passive fmri mapping of language function for pediatric ... · their clinical evaluation at boston...
TRANSCRIPT
E
PfV
RCTS
a
b
c
d
RA
h0
pilepsy Research (2014) 108, 1874—1888
j ourna l h om epa ge: www.elsev ier .com/ locate /ep i lepsyres
assive fMRI mapping of language functionor pediatric epilepsy surgical planning:alidation using Wada, ECS, and FMAER
alph O. Suareza,∗, Vahid Taimouria, Katrina Boyerc,d,lemente Vegac,d, Alexander Rotenbergd, Joseph R. Madsenb,obias Loddenkemperd, Frank H. Duffyd, Sanjay P. Prabhua,imon K. Warfielda
Department of Radiology, Boston Children’s Hospital, Harvard Medical School, Boston, MA, USADepartment of Neurosurgery, Boston Children’s Hospital, Harvard Medical School, Boston, MA, USADepartment of Psychology, Boston Children’s Hospital, Harvard Medical School, Boston, MA, USADepartment of Neurology, Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA
eceived 7 May 2014; received in revised form 7 September 2014; accepted 13 September 2014vailable online 28 September 2014
KEYWORDSLanguage;Functional MRI;Pre-surgical planning;Pediatric epilepsy;Functional mapping
Summary In this study we validate passive language fMRI protocols designed for clinical appli-cation in pediatric epilepsy surgical planning as they do not require overt participation frompatients. We introduced a set of quality checks that assess reliability of noninvasive fMRI map-pings utilized for clinical purposes. We initially compared two fMRI language mapping paradigms,one active in nature (requiring participation from the patient) and the other passive in nature(requiring no participation from the patient). Group-level analysis in a healthy control cohortdemonstrated similar activation of the putative language centers of the brain in the inferiorfrontal (IFG) and temporoparietal (TPG) regions. Additionally, we showed that passive languagefMRI produced more left-lateralized activation in TPG (LI = +0.45) compared to the active task;with similarly robust left-lateralized IFG (LI = +0.24) activations using the passive task. We vali-
dated our recommended fMRI mapping protocols in a cohort of 15 pediatric epilepsy patients bydirect comparison against the invasive clinical gold-standards. We found that language-specificTPG activation by fMRI agreed to within 9.2 mm to subdural localizations by invasive functional mapping in the same patients, and language dominance by fMRI agreed with Wada test resultsat 80% congruency in TPG and 73% congruency in IFG. Lastly, we tested the recommendedpassive language fMRI protocols in a cohort of very young patients and confirmed reliable∗ Corresponding author. Tel.: +1 617 355 2755; fax: +1 617 730 4644.E-mail address: [email protected] (R.O. Suarez).
ttp://dx.doi.org/10.1016/j.eplepsyres.2014.09.016920-1211/© 2014 Elsevier B.V. All rights reserved.
Passive fMRI mapping of language function for pediatric epilepsy surgical planning 1875
language-specific activation patterns in that challenging cohort. We concluded that languageactivation maps can be reliably achieved using the passive language fMRI protocols we proposedeven in very young (average 7.5 years old) or sedated pediatric epilepsy patients.© 2014 Elsevier B.V. All rights res
(llsa
lpgaHastctfmwTaib
fnqivart
•
•
•
•
cortex, depending on the stimulus mode used).• Language-specific activation—–confirmation of activation
in the putative language centers of the inferior frontal
Introduction
The mapping of language is important in pediatric patientswho will undergo resection surgery near cortical regionsessential for language function. However, due to youngage, pediatric patients are typically unable to comply withcomplicated functional mapping protocols—–particularlyinvasive clinical gold-standards, as well as non-invasiveimagining techniques more easily used with adult popu-lations. A trend presently in the field promotes resectionsurgery to cure or prevent disease progression inincreasingly younger epilepsy patients. This creates anurgent demand for reliable functional language map-ping methods in characteristically incompliant patients,those often ineligible to undergo the invasive func-tional mapping gold-standards. In this article, we addressthis need by introducing functional magnetic resonanceimaging (fMRI) methodologies for mapping of languagein pediatric patients. Our approach evokes languageactivation without the need of overt compliance from thepatient, and as such is designed for presurgical planning inpediatric patients.
The goal of brain resection is maximal removal of dis-eased tissue while at the same time minimizing potentialdamage to the eloquent brain systems. Resection surgerycan raise the risk of damaging one or both languagecenters of the brain, resulting in permanent postsurgi-cal language deficit (Duffau et al., 2009; Loiselle et al.,2012; Matsuda et al., 2012). Determining the lateralizationand localization of the essential language centers in rela-tion to the surgical site therefore provides surgeons witha valuable tool for evaluating risk of postoperative mor-bidity associated with language deficit (Abou-Khalil, 2007;Sharan et al., 2011). Among the initial considerations inthis evaluation is establishing the language dominance ofthe patient. Language dominance is of particular impor-tance in patients for whom the planned surgical site is inthe left cerebral hemisphere, as the vast majority of thepopulation is left hemisphere dominant for language—–onthe order of 90% for the general healthy population (Wadaand Rasmussen, 1960; Binder et al., 1996; Knecht et al.,2000). A planned resection in the same cerebral hemisphereas the patient’s language dominance will consequentlyincrease the risk of postsurgical language impairment.In such cases, a more precise localization of eloquentcortical regions is often determined by awake func-tional mapping in the operating room. The clinicalgold-standards for determination of language lateral-ization and language localization are the intracarotidsodium amobarbiatal procedure (Wada test) and direct
eletrocortical stimulation (ECS) mapping, respectively(Wada and Rasmussen, 1960; Ojemann, 1993). Func-tional MRI of language is a noninvasive alternative usedwith increasing frequency as a noninvasive alternative werved.
Beers et al., 2012; Garrett et al., 2012). Clinical fMRIanguage mapping has the potential for providing analogousanguage mapping measures to that of the clinical gold-tandards noninvasively, and therefore is well suited forpplication in the youngest patient populations.
In previous work we demonstrated that an active fMRIanguage tasks, such as vocalized antonym-generation,roduces lateralization and localization of cortical lan-uage processing highly congruent with gold-standards indult epilepsy surgical patients (Suarez et al., 2009, 2010).owever, the mapping of language in pediatrics presentsdditional challenges. Children can be expected to havehorter attention spans to follow complicated cognitiveasks, and less tolerance for holding still during prolonged oromplicated paradigms. The ideal pediatric fMRI paradigmherefore should be simple for the patient to perform andast to acquire. In this study, we compare fMRI languageappings acquired using the active antonym-generation taske previously validated in adults (Suarez et al., 2008, 2009;ie et al., 2009), to a passive listening paradigm more ide-lly suited for application in pediatric clinical populations ast does not require any overt patient participation and cane acquired in approximately 7 min.
Given the clinically relevant information obtained fromMRI language maps in epilepsy surgical candidates, it isecessary to clearly define image-based indicators of mapuality. We therefore defined a list of quality checks (QC)n order to evaluate the reliability of the language acti-ation maps obtained for clinical applications. This novelpproach defines a systematic protocol for identifying, cor-ecting, or if necessary, rejecting activation maps based onhe following QC measures:
Patient motion—–quantitative assessment of head motionfrom translation and rotation during the full fMRI timeseries.
Functional overlay to structural underlay alignment—–quantitative assessment of proper alignment between thefunctional map and the underlying anatomical referenceframe (typically T1-weighted structural images).
Behavioral-specific primary motor activation—–confirmation of requisite primary motor cortex activationconsistent with patient compliance for the behavioraltask.
Stimulus-specific primary sensory activation—–confirmation of requisite primary sensory cortexactivation consistent with stimulus perception bythe patient (primary auditory cortex or primary visual
and temporoparietal cortices.
These measures represent quantifiable features whichhile practical in nature are often overlooked or neglected
1
ibofetrns
alracsbtratvafigms
cdwsofrgIbpda(t(sfeif(
c(cppuomfwy
M
S
Wd1na
Ogyio(p(ocsS1fI
I
AarMtilw
3eotTa7
thCa
aiasFl
876
n standard fMRI practices. In particular, the use ofehavioral-specific and sensory-specific activations areften ignored, or actively eliminated by intricately designedMRI subtraction schemes (Binder et al., 2008; Szenkovitst al., 2012; Stoppelman et al., 2013). However, we arguehat if such non-language sensory-specific activation is notobustly apparent in the fMRI activation pattern, the diag-ostic value of the language-specific activation pattern alsohould not be relied upon.
In this study, we initially perform a comparative study ofctive language fMRI (previously validated) against passiveanguage fMRI (no yet validated); we recruited a healthy,ight-handed control cohort of volunteers age-matched to
cohort of presurgical pediatric epilepsy patients. Thisontrol cohort was chosen as it can be expected to demon-trate typical language activation patterns—–characterizedy left-dominant activation of the putative language cen-ers in the left inferior frontal and left temporoparietalegions, classically referred to as Broca’s and Wernicke’sreas, respectively (Broca, 1861; Wernicke, 1874). We testhe hypothesis that our passive language fMRI task will acti-ate both language centers of the brain similarly as thective task. We then validate the passive language fMRIndings in surgical patients by comparisons to the clinicalold-standards. Finally, we illustrate the use of our recom-ended methods in a challenging cohort of very young and
edated presurgical patients (mean age 7.5 years).Pediatric temporal lobe epilepsy patients who undergo
linically-indicated Wada test and/or implantation of sub-ural electrodes with ECS mapping are a unique cohortith whom to validate language mappings done noninva-
ively by fMRI. The clinical gold-standard for lateralizationf language is the Wada test, and ECS is the gold-standardor the localization of essential language regions. As such,esults from these invasive modalities typically serve as theround-truth for the evaluation of noninvasive fMRI results.n particular, validation of fMRI language lateralization isased on comparison against Wada testing in the sameatient (Binder et al., 1996; Suarez et al., 2009), and vali-ation of fMRI language localization is based on comparisongainst ECS functional mapping performed on the patientPetrovich Brennan et al., 2007; Suarez et al., 2010). In addi-ion, the frequency modulated auditory evoked responseFMAER) recorded from scalp electrodes has been previouslyhown to identify key aspects of language-level processingrom auditory stimuli (Green et al., 1979, 1980; Stefanatost al., 1989). More recently, we have recorded FMAER fromntracranial implanted electrode grids to localize corticalunction associated with language-level auditory processingDuffy et al., 2013).
The goal of this study is to extend clinical fMRI mapping ofognitive function for epilepsy surgical applications in adultsBranco et al., 2006; Suarez et al., 2008—2010) to morehallenging populations, and to demonstrate feasibility oferforming reliable fMRI language mapping in presurgicalediatric epilepsy patients. We perform a validation studysing a cohort of 15 surgical patients who undergo both pre-perative language fMRI and invasive functional language
apping in the clinic. Finally, we present passive languageMRI results from an additional cohort of 6 patients whoere too young to receive invasive mapping (mean age: 7.5ears), two of whom were sedated during scanning.
hao2
R.O. Suarez et al.
ethods
tudy participants and consent
e recruited 25 right-handed volunteers (hand dominanceetermined by The Edinburgh handiness inventory, Oldfield,971) with no history of neurologic disorders to participate inoninvasive MR imaging and language mapping, 14 femalesnd 11 males mean age of 14.2 years (SD: 3.5 years).
The total patient cohort consisted of two populations.ne patient cohort consisted of 15 pediatric epilepsy sur-ical patients, 8 females and 7 males mean age of 14.6ears (SD: 3.0 years), all of whom underwent clinically-ndicated invasive testing to map language function as partf their clinical evaluation at Boston Children’s HospitalBCH). The second patient cohort consisted of youngeratients, 4 female and 2 males, mean age of 7.5 yearsSD: 1.8 years). All patients were diagnosed with pediatricnset epilepsy, and underwent comprehensive neuropsy-hological evaluation as part of the process for determiningurgical candidacy: Verbal Comprehension Index (AVG: 92.9,D: 25.1); Perceptual Reasoning Index (AVG: 97.8, SD:3.8). Informed signed consent and institutional approvalor medical records reviews was obtained in accordance withnstitutional Review Board (IRB) regulations.
maging protocols
ll of the participants in our study, 46 in total (25 controlsnd 21 patients) attended MRI imaging to acquire high-esolution T1-weighted anatomical images and functionalRI imaging (3-Tasla Trio scanner, Siemens, Germany); 5 of
he 15 patients also received computer tomography (CT)maging (General Electric Medical Systems, USA) in order toocalize indwelling cortical electrode strip and grid positionsith respect to the surrounding brain anatomy.
We acquired high-resolution T1-weighted images usingD magnetization prepared rapid acquisition gradientcho (MPRAGE). Typical imaging parameters consistedf: 24 cm field-of-view (FOV), 1.0 mm contiguous slicehickness, sagittal slices covering the entire head,R/TE = 1410 ms/2.27 ms, matrix 256 × 256, TI = 800 ms, and
flip angle = 9-degrees; scan duration was approximately min.
For 5 of the 15 patients, those who had intracranial elec-rodes implanted for seizure localization, CT scans of theead were acquired shortly after strip and grid implantation;T imaging was carried out using a tube voltage of 125 kV,nd voxel size of 0.5 × 0.5 × 0.625 mm3, with FOV = 500 mm.
We compared two language paradigms designed to bedministered in less than 7 min, and which use simple behav-oral tasks easy to perform by pediatric subjects—–vocalizedntonym-generation presented in visual mode and pas-ive story narrative paradigm presented in auditory mode.unctional MRI imaging was implemented in blood-oxygen-evel-dependent (BOLD) acquisitions using a 32-channel
ead coil configured to accommodate MRI-compatible visualnd auditory presentation equipment (Resonance Technol-gy, Inc., USA). Typical scanning parameters consisted of:4 cm FOV; image matrix 128 × 128, between 3.0 and 4.0 mm
epsy
pbflWppvtverpFpo
F
PQfasmcoifwdtF
Passive fMRI mapping of language function for pediatric epil
thick contiguous axial slices in order to cover the entirebrain; TR/TE = 2500 ms/31 ms; and a flip angle = 90-degrees.
Active language fMRIVisually presented, vocalized, antonym-generation task pre-sented in a rapid, event-related fMRI paradigm consisting ofpresentation of antonym cue words (e.g., UP, LEFT, OPEN,etc) presented as black text on white background for 2 seach with a pseudo-random inter-stimulus interval of 8 s onaverage. Between each cue word, a fixation point was pre-sented at center of the monitor; participants were asked toarticulate a word having the opposite meaning as each ofthe cue words presented (e.g., DOWN, RIGHT, CLOSE, etc.)without moving their head. Visual stimulus was presentedusing MRI-compatible video goggles (Resonance Technology,Inc., USA). Total scan duration: 5.5 min with a TR of 2.5 s.
Passive language fMRIAuditory presentation of a children’s story narrative readby a female, monotone voice presented in a standard fMRIblocked design interleaving 40 s intervals of passive listen-ing to the story narrative contrasted with 40 s intervals ofsilence. Participants were asked to listen to the soundsthey hear holding their head still while fixated on the fix-ation point at center of the monitor. Auditory stimulus waspresented using MRI-compatible headphones (ResonanceTechnology, Inc., USA). Total scan duration: 6.5 min with aTR of 2.5 s.
BOLD contrasts, fMRI thresholds, and overlaysStandard preprocessing of fMRI volumes was done includ-ing low band pass filtering, spatial smoothing with 8 mm
kernel, and rigid alignment of all fMRI BOLD time-seriesacquisition volumes (128 for antonym-generation task and155 for passive story narrative task). Highly activated corti-cal regions were then identified in fMRI images by statisticalFQf
Fig. 1 Motion parameters from SPM5 rigid alignments of fMRI timeach volume, top panels show x, y, z translations and bottom pandepict similar data for two representative subjects, left panel showsshows a representative scan with satisfactory amount of motion.
surgical planning 1877
arametric mapping (SPM5 Wellcome Laboratories, UK)ased on the statistical correlation between BOLD signaluctuations and the stimulus presentation paradigm used.e generated BOLD contrasts for antonym-generation com-ared to rest (silent viewing of fixation point), and forassive story narrative presentation compared to rest. Acti-ation maxima in fMRI maps were identified after setting ahreshold based on the uncorrected P statistic—–all fMRI acti-ation maps were thresholded at 0.05 or better, to avoidxcessive false negatives in family-wise-error frequencyate; however, lateralization calculations were done inde-endent of any threshold setting (Suarez et al., 2008, 2009).unctional MRI activation maps were rigidly aligned, resam-led to high-resolution T1-weighted voxel dimensions, andverlaid onto T1-weighted anatomical volume.
MRI QCs
articipant motionuantitative assessment of patient motion was performed
or each fMRI session acquired. This QC evaluates themount of translations and rotations performed duringtandard rigid alignment of BOLD time-series volumes to theean BOLD signal using rigid alignment. We define unac-
eptable amounts of x, y, z translations as those equal tor greater than twice the largest dimension of the fMRImage voxels, when occurring for more than one third of theull time series. That is, translations greater than 6.6 mmith respect to any Cartesian direction. Unacceptableegree of rotation was defined as rigid alignments greaterhan 1.0-degree with respect to any rotational axis. Seeig. 1.
unctional overlay to structural underlay alignmentuantitative assessment of the alignment between fMRI
unctional overlay and anatomical T1-weighted underlay
e-series volumes and the BOLD mean, plotted as a function ofels show pitch, roll, and yaw rotations. Left and right panels
a scan with unaccepted amount of subject motion, right panel
1878 R.O. Suarez et al.
Fig. 2 Quantitative assessment of functional overlay toanatomical underlay alignment. Top panel depicts a rep-resentative subject scan illustrating poor alignment—–noteunacceptable alignment errors visible in the sagittal and axialplanes (green arrows). Conversely, the bottom panel illustratesa representative example depicting proper alignment. (Forit
rQmlatvaaAawvma
BpAisolsvcc
Fig. 3 Illustration of quality checks to confirm expectedsensory-specific (top panels) and motor-specific (bottom panels)fMRI activation patterns from the active language task. A repre-sentative participant scan showing an unacceptable absence ofrobust, bilateral activation of primary visual cortex is shown inthe top left panel; conversely, the top right panel shows a rep-resentative scan with green arrows highlighting the expectedactivation pattern consistent with proper sensory perception ofthe visual cue words. The bottom left panel shows a scan withunacceptable absence of robust, bilateral activation of primarymotor cortex; conversely, the bottom right panel shows an acti-vation pattern consistent with articulation of antonym responsewords and thus compliance for the active task. (For interpreta-ti
(pttavatw
SPo
nterpretation of the references to color in this figure legend,he reader is referred to the web version of this article.)
eference frames was performed on all acquisitions. ThisC required confirmation of proper alignment between theean BOLD volume and the anatomical T1-weighterd under-
ay volume, see Fig. 2. We used a standard rigid alignmentlgorithm based on mutual information in order to alignhe BOLD time series mean to the T1-weighted referenceolume. We required accurate fMRI to anatomical volumelignment precision to within the order of T1-weightednatomical voxel size in each plane (0.5 × 0.5 × 0.625 mm3).ny misalignment errors were by this means identified,nd if necessary corrected. Often, alignment improvementsere achieved by aligning the fMRI mean to skull-strippedersions of the T1-weighted anatomical reference usingethods previously described (Grau et al., 2004; Weisenfeld
nd Warfield, 2009).
ehavioral-specific primary motor and sensory-specificrimary visual activationntonym-generation fMRI activation maps were inspected
n order to confirm requisite sensory-specific and motor-pecific activation. This QC confirms robust, focal activationf the primary visual centers bilaterally in the occipitalobes consistent with sensory processing of visually pre-
ented antonym cue words. Additionally, as the task requiredocalized responses, we inspected all activation maps toonfirm robust, bilateral, focal activation of primary motorortex consistent with articulation of the response wordssrec
ion of the references to color in this figure legend, the readers referred to the web version of this article.)
often this activation was accompanied by activation of sup-lementary motor cortex in the midline). The presence ofhis activation pattern confirms the participant perceivedhe visual stimulus presented and consistently vocalized thentonym response words; in this way confirming successfulisual stimulus delivery and participant compliance for thective task. See Fig. 3. Absence of these activation pat-erns classified the fMRI map as unreliable and thereforeas excluded from further analysis.
ensory-specific primary auditory activationassive language fMRI activation maps were inspected inrder to confirm sensory-specific activation from aurally pre-
ented story narrative. This QC required confirmation ofobust, focal activation of the primary auditory cortex bilat-rally in the middle portion of the superior temporal gyrusonsistent with proper auditory perception of the auditory
Passive fMRI mapping of language function for pediatric epilepsy surgical planning 1879
Fig. 4 Illustration of quality check to confirm the expectedsensory-specific fMRI activation pattern from passive listeningto story narrative language task. A representative participantscan showing unacceptable absence of robust, bilateral acti-vation of primary auditory cortex is shown in the left panel;conversely, the right panel shows a representative scan con-firming robust, bilateral activation of primary auditory cortex(green arrows) consistent with proper sensory perception of theauditory stimulus. (For interpretation of the references to color Fig. 5 A representative fMRI language activation map
(p < 0.001, uncorrected) from the visually presented antonym-generation task, note the expected language-specific activationpeaks in the putative language centers of the inferior frontalgyrus and temporoparietal regions (white circles). Note theclear activation asymmetry across the cerebral hemispheres,in this case denoting typically left-lateralized language for thisp
C
WgpsTaEtpipc
Vp
Tplopwas well as clinically-indicated lateralization of language
in this figure legend, the reader is referred to the web versionof this article.)
stimulus. Absence of this sensory-specific activation pat-tern rendered the fMRI map unreliable and therefore wasexcluded from further analysis. See Fig. 4.
Language-specific activationAll fMRI activation maps were then inspected in order to con-firm robust, local activation maxima in regions of the brainconsistent with the putative language centers of the: (1)inferior frontal gyrus (IFG), and (2) temporoparietal regions(TPG). This activation is typically lateralized favoring onecerebral hemisphere over the other, often the left (Wada andRasmussen, 1960; Binder et al., 1996; Knecht et al., 2000),see Fig. 5. The x, y, z coordinates of IFG and TPG local acti-vation peaks were identified and recorded for localizationvalidation.
Lateralization of language fMRI
We quantified language activation asymmetry using analgorithm independent of fMRI threshold. We previouslydescribed this approach in detail (Branco et al., 2006;Suarez et al., 2008, 2009). Briefly, a unique laterality index(LI) was calculated based on the T-squared weighted sumsof the un-thresholded positive fMRI activation correlationvalues recorded within the two language-specific regions-of-interest (ROIs): (1) the IFG, and (2) the TPG regions. SeeFig. 5. This calculation yielded a unique LI value for eachROI in the range between −1.0, 0.0, and +1.0, where nega-tive LI values indicate right-lateralized activation, positive
LI values indicate left-lateralized activation, and absoluteLI values less than or equal to 0.1 indicate no activationasymmetry (Suarez et al., 2008, 2009).fgu
articipant.
omparison of fMRI tasks in healthy controls
e compared the previously validated vocalized antonym-eneration language task (Suarez et al., 2008, 2009) toassive language fMRI in the control cohort of 25 individuals,ee Table 1. Subject and group region-specific LIs for IFG andPG were calculated in order to evaluate language-specificctivation asymmetry trends for the two task conditions.ach subject’s LI were plotted and the mean LI for the groupabulated and compared across task conditions. Active andassive language fMRI paradigms were judged based on abil-ty to activate the known language centers and demonstratereponderance for left-dominant language processing in theontrol cohort.
alidation of fMRI in temporal lobe epilepsyatients
he validation of passive language fMRI was carried out onatient cohort 1, made up of 15 pediatric-onset temporalobe epilepsy surgical patients all of whom had some formf invasive mapping for comparisons against fMRI activationatterns. See Table 2. As part of their standard surgicalorkup, these patients underwent fMRI language mapping
unction by Wada testing and/or localization of eloquent lan-uage cortex by ECS. Five of these 15 patients additionallynderwent recording of subdural FMAER evoked responses,
1880
Table 1 Healthy control cohort consisting of 25 volunteersage-matched and sex-matched to pediatric epilepsy surgicalpatients, all strong right-handers by the Edinburgh handed-ness inventory (Oldfield, 1971).
Subject Sex Age Handedness
c01 M 11 +100c02 M 8 +90c03 F 16 +100c04 F 10 +70c05 M 11 +85c06 M 10 +100c07 M 17 +100c08 F 8 +100c09 F 18 +85c10 M 16 +60c11 F 17 +100c12 F 16 +85c13 M 17 +75c14 Ll 17 +90c15 F 17 +100c16 M 16 +70c17 11 +90c18 F 12 +100c19 F 9 +85c20 F 16 +90c21 M 12 +100c22 M 16 +85c23 F 15 +100c24 F 19 +90c25 M 19 +100
Mean age (years) 14.2
wpif1fats
T
Ilbf5npfpt
N
ScwcrIt
W
CdtBolaamalpaile
D
Ceplste8iRsa
pfatodia61ewaered positive for essential language function if disruption ofthe behavioral task was consistently observed upon two orthree separate stimulations; the unique number indentify-
hich were used to validate auditory evoked activation byassive fMRI. Thirteen of these 15 patients had clinically-ndicated language dominance determined by Wada testing,or comparison against fMRI LIs for IFG and TPG. Five of these5 patients had ECS sites confirmed for essential languageunction and were used to compare against fMRI languagectivation. Language localizations were assessed based onhe Cartesian distance mismatch between positive electrodeites by found by ECS and fMRI activation peaks.
esting of passive fMRI in cohort 2
n a second patient cohort (cohort 2), we tested passiveanguage fMRI in a challenging population of patients whoecause of their young age were not candidates for invasiveunctional mapping. Cohort 2 consisted of 6 patients (age, 8, or 9 years), two of who generally undergo MRI scan-ing under clinically-indicated sedation: patients p20 and21; age 5 and 8 years. Please see Table 2. Passive languageMRI activation patterns in cohort 2 were used as compellingroof-of-concept in the most challenging pediatric popula-
ion. iR.O. Suarez et al.
europsychological measures
ixteen of twenty-one patients underwent recent neuropsy-hological evaluation as part of their standard presurgicalorkup. All measures were administered by a licensedlinical neuropsychologist or by a postdoctoral fellow in neu-opsychology under the supervision of a licensed clinician.ntellectual functioning (VCI and PRI) were assessed usinghe Wechsler Intelligence Scale for Children—–fourth edition.
ada testing procedures
linically-indicated intracarotid sodium amobarbital proce-ure (Wada) was conducted in 13 of 21 patients as part ofhe standard epilepsy surgery candidacy determination atCH. The procedure involved administration of 100—150 mgf amobarbital to the presumed lesional hemisphere, fol-owed by administration to the nonlesional hemispherepproximately 30 min later; neurobehavioral evaluation wasdministered immediately after each injection. Effects ofedication were confirmed by contralateral hemiparesis
nd EEG slowing at the target hemisphere. A battery ofanguage tests was completed including single-word com-rehension, confrontation naming, repetition of phrases,nd recitation of complex verbal sequences such as recit-ng months of the year in reverse order. Region-specificateralization outcome was determined for receptive andxpressive language functions separately for each patient.
irect electrocortical stimulation mapping (ECS)
linically-indicated functional language mapping by directlectrocortical stimulation was performed in 5 of the 21resurgical patients; ECS mapping was carried out uti-izing electrode strips and grids previously implanted fortandard phase II monitoring at BCH. The implanted elec-rodes consisted of stainless-steel or platinum—iridium alloylectrode discs 4 mm in diameter arranged in grids of: 8 × 8,
× 4, or 8 × 2, or strips of: 8 × 1, 6 × 1, or 4 × 1, with annter-electrode spacing of 10 mm center-to-center (AdTech,acine, USA). After implantation, strips and grids wereutured to the dural membrane. Patients p01, p02, p08, p11,nd p12 received ECS functional mapping of language.
To map the regions essential to language by ECS,atients were awaken after craniotomy and asked to per-orm behavioral language tasks designed to test receptivend expressive language functions, including overt repeti-ion of words, simple phrases, or sentences, and recitationf sequences such as reciting months of the year, orays of the week. While patients performed these behav-oral tasks, brief electrical currents were systematicallypplied to the strip and grid electrodes using an IRES00 CH electrical stimulator (Micromed, Italy; 50 Hz, 4e7 s,.5e10 mA). Intensity of the stimulations was adjusted forach patient in order to maximize the stimulation effecthile at the same time minimizing the occurrence offter-discharges. Electrode stimulation sites were consid-
ng that electrode site was recorded. Stimulations that were
Passive fM
RI m
apping of
language function
for pediatric
epilepsy surgical
planning
1881
Table 2 Demographics and description of pediatric epilepsy patient cohorts; verbal comprehension index (VCI) and perceptual reasoning index (PRI) are shown. Data notshown were not available in those patients. Shown are the two patient cohorts used in the study, including their etiology and pathology data if available.
Patient Sex Handedness Age Onset age VCI IQ PRI IQ Epileptic localization Etiology Pathology
p01 M R 16 12 53 70 Left temporal Neurofibromatosis type 1 Hippocampal sclerosisp02 M R 17 7 — — Left tempo-
ral/frontal/parietalFocal cortical dysplasia Cortical dysplasia
p03 F R 17 13 149 115 Left temporal Temporal lope epilepsy No biopsyp04 M R 16 9 116 119 Right posterior
quadrantUnknown No biopsy
p05 F L 13 6 93 106 Left temporal Mesial temporal lobesclerosis
Mild mesial temporalsclerosis
p06 M L 10 1 71 88 Left hemisphere Vascular insult Meningoencephalitisp07 F R 19 6 72 86 Right mesial temporal Focal cortical dysplasia Mild malformationp08 F R 16 13 108 98 Left temporal Subcortical heterotopia Palmini grade Iap09 M R 17 15 100 88 Left temporal Focal cortical dysplasia Sclerosisp10 F L 12 0 75 102 Left frontotemporal Vascular insult Cortical dysplasiap11 F R 13 3 — — Left temporal Unknown Cortical dysplasia
type IIIdp12 M L 12 5 — — Left mesial temporal Unknown No biopsyp13 M R 15 4 102 110 Left temporal/frontal Vascular insult Mild gliosisp14 F R 17 8 87 92 Left temporal Temporal lope epilepsy cortical dysplasiap15 F R 9 9 89 100 Left temporal Temporal lope epilepsy Cortical dysplasia
type IIIa
Mean age cohort 1 (years) 14.6 7.4p16 F R 9 8 87 92 Left temporal Temporal lope epilepsy No biopsyp17 F R 9 2 — 46 Left anterior
subcorticalLeft periventricular cyst No biopsy
p18 M R 9 8 — 61 Left frontal Rasmussen’s encephalitis Moderate Tlymphocyte
p19 F L 5 3 100 108 Left temporal Focal cortical dysplasia No biopsyp20 M R & L 8 1 — — Temporal/bilateral
occipitalMesial temporal lobesclerosis
No biopsy
p21 F L 5 2 — — Left hemisphere Rasmussen’s encephalitis No biopsy
Mean age cohort 2 (years) 7.5 4
1 R.O. Suarez et al.
aootttppcpc
F(
Spd2cuotwuwlaTewesna(mwsaf
Pf
IEsr2msciWstAteL
Table 3 Control cohort language lateralization by fMRIusing activate and passive fMRI tasks. Listed are group meanlateralization indices (LI) in the inferior frontal (IFG) andtemporoparietal (TPG) language regions. Positive LI indicateleft (L) lateralization; (−) denotes omitted scans.
Subject Active task Passive task
IFG TPG IFG TPG
c01 L L L Lc02 L L L Lc03 L L L Lc04 — — R Lc05 L L L Lc06 L L L Lc07 — — R Rc08 — — R Lc09 L R R Rc10 R R — —c11 L Bilateral L Lc12 L R R Lc13 L L L Lc14 L R L Lc15 L L L Lc16 L L L Lc17 — — R Lc18 L L R Lc19 L L L Lc20 — — — —c21 L Bilateral L Lc24 L L L Bilateralc25 — — L L
Mean LI value +0.38 +0.15 +0.24 +0.45
mpS
R
C
BcerQl
iaL
882
ccompanied by after-discharges were screened for theccurrence of spread to other electrodes and when thisccurred the respective stimulation was excluded from fur-her analysis. We measured the Cartesian distance betweenhe ECS electrodes confirmed for language function andhe localization of peak fMRI TPG activation for languagerocessing. Note that cortical stimulation sites in eachatient were by necessity limited to the specific electrodeoverage clinically-indicated in each patient for the pur-ose of epilogenic foci localization and not necessarily foromprehensive functional mapping.
requency modulated auditory evoked responsesFMAER)
ubdural FMAER responses were recorded in 5 of the 15atients in cohort 1. The FMAER procedures used wereescribed in detail in a previous publication (Duffy et al.,013). Briefly, we administered aurally presented stimulusonsisting of a carrier sine wave at 1000 Hz frequency mod-lated by a slower 10 Hz sine wave, causing the frequencyf the carrier wave to deviate between 960 and 1060 Hz athe 10 Hz rate and thus producing a frequency modulationarbling tone. Next, a 10 Hz sine wave was amplitude mod-lated by a slower 4 Hz sine wave such that the warblingas sinusoidally turned on and off at 4 Hz rate. Each stimu-
us onset was used to trigger separate epochs used for signalveraging and generation of the averaged evoked response.ypically 300—1000 ms trigger pulses were averaged over anpoch duration of 1000 ms (Green et al., 1979, 1980). Dataere recorded directly from the cortex utilizing indwellinglectrodes. Recordings that contained continual or frequenteizure discharges or electrodes that had not been placedear or over the temporal regions were not included in thenalysis. Signal averaging was performed using BESA GmbHv6.0, Germany). Only the electrode(s) demonstrating theaximal evoked responses following 4 Hz FMAER stimulusere recorded for our analysis; we measured the Carte-
ian distance between the uniquely numbered electrodesnd the localization of peak fMRI primary auditory activationor language stimuli.
reparation of 3D surfaces for electrode grid andMRI map co-registrations
n order to produce co-registrations and 3D models ofCS, FMAER, and fMRI mapping results for comparativetudy in the same patient, we utilized an alignment algo-ithm we previously described in detail (Taimouri et al.,013). Briefly, we created a patient-specific 3D geometricodel of the cortical brain surface by means of automatic
egmentation of the T1-weighted anatomical images. Theo-registration of T1-weighed anatomical volumes and CTmages was achieved by mutual information rigid alignment.e then projected the grid electrodes onto the cortical
urface in the perpendicular direction, that is, normal tohe plastic sheet on which the electrodes were mounted.
unique normal vector was determined for the strip elec-rodes by minimizing the sum squared distances betweenlectrodes and the plane crossing all of the strip electrodes.astly, we colored the 3D brain surface by fMRI activation
6++F
aps in order to measure the Cartesian distance betweenositive electrode localization and local fMRI maxima.ee Fig. 7.
esults
omparison of fMRI tasks
oth of the fMRI language tasks we tested in the controlohort demonstrated robust activation consistent with thexpressive language region (IFG) and the receptive languageegion (TPG) in all of the subject scans with satisfactoryCs (38 out of 50): 17 active language fMRI and 21 passive
anguage fMRI.In the control cohort, the active task demonstrated 94%
ncidence of left-lateralized IFG activation (mean LI: +0.38),nd 74% incidence of left-lateralized TPG activation (meanI: +0.15), compared to the passive task which demonstrated7% incidence of left-lateralized IFG activation (mean LI:0.24), and 86% incidence of left-lateralized TPG (mean LI:
0.45). These comparisons were summarized in Table 3 andig. 6.
Passive fMRI mapping of language function for pediatric epilepsy surgical planning 1883
Fig. 6 Control cohort lateralization indices (LI) for the inferior frontal gyrus (IFG) and temporoparietal (TPG) language regions,using activate language fMRI (top panel) and passive language fMRI (bottom panel). Positive LI values indicate left-lateralization,
itude
V
Li(esaT
tiTg
negative LI indicates right-lateralization, and LI values of magn
Validation of passive language fMRI
Patient cohort 1 received passive language fMRI and inva-sive functional mapping by Wada testing and/or ECS. Twoof the fMRI scans were re-acquired due to lack primaryauditory activation in the first session (we suspected pooror no audio stimulus being delivered to the patient asa result of user error in the first attempt—–the patientfailed to report the problem). Patient p04 did not demon-strate the expected bilateral primary auditory activation(only observed in the left side), however, the lack of rightactivation was attributed to the presence of large lesionencompassing the primary auditory cortex of that side ofthe brain and therefore that fMRI scan was not repeated
in that case. All of the fMRI scans acquired in the patientcohort with satisfactory QCs (15 of 15, after 2 scans wereredone) demonstrated local fMRI activation peaks in IFG andTPG consistent with language-specific activation.tmsp
0.1 or less indicate bilateral activation.
alidation of passive language fMRI lateralization
anguage lateralization by fMRI in the TPG ROI agreed withnvasive clinical testing for receptive language dominanceWada testing and ECS mapping) in 11 of 15 patients. Gen-rally when fMRI was compared against these referencetandards, we observed 80% congruency in TPG languagectivation, and 73% congruency in IFG activation. Seeable 4.
The fMRI language lateralization of IFG agreed with Wadaest language dominance for expressive language processingn 9 of 13 patients (70% fMRI to Wada test congruency); fMRIPG lateralization agreed with Wada test for receptive lan-uage dominance in 10 of 13 patients (77% fMRI to Wada
est congruency). Of 5 patients who received ECS languageapping, 4 showed dominant TPG activation by fMRI in theame cerebral hemisphere as essential receptive languagerocessing localized by ECS (80% fMRI to ECS congruency).
1884 R.O. Suarez et al.
Table 4 Congruency between region-specific language lateralization by invasive clinical standards compared to fMRI. Shownare left, right (L, R), or bilateral expressive and receptive language dominance as determined by Wada testing; hemisphericallocalization of receptive language function as determined by direct eletrocortical stimulation mapping (ECS); and languagelateralization of the inferior frontal (IFG) and temporoparietal (TPG) language regions as determined by passive language fMRI.Bold lettering denotes instances of fMRI incongruity with the clinical reference standards.
Patient Wada lateralization ECS fMRI lateralization
Expressive Receptive IFG TPG
p01 L L Left TPG L Lp02 L Bilateral Left TPG R Rp03 L L — L Rp04 L L — R Lp05 Bilateral Bilateral — L Rp06 R R — R Rp07 L L — R Lp08 L L Left TPG L Lp09 L L — L Lp10 R R — R Rp11 — — Left TPG R Lp12 — — Left TPG L Lp13 Bilateral L — Bilateral Lp14 L L — L Lp15 L L — L L
11/15 (73%) 12/15 (80%)
V
Ilpamcl5omptictafi
T
Ffnsaotvt
Table 5 The cortical Cartesian distance mismatch betweenFMAER and primary auditory activation from passive lan-guage fMRI (FMAER—–auditory fMRI), and between ECSlanguage sites and TPG activation by passive language fMRI(ECS—–language fMRI). Cases which did not have sufficientelectrode coverage in regions identified as dominant forreceptive language by fMRI are denoted with (*) and wereexcluded from the calculation of mean value.
Patient Mismatch distance (mm)
FMAER—–auditory fMRI ECS—–language fMRI
p01 13.3 >50.0*
p02 22.3 >50.0*
p08 10.6 12.1p11 8.1 7.3p12 17.6 8.1
wF
D
Igp
Cases congruent with clinical standards
alidation of fMRI localization
n patient cohort 1, 5 of 15 patients received both ECSanguage mapping and FMAER localization. In the com-arison between FMAER localizations and primary auditoryctivation from passive language fMRI, the mean distanceismatch observed was 14.4 mm (SD: 5.7 mm). In the
omparison between ECS language localizations and TPGanguage activation from passive language fMRI only 2 of the
patients (p01 and p02) had sufficient electrode coveragever the TPG activation region and as a result the distanceismatch could not be confirmed in those cases. In patient01, TPG activation by fMRI localized over 50 mm away fromhe nearest electrode. In patient p02, TPG activation local-zed to the right hemisphere, where there was no electrodeoverage. With p01 and p02 results excluded, the mean dis-ance mismatch between receptive language ECS electrodesnd fMRI TPG activation was 9.2 mm (SD: 2.6 mm). Thesendings are summarized in Table 5 and Fig. 7.
esting of passive fMRI in cohort 2
or 6 of 6 patients in cohort 2, we recorded robust bilateralMRI activation of primary auditory cortex consistent withormal perception of the auditory narrative stimulus pre-ented. In 6 of 6 patients in cohort 2 we detected lateralizedctivation consistent with typical left-lateralized processing
f the putative temporoparietal language center; two ofhese patients (p17 and p20) demonstrated lateralized acti-ation consistent with typical left-lateralized processing ofhe putative frontal language center. Patients p20 and p21a
pe
Mean (mm) 14.4 9.2
ere sedated during passive language fMRI recording. Seeig. 8.
iscussion
n this study we defined, validated, and tested passive lan-uage fMRI imaging protocols which do not require anyarticipation from the patient and are designed for reliable
pplication in pediatric epilepsy applications.We initially compared two fMRI language mappingaradigms: a previously validated protocol for adults (Suarezt al., 2008, 2009) which is active in nature (vocalized
Passive fMRI mapping of language function for pediatric epilepsy
Fig. 7 Three-dimensional renderings of cortical brain sur-faces from the 5 participants in our patient cohort whounderwent implantation of intracranial electrode strips andgrids. Depicted on the cortical surfaces are the intracranialelectrodes (black dots), and the cortical activation maps frompassive language fMRI (p < 0.001 uncorrected). Note fMRI activa-tion patterns consistent with primary auditory activation. NotefMRI activation patterns consistent with temporoparietal recep-tive language processing. The electrodes of maximal frequencymodulated auditory evoked responses (FMAER) are denoted byblue circles, electrode locations for essential language functionby direct eletrocortical stimulation (ECS) are denoted by greendots. (For interpretation of the references to color in this figure
s8ciw1iw9
fyiboaplap
ctmfd2fon2iec2apqbitt
ms1iouft1asccl(
legend, the reader is referred to the web version of this article.)
antonym-generation); and passive listening to a story nar-rative. Group-level analysis demonstrated that both tasksproduced consistent activation in the two main languagecenters. We found that the active task more robustly acti-vated the expressive language region (IFG) compared tothe receptive language region (TPG), while the passive taskmore robustly activated TPG compared to IFG. Addition-ally, more left-lateralized activation was observed in IFG(LI = +0.38) compared to TPG (LI = +0.15) using the activetask, whereas more left-lateralized activation was observedin TPG (LI = +0.45) compared to IFG (LI = +0.24) using thepassive task.
We then validated the passive language fMRI in cohort1, made up of 15 presurgical patients with invasive func-tional mapping results. The validation approach comparedlanguage dominance by Wada testing and ECS, against
functional lateralization by fMRI; FMAER peak responselocalizations were compared against auditory-specific acti-vation by fMRI; and ECS language localizations againstlanguage-specific TPG activation by fMRI. We confirmedsaoc
surgical planning 1885
trong fMRI congruency to the invasive clinical procedures:0% congruency with respect to TPG activation, and 73%ongruency with respect to IFG activation. In the compar-son between fMRI and FMAER for the 5 patients tested,e found a mean FMAER to fMRI mismatch distance of only4.4 mm. In the comparison between ECS and TPG activationn the 3 patients tested with sufficient electrode coverage,e observed a mean ECS to fMRI mismatch distance of only.2 mm.
Lastly, we tested the recommended passive languageMRI and QC protocols in a challenging cohort of veryoung patients, cohort 2, two of who were sedated dur-ng scanning. Reliability evaluation of these fMRI maps wasased on our recommended quality checks (QC)—–primarilyn detection of expected sensory-specific primary auditoryctivation patterns, and detection of lateralized activationatterns in the putative language regions consistent withanguage-specific cortical processing. We confirmed reliablectivation patterns from passive language fMRI in 6 of 6atients in this challenging cohort.
There is a recent trend which advocates increased indi-ation for surgical intervention in younger epilepsy patientshan was true in the past. It is now understood that inost cases, early and minimally invasive surgery is essential
or successful management of epilepsy progression in chil-ren (Holthausen et al., 2013; Glauser and Loddenkemper,013). Accordingly, children who in the past were excludedrom epilepsy surgery, for example due to early onset ager presentation of multiple lesions, are now in greaterumbers becoming surgical candidates (Beier and Rutka,013; Holthausen et al., 2013). While the gradually widen-ng spectrum of indications for surgery in children is shownffective for preventing later-stage development of moreatastrophic disease manifestations (Holthausen et al.,013), it places greater than before emphasis on reli-ble presurgical language mapping in increasingly youngeratients. Because passive language fMRI can be administereduickly and simply (7 min scan which does not require anyehavioral participation from the patient) and is thereforedeal for use with younger patients who may not be expectedo comply with the more complicated behaviorally activeasks.
Localization of neurophysiological activity in the brain isost reliably mapped by direct mapping techniques applied
ubdurally to the cortical surface, such as ECS (Ojemann,993). Intracranial electrode strips and grids implanted fornvasive monitoring in presurgical epilepsy patients henceffers a unique opportunity to perform activation studiessing the same averaged evoked potential paradigms usedor decades with scalp electroencephalography and magne-oencephalography (Green et al., 1979, 1980; Reite et al.,978; Stefanatos et al., 1989). The frequency modulateduditory evoked response (FMAER) techniques used in thistudy localize sensory-specific activation of the auditoryortex which in addition to processing simple sensory per-eption of the stimulus also holds promise as a proxy foranguage-level processing from the FM stimulus presentedDuffy et al., 2013). In this study, we found that sensory-
pecific fMRI activation produced by auditory presentation ofstory narrative agreed to within 15 mm with the electrodesf peak FMAER responses. This close proximity providesompelling validation for fMRI localization of auditory-level
1886 R.O. Suarez et al.
Fig. 8 Passive language fMRI activation patterns recorded in patient cohort 2 (p < 0.001, uncorrected), made up of young patients(mean age of 7.5 years). Six of six fMRI activation patterns demonstrated bilateral primary auditory activation in the middle portionof the superior temporal gyrus consistent with typical perception of auditory narrative stimulus. Six of six fMRI maps demonstratedlateralized activation of the temporoparietal language region consistent with typical left-dominant language-specific activation;p d acw
pl
Etc(iwtwgrbstpFeplEw
lmwt
cgawaaenpvssmbvslc2p2twn
atients p17 and p20 additionally demonstrated left-lateralizeere sedated during fMRI scanning.
rocessing of the language-specific stimulus in the passiveanguage paradigm.
Invasive functional mapping of language processing byCS, in contrast to the evoked FMAER responses, is a deac-ivation technique which aims to identify and localize theortical sites which are essential for language functionOjemann, 1993). Application of electric currents to themplanted electrodes in effect produces a temporary lesionhich momentarily ceases cortical function at the applica-
ion volume. By electrically deactivating the cortical siteshich if damaged during resection cause similar postsur-ical language deficits, the ECS technique is currently theeference standard for localizing essential language in therain. However, ECS functional language mapping is an inva-ive procedure requiring adequate electrode coverage overhe essential language regions, awake surgery, and strictatient task compliance intra-operatively after craniotomy.or these reasons, functional language mapping by ECS isxtremely limited or not feasible in very young or incom-liant patients. We found that in most cases, receptiveanguage activation by fMRI acquired passively agreed withCS localization of essential receptive language function toithin 10 mm.
The novel set of QCs we recommend for clinical fMRI
anguage mappings were designed to rule out many com-on errors which if not identified (and if possible corrected)ould render the fMRI activation maps unreliable, but whicho our knowledge have never been used to systematically
ifap
tivation of the frontal language region. Patients p21 and p20
lassify fMRI maps as unreliable. When assessing fMRI lan-uage maps, we recommend close attention to non-languagectivation patterns, such as sensory or motor activationhich while not directly associated with language processingre nevertheless required features consistent successfulpplication of a particular fMRI language protocol. Forxample, visual words or auditory presentation of a storyarrative by necessity require initially the sensory-specificerception of the stimulus by the patient (e.g., primaryisual cortex, or primary auditory cortex, depending on thetimulus mode used). Therefore, the absence of this sensory-pecific activation indicates that the resulting activationap is inconsistent with the fMRI paradigm and should note relied upon to correctly delineate language-specific acti-ation in the clinic. In prior studies by others, subtractionchemes have been proposed in order to eliminate non-anguage activation, typically by carefully defined baselineonditions and subtraction methodologies (Binder et al.,008; Szenkovits et al., 2012; Stoppelman et al., 2013), orrecisely timed MRI scanner sampling schemes (Birn et al.,004; Preibisch et al., 2003). However, it is not clear howo define the most appropriate choice of baseline condition,hich eliminates all of the non-language activation whileot at all impacting language-specific activation patterns. It
s also not clear that non-language activation is irrelevantor the analysis of language-specific activation patterns. Wedditionally point out that while a myriad of factors couldotentially lead to alignment errors when creating fMRI map
epsy
B
B
B
D
D
G
G
G
G
G
H
K
L
M
Passive fMRI mapping of language function for pediatric epil
overlies, often these alignment errors are not detected orare ignored; thereby introducing localization errors in acti-vation maps.
By applying the recommend QCs systematically to thefMRI activation maps in our control population, we wereable to determine that while overt antonym-generation taskcan more robustly activate frontal language processing inIFG, it also has a higher likelihood of poor task compli-ance or becoming contaminated by excessive patient motioncompared to the passive narrative task. This indicates that inmost pediatric temporal lobe epilepsy surgical candidates,the fMRI mapping of essential language can proceed withpassive language fMRI. This protocol similarly holds strongpromise for routine use with sleeping, sedated, or uncon-scious patients in whom language mapping would otherwisenot be possible. We presented compelling evidence that ourrecommended fMRI language mapping protocols will nonin-vasively produce language maps which are equivalent withthe invasive clinical procedures.
Conclusions
We have shown that our recommended passive languagefMRI paradigm can be applied for reliable determination oflanguage dominance and language functional localization inchildren as young as 5—9 years of age, including patients whoare sedated. As such, the fMRI protocols we present effec-tively meet an increasing demand for reliable noninvasivefunctional mapping of language in very young presurgicalepilepsy patients.
Acknowledgements
This work was supported in part by National Institutesof Health R01 NS079788, R42 MH086984, National Multi-ple Sclerosis Society Pilot Grant PP1625, and United StatesDepartment of Defense award MS120015.
References
Abou-Khalil, B., 2007. An update on determination of language dom-inance in screening for epilepsy surgery: the Wada test andnewer noninvasive alternatives. Epilepsia 48 (Mar (3)), 442—455,Epub 2007 Feb 21. Review. PubMed PMID: 17319925.
Beers, C.A., Federico, P., 2012. Functional MRI applications inepilepsy surgery. Can. J. Neurol. Sci. 39 (May (3)), 271—285,Review. PubMed PMID: 22547506.
Beier, A.D., Rutka, J.T., 2013. Hemispherectomy: historicalreview and recent technical advances. Neurosurg. Focus 34(Jun (6)), E11, http://dx.doi.org/10.3171/2013.3.FOCUS1341,PubMed PMID: 23724835.
Binder, J.R., Swanson, S.J., Hammeke, T.A., Morris, G.L., Mueller,W.M., Fischer, M., Benbadis, S., Frost, J.A., Rao, S.M., Haughton,V.M., 1996. Determination of language dominance using func-tional MRI: a comparison with the Wada test. Neurology 46 (Apr(4)), 978—984.
Binder, J.R., Swanson, S.J., Hammeke, T.A., Sabsevitz, D.S.,2008. A comparison of five fMRI protocols for mapping speech
comprehension systems. Epilepsia 49 (Dec (12)), 1980—1997,http://dx.doi.org/10.111/j.1528-1167.2008.01683.x, Epub2008 May 29. PubMed PMID: 18513352; PubMed Central PMCID:PMC2645716.O
surgical planning 1887
irn, R.M., Cox, R.W., Bandettini, P.A., 2004. Experimentaldesignsand processing strategies for fMRI studies involving overt verbalresponses. NeuroImage 23 (Nov (3)), 1046—1058.
ranco, D.M., Suarez, R.O., Whalen, S., O‘Shea, J.P., Nelson, A.P.,da Costa, J.C., Golby, A.J., 2006. Functional MRI of memory inthe hippocampus: laterality indices may be more meaningful ifcalculated from whole voxel distributions. NeuroImage 32 (Aug(2)), 592—602, Epub 206 Jun 13. PubMed PMID: 16777435.
roca, P.P., 1861. Perte de la parole ramolissement chronique etdestruction partielle du lobe antérieur gauche de cerveau. Bul-letins de la Société d’anthropologie de Paris 2, 235—238.
uffau, H., Moritz-Gasser, S., Gatignol, P., 2009. Functionaloutcome after language mapping for insular World HealthOrganization Grade II gliomas in the dominant hemisphere: expe-rience with 24 patients. Neurosurg. Focus 27 (Aug (2)), E7,http://dx.doi.org/10.3171/2009.5.FOCUS0938, PubMed PMID:19645563.
uffy, F.H., Eksioglu, Y.Z., Rotenberg, A., Madsen, J.R., Shankar-dass, A., Als, H., 2013. The frequency modulated auditoryevoked response (FMAER), a technical advance for study ofchildhood language disorders: cortical source localization andselected case studies. BMC Neurol. 13, 12, http://dx.doi.org/10.1186/1471-2377-13-12, PubMed PMID: 23351174; PubMedCentral PMCID: PMC3582442.
arrett, M.C., Pouratian, N., Liau, L.M., 2012. Use of lan-guage mapping to aid in resection of gliomas in eloquentbrain regions. Neurosurg. Clin. N. Am. 23 (Jul (3)), 497—506,http://dx.doi.org/10.1016/j.nec.2012.05.003, Review. PubMedPMID: 22748661; PubMed Central PMCID: PMC3618998.
lauser, T.A., Loddenkemper, T., 2013. Management of childhoodepilepsy. Continuum. (Minneap. Minn.) 19 (Jun (3 Epilepsy)),656—681, http://dx.doi.org/10.1212/01.CON.0000431381.29308.85, PubMed PMID: 23739103.
rau, V., Mewes, A.U., Alcaniz, M., Kikinis, R., Warfield, S.K., 2004.Improved watershed transform for medical image segmentationusing prior information. IEEE Trans. Med. Imaging 23, 447—458.
reen, G.G.R., Kay, R.H., Rees, A., 1979. Responses evoked byfrequency—modulated sounds recorded from the human scalp.J. Physiol. 296, 21P—22P [PubMed].
reen, G.G.R., Rees, A., Stefanatos, G.A., 1980. A method forrecording evoked responses to frequency modulated sounds inman. J. Physiol. 307, 10p.
olthausen, H., Pieper, T., Kudernatsch, M., 2013. Towardsearly diagnosis and treatment to save children from cata-strophic epilepsy—–focus on epilepsy surgery. Brain Dev. (Jun),http://dx.doi.org/10.1016/j.braindev.2013.05.003, pii: S0387-7604(13)00166-6. [Epub ahead of print] PubMed PMID: 23791480.
necht, S., Dräger, B., Deppe, M., Bobe, L., Lohmann, H., Flöel, A.,Ringelstein, E.B., Henningsen, H., 2000. Handedness and hemi-spheric language dominance in healthy humans. Brain 123 (Dec(Pt 12)), 2512—2518, PubMed PMID: 11099452.
oiselle, M., Rouleau, I., Nguyen, D.K., Dubeau, F., Macoir, J.,Whatmough, C., Lepore, F., Joubert, S., 2012. Comprehension ofconcrete and abstract words in patients with selective anteriortemporal lobe resection and in patients with selective amygdalo-hippocampectomy. Neuropsychologia 50 (Apr (5)), 630—639,http://dx.doi.org/10.1016/j.neuropsychologia.2011.12.023,Epub 2012 Jan 8. PubMed PMID: 22245005.
atsuda, R., Coello, A.F., De Benedictis, A., Martinoni,M., Duffau, H., 2012. Awake mapping for resection ofcavernous angioma and surrounding gliosis in the left dominanthemisphere: surgical technique and functional results: clinicalarticle. J. Neurosurg. 117 (Dec (6)), 1076—1081, http://dx.doi.org/10.3171/2012.9.JNS12662, Epub 2012 Oct 5. PubMed
PMID: 23039148.jemann, G.A., 1993. Functional mapping of cortical language areasin adults. Intraoperative approaches. Adv. Neurol. 63, 155—163.
1
O
P
P
R
S
S
S
S
S
S
S
T
T
W
W
888
ldfield, R.C., 1971. The assessment and analysis of handedness:the Edinburgh inventory. Neuropsychologia 9 (Mar (1)), 97—113.
etrovich Brennan, N.M., Whalen, S., de Morales Branco, D., O‘shea,J.P., Norton, I.H., Golby, A.J., 2007. Object naming is a moresensitive measure of speech localization than number counting:converging evidence from direct cortical stimulation and fMRI.NeuroImage 37 (Suppl 1), S100—S108.
reibisch, C., Raab, P., Neumann, K., Euler, H.A., vonGudenberg,A.W., Gall, V., et al., 2003. Event related fMRI for the suppres-sion of speech-associated artifacts in stuttering. NeuroImage 19(Jul (3)), 1076—1084.
eite, M., Edrich, J., Zimmerman, J.T., Zimmerman, J.E., 1978.Human magnetic auditory evoked fields. Electroencephalogr.Clin. Neurophysiol. 45 (Jul (1)), 114—117, PubMed PMID: 78814.
haran, A., Ooi, Y.C., Langfitt, J., Sperling, M.R., 2011. Intrac-arotid amobarbital procedure for epilepsy surgery. EpilepsyBehav. 20 (Feb (2)), 209—213, http://dx.doi.org/10.1016/j.yebeh.2010.11.013, Epub 2010 Dec 28. Review. PubMed PMID:21190900.
tefanatos, G.A., Green, G.G.R., Ratcliff, G.G., 1989. Neu-rophysiological evidence of auditory channel anomalies indevelopmental dysphasia. Arch. Neurol. 46 (August), 871—875.
toppelman, N., Harpaz, T., Ben-Shachar, M., 2013. Do not throw outthe baby with the bath water: choosing an effective baselinefor a functional localizer of speech processing. Brain Behav. 3(May (3)), 211—222, http://dx.doi.org/10.1002/brb3.129, Epub2013 Feb 17. PubMed PMID: 23785653; PubMed Central PMCID:PMC3683281.
uarez, R.O., Whalen, S., O’Shea, J.P., Golby, A.J., 2008. A surgi-cal planning method for functional MRI assessment of languagedominance: influences from threshold, region-of-interest, andstimulus mode. Brain Imaging Behav. 2 (2), 59—73.
uarez, R.O., Whalen, S., Nelson, A.P., Tie, Y., Meadows, M.E.,Radmanesh, A., Golby, A.J., 2009. Threshold-independentfunctional MRI determination of language dominance: avalidation study against clinical gold standards. Epilepsy
W
R.O. Suarez et al.
Behav. 16 (Oct (2)), 288—297, http://dx.doi.org/10.1016/j.yebeh.2009.07.034, Epub 2009 Sep 4. PubMed PMID: 19733509;PubMed Central PMCID: PMC2758322.
uarez, R.O., Golby, A., Whalen, S., Sato, S., Theodore, W.H.,Kufta, C.V., Devinsky, O., Balish, M., Bromfield, E.B., 2010.Contributions to singing ability by the posterior portion ofthe superior temporal gyrus of the non-language-dominanthemisphere: first evidence from subdural cortical stimula-tion, Wada testing, and fMRI. Cortex 46 (Mar (3)), 343—353,http://dx.doi.org/10.1016/j.cortex.2009.04.010, Epub 2009May 18. PubMed PMID: 19570530; PubMed Central PMCID:PMC2821975.
zenkovits, G., Peelle, J.E., Norris, D., Davis, M.H., 2012. Indi-vidual differences in premotor and motor recruitment duringspeech perception. Neuropsychologia 50 (Jun (7)), 1380—1392,http://dx.doi.org/10.1016/j.neuropsychologia.2012.02.023,Epub 2012 Apr 12. PubMed PMID: 22521874.
aimouri, V., Akhondi-Asl, A., Tomas-Fernandez, X., Peters, J.M.,Prabhu, S.P., Poduri, A., Takeoka, M., Loddenkemper, T., Bergin,A.M., Harini, C., Madsen, J.R., Warfield, S.K., 2014. Electrodelocalization for planning surgical resection of the epileptogeniczone inpediatric epilepsy. Int. J. Comput. Assist. Radiol. Surg 9(1), 91—105, Epub 2013 Jun 23. PubMed PMID: 23793723; Pub-MedCentral PMCID: PMC3955988.
ie, Y., Suarez, R.O., Whalen, S., Radmanesh, A., Norton, I.H.,Golby, A.J., 2009. Comparison of blocked and event-related fMRIdesigns for pre-surgical language mapping. NeuroImage 47 (Aug(Suppl 2)), T107—T115, PubMed PMID: 19101639; PubMed CentralPMCID: PMC3036974.
ada, J., Rasmussen, T., 1960. Intracarotid injection of sodiumamytal for the lateralization of cerebral speech dominance. J.Neurosurg. 17, 266—282.
eisenfeld, N.I., Warfield, S.K., 2009. Automatic segmentation of
newborn brain MRI. NeuroImage 47 (Aug (2)), 564—572.ernicke, K., 1874. Der aphasische Symptomencomplex. Einepsychologische Studie auf anatomischer Basis. M. Crohn undWeigert, Breslau.