Magnetic Resonance Imaging of Neonatal Encephalopathy at 4.7 Tesla: Initial Experiences

DOI: 10.1542/peds.2006-1499; originally published online November 13, 2006;Pediatrics

J. RobertsonNicolaLombard, Wui K. Chong, John S. Wyatt, Ernest B. Cady, Roger J. Ordidge and

Enrico De Vita, Alan Bainbridge, Jeanie L. Y. Cheong, Cornelia Hagmann, RosarieExperiences

Magnetic Resonance Imaging of Neonatal Encephalopathy at 4.7 Tesla: Initial  

  http://pediatrics.aappublications.org/content/early/2006/11/13/peds.2006-1499.citation

located on the World Wide Web at: The online version of this article, along with updated information and services, is

of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2006 by the American Academy published, and trademarked by the American Academy of Pediatrics, 141 Northwest Pointpublication, it has been published continuously since 1948. PEDIATRICS is owned, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from

ARTICLE

Magnetic Resonance Imaging of NeonatalEncephalopathy at 4.7 Tesla: Initial ExperiencesEnrico De Vita, PhDa,b, Alan Bainbridge, PhDa, Jeanie L. Y. Cheong, FRACPc, Cornelia Hagmann, MDc, Rosarie Lombard, ANNP, MScd,

Wui K. Chong, MD, FRCRe, John S. Wyatt, FRCPCHc, Ernest B. Cady, FInstP, BSca,b, Roger J. Ordidge, PhDb, Nicola J. Robertson, FRCPCH, PhDc

aDepartment of Medical Physics and Bio-Engineering and dNeonatal Intensive Care Unit, Elizabeth Garrett Anderson Hospital, University College London HospitalsNational Health Service Foundation Trust, London, United Kingdom; bDepartment of Medical Physics and Bio-Engineering, cCentre for Perinatal Brain Research, Institutefor Women’s Health, and eRadiology and Physics Unit, Institute of Child Health, University College London, London, United Kingdom

The authors have indicated they have no financial relationships relevant to this article to disclose.

ABSTRACT

OBJECTIVES. The goals were to develop safe 4.7-T MRI examination protocols fornewborn infants and to explore the advantages of this field strength in neonatalencephalopathy.

METHODS.Nine ventilated newborn infants with moderate or severe encephalopathywere studied at 4.7 T, with ethical approval and informed parental consent. Thecustom-made, 4.7-T-compatible, neonatal patient management system includedacoustic noise protection and physiologic monitoring. An adult head coil was used.Acquisition parameters for T2-weighted fast spin echo MRI and a variety ofT1-weighted methods were adapted for MRI of neonatal brain at 4.7 T. The pulsesequences used had a radiofrequency specific absorption rate of �2 W/kg.

RESULTS. Physiologic measures were normal throughout each scan. T2-weighted fastspin echo imaging provided better anatomic resolution and gray/white mattercontrast than typically obtained at 1.5 T; T1-weighted images were less impressive.

CONCLUSIONS.With appropriate safety precautions, MRI of newborn infants under-going intensive care is as feasible at 4.7 T as it is at 1.5 T; our initial studiesproduced T2-weighted fast spin echo images with more detail than commonlyobtained at 1.5 T. Although T1-weighted images were not adequately informative,additional pulse sequence optimization may be advantageous. A smaller neonatalhead coil should also permit greater flexibility in acquisition parameters and evenmore anatomic resolution and tissue contrast. In neonatal encephalopathy, inter-pretation of the T2-weighted pathologic detail in combination with comprehensiveneurodevelopmental follow-up should improve prognostic accuracy and enablemore patient-specific therapeutic interventions. In addition, more precise relation-ships between structural changes and functional impairment may be defined.

www.pediatrics.org/cgi/doi/10.1542/peds.2006-1499

doi:10.1542/peds.2006-1499

KeyWordsmagnetic resonance imaging, brain,neonatal encephalopathy, high-field MRI,developmental outcomes

AbbreviationsFOV—field of viewFSE—fast spin echoGWMC—gray/white matter contrastIR—inversion recoveryMDEFT—modified driven-equilibriumFourier transformNE—neonatal encephalopathySAR—specific absorption rateSNR—signal/noise ratioTE—echo timeTI—inversion timeTR—repetition time

Accepted for publication Jul 18, 2006

Address correspondence to Nicola J.Robertson, FRCPCH, PhD, Department ofPaediatrics and Child Health, UniversityCollege London, 5 University St, London WC1E6JJ, United Kingdom. E-mail: [email protected]

PEDIATRICS (ISSN Numbers: Print, 0031-4005;Online, 1098-4275). Copyright © 2006 by theAmerican Academy of Pediatrics

e1812 DE VITA et al by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from

MRI AT 1.5 T characterizes brain development andperinatal brain injury with detail surpassing that

provided by cranial ultrasonography and computed to-mography.1,2 MRI has a particularly important diagnosticand prognostic role for infants with neonatal encepha-lopathy (NE); conventional T1-weighted and T2-weighted MRI performed at 5 to 14 days of age providesthe most specific means of predicting the pattern ofneuromotor outcomes and is a recommended standardof clinical care.3 NE has a wide range of prenatal, intra-partum, and postpartum risk factors, and MRI is partic-ularly useful in diagnosing cerebral developmental ab-normalities, focal lesions, stroke, and metabolic disordersthat may present with NE.

In the past 3 decades, there has been a dramatic increasein clinical MRI field strengths, from 15 mT to 9.4 T.4

There are many anticipated benefits of MRI at higher fieldstrengths, such as increased signal/noise ratio (SNR) andgreater sensitivity to susceptibility-related contrast mecha-nisms. Properly harnessed, these benefits can result in in-creased spatial resolution and tissue contrast or shorterexamination times. However, technical and safety issuemust also be considered; greater installation difficulties,increased radio-frequency power deposition, more acousticnoise, and risk of peripheral nerve stimulation from thepulsed magnetic-field gradients are associated to higherMRI field strengths. Because of increased susceptibility-induced geometric distortion and signal dropout, increasedradiofrequency field inhomogeneity, and altered relax-ation times, the anticipated tissue contrast improvementmay not be achieved. Given the high cost and complexityof high-field MRI systems, it is important to assess theirbenefits and limitations for clinical practice.

MRI scanners operating at 7 T and 8 T have been usedto study adult subjects for several years.5,6 However, MRIof term and preterm infants at �3 T is relatively re-cent7–11 and raises particular safety concerns, becausethese subjects are unable to report unpleasant sensationsand may be more sensitive to magnetic fields, tissueheating, gradient switching, and acoustic noise.12 Con-servative safety limits and continuous physiologic mon-itoring are essential.

Our objectives were to develop safe effective protocolsto study newborn infants at 4.7 T and to assess whetherconventional MRI at this field strength is advantageous forinfants with moderate or severe NE. Our report describessafety precautions and initial experiences of pulse sequenceoptimization, and we present 4.7-T neonatal brain images.Proton magnetic resonance spectroscopy at 4.7 T was alsoperformed and was reported elsewhere.13

METHODS

Subjects and Patient HandlingNine subjects were recruited consecutively between Mayand December 2004, including 5 infants with eithermoderate or severe encephalopathy at birth and 4 in-

fants with normal neurologic status at birth and subse-quent development of abnormal neurologic signs or sei-zures during the neonatal period. Infants were includedif there were both abnormal neurologic examinationresults14 and abnormal background activity or seizureson the amplitude-integrated electroencephalogram. Eth-ical approval was granted by our hospital ethics commit-tee (reference no. 03/0292), and informed parental con-sent was obtained.

Infants were studied in a perspex pod lined withsound-absorber (Sonex; Illbruck, Minneapolis, MN),ventilated with a MRI-compatible ventilator (BabyPacB100; Pneupac, Watford, United Kingdom), and moni-tored with electrocardiography, pulse oximetry (8600FO;Nonin Medical, Plymouth, MN), and skin temperaturemeasurements, in the presence of 2 neonatologists ex-perienced in MRI. Maintenance fluids were adjusted byusing remote infusion pumps.

Two infants (subjects 3 and 4 in Table 1) had under-gone whole-body cooling to a rectal temperature of33.5°C for 72 hours, as part of a clinical therapeutic trial(Whole-Body Hypothermia for the Treatment of Peri-natal Asphyxial Encephalopathy; TOBY trial, www.npeu.ox.ac.uk/toby). At the time of MRI, these infants werenormothermic.

Clinical Measurements and Neurodevelopmental OutcomesClinical measurements included cord pH, base excess,Apgar scores, and amplitude-integrated electroencepha-lography; details of these are shown in Table 1. At 10 to13 months of age, all 7 surviving subjects underwent astandard neurologic assessment15 and a Griffiths devel-opmental assessment16 performed by an experiencedneonatologist. Infants were classified as having normaloutcome (normal neurologic assessment results and aGriffiths quotient of �85), mild/moderate outcome(neuromotor signs but no functional difficulties or aGriffiths quotient between 75 and 84 in �1 subscale), orsevere outcome (functional motor or sensory deficits ora Griffiths quotient of �75 in �1 subscale) (Table 1).

MRI SafetyBefore the study, electrocardiographic and skin temper-ature measurement equipment was tested for MRI-in-duced heating at different positions within the MRI coilused in this study. Pulsed magnetic field gradient acous-tic noise was reduced by using trimmed ear plugs (at-tenuation: �24 dB; Earsoft; Aearo, Indianapolis, IN) andMinimuffs (attenuation: 7 dB; Natus, San Carlos, CA), inaddition to the sound-absorbing pod lining, to maintainthe estimated peak sound pressure level within the pa-tient’s ear at �90 dB, as recommended by the AmericanAcademy of Pediatrics.12 The radiofrequency power spe-cific absorption rate (SAR) was monitored and MRI se-quences were adjusted to ensure that the rate was belowthe Medical Devices Agency (London, United Kingdom)

PEDIATRICS Volume 118, Number 6, December 2006 e1813 by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from

TABLE1

ClinicalDetailsof

theInfantsStud

ied

Infant

No.

Birth

Gestatio

nal

Age,

wk

�d

Birth

Weigh

t,g

Delivery

Cord

pHa

Base

Excess,

mmol/L

Apgar

Scores

at1/5/10

min

Stud

yGPA

,wk

�d

EEGResults

Diagn

osis

TOBY

Trial

Inclusion

Neurodevelopm

ental

Outcomeat

10–13mo

140

�3

3060

EmCS;thick,m

econium-stained

liquor

6.95

�18

0/NA/NA

41�

3Abnormalbackground

activity,noseizures

PerinatalHI,Sarnatstage

2No

Normal

238

�4

2660

SVD;thick,m

econium-stained

liquor

6.7

�27

6/NA/NA

39�

3Seizures,mildlyabnormal

background

activity

PerinatalHI,Sarnatstage

2No

Normal

337

�5

2428

CordprolapseandEm

CS7.37

�2.6

3/5/NA

38�

5Abnormalbackground

activity

IUGR,perinatalHI,Sarnatstage

1Cooled

for72h

Mild/m

oderate

439

�4

3280

EmCS

7.12

�9.8

0/2/2

41�

4Burstsuppression

PerinatalHI,Sarnatstage

2Cooled

for72h

Severe

538

�0

3390

EmCS;thick,m

econium-stained

liquor

6.92

�20.3

4/7/NA

39�

0Statusepilepticu

sPerinatalHI,Sarnatstage

2No

tcooled

Severe

641

�6

3800

Prolongedsecond

stage;Em

CSNA

NA

9/10/NA

42�

3Left-sided

seizures

Leftmiddlecerebralartery

infarction

NoMild/m

oderate,right

hemiplegia

739

�0

3120

SVD,admitted

onday3with

SBRlevel

of670

�mol/L(39mg/dL)

NA

NA

9/10/NA

43�

0Seizures,diffuseabnormal

corticalactivity

Acutebilirubinencephalopathy,

G6PD

deficiency

NoSevere

841

�2

3760

Ventouseextraction

6.45

(capillary)

�34

8/9/NA

41�

5Seizures,little

cortical

activity

Subgalealhem

orrhage

NoDied

927

�1

1320

Chorioam

nionitisand

EmCS

6.99

(capillaryafterH

I)�1.5

2/6/8

42�

6Lowvoltage/slow

background

activity

Chylothoraces,CLD,NEC,cardiac

arrestatGPAof

�41

wk

NoDied

CLDindicateschroniclung

disease;EEG,electro

encephalogram;EmCS,emergencycesarean

section;GP

A,gestationalpluspostnatalage;G6PD,glucose-6-phosphatedehydrogenase;HI,hypoxia-ischemia;IUGR

,intrauterinegrow

threstriction;NEC,necrotizing

enterocolitis;NA,notavailable;SBR,serumbilirubin;SVD

,spontaneous

vaginaldelivery;TO

BY,w

hole-bodyhypothermiaforthe

treatmentofperinatalasphyxialencephalopathy.

aFrom

theum

bilicalcordunlessotherwise

stated.

e1814 DE VITA et al by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from

recommendation of 2 W/kg for fetuses.17 As a conse-quence, fewer slices than desired were imaged in someinstances.

MRI MethodsA Surrey Medical Imaging Systems MR5000 4.7-T sys-tem (supported by Philips, Noord, Netherlands) wasused with the 28-cm diameter, birdcage, adult head coilsupplied by the manufacturer. Coil tuning and matchingwere optimized by using a loading ring (inner diameter:152 mm; outer diameter: 193 mm), filled with dopedsaline solution (300 mmol/L NaCl, 18 mmol/L MnCl2),positioned around the head.

MRI sequences were selected on the basis of previousimaging experience of adult brain at 4.7 T and typicalneonatal protocols at lower field strengths. MRI acqui-sition parameters were first estimated from the knownfield dependences of T1 and T2 in adult brain, comparedwith neonatal values at 1.5 T. After the first acquisitions,the parameters were adjusted according to measuredestimates of relaxation times, to improve image quality(eg, for the first patient, fast spin echo [FSE] imaging wasperformed with 3 different effective echo time [TE] val-ues).

T1-weighted MRI used a conventional 8- or 10-slicesingle spin echo sequence with acquisition bandwidth of25 kHz, TE of 13 milliseconds, repetition time (TR) of800 milliseconds, 2 averages, slice thickness of 5 mm,and interslice gap of 1 mm. For the first 4 subjects, weused a field of view (FOV) of 160 mm � 160 mm, datamatrix of 256 � 128, and nominal in-plane resolution of0.63 mm � 1.25 mm, which yielded a total acquisitiontime of 204 seconds; subsequently, the FOV was 230 mm� 172.5 mm and data matrix was 512 � 192, whichresulted in an improved in-plane resolution of 0.45 mm� 0.9 mm but increased the acquisition time to 307seconds.

T2-weighted MRI used a 7- or 8-slice 8-echo FSEsequence, which was optimized previously at 4.7 T foradult brain,18,19 with FOV of 240 mm � 360 mm, datamatrix of 512 � 768, in-plane pixel dimensions of 0.47mm � 0.47 mm, acquisition bandwidth of 50 kHz, ef-fective TE of 88 milliseconds (echo spacing: 22 millisec-onds), TR of 3.5 seconds, slice thickness of 2.0 mm, andinterslice gap of 0.7 mm; the acquisition time was 340seconds. The following sequences were also used: (1) for3 patients, an inversion recovery (IR)-FSE sequencewith TR of 5000 milliseconds, TE of 22 milliseconds, andinversion time (TI) of 1400 milliseconds, with 8 slicescovering the basal ganglia and with FOV, data matrix,slice thickness/gap, and spatial resolution as for T2-weighted FSE imaging; (2) for 2 patients, an adapted,3-dimensional, modified driven-equilibrium Fouriertransform (MDEFT) sequence20 with TR of 13 millisec-onds, TE of 4 milliseconds, TI of 367 milliseconds, delaybetween saturation and inversion pulses of 206 millisec-

onds, nominal flip angle of 18.8°, acquisition bandwidthof 25 kHz, data matrix of 256 � 224 � 176, isotropic1-mm spatial resolution, and acquisition time of 12 min-utes; (3) for 5 patients (successful for only 3) for T2relaxometry, a spin echo sequence similar to that forT1-weighted imaging, with FOV of 160 mm � 160 mmand TR of 825 milliseconds but with only 5 axial slices, asingle signal average, and TE of 14, 50, 100, and 140milliseconds. From the latter, T2 maps were calculated(pixel by pixel) by using Matlab (MathWorks, Natick,MA); the signal amplitudes (S) for the 4 TE values werefitted to the monoexponential equation S(TE) �S(0)exp(�TE/T2). T2 values were measured over re-gions of interest in white matter and cortical and deepgray matter (Fig 1E). Because quantitative T2 relaxom-etry measurements were obtained for only 3 infants, itwas inappropriate to compare these results with out-comes. Therefore, we present only a representative T2map for a single subject.

Image Interpretation and Quality AssessmentMRI scans were interpreted qualitatively by a pediatricneuroradiologist (W.K.C.) experienced in neonatal neu-rologic MRI at �1.5 T. Gray/white matter contrast(GWMC) was assessed on the basis of the ease of distin-guishing the cortical ribbon from adjacent white matter.

RESULTS

SubjectsThe clinical details and neurodevelopmental outcomesare detailed in Table 1.

Physiologic Monitoring SafetyNo radiofrequency heating or other adverse effects weredetected during MRI.

Patient HandlingPeak inspiratory and end-inspiratory ventilator pres-sures decreased �10% in the magnet bore, and theventilator was adjusted to ensure appropriate constantpressure. Physiologic indices were stable during study,and there were no MRI-related adverse events. The totalexamination time was 2 to 2.5 hours (including 1–1.5hours for magnetic resonance spectroscopy).

MRI

General FindingsRepresentative 4.7-T MRI brain scans for 4 of the 9infants are displayed in Figs 1 to 4. T2-weighted FSE MRIgave remarkable GWMC and excellent structural detail;the small voxel dimensions and brief echo train enabledgood delineation of small structures, with minimal blur-ring and partial-volume effects. All representative im-ages used 2-mm slice thickness. White matter appearedmore heterogeneous than generally seen at 1.5 T. T1-

PEDIATRICS Volume 118, Number 6, December 2006 e1815 by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from

weighted GWMC was not as good. Some images showedrespiration artifacts.

Perinatal Hypoxic-Ischemic InjuryImages for infant 2 (normal outcome) revealed a ceph-alhematoma over the right posterior cranium (Fig 1). Inthe T2-weighted image (Fig 1A), vascular features andperiventricular spaces were particularly well defined,compared with typical 1.5-T images. There was someevidence of capsular edema (slightly worse on the left),slight thalamic and basal ganglion swelling, and promi-nence of the deep white matter and periatrial veins. Spinecho T1-weighted MRI (Fig 1B) provided worse GWMC;

with IR-FSE MRI (Fig 1C), GWMC was good, althoughwhite matter appeared relatively featureless and thethalami and basal ganglia were poorly delineated, be-cause of lower SNR resulting from the 1400-millisecondTI (which maximizes GWMC while minimizing cerebro-spinal fluid signal). Figure 1D shows a 3-dimensionalMDEFT image. With the lower SNR associated with1-mm isotropic resolution, no prominent white matterfeatures were seen; however, GWMC was good, and theposterior limb of the internal capsule and other deepgray matter structures were well delineated. A quanti-tative T2 map and the regions of interest from which T2values were measured are shown in Fig 1E. The cortical

FIGURE 1Images for infant 2, who was studied because of suspected perinatal hypoxic-ischemic cerebral injury but had a normal outcome. A, High-resolution, axial, T2-weighted, 4.7-T FSE MRI(voxels of 0.47mm� 0.47mm� 2mm) at the level of the basal ganglia. GWMC is very good. The periatrial and deepwhitematter veins are particularly apparent. Themyelinated andunmyelinated parts of the posterior limbs of the internal capsule are clearly delineated. B, T1-weighted spin echo MRI, which provided poor contrast, compared with the excellentGWMC obtained with T2-weighted FSE MRI. However, in this image (voxels of 0.45 mm � 0.90 mm � 5 mm; positioned slightly superior to A), a cephalhematoma over the rightposterior cranium (bottom left) can be observed clearly. The slightly increased overall signal intensity in the middle of the image, relative to the periphery, is a radiofrequency fieldinhomogeneity effect. C, High-resolution, axial, IR-FSE MRI (voxels of 0.47 mm � 0.47 mm � 2 mm), at the same level as B, showing good GWMC and cerebrospinal fluid signalsuppression. Despite these attributes, the relatively low SNR does not facilitate delineation of different nuclei within the lentiform nucleus or the thalamus. However, both anterior andposterior limbs of the internal capsule are visible. D, 3-dimensional MDEFT image (voxels of 1 mm� 1 mm� 1 mm) with a slice position similar to that of B and C; the voxel volumefor this protocol was one half that of the spin echo T1-weighted image (B) and approximately twice that of the IR-FSE image (C). GWMC is good, and both anterior and posterior limbsof the internal capsule are identifiable. E, Quantitative T2map (voxels of 0.63mm� 1.25mm� 5mm), at a similar position as B to D. There is good GWMC. The regions of interest usedfor T2 estimation are shown.

e1816 DE VITA et al by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from

gray matter T2 was 82 � 5 milliseconds (mean � SD),white matter (frontal and occipital) T2 was 128 � 4milliseconds, basal ganglion T2 was 90 � 2 milliseconds,and thalamic T2 was 72 � 1 milliseconds.

T2-weighted FSE MRI scans for infant 4 (severe out-come) are shown in Fig 2. The anatomic detail wasremarkable; visible structures included the septal nuclei,the claustra separating external and extreme capsules,the internal and external medullary laminae of the globipallidi, and conspicuous hippocampal internal architec-ture (Fig 2, A and B). Focal abnormalities, consistentwith severe, acute, term, hypoxic-ischemic injury, couldbe seen in the posterior putamina, ventrolateral thalami,and intervening internal capsule. T1-weighted MRIshowed abnormal increased signal in the lentiform nu-clei and thalami (Fig 2C); there was little GWMC, how-ever.

Middle Cerebral Artery InfarctionT2-weighted images for infant 6 (mild/moderate out-come) demonstrated extensive left-hemispheric involve-ment, left-sided cortical swelling, and reduced left-sidedparietal GWMC (Fig 3). Involvement of the left caudatenucleus and anterior putamen was also noted. The MRIabnormalities were typical of extensive, acute, left mid-dle cerebral artery territory infarction. T1-weighted MRI(not shown) provided no additional useful information.MRI angiography was not performed, but the proximalsegments of both middle cerebral arteries appeared nor-mal and symmetrical, whereas the peripheral brancheson the left appeared attenuated.

Acute Bilirubin EncephalopathyBilateral and symmetrical signal abnormalities of theglobi pallidi were revealed on T2-weighted images forinfant 7 (severe outcome resulting from kernicterus)(Fig 4). The adjacent parts of the caudate, putamina, andcapsular white matter also appeared involved, but to alesser degree.

FIGURE 2Images for infant 4, who had perinatal hypoxic-ischemic cerebral injury and a severe outcome. A and B, High-resolution T2-weighted FSEMRIs (voxels of 0.47mm� 0.47mm� 2mm)fromneighboring slices at the level of the basal ganglia. As in Fig 1A, there is excellent GWMCanddemonstration of fine anatomic detail. In A, arrows indicate the anterior cerebral arterybranch (a), the septal nuclei (b) and fornix (c), the claustrum separating the extreme and external capsules (d), the medullary lamina separating the external and internal pallidus (e), aparenchymal vessel in the thalamic region (f), and the hippocampus, with demonstrable internal architecture (g). In B, arrows indicate a vein bordering the left anterior horn (a), the leftmedullary lamina (b), an apparently abnormal hypointense posterior part of the left internal globus pallidus (c), lateral (d) and medial (e) geniculate bodies, the superior colliculus (f),and symmetrical edema of the periventricular hypothalamic tissues near thewalls of the third ventricle (g). C, T1-weighted spin echoMRI at the level of the basal ganglia (voxels of 0.63mm � 1.25 mm � 5 mm). Despite poor GWMC, abnormally increased lentiform nucleus (a and b) and thalamic (c) signal intensities were noted. We speculate that such T1hyperintensity is caused by the presence of methemoglobin as a consequence of hemorrhagic transformation after ischemic injury. Radiofrequency field inhomogeneity results inslightly decreased overall signal intensity in the middle of the image, relative to the periphery.

FIGURE 3High-resolution, T2-weighted, FSEMRI (voxels of 0.47mm� 0.47mm� 2mm) for infant6, who had a mild/moderate outcome, showing MRI abnormalities typical of extensive,acute, leftmiddle cerebral artery infarction. The image shows extensive, left-hemispheric,cortical swelling, greatly reduced parietal GWMC, and peri-Sylvian cortex abnormality.Involvement of the left caudate nucleus and anterior putamen was also evident. T1-weighted MRI (not shown) was not informative in this case.

PEDIATRICS Volume 118, Number 6, December 2006 e1817 by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from

DISCUSSION

SafetyOur initial experience demonstrated that, with appropri-ate precautions, newborn infants undergoing intensivecare could be studied safely at 4.7 T.

T2-Weighted ImagingAppropriate imaging sequence selection and acquisitionparameter optimization accounting for the differentbrain-water relaxation times at 4.7 T produced T2-weighted images with high spatial resolution (pixels of0.47 mm � 0.47 mm � 2.0 mm) and excellent GWMCwithin 6 minutes (Figs 1A, 2A, 2B, 3, and 4). Imageclarity resulted not only from the small nominal in-slicepixel dimensions but also from the small slice thicknessand consequent reduced partial-volume effect.

T1-Weighted ImagingConventional T1-weighted spin echo imaging did notprovide good GWMC at 4.7 T. This was probably attrib-utable to increased brain-water T1, compared with, forexample, 1.5 T,21,22 which typically is addressed by using3-dimensional fast low-angle shot imaging,23 its magne-tization-prepared counterpart,24,25 or MDEFT sequenc-es.20,26,27 However, pulse-sequence optimization for neo-natal brain is further complicated by the fact that most ofthe newborn’s white matter is unmyelinated and char-acterized by an increased water content close to thewater content of grey matter.8,28 This has ramificationsfor the selection of optimal pulse sequences and acqui-sition parameters at various field strengths. Indeed, arecent relaxation time study suggested that T2-weightedimaging might provide better neonatal brain GWMC

than T1-weighted imaging, even at 3 T.8 Furthermore,the differences in signal intensity between the braincenter and periphery evident in Figs 1B and 2C areattributable to radiofrequency field inhomogeneity. Thelatter has both signal-excitation effects (flip-angle devi-ation from nominal values of 10%–15% in this study;signal amplitude is proportional to the cubed sine of theflip angle in the spin echo sequence used) and detectioneffects (spatial variation in MRI coil sensitivity). Thesecould complicate radiologic interpretation of the T1-weighted images, potentially leading to incorrect diag-noses. Although spatial detection effects could be mea-sured and corrected empirically, flip-angle deviationscould be reduced only by using, for example, adiabaticradiofrequency pulses or fundamentally different se-quences. Because our FSE sequence combined spin ech-oes and stimulated echoes, radiofrequency field inhomo-geneity did not produce significant effects in our T2-weighted images.18,19

Fluid-attenuated IR imaging has also been used toimage neonatal brain29,30 and, because of FSE resilienceto radiofrequency field inhomogeneities, we started toevaluate IR-FSE techniques for T1-dependent imaging at4.7 T. Initial, high-resolution images showed promisingGWMC and good cerebrospinal fluid suppression (Fig1C) although, because of the chosen TI, the overall SNRwas lower and the acquisition time longer than for T2-weighted FSE imaging.

We also evaluated 3-dimensional MDEFT imaging, amodified version (combining adiabatic and nonadiabaticpreparation pulses) of which proved robust against ra-diofrequency field inhomogeneities and delivered goodtissue contrast in 4.7-T adult studies.20 The adaptation of3-dimensional MDEFT sequence parameters for neona-tal brain25 yielded GWMC (Fig 1D) sufficient for diag-nostic evaluation. The possibility of reprocessing the3-dimensional data to produce images with any desiredorientation presents obvious diagnostic benefits. Withaccurate evaluation of neonatal T1 values, additionalsequence optimization should be possible, with the po-tential for improved resolution.

Quantitative T2 MapsQuantitative relaxation-time maps may help eliminateradiofrequency field inhomogeneity artifacts and mayprovide more objective, observer-independent interpre-tation than qualitative image examination. For example,in a recent study of infants �5 days of age with NE, basalganglion T2 was prognostic for neurodevelopmentaloutcomes.31 Nevertheless, quantitative T1 and T2 imag-ing is not yet routinely available on commercial MRIsystems. Our T2 maps displayed good GWMC (Fig 1E)and confirmed decreased brain-water T2 values athigher field strength.8,21 For example, at 4.7 T, infant 2(with a suspected hypoxic-ischemic injury but normal1-year outcome) had thalamic and white matter T2 val-

FIGURE 4High-resolution, T2-weighted, FSEMRI (voxels of 0.47mm� 0.47mm� 2mm) for infant7, who had acute bilirubin encephalopathy and a severe outcome. Bilateral and symmet-rical signal abnormalities of the globi pallidi (gp) are demonstrated. The adjacent parts ofthe caudate (c), putamina (p), and capsular white matter also appeared involved, but toa lesser degree.

e1818 DE VITA et al by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from

ues of 72 and 128 milliseconds, respectively, comparedwith 123 and 210 milliseconds at 3 T8 and 141 and 228milliseconds at 2.35 T.32 Additional optimization of thequantitative T2 sequence may provide higher resolutionand/or acquisition of more slices.

Limitations of the Study

SARIn this study, the number of slices per acquisition waslimited by the requirement of keeping the SAR �2W/kg. Therefore, it was not possible to examine thewhole brain with FSE imaging in a single scan lasting �6minutes. Implementation of variable-flip-angle FSEtechniques, such as hyperechoes33 or transition intopseudo-steady state,34 which reduce SAR greatly withminimal SNR loss, should increase the number of slicesobtainable within recommended safety limits.

Radiofrequency Coil DimensionsThe (adult) head coil used for this study had an internaldiameter of 28 cm, whereas the newborn head diameteris only �12 cm. Because SNR increases with the sample/coil volume ratio and smaller coils require less radiofre-quency power, there is the potential to increase signifi-cantly the image quality and the number of slices byusing a smaller neonatal coil.35–37

Pulse Sequence OptimizationAlthough T1-weighted spin echo and T2-weighted FSEtechniques were used in all studies, the other sequences,yielding T1 contrast images and T2 maps, were used less.Although IR-FSE and MDEFT image quality was accept-able and these sequences provided a useful alternative toT1-weighted spin echo imaging, acquisition parametersmust be optimized. For this purpose, T1 and T2 in neo-natal brain must be evaluated more comprehensively at4.7 T. We are investigating T1-mapping methods that areresilient to flip-angle inhomogeneities and show prom-ise for neonatal application, including a modified fastlow-angle shot sequence38 and T1 by multiple readoutpulses.39

Motion SensitivityThe smaller slice thicknesses and higher in-slice resolu-tion used for 4.7-T FSE imaging necessitate perfect headimmobilization, to minimize subject movement. Somemotion artifacts were seen in a few neonatal images (eg,Fig 3, top left, and Fig 4, top right). Respiratory andcardiac gating may thus be beneficial.

Image InterpretationEthical permission was granted only to study ventilatedinfants undergoing intensive care; at this stage, permis-sion for the study of healthy, age-matched, control in-fants was withheld. Lack of normative data is common

for any novel modality. The appearance of pathologicbrain tissue at 4.7 T requires careful appraisal, and thefull clinical implications of the images presented hereremain uncertain. Indeed, the first high-resolution,4.7-T, adult, FSE images caused concern initially becausethe Virchov-Robin spaces appeared excessively promi-nent even in young healthy subjects.18 Therefore, it isvery important to study control infants at 4.7 T. Further-more, for clinical reasons, it was not possible to comparethe 4.7-T images with images acquired for the samesubjects with similar pulse sequences at 1.5 T and thus tocarry out a quantitative field-strength comparison.

We studied 5 infants because of suspected perinatalhypoxic-ischemic injury; 2 had normal outcomes, 1mild/moderate, and 2 severe. Abnormalities were notedin the T1-weighted and T2-weighted images for the in-fants with severe outcomes (compare Figs 1 and 2).Conventional MRI at 1.5 T is currently the investigationmethod of choice but not until the end of the first weekafter birth, at which time a close correlation exists be-tween the pattern of MRI abnormalities and neurode-velopmental outcomes.3,40–44 Infants with acute insultsare likely to sustain basal ganglion and thalamic le-sions,45 and abnormal signal intensity in the posteriorlimb of the internal capsule is used frequently as a pre-dictor of adverse outcomes.43 The increased anatomicresolution and GWMC provided by 4.7-T imaging maylead to more-accurate prognoses, may improve thera-peutic interventions (eg, whole-body cooling), and maydefine brain structure-function relationships for thishigh-risk population.

Two infants (infants 3 and 4) underwent whole-bodycooling at 33.5°C for 72 hours, as part of a clinical trial.Brain-water T1 and T2 values would change duringcerebral hypothermia, resulting transiently in alteredGWMC. However, these infants were normothermic atthe time of MRI. The effect of this hypothermic treat-ment on the MRI pattern of injury after perinatal hy-poxia-ischemia is unclear, although reports suggest re-duced rates of moderate/severe neurologic impairmentand death.46,47 A recent NE MRI study of both whole-body and selective head cooling demonstrated fewerbasal ganglion lesions in infants who had been cooledand who, at the time of random assignment, had mod-erate, as opposed to severe, amplitude-integrated elec-troencephalographic abnormalities.48 Differential corti-cal protection was seen in another whole-body coolingstudy, although electroencephalographic entry criteriawere not used.49

CONCLUSIONSWith attention to potential hazards and patient han-dling, newborn infants undergoing intensive care can bestudied at 4.7 T with excellent results; T2-weighted FSEMRI brain images showed remarkable anatomic detailand tissue contrast. On 4.7-T T1-weighted images,

PEDIATRICS Volume 118, Number 6, December 2006 e1819 by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from

GWMC was worse than that at 1.5 T. Consequently,abnormalities would have to change T1 substantially tobe noticed visually. From our small cohort of infants, wehave not gained adequate experience to be certainwhether the T1 change required for detection is signifi-cantly greater than that at 1.5 T. The MRI examinationtime was not reduced, compared with imaging at 1.5 T,because of sequence adjustments required to ensure thatthe SAR was �2 W/kg. The increased T2-weighted im-age detail, however, may lead to more accurate prog-noses, may improve therapeutic interventions, and mayenable definition of more precise structure-function re-lationships in NE. Neonatal brain image quality at 4.7 Tcould be improved with small neonatal imaging coils andfurther pulse sequence development and optimization.Areas of future interest at high field include the devel-opment of sequences for 3-dimensional imaging (forvoxel or deformation-based morphometric analysis) andfor the measurement of apparent diffusion coefficientsand diffusion anisotropy. Meaningful interpretation ofpathologic 4.7-T images will require the study of normalinfants and detailed long-term neurodevelopmental fol-low-up to define relationships between subtle MRI ab-normalities and specific functional impairments.

ACKNOWLEDGMENTSWe thank the Special Trustees of the Middlesex Hospitalfor funding the Imaging Fellows and the ventilator andthe Wellcome Trust for funding the 4.7-T MRI systemand associated infrastructure.

We thank Dr David Thomas for providing the pro-gram to optimize MDEFT parameters and Sahan Thala-yasingam for helping to modify the neonatal examina-tion pod for 4.7 T.

REFERENCES1. McArdle CB, Richardson CJ, Hayden CK, Nicholas DA, Amparo

EG. Abnormalities of the neonatal brain: MR imaging, part II:hypoxic-ischaemic brain injury. Radiology. 1987;163:395–403

2. Barkovich AJ, Westmark K, Partridge C, Sola A, Ferriero DM.Perinatal asphyxia: MR findings in the first 10 days. AJNR Am JNeuroradiol. 1995;16:427–438

3. Ment LR, Bada HS, Barnes P, et al. Practice parameter: neuro-imaging of the neonate: report of the Quality Standards Sub-committee of the American Academy of Neurology and thePractice Committee of the Child Neurology Society. Neurology.2002;58:1726–1738

4. Hu X, Norris DG. Advances in high-field magnetic resonanceimaging. Annu Rev Biomed Eng. 2004;6:157–184

5. Robitaille PM, Abduljalil AM, Kangarlu A, et al. Human mag-netic resonance imaging at 8 T. NMR Biomed. 1998;11:263–265

6. Adriany G, Yacoub E, Tkac I, et al. 7T vs 4T: preliminary B1,SNR, SAR comparison in the human head. In: Proceedings of the8th Scientific Meeting of the International Society for Magnetic Reso-nance in Medicine, Denver, Colorado. Berkeley, CA: InternationalSociety for Magnetic Resonance in Medicine; 2000:147

7. Rutherford M, Malamateniou C, Zeka J, Counsell S. MR im-aging of the neonatal brain at 3 Tesla. Eur J Paediatr Neurol.2004;8:281–289

8. Williams LA, Gelman N, Picot PA, et al. Neonatal brain: re-gional variability of in vivo MR imaging relaxation rates at 3.0T: initial experience. Radiology. 2005;235:595–603

9. Malamateniou C, Serena JC, Allsop JM, et al. Tortuosity anddiameter differences of neonatal cerebral vasculature betweenterm and preterm infants at term: an MRA study at 3 Tesla. In:Proceedings of the 13th Scientific Meeting of the International Societyfor Magnetic Resonance in Medicine, Miami, Florida. Berkeley, CA:International Society for Magnetic Resonance in Medicine;2005:641

10. Gerig G, Weili L, Vetsa YSK, Gilmore JH. Assessing whitematter growth trajectory of early neonatal development by 3TMR-DTI. In: Proceedings of the 13th Scientific Meeting of the Inter-national Society for Magnetic Resonance in Medicine, Miami, Florida.Berkeley, CA: International Society for Magnetic Resonance inMedicine; 2005:293

11. Foran AM, Fitzpatrick J, Schmitz S, Hajnal J, Edwards AD.Cardiac MRI at 3.0T in preterm infants. Pediatr Res. 2005;58:376

12. American Academy of Pediatrics, Committee on Environmen-tal Health. Noise: a hazard for the fetus and newborn. Pediatrics.1997;100:724–727

13. De Vita E, Bainbridge A, Cheong JLY, et al. Localised 4.7 Teslaproton magnetic resonance spectroscopy in neonatalencephalopathy: implementation, safety, and preliminary in-terpretation of results. Imaging Decisions. 2005;9:31–41

14. Sarnat HB, Sarnat MS. Neonatal encephalopathy followingfetal distress. Arch Neurol. 1976;33:696–705

15. Amiel Tison C, Stewart A. Follow up studies during the firstfive years of life: a pervasive assessment of neurological func-tion. Arch Dis Child. 1989;64:496–502

16. Griffiths R. The Abilities of Babies: A Study in Mental Measurement.Rev ed. Amersham, United Kingdom: Association for Researchin Infant and Child Development; 1986

17. Medical Devices Agency. Guidelines for Magnetic ResonanceEquipment in Clinical Use. London, United Kingdom: MedicalDevices Agency; 2002

18. De Vita E, Thomas DL, Roberts S, et al. High resolution MRI ofthe brain at 4.7 Tesla using fast spin echo imaging. Br J Radiol.2003;76:631–637

19. Thomas DL, De Vita E, Roberts S, Turner R, Yousry TA, Or-didge RJ. High-resolution fast spin echo imaging of the humanbrain at 4.7 T: implementation and sequence characteristics.Magn Reson Med. 2004;51:1254–1264

20. Thomas DL, De Vita E, Deichmann R, Turner R, Ordidge RJ. 3DMDEFT imaging of the human brain at 4.7 T with reducedsensitivity to radiofrequency inhomogeneity. Magn Reson Med.2005;53:1452–1458

21. Jezzard P, Duewell S, Balaban RS. MR relaxation times inhuman brain: measurement at 4 T. Radiology. 1996;199:773–779

22. Ross JS. The high-field-strength curmudgeon. AJNR Am J Neu-roradiol. 2004;25:168–169

23. Frahm J, Haase A, Matthaei D. Rapid three-dimensional NMRimaging using the FLASH technique. J Comput Assist Tomogr.1986;10:363–368

24. Mugler JP III, Brookeman JR. Rapid three-dimensional T1-weighted MR imaging with the MP-RAGE sequence. J MagnReson Imag. 1991;1:561–567

25. Deichmann R, Good CD, Josephs O, Ashburner J, Turner R.Optimization of 3D MP-RAGE sequences for structural brainimaging. NeuroImage. 2000;12:112–127

26. Lee J-H, Garwood M, Menon R, et al. High contrast and fastthree-dimensional magnetic resonance imaging at high fields.Magn Reson Med. 1995;34:308–312

27. Deichmann R, Schwarzbauer C, Turner R. Optimisation of the

e1820 DE VITA et al by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from

3D MDEFT sequence for anatomical brain imaging: technicalimplications at 1.5 and 3 T. NeuroImage. 2004;21:757–767

28. McArdle CB, Richardson CJ, Nicholas DA, Mirfakhraee M,Hayden CK, Amparo EG. Developmental features of the neo-natal brain: MR imaging, part I: gray-white matter differenti-ation and myelination. Radiology. 1987;162:223–229

29. Ashikaga R, Araki Y, Ono Y, Nishimura Y, Ishida O. Appear-ance of normal brain maturation on fluid-attenuated inver-sion-recovery (FLAIR) MR images. AJNR Am J Neuroradiol.1999;20:427–431

30. Iwata O, Iwata S, Tamura M, et al. Periventricular low inten-sities on fluid attenuated inversion recovery imaging in thenewborn infant: relationships to the clinical data and long-term outcome. Pediatr Int. 2004;46:150–157

31. Shanmugalingam S, Thornton JS, Iwata O, et al. Comparativeprognostic utilities of early quantitative magnetic resonanceimaging spin-spin relaxometry and proton magnetic resonancespectroscopy in neonatal encephalopathy. Pediatrics. 2006;118:1467–1477

32. Thornton JS, Amess PN, Penrice J, Chong WK, Wyatt JS,Ordidge RJ. Cerebral tissue water spin-spin relaxation times inhuman neonates at 2.4 Tesla: methodology and the effects ofmaturation. Magn Reson Imaging. 1999;17:1289–1295

33. Hennig J, Scheffler K. Hyperechoes. Magn Reson Med. 2001;46:6–12

34. Hennig J, Weigel M, Scheffler K. Multiecho sequences withvariable refocusing flip angles: optimization of signal behaviorusing smooth transitions between pseudo steady states(TRAPS). Magn Reson Med. 2003;49:527–535

35. Erberich SG, Friedlich P, Seri I, Nelson MD Jr, Bluml S. Func-tional MRI in neonates using neonatal head coil and MRcompatible incubator. NeuroImage. 2003;20:683–692

36. Whitby EH, Griffiths PD, Lonneker-Lammers T, et al. Ultrafastmagnetic resonance imaging of the neonate in a magneticresonance-compatible incubator with a built-in coil. Pediatrics.2004;113(2). Available at: www.pediatrics.org/cgi/content/full/113/2/e150

37. Bluml S, Friedlich P, Erberich S, Wood JC, Seri I, Nelson MD Jr.MR imaging of newborns by using an MR compatible incubatorwith integrated radiofrequency coils. Radiology. 2004;231:594–601

38. Deichmann R. Fast high-resolution T1 mapping of the humanbrain. Magn Reson Med. 2005;54:20–27

39. Gelman N, Ewing JR, Gorell JM, Spickler EM, Solomon EG.Interregional variation of longitudinal relaxation rates in hu-man brain at 3.0 T: relation to estimated iron and water con-tents. Magn Reson Med. 2001;45:71–79

40. Robertson NJ, Wyatt JS. The magnetic resonance revolution inbrain imaging: impact on neonatal intensive care. Arch DisChild. 2004;89:F193–F197

41. Barkovich AJ. MR of the normal neonatal brain: assessment ofdeep structures. AJNR Am J Neuroradiol. 1998;19:1397–1403

42. Rutherford M, Pennock J, Schwieso J, Cowan F, Dubowitz L.Hypoxic-ischaemic encephalopathy: early and late magneticresonance imaging findings in relation to outcome. Arch DisChild Fetal Neonatal Ed. 1996;75:F145–F151

43. Rutherford MA, Pennock JM, Counsell SJ, et al. Abnormalmagnetic resonance signal in the internal capsule predicts poorneurodevelopmental outcome in infants with hypoxic-ischemic encephalopathy. Pediatrics. 1998;102:323–328

44. Rutherford M, Ward P, Allsop J, Malamatentiou C, Counsell S.Magnetic resonance imaging in neonatal encephalopathy.Early Hum Dev. 2005;81:13–25

45. Cowan FM. Outcome after intrapartum asphyxia in term in-fants. Semin Neonatol. 2000;5:127–140

46. Shankaran S, Laptook AR, Ehrenkranz RA, et al. Whole-bodyhypothermia for neonates with hypoxic-ischemic encephalop-athy. N Engl J Med. 2005;353:1574–1584

47. Gluckman PD, Wyatt JS, Azzopardi D, et al. Selective headcooling with mild systemic hypothermia after neonatalencephalopathy: multicentre randomised trial. Lancet. 2005;365:663–670

48. Rutherford MA, Azzopardi D, Whitelaw A, et al. Mild hypo-thermia and the distribution of cerebral lesions in neonateswith hypoxic-ischemic encephalopathy. Pediatrics. 2005;116:1001–1006

49. Inder TE, Hunt RW, Morley CJ, et al. Randomized trial ofsystemic hypothermia selectively protects the cortex on MRI interm hypoxic-ischemic encephalopathy. J Pediatr. 2004;145:835–837

PEDIATRICS Volume 118, Number 6, December 2006 e1821 by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from

DOI: 10.1542/peds.2006-1499; originally published online November 13, 2006;Pediatrics

J. RobertsonNicolaLombard, Wui K. Chong, John S. Wyatt, Ernest B. Cady, Roger J. Ordidge and

Enrico De Vita, Alan Bainbridge, Jeanie L. Y. Cheong, Cornelia Hagmann, RosarieExperiences

Magnetic Resonance Imaging of Neonatal Encephalopathy at 4.7 Tesla: Initial  

ServicesUpdated Information &

/peds.2006-1499.citationhttp://pediatrics.aappublications.org/content/early/2006/11/13including high resolution figures, can be found at:

References

/peds.2006-1499.citation#ref-list-1http://pediatrics.aappublications.org/content/early/2006/11/13at:This article cites 43 articles, 17 of which can be accessed free

Permissions & Licensing

mlhttp://pediatrics.aappublications.org/site/misc/Permissions.xhttables) or in its entirety can be found online at: Information about reproducing this article in parts (figures,

Reprints http://pediatrics.aappublications.org/site/misc/reprints.xhtml

Information about ordering reprints can be found online:

rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.Grove Village, Illinois, 60007. Copyright © 2006 by the American Academy of Pediatrics. All and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elkpublication, it has been published continuously since 1948. PEDIATRICS is owned, published, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

by guest on June 12, 2013pediatrics.aappublications.orgDownloaded from