Recommendations for the use of structural magnetic resonance imaging in the care of patients with epilepsy: A consensus report from the International League Against Epilepsy Neuroimaging Task Force

1054

wileyonlinelibrary.com/journal/epi Epilepsia. 2019;60:1054–1068.

Wiley Periodicals, Inc.

Received: 11 May 2018

Revised: 23 April 2019

Accepted: 24 April 2019

DOI: 10.1111/epi.15612

SPECIAL REPORT

Recommendations for the use of structural magnetic resonance

imaging in the care of patients with epilepsy: A consensus report

from the International League Against Epilepsy Neuroimaging

Task Force

AndreaBernasconi

FernandoCendes

William H.Theodore

Ravnoor S.Gill

Matthias J.Koepp

Robert EdwardHogan

Graeme D.Jackson

PaoloFederico

AngeloLabate

Anna ElisabettaVaudano

IngmarBlümcke

PhilippeRyvlin

NedaBernasconi

Neuroimaging of Epilepsy Laboratory,McConnell Brain Imaging Centre,Montreal Neurological Institute and Hospital,McGill University, Montreal,

Quebec, Canada

Department of Neurology,University of Campinas, Campinas, Brazil

Clinical Epilepsy Section,National Institutes of Health, Bethesda, Maryland

Institute for Neurology,University College London, London, UK

Department of Neurology,Washington University School of Medicine, St Louis, Missouri

Florey Institute of Neuroscience and Mental Health,University of Melbourne, Heidelberg, Victoria, Australia

Hotchkiss Brain Institute,University of Calgary, Calgary, Alberta, Canada

Institute of Neurology,University of Catanzaro, Catanzaro, Italy

Neurology Unit,Azienda Ospedaliero Universitaria,University of Modena and Reggio Emilia, Modena, Italy

Department of Neuropathology,University Hospital Erlangen, Erlangen, Germany

Clinical Neurosciences,Lausanne University Hospital, Lausanne, Switzerland

Andrea Bernasconi and Neda Bernasconi contributed equally to this work.

Correspondence

Andrea Bernasconi, Neuroimaging of

Epilepsy Laboratory, McConnell Brain

Imaging Centre and Montreal Neurological

Institute and Hospital, McGill University,

3801 University Street, Montreal, Quebec

H3A 2B4, Canada.

Email: [email protected]

Abstract

Structural magnetic resonance imaging (MRI) is of fundamental importance to the

diagnosis and treatment of epilepsy, particularly when surgery is being considered.

Despite previous recommendations and guidelines, practices for the use of MRI

are variable worldwide and may not harness the full potential of recent techno-

logical advances for the benefit of people with epilepsy. The International League

Against Epilepsy Diagnostic Methods Commission has thus charged the 2013‐2017

Neuroimaging Task Force to develop a set of recommendations addressing the fol-

lowing questions: (1) Who should have an MRI? (2) What are the minimum re-

quirements for an MRI epilepsy protocol? (3) How should magnetic resonance (MR)

images be evaluated? (4) How to optimize lesion detection? These recommenda-

tions target clinicians in established epilepsy centers and neurologists in general/

district hospitals. They endorse routine structural imaging in new onset generalized

1055

BERNASCONI EtAl.

INTRODUCTION

Since its inception in the early 1980s, steady advances in

magnetic resonance imaging (MRI) technology have led to

dramatic improvements in the ability to obtain high‐qual-

ity detailed information about the brain, thereby providing

insights into disease processes. Computational approaches

and novel quantitative MRI acquisition and postprocessing

techniques have emerged to study neuroanatomy, yielding

increasingly sophisticated markers of tissue microstructural

integrity. In epileptology, MRI has revolutionized our abil-

ity to detect lesions, shifting the field from prevailing elec-

troclinical correlations to a multidisciplinary approach. In

particular, this technique has become fundamental in the

management of drug‐resistant epilepsy, as the identification

of a clear‐cut lesion on structural MRI is associated with fa-

vorable seizure outcome after surgery.

The rapid pace of technical advances and developments in

neuroimaging has not systematically translated into clinical

care. This is due to a number of reasons, including variabil-

ity in economic resources and technical infrastructures, diffi-

culty of performing prospective randomized controlled trials

to assess level of evidence and added value of a given test, and

lack of standardized image acquisition protocols and postpro-

cessing methods. Collectively, these factors may slow down or

impede timely validation of imaging markers and assessment

of generalizability, thus creating a sense of disconnect between

research and clinical practice. Over the years, the International

League Against Epilepsy (ILAE) has thus produced consensus

recommendations on the use of MRI in the diagnosis and man-

agement of people with epilepsy. The first was published in

1997,

followed by guidelines focused on patients with drug‐

resistant epilepsy

and functional neuroimaging

published in

the 1998 and 2000, respectively. In 2009, the subcommittee

for pediatric neuroimaging recommended structural MRI as

the examination of choice in recent onset epilepsy.

In 2015,

the Task Force Report for the ILAE Commission of Pediatrics

recommended neuroimaging at all levels of care for infants

presenting with epilepsy, with level A recommendation for

structural MRI as standard investigation.

Despite previous ILAE recommendations and guide-

lines, practices on the use of MRI are still variable world-

wide and do not harness the full potential of technological

advances for the benefit of people with epilepsy. The ILAE

Diagnostic Methods Commission has thus charged the

2013‐2017 Neuroimaging Task Force to formulate a new

consensus recommendation for the use of MRI in epilepsy

answering the following key questions: (1) Who should

have an MRI? (2) What are the minimum requirements for

an MRI epilepsy protocol? (3) How should magnetic res-

onance (MR) images be evaluated? (4) How to optimize

lesion detection? As the ultimate purpose of this recom-

mendation is to standardize epilepsy diagnostic imaging

in outpatient clinics and specialized surgery centers alike,

and focal epilepsy alike and describe the range of situations when detailed assess-

ment is indicated. The Neuroimaging Task Force identified a set of sequences, with

three‐dimensional acquisitions at its core, the harmonized neuroimaging of epilepsy

structural sequences—HARNESS‐MRI protocol. As these sequences are available

on most MR scanners, the HARNESS‐MRI protocol is generalizable, regardless of

the clinical setting and country. The Neuroimaging Task Force also endorses the use

of computer‐aided image postprocessing methods to provide an objective account of

an individual's brain anatomy and pathology. By discussing the breadth and depth of

scope of MRI, this report emphasizes the unique role of this noninvasive investiga-

tion in the care of people with epilepsy.

KEYWORDS

adults, epilepsy, pediatrics, structural magnetic resonance imaging

Key Points

• Practices for the use of structural MRI are variable

worldwide and may not harness the full potential

of technological advances for the benefit of peo-

ple with epilepsy

• The Neuroimaging Task Force recommends use of

the Harmonized Neuroimaging of Epilepsy Structural

Sequences (HARNESS‐MRI) protocol with iso-

tropic, millimetric 3D T1 and FLAIR images, and

high‐resolution 2D submillimetric T2 images

• Use of the HARNESS‐MRI protocol standardizes

best‐practice neuroimaging of epilepsy in outpa-

tient clinics and specialized surgery centers alike

1056

BERNASCONI EtAl.

the categorization of these questions is intentionally broad

and independent from the clinical definition of drug‐resis-

tance and nonlesional MRI. Despite American Academy of

Neurology guidelines recommending referral for surgical

evaluation to specialized centers and ILAE recommenda-

tions defining refractory epilepsy (eg, failure to respond to

two adequately tried medications),

7,8

often these criteria

are not applied by the treating physicians, and on average

adult patients who do get surgery have had intractable epi-

lepsy for 20years or more.

9‒11

Moreover, the terminology

“nonlesional MRI” is currently ill‐defined and depends on

multiple factors, including the type of imaging, the reader's

expertise, and the use of postprocessing.

12,13

MATERIALS AND METHODS

The current recommendations derive from the following

considerations, with the aim of providing a consensus view

on the role of structural MRI in epilepsy. First, they build

upon previous ILAE neuroimaging reports. Second, they

derive from clinical protocols conducted at the institutions

of the members of the Neuroimaging Task Force with basic

sequences available on most MR scanners and thus general-

izable to many centers, regardless of the clinical setting and

country. Third, they consider review papers, evidence‐based

guidelines, and reports on the role of structural MRI in the

diagnosis and management of seizure disorders,

14‒25

with

particular attention to studies that meet at least some stand-

ards for evidence classification. These sources of informa-

tion were complemented by a literature review based on an

Ovid MEDLINE query between 2002 and 2018. The search

strategy and list of 67 identified publications are detailed in

Material S1. Our recommendations, which take into account

clinical indications, new developments in MRI hardware and

sequences, and research findings, are intended to be primarily

applicable to adult patients; the overall principles, however,

are generalizable to children. Also, they are intentionally

broad to assist clinicians in established epilepsy surgery cent-

ers and general neurology clinics alike. Implementation of

such recommendations necessarily will vary depending on

available resources and organization of care. Ideally, in the

developed world, only centers meeting appropriate standards

should image patients with epilepsy. In resource‐limited set-

tings where technical infrastructure and specialist training

may not be available, epilepsy care must still be provided;

these recommendations are thus an essential resource to per-

suade local health organizations to provide or improve both

training and access to MRI services.

In the following paragraphs, Neuroimaging Task Force

recommendations on the use of MRI pertain to the proposed

Harmonized Neuroimaging of Epilepsy Structural Sequences

(HARNESS‐MRI) protocol (as described in Section 2.2.2).

2.1

Who should have an MRI?

Once the first seizure occurs, recurrence will depend on nu-

merous factors. Compared to patients in whom the cause is

unknown, the rate of seizure recurrence increases twofold in

those with a lesion on MRI, from 10% to 26% at 1year and

from 29% to 48% at 5years.

Numerous studies have related

the presence and types of MRI abnormalities to clinical out-

comes. In a cohort of 764 patients undergoing MRI at the

time or soon after a first seizure, 23% had a potentially epi-

leptogenic lesion, including stroke, trauma, a developmental

abnormality, or a tumor.

Another showed that patients with

focal epilepsy and unremarkable MRI have a 42% chance

to have their seizures controlled with antiepileptic drugs,

whereas this is true in 54% of cases with poststroke epilepsy;

conversely, seizure control with medication was achieved in

<10% of patients with hippocampal sclerosis on MRI.

2.1.1

First seizure

Data from the World Health Organization show that com-

puted tomography (CT) is widely available in hospitals

worldwide.

Evidence‐based guidelines of the therapeutics

and technology assessment subcommittee of the American

Academy of Neurology

recommend immediate noncontrast

CT in emergency patients presenting with a first seizure to

guide appropriate acute management, especially in those with

abnormal neurological examination, predisposing history, or

focal seizure onset. In these situations, there is great poten-

tial for pathology that may require immediate management,

such as a hemorrhage or large mass. Notably, noncontrast CT

can detect some tumors, large arteriovenous malformations,

stroke, and calcified lesions. CT with contrast is indicated

in cases with suspicion for infection or small neoplasms (in-

cluding metastases),

if MRI is unavailable.

In accordance with a recent ILAE publication,

the

Neuroimaging Task Force advises that the HARNESS‐MRI

protocol should be done soon after the first seizure, if re-

sources allow; this will help establish a syndromic defini-

tion and guiding management. MRI has high sensitivity and

specificity

for developmental cortical malformations, in-

cluding focal cortical dysplasia (FCD), and mesiotemporal

sclerosis, a group of prevalent structural lesions associated

with increased risk of drug resistance.

32‒34

Notably, an early

MRI is particularly important in young children, as ongoing

myelination may mask the appearance of FCDon later scans;

in these cases, conclusions may be misleading with respect to

diagnosis and appropriateness of surgical treatment.

2.1.2

Newly diagnosed epilepsy

The identification of a structural lesion in recent onset epi-

lepsy is a strong indicator of drug resistance and should be

1057

BERNASCONI EtAl.

an incentive to strictly adhere to the ILAE criteria for drug

resistance.

In other words, once a lesion is discovered on

MRI, a patient should be referred to a specialized epilepsy

surgery center to evaluate surgical candidacy.

While, a non-

progressive brain lesion may be associated with response to

antiepileptic drugs, a recent prospective longitudinal cohort

study showed that patients with mild mesial temporal lobe

epilepsy (TLE) and hippocampal sclerosis seen on MRI early

in the course of the disease have three times higher likelihood

of becoming refractory than those without such lesion.

meta‐analysis showed that odds of becoming seizure‐free

after surgery were 2.5 times higher in patients with MRI‐de-

fined lesions.

Moreover, >60% of patients with drug‐resist-

ant frontal lobe epilepsy achieve postsurgical seizure freedom

if operated within 5years of disease onset compared to only

30% when surgery is delayed.

This body of evidence should

become knowledge for every practicing neurologist, because

epilepsy surgery remains largely underutilized, with only a

fraction of patients being evaluated in specialized tertiary

centers.

9,11,40,41

Moreover, drug‐resistant epilepsy is asso-

ciated with increased risk of injury and mortality, affective

disturbances, and cognitive decline.

Deferring surgery may

thus cost the patient chances of seizure freedom, cognitive

benefits, and years of life expectancy.

There is currently insufficient evidence to recommend

the systematic use of MRI in patients with genetic general-

ized syndromes, such as juvenile myoclonic epilepsy, and

self‐limited drug‐responsive syndromes, such as childhood

epilepsy with centrotemporal spikes. Although neuroimag-

ing studies demonstrate structural and functional anomalies

in these epilepsies,

5,43

their prognostic value remains to be

determined. Notably, focal epilepsy may mimic generalized

syndromes; in these cases, the HARNESS‐MRI protocol is

recommended in the presence of atypical features such as ab-

normal neurologic development, cognitive decline, difficult‐

to‐treat seizures, or focal interictal epileptic spikes.

The Neuroimaging Task Force acknowledges that in re-

source‐limited areas MRI may not be readily obtainable

; in

this scenario, a CT scan would be the examination of choice

awaiting future availabilities.

2.1.3

The importance of repeating the MRI

The MRI should be repeated using the HARNESS‐MRI

protocol if images from a previous examination are not

available or the type and quality of previous acquisitions

are suboptimal. Relying on a written radiological report

may be insufficient, as putative anomalies may have been

overlooked due to poor image quality or lack of the read-

er's expertise in neuroimaging of epilepsy.

Importantly,

images should be evaluated in light of the evolving elec-

troclinical picture, particularly an unexplained increase in

seizure frequency (ie, not related to toxic‐metabolic factors,

medication compliance, etc), rapid cognitive decline, or ap-

pearance/worsening of neuropsychiatric symptoms. Given

the evidence for progressive brain atrophy developing over

1‐3years in both patients with refractory and patients with

well‐controlled seizures,

37,44‒46

repeated MRI may have

prognostic value. In drug‐resistant TLE, progressive atro-

phy of the neocortex and mesiotemporal lobe structures is

associated with poor outcome after surgery.

44,47

Finally,

the diagnostic yield depends heavily upon logistics, includ-

ing image resolution, magnetic field strength, number of

phased‐array head coils, and expertise of the reader.

is thus utterly important to repeat the examination with an

optimized protocol,

particularly in patients with drug‐re-

sistant epilepsy and previous “normal” MRI, as this may

reveal a lesion in 30%‐65% of cases

49‒51

; when MRI is

combined with image postprocessing, sensitivity may be as

high as 70%,

thereby significantly improving clinical de-

cisionmaking. Notably, imaging in the first year of life may

be helpful in identifying FCD associated with very subtle

signal changes on later images of the postmyelinated, ma-

tured brain and should be retained for comparison.

2.2

What are the minimum requirements

for an epilepsy protocol?

It is the consensus of the Neuroimaging Task Force that

neuroimaging workup of patients with epilepsy requires a

minimum set of MRI basic sequences that are available on

most MR scanners, and thus generalizable, regardless of

the clinical setting and country. Beyond the Neuroimaging

Task Force, previous independent expert opinion has un-

derlined the importance of high spatial resolution and

image contrast with complete brain coverage to optimally

appraise brain anatomy, the interface between gray mat-

ter and white matter, and signal anomalies. In particular,

three‐dimensional (3D) sequences with isotropic voxels

(ie, cube‐shaped voxels of identical length on each side or

image plane) of 1mm or less dramatically reduce partial

volume effects, a phenomenon resulting from the presence

of multiple tissue types within a given voxel. Notably, par-

tial volume is detrimental when looking for subtle cortical

dysplasia, as it mimics tissue blurring, a cardinal feature of

these lesions.

2.2.1

Previous MRI protocols:

Summary and limitations

The original guidelines established two decades ago by

the ILAE proposed T1‐ and T2‐weighted MRI with the

minimum slice thickness possible, acquired in two or-

thogonal planes (axial and coronal), and a 3D volumet-

ric T1‐weighted acquisition. To obtain 2D images with

whole‐brain coverage in a clinically acceptable time, it was

1058

BERNASCONI EtAl.

necessary to apply interslice gaps of 3‐5 mm. Moreover,

epilepsy protocols were divided according to clinical syn-

dromes into temporal and extratemporal with a series of

coronal, axial, and sometimes sagittal cuts, a strategy still

in practice in many institutions. Initial volumetric 3D se-

quences, obtained on 1.5T scanners, were only possible for

T1‐weighted sequences, with slice thickness varying be-

tween 1 and 3mm, rarely acquired with isotropic voxels,

because of either time or hardware constraints. Notably,

whereas in 3D acquisitions with isotropic voxels, thick-

ness and resolution are interchangeable quantities, for 2D

images the in‐plane voxel dimension (not slice thickness)

defines image resolution. To achieve finer in‐plane resolu-

tions (≤1mm), one had to reduce the size of the field of

view or introduce interslice gaps, thus sacrificing whole‐

brain coverage, with the risk of missing lesions.

2.2.2

Harmonized Neuroimaging of

Epilepsy Structural Sequences

The advent of high‐field magnets at 3 T, combined with the

use of multiple phased arrays instead of conventional quad-

rature coils, has resulted in accelerated image acquisition,

improved signal‐to‐noise ratio, and increased image contrast.

Importantly, 3D MR images with isotropic voxel resolution

and no interslice gap eliminate the need for syndrome‐spe-

cific protocols, as images can be reformatted and inspected

in any plane with equal resolution. Additional considerations

for optimal imaging include comfortable padding of the head

with foam cushions to minimize motion artifacts and center-

ing the head in the coil prior to starting the acquisition. Head

positioning can be verified on the scout image (or “local-

izer”) done at the beginning of the session. Any tilt or rotation

should be corrected for planning of the subsequent sequences

and later side‐by‐side analysis of brain structures; this is par-

ticularly important when acquiring 2D coronal T2‐weighted

images, as specified below. Sedation‐related recommenda-

tions have been discussed in a special report published by

the ILAE subcommittee for pediatric neuroimaging in 2009.

The Neuroimaging Task Force proposes HARNESS‐

MRI, a core structural MRI protocol comprising three acqui-

sitions. The HARNESS‐MRI protocol is applicable to adults

and children alike. It is time‐effective, as each sequence lasts

7‐10minutes, for a total time not exceeding 30minutes when

using multiple phased‐array coils (8, 12, or 32 channels)

with accelerated parallel imaging (eg, GRAPPA, ASSET,

SENSE). Table 1 presents key points regarding the protocol.

The HARNESS‐MRI protocol is optimized for 3T scanners,

if available. Notably, although it is possible to obtain this pro-

tocol on new generations of 1.5T systems, the overall image

quality may be inferior.

Suggested acquisition parameters for the HARNESS‐

MRI protocol on a 3T scanner are shown in Material S2. The

Neuroimaging Task Force recommends all patients in whom

previous investigations were unremarkable to undergo a repeated

scan using the HARNESS‐MRI protocol. Even in patients in

whom seizures are associated with other conditions, such as

head trauma, neurodegenerative disorders, multiple sclerosis, or

alcoholism, the HARNESS‐MRI protocol can be used, as it con-

tains basic sequences that are available on most MR scanners.

High‐resolution 3D T1‐weighted MRI

The magnetization‐prepared rapid gradient‐echo (MP‐

RAGE) sequence (Figure 1), as well as the equivalent 3D

spoiled gradient echo and 3D turbo field echo protocols with

isotropic millimetric voxel resolution (ie, 1×1×1mm

, no

interslice gap) are the most popular 3D T1‐weighted gradient

echo sequences. They allow for optimal evaluation of brain

anatomy and morphology.

High‐resolution 3D fluid‐attenuated inversion recovery

This 3D fluid‐attenuated inversion recovery (FLAIR) se-

quence (named CUBE, VISTA or SPACE, depending on the

MR vendor) is best suited for assessing signal anomalies, in

particular hyperintensities related to gliosis and increased

extracellular space (Figure 1). Compared to conventional

T2‐weighted contrasts, the nulling of cerebrospinal fluid

(CSF) signal enhances the visibility of hyperintense corti-

cal lesions. This acquisition should also be acquired with

isotropic millimetric voxel resolution (ie, 1×1×1mm

)

and no interslice gap. Because limbic structures are inher-

ently hyperintense,

FLAIR may not be sensitive to detect

very subtle hippocampal sclerosis. Moreover, FLAIR im-

ages are not sensitive to epilepsy‐associated pathology in

neonates and infants before 24months, as myelination is

not yet complete.

TABLE 1 Key points summarizing the main advantages of the

HARNESS‐MRI protocol

High‐contrast, 3D sequences with isotropic voxels (ie, identical

dimensions across planes)

Can be obtained on 1.5T and 3T scanners

Applicable to adults and children

Provide complete brain coverage

No need for operator‐dependent slice angulations

Images may be reformatted in any plane without loss of

resolution

Greatly reduce partial volume effect (ie, multiple tissue types

present within a given voxel)

Provide improved signal‐to‐noise ratio and tissue contrast

Allow for accelerated image acquisition (GRAPPA, ASSET,

SENSE)when using multiple phased-arrays head coils

Abbreviations: 3D, three‐dimensional; HARNESS‐MRI, Harmonized

Neuroimaging of Epilepsy Structural Sequences;GRAPPA, generalized

autocalibrating partial parallel acquisition; ASSET, array coil spatial sensitivity

encoding; SENSE, sensitivity encoding.

1059

BERNASCONI EtAl.

High in‐plane resolution 2D coronal T2‐weighted MRI

This turbo spin echo sequence is the examination of choice

for assessing the hippocampal internal structure, given that

images are acquired perpendicular to the long axis of the

hippocampus and using submillimetric voxel resolution (eg,

0.4×0.4×2mm, no interslice gap; Figure 2). Notably, the

densely myelinated molecular layer appearing as a dark rib-

bon inside the hippocampus allows discriminating Cornu

Ammonis (CA) subfields from the dentate gyrus.

When a tumor, vascular malformation, or infectious pro-

cess is suspected, the HARNESS‐MRI protocol should be

complemented by T1 MRI with gadolinium to look for con-

trast enhancement and susceptibility‐weighted imaging and

T2* contrasts sensitive to venous blood, hemorrhage, iron

deposits, and calcifications.

2.3

How should MR images be evaluated?

To embrace the multidisciplinary facets of disease diagnos-

tics, epileptologists should be given the opportunity to train

and receive continued medical education in neuroimaging.

Even with an appropriate MRI protocol, the interpretation

strongly depends on the reader's expertise in imaging of

epilepsy.

Notably, in‐depth inspection, particularly when

dealing with small cortical dysplasias or subtle hippocampal

sclerosis, requires significant time investment. Importantly,

optimal sensitivity for lesion detection is achieved when the

reader has access to a detailed description of the electroclini-

cal findings, including the suspected hemisphere and lobe,

information oftentimes missing in the radiology requisition.

In some cases, particularly at disease onset, it may be diffi-

cult to establish the exact syndromic classification. In light

of new electroclinical data or information derived from any

other test, the epileptologist may be best positioned to evalu-

ate previous scans or decide to repeat them, if necessary.

Because of the large number of MRI cuts, instead of in-

specting the original native high‐resolution format, some

radiologists may decide to inspect images that have been

reconstructed into thicker slices. For instance, 1 mm

iso-

tropic resolution T1 or FLAIR may be reformatted at 3‐mm

thickness, at times with interslice gaps that further reduce the

number of slices to inspect, from approximately 170 to less

than 50. This process is detrimental and counteracts the pur-

pose of 3D MRI, as it generates lower resolution images and

accentuates partial volume effects, potentially masking sub-

tle lesions (Figure 3). Visualization techniques, such as the

widely used clinical picture archiving and communication

systems (PACS) as well as several freely available imaging

platforms, have greatly facilitated the inspection of 3D MRI

by allowing time‐effective simultaneous inspection of images

in all three orthogonal planes (coronal, axial, and sagittal).

These platforms also allow viewing different MRI contrasts

side by side and evaluating both morphology and signal, as

co‐occurring anomalies increase diagnostic confidence.

FIGURE 1 HARNESS‐MRI (Harmonized Neuroimaging of Epilepsy Structural Sequences) three‐dimensional (3D) protocol at 3 T. T1‐

weighted and fluid‐attenuated inversion recovery (FLAIR) images are shown, with representative axial, coronal, and sagittal cuts with millimetric

resolution

1060

BERNASCONI EtAl.

The following paragraphs give a short overview on the

main criteria for the visual inspection of prevalent epilepto-

genic lesions associated with drug‐resistant epilepsy.

2.3.1

Visual MRI analysis in temporal

lobe epilepsy

In temporal lobe epileps (TLE), the most frequent histopatho-

logical finding is mesiotemporal sclerosis (MTS) character-

ized by cell loss and astrocytic gliosis.

These features are

not limited to the hippocampus, but are often found in the

amygdala, entorhinal cortex, temporopolar cortex, and the

temporal lobe.

On MRI, typical MTS is characterized by

anomalies more easily appreciated in the hippocampus proper,

including atrophy, loss of internal structure, and decreased T1

and increased T2 signal intensity. Additional features may in-

clude atrophy of the ipsilateral fornix, mammillary body, and

temporal lobe, particularly the pole. Inspection of coronal

sections allows for side‐by‐side comparison of asymmetry in

volume, shape, and signal, whereas sagittal images provide a

complete anteroposterior view, facilitating appraisal of pat-

terns of signal distribution within the hippocampus and para-

hippocampus. Field strengths at 3 T (and above) allow visual

evaluation of the internal architecture of the hippocampus

and thus better appreciation of subtle volume loss within in-

dividual subfields, particularly CA1, and CA4–dentate gyrus.

In addition, the molecular layer, a band of white matter run-

ning through the CA regions and dentate gyrus, may become

thin and blurred, a characteristic seen on T2‐weighted images

(Figure 4A). Besides atrophy and signal changes, about 40%

of patients with TLE present with malrotation characterized

by an abnormally round and vertically orientated hippocam-

pus, and a deep collateral sulcus.

This neurodevelopmental

shape variant occurs more frequently in the left hemisphere

and may be misinterpreted as hippocampal atrophy. Although

it is more prevalent in patients than in healthy controls, its re-

lation to epileptogenicity remains unclear.

Encephaloceles of the temporal pole

60‒62

and parahippo-

campal dysplasia

may be underdiagnosed, treatable causes

of refractory TLE. Encephaloceles present as a herniation

of brain tissue through a defect in the skull base, often the

greater wing of the sphenoid bone. Their detection is facil-

itated by high‐resolution 3D sequences and signal hyperin-

tensity on T2 and FLAIR; high‐resolution CT confirms the

bony defects in the inner table of the skull. Parahippocampal

dysplasia is characterized by prevailing white matter signal

anomalies, without apparent increased in cortical thickness.

Because of the presence of nearby blood vessels, this lesion

FIGURE 2 HARNESS‐MRI (Harmonized Neuroimaging of Epilepsy Structural Sequences) two‐dimensional (2D) protocol at 3 T. Coronal

T2‐weighted images at submillimetric in‐plane resolution cover the entire extent of the temporal lobes and hippocampi. Representative cuts at the

level of the hippocampal head (Ant), body (Mid), and tail (Post) are shown. Slices are acquired perpendicular to the long axis of the hippocampus

as shown in the sagittal view to optimize the evaluation of the hippocampal internal structure. In the magnified panel, one can appreciate the

densely myelinated molecular layer of the Cornu Ammonis (CA) and dentate gyrus fused across the hippocampal sulcus appearing as a dark ribbon,

which allows discriminating these compartments

1061

BERNASCONI EtAl.

may be mislabeled as flow or partial volume artifacts, if the

MRI cuts are thick. An in‐depth inspection of the temporal

lobe should also include the periventricular zone, in search of

nodular heterotopia, a cortical malformation often associated

with drug‐resistant TLE.

2.3.2

Visual MRI analysis of focal

cortical dysplasia

Focal cortical dysplasia(FCD) isa prevalent cause of medi-

cally intractable epilepsy and among the most frequent histo-

logical finding in patients undergoing epilepsy surgery.

The

past decades have witnessed numerous attempts to provide

a histological grading system. Currently, FCD are classified

into three types (I‐III) and several subtypes (eg, types IIA and

IIB) based on a combination of architectural alterations of

cortical layers either alone (type I, type III) or together with

cell overgrowth and morphological aberrations, including

giant dysmorphic neurons (type IIA) and balloon cells (type

IIB).

Gliosis and demyelination are also seen in the lesion

and underlying white matter. The MRI signature of FCD type

I remains unclear. Conversely, FCD type II is mainly char-

acterized by increased cortical thickness and blurring of the

gray‐white matter interface on T1‐weighted MRI in 50%‐90%

of cases. Analysis of T2‐weighted MRI, particularly FLAIR,

reveals gray matter hyperintensity in up to 100% of patients.

In many patients, however, FCD type II features may be

very subtle, and the MRI may be consequently reported as

unremarkable (Figure 4B).

In these cases, the inspection of

axial slices allows for side‐by‐side comparisons in search for

asymmetries in sulcogyral patterns. This is particularly im-

portant, as small FCD lesions may be preferentially located

at the bottom of deep sulci.

The transmantle sign, a fun-

nel‐shaped signal extending from the ventricular wall to the

neocortex harboring the lesion, may be the first feature to

attract the observer's attention toward a small FCD lesion,

underlying the importance of systematical inspection of the

white matter.

FIGURE 3 Image resampling versus original resolution. Axial 3T three‐dimensional (3D) fluid‐attenuated inversion recovery (FLAIR)

images of a patient with histologically proven focal cortical dysplasia type II. Upper panels: The radiological evaluation was initially done

on images reconstructed from the original 3D high‐resolution 1‐mm acquisition into 3‐mm thickslabs. This examination was reported as

unremarkable. Lower panels: The repeated inspection of the original (native) 3D high‐resolution 1‐mm isotropic images revealed the initially

overlooked subtle dysplasia characterized by blurring of the lesional boundaries (seen on all the slices, as indicated by the arrows) and a minute

transmantle sign (arrowheads)

1062

BERNASCONI EtAl.

2.4

How to optimize lesion detection with

MRI postprocessing?

Despite technical advances, routine visual MRI inspection

does not permit a diagnosis with a sufficient degree of con-

fidence in 30%‐50% of cases, or is simply unremarkable,

even though a lesion is found on histology.

This clinical

conundrum, currently one of the main barriers to effective

epilepsy surgery, has motivated the development of com-

puter‐aided methods aimed at quantitatively analyzing mor-

phology and signal of 3D MR images.

12,67‒69

However, there

are a number of basic steps in data preparation, namely,

correction for image intensity nonuniformities, registration,

and tissue segmentation, that need to be carefully evalu-

ated by the user, as their quality greatly influences final

results. For instance, subject motion negatively impacts tis-

sue segmentation and leads to artifacts that mimic lesions,

including atrophy. Another important point is performance

evaluation. Ideally, metrics derived from MRI postprocess-

ing should be sensitive and specific (ie, identify correctly

affected and unaffected subjects, respectively), and repro-

ducible (ie, consistent between repeated measures). Such

rigorous standards are essential to guarantee clinical valid-

ity of these advanced image analysis techniques.

52,70

The following paragraphs give a short overview of image

analysis methods for the detection of MTS and FCD. Theuse of

these algorithms is encouraged, as there is mounting evidence

for their ability to reveal subtle lesions that previously eluded

visual inspection, particularly when applied to 3D millimetric

or submillimetric isotropic multicontrast images.

52,71‒74

2.4.1

Volumetry and shape modeling of

mesiotemporal lobe structures

Manual volumetry performed on T1‐weighted anatomical MRI

has shown increased sensitivity of detecting hippocampal at-

rophy compared to visual MRI, particularly when values are

corrected for head size and normalized with respect to the distri-

bution in healthy controls. Volumetry of the entorhinal cortex,

amygdala, and temporopolar region, as well as the thalamus,

may lateralize the seizure focus, particularly in patients with

normal hippocampal volume.

Importantly, the degree of MRI

volume loss has been shown to correlate with the degree of cell

loss on surgical specimens.

Thus, hippocampal volumetry

FIGURE 4 The magnetic resonance imaging (MRI) spectrum of epileptogenic lesions. A, Coronal T1‐ and T2‐weighted 3T MRI in two cases

with drug‐resistant temporal lobe epilepsy and histologically confirmed hippocampal sclerosis. In the MRI‐positive case, the right hippocampus is

clearly atrophic and shows T1 hypo‐ and T2 hyperintensity (arrows). In the case initially reported as “MRI‐negative,” careful examination of the

T2‐weighted MRI shows a subtle T2 signal hyperintensity across the left CA 1‐3 regions. Moreover, compared to the contralateral side, the dark

ribbon representing the molecular layer is blurred, making the distinction between the CA subfields and the dentate gyrus difficult to appreciate (see

magnified panel). B, Axial T1‐ and T2‐weighted 3T MRI in two cases with drug‐resistant left frontal lobe epilepsy and histologically confirmed

focal cortical dysplasia type II. In the MRI‐positive case, there is cortical thickening and blurring of the gray‐white matter transition in the left

superior frontal gyrus (arrows). In the case initially reported as “MRI‐negative,” reexamination of the FLAIR images shows a subtle blurring at the

bottom of a sulcus (arrowhead), which is difficult to discern on T1‐weighted images

1063

BERNASCONI EtAl.

is part of the minimal requirement when considering epilepsy

surgery to lateralize the focus and establish whether the con-

tralateral structures are normal. Bilateral mesial temporal lobe

atrophy raises concerns of markedly reduced chance of seizure

freedom after surgery

and an increased risk of memory im-

pairment.

Over the years, steady technical advances have

propelled the design of automated algorithms yielding segmen-

tation of the whole hippocampus (eg, Sone et al,

Hosseini

etal,

Kim et al

), and more recently hippocampal subfields,

thereby creating a solid basis for broad translation (Figure 5).

Several US Food and Drug Administration (FDA)–approved

commercial software packages are currently available for rou-

tine use in clinical practice and provide an automated report

that details the volume and percentile of each parcellated corti-

cal region compared to a normative database. They have been

used to lateralize hippocampal atrophy in TLE patients with

accuracy rates that exceed visual inspection.

Notably, hip-

pocampal labels may be used to examine structural alterations

through statistical parametric surface shape modeling,

82,83

fur-

ther increasing sensitivity.

2.4.2

Hippocampal T2 relaxometry

Compared to visual analysis of T2‐weighted MRI, T2 relax-

ometry,

84,85

a sequence providing quantitative estimates of the

T2‐weighted signal, yields increased sensitivity for detecting

FIGURE 5 Hippocampal subfield volumetry in temporal lobe epilepsy. Coronal T1‐ and T2‐weighted MRI at the level of hippocampal

body and 4‐μm‐thick paraffin‐embedded histology sections with NeuN immunohistochemistry at a comparable level in two patients with right

temporal focus. A, Volumes of subiculum (Sub; green), CA 1‐3 (red), and CA4–dentate gyrus (DG; blue) are >3 standard deviations (SD) below

themean of healthy controls, and pathology shows severe panhippocampal neuronal loss. B, Volumetry detected subtle CA1‐3 atrophy (−2.2 SD),

and histology shows CA1 minimal neuronal loss. In this “MRI‐negative” patient, conventional whole‐hippocampal volumetry was unremarkable,

highlighting the value of subfield volumetry. Scale bars=2mm

1064

BERNASCONI EtAl.

mesiotemporal gliosis.

Importantly, it correctly lateralizes

the focus in up to 80% of patients with normal hippocampal

volume.

Measurement of T2 relaxation times can be done by

placing a manually or automatically generated region of interest

within the hippocampus,

carefully avoiding the adjacent CSF.

2.4.3

Texture analysis

Voxel‐based modeling of gray‐white matter blurring and gray

matter intensity, derived from 3D T1 MRI, assists visual in-

spection and increases sensitivity for the detection of FCD

type II up to 40% relative to conventional MRI (Figure 6).

Analysis of these maps can be done either by normalizing (z

scoring) data within the same brain

or by comparing fea-

tures to a group of healthy controls.

Surface‐based methods

improve intersubject anatomical correspondence and allow

for multivariate analysis of MRI contrasts and features to un-

veil latent tissue properties not readily identified on a single

modality.

2.4.4

Fully automated lesion

detection techniques

Over the past 15years, a number of algorithms have been

developed for automated FCD detection. These methods

were initially based on morphology and signal derived

from 3D T1‐weighted MRI. More recent tools have incor-

porated 3D FLAIR.

90,91

A recent publication showed class

II evidence that machine learning of MRI patterns accu-

rately identifies FCD type II in >70% of patients in whom

the lesion had been overlooked by routine clinical visual

inspection.

FIGURE 6 Texture analysis of “MRI‐negative” focal cortical dysplasia. Three‐dimensional (3D) T1 and FLAIR axial, sagittal, and coronal

views in a patient with right frontal lobe epilepsy initially reported as “MRI‐negative” are shown. The last columnshows cuts of the 3D gradient

map obtained from the T1‐weighted MRI, which calculates the rate of change of intensities, thereby modeling blurring at the interface between the

gray and white matter. In regions of normal transition, the gradient is expected to be steep, thus appearing hyperintense. In regions of blurring, the

gradient is expected to be less steep, thus appearing hypointense. In this case, there is a clear breakdown in the gradient, with a hypointense region

within the right orbitofrontal region (outlined by the dashed rectangles). The reinspection of the T1‐ and T2‐weighted images, informed by the

texture map, reveals an extensive blurring in the same area initially overlooked by conventional radiological examination

1065

BERNASCONI EtAl.

CONCLUSION

MRI provides a unique, versatile, and noninvasive tool

for brain‐wide evaluation of patients with epilepsy.

Notwithstanding the relentless progress in hardware and ac-

quisition techniques, as well as methods for computational

analysis, any guideline is difficult to implement when re-

sources are scarce, and where technical infrastructure and

specialist training may not be available. The Neuroimaging

Task Force believes, nevertheless, that the proposed rec-

ommendations set a tangible basis for a consistent use of

structural MRI in epilepsy. By revealing lesions unseen

by conventional neuroradiology, the HARNESS‐MRI pro-

tocol combined with postprocessing has the potential to

transform MRI‐negative into MRI‐positive, thereby offer-

ing the life‐changing benefits of epilepsy surgery to more

patients.

Because of the transforming role of MRI in modern

epileptology, the forthcoming competency‐based ILAE

educational curriculum requires neurologists and epi-

leptologists to train in neuroimaging.

With the goal

of optimally meeting the needs of people with epilepsy,

the learning objectives will include acquiring a range

of skills, from basic MRI visual evaluation to advanced

training in image postprocessing. Notably, such train-

ing may also provide a unique opportunity to optimize

skills in neuroimaging of epilepsy for neuroradiologists.

Achieving this goal will require a combined effort from

ILAE and its regional chapters, medical societies, and

academies, universities, and centers that offer epilepsy

fellowship training. Concretesteps toward this objective

are the ILAE‐endorsed courses on neuroimaging of epi-

lepsy currently offered around the globe and online edu-

cational platforms.

DISCLOSURE

None of the authors has any conflict of interest to disclose.

We confirm that we have read the Journal's position on issues

involved in ethical publication and affirm that this report is

consistent with those guidelines.

ORCID

Andrea Bernasconi https://orcid.org/0000-0001-9358-5703

Fernando Cendes https://orcid.org/0000-0001-9336-9568

William H. Theodore https://orcid.org/0000-0002-4669-5747

Angelo Labate https://orcid.org/0000-0002-8827-7324

Philippe Ryvlin https://orcid.

org/0000-0001-7775-6576

REFERENCES

1. Jones AL, Cascino GD. Evidence on use of neuroimaging for sur-

gical treatment of temporal lobe epilepsy: a systematic review.

JAMA Neurol. 2016;73:464–70.

2. Commission on Neuroimaging of the International League Against

Epilepsy. Recommendations for neuroimaging of patients with ep-

ilepsy. Epilepsia. 1997;38:1255–6.

3. Commission on Neuroimaging of the International League Against

Epilepsy. Guidelines for neuroimaging evaluation of patients

with uncontrolled epilepsy considered for surgery. Epilepsia.

1998;39:1375–6.

4. Neuroimaging Subcommision of the International League Against

Epilepsy. Commission on Diagnostic Strategies: recommendations

for functional neuroimaging of persons with epilepsy. Epilepsia.

2000;41:1350–6.

5. Gaillard WD, Chiron C, Cross JH, et al. Guidelines for imag-

ing infants and children with recent‐onset epilepsy. Epilepsia.

2009;50:2147–53.

6. Wilmshurst JM, Gaillard WD, Vinayan KP, et al. Summary of

recommendations for the management of infantile seizures: Task

Force Report for the ILAE Commission of Pediatrics. Epilepsia.

2015;56:1185–97.

7. Engel J Jr, Wiebe S, French J, et al. Practice parameter: temporal

lobe and localized neocortical resections for epilepsy. Epilepsia.

2003;44:741–51.

8. Kwan P, Arzimanoglou A, Berg AT, et al. Definition of drug re-

sistant epilepsy: consensus proposal by the ad hoc Task Force

of the ILAE Commission on Therapeutic Strategies. Epilepsia.

2010;51:1069–77.

9. Jehi L, Mathern GW. Who's responsible to refer for epilepsy sur-

gery? We all are! Neurology. 2015;84:112–3.

10. Haneef Z, Stern J, Dewar S, Engel J Jr. Referral pattern for epilepsy

surgery after evidence‐based recommendations: a retrospective

study. Neurology. 2010;75:699–704.

11. Roberts JI, Hrazdil C, Wiebe S, et al. Neurologists’ knowledge of

and attitudes toward epilepsy surgery: a national survey. Neurology.

2015;84:159–66.

12. Bernasconi A, Bernasconi N, Bernhardt BC, Schrader D.

Advances in MRI for ‘cryptogenic’ epilepsies. Nat Rev Neurol.

2011;7:99–108.

13. So EL, Lee RW. Epilepsy surgery in MRI‐negative epilepsies. Curr

Opin Neurol. 2014;27:206–12.

14. Duncan JS, Winston GP, Koepp MJ, Ourselin S. Brain im-

aging in the assessment for epilepsy surgery. Lancet Neurol.

2016;15:420–33.

15. Craven IJ, Griffiths PD, Bhattacharyya D, et al. 3.0 T MRI of 2000

consecutive patients with localisation‐related epilepsy. Br J Radiol.

2000;2012(85):1236–42.

16. Craven I, Griffiths PD, Hoggard N. Magnetic resonance imaging of

epilepsy at 3 Tesla. Clin Radiol. 2011;66:278–86.

17. Jayakar P, Gaillard WD, Tripathi M, et al. Diagnostic test utili-

zation in evaluation for resective epilepsy surgery in children.

Epilepsia. 2014;55:507–18.

18. Griffiths PD, Coley SC, Connolly DJ, et al. MR imaging of patients

with localisation‐related seizures: initial experience at 3.0T and

relevance to the NICE guidelines. Clin Radiol. 2005;60:1090–9.

19. Stylianou P, Kimchi G, Hoffmann C, Blat I, Harnof S. Neuroimaging

for patient selection for medial temporal lobe epilepsy surgery. Part

2: Functional neuroimaging. J Clin Neurosci. 2016;23:23–33.

1066

BERNASCONI EtAl.

20. Stylianou P, Hoffmann C, Blat I, Harnof S. Neuroimaging for pa-

tient selection for medial temporal lobe epilepsy surgery. Part 1:

Structural neuroimaging. J Clin Neurosci. 2016;23:14–22.

21. Duncan J. The current status of neuroimaging for epilepsy. Curr

Opin Neurol. 2009;22:179–84.

22. Wellmer J, Quesada CM, Rothe L, Elger CE, Bien CG, Urbach H.

Proposal for a magnetic resonance imaging protocol for the detec-

tion of epileptogenic lesions at early outpatient stages. Epilepsia.

2013;54:1977–87.

23. Krumholz A, Wiebe S, Gronseth GS, et al. Evidence‐based guide-

line: management of an unprovoked first seizure in adults: report

of the Guideline Development Subcommittee of the American

Academy of Neurology and the American Epilepsy Society.

Neurology. 2015;84:1705–13.

24. Von Oertzen J, Urbach H, Jungbluth S, et al. Standard magnetic

resonance imaging is inadequate for patients with refractory focal

epilepsy. J Neurol Neurosurg Psychiatry. 2002;73:643–7.

25. Ruber T, David B, Elger CE. MRI in epilepsy: clinical standard and

evolution. Curr Opin Neurol. 2018;31:223–31.

26. Hakami T, McIntosh A, Todaro M, et al. MRI‐identified pathology

in adults with new‐onset seizures. Neurology. 2013;81:920–7.

27. Semah F, Picot MC, Adam C, et al. Is the underlying cause of

epilepsy a major prognostic factor for recurrence? Neurology.

1998;51:1256–62.

28. World Health Organization. Global maps for diagnostic imaging.

2014. [Cited 2019 Apr 17]. Available at: https ://www.who.int/

diagn ostic_imagi ng/colla borat ion/global_collab_maps/en/.

29. Harden CL, Huff JS, Schwartz TH, et al. Reassessment: neuro-

imaging in the emergency patient presenting with seizure (an ev-

idence‐based review): report of the Therapeutics and Technology

Assessment Subcommittee of the American Academy of

Neurology. Neurology. 2007;69:1772–80.

30. Middlebrooks EH, Ver Hoef L, Szaflarski JP. Neuroimaging in ep-

ilepsy. Curr Neurol Neurosci Rep. 2017;17:32.

31. Scheffer IE, Berkovic S, Capovilla G, et al. ILAE classification

of the epilepsies: position paper of the ILAE Commission for

Classification and Terminology. Epilepsia. 2017;58:512–21.

32. Wiebe S, Jette N. Pharmacoresistance and the role of surgery in

difficult to treat epilepsy. Nat Rev Neurol. 2012;8:669–77.

33. Berg AT, Berkovic SF, Brodie MJ, et al. Revised terminology and

concepts for organization of seizures and epilepsies: report of the

ILAE Commission on Classification and Terminology, 2005‐2009.

Epilepsia. 2010;51:676–85.

34. Blumcke I, Spreafico R, Haaker G, et al. Histopathological find-

ings in brain tissue obtained during epilepsy surgery. N Engl J

Med. 2017;377:1648–56.

35. Eltze CM, Chong WK, Bhate S, Harding B, Neville BG, Cross

JH. Taylor‐type focal cortical dysplasia in infants: some MRI le-

sions almost disappear with maturation of myelination. Epilepsia.

2005;46:1988–92.

36. Engel J Jr. What can we do for people with drug‐resistant epilepsy?

The 2016 Wartenberg Lecture. Neurology. 2016;87:2483–9.

37. Labate A, Aguglia U, Tripepi G, et al. Long‐term outcome of mild

mesial temporal lobe epilepsy: a prospective longitudinal cohort

study. Neurology. 2016;86:1904–10.

38. Tellez‐Zenteno JF, Hernandez Ronquillo L, Moien‐Afshari F,

Wiebe S. Surgical outcomes in lesional and non‐lesional ep-

ilepsy: a systematic review and meta‐analysis. Epilepsy Res.

2010;89:310–8.

39. Simasathien T, Vadera S, Najm I, Gupta A, Bingaman W, Jehi L.

Improved outcomes with earlier surgery for intractable frontal lobe

epilepsy. Ann Neurol. 2013;73:646–54.

40. Cloppenborg T, May TW, Blumcke I, et al. Trends in epilepsy sur-

gery: stable surgical numbers despite increasing presurgical vol-

umes. J Neurol Neurosurg Psychiatry. 2016;87:1322–9.

41. Burneo JG, Shariff SZ, Liu K, Leonard S, Saposnik G, Garg AX.

Disparities in surgery among patients with intractable epilepsy in a

universal health system. Neurology. 2016;86:72–8.

42. Wiebe S. Burden of intractable epilepsy. Adv Neurol. 2006;97:1–4.

43. Nickels KC, Zaccariello MJ, Hamiwka LD, Wirrell EC. Cognitive

and neurodevelopmental comorbidities in paediatric epilepsy. Nat

Rev Neurol. 2016;12:465–76.

44. Bernhardt BC, Kim H, Bernasconi N. Patterns of subregional

mesiotemporal disease progression in temporal lobe epilepsy.

Neurology. 2013;81:1840–7.

45. Caciagli L, Bernasconi A, Wiebe S, Koepp MJ, Bernasconi N,

Bernhardt BC. A meta‐analysis on progressive atrophy in in-

tractable temporal lobe epilepsy: time is brain? Neurology.

2017;89:506–16.

46. Bernhardt BC, Worsley KJ, Kim H, Evans AC, Bernasconi A,

Bernasconi N. Longitudinal and cross‐sectional analysis of at-

rophy in pharmacoresistant temporal lobe epilepsy. Neurology.

2009;72:1747–54.

47. Bernhardt BC, Bernasconi N, Concha L, Bernasconi A. Cortical

thickness analysis in temporal lobe epilepsy: reproducibility and

relation to outcome. Neurology. 2010;74:1776–84.

48. Winston GP, Micallef C, Kendell BE, et al. The value of repeat

neuroimaging for epilepsy at a tertiary referral centre: 16years of

experience. Epilepsy Res. 2013;105:349–55.

49. Kokkinos V, Kallifatidis A, Kapsalaki EZ, Papanikolaou N,

Garganis K. Thin isotropic FLAIR MR images at 1.5T increase

the yield of focal cortical dysplasia transmantle sign detection in

frontal lobe epilepsy. Epilepsy Res. 2017;132:1–7.

50. Mellerio C, Labeyrie MA, Chassoux F, et al. 3T MRI improves

the detection of transmantle sign in type 2 focal cortical dysplasia.

Epilepsia. 2014;55:117–22.

51. Knake S, Triantafyllou C, Wald LL, et al. 3T phased array MRI im-

proves the presurgical evaluation in focal epilepsies: a prospective

study. Neurology. 2005;65:1026–31.

52. Hong SJ, Kim H, Schrader D, Bernasconi N, Bernhardt BC,

Bernasconi A. Automated detection of cortical dysplasia type II in

MRI‐negative epilepsy. Neurology. 2014;83:48–55.

53. Adler S, Hong SJ, Liu M, et al. Topographic principles of cortical

fluid‐attenuated inversion recovery signal in temporal lobe epi-

lepsy. Epilepsia. 2018;59:627–35.

54. Bernasconi N, Bernasconi A. Epilepsy: imaging the epileptic

brain—time for new standards. Nat Rev Neurol. 2014;10:133–4.

55. Blumcke I, Thom M, Aronica E, et al. International consensus

classification of hippocampal sclerosis in temporal lobe epilepsy:

a task force report from the ILAE Commission on Diagnostic

Methods. Epilepsia. 2013;54:1315–29.

56. Thom M. Review: Hippocampal sclerosis in epilepsy: a neuropa-

thology review. Neuropathol Appl Neurobiol. 2014;40:520–43.

57. Kulaga‐Yoskovitz J, Bernhardt BC, Hong SJ, et al. Multi‐contrast

submillimetric 3 tesla hippocampal subfield segmentation protocol

and dataset. Sci Data. 2015;2:150059.

58. Bernasconi N, Kinay D, Andermann F, Antel S, Bernasconi A.

Analysis of shape and positioning of the hippocampal formation:

1067

BERNASCONI EtAl.

an MRI study in patients with partial epilepsy and healthy controls.

Brain. 2005;128:2442–52.

59. Tsai MH, Vaughan DN, Perchyonok Y, et al. Hippocampal mal-

rotation is an anatomic variant and has no clinical significance in

MRI‐negative temporal lobe epilepsy. Epilepsia. 2016;57:1719–28.

60. Abou‐Hamden A, Lau M, Fabinyi G, et al. Small temporal pole en-

cephaloceles: a treatable cause of “lesion negative” temporal lobe

epilepsy. Epilepsia. 2010;51:2199–202.

61. Toledano R, Jimenez‐Huete A, Campo P, et al. Small temporal pole

encephalocele: a hidden cause of “normal” MRI temporal lobe ep-

ilepsy. Epilepsia. 2016;57:841–51.

62. Saavalainen T, Jutila L, Mervaala E, Kälviäinen R, Vanninen

R, Immonen A. Temporal anteroinferior encephalocele: an un-

derrecognized etiology of temporal lobe epilepsy? Neurology.

2015;85:1467–74.

63. Pillay N, Fabinyi GC, Myles TS, Fitt GJ, Berkovic SF, Jackson

GD. Parahippocampal epilepsy with subtle dysplasia: a cause of

“imaging negative” partial epilepsy. Epilepsia. 2009;50:2611–8.

64. Thompson SA, Kalamangalam GP, Tandon N. Intracranial evalu-

ation and laser ablation for epilepsy with periventricular nodular

heterotopia. Seizure. 2016;41:211–6.

65. Blumcke I, Thom M, Aronica E, et al. The clinicopathologic spec-

trum of focal cortical dysplasias: a consensus classification pro-

posed by an ad hoc task force of the ILAE Diagnostic Methods

Commission. Epilepsia. 2011;52:158–74.

66. Besson P, Andermann F, Dubeau F, Bernasconi A. Small focal cor-

tical dysplasia lesions are located at the bottom of a deep sulcus.

Brain. 2008;131:3246–55.

67. Martin P, Bender B, Focke NK. Post‐processing of structural

MRI for individualized diagnostics. Quant Imaging Med Surg.

2015;5:188–203.

68. Kini LG, Gee JC, Litt B. Computational analysis in epilepsy neu-

roimaging: a survey of features and methods. Neuroimage Clin.

2016;11:515–29.

69. Bernasconi A, Bernasconi N. Unveiling epileptogenic lesions:

the contribution of image processing. Epilepsia. 2011;52(Suppl

4):20–4.

70. Gaillard WD, Cross JH, Duncan JS, Stefan H, Theodore WH, Task

Force on Practice Parameter Imaging Guidelines for International

League Against Epilepsy, Commission for Diagnostics. Epilepsy

imaging study guideline criteria: commentary on diagnos-

tic testing study guidelines and practice parameters. Epilepsia.

2011;52:1750–6.

71. Bernasconi A, Antel SB, Collins DL, et al. Texture analysis and

morphological processing of magnetic resonance imaging assist

detection of focal cortical dysplasia in extra‐temporal partial ep-

ilepsy. Ann Neurol. 2001;49:770–5.

72. Colliot O, Bernasconi N, Khalili N, Antel SB, Naessens V,

Bernasconi A. Individual voxel‐based analysis of gray matter in

focal cortical dysplasia. Neuroimage. 2006;29:162–71.

73. Wagner J, Weber B, Urbach H, Elger CE, Huppertz HJ.

Morphometric MRI analysis improves detection of focal cortical

dysplasia type II. Brain. 2011;134:2844–54.

74. Huppertz HJ, Kurthen M, Kassubek J. Voxel‐based 3D MRI analy-

sis for the detection of epileptogenic lesions at single subject level.

Epilepsia. 2009;50:155–6.

75. Goubran M, Bernhardt BC, Cantor‐Rivera D, et al. In vivo MRI

signatures of hippocampal subfield pathology in intractable epi-

lepsy. Hum Brain Mapp. 2016;37:1103–19.

76. Aghakhani Y, Liu X, Jette N, Wiebe S. Epilepsy surgery in pa-

tients with bilateral temporal lobe seizures: a systematic review.

Epilepsia. 2014;55:1892–901.

77. Sone D, Sato N, Maikusa N, et al. Automated subfield volumetric

analysis of hippocampus in temporal lobe epilepsy using high‐reso-

lution T2‐weighed MR imaging. Neuroimage Clin. 2016;12:57–64.

78. Hosseini MP, Nazem‐Zadeh MR, Pompili D, et al. Comparative

performance evaluation of automated segmentation methods of

hippocampus from magnetic resonance images of temporal lobe

epilepsy patients. Med Phys. 2016;43:538.

79. Kim H, Mansi T, Bernasconi N, Bernasconi A. Surface‐based

multi‐template automated hippocampal segmentation: application

to temporal lobe epilepsy. Med Image Anal. 2012;16:1445–55.

80. Caldairou B, Bernhardt BC, Kulaga‐Yoskovitz J, Kim H,

Bernasconi A, Bernasconi N. A Surface Patch‐Based Segmentation

Method for Hippocampal Subfields. Cham, Switzerland: Springer

International Publishing; 2016:379–87.

81. Farid N, Girard HM, Kemmotsu N, et al. Temporal lobe epilepsy:

quantitative MR volumetry in detection of hippocampal atrophy.

Radiology. 2012;264:542–50.

82. Kim H, Bernhardt BC, Kulaga‐Yoskovitz J, Bernasconi A,

Bernasconi N. Multivariate hippocampal subfield analysis of local

MRI intensity and volume: application to temporal lobe epilepsy.

Med Image Comput Comput Assist Interv. 2014;17:170–8.

83. Bernhardt BC, Bernasconi A, Liu M, et al. The spectrum of struc-

tural and functional imaging abnormalities in temporal lobe epi-

lepsy. Ann Neurol. 2016;80:142–53.

84. Jackson GD, Kuzniecky RI, Cascino GD. Hippocampal scle-

rosis without detectable hippocampal atrophy. Neurology.

1994;44:42–6.

85. Kosior RK, Lauzon ML, Federico P, Frayne R. Algebraic T2 esti-

mation improves detection of right temporal lobe epilepsy by MR

T2 relaxometry. Neuroimage. 2011;58:189–97.

86. Peixoto‐Santos JE, Kandratavicius L, Velasco TR, et al.

Individual hippocampal subfield assessment indicates that

matrix macromolecules and gliosis are key elements for the

increased T2 relaxation time seen in temporal lobe epilepsy.

Epilepsia. 2017;58:149–59.

87. Bernasconi A, Bernasconi N, Caramanos Z, et al. T2 relaxometry

can lateralize mesial temporal lobe epilepsy in patients with nor-

mal MRI. Neuroimage. 2000;12:739–46.

88. Winston GP, Vos SB, Burdett JL, Cardoso MJ, Ourselin S, Duncan

JS. Automated T2 relaxometry of the hippocampus for temporal

lobe epilepsy. Epilepsia. 2017;58:1645–52.

89. Hong S‐J, Bernhardt BC, Caldairou B, et al. Multimodal

MRI profiling of focal cortical dysplasia type II. Neurology.

2017;88:734–42.

90. Gill RS, Hong SJ, Fadaie F, et al. Automated detection of epilepto-

genic cortical malformations using multimodal MRI. In: Cardoso

MJ, Arbel T, Carneiro G, et al, eds. Deep Learning in Medical

Image Analysis and Multimodal Learning for Clinical Decision

Support: Third International Workshop, DLMIA 2017, and 7th

International Workshop, ML‐CDS 2017, Held in Conjunction

with MICCAI 2017, Québec City, QC, Canada, September

14, Proceedings. Cham, Switzerland: Springer International

Publishing; 2017:349–56.

91. Adler S, Wagstyl K, Gunny R, et al. Novel surface features for auto-

mated detection of focal cortical dysplasias in paediatric epilepsy.

Neuroimage Clin. 2017;14:18–27.

1068

BERNASCONI EtAl.

92. Blumcke I, Arzimanoglou A, Beniczky S, Wiebe S. Roadmap for a

competency‐based educational curriculum in epileptology: report

of the Epilepsy Education Task Force of the International League

Against Epilepsy. Epileptic Disord. 2019;21(2):129–40.

SUPPORTING INFORMATION

Additional supporting information may be found online in

the Supporting Information section at the end of the article.

How to cite this article: Bernasconi A, Cendes F,

Theodore WH, etal. Recommendations for the use of

structural magnetic resonance imaging in the care of

patients with epilepsy: A consensus report from the

International League Against Epilepsy Neuroimaging

Task Force. Epilepsia. 2019;60:1054–1068. https ://doi.

org/10.1111/epi.15612