC perfusion is the most widely validated advanced MRI technique for glioma grading and treatment response assessment. The rCBV (relative cerebral blood volume, normalised to contralateral white matter, typically in the centrum semiovale) reflects tumour vascularity and neoangiogenesis.

rCBV thresholds for grading (approximate; platform- and leakage-correction-dependent):

• Low-grade glioma (grade 2): rCBV typically < 1.5–2.0
• Grade 3 glioma: rCBV 2.0–3.5
• GBM: rCBV typically > 3.5 (often 5–10)
• Post-treatment change / radionecrosis: rCBV < 2.0
• Pseudoprogression (viable tumour mixed with treatment effect): variable; often rCBV 2.0–3.5
• True progression: rCBV ≥ 2.0; increases over time on serial measurements [9, 10]

The rCBV cutoff for pseudoprogression vs progression: the most validated threshold for distinguishing pseudoprogression from early true progression in post-chemoradiation GBM is rCBV ≥ 2.0 relative to contralateral WM. Sensitivity and specificity are approximately 70–80% across published series [9, 10]. The RANO 2.0 NEPA criteria [2] incorporate rCBV at this threshold as one of three criteria for non-enhancing progression.

ROI placement: the rCBV measurement must be placed in the highest-perfusion region within the tumour (hot spot), not in necrotic or cystic areas. The standard is the maximum rCBV within the enhancing component or the highest T2/FLAIR signal region for non-enhancing tumours. The contralateral reference ROI is placed in the normal-appearing centrum semiovale (see DSC sequence page for ROI placement protocol).

rCBV in oligodendroglioma: oligodendrogliomas have characteristically elevated rCBV (often > 2.0) compared with astrocytomas of equivalent grade due to their "chicken-wire" capillary network. This means that rCBV elevation in an oligodendroglioma does not necessarily indicate high-grade transformation, and the rCBV threshold must be interpreted in the context of the known or suspected molecular subtype.

rCBV and IDH mutation: IDH-mutant gliomas have lower rCBV than IDH-wildtype gliomas of equivalent morphological appearance. An infiltrating non-enhancing glioma with low rCBV (< 1.75) is more likely IDH-mutant; rCBV > 2.0 in a non-enhancing glioma suggests IDH-wildtype GBM or oligodendroglioma.

DSC technical requirements for glioma: pre-bolus leakage correction (either pre-bolus GBCA or mathematical model correction) is mandatory for reliable rCBV in GBM because the enhancing tumour leaks GBCA into the extravascular space, contaminating T2* signal. Uncorrected rCBV in GBM is systematically underestimated. See the DSC perfusion deep dive pages for full technical protocol.

4.8 MR Spectroscopy (MRS)

MRS provides metabolite ratios reflecting tissue cellular composition:

Key metabolites in glioma:

Diagnostic spectral ratios:

• Cho/NAA ratio: most used for grading. Values > 2.0 suggest high-grade; > 3.0 is highly suggestive of grade 4. Grade 2 gliomas typically show Cho/NAA 1.5–2.0.
• Cho/Cr ratio: elevated in tumour vs. normal brain; less reliable than Cho/NAA for grading.
• Lac/Cr and Lip/Cr: elevation indicates necrosis, hypoxia, or high-grade behaviour.

2-HG detection: IDH mutation generates 2-HG as an oncometabolite. Advanced MRS techniques (2D PRESS, TE optimisation at ~97 ms for 2-HG resonance at 2.25 ppm) can detect 2-HG in IDH-mutant gliomas with approximately 80–90% sensitivity [11]. This non-invasive molecular marker is increasingly used for: (a) pre-operative IDH status estimation when biopsy is deferred; (b) monitoring treatment response in IDH-mutant tumours (2-HG reduction with treatment indicates IDH inhibitor response). 2-HG MRS is technically demanding and not widely clinically available; it requires dedicated voxel placement, field strength optimisation, and specialised post-processing.

Spectroscopy limitations: MRS voxel size (typically 1–4 cm³ for single voxel; smaller for MRSI) means that a mixed tumour-oedema-necrosis region will produce blended spectra. The spectroscopy voxel must be placed in the maximally cellular (highest rCBV, highest T2* signal reduction) region to reflect the highest-grade component. Avoiding necrosis (which produces only lipid/lactate) is essential for meaningful spectra.

5. Glioma-Specific MRI Profiles by Tumour Entity

5.1 Glioblastoma, IDH-Wildtype (WHO Grade 4)

GBM is the most common primary malignant brain tumour in adults (peak 55–65 years) and the most frequent glioma encountered in daily neuroradiology practice.

Classic MRI appearance (present in approximately 70–80% of cases):

• T1 pre-contrast: heterogeneous hypointensity with T1-bright foci (haemorrhage, protein)
• T2: heterogeneous; bright infiltration with dark blood products; central heterogeneous signal (necrosis mixed with debris)
• FLAIR: extensive, poorly marginated hyperintensity extending far beyond the enhancing rim; subependymal spread common
• DWI/ADC: restricted diffusion in the enhancing rim (cellularity); elevated ADC in the necrotic core; intermediate ADC in the peritumoral FLAIR zone (variable infiltration vs. oedema)
• SWI: haemorrhage foci within and around the tumour; TVASN (tumour-associated susceptibility) in the enhancing rim; satellite dark foci
• Post-contrast T1: thick, irregular ring enhancement around central necrosis; inhomogeneous with enhancing nodular components; may be multifocal (5–10% of cases)
• rCBV: markedly elevated (typically > 4–5); hot spot in the enhancing rim
• MRS: high Cho/NAA; lactate and lipids in the necrotic zone

"Butterfly GBM": bilateral synchronous involvement of the corpus callosum (the "butterfly" shape) from midline spread via the corpus callosum. Found in 5–10% of GBMs; indicates bilateral hemispheric involvement. Differential: primary CNS lymphoma (PCNSL) — which also crosses midline but tends to be T2-iso/hypointense (dense cellularity) and homogeneously enhancing without ring pattern.

Non-enhancing GBM: approximately 15–25% of IDH-wildtype GBMs do not enhance at initial presentation. These are most common in the pre-MGMT methylated setting. Imaging shows infiltrating T2/FLAIR hyperintensity ± restricted diffusion, with elevated rCBV despite absent enhancement. The T2-FLAIR mismatch sign is absent in GBM IDH-wildtype.

GBM mimics requiring differential consideration (discussed in Section 7): metastasis (solitary or multiple), PCNSL, tumefactive demyelination, brain abscess, subacute infarct with luxury perfusion.

5.2 Astrocytoma, IDH-Mutant (WHO Grade 2/3/4)

IDH-mutant astrocytomas are the most common lower-grade diffuse gliomas in adults and have a significantly better prognosis than GBM. The WHO 2021 grading system for IDH-mutant astrocytoma is based on: grade 2 (no necrosis, no microvascular proliferation); grade 3 (mitoses, but no necrosis/MVP); grade 4 (necrosis and/or MVP present, or CDKN2A/B homozygous deletion regardless of histology).

Grade 2 IDH-mutant astrocytoma MRI:

• T1: hypointense (no calcification)
• T2: homogeneous, well-defined (but not truly circumscribed — infiltrative at histology) hyperintensity
• FLAIR: the T2-FLAIR mismatch sign [5] — lesion is bright on T2 but shows relative hypointensity on FLAIR compared with the T2 extent. This reflects the homogeneous myxoid/mucinous matrix producing very long T1 approaching FLAIR null
• DWI/ADC: no restriction; ADC > 1.3–1.5 × 10⁻³ mm²/s (high)
• No enhancement (> 95% of grade 2 cases)
• rCBV: typically < 1.5; low rCBV supports grade 2

Grade 3 IDH-mutant astrocytoma MRI: variable enhancement (40–60% enhance); rCBV intermediate (1.5–3.0); may show incomplete ring or nodular enhancement. The T2-FLAIR mismatch may be partially preserved in mixed grade 2/3 areas but is typically lost in the enhancing zone.

Grade 4 IDH-mutant astrocytoma MRI: ring enhancement, central necrosis, elevated rCBV — morphologically similar to GBM IDH-wildtype. Distinguishing features favouring IDH-mutant grade 4 astrocytoma over GBM IDH-wildtype: (a) younger patient; (b) prior history of lower-grade lesion; (c) more compact, better-defined mass; (d) less extensive satellite spread; (e) partial T2-FLAIR mismatch in the periphery. None of these features is reliable enough to replace molecular testing.

Evolution and malignant transformation: serial MRI of IDH-mutant astrocytoma shows slow growth over years. New enhancement developing in a previously non-enhancing IDH-mutant astrocytoma indicates grade transformation (anaplastic transformation from grade 2 to grade 3/4). This is a critical follow-up endpoint; loss of T2-FLAIR mismatch and new restriction on DWI are early signs preceding enhancement development.

5.3 Oligodendroglioma, IDH-Mutant and 1p/19q-Codeleted (WHO Grade 2/3)

Oligodendrogliomas are predominantly cortical/subcortical tumours (frontal lobe most common, approximately 50–60% of cases) with characteristic MRI features reflecting their calcium content and cortical preference.

MRI characteristics:

• Location: cortical or subcortical; often involves the cortex directly; frontal > temporal > parietal
• T1: predominantly hypointense; discrete T1-bright foci from calcification (mild T1 shortening) or haemorrhage
• T2: heterogeneous hyperintensity; less homogeneous than grade 2 astrocytoma; internal heterogeneity from calcium and vessels
• FLAIR: FLAIR positive (unlike grade 2 astrocytoma which shows T2-FLAIR mismatch); this reflects the cellular heterogeneity and calcium content
• SWI: the most characteristic feature — phase-bright (diamagnetic) foci within the lesion indicating calcification; magnitude-dark from T2* effect; this finding is present in 70–90% of cases and strongly supports oligodendroglioma diagnosis in the appropriate clinical context [12]
• Enhancement: grade 2 — minimal or absent; grade 3 — variable nodular or patchy enhancement
• rCBV: characteristically elevated relative to grade due to the "chicken-wire" capillary network. rCBV > 2.0 is common even in grade 2 oligodendroglioma. This means elevated rCBV should not automatically trigger grading upgrading in a known oligodendroglioma.
• Cortical scalloping: erosion of the inner table of the calvarium from chronic slow growth is pathognomonic but seen in < 20% of cases

Grade 2 vs grade 3 oligodendroglioma differentiation on imaging: grade 3 (anaplastic oligodendroglioma) shows new or increased enhancement, higher rCBV relative to prior, and DWI restriction compared with grade 2. The distinction is primarily molecular and histopathological; imaging changes are secondary markers.

5.4 Diffuse Midline Glioma, H3 K27-Altered (WHO Grade 4)

Diffuse midline glioma (formerly DIPG in the paediatric brainstem context; now a unified entity under WHO 2021) is a WHO grade 4 tumour regardless of histological appearance.

Locations: pons (most common, paediatric); thalamus (both children and adults); spinal cord; other midline structures (hypothalamus, pineal region, cerebellum).

MRI in DIPG/brainstem diffuse midline glioma:

• T2/FLAIR: expansion of the pons by > 50% with poorly defined T2 hyperintensity. The "garland" or "engulfment" of the basilar artery without encasement is characteristic
• T1 pre-contrast: diffuse T1-hypointensity replacing normal pons T1 signal
• Enhancement: variable; 30–70% show enhancement at diagnosis. Enhancement pattern does not reliably correlate with grade or outcome in DMG. Lack of enhancement at diagnosis does not indicate lower grade
• DWI: restricted diffusion in densely cellular regions; correlates with poorer prognosis
• rCBV: often moderately elevated; marked elevation predicts poor response to radiotherapy

Thalamic diffuse midline glioma (adult): bilateral thalamic infiltration, T2 hyperintensity, variable enhancement. Can be mistaken for thalamic gliosis, CNS vasculitis, or metabolic disease. DWI restriction and rCBV elevation support malignant behaviour.

At progression: DMG progresses rapidly with increased enhancement, mass effect, spread to cerebellum, midbrain, and spinal cord (posterior fossa spread); may develop ring necrosis pattern resembling GBM.

5.5 Pilocytic Astrocytoma (WHO Grade 1)

Pilocytic astrocytoma is the most common paediatric brain tumour (peak 5–15 years) and is typically located in the cerebellum (60%), optic pathway, hypothalamus, or brainstem.

Classic MRI appearance ("cyst with mural nodule"):

• T1: well-defined hypointense cyst with T1-isointense mural nodule
• T2: cyst is very T2-bright (near-CSF); mural nodule is T2-intermediate to hyperintense
• FLAIR: cyst partially suppressed (near-CSF T1); mural nodule FLAIR-positive
• Enhancement: intensely and homogeneously enhancing mural nodule — the definitive diagnostic feature. The cyst wall may or may not enhance; enhancing wall indicates the cyst is tumour-associated rather than a simple cyst
• DWI/ADC: no restriction in the nodule (despite intense enhancement); high ADC; this non-restriction despite enhancement is a key feature distinguishing PA from high-grade tumours
• rCBV: variable; may be elevated (particularly at the mural nodule) despite WHO grade 1 — this is the primary pitfall when using rCBV in the paediatric posterior fossa context
• MRS: mild Cho elevation; no lactate or lipids in uncomplicated PA

Solid pilocytic astrocytoma (hypothalamic, brainstem, optic pathway): no cyst component; solid, intensely enhancing mass; may be difficult to distinguish from other enhancing tumours in these locations. The absence of necrosis, low ADC, and absence of haemorrhage differentiate from high-grade entities.

BRAF fusion molecular context: approximately 70% of PAs harbour BRAF-KIAA1549 fusion. BRAF inhibitors (e.g., dabrafenib) are now approved for BRAF-fusion paediatric low-grade glioma — making molecular characterisation important for treatment planning. MRI cannot identify BRAF status but the classic cystic-mural nodule morphology in a child < 18 years is highly predictive.

5.6 Diffuse Paediatric Low-Grade Gliomas (MAPK pathway-altered)

The WHO 2021 classification introduced several new paediatric-specific glioma entities distinct from adult diffuse gliomas:

Diffuse low-grade glioma, MAPK pathway-altered: a heterogeneous group of low-grade (grade 1–2) diffuse gliomas in children and adolescents with BRAF, RAF1, NTRK, or other MAPK pathway alterations. They lack IDH mutation (which is characteristic of adult diffuse gliomas) and have generally better prognosis.

MRI appearance: similar to adult diffuse low-grade glioma — T2 hyperintense, infiltrative, non-enhancing or minimally enhancing. However, the FLAIR signal is typically more consistently positive (unlike adult IDH-mutant astrocytoma T2-FLAIR mismatch, which is more prevalent in IDH-mutant adult tumours). Location in children often involves the optic pathway, brainstem, and posterior fossa — locations atypical for adult IDH-mutant astrocytomas.

6. RANO Response Assessment Criteria

6.1 RANO 2.0 Framework

RANO 2.0 [2] updated the original 2010 RANO criteria [13] for treatment response assessment in GBM. The key changes from original RANO:

Volumetric enhancement measurement: RANO 2.0 mandates 3D volume of enhancing tumour (from 3D T1 post-contrast using automated or semi-automated segmentation tools) rather than 2D bidimensional product. This reduces measurement variability and captures irregular tumours more accurately.

NEPA (Non-Enhancing Progression Assessment): RANO 2.0 introduced formal criteria for non-enhancing progression (NEP) — distinct from enhancing progression. NEP is defined by at least two of three criteria: (a) rCBV ≥ 2.0 in the T2/FLAIR region; (b) ADC ≤ 1.36 × 10⁻³ mm²/s in the T2/FLAIR region; (c) MRSI Cho/NAA ≥ 1.5. These criteria aim to capture the clinically relevant phenotype of non-enhancing tumour progression, particularly in IDH-mutant gliomas that may progress without developing enhancement.

T2-FLAIR mismatch re-evaluation: the RANO 2.0 document acknowledges the T2-FLAIR mismatch as a specific imaging marker that should be tracked longitudinally — loss of mismatch may indicate grade transformation before enhancement develops.

6.2 RANO Categories

Pseudoprogression: within 12 weeks of completing concurrent chemoradiation (Stupp protocol), new or increased enhancement may represent treatment-induced inflammatory/vascular change rather than true tumour progression. RANO 2.0 [2] recommends: (a) continue treatment if clinical status is stable; (b) use multiparametric imaging (rCBV, ADC, spectroscopy) to support decision; (c) repeat imaging in 4–8 weeks. Pseudoprogression is more common in MGMT-methylated GBM (approximately 30–40% incidence) than in unmethylated GBM.

Pseudoresponse: bevacizumab (anti-VEGF) rapidly reduces enhancement and peritumoral oedema by reducing vascular permeability — this appears as dramatic "response" on standard RANO criteria but does not reflect actual tumour cell kill. The T2/FLAIR zone does not decrease proportionally, and non-enhancing progression (NEPA criteria) is the relevant assessment endpoint. rCBV normalises or decreases on bevacizumab even in progressive tumour, reducing its reliability for monitoring on anti-angiogenic therapy.

6.3 T1 Subtraction in Post-Treatment Assessment

In the post-treatment setting, pre-contrast T1 signal may be elevated due to:

• Treatment-related T1 shortening (radiation-induced protein deposition)
• Gadolinium deposition from prior injections (particularly in patients with many prior enhanced studies)
• Haemorrhagic transformation
• Calcification in treated tumour

In these settings, enhancement may be simulated or masked on standard post-contrast T1. Subtraction (post minus pre T1, registered and matched) reveals true new enhancement against elevated T1 baseline. This technique is particularly important in the post-Stupp protocol evaluation.

7. Differential Diagnosis

7.1 GBM vs. Solitary Brain Metastasis

This is the most clinically important glioma differential. Both present as ring-enhancing mass lesions with surrounding T2/FLAIR hyperintensity. Key differentiating features:

The peritumoral ADC value is the most reliable single imaging differentiator: < 1.1–1.2 × 10⁻³ mm²/s in the peritumoral zone strongly favours GBM over metastasis [14].

7.2 GBM vs. PCNSL (Primary CNS Lymphoma)

PCNSL peaks in immunocompromised patients (AIDS-defining) and immunocompetent elderly. The ring-enhancing pattern of PCNSL in immunocompromised patients may resemble GBM; the classic PCNSL in immunocompetent patients is solid-enhancing.

The ADC in PCNSL is the most useful differentiator: ADC < 0.7 × 10⁻³ mm²/s in a solid-enhancing lesion is highly specific for lymphoma. GBM enhancing tissue does not reach this level of ADC restriction.

7.3 GBM vs. Tumefactive Demyelination

Tumefactive MS (> 2 cm T2/FLAIR lesion with enhancement) is the primary demyelinating mimic:

The open ring sign and non-restriction within the enhancement zone are the key MRI differentiators. However, the open ring sign is not pathognomonic for MS — GBM can show an incomplete ring.

7.4 GBM vs. Brain Abscess

This is the most clinically urgent differential: an unsuspected abscess treated as tumour (withholding antibiotics and anti-abscess drainage) can be fatal. DWI central restriction is the essential discriminating criterion.

8. Imaging Pitfalls

8.1 Technical Pitfalls

Pseudoprogression misdiagnosed as true progression: within 12 weeks post-Stupp, new or increased enhancement without multiparametric corroboration (stable rCBV < 2.0; stable or decreasing restriction; no new clinical deficit) should not trigger treatment cessation. Always integrate rCBV and ADC before changing treatment based on enhancement increase alone.

T2-FLAIR mismatch false positive: the mismatch sign has moderate sensitivity. A similar pattern can be seen in: (a) pilocytic astrocytoma (also myxoid); (b) low-grade DNET (dysembryoplastic neuroepithelial tumour) which shares the cortical location; (c) extensive reactive demyelination adjacent to tumour. The sign is most reliable in the subcortical WM; cortical lesions may produce artefactual partial mismatch from partial volume with CSF.

Bevacizumab "pseudoresponse" misinterpreted as response: see Section 6.2. When bevacizumab is in the treatment regimen, track T2/FLAIR volume and apply NEPA criteria; do not rely on enhancement reduction as a response marker.

rCBV in oligodendroglioma misinterpreted as high grade: oligodendroglioma characteristically has elevated rCBV at grade 2 due to its capillary network. rCBV > 2.0 in a known or suspected oligodendroglioma should not trigger clinical alarm without corroborating clinical or other imaging findings.

Radiation necrosis vs. GBM recurrence at the same anatomical location: this is the most common and most diagnostically challenging post-treatment pitfall. The imaging pattern of radiation necrosis (at 6–24 months post-radiation) — thick ring enhancement, T2 heterogeneity, "soap bubble" or "Swiss cheese" central T2 pattern — can be indistinguishable from tumour recurrence on standard MRI. The distinction requires: rCBV < 2.0 (necrosis) vs. ≥ 2.0 (recurrence); ADC high in central zone; MRS lipid/lactate dominant (necrosis) vs. Cho/NAA elevated (recurrence). In persistent ambiguity, consider PET (FET/FDOPA preferred over FDG in this context) or biopsy.

Pre-contrast T1 hyperintensity masking or simulating enhancement: always acquire and review the pre-contrast T1 before interpreting the post-contrast T1. Use subtraction imaging. Do not diagnose "enhancement" without pre-contrast comparison.

8.2 Interpretive Pitfalls

Non-enhancing GBM misclassified as low-grade: as discussed in Section 4.6, approximately 20% of GBMs do not enhance at presentation. An infiltrating non-enhancing lesion with elevated rCBV (> 2.0) and IDH-wildtype molecular profile is GBM regardless of enhancement status.

Low rCBV in high-grade glioma after anti-VEGF: bevacizumab normalises rCBV values, making them unreliable for grading or progression assessment. After anti-VEGF therapy, rCBV-based criteria must be interpreted with caution.

Cortical spreading depiction on FLAIR: diffuse superficial FLAIR hyperintensity in glioma may represent: (a) leptomeningeal spread (serious — indicate tumour seeding); (b) post-contrast FLAIR artefact (gadolinium in CSF — timing artefact; see FLAIR sequence page); (c) pial infiltration by the tumour. Always acquire FLAIR before contrast or at > 30 minutes post-contrast to distinguish gadolinium artefact from true leptomeningeal disease.

FLAIR underestimation in the posterior fossa: standard FLAIR is unreliable for posterior fossa lesions due to CSF pulsation and B1 inhomogeneity. For DMG (H3 K27-altered) in the pons, use T2 TSE as the primary sequence for extent assessment, not FLAIR.

9. Specific Protocol Modifications for Glioma Imaging

9.1 New Diagnosis vs. Post-Treatment Follow-Up

New diagnosis protocol additions:

• DSC perfusion (mandatory for grading and surgical planning target identification)
• 3D isotropic post-contrast T1 at ≤ 1 mm (RANO 2.0 baseline)
• SWI for calcification (oligodendroglioma) and haemorrhage
• Spectroscopy (optional; 2-HG if IDH-mutant suspected)
• DWI with high b-value (b = 1500–2000 for cellular tumour vs. necrosis characterisation)
• Consider fMRI and DTI for pre-surgical functional and tract mapping (addressed in dedicated child pages)

Post-treatment follow-up protocol (per RANO 2.0):

• 3D pre- and post-contrast T1 (identical geometry; subtraction)
• T2 TSE (same geometry as baseline)
• 3D FLAIR (same geometry as baseline)
• DWI (b=1000, ADC map)
• DSC perfusion (rCBV for pseudoprogression assessment)
• Same scanner, same field strength, same protocol whenever possible for serial comparison

9.2 Surgical Planning Context

For pre-surgical GBM imaging, the radiologist must additionally identify and report:

• Relationship to eloquent cortex (primary motor, supplementary motor, Broca's, Wernicke's areas)
• Proximity to corticospinal tract (DTI — dedicated page)
• Eloquent cortex involvement by tumour (BOLD fMRI — dedicated page)
• Vascular territory relationship for surgical approach planning
• Degree of mass effect and midline shift (quantified on axial T2/FLAIR)
• Evidence of uncal herniation (downward displacement of the uncus through the tentorial notch on coronal T2)

10. Evidence Gaps and Ongoing Debate

rCBV threshold for pseudoprogression: the most validated threshold is rCBV ≥ 2.0, but sensitivity and specificity across published series vary widely (60–90%) reflecting: different leakage correction methods, different scanner protocols, different timing of rCBV acquisition, and heterogeneous patient populations. No multicentre standardised protocol has been prospectively validated in a phase 3 trial setting.

T2-FLAIR mismatch standardisation: the sign has not been formally standardised for scoring — different publications use different qualitative criteria for what constitutes "mismatch." A quantitative FLAIR/T2 signal ratio threshold has been proposed but not widely validated. Inter-rater reliability for the sign is moderate at best.

2-HG MRS clinical deployment: technically feasible at research centres but requires optimised voxel placement, dedicated echo times, and post-processing. No clinical standard exists for acquisition or reporting. The clinical utility for guiding treatment decisions (IDH inhibitor response monitoring) is under active investigation in clinical trials.

Radiomics and AI for molecular subtype prediction: multiple groups have published machine learning models predicting IDH mutation status, 1p/19q codeletion, and MGMT methylation from conventional MRI with 75–90% accuracy in single-centre retrospective series. No prospectively validated, multi-centre, FDA/EMA-cleared algorithm is available for clinical use at the time of writing.

Limitations of the RANO 2.0 NEPA criteria: the three NEPA criteria (rCBV ≥ 2.0, ADC ≤ 1.36, Cho/NAA ≥ 1.5) have been incorporated into RANO 2.0 but are based on limited retrospective data. Prospective validation in large cohorts is ongoing. Not all centres have access to all three criteria (particularly MRSI), creating inequity in criterion application.

11. Bibliography

A. Guidelines / Consensus / Society Recommendations

B. Systematic Reviews / Meta-analyses

C. Important Prospective / Original Studies

D. Technical MRI Papers

E. Landmark Historical References

End of document — Cerebral Gliomas MRI Deep Dive — MRIninja v1.0 — May 2026