MRI Brain for Epilepsy — Dedicated Child Protocol

MRIninja Knowledge Base | Child Page — Pathology-Specific Protocol Parent page: MRI Brain Generic Standard Protocol Version 1.0 — May 2026

Prerequisite: This page assumes full familiarity with the MRI Brain Generic Standard Protocol on MRIninja. Generic sequence theory, universal fat suppression principles, standard slice positioning, and generic artefact management are not repeated here. This page documents exclusively what changes, what is added, and what is interpreted differently when the clinical question is epilepsy.

1. Executive Summary

Epilepsy MRI is not brain MRI performed with a different question in mind — it is a fundamentally different examination that requires a dedicated protocol, dedicated plane geometry, dedicated post-processing, and a disease-specific semiological framework. The generic brain MRI protocol, designed for general neurological assessment, consistently fails to detect the most surgically important epileptogenic lesions: focal cortical dysplasia (FCD), subtle hippocampal sclerosis, and small cortical heterotopia. The stakes are high — approximately 30% of epilepsy patients have drug-resistant disease, and surgery can achieve seizure freedom in 60–80% of carefully selected patients with a lesion-positive MRI [1]. Every MRI-negative case that is in fact MRI-positive with the correct protocol represents a missed surgical opportunity.

The HARNESS-MRI (Harmonized Neuroimaging of Epilepsy Structural Sequences) protocol, endorsed by the International League Against Epilepsy (ILAE) Neuroimaging Task Force in a 2019 consensus report [1], defines the minimum standard for structural epilepsy MRI. It is built around three-dimensional isotropic acquisitions as its core — not thin-section 2D — with high-resolution T2-weighted coronal sequences targeted to the hippocampus as the essential supplementary component.

1.1 Added Value Over the Generic Protocol

The generic brain MRI protocol uses standard axial 2D FLAIR, standard axial DWI, standard axial T2, and a volumetric T1. This covers the broad neurological differential adequately but is not calibrated for epilepsy-specific lesion detection. The dedicated epilepsy protocol adds: For sequence-level protocol optimisation, vendor terminology and artefact management, see the dedicated MRIninja page FLAIR Sequence.

• 3D isotropic FLAIR (SPACE/CUBE/VISTA at 1 mm isotropic) replacing or supplementing 2D axial FLAIR — enabling multiplanar reconstruction along any epileptogenic lesion plane, which is the single most important technical change for FCD detection
• High-resolution oblique coronal T2-weighted FSE perpendicular to the hippocampal long axis — mandatory for mesial temporal sclerosis (MTS) detection and grading; this plane is absent from the generic protocol
• 3D T1 volumetric — retained from the generic protocol but used for cortical thickness analysis and morphometric post-processing
• SWI — retained and specifically interpreted for cavernoma, cortical haemosiderin, and dysplastic cortical calcification
• Double inversion recovery (DIR) — optional but highly valuable addition for FCD conspicuity
• Removal of sequences that are not diagnostically relevant to uncomplicated epilepsy workup: DWI and standard axial T2 can be deferred or abbreviated when the sole question is structural epilepsy workup in a stable patient (though both should be retained in the first MRI for full neurological survey)

1.2 Limits of the Dedicated Protocol

Even the optimised HARNESS protocol leaves approximately 15–30% of drug-resistant epilepsy patients MRI-negative in routine clinical practice [1]. The principal limitations are:

• FCD type I (subtle cortical dysmaturation without clear transmantle sign) may be invisible even on 3T HARNESS protocol
• Bilateral hippocampal sclerosis can be extremely difficult to detect by visual inspection; quantitative volumetry is required
• Extratemporal epilepsy involving the insular or frontal opercular cortex is difficult to assess due to complex sulcal anatomy
• 3T is the minimum acceptable field strength for diagnostic epilepsy MRI; 1.5T is insufficient for FCD detection in most settings [1]
• Post-processing (VBM, surface-based morphometry, MELD classifier) substantially increases detection yield but is not available in all clinical environments

The following sub-scenarios require specialised protocols beyond this page: post-surgical epilepsy MRI; paediatric epilepsy with developmental cortical anomalies (separate protocol considerations); functional MRI (fMRI) for pre-surgical language and memory lateralisation; DTI tractography for surgical planning; ictal/peri-ictal MRI for acute seizure assessment.

2. Clinical Context and Pre-Test Information

2.1 Clinical Presentation Relevant to MRI

The epilepsy MRI is requested across a spectrum of clinical scenarios that materially change the protocol targets:

New-onset seizures: the first MRI in a patient with newly diagnosed epilepsy should be a full neurological survey first (generic protocol modified to include the epilepsy-specific additions). The primary question is whether a structural cause is present and, if so, what type. The ILAE recommends structural MRI for all patients with new-onset focal epilepsy and for selected patients with generalised epilepsy [1].

Drug-resistant focal epilepsy (pre-surgical workup): the most demanding indication. Here the dedicated HARNESS protocol is mandatory. The primary question is lesion identification and lateralisation for surgical planning. Even a single small FCD that is confidently localised on MRI — and concordant with EEG, clinical semiology, and PET/SPECT data — can guide definitive surgical treatment.

Known epilepsy, previously negative MRI: re-imaging at 3T with the full dedicated protocol and post-processing is the recommended next step when prior 1.5T or non-dedicated MRI was negative in a patient with clinical evidence of focal epilepsy.

Progressive or changing seizure pattern: re-imaging required to exclude acquired lesion (low-grade glioma masquerading as epilepsy, early VGKCC/LGI1 autoimmune encephalitis, progressive FCD).

2.2 Pre-Test Information the Radiologist and Technologist Must Know

The following information materially changes protocol design or image interpretation:

• EEG localisation: the single most important pre-test datum. If EEG shows a left temporal focus, the coronal T2 and volumetric analysis must be scrutinised specifically at the left hippocampus. If EEG lateralisation is uncertain or bilateral, both sides require equal attention. Without EEG data, the exam is interpreted as a survey.
• Seizure semiology: temporal lobe automatisms → focus on mesial temporal structures; déjà vu, olfactory aura → amygdala/entorhinal cortex; tonic-clonic with motor asymmetry → extratemporal cortex
• Age at epilepsy onset: early onset (<5 years) raises the probability of cortical developmental anomalies, focal cortical dysplasia type I, and polymicrogyria; onset in adolescence or early adulthood favours MTS, FCD type II
• Prior surgery or temporal lobe resection: post-operative MRI requires modified protocol and specific semiological framework (not covered here)
• Febrile seizures in childhood: strong predictor of MTS in temporal lobe epilepsy
• Family history: suggests genetic generalised epilepsy (structural MRI often normal) or structural anomalies (tuberous sclerosis, focal cortical dysplasia)
• Medications: phenytoin can cause cerebellar atrophy; vigabatrin can cause reversible MRI signal changes in the brainstem and basal ganglia (relevant to interpretation)
• Prior MRI reports: are prior examinations truly negative, or performed with a suboptimal protocol?
• Duration of epilepsy: long-standing drug-resistant epilepsy → greater probability of secondary generalised atrophic changes that should not be misinterpreted as the primary lesion

2.3 Differential Diagnosis Landscape

The protocol must be designed to discriminate the major epileptogenic lesion types, whose MRI signatures differ substantially:

• Mesial temporal sclerosis (MTS): hippocampal volume loss and T2/FLAIR hyperintensity — requires oblique coronal T2
• Focal cortical dysplasia (FCD) type I/II/III: cortical thickening, blurring of the grey-white junction, transmantle sign, FLAIR hyperintensity — requires 3D FLAIR and dedicated post-processing
• Tuberous sclerosis complex: cortical tubers (FLAIR hyperintense), subependymal nodules, white matter radial migration lines — requires T1 and FLAIR
• Cavernous malformation: susceptibility blooming on SWI/GRE, "popcorn" appearance — requires SWI
• Cortical heterotopia (periventricular, band, nodular): signal identical to grey matter on all sequences — requires 3D T1 for morphometry
• Low-grade glioma: FLAIR hyperintensity, mass effect, occasional cortical involvement — T1 + FLAIR + post-contrast T1
• Cortical/hippocampal DNET, ganglioglioma: T2/FLAIR hyperintense cortical/juxtacortical lesion, often temporal — FLAIR and post-contrast T1
• Rasmussen encephalitis: unilateral progressive cortical atrophy and FLAIR signal changes — serial imaging
• Autoimmune encephalitis (LGI1, NMDAR, VGCC): T2/FLAIR hyperintensity in the limbic system — may not be visible on standard structural protocol; clinical and CSF context essential

3. Indications, Appropriateness and Imaging Pathway

3.1 When the Dedicated Protocol Is Indicated

The dedicated epilepsy protocol is indicated for [1, 2]:

• All patients with drug-resistant focal epilepsy being evaluated for surgical treatment
• All patients with new-onset focal epilepsy where structural MRI has not yet been performed at 3T with the dedicated protocol
• Patients with a prior MRI-negative result on a non-dedicated protocol, when focal epilepsy is clinically established by EEG
• Patients where the clinical seizure semiology and/or EEG suggest a specific anatomical focus, warranting targeted high-resolution imaging of that region
• Paediatric patients with drug-resistant epilepsy regardless of prior MRI findings

The ACR Appropriateness Criteria designate MRI with epilepsy protocol as the most appropriate imaging study for drug-resistant epilepsy presurgical evaluation [2].

3.2 When the Generic Master Protocol Is Sufficient

The generic brain MRI protocol is sufficient for:

• Single unprovoked seizure in an adult with no focal EEG features and normal neurological examination — structural exclusion purpose, not epilepsy workup
• Generalised epilepsy with clinical and EEG features fully consistent with a genetic generalised epilepsy syndrome (juvenile myoclonic epilepsy, childhood absence epilepsy) — in which a normal MRI is the expected and clinically useful finding
• Routine surveillance in known structural epilepsy without clinical change, where the question is simply "is there any change from prior study?" rather than primary lesion detection

3.3 When Further Sub-Specialised Protocols Are Required

• MRI-negative drug-resistant epilepsy: proceed to 7T MRI (if available) and/or computer-aided post-processing (MELD classifier, FreeSurfer cortical thickness analysis) before concluding MRI-negative
• Pre-surgical language and memory lateralisation: add fMRI language battery and Wada test coordination
• Invasive EEG planning: add DTI tractography for white matter map
• Suspected autoimmune limbic encephalitis: add post-contrast T1 (for enhancement) and consider dedicated CSF/serological workup before interpreting the MRI
• FDG-PET or ictal SPECT: when MRI is negative or non-lateralising in confirmed focal epilepsy

3.4 Red Flags Modifying Urgency or Protocol

4. Dedicated Protocol Design

4.1 Protocol Delta vs the Master Protocol

4.2 Mandatory Dedicated Sequences

4.3 Conditional and Advanced Sequences

4.4 Rationale per Disease-Specific Sequence

Oblique Coronal T2 Perpendicular to the Hippocampal Long Axis

This is the most disease-specific sequence in the entire epilepsy protocol and the one most consistently absent from generic brain MRI. The hippocampus lies in the medial temporal lobe at approximately 20–30° from the body axial plane, oriented anterolaterally to posteromedially. Standard axial slices cut the hippocampus at an oblique angle, producing sections through multiple hippocampal subfields simultaneously and generating partial volume artefacts that simulate or obscure volume asymmetry. The oblique coronal section perpendicular to the long axis of the hippocampus — planned from the sagittal localiser by drawing a line perpendicular to the hippocampal body — provides true cross-sections of each subfield (CA1, CA2-3, CA4/DG, subiculum) at each anatomical level from head to tail.

On this sequence, MTS manifests as: (i) volume loss — asymmetric reduction of the hippocampal cross-sectional area; (ii) T2 signal hyperintensity — proportional to the degree of gliosis; (iii) loss of internal architecture — the normal digitations of the hippocampal head are absent; (iv) loss of the dark CA1 band normally visible on high-resolution T2 (the T2-hypointense CA1-CA3 layer in normal hippocampus).

The HARNESS-MRI specification requires 2–3 mm slice thickness at ≤ 0.4 × 0.4 mm in-plane resolution, no gap, on this plane [1]. This is non-negotiable — at 3 mm or greater, early volume asymmetry is not reliably detected.

Pitfall: the amygdala is located directly anterior and superior to the hippocampal head and is not a hippocampal structure; volume asymmetry of the amygdala on coronal sequences is normal (the right amygdala is naturally 10–15% larger than the left in most individuals). Do not confuse amygdala asymmetry with hippocampal pathology.

3D Isotropic FLAIR (SPACE/CUBE/VISTA)

The 3D isotropic FLAIR acquisition at 1 mm isotropic is the core sequence for FCD detection. Its critical advantage over 2D FLAIR is not contrast or sensitivity per se, but the ability to reformat the dataset in any plane oblique to the lesion — allowing sections perpendicular to the suspected cortical dysplasia, which dramatically improves the conspicuity of the transmantle sign (a funnel-shaped or linear FLAIR hyperintensity extending from the cortex to the ventricle) and of subtle cortical T2/FLAIR signal changes.

FCD type II (balloon cell dysplasia) is the most reliably detected subtype: FLAIR hyperintensity both in the dysplastic cortex and in the immediately underlying white matter (blurring of the grey-white junction) is the characteristic appearance. FCD type I (architectural dysplasia without balloon cells) may produce no FLAIR signal change — its detection relies predominantly on the 3D T1 sequence and morphometric analysis. FCD type III (FCD associated with adjacent principal lesion) depends on the primary lesion type.

Pitfall: 3D FLAIR sequences are significantly more sensitive to patient motion than 2D FLAIR because the entire k-space is acquired over a 6–8 minute acquisition. Even small motion produces banding artefacts across the entire brain that can simulate or obscure cortical signal changes. Motion-insensitive acquisition schemes (PROPELLER/BLADE reconstruction) reduce but do not eliminate this vulnerability.

DIR (Double Inversion Recovery)

DIR uses two sequential inversion recovery pulses — one to null white matter signal (at TI_WM) and one to null CSF signal (at TI_CSF). The result is an image in which only grey matter is visible against a dark background. This dramatically improves the contrast between dysplastic grey matter and the surrounding WM, particularly for FCD type IIa and IIb where the grey-white junction blurring is the key sign. Published sensitivity for FCD detection with DIR ranges from 50–88% (varying by reader experience) [4], compared to approximately 44–67% for MPRAGE alone in the same series. The combination of 3D FLAIR + DIR + MPRAGE provides higher sensitivity than any single sequence.

Pitfall: DIR images show artefactually elevated cortical signal at cortical folding depth (sulcal fundus) due to partial volume effects between grey matter and CSF on the sulcal bank. This can simulate subtle FCD at sulcal fundus sites. Correlation with the 3D FLAIR and MPRAGE is mandatory before interpreting DIR signal as pathological.

4.5 Dedicated Planes, FOV, Resolution and Coverage

Oblique coronal for hippocampus:

• Planned from the sagittal T1 or FLAIR localiser
• Draw a line perpendicular to the long axis of the hippocampal body
• The angle is typically 10–25° from the true coronal, varying with patient anatomy
• Coverage: from the amygdala anteriorly to the posterior hippocampal tail (typically 50–60 mm craniocaudal extent)
• Slice thickness: 2–3 mm, no gap
• Target in-plane resolution: ≤ 0.4 × 0.4 mm
• Both hippocampi must be fully covered in every section

3D FLAIR:

• Acquired in the sagittal plane for maximum head coverage within one 3D slab
• Isotropic 1 mm voxels; post-hoc reformats in coronal and axial planes are routine
• The key post-hoc reformat is along and perpendicular to any suspected FCD — this requires the clinical information (EEG lateralisation) to be conveyed to the technologist before acquisition

3D T1:

• Acquired in the sagittal plane; same as the generic protocol
• No change from master protocol — 1 mm isotropic MPRAGE/BRAVO
• Used identically for FCD morphometry and for heterotopia detection

4.6 Contrast Strategy

Gadolinium-based contrast is not part of the standard epilepsy protocol for the core structural epilepsy workup (FCD, MTS, heterotopia, cavernoma, TSC). None of these lesions require contrast for their primary detection or characterisation.

Post-contrast T1 is added as a conditional sequence in the following specific scenarios:

• New-onset seizure in a patient over 40 years, where a mass lesion (low-grade glioma, metastasis) must be excluded
• Clinical or laboratory suspicion of autoimmune limbic encephalitis (VGKCC/LGI1/CASPR2 antibodies), where hippocampal or amygdala enhancement may be present in the acute phase
• EEG-concordant lesion on structural imaging that has an atypical appearance suggesting DNET or ganglioglioma, where enhancement would be diagnostically relevant
• Any MRI finding that is morphologically atypical for the expected epileptogenic lesion type

When contrast is added, pre-contrast 3D T1 is always acquired first (for subtraction) and the post-contrast 3D T1 uses identical parameters (as in the generic master protocol).

4.7 Sequence Matching, Reproducibility and Follow-Up

For serial epilepsy MRI (pre-surgical workup, disease monitoring), the oblique coronal T2 must be acquired with identical prescription (identical angulation to the hippocampal long axis, identical slice positions) at every examination. Even small changes in angulation produce apparent hippocampal volume changes that are artifactual. In practice, the coronal angulation angle in degrees from the ACPaC (anterior commissure-posterior commissure plane) should be documented in the first report and reproduced at each subsequent examination.

The 3D FLAIR and 3D T1 prescriptions are inherently reproducible because they are sagittal whole-brain acquisitions that do not depend on lesion-specific prescription. However, the post-hoc reformats should be performed in identical oblique planes for serial comparison.

5. MRI Semiotics — Disease-Specific Imaging Findings

5.1 Direct Signs

Mesial Temporal Sclerosis (MTS)

MTS produces a characteristic combination of findings on the oblique coronal T2 and FLAIR sequences that together have high diagnostic confidence when present [3, 5, 6]:

• Hippocampal volume loss (primary sign): asymmetric reduction of hippocampal cross-sectional area on coronal T2, best quantified by comparing the cross-sectional area or measuring the cranial-caudal diameter bilaterally. Even experienced neuroradiologists have a threshold of approximately 10–15% volume reduction for reliable visual detection; smaller asymmetries require volumetric quantification.
• T2/FLAIR signal hyperintensity (primary sign): increased T2 and FLAIR signal in the hippocampal body and head, reflecting gliosis and free water accumulation secondary to neuronal loss. The signal is diffusely elevated throughout the hippocampal formation in ILAE type 1 HS; more focally elevated in type 2 and 3.
• Loss of hippocampal internal architecture: the normal hippocampal digitations of the head are absent; the normal stratification of the CA1-CA3 layers (visualised as a T2-hypointense band on high-resolution coronal T2) is lost.
• Hippocampal head folding loss: the normal complex infolded morphology of the hippocampal head is simplified and smooth in MTS.

On the 3D T1 MPRAGE, volume loss is visible as a thin and atrophic hippocampal formation compared to the contralateral side, with a small hippocampal body relative to the parahippocampal white matter.

Focal Cortical Dysplasia (FCD)

MRI semiotics of FCD depend critically on the histological subtype [7, 8]:

FCD type II (Taylor's type, with balloon cells):

• Cortical thickening: the dysplastic cortex appears thicker than the adjacent normal cortex on 3D T1 and FLAIR; normal cortical thickness is 2–5 mm, with FCD often showing 5–8 mm
• Grey-white junction blurring: the sharp demarcation between cortical grey matter and subcortical white matter on T1 is indistinct in FCD, producing a zone of intermediate signal at the junction
• Transmantle sign: a funnel- or column-shaped FLAIR hyperintensity extending from the lesion in the cortex through the white matter toward the lateral ventricle — this is highly specific for FCD type IIb (with balloon cells) and its detection on 3D FLAIR with multiplanar reformats is the major diagnostic advance of the HARNESS protocol
• Cortical and subcortical FLAIR hyperintensity: both the dysplastic grey matter and the immediately underlying white matter show elevated T2/FLAIR signal in FCD type IIb
• Gyral pattern abnormality: the affected gyrus or sulcus may show simplified morphology, abnormal depth, or absent normal sulcal pattern

FCD type I (architectural dysplasia only):

• Often FLAIR-negative or with only subtle signal change
• Primary sign: cortical thickening on 3D T1 → requires morphometric post-processing
• May show subtle T2 signal change in the overlying cortex on 3D FLAIR if acquisitions are optimal

FCD type III (associated with adjacent hippocampal sclerosis or other lesion):

• The cortical changes of FCD type I or II are present in the cortex adjacent or contiguous with the principal lesion (hippocampus, scar, vascular malformation)
• May be subtle and require knowledge of the primary lesion location for targeted search

Cavernous Malformation

• Classic "popcorn" appearance on SWI/GRE: central T1/T2 heterogeneous signal (mixed haemorrhagic products at different stages) surrounded by a complete haemosiderin ring (blooming on SWI)
• The haemosiderin ring distinguishes cavernoma from other lesions; AVM lacks the complete ring
• May have surrounding cortical gliosis visible as FLAIR signal change

Cortical/Band/Periventricular Heterotopia

• Signal identical to grey matter on ALL sequences (T1, T2, FLAIR) — this is the defining feature
• Periventricular nodular heterotopia: nodules lining the ventricular wall, convex toward the ventricle, grey-matter signal on T1 and T2
• Band heterotopia (double cortex): a band of grey matter signal interposed between the cortex and the white matter; may produce a "double cortex" appearance on axial T1

Tuberous Sclerosis Complex

• Cortical tubers: FLAIR hyperintense, T1 hypointense cortical lesions at the grey-white junction; may calcify in older patients (becoming T1/SWI signal change); do not enhance unless giant cell astrocytoma develops
• Subependymal nodules (SEN): along the ventricular wall, T1 slightly hyperintense (calcium), FLAIR hypointense; may calcify (SWI blooming) or progressively enlarge to SEGA
• Subependymal giant cell astrocytoma (SEGA): near the foramen of Monro; T1 bright (calcium + protein), T2 heterogeneous, enhances on post-contrast T1 — the one TSC lesion requiring post-contrast T1

5.2 Indirect and Secondary Signs of MTS

• Ipsilateral temporal lobe atrophy: reduction of overall temporal lobe volume ipsilateral to MTS
• Choroidal fissure dilatation: the CSF space between the hippocampus and the temporal stem is enlarged on the atrophic side
• Collateral sulcus dilatation: the temporal horn of the lateral ventricle is enlarged
• Fornical atrophy: the fornix is atrophic ipsilateral to MTS — visible as asymmetric volume reduction of the fornicular columns on coronal or sagittal sequences
• Mammillary body atrophy: the ipsilateral mammillary body is smaller, reflecting Papez circuit neuronal loss
• Contralateral thalamic atrophy: in long-standing MTS, the ipsilateral dorsomedial thalamic nucleus shows volume loss and T2/FLAIR signal change

These secondary signs increase diagnostic confidence when the primary hippocampal signs are subtle or equivocal. Their presence confirms the ipsilateral side as the affected side even when hippocampal volume asymmetry is borderline.

5.3 Severity, Extent and Activity Assessment

For MTS, severity is assessed by combining:

• Degree of volume loss (mild: asymmetry 10–15%; moderate: 15–25%; severe: > 25%)
• Degree of T2/FLAIR signal change (mild: subtle; moderate: clearly visible; severe: markedly hyperintense)
• Presence and extent of secondary signs (fornical atrophy, temporal lobe atrophy, contralateral thalamic change)

For FCD, severity relates to the surgical relevance:

• Size of the dysplastic cortex (larger lesions are more readily detected)
• Presence of transmantle sign (strongly predicts complete surgical resection possibility)
• Proximity to eloquent cortex (motor strip, language cortex) — determines whether resection is feasible
• Extent of FLAIR hyperintensity in the white matter

5.4 Validated Classification and Grading Systems

ILAE Hippocampal Sclerosis Classification (2013)

The ILAE Neuroimaging Task Force published a pathological classification of hippocampal sclerosis in 2013 that has been correlated with MRI findings and surgical outcomes [3, 5, 6]:

The MRI-based radiological classification proposed by Okromelidze et al. (2024) [5] aligns with the ILAE histopathological classification and provides specific MRI criteria for each type, with the aim of improving pre-surgical communication.

ILAE Focal Cortical Dysplasia Classification (2022)

The ILAE revised FCD classification (2022) [7] is pathology-based but has direct MRI correlates:

• FCD type Ia/Ib: radial/tangential laminar architectural dysplasia — MRI often negative; primary detection by morphometry
• FCD type IIa: dysmorphic neurons without balloon cells — FLAIR signal less prominent; grey-white blurring the key sign
• FCD type IIb: dysmorphic neurons with balloon cells — classic transmantle sign; FLAIR hyperintensity prominent; most MRI-visible type
• FCD type III: secondary to adjacent hippocampal sclerosis, vascular malformation, tumour, or scar; MRI shows both the primary lesion and the FCD component in adjacent cortex

5.5 Differential Diagnosis on MRI

5.6 Mimickers, Pseudolesions and Normal Variants

Normal hippocampal asymmetry: minor right-left hippocampal asymmetry (< 10% volume difference) is common in normal subjects. Do not diagnose MTS in the absence of signal change and/or secondary signs.

Normal hippocampal head cysts (hippocampal sulcus remnants): small CSF-signal cysts at the hippocampal head are a normal variant (unfused hippocampal sulcal remnants). They appear as T2/FLAIR bright and T1 dark focal lesions, following CSF signal on all sequences — distinguished from MTS by CSF signal (not tissue signal) and absence of volume asymmetry or signal change in the hippocampus itself.

Post-ictal transient signal changes: in the peri-ictal period, the hippocampus and temporal cortex can show transient T2/FLAIR hyperintensity and restricted diffusion that simulates MTS or encephalitis. Serial imaging after the post-ictal period resolves these changes; MTS is persistent.

FCD mimickers on 3D FLAIR: motion artefacts in the cortex (particularly at the vertex and parasagittal cortex) can produce apparent cortical signal changes. Correlation with the 3D T1 (which is less motion-sensitive) distinguishes artefact from true cortical lesion.

Mega cisterna magna and prominent Virchow-Robin spaces: these produce T2/FLAIR bright spaces that should not be mistaken for cortical lesions; they follow CSF signal precisely on FLAIR (attenuated).

6. Reporting Framework Specific to Epilepsy

6.1 Structured Reporting Template

Indication and clinical question: state the seizure type (focal, generalised, unclassified), EEG lateralisation or localisation if known, and the specific question (lesion detection, pre-surgical evaluation, change from prior MRI).

Technique (disease-specific items only): state field strength (3T mandatory for dedicated epilepsy workup), that the hippocampal oblique coronal T2 was performed with stated angulation, that 3D isotropic FLAIR was acquired, whether DIR was performed, whether contrast was administered (and if so, why). State coil used if non-standard.

Comparison: document any prior epilepsy MRI with date, field strength, and whether the prior study used a dedicated epilepsy protocol.

Findings — hippocampal assessment (always first if temporal lobe focus suspected):

• Right hippocampus: volume, T2 signal, internal architecture
• Left hippocampus: volume, T2 signal, internal architecture
• Symmetry assessment: symmetric / right > left / left > right (qualitative); volumetric if available
• Secondary signs of MTS if present: fornical atrophy, temporal horn dilatation, temporal lobe volume

Findings — cortical assessment (FCD screen):

• 3D FLAIR: any focal cortical or juxtacortical signal change (location, size, transmantle sign)
• 3D T1: grey-white junction blurring, cortical thickening, sulcal pattern abnormality
• DIR (if performed): any cortical signal change not visible on other sequences

Findings — other epileptogenic lesions:

• SWI: cavernoma or cortical haemosiderin (location, blooming grade, associated signal changes)
• Heterotopia: location, signal characteristics
• TSC features: tubers, SEN, SEGA

Impression — aligned with the clinical question:

• Is the MRI positive or negative for an epileptogenic lesion?
• If positive: lesion type, location (lobe, subregion, laterality), size, relevant surgical landmarks
• If MTS: side, severity, secondary signs, ILAE classification if possible
• If FCD: type (based on MRI signature), location, transmantle sign present/absent, relationship to eloquent cortex
• Concordance with provided EEG lateralisation (state if concordant, discordant, or EEG data not provided)

Limitations: state if full HARNESS protocol was not performed; if motion limited 3D FLAIR quality; if post-processing was not available; if the study was performed at 1.5T.

Recommendations: if MRI-negative in drug-resistant focal epilepsy — state that post-processing (FreeSurfer/MELD) and/or 7T MRI are the recommended next steps; do not simply report "no epileptogenic lesion identified" without this context.

6.2 Mandatory Disease-Specific Reporting Checklist

• [ ] Hippocampal volume: symmetric vs. asymmetric; which side if asymmetric
• [ ] Hippocampal T2 signal: normal vs. hyperintense; which side and which segment
• [ ] Hippocampal internal architecture: preserved vs. lost
• [ ] Secondary MTS signs: fornical atrophy, temporal horn dilatation, collateral sulcal dilatation
• [ ] Cortical grey-white junction: sharp vs. blurred (3D T1)
• [ ] FLAIR cortical signal: normal vs. focal abnormality (location, transmantle sign)
• [ ] DIR signal if performed: any additional cortical finding not on FLAIR
• [ ] SWI: cavernoma/haemosiderin present or absent
• [ ] Heterotopia: periventricular or cortical nodules with grey matter signal
• [ ] TSC features: tubers, SEN, SEGA
• [ ] Concordance with EEG lateralisation stated explicitly
• [ ] Post-processing performed or recommended

6.3 Critical Findings and Communication

Direct communication to the referring epileptologist/neurologist is required when:

• A previously MRI-negative patient is found positive for an epileptogenic lesion — this changes the surgical pathway immediately
• A lesion is identified adjacent to the motor or language cortex — neurosurgical planning implications
• An unexpected mass lesion (glioma, metastasis) is found — changes the diagnostic and treatment pathway entirely
• A SEGA is identified in a TSC patient — foramen of Monro obstruction risk requires immediate neurosurgical referral

6.4 Common Reporting Errors

7. Technical Pitfalls and Disease-Specific Optimisation

7.1 Technical Pitfalls Specific to Epilepsy

Motion on 3D FLAIR is the single most common cause of non-diagnostic epilepsy MRI. The 6–8 minute 3D FLAIR acquisition is exquisitely sensitive to motion: even 1–2 mm of head displacement between k-space segments produces banding artefacts across the entire brain that can simulate cortical signal changes or obscure true ones. Mitigation: place the 3D FLAIR sequence first in the protocol (before patient fatigue); use head immobilisation padding; consider PROPELLER/BLADE reconstruction if available; repeat acquisition if banding is visible across cortical regions of interest.

Incorrect oblique coronal plane for hippocampal assessment: if the coronal T2 is not perpendicular to the hippocampal long axis, the hippocampal cross-sections are oblique, producing artifactual volume asymmetry and false T2 signal changes. This is the most common technical cause of both false-positive and false-negative MTS diagnosis. Mitigation: always plan from the sagittal T1 or FLAIR by drawing the perpendicular line directly on the hippocampal body visible on the sagittal image.

Partial volume at hippocampal head/amygdala interface: the amygdala sits directly anterior to the hippocampal head. In sections that are slightly anterior to the true head, the amygdala signal is sampled rather than hippocampal signal, producing apparent T2 heterogeneity that mimics hippocampal signal change. Mitigation: identify the hippocampal head landmarks (the hippocampal digitations are visible on true coronal sections but absent at the amygdalar level) and restrict signal analysis to sections that contain hippocampal tissue.

1.5T field strength: the ILAE consensus [1] and published evidence [9] consistently document that 1.5T is insufficient for reliable FCD detection and for subtle MTS. If epilepsy MRI is performed at 1.5T, this limitation must be explicitly stated in the report. Re-imaging at 3T should be recommended for any MRI-negative result at 1.5T in drug-resistant focal epilepsy.

7.2 Sequence-Specific Disease Pitfalls

3D FLAIR: T2-shine-through from prominent perivascular spaces can simulate focal cortical signal change, particularly in the temporal pole and insular region. These spaces follow CSF signal on other sequences (T1 dark, T2 bright); they do not show grey-white blurring on T1.

Oblique coronal T2: white matter hyperintensities related to migraine, small vessel disease, or ageing can be adjacent to the hippocampus and should not be misinterpreted as hippocampal gliosis. These hyperintensities are typically in the white matter, not in the hippocampal formation itself, and do not produce volume loss.

DIR: the most common DIR artefact is the "cortical false positive" at gyral crests and sulcal fundi, where partial volume between grey matter and CSF produces apparent grey matter signal elevation. This artefact follows the sulcal anatomy symmetrically — it is characteristically bilateral and geometrically regular. True FCD is focal, asymmetric, and does not correspond to expected partial volume geometry.

SWI: signal blooming on SWI from dental amalgam, surgical hardware, or vascular structures can simulate small cavernomas. The practical discriminator is that artefactual blooming does not show the complete haemosiderin ring with central heterogeneous signal characteristic of cavernoma. Always correlate SWI with T1 — cavernomas are T1-bright centrally; artefacts are not.

7.3 When the Exam Is Non-Diagnostic for This Question

The dedicated epilepsy MRI protocol is non-diagnostic when:

• 3D FLAIR quality is non-diagnostic due to motion → repeat with better immobilisation; consider procedural sedation for paediatric patients
• The study was performed at 1.5T → repeat at 3T
• Post-processing was not available → refer to a tertiary epilepsy centre with post-processing capability
• MRI-negative epilepsy remains unexplained after HARNESS protocol → next step is 7T MRI (if available) and/or MELD deep learning classifier; consider invasive EEG evaluation
• The clinical question requires ictal imaging → imaging during or immediately after a seizure is not routinely achievable; consider ASL and PET

8. MRI Technologist Pearls Specific to Epilepsy

8.1 Disease-Specific Positioning and Coil Tricks

Head positioning for the epilepsy protocol should place the head in the coil with the chin slightly tucked — this minimises the hippocampal oblique angulation relative to the scanner plane, making oblique coronal prescription more intuitive on the localiser. Foam padding bilaterally at the temples reduces minor motion from swallowing more effectively than forehead padding alone.

If the patient is known to have frequent micro-seizures or myoclonus, acquire the 3D FLAIR first, before the patient becomes fatigued. Even brief myoclonic jerks during the 3D FLAIR acquisition produce irreversible banding artefacts.

8.2 Sequence Order Logic

Recommended acquisition order for the dedicated epilepsy protocol:

1. Three-plane localiser
2. 3D FLAIR ← first priority; patient is freshest; least motion; longest and most motion-sensitive acquisition
3. Oblique coronal T2 perpendicular to hippocampus ← plan from the sagittal view of the 3D FLAIR or T1; second priority
4. 3D T1 MPRAGE/BRAVO ← less motion-sensitive; can tolerate minor motion
5. SWI ← stable and motion-tolerant
6. DIR (if performed) ← after T1 to allow DIR TI optimisation from T1 signal
7. DWI, post-contrast T1 (if indicated) ← last

The oblique coronal T2 must be planned from a sequence that already shows the hippocampus clearly (the 3D FLAIR or 3D T1 sagittal reformat). This means the technologist must perform an interactive planning step after the 3D FLAIR acquisition before proceeding to the coronal T2. This is the step most frequently skipped in non-specialist departments.

8.3 Fast Salvage Version of the Dedicated Protocol

These four sequences in approximately 18–23 minutes constitute the minimum clinically useful epilepsy MRI. DIR can be added if the time window extends to 30 minutes.

8.4 Disease-Specific Avoidable Errors

9. Quality Control Checklist for the Dedicated Protocol

• [ ] 3D FLAIR: isotropic 1 mm coverage of whole brain; no motion banding artefacts over relevant cortical regions
• [ ] Oblique coronal T2: perpendicular to hippocampal long axis (verified on sagittal localiser view); both hippocampi fully covered from head to tail; ≤ 3 mm slice thickness; no gap; ≤ 0.4 × 0.4 mm in-plane resolution
• [ ] 3D T1 MPRAGE: isotropic 1 mm; whole brain coverage; adequate grey-white junction contrast
• [ ] SWI: full brain coverage; phase images available in addition to magnitude
• [ ] DIR (if acquired): adequate grey matter-white matter-CSF differentiation on spot check
• [ ] EEG lateralisation data provided to reading radiologist
• [ ] Field strength documented (3T or 1.5T) in the protocol notes
• [ ] Post-processing performed (yes/no; method) or referred to tertiary centre
• [ ] Oblique coronal T2 angulation documented for serial study reproducibility

11. Evidence Gaps and Ongoing Debate

Sensitivity ceiling for FCD detection: even with the full HARNESS protocol at 3T, approximately 15–30% of histologically confirmed FCD cases remain MRI-negative. Whether this reflects resolution limitations of 3T MRI, insufficient post-processing, or truly "invisible" FCD type I is unresolved. The relative contribution of 7T MRI versus advanced post-processing versus AI-based classifiers (MELD, DeepFCD) in closing this gap is an active research area without definitive comparative data.

DIR vs. 3D FLAIR for FCD detection: no large prospective study has definitively established whether DIR adds clinically significant diagnostic value over 3D FLAIR alone in a routine department context (as opposed to dedicated epilepsy centres with expert readers). Published sensitivity figures for DIR vary widely (50–88%) and are highly reader-dependent, raising questions about its generalisability in non-expert settings.

AI-assisted FCD detection (MELD classifier): the MELD (Multi-center Epilepsy Lesion Detection) classifier [10] has demonstrated detection of previously missed FCD lesions in MRI-negative epilepsy on population-level data. Its clinical implementation and false-positive rate in routine practice remain active areas of investigation. The classifier requires standardised HARNESS protocol inputs; non-standard protocols produce unreliable outputs.

HS ILAE subtyping by MRI: the proposed MRI-based ILAE HS classification [5] has not yet been prospectively validated in large multicentre series for inter-reader reliability or predictive value for surgical outcome. Whether MRI-based HS subtyping adds surgical planning value beyond the binary MTS positive/negative determination is not established.

Contrast in autoimmune epilepsy: the role of post-contrast T1 in initial epilepsy workup for autoimmune limbic encephalitis remains debated. Enhancement in the acute phase of LGI1 encephalitis is inconsistently reported; most limbic encephalitis cases are not enhancing on structural MRI. Guidelines do not mandate contrast for routine epilepsy workup, but clinical acuity may justify it.

1.5T vs. 3T equivalence for HARNESS: the ILAE task force endorses the HARNESS protocol for both 1.5T and 3T, acknowledging that 1.5T with optimised protocol is better than non-dedicated 3T. However, the diagnostic yield for FCD detection at 1.5T versus 3T in comparative studies consistently favours 3T. The minimum acceptable field strength for drug-resistant epilepsy presurgical evaluation in current practice is a matter of institutional capability rather than consensus threshold.

12. Evidence-Based References

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 — MRI Brain Epilepsy Child Protocol — MRIninja v1.0 — May 2026

Parent page: MRI Brain Generic Standard Protocol

Future child pages building on this page: post-surgical epilepsy MRI; paediatric cortical malformation MRI; autoimmune limbic encephalitis MRI; pre-surgical fMRI and DTI protocol.