presurgical functional mr imaging of language and motor · revision received october 30; accepted...

Presurgical Functional MRImaging of Language and MotorFunctions: Validation withIntraoperative Electrocortical Mapping1

Alberto Bizzi, MDValeria Blasi, MDAndrea Falini, MDPaolo Ferroli, MDMarcello Cadioli, PhDUgo Danesi, PhDDomenico Aquino, PhDCarlo Marras, MDDario Caldiroli, MDGiovanni Broggi, MD

Purpose: To prospectively determine the sensitivity and specificityof functional magnetic resonance (MR) imaging for map-ping language and motor functions in patients with a focalmass adjacent to eloquent cortex, by using intraoperativeelectrocortical mapping (ECM) as the reference standard.

Materials andMethods:

The ethics committee approved the study, and patientsgave written informed consent. Thirty-four consecutivepatients (16 women, 18 men; mean age, 43.2 years) wereincluded who met the following three criteria: They had afocal mass in or adjacent to eloquent cortex of the languageor motor system, they had the ability to perform the func-tional MR imaging task, and they had to undergo surgerywith intraoperative ECM. Functional MR imaging withverb generation (n � 17) or finger tapping of the contralat-eral hand (n � 17) was performed at 1.5 T with a blockdesign and an echo-planar gradient-echo T2*-weighted se-quence. Cortex essential for language or hand motor func-tions was mapped with ECM. A site-by-site comparisonbetween functional MR imaging and ECM was performedwith the aid of a neuronavigational device. Sensitivity andspecificity were calculated according to task performed,histopathologic findings, and tumor grade. Exact 95% con-fidence intervals were calculated for each sensitivity andspecificity value.

Results: For 34 consecutive patients, there were 28 with gliomas,two with metastases, one with meningioma, and threewith cavernous angiomas. A total of 251 cortical sites weretested with ECM; overall functional MR imaging sensitivityand specificity were 83% and 82%, respectively. Sensitiv-ity (65%) was lower and specificity (93%) was higher inWorld Health Organization grade IV gliomas comparedwith grade II (sensitivity, 93%; specificity, 79%) and III(sensitivity, 93%; specificity, 76%) gliomas. At 3 monthsafter surgery, language proficiency was unchanged in 15patients; functionality of the contralateral arm was un-changed in 14 patients and improved in one patient.

Conclusion: Functional MR imaging is a sensitive and specific methodfor mapping language and motor functions.

� RSNA, 2008

1 From the Departments of Neuroradiology (A.B., U.D.,D.A.), Neurosurgery (P.F., C.M., G.B.), and Neuroanaes-thesiology (D.C.), Fondazione IRCCS Istituto NeurologicoCarlo Besta, Via Celoria, 11, 20133 Milan, Italy; and De-partment of Neuroradiology and CERMAC, Universita Vita-Salute San Raffaele, Milan, Italy (V.B., A.F., M.C.). Re-ceived July 11, 2007; revision requested September 4;revision received October 30; accepted January 15, 2008;final version accepted February 19. Address correspon-dence to A.B. (e-mail: [email protected] ).

� RSNA, 2008

ORIGINALRESEARCH

�NEURORADIOLOGY

Radiology: Volume 248: Number 2—August 2008 579

Note: This copy is for your personal, non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, use the Radiology Reprints form at the end of this article.

Surgery in eloquent cortex remainsa challenge because of the highrisk of sensorimotor and language

deficits. Although morphologic landmarksfor mapping hand, foot, and face motorfunctions are available at conventionalmagnetic resonance (MR) imaging, map-ping language function is more difficultbecause of the lack of reliable surfacelandmarks (1) and may be less accurateas a result of high individual variability(2,3). In addition, normal sulcal anat-omy often is not recognizable because oftumor growth.

The classic procedure for languagelocalization is intraoperative electrocor-tical mapping (ECM) in awake patients(2,4,5). ECM is used to extend the indi-cations of surgery within eloquent ar-eas, decrease the risk of sequelae, andincrease the quality of tumor resectionwith an effect on survival (6). Research-ers have claimed that ECM is effective,particularly in low-grade gliomas (7).This procedure has shown substantialinterindividual variability in language lo-calization in patients with a focal lesion(2). Localization of speech is even moreproblematic in patients who are fluentin different languages (8,9).

The motor cortex usually is mappedwhile the patient receives a general an-esthetic. When evaluation of language isperformed, however, the patient mustbe awake and cooperative, and the pro-cedure may become time consuming

(6). Functional MR imaging has beenproposed for presurgical planning in pa-tients with a focal mass adjacent to elo-quent cortex (10). Functional MR imag-ing is based on the blood oxygen level–dependent (BOLD) effect: elevation ofMR signal caused by the increased ra-tio between oxygenated and deoxy-genated hemoglobin secondary to on-set of local brain activity. Deoxygen-ated hemoglobin has paramagneticproperties, and its decrement in-creases the MR signal.

Investigators in studies (10,11) haveproved functional MR imaging to behighly sensitive in the detection of sen-sorimotor activation in healthy subjectsand feasible in patients with lesionsaround the central sulcus. Researchersin other studies (10,12–16) have em-phasized a good spatial correlation be-tween motor functional MR imaging andintraoperative ECM results. FunctionalMR imaging has been proved to be atleast as effective as the Wada test in de-termining language dominance (17,18).Investigators in fewer studies have ad-dressed validation of functional MR im-aging in mapping language cortex. Someresearchers (19) have focused on feasi-bility, the choice of the most appropri-ate task, and the spatial relationship be-tween the lesion and activated corticalareas. However, validation studies haveinvolved small samples (20–26). Meth-ods for comparing activation foci havebeen qualitative and subjective (20–23,25,27) rather than quantitative and ob-jective, with a few exceptions (24,26). Iffunctional MR imaging is to be used as areliable and accurate tool for planningand performing function-preserving sur-gery, its results must be validated andcorrelated with clinical outcome.

Thus the purpose of our study wasto prospectively determine the sensitiv-ity and specificity of functional MR im-

aging for mapping language and motorfunctions in patients with a focal massadjacent to eloquent cortex, by usingintraoperative ECM as the referencestandard.

Materials and Methods

PatientsFunctional MR examinations and intraop-erative ECM were performed in 34 con-secutive patients (16 women, 18 men;mean age, 43.2 years; range, 20–69years). Seventeen patients were testedwith verb generation (VGEN); the other17 patients were tested with finger tap-ping. The local institutional ethics reviewboard approved the study; informed con-sent was obtained. Inclusion criteria wereas follows: The patients had a focal massin or adjacent to at least one eloquentarea of the motor or language systems.They had the ability to perform the func-tional MR imaging task. They had to un-dergo intraoperative ECM.

MR ImagingMR imaging studies were performed at1.5 T (Vision or Avanto, Siemens, Er-

Published online before print10.1148/radiol.2482071214

Radiology 2008; 248:579–589

Abbreviations:BOLD � blood oxygen level–dependentECM � electrocortical mappingFN � false negativeFP � false positiveGBM � glioblastoma multiformeTN � true negativeTP � true positiveVGEN � verb generationWHO � World Health Organization

Author contributions:Guarantor of integrity of entire study, A.B.; study con-cepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manu-script revision for important intellectual content, all au-thors; manuscript final version approval, all authors;literature research, A.B., V.B., P.F., C.M., D.C., G.B.; clini-cal studies, A.B., V.B., A.F., P.F., C.M., D.C., G.B.; experi-mental studies, A.B., V.B., M.C., U.D., D.A.; statisticalanalysis, A.B., V.B.; and manuscript editing, A.B., V.B.,A.F., P.F., M.C., C.M., G.B.

Authors stated no financial relationship to disclose.

Advances in Knowledge

� The diagnostic performance offunctional MR imaging maychange according to the grade ofthe glioma: Sensitivity was higherand specificity was lower inWorld Health Organization gradeII and III gliomas than in glioblas-toma multiforme, particularly forfunctional MR imaging of lan-guage.

� In patients with Rolando area tu-mors, the sensitivity and specific-ity of functional MR imaging arehigher (88% and 87%, respec-tively) than in patients with amass near language cortical areas(80% and 78%, respectively).

Implication for Patient Care

� In patients with a focal mass adja-cent to eloquent cortex, functionalMR imaging can help to localizethe center of functional cortexwithin the gyrus that needs to bepreserved during surgery.

NEURORADIOLOGY: Presurgical Functional MR Imaging Bizzi et al

580 Radiology: Volume 248: Number 2—August 2008

langen, Germany; or Intera, Philips,Best, the Netherlands). In addition todaily standardized quality assurancevendor recommendations, a 30-minuteimager stability test with an echo-planargradient-echo sequence and signal-to-noise ratio in a water phantom weremeasured with the three MR imagersmonthly. Actual stability and signal driftwere within these limits: �0.01% and�0.2%, respectively. Functional MR

imaging was performed during VGEN oralternate finger tapping of the hands byusing a T2*-weighted BOLD echo-pla-nar gradient-echo sequence (repetitiontime msec/echo time msec, 3000/52;field of view, 240 mm2; matrix, 128 �128; resolution, 1.9 � 1.9 � 4 mm;number of sections, 25; and no intersec-tion gap). A series of 76 volumes thatincluded four initial dummy images wasobtained in the transverse plane parallel

to the anteroposterior commissuralline. High-spatial-resolution anatomicimaging was performed by using a volu-metric T1-weighted three-dimensionalgradient-echo sequence (repetitiontime msec/echo time msec/inversiontime msec, 1640/2.28/552; flip angle,12°; number of sections, 160; and iso-volumetric resolution, 1.0 mm3).Head motion was restrained withfoam cushions and straps.

Table 1

Demographics, Location of Space-occupying Mass, Neuropathologic Diagnosis, Functional MR Imaging Paradigm, and Resultsof Correlation of Functional MR Imaging with ECM

Patient No./Sex/Age (y) Side Location Pathologic Finding

WHOGliomaGrade Paradigm*

No. of TPTags

No. of TNTags

No. of FPTags

No. of FNTags

Total No.of Tags

1/F/29 Left Frontal, parietal lobes Astrocytoma I Finger tapping 4 2 0 0 62/M/29 Left Frontal lobe Astrocytoma II VGEN 1 2 4 0 73/M/33 Left Frontal lobe Astrocytoma II VGEN 3 1 0 0 44/F/38 Left Frontal, temporal lobes; insula Astrocytoma II VGEN 2 5 0 0 75/M/43 Left Frontal lobe, insula Astrocytoma II VGEN 3 4 0 1 86/M/38 Left Frontal, temporal lobes; insula Astrocytoma II VGEN 1 2 0 0 37/M/33 Left Frontal, temporal lobes; insula Oligoastrocytoma II VGEN 2 3 0 1 68/F/27 Left Temporal lobe, insula Oligoastrocytoma II VGEN 7 3 1 0 119/M/35 Left Frontal lobe Oligoastrocytoma II VGEN 3 5 0 0 8

10/F/28 Left Frontal lobe Oligoastrocytoma II VGEN 1 3 4 0 811/F/49 Left Temporal, parietal lobes Astrocytoma II Finger tapping 5 0 0 1 612/M/42 Right Frontal lobe Astrocytoma II Finger tapping 3 4 0 0 713/M/40 Left Frontal, temporal lobes Astrocytoma II Finger tapping 1 4 0 0 514/F/36 Right Frontal, parietal lobes Astrocytoma II Finger tapping 1 6 0 0 715/M/56 Right Parietal lobe Oligoastrocytoma II Finger tapping 3 4 2 0 916/F/20 Left Temporal, parietal lobes Oligoastrocytoma II Finger tapping 1 0 2 0 317/F/50 Left Temporal lobe, insula Anaplastic astrocytoma III VGEN 3 1 3 0 718/M/68 Left Temporal lobe Anaplastic astrocytoma III VGEN 0 5 1 0 619/M/52 Left Temporal lobe, insula Anaplastic oligodendroglioma III VGEN 1 6 2 0 920/F/58 Left Frontal, temporal lobes; insula Anaplastic oligoastrocytoma III VGEN 6 4 0 1 1121/F/48 Left Parietal lobe Anaplastic oligoastrocytoma III VGEN 3 7 2 0 1222/M/31 Left Frontal lobe Anaplastic oligoastrocytoma III Finger tapping 1 3 0 0 423/M/47 Left Temporal lobe GBM IV VGEN 4 5 0 2 1124/M/35 Left Temporal lobe, insula GBM IV VGEN 5 1 1 5 1225/M/48 Left Temporal lobe GBM IV VGEN 4 5 0 2 1126/M/37 Right Parietal lobe GBM IV Finger tapping 1 5 1 0 727/M/69 Right Frontal lobe GBM IV Finger tapping 2 6 0 0 828/F/68 Left Frontal, parietal lobes GBM IV Finger tapping 1 6 0 0 729/F/56 Right Frontal lobe Meningioma . . . Finger tapping 5 3 0 2 1030/M/63 Left Frontal lobe Metastasis . . . Finger tapping 0 4 0 1 531/F/43 Left Frontal lobe Metastasis . . . Finger tapping 0 2 4 1 732/F/36 Left Frontal lobe Cavernous angioma . . . Finger tapping 2 2 0 0 433/F/53 Left Frontal lobe Cavernous angioma . . . Finger tapping 2 6 0 0 834/F/30 Right Frontal lobe Cavernous angioma . . . Finger tapping 3 4 0 0 7

Note.—For all patients, the total number of TP tags was 84, that of TN tags was 123, that of FP tags was 27, that of FN tags was 17, and the total for all tags combined was 251. GBM � glioblastomamultiforme, WHO � World Health Organization.

* Finger tapping was performed with contralateral hand.



Functional MR Imaging Paradigmsand AnalysisAll patients were trained briefly 15 min-utes before the study; their ability toperform the task correctly was verifiedbefore and after the study by two neu-

rologists (A.B., V.B., with 10 and 7years of experience in functional MRimaging, respectively). For the languagetask, patients were asked to performVGEN silently in response to hearing anoun. For the motor task, patients were

asked to perform alternated finger tap-ping with 2-Hz frequency after acousticand visual cues. During rest periods,patients were asked to mentally countfrom zero to 10 iteratively. A block de-sign with nine 24-second alternating pe-riods was used.

For functional data analysis, an up-dated software package for implement-ing statistical parametric mapping forneuroimaging data (SPM2; WellcomeDepartment of Imaging Neuroscience,University College London, London, En-gland [http://www.fil.ion.ucl.ac.uk/spm])was used. For every functional session,six-parameter rigid-body realignmentwas applied. Smoothing of the realigneddata with a Gaussian kernel by using fullwidth at half maximum (6 � 6 � 6 mm)was performed. We used modeling ofthe expected hemodynamic responsefunction with a block design, convolvedwith the canonic hemodynamic responsefunction of the software package. To avoidloss of signal, a high-pass filter with a fre-quency of 1⁄96 second, which is half the fre-quency of the paradigm used (ie, 1⁄48 sec-ond), and a low-pass filter were applied tothe time series. Application of a P value of.001 uncorrected threshold (minimum of5-voxel clustering) was used to estimate ttest maps. Coregistration of the t test mapswith the anatomic image was performed inorder to have the anatomic localization ofthe functional foci.

Intraoperative ECM: Reference StandardECM of the language cortex was per-formed during asleep-awake anesthesia(28). Image guidance (Stealth StationTreon; Medtronic Surgical NavigationTechnologies, Louisville, Colo) was used inall patients for the surgical approach.During the awake phase, image-guidedstimulation (alternating current of 60Hz, 2 msec, 1–8 mA peak) with a bipo-lar electric probe of a cortical stimulator(Ojemann, model OCS-1, Radionics,Burlington, Mass; Nimbus, Newmedic/Hemodia, Toulouse, France) was per-formed. The current amplitude wasprogressively increased by 1 mA. Bipha-sic square-wave pulses of 2 msec at 60Hz, with maximal duration of a se-quence of pulses of 2 seconds, wereused for the stimulation of the motor

Figure 1

Figure 1: Transverse functional MR images of language production during VGEN derived from series ofT2*-weighted echo-planar MR images (3000/52, 24 � 24-cm field of view, 128 � 128 matrix, 4-mm sectionthickness, no intersection gap) show WHO grade II oligoastrocytoma infiltrating left insula and temporal lobein 27-year-old woman (patient 8). Functional MR threshold maps (P � .001, uncorrected) were overlaid onT1-weighted gradient-echo MR images (1640/2.28/552, 12° flip angle, 160 contiguous sections, 1.0-mm3

isotropic resolution). Yellow and red areas indicate significant voxels with decreasing power. ECM tag loca-tions are indicated in blue if positive and in green if negative; sphere with 10-mm radius indicates distance.Overlapping of blue ECM sphere with functional MR image focus is counted as TP tag; overlapping of greenECM sphere with functional MR image focus is counted as FP tag; no overlapping of blue ECM sphere iscounted as TP (FN) tag; no overlapping of green ECM sphere is counted as TN tag. Comparison of functionalMR image with electrocortical map resulted in seven TP, three TN, one FP, and no FN tags.



function during general anesthesia andfor stimulation of the language functionduring asleep-awake anesthesia. Lan-guage mapping was performed by usingthe largest current that did not produceafterdischarge (2,29) in the range be-tween 4–10 mA. The entire exposedcortex was stimulated, including thepreplanned area of resection. Two neu-rosurgeons (P.F. and G.B., with 14 and35 years of experience, respectively),who were blinded to the results of func-tional MR imaging, performed ECM inall patients.

A continuous multichannel electro-myographic recording was used for thedetection of motor responses togetherwith clinical assessment of movement.Arrest of speech, random answering, orperseveration to stimulation were con-sidered positive sites if confirmed atleast twice. Errors were classified intra-operatively by a trained neurologist(D.C.) with experience in evaluation ofspeech disorders. He was monitoring thepatient’s response and immediately in-formed the surgeon of any error. A num-bered tag identified each stimulated area.Before starting tumor removal, location ofeach ECM tag on three-dimensional MRimages was demonstrated with the aid ofthe image guidance system and screenshotswere archived.

The comparison between ECM-testedareas and functional MR imaging–acti-vated foci was made only on the ex-posed cortical surface. Therefore, func-tional MR imaging–activated foci awayfrom the craniotomy were not testedand validated with ECM.

Measurement of Clinical OutcomeNeurologic examination was performedby one neurosurgeon (C.M.) with 12years of experience daily during the first7 days and at 3 months. Muscle functionof both upper limbs was evaluated andassigned a grade according to the scaleof grades 0–5 reported by De Jong (30).A grade 4 motor deficit of the upperlimb was classified as mild, a grade 2 or3 deficit was classified as moderate, anda grade 0 or 1 deficit was classified assevere. Language performance was de-termined by evaluation of verbal fluency,denomination, and comprehension of sim-

ple objects and categories. Aphasia wasconsidered mild when the patient hadrare anomia, paraphasia, or compre-hension deficits, or all three, that didnot affect patient ability to communi-cate. Aphasia was considered moderatewhen the same deficits occurred withhigher frequency and slowed the patient’sability to communicate. Aphasia was con-sidered severe when these deficits im-paired the patient’s communication andability to perform daily activities.

Statistical AnalysisThe anatomic position of each recordedECM site was recorded on the three-di-mensional functional MR image data set,and a 1-cm-radius sphere was overlaid oneach tag. Electrocortical maps and func-tional MR images were considered to

match when the functional MR image fo-cus was within the volume defined by thesphere (ie, the distance between the twowas 1 cm or less). For each patient, thenumbers of true-positive (TP), true-nega-tive (TN), false-positive (FP), and false-negative (FN) tags were computed. Toaddress the issue of different numbersfrom each patient, it was assumed thatthe positive or negative response fromeither the index or the reference test inone cortical site had no effect on the re-sponse in another site. In other words,each electrocortical stimulation was con-sidered independent. The same assump-tion was made for statistical analysis insimilar published studies (24,21).

Sensitivity and specificity were cal-culated according to the task performed, histopathologic findings, and

Figure 2

Figure 2: Validation of functional MR imaging with ECM in same study as in Figure 1. Intraoperative ECMinterrupted language production during VGEN in five cortical sites in left inferior frontal gyrus and dorsolateralprefrontal cortex, as illustrated in intraoperative optic microscopic view (bottom right). Silk string identifiesfrontal eloquent area. Other eloquent sites were found in posterior temporal lobe (not shown). With aid of neu-ronavigational device, position of each ECM tag on coronal (top left), sagittal (top right), and axial (bottom left)views of presurgical functional MR imaging data set was determined.



tumor grade. For language, sensitivityand specificity also were computed inthree main locations (left inferior fron-tal gyrus or Broca area, left dorsolateralprefrontal cortex, and middle and supe-rior temporal gyri or Wernicke area)according to landmarks established byNaidich et al (31). Exact 95% confi-dence intervals that were based on bi-nomial distribution were calculated foreach sensitivity and specificity valuewith software (SAS, version 8.2; http://www.sas.com).

Results

For the 34 consecutive patients recruitedfrom March 2002 until April 2007, dataare reported in Table 1. Twenty-eight hadgliomas (16 low grade and 12 high grade),two had metastases, one had a meningi-oma, and three had cavernous angiomas(Table 1; Figs 1–5). Time between func-tional MR imaging and intraoperativeECM was within 3 weeks. All patients whounderwent funtional MR imaging and ECMwere included in this study (Fig 6).

Patient Clinical OutcomeReversible postoperative deficits oc-curred in three patients and resolvedwithin 72 hours. Postoperative deficitsof variable intensity were still present infour patients 3 months after surgery(Table 2): The permanent complicationrate was 11%. Deficits were mild in twopatients, with aphasia and contralateralarm hemiparesis in one each; moder-ate contralateral arm hemiparesis waspresent in a third patient. Intraopera-tive bleeding with temporary occlusionof distal branches of the left middle ce-rebral artery resulted in a large infarct,with permanent severe aphasia, in thefourth patient. The total complicationrate (reversible plus permanent, mild,and severe) was 20%.

Functional MR Imaging PerformanceThe three main cortical areas of the lan-guage system were mapped with func-tional MR imaging in 17 of 17 patients.Additional areas were mapped in theanterior cingulate and supplementarymotor area in most patients. Finger tap-ping evoked a BOLD response in theprimary motor cortex and in the supple-mentary motor area ipsilateral to themass in all 17 patients.

A total of 251 ECM sites were tested:141 in patients evaluated with VGENand 110 in patients evaluated with fin-ger tapping. Speech arrest or falteringwas induced in 16 of 17 patients duringasleep-awake anesthesia. Muscle con-traction of the contralateral hand wasevoked in 17 of 17 patients. Overall sen-sitivity and specificity of functional MRimaging in 34 patients were 83% and82%, respectively.

For hand motor function alone, sen-sitivity and specificity were 88% and87%, respectively. For language, sensi-tivity and specificity were 80% and78%, respectively (Table 3). In 10 of 17patients evaluated with VGEN, the tu-mor was abutting the cortical surface.Therefore, it was possible to performECM: At all sites, results were negativewith either functional MR imaging orECM (Table 3).

GBM showed a substantial lowersensitivity and higher specificity thanWHO grade II and III gliomas (Table 4).

Figure 3

Figure 3: Transverse functional MR images derived from series of T2*-weighted echo-planar MR images(3000/52, 24 � 24-cm field of view, 128 � 128 matrix, 4-mm section thickness, no intersection gap) showWHO grade II oligoastrocytoma in right parietal lobe in 56-year-old man (patient 15). Functional MR thresh-old maps (P � .001, uncorrected) were overlaid on T1-weighted gradient-echo MR images (1640/2.28/552,12° flip angle, 160 contiguous sections, 1.0-mm3 isotropic resolution). White, yellow, and red areas indicatesignificant voxels with decreasing power. ECM tag locations are indicated in blue if positive and in green ifnegative; sphere with 10-mm radius indicates distance. Comparison of functional MR image with electrocorti-cal map resulted in three TP, four TN, two FP, and no FN tags.



These differences reflect the highernumber of FN tags and lower number ofFP tags found in GBM.

In patient 18, ECM did not induceany speech arrest or faltering. The pa-tient’s ability to speak was monitoredduring tumor resection. The patient didnot develop any speech deficit during orafter surgery; therefore, it was con-cluded that there was no eloquent areain the exposed cortex, despite the find-ing of one positive functional MR imagefocus that was interpreted as FP.

Discussion

In our study, overall sensitivity andspecificity for mapping language andmotor functions with functional MR im-aging in patients with a focal brain masswere greater than 80%; in addition, wedemonstrated that sensitivity and speci-ficity may change with a higher gliomagrade: Sensitivity was lower and speci-ficity was higher in GBM than in WHOgrade II or III glioma. There are twopossible explanations for the higher rateof FN tags in GBM. GBM is an undiffer-entiated tumor with a rich abnormalvasculature caused by angiogenesis. Neu-rovascular uncoupling has been de-scribed in higher-grade tumors, and itcould lead to loss of the BOLD response(32,33). A larger brain shift occurring inGBM, as the dura is exposed, might bean alternative explanation.

Despite our results, ECM remainsthe reference standard for intraopera-tive decisions when the BOLD responseis shown in the proximity of a mass.Intraoperative ECM has been used ex-tensively, and it has been validated byclinical outcome. If eloquent sites arerespected, the risk of permanent defi-cits is low (2,5).

Patient outcome for permanent se-quelae was 11% in this series of 34 pa-tients, with one patient with severemorbidity due to surgical complicationsindependent of function localization tech-niques and three patients with moderateor mild permanent postoperative deficits.Other studies have shown a low rate ofpostoperative deficits in patients who hadpresurgical functional MR imaging (32).Neither functional MR imaging nor ECM

showed eloquent tissue within the tumorwhen the mass was near the surface andit could be evaluated with ECM. Roux etal (24) showed functional MR imaging in-tratumoral activations in six patients.

Evidence that functional MR imag-ing may have a favorable effect on mea-surement of clinical outcome has notbeen fully established so far. The rela-tionship between the function testedwith functional MR imaging and clinicaloutcome often is indirect. A randomizedcontrolled trial in a large number of pa-tients might be needed to show an ef-

fect. Although there is no clinical evi-dence yet that use of functional MR im-aging has an effect on patient outcome,findings in recent studies support its ef-fect on therapeutic planning (34,35).

The ideal mapping method shouldbe both sensitive and specific. A tech-nique with high sensitivity will have alow rate of FN foci, and it will rarelymiss functioning cortex. With high spec-ificity, the rate of FP sites will stay low.FN sites might eventually lure the neu-rosurgeon to perform a larger resectionwith a higher risk of permanent neuro-

Figure 4

Figure 4: Validation of functional MR imaging with ECM in same study as in Figure 3. Intraoperative ECMevoked hand motion in three cortical sites, as illustrated in intraoperative optic microscopic view (bottom right). Twosilk strings identify primary motor cortex (tag 100 ) and primary sensory cortex (tags 31 and 32 ) separated by centralsulcus. With aid of neuronavigational device, position of each ECM tag on coronal (top left), sagittal (top right), andaxial (bottom left) views of presurgical functional MR imaging data set was determined.

Table 2

Patients’ Clinical Outcome according to Function Tested and Time before and afterSurgery

TaskNo. ofPatients

No. withPreoperativeDeficits

No. with ReversiblePostoperativeDeficits

Postoperative Deficits at3 Months

No. With No. Without

VGEN 17 2 1 2 15Finger tapping of contralateral hand 17 7 2 2 15

Total 34 9 3 4 30



logic sequelae. FP sites also are undesir-able because they may discourage moreextensive resection or surgery at all.

It is important to emphasize thatthere are important physiologic differ-

ences between ECM and functional MRimaging. ECM is a disruption methodthat causes an arrest of function whenan essential area is stimulated. Func-tional MR imaging is an activation-based

method, and a statistically significantBOLD response appears in all areas thatare functionally active, but not neces-sarily essential, during the execution ofa particular task. Compared with ECM,a larger number of active cortical fociare then expected with functional MRimaging. Future clinical outcome studieswill have to determine whether inciden-tal damage to areas with false-positivefoci (positive at functional MR imagingand negative at intraoperative ECM) oc-curring during surgery will cause tran-sient or permanent functional deficits.

Functional MR maps overlaid onthe three-dimensional MR images wereavailable for display with the neuronavi-gational system in the operating room.The location of all stimulated ECM tagson the functional MR image data set wasrecorded. This procedure is importantin the determination of true and falseresults, and it may also reduce opera-tor-dependent bias. The objectivity ofthe method is greater if it is comparedwith the method that uses overlaid digi-tized intraoperative ECM photographsoff-line (21,22,24).

Sensitivity and specificity of func-tional MR imaging were higher in map-

Figure 5

Figure 5: Transverse functional MR images of language production during VGEN derived from series ofT2*-weighted echo-planar MR images (3000/52, 24 � 24-cm field of view, 128 � 128 matrix, 4-mm sectionthickness, no intersection gap) show grade IV GBM in left temporal lobe in 48-year-old man (patient 25).Functional MR threshold maps (P � .001, uncorrected) were overlaid on T1-weighted gradient-echo MRimages (1640/2.28/552, 12° flip angle, 160 contiguous sections, 1.0-mm3 isotropic resolution). Yellow andred areas indicate significant voxels with decreasing power. ECM tag locations are indicated in blue if positiveand in green if negative; sphere with 10-mm radius indicates distance. Comparison of functional MR imagewith electrocortical map resulted in four TP, five TN, no FP, and two FN tags.

Figure 6

Figure 6: Flow diagram of study about diag-nostic accuracy.



ping the primary motor cortex than inmapping the language system. This re-sult was expected, and it is consistentwith results in other published studies.Functional MR image sensitivity was85% in 103 patients (36) and 97% in125 patients (37) with tumor in the pri-mary motor cortex. Finger tapping is areliable and robust paradigm to localizethe hand representation in primary mo-tor cortex. Dislocation of the eloquentmotor cortex by para-Rolando tumors,as confirmed by using functional MR im-aging and ECM, occurred in several ofour patients.

There are few studies in which sen-sitivity and specificity of functional MRimaging of language were measured byusing ECM as the reference standard

(21,22,24,25). In only two studies wasa site-by-site correlation with a largenumber of tags performed (21,24).FitzGerald et al (21) evaluated 140 sitesin 11 patients with five language tasksand found a sensitivity of 81% anda specificity of 53%. Sensitivity washigher for results from all tasks com-bined than for results from a single task.VGEN was the single task with the high-est sensitivity. In the current study, onlythe results of VGEN were correlatedwith findings at ECM, because VGENwas found to be the most robust of ourbattery of language tasks (3) and it wasdecided it would have been impracticaland too time consuming to repeat ECMfor multiple tasks in the operatingroom. Roux et al (24) correlated 426

ECM tags in 14 patients. Sensitivity andspecificity were 59% and 97%, respec-tively, with two tasks combined (VGENand naming), with a P value of less than.005. Decreasing the analysis thresholdimproved the sensitivity to 66% and de-creased specificity to 91%. The sensitiv-ity in the study of Roux et al (24) is lowerthan that in the study of FitzGeraldet al (21) and that in our study. Differ-ences in sensitivity values and specificityvalues among these studies may be ex-pected because of differences in patientpopulation (tumor type and location),craniotomy size, functional MR imageparadigms, and methods used to com-pare functional MR images and electro-cortical maps.

Sensitivity was higher in the frontal

Table 3

Results for Task, Three Main Language Cortical Areas, and Stimulation over Superficial Lesion

Statistical and Tag Data Finger Tapping of Contralateral Hand VGEN Broca Area Left Dorsolateral Prefrontal Cortex Wernicke Area Superficial Lesion

StatisticSensitivity (%)* 88 (73, 96) 80 (68, 89) 100 (78, 100) 89 (71, 98) 64 (42, 82) UndeterminedSpecificity (%)* 87 (77, 94) 78 (67, 86) 68 (43, 87) 58 (28, 85) 85 (65, 96) 100 (86, 100)

No. of tagsTP 35 49 12 24 16 0TN 61 62 13 7 22 20FP 9 18 6 5 4 0FN 5 12 0 3 9 0Total 110 141 31 39 51 20

Note.—Sensitivity was calculated as TP/(TP � FN), and specificity was calculated as TN/(TN � FP).

* Numbers in parentheses are 95% confidence intervals.

Table 4

Results for WHO Glioma Grade and Mass Type

Statistical and Tag DataWHO Glioma Grade

Metastasis Meningioma Cavernous Angioma All MassesI and II III IV All

StatisticSensitivity (%)* 93 (81, 99) 93 (68, 100) 65 (44, 83) 85 (75, 92) 0 (0, 78) 71 (29, 96) 100 (65, 100) 83 (74, 90)Specificity (%)* 79 (66, 88) 76 (59, 89) 93 (78, 99) 82 (74, 88) 60 (26, 88) 100 (37, 100) 100 (78, 100) 82 (75, 88)

No. of tagsTP 41 14 17 72 0 5 7 84TN 48 26 28 102 6 3 12 123FP 13 8 2 23 4 0 0 27FN 3 1 9 13 2 2 0 17Total 105 49 56 210 12 10 19 251

Note.—Sensitivity was calculated as TP/(TP � FN), and specificity was calculated as TN/(TN � FP).

* Numbers in parentheses are 95% confidence intervals.



areas (inferior frontal gyrus and dorso-lateral prefrontal cortex) of languageproduction than in posterior temporalgyri. This result is of interest, but it mayhave been influenced by the use of atask of speech production in this series.Lurito et al (22) focused on functionalMR image mapping of receptive lan-guage in three patients with gliomas inthe temporal and parietal lobes. Theycorrelated 10 ECM tags and foundthat all, except one located within theboundary of the tumor in the left pari-etal lobe, had been mapped by usingfunctional MR imaging. They concludedthat their functional MR imaging para-digm was good but not perfect (22).

Sensitivity and specificity were thehighest in patients with cavernous angi-omas. Pouratian et al (23) reported sim-ilar results in three patients with cav-ernous angioma.

There were limitations to our study.Intrinsic limitations were caused by theimaging modality used: MR imagingis affected by geometric distortionscaused by inhomogeneity of the mag-netic field within the head. Errors oc-curring during coregistration of ana-tomic and functional MR images add upto errors occurring during neuronaviga-tional registration. These errors maysum up to a few millimeters. In ourstudy, only the most robust languageparadigm was used intraoperatively.The use of additional tasks might im-prove sensitivity of functional MR imag-ing, thus reducing the risk of FN results.

In conclusion, functional MR imag-ing is a sensitive and specific method forlocalization of the eloquent cortex of thelanguage and motor systems. Sensitivityof functional MR imaging was higher incavernous angiomas, followed by WHOgrade II and III gliomas, and then GBM.

Acknowledgments: The professional assis-tance of Carlo L. Solero, MD, Giovanni Tringali,MD, and Roberto Cordella, MSc, in the operat-ing room and the assistance of Chiara Falcone,MSc, for statistical analyses are gratefully ac-knowledged.

References1. Quinones-Hinojosa A, Ojemann SG, Sanai N,

Dillon WP, Berger MS. Preoperative corre-

lation of intraoperative cortical mappingwith magnetic resonance imaging landmarksto predict localization of the Broca area.J Neurosurg 2003;99:311–318.

2. Ojemann G, Ojemann J, Lettich E, BergerM. Cortical language localization in left,dominant hemisphere: an electrical stimula-tion mapping investigation in 117 patients.J Neurosurg 1989;71:316–326.

3. Cerliani L, Mandelli ML, Aquino D, Filosa G,Bruzzone MG, Bizzi A. Preoperative map-ping of language with fMRI: frequency mapsand comparison of three tasks in healthysubjects [abstr]. In: Proceedings of the Fif-teenth Annual Meeting of the InternationalSociety for Magnetic Resonance in Medicine.Berkeley, Calif: International Society forMagnetic Resonance in Medicine, 2007; 206.

4. Berger MS, Rostomily RC. Low gradegliomas: functional mapping resection strat-egies, extent of resection, and outcome.J Neurooncol 1997;34:85–101.

5. Ojemann JG, Ojemann GA, Lettich E. Corti-cal stimulation mapping of language cortexby using a verb generation task: effects oflearning and comparison to mapping basedon object naming. J Neurosurg 2002;97:33–38.

6. Duffau H, Lopes M, Arthuis F, et al. Contri-bution of intraoperative electrical stimula-tions in surgery of low grade gliomas: a com-parative study between two series without(1985–96) and with (1996–2003) functionalmapping in the same institution. J NeurolNeurosurg Psychiatry 2005;76:845–851.

7. Keles GE, Lamborn KR, Berger M. Low-grade hemispheric gliomas in adults: a criti-cal review of extent of resection as a factorinfluencing outcome. J Neurosurg 2001;95:735–745.

8. Bello L, Acerbi F, Giussani C, et al. Intraop-erative language localization in multilingualpatients with gliomas. Neurosurgery 2006;59:115–125.

9. Roux FE, Tremoulet M. Organization of lan-guage areas in bilingual patients: a corticalstimulation study. J Neurosurg 2002;97:857–864.

10. Jack CR Jr, Thompson RM, Butts RK, et al.Sensory motor cortex: correlation of presur-gical mapping with functional MR imagingand invasive cortical mapping. Radiology1994;190:85–92.

11. Pujol J, Conesa G, Deus J, et al. Presurgicalidentification of primary sensorimotor cor-tex by functional magnetic resonance imag-ing. J Neurosurg 1996;84:7–13.

12. Mueller WM, Yetkin FZ, Hammeke TA,et al. Functional magnetic resonance imaging

mapping of the motor cortex in patients withcerebral tumors. Neurosurgery 1996;39:515–520.

13. Yetkin FZ, Mueller WM, Morris GL, et al.Functional MR activation correlated with in-traoperative cortical mapping. AJNR Am JNeuroradiol 1997;18:1311–1315.

14. Roux FE, Boulanouar K, Ranjeva JP, et al.Usefulness of motor functional MRI corre-lated to cortical mapping in Rolandic low-grade astrocytomas. Acta Neurochir (Wien)1999;141:71–79.

15. Fandino J, Kollias SS, Wieser HG, ValavanisA, Yonekawa Y. Intraoperative validation offunctional magnetic resonance imaging andcortical reorganization patterns in patientswith brain tumors involving the primary mo-tor cortex. J Neurosurg 1999;91:238–250.

16. Lehericy S, Duffau H, Cornu P, et al. Corre-spondence between functional magnetic res-onance imaging somatotopy and individualbrain anatomy of the central region: compar-ison with intraoperative stimulation in pa-tients with brain tumors. J Neurosurg 2000;92:589–598.

17. Binder JR, Swanson SJ, Hammeke TA, et al.Determination of language dominance usingfunctional MRI: a comparison with the Wadatest. Neurology 1996;46:978–984.

18. Sabbah P, Chassoux F, Leveque C, et al.Functional MR imaging in assessment of lan-guage dominance in epileptic patients. Neu-roimage 2003;18:460–467.

19. Stippich C, Rapps N, Dreyhaupt J, et al.Localizing and lateralizing language in pa-tients with brain tumors: feasibility of rou-tine preoperative functional MR imaging in81 consecutive patients. Radiology 2007;243:828–836.

20. Carpentier A, Pugh KR, Westerveld M, et al.Functional MRI of language processing: de-pendence on input modality and temporallobe epilepsy. Epilepsia 2001;42:1241–1254.

21. FitzGerald DB, Cosgrove GR, Ronner S,et al. Location of language in the cortex: acomparison between functional MR imagingand electrocortical stimulation. AJNR Am JNeuroradiol 1997;18:1529–1539.

22. Lurito JT, Lowe MJ, Sartorius C, MathewsVP. Comparison of fMRI and intraoperativedirect cortical stimulation in localization ofreceptive language areas. J Comput AssistTomogr 2000;24:99–105.

23. Pouratian N, Bookheimer SY, Rex DE, MartinNA, Toga AW. Utility of preoperative func-tional magnetic resonance imaging for iden-tifying language cortices in patients with vas-cular malformations. J Neurosurg 2002;97:21–32.



24. Roux FE, Boulanouar K, Lotterie JA, MejdoubiM, LeSage JP, Berry I. Language functionalmagnetic resonance imaging in preoperativeassessment of language areas: correlationwith direct cortical stimulation. Neurosur-gery 2003;52:1335–1345.

25. Ruge MI, Victor J, Hosain S, et al. Concor-dance between functional magnetic reso-nance imaging and intraoperative languagemapping. Stereotact Funct Neurosurg 1999;72:95–102.

26. Rutten GJ, Ramsey NF, van Rijen PC,Noordmans HJ, van Veelen CW. Develop-ment of a functional magnetic resonance im-aging protocol for intraoperative localizationof critical temporoparietal language areas.Ann Neurol 2002;51:350–360.

27. Grummich P, Nimsky C, Pauli E, BuchfelderM, Ganslandt O. Combining fMRI and MEGincreases the reliability of presurgical lan-guage localization: a clinical study on the dif-ference between and congruence of both mo-dalities. Neuroimage 2006;32:1793–1803.

28. Picht T, Kombos T, Gramm HJ, Brock M,Suess O. Multimodal protocol for awake cra-

niotomy in language cortex tumour surgery.Acta Neurochir (Wien) 2006;148:127–137.

29. Walker JA, Quinones-Hinojosa A, BergerMS. Intraoperative speech mapping in 17 bi-lingual patients undergoing resection of amass lesion. Neurosurgery 2004;54:113–117.

30. De Jong RN. The neurologic examination:incorporating the fundamentals of neuro-anatomy and neurophysiology. Hagerstown,Md: Harper & Row, 1979.

31. Naidich TP, Hof PR, Gannon PJ, Yousry TA,Yousry I. Anatomic substrates of language:emphasizing speech. Neuroimaging Clin NAm 2001;11:305–341.

32. Holodny AI, Schulder M, Liu WC, Wolko J,Maldjian JA, Kalnin AJ. The effect of braintumors on BOLD functional MR imaging ac-tivation in the adjacent motor cortex: impli-cations for image-guided neurosurgery.AJNR Am J Neuroradiol 2000;21:1415–1422.

33. Ulmer JL, Hacein-Bey L, Mathews VP, et al.Lesion-induced pseudo-dominance at func-

tional magnetic resonance imaging: implica-tions for preoperative assessments. Neuro-surgery 2004;55:569–579.

34. Petrella JR, Shah LM, Harris KM, et al. Pre-operative functional MR imaging localizationof language and motor areas: effect on ther-apeutic decision making in patients with po-tentially resectable brain tumors. Radiology2006;240:793–802.

35. Medina LS, Bernal B, Dunoyer C, et al. Sei-zure disorders: functional MR imaging fordiagnostic evaluation and surgical treat-ment—prospective study. Radiology 2005;236:247–253.

36. Krings T, Reinges MH, Erberich S, et al.Functional MRI for presurgical planning:problems, artefacts, and solution strategies.J Neurol Neurosurg Psychiatry 2001;70:749–760.

37. Hirsch J, Ruge MI, Kim KH, et al. An inte-grated functional magnetic resonance imag-ing procedure for preoperative mapping ofcortical areas associated with tactile, motor,language, and visual functions. Neurosurgery2000;47:711–721.



Radiology 2008 This is your reprint order form or pro forma invoice

(Please keep a copy of this document for your records.)

Reprint order forms and purchase orders or prepayments must be received 72 hours after receipt of form either by mail or by fax at 410-820-9765. It is the policy of Cadmus Reprints to issue one invoice per order.

Please print clearly.

Author Name _______________________________________________________________________________________________ Title of Article _______________________________________________________________________________________________ Issue of Journal_______________________________ Reprint # _____________ Publication Date ________________ Number of Pages_______________________________ KB # _____________ Symbol Radiology Color in Article? Yes / No (Please Circle) P lease include the journal name and reprint number or manuscript number on your purchase order or other correspondence.

Order and Shipping Information Reprint Costs (Please see page 2 of 2 for reprint costs/fees.) ________ Number of reprints ordered $_________ ________ Number of color reprints ordered $_________ ________ Number of covers ordered $_________ Subtotal $_________ Taxes $_________ (Add appropriate sales tax for Virginia, Maryland, Pennsylvania, and the District of Columbia or Canadian GST to the reprints if your order is to be shipped to these locations.) First address included, add $32 for each additional shipping address $_________

TOTAL $_________

Country _________________________________________

Quantity __________________ Fax __________________

Shipping Address (cannot ship to a P.O. Box) Please Print Clearly Name ___________________________________________ Institution _________________________________________ Street ___________________________________________ City ____________________ State _____ Zip ___________ Country ___________________________________________ Quantity___________________ Fax ___________________ Phone: Day _________________ Evening _______________ E-mail Address _____________________________________ Additional Shipping Address* (cannot ship to a P.O. Box) Name ___________________________________________ Institution _________________________________________ Street ___________________________________________ City ________________ State ______ Zip ___________

Phone: Day ________________ Evening ______________ E-mail Address ____________________________________ * Add $32 for each additional shipping address

P ayment and Credit Card Details Enclosed: Personal Check ___________ Credit Card Payment Details _________ Checks must be paid in U.S. dollars and drawn on a U.S. Bank. Credit Card: __ VISA __ Am. Exp. __ MasterCard Card Number __________________________________ E xpiration Date_________________________________ Signature: _____________________________________ Please send your order form and prepayment made payable to: Cadmus Reprints P.O. Box 751903 Charlotte, NC 28275-1903 Note: Do not send express packages to this location, PO Box.

FEIN #:541274108

Invoice or Credit Card Information Invoice Address Please Print Clearly Please complete Invoice address as it appears on credit card statement Name ____________________________________________ Institution ________________________________________ Department _______________________________________ Street ____________________________________________ City ________________________ State _____ Zip _______ Country ___________________________________________ Phone _____________________ Fax _________________ E-mail Address _____________________________________ Cadmus will process credit cards and Cadmus Journal

Services will appear on the credit card statement. If you don’t mail your order form, you may fax it to 410-820-9765 with

your credit card information.

Signature __________________________________________ Date _______________________________________ Signature is required. By signing this form, the author agrees to accept the responsibility for the payment of reprints and/or all charges described in this document. RB-9/26/07

Page 1 of 2

Black and White Reprint Prices

Domestic (USA only) # of

Pages 50 100 200 300 400 500

1-4 $221 $233 $268 $285 $303 $323 5-8 $355 $382 $432 $466 $510 $544 9-12 $466 $513 $595 $652 $714 $775

13-16 $576 $640 $749 $830 $912 $995 17-20 $694 $775 $906 $1,017 $1,117 $1,22021-24 $809 $906 $1,071 $1,200 $1,321 $1,47125-28 $928 $1,041 $1,242 $1,390 $1,544 $1,68829-32 $1,042 $1,178 $1,403 $1,568 $1,751 $1,924

Covers $97 $118 $215 $323 $442 $555

International (includes Canada and Mexico) # of

Pages 50 100 200 300 400 500

1-4 $272 $283 $340 $397 $446 $506 5-8 $428 $455 $576 $675 $784 $884 9-12 $580 $626 $805 $964 $1,115 $1,278

13-16 $724 $786 $1,023 $1,232 $1,445 $1,65217-20 $878 $958 $1,246 $1,520 $1,774 $2,03021-24 $1,022 $1,119 $1,474 $1,795 $2,108 $2,42625-28 $1,176 $1,291 $1,700 $2,070 $2,450 $2,81329-32 $1,316 $1,452 $1,936 $2,355 $2,784 $3,209

Covers $156 $176 $335 $525 $716 $905 Minimum order is 50 copies. For orders larger than 500 copies, please consult Cadmus Reprints at 800-407-9190. Reprint Cover Cover prices are listed above. The cover will include the publication title, article title, and author name in black. Shipping Shipping costs are included in the reprint prices. Domestic orders are shipped via UPS Ground service. Foreign orders are shipped via a proof of delivery air service. Multiple Shipments Orders can be shipped to more than one location. Please be aware that it will cost $32 for each additional location. Delivery Your order will be shipped within 2 weeks of the journal print date. Allow extra time for delivery.

Color Reprint Prices

Domestic (USA only) # of

Pages 50 100 200 300 400 500

1-4 $223 $239 $352 $473 $597 $719 5-8 $349 $401 $601 $849 $1,099 $1,3499-12 $486 $517 $852 $1,232 $1,609 $1,992

13-16 $615 $651 $1,105 $1,609 $2,117 $2,62417-20 $759 $787 $1,357 $1,997 $2,626 $3,26021-24 $897 $924 $1,611 $2,376 $3,135 $3,90525-28 $1,033 $1,071 $1,873 $2,757 $3,650 $4,53629-32 $1,175 $1,208 $2,122 $3,138 $4,162 $5,180

Covers $97 $118 $215 $323 $442 $555

International (includes Canada and Mexico)) # of

Pages 50 100 200 300 400 500

1-4 $278 $290 $424 $586 $741 $904 5-8 $429 $472 $746 $1,058 $1,374 $1,6909-12 $604 $629 $1,061 $1,545 $2,011 $2,494

13-16 $766 $797 $1,378 $2,013 $2,647 $3,28017-20 $945 $972 $1,698 $2,499 $3,282 $4,06921-24 $1,110 $1,139 $2,015 $2,970 $3,921 $4,87325-28 $1,290 $1,321 $2,333 $3,437 $4,556 $5,66129-32 $1,455 $1,482 $2,652 $3,924 $5,193 $6,462

Covers $156 $176 $335 $525 $716 $905 Tax Due Residents of Virginia, Maryland, Pennsylvania, and the District of Columbia are required to add the appropriate sales tax to each reprint order. For orders shipped to Canada, please add 7% Canadian GST unless exemption is claimed. Ordering Reprint order forms and purchase order or prepayment is required to process your order. Please reference journal name and reprint number or manuscript number on any correspondence. You may use the reverse side of this form as a proforma invoice. Please return your order form and prepayment to: Cadmus Reprints P.O. Box 751903 Charlotte, NC 28275-1903 Note: Do not send express packages to this location, PO Box. FEIN #:541274108

Reprint Order Forms and purchase order or prepayments must be received 72 hours after receipt of form.

Please direct all inquiries to:

Rose A. Baynard 800-407-9190 (toll free number) 410-819-3966 (direct number) 410-820-9765 (FAX number)

[email protected] (e-mail)

Page 2 of 2

Radiology 2008

mailto:[email protected]

presurgical functional mr imaging of language and motor · revision received october 30; accepted...

Documents