Fetal Magnetic Resonance Imaging at 3.0 T - pedrad.org MRI/Fetal MRI 3T 2011.pdf · Fetal Magnetic...

Fetal Magnetic Resonance Imaging at 3.0 TRobert C. Welsh, PhD,* Ursula Nemec,Þ and Moriah E. Thomason, PhDþ§

Abstract: Magnetic resonance imaging (MRI) has been used to imagethe in utero fetus for the past 3 decades. Although not as commonplaceas other patient-oriented MRI, it is a growing field and demonstratinga role in the clinical care of the fetus. Indeed, the body of literature in-volving fetal MRI exceeds 3000 published articles. Indeed, there is in-terest in accessing even the healthy fetus with MRI to further understandthe development of humans during the fetal stage. On the horizon is fetalimaging using 3.0-T clinical systems. Although a clear path is not neces-sarily determined, experiments, theoretical calculations, advances in pulsesequence design, new hardware, and experience from imaging at 1.5 Thelp define the path.

Key Words: Fetal, MRI, 3.0 T

(Top Magn Reson Imaging 2011;22: 119Y131)

Imaging of the human body has undergone incredible ad-vances in the near century since World War I. Heavily influ-

enced by Nobel Laureate Maria Curie, radiological science andits application in medicine, in particular, were elevated from‘‘notoriously inadequate’’ to a point where ‘‘no surgeon wouldthink of removing a projectile without the precise knowledgeof its location’’ by the end of the war.1 Essentially, advances inimaging have transformed medical sciences such that we canpeer into the body and reconcile with great detail tissues, organs,chemical constitution, flow, and action. In confirmation of thesignificance of these achievements, Paul Lauterbur and PeterMansfield received the Nobel Prize in Medicine in 2003 for theirdiscoveries concerning magnetic resonance imaging (MRI). Mag-netic resonance imaging is among the most notable of the ad-vanced imaging sciences, with privileged position as being lowrisk and being a flexible enough modality to acquire a widevariety of information within 1 medical imaging system.

It has been close to 30 years since one of the very firstMRI (nuclear magnetic resonance imaging) images of preg-nancy was acquired at 0.04 T.2 Since that time, fetal imaging hasexperienced slow, but steady, growth, predominantly motivatedby apparent clinical need. Clinical MRI is most frequently per-formed on 1.5-T MRI systems, but since 2000, there has been amarked increase in clinical imaging performed at higher field,3.0 T. The shift from 1.5-T to 3.0-T medical imaging in the fetushas developed more gradually, and therefore, data from fetalimaging at 3.0 T is becoming available just now. Certainly, med-ical imaging of the developing fetus presents a critical oppor-tunity for both medical and research sciences, and there can beadvantages in moving to a higher field strength. However, de-velopment of this opportunity must be undertaken with care, and

unfortunately, this effort is heavily laden with extreme techni-cal challenges, some of which will be exaggerated in the shiftto a higher MR field strength.

Imaging of the human body from postnatal to adult has itsown list of constraints, what obviously complicates matters forfetal imaging is that it must be done in utero. Thus, imaging isseverely limited by the complications of doing abdominal im-aging and dealing with structures that can undergo motion thatcannot be controlled without the administration of sedatives.3

In addition, there is the critical need to guarantee the safety ofthe fetus and mother by preventing absorption of too much ra-diofrequency (RF) energy. That is, when one images the fetus,special consideration must be paid to the issue of specific ab-sorption rate (SAR). Yet, even within the difficult arena set bysuch constraints, there have been recent advances and demon-strations of proof of principle of fetal imaging at 3.0 T.4

Others4Y13 have written on the topic of fetal imaging in gene-ral, and thus, to be parsimonious, this article will concentrate onrecent developments in a handful of areas of MR research andthe potential for application for fetal imaging.

HISTORICAL PERSPECTIVETo put the eminent emergence of fetal imaging at 3.0 T into

perspective, we can examine the recent history of medical andscientific literature with respect to MRI in humans at relativefield strengths. A literature searchwas performed in pubmed.org,searching for terms related to ‘‘magnetic resonance imaging’’with respect to humans (ie, not animals) in 2 leading radiologyjournals (Radiology and American Journal of Neuroradiology)as well as the 2 leading journals reporting neuroimaging-basedneuroscience (NeuroImage and Human Brain Mapping). Thisresults in a total of 16,694 abstracts. In addition, only thoseabstracts that reported a field strength of either 1.5 T or 3.0 Twere included. The search resulted in a total of 1476 publica-tion records (È91% of published articles do not report fieldstrength in abstracts).

The total growth of MRI-related publications regardlessof journal of origin has grown steadily at approximately 1%per year from 1985 until about 2000 when more 3.0-T systemsstarted to turn on. In addition, with the advent of activationstudies based on blood oxygen level dependence (BOLD), therehas been a dramatic increase in publications (Fig. 1). Further ex-amining this growth, we see that although neuroscience studiesare mainly based on 3.0-T systems, both neuroscience and med-ical publications have seen a growth of approximately 9% to10% per year since 2000 (Fig. 2).

Fetal imaging has been occurring since the early days ofMRI.2 However, it appears that this has happened more con-servatively in the United States/North America, with a steadylinear trend of approximately 3.5% per year. On the other hand,interest in fetal imaging is growing quadratically in Europe/elsewhere (Fig. 3.), with a little more than 200 publications onfetal imaging in 2011 alone. In 2005, General Electric announcedthe installation of its 250th Signa 3.0T system. The installa-tion rate of 3.0-T systems has grown such that Siemens in-stalled roughly 300 3.0-T systems (Dr Keith Heberlein, SiemensMedical Solutions, personal communication) in 2011.

ORIGINAL ARTICLE

Top Magn Reson Imaging & Volume 22, Number 3, June 2011 www.topicsinmri.com 119

From the *Departments of Radiology and Psychiatry, University of Michigan,Ann Arbor, MI; †Department of Radiology, Division of Neuroradiologyand Musculoskeletal Radiology, Medical University of Vienna, Vienna,Austria; and ‡Department of Pediatrics,Wayne State University School ofMedicine, and §Merrill Palmer Skillman Institute for Child and FamilyDevelopment, Wayne State University, Detroit, MI.

The authors declare no conflict of interest.Reprints: Robert C. Welsh, PhD, Medical Science I, Room 3208C,

Department of Radiology, 1301 Catherine St, Ann Arbor, MI 48109-5667(e-mail: [email protected]).

Copyright * 2012 by Lippincott Williams & Wilkins

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

The foregone conclusion of these observations regardingfetal publication rates and expanding MR system installationbases is that fetal imaging at 3.0 Twill undergo major growth inthe next 1 to 5 years. However, this can only be determined bythe scientific interest and medical indication, medical necessity,and medical benefit. Benefits of imaging at 3.0 T comparedwith 1.5 T have to outweigh the potential risks.

PHYSICS BASICSTwo intrinsic physical parameters are directly influenced

by moving to a higher field, namely, the signal-to-noise ratio(SNR) and the longitudinal relaxation time of protons (ie, watermolecules).

By doubling the strength of the main field (B0), an increasein SNR is realized. This increased SNR is rooted in having twicethe available polarization of the water molecule protons. Thepolarization is a function of the Boltzmann factor (k is theBoltzmann constant), that is,

PðB0Þ ¼ ej$E=kT

which, at room temperature (T), and relatively low magneticfield strength such that the exponential factor is small, the for-mula can be expanded to the first 2 terms:

PðB0Þ;1j$E=kT

where the energy difference of the magnetic states is directlyproportional to B0, that is, $E B0, and thus, the polarization at3.0 T is twice that at 1.5 T:

Pð3:0TÞ,2qPð1:5TÞ

As field strength increases, the tissue longitudinal relaxa-tion time T1 varies. It has been observed at low magnetic fieldsthat T1 for both gray matter (GM) and white matter (WM) is anexponential function varying with magnetic field B0, namely,T1ðB0Þ ¼ AqBB

0 , where A and B are derived from data. How-ever, this functional form was derived for main field strengthsless than 2.5 T, with most of the measurements at quite lowfields.14

FIGURE 1. Relative publication rate of all nonYanimal-relatedMRI literature from 1985 to 2011, unit-normalized by publicationsin 2011. In approximately 2007Y2008, publications based on3.0 T systems started to dominate the literature.

FIGURE 2. Relative publication rates between 1.5 T and 3.0 T systems for neuroscience (left)- and medical-oriented (right) journals from1985 to 2011. Note that before the discovery of BOLD, there were no MRI-based neuroscience articles in the pubmed.org search results.

FIGURE 3. Fetal (foetus, fetal, fetus) MRI publications in theyear range of 1985Y2011. The United States/North Americapublication rate increases linearly, whereas that of Europe/Othersuggests a quadratic increase.

Welsh et al Top Magn Reson Imaging & Volume 22, Number 3, June 2011

120 www.topicsinmri.com * 2012 Lippincott Williams & Wilkins


By taking recent T1 measurements from various laborato-ries around the world, a linear fit can be achieved for the rangeof 1.5 T to 8.0 T. For the adult brain, we can expect an ap-proximate 18% lengthening of GM T1 and an approximate 31%lengthening of WM T1. Furthermore, relaxation rates vary inthe fetal brain compared with the adult brain because of a highwater content and a lower cell density.15 Almajeed et al16 esti-mated that at 1.5 T, WM longitudinal relaxation time T1 is ap-proximately 1600 milliseconds. Although not explicitly measuredin utero, Williams et al17 made relaxation rate measurements ofin vivo brain tissue from a cohort of preterm infants (a surro-gate for fetal brain) at 3.0 T. The range of in vivo neonate mea-surements by Williams et al17 at 3.0 T thus appears consistentwith the smaller value of in utero T1 estimated by Almajeedet al16 at 1.5 T, assuming a similar lengthening trend as seen inadults (Fig. 4). In principle, T1 changes will directly manifestin changing contrast between tissue types when going from 1.5 Tto 3.0 T, unless of course a change in repetition time (TR) isimplemented. However, with the implied overlap of TGM1 andTWM1 for in utero brain tissue, a loss of GM/WM contrast canbe expected in the fetal brain. Indeed, most investigators willfind themselves examining cortical structures with T2-weightedimaging given that T2 is relatively field strength independent.17

WHY 3.0 T?The heading of this section is an obvious question with a

not-so-simple answer. Indeed, human imaging (and spectros-copy) has been advanced with the emergence of 3.0-T scanners.Around the world, 3.0-T scanners are becoming less of the ex-ception in clinical environments (see Fig. 2 on increased 3.0-Tpublications). However, a benefit-risk assessment has to be donefor fetal imaging. In terms of advantage, there is the marked in-crease in SNR, which gives rise to a myriad of benefits, namely,the possibility for faster scanning, higher spatial resolution, im-proved BOLD signal, but also accompanying a move from 1.5 Tto 3.0 T are technical challenges, namely, susceptibility artifacts,changes in longitudinal relaxations times, RF (also referred toas B1 at times, which is the strength of the field produced bythe transmitted RF energy for manipulation of the water pro-ton spins, it produces the tilting of the spins) penetration issues,and RF deposition. As with any imaging protocol determination,a balance has to be struck based on the constraints determinedby the physics and medical/scientific need of the situation.

Where we can see benefit at 3.0 T is the increased SNR foruse in MR spectroscopy (MRS), the improvement in contrast-to-noise ratio (CNR) in blood oxygen level dependence for bothactivation and resting state scans and the overall improvement inSNR for higher resolution anatomic imaging (including diffu-sion tensor imaging [DTI]). Just as at 1.5 T, fetal movement isa major consideration and incremental increase in SNR can alsobe used for faster scan sequences (within the limits of SAR). Inaddition, as outlined below, for BOLD imaging, in addition to theincreased SNR, there is the added benefit of maximal BOLDsensitivity at a shorter echo time expected caused by the short-ening of the relaxation time T*

2 .All in all, the technical challenges for fetal imaging at 3.0 T

are not unlike the challenges when other human imaging movedfrom 1.5 T to 3.0 T. The problems presented in that move weredeconstructed and solved.18Y32 This is anticipated to similarlyoccur with fetal imaging. Indeed, as briefly described below, ad-vances have already been made that will be directly applicableto fetal imaging at 3.0 T.

FETAL SAFETY AND SARAt the heart of fetal MRI is careful consideration of safety

of the methodology to the developing fetus. This key issue islikely to dictate how MR is used to expand scientific and med-ical understanding of the fetal period. Because of the centralityof this topic to the current discussion, it is important to articu-late exactly what is meant when we talk about ‘‘safety.’’

When we consider risks, what kinds of impacts are we con-cerned with specifically?

How well do we understand the risks, and can the risk beaccurately quantified?

When we talk about safety in fetal MRI, or any MRI forthat matter, the main concern is energy deposition and the as-sociated rise in temperature. Much literature has been publishedon the topic of energy deposition in the pregnant woman andfetus during MRI. The SAR is a measure of the heating causedby RF energy deposition in the body. For typical imaging

FIGURE 4. T1 relaxation in human (adult) brain at varyingmagnetic field strengths B0.

TABLE 1. Calculated SAR Values for Body, Local 1 g andLocal 10 g

Regulatory Limit SARBody 1.5 T SARBody 3.0 T

FDA (SAR1 g e8 W/kg) 0.68 W/kg 0.73 W/kgICNIRP (SAR10 g e10 W/kg) 1.18 W/kg 1.23 W/kgNRPB (fetal SAR10 g e4 W/kg) 0.87 W/kg 0.76 W/kg

Values extracted from Table 2 of Hand et al.38 NRPB indicates Na-tional Radiological Protection Board, now known as the Health Protec-tion Agency (UK).

TABLE 2. Signal to Noise Measured at Different Field Strengths

Sequence

1.5 T 3.0 T

SNR SAR SNR SAR

DTI/DWI 3.91 0.26 8.45 0.27SWI 5.72 0.01 13.33 0.06T2 5.32 1.14 6.16 0.53EPI 14.74 0.02 11.64 0.21

EPI indicates echo planar imaging; SWI, susceptibility weightedimaging.

Top Magn Reson Imaging & Volume 22, Number 3, June 2011 Fetal Magnetic Resonance Imaging at 3.0 T

* 2012 Lippincott Williams & Wilkins www.topicsinmri.com 121


sequences on commercially available scanners, SAR will havethe following relationship33:

SAR ¼ B205

2$f

where B0 is the main magnetic field (1.5 Tor 3.0 T),5 is the flipangle, and $N is the RF bandwidth.

In theUnited States, the Food andDrugAdministration (FDA)has defined exposure limits and states: (http://www.fda.gov/MedicalDevices/DeviceRegulationandGuidance/GuidanceDocuments/ucm107721.htm):

‘‘Current FDA guidance limits SARwhole-body exposure to4.0W/kg for patients with normal thermoregulatory function and1.5 W/kg for all patients, regardless of their condition.’’

The official position of the American College of Ra-diology34 states:

‘‘Present data have not conclusively documented any del-eterious effects of MRI exposure on the developing fetus.Therefore, no special consideration is recommended for the firstversus any other trimester in pregnancy.’’

And goes on to further declare:‘‘Pregnant patients can be accepted to undergo MR scans at any

stage of pregnancy if, in the determination of a level 2MR personnel-designated attending radiologist, the risk-benefit ratio to the patient warrants that the study beperformed. The radiologist should confer with the refer-ring physician and document the following in the radi-ology report or the patient’s medical record:’’

The information requested from the MR study cannot beacquired via nonionizing means (eg, ultrasonography).The data are needed to potentially affect the care of thepatient or fetus during the pregnancy.The referring physician does not feel that it is prudent towait until the patient is no longer pregnant to obtain thesedata.

The Fetal Magnetic Resonance Study Group of the Ital-ian Society of Medical Radiology recommends35

FIGURE 5. Acquired MRS at 3.0 T in a male fetus, 31-week and 0/7 days gestational age. Total acquisition time 2 minutes 8 seconds.Data from Wayne State University.

FIGURE 6. Motion parameters for MRS in adult without (left) and with (right) motion correction. Translations in X, Y, Z are depictedby solid, dashed, and dot-dashed lines, respectively. Rotational angles in pitch, roll, and yaw are depicted by solid, dashed, and dot-dashedlines, respectively. Data courtesy of Dr Brian Keating and Dr Thomas Ernst, University of Hawaii.




http://www.fda.gov/MedicalDevices/DeviceRegulationandGuidance/GuidanceDocuments/ucm107721.htm



‘‘The use of magnetic fields with intensity higher than 1.5 Tis currently not considered a routine approach, as this still needsto be validated by specific research protocols.’’

With respect to pregnancy, the British National HealthService (http://www.nhs.uk/Conditions/MRI-scan/Pages/Who-can-use-it.aspx) states:

‘‘Although there is no evidence that MRI scans pose a riskduring pregnancy, as a precaution, scanning is not usuallyrecommended during the first 3 months of pregnancy.’’

What is obviously paramount is that assessment of heatgenerated in the pregnant woman and/or the fetus induced byRF energy deposition needs to be made to ensure the safety ofthe woman and the fetus. Although is it clear that MRI and MRSdo not use ionizing radiation, it is possible to deposit enoughRF energy that results in local tissue heating. As previously dis-cussed, the energy for given MR pulse sequences can be esti-mated by the calculation of the SAR.

Direct measures of localized heating are difficult, and MRthermography is only accurate to T2.0-C.36 An experiment de-signed to measure localized heating and temperature changes was

carried out in 2001 using a piglet model36 by scanning pregnantpigs. The imaging protocol was carried out using a half-Fourier-acquisition-single-shot-turbo-spin-echo (HASTE). Fiber-optictemperature probes were positioned into the pig fetal brain, ab-domen, and amniotic fluid. They performed their regular humanfetal imaging protocol, but at a rate 3 to 5 times faster (15minutesfor human fetus, whereas 3Y5 minutes for pig fetus). Using thedirect measures of localized heating, temperature measurementsduring scanning did not fluctuate beyond that measured beforeor after scanning. The conclusion of this direct measure exper-iment was that, at 1.5 T, the HASTE protocol used did not de-posit a significant amount of energy in the pregnant pig as tocause any detectable temperature rise, and thus, scanning withinFDA guidelines is sufficient protection.

Perhaps more germane to conclusions we can draw abouthuman fetal MR, by implementing a realistic physical model ofa pregnant woman, Hand et al37 calculated spatial distributionsof SAR. The approach used was to examine a realistic model ofa body coil and under the conditions of a continuous transverseB1 field of 1 KT, then estimate SAR with a finite element model.These calculations were used to predict SAR averaged over

FIGURE 7. Time averaged spectra from adult brain as participant moved. Left: without prospective motion correction. Right: With motioncorrection for MRS voxel position. Data courtesy of Dr Brian Keating and Dr Thomas Ernst, University of Hawaii.

FIGURE 8. Prescribed motion for the 2 MRS dynamic shim experiments. In both MRS spectra acquisitions, the participant movedthe head in a prescribed manner. Prospective motion correction was implemented in both acquisitions for MRS voxel position. Left:dynamic B0 shim not implemented. Right: dynamic B0 shim implemented. Data courtesy of Dr Brian Keating and Dr Thomas Ernst.




http://www.nhs.uk/Conditions/MRI-scan/Pages/Who-can-use-it.aspx

http://www.nhs.uk/Conditions/MRI-scan/Pages/Who-can-use-it.aspx

the whole body (SARBody), as well as 2 local measures of SARin 1 g of tissue and 10 g of tissue (SAR1 g and SAR10 g) in bothmother and fetus at both 1.5 T (64 MHz) and 3.0 T (127 MHz).In addition, the calculated values of SAR1 g and SAR10 g canthen be scaled with SARBody to the limits of SARBody set by theFDA38 and the International Committee on Non-Ionizing Ra-diation Protection (ICNIRP).39 When scaled to the whole-bodysafety limits of 4 W/kg and 2 W/kg, respectively, for FDA andICNIRP limits, the SAR1 g and SAR10 g exceed safety limits.Inverting the problem and limiting the local SAR limits withinsafety guidelines, that is, SAR1 g and SAR10 g, to determinemaximum SARBody, Hand et al41 conclude the following limits:

Others have also made recommendations that add to theabove, Gowland and De Wilde40 specify that women with ther-moregulation issues should only be scannedwith care and, unlessclinically urgent, women who are febrile should not be scanned.

Hand et al41 further refined their calculation in 2010 byalso incorporating a thermal model of heat conduction. Theirfindings were that, under the condition of SARBody less than orequal to 2W/kg, fetalwhole-body SARwas limited to 1.24W/kgand 1.14 W/kg at 1.5 T and 3.0 T, respectively, and that aver-age fetal temperature remained below 38-C, although it wasrecommended to not have continuous exposure to RF for morethan 7.5 minutes. Also in 2010, Kikuchi et al42 used a modelof RF illumination and a thermal model of both a pregnantwoman and a woman that was not pregnant. Although only cal-culated at a lower main field of B0 = 1.5 T (64 MHz), theyconcluded that the safety of the fetus was overestimated andrecommend a exposure time limit of 40minutes when SARBody =2.0 W/kg and 10 minutes at SARBody = 4.0 W/kg. Also, theydo recommend a more complete modeling of the thermal prop-erties of amniotic fluid.

What has yet to be established though is the heat loadingof the pregnant woman and fetus under the conditions of amultitransmit system. Although limited to anatomic models thatdid not include a pregnant woman, it was determined that inworst-case scenarios with multitransmit systems, there can bewidely spatially variant local SAR at 3.0 T, which can also de-pend on positioning of the patient. Caution should be had withimaging pregnant women with such systems.43 In addition, theyrecommended that any guidelines developed include attentionto local SAR calculation and not only SARBody. Indeed, recently,the feasibility of individualized SAR models30 at 3.0 T wasreported.

Considering the relationship betweenSARand field strength,fetal imaging protocols using 3.0 T necessarilymake adjustmentsto acquisition parameters that will lessen SAR, but these some-times come at the cost of reduced SNR. A recent MR protocolenacted at Wayne State University, in partnership with the Peri-natology Research Branch of National Institute of Child Healthand Human Development/National Institutes of Health/Departmentof Health and Human Services, has examined these trade-offs insearch of the optimal balance. A summary of SNR and SAR at1.5 T and 3.0 T measured in 8 women scanned at either field ispresented in Table 1. This preliminary report suggests that, forfetal imaging, one can move from imaging at 1.5 to 3 T and keepSAR levels rather steady while gaining in SNR for most measures.

MAGNETIC RESONANCE SPECTROSCOPYMagnetic resonance spectroscopy can directly benefit from

an increased signal caused by a high polarization at 3.0 T, as wellas a higher spectral resolution. However, an obvious issue is fetalmotion.4 Indirect sedation of the fetus is a solution,3 althoughnot necessarily desirable. Typical MRS acquisitions can last on

FIGURE 9. Time-averaged spectra from adult brain as participant moved. Voxel position updated prospectively. Left: without dynamicB0 shim. Right: with dynamic B0 shim. Data courtesy of Dr Brian Keating and Dr Thomas Ernst, University of Hawaii.

FIGURE 10. Adult T*2 data taken from Kruger et al64 (diamond

and triangle). T*2 ranges (in gray) taken from Rivkin et al.67

Data in red calculated by shifting ranges in gray by measureddifference from 1.5 T to 3.0 T.




the order of minutes, and if there is large fetal motion, obtainedspectra can become useless,44 and even with smaller amplitudemovements, spectra can be broadened and metabolite resolutionmay be lost. Many others have reviewed the use of acquiringMRS in the fetus since first investigating lung maturity45 to morerecent assessments of restricted fetal brain growth by the detectionof lactate.46,47 Thus, this article will focus on recent developmentsto improve the quality of MRS data and the feasibility at 3.0 T.

Here we provide preliminary data from a representativeMRS scan in a 31-week gestational age fetus. Acquired atWayneState University, the spectrum shown in Figure 5, along with adepiction of the MRS voxel of interest, reveals characteristicpeaks but alsovariation that is somewhat challenging to interpret.Compared with typical adult spectra, the lines have the appear-ance of broadening. This can be caused by an unoptimized shim,size of MRS voxel, or motion. As shown in the section on BOLDbelow in Figure 11, fetal motion during a 2-minute run can bequite extensive, resulting in the tissue of interest moving out-side of the MRS voxel, resulting in a degradation of the MRSshim, or contamination, such as by lipids.

Ideally, the solution would be to be able to track the tissueof interest and either reject shot-to-shot spectra before averag-ing or do prospective motion correction on the MRS voxel ofinterest. During the past few years, there has indeed been interestand developments of such techniques. Earlier methods exam-ined motion to discard corrupted acquisitions caused by mo-tion.48 However, such a technique does not mitigate the necessityto repeat the MRS exam, or at least take sufficient spectra as tohave a satisfactory SNR after editing. Keating et al49 demon-strated in 2010 that prospective motion correct can be used by

the use of navigators to update the location of the MRS voxel,that is, to have it move in such a manner as to track the tissueof interest. In Figure 6, we show data from their laboratory (DrThomas Ernst and Dr Brian Keating, The University of HawaiiNeuroscience and MRI Research Program) for a typical exper-iment in which a participant has been trained to execute a pre-scribed motion of his or her head in the scanner while spectraare being acquired. In comparisonwith themovement plot shownin Figure 11 in the next section covering BOLD, the extent ofmotion is comparable between the test (adult) participant andthe fetal movement. In Figure 7, the spectra without motion cor-rection and with motion correction from the prescribed movementexperiment from Keating are shown. The data were analyzedin LCModel 6.1 (http://s-provencher.com/pages/lcmodel.shtml),and summarymetrics report a full-width-half-maximum (FWHM)line width of 0.033 ppm for the uncorrected data and 0.029 ppmfor the corrected data and an improvement of SNR from 25 to 27.

An obvious caveat of course is that further line narrow-ing could be achieved with an update of the B0 shim. In 2011,Keating and Ernst50 implemented a method to update the B0

shim caused by motion during an MRS acquisition. Executinga similar experiment to demonstrate the improved MRS linewidth, adult participants executed a prescribed head movementwhile MRS images were acquired without and with dynamicshim correction. To see the factored improvement caused by onlythe shim correction, motion correction for voxel position up-dating was done during both conditions. Here again data fromtheir laboratory are shown in Figures 8 (movement parameters)and 9 (spectra without and with dynamic B0 shim updating). Ascan be clearly seen, the adult participant executed a drastic pitch

FIGURE 11. Rigid body-derived movement parameters of the 2 different fetuses at 3.0 T. Changes in X, Y, Z are depicted by solid,dashed, and dot-dashed lines, respectively. Rotational angles in pitch, roll, and yaw are depicted by solid, dashed, and dot-dashed lines,respectively. Data from Wayne State University.

FIGURE 12. Six sequential axial frames from fetal echo planar imaging (EPI) series with repetition time TR = 2.0 seconds. Red line tohighlight hemispheric fissure and extent of rapid rotation. Data from Wayne State University.




http://s-provencher.com/pages/lcmodel.shtml

of the head as well as a marked translation. These movementvelocities and amplitudes are typically seen in the fetus (againrefer to Figure 11 in the BOLD section). With both prospectivemotion correction for updating the location of the MRS voxeland with dynamic B0 shim, quality spectra can be obtainedwithout the need to remove bad shot-to-shot spectra. When ex-amining the effect of dynamic shim alone, in the above instanceof the method, the full-width-half-maximum improved from0.076 ppm to 0.048 ppm and the SNR improved from 22 to 25.

With methods such as these and those being developed atother laboratories around the world49Y53 to mitigate tissue mo-tion and changes in B0 shim caused by motion, MRS shouldbecome less of a burden and more reliable in the fetus and thusbecause of the increased SNR from the high polarization be-comes more beneficial at 3.0 T compared with 1.5 T.

BLOOD OXYGEN LEVEL DEPENDENCEWithin a decade of the discovery of BOLD54 in 1990

and the first demonstration of detecting brain activity withBOLD,55 the first fetal functional MRI (fMRI) experiment wasconducted.56 Ostensibly, the auditory cortex was activated byplaying nursery rhymes. This very first fetal fMRI was performedat a magnetic field strength of 0.5 T. A direct benefit of un-dertaking activation fMRI and/or resting-state functional con-nectivity MRI57Y60 at 3.0 T compared with 1.5 T is the improvedSNR. However, what needs to be mitigated of course is thehigher level of susceptibility artifact. All materials and mediums(ie, tissue) have an intrinsic property known as magnetic sus-ceptibility W or the tendency of the medium to be magnetized bythe application of an external magnetic field.61 The suscepti-bility of soft tissue (cortex) is on order of WTissue = j9.05 ppm,whereas cortical bone has WBone = j8.86 ppm.62 At boundariesbetween tissue and bone, there will be a gradient between theeffective fields, resulting in a gradient of the local resonancefrequency of the water protons. Although the susceptibilities arequite close, the difference yet can yield drastic image artifacts,especially when echo time (TE) is increased and being moremanifest at 3.0 T compared with 1.5 T. Most susceptibility arti-facts in BOLD images are caused by an air/tissue and/or an air/bone boundary (WAir = 0.36 ppm), which of course is not pres-ent in the fetus.

Kruger et al63 has shown that an expected increase inBOLD CNR by a factor of 2.0 to 3.4 can be expected when im-

aging adult brains between 1.5 T and 3.0 T. An unknown factoris the value of T*

2 and $R*2 in the fetus. However, in an

experiment examining hypoxia in fetal sheep at 1.5 T and 3.0T,Wedegartner et al64 demonstrated twice the sensitivity of $R*

2for varying oxygen saturation at 3.0 T compared with 1.5 T.Typical echo times for adult brain at 3.0 T is 30 millisec-onds,63 and the CNR is an exponential function of the relaxa-

tion rate R*2 qðR*

2 j1=T*2 Þ: CNRejTEqR

*2 . It has been shown

that besides R*2 potentially varying by cortical area, R*

2 alsoroughly increases linearly by up to approximately 12% fromage 8 years until middle age (È50 y).65 Such a decrease in R*

2in fetus compared with adult brain would then imply that a longerecho time is needed to maximize CNR in the fetus. Rivkin et al67

measured T*2 in preterm newborns at 1.5 T. We can use these

data in conjunction with the measured shortening of T*2 as ob-

served by Kruger et al63 to arrive at an estimated range of 115milliseconds to 135 milliseconds at 3.0 T for fetal brain T*

2 inthe middle third trimester (Fig. 10). However, this should beempirically verified.

Finally, it is typical to choose a magnetization flip angleto optimize the signal intensity (Ernst angle),61 however, it wasrecently shown by Gonzalez-Castillo et al67 that in a real-worldsituation where physiological noise dominates system/thermalnoise and SNR is high, a much smaller flip angle can be chosenwhile still realizing a high BOLDCNR. Indeed, they showed thatalthough the Ernst condition may result in a calculated flip angleof 77 degrees, their real-world calculation leads to a flip anglejust shy of 8 degrees. Given that RF energy absorbed per unittime is given by

PðB1ÞòB21

where B1 is the applied RF field to generate the specifiedflip angle and that B1ò$5, with $5 being the flip angle, then ifthe time that the RF field that is applied is held constant, thepower at flip angle $51 can be related to that at $52, with therelationship

Pð52Þ ¼5

22

521

!qPð51Þ

Thus, by theoretically going from a flip angle of 77degrees to 8 degrees, the average RF power caused by slice

FIGURE 13. Approximate angle of rotation frame-to-frame duringEPI sequence of Figure X. Data from Wayne State University.

FIGURE 14. Determination of b factor to be used in fetal DWI/DTIto achieve same signal suppression as observed in adults.




excitation should drop by a factor of at least 90, and thus a re-duced SAR can be realized.

Where this can have a direct application with time-seriesdata is shortening of the TR to get a better assessment of fetalmovement. Shown in Figure 11 is the fetal head movement from2 different gradient-echo echo-planar-image runs with a TR =2000 milliseconds at 3.0 T from M. Thomason at Wayne StateUniversity. The fetal motion in the left plot shows a quite rea-sonable and acceptable fetal movement requiring little correction,whereas that on the right depicts a more active fetus, executinga large roll and yaw of its head about two thirds of the waythrough the time-series data. Figures 8 and 12 further demon-strate the severity of fetal movement. In just viewing a few axialslices of the fetal brain (Fig. 12 and plotted in Fig. 13), it canbe seen that the fetus executes a yaw maneuver in a short timeand indeed returning relatively close to the original position dur-ing the course of just a few TRs. Typically, such wild movementwould be cause to reject a data set such as this in the context ofan adult healthy volunteer in a typical neuroscience experiment.

Head velocities of up to 10 mm per second and 10 degreesper second are not unexpected when imaging the fetal brain. Theissue then is that the brain does not, in a sense, move as a rigidbody as the acquired planes are not necessarily sampling paral-lel planes of tissue. An advantage of using a smaller flip angle

while still achieving the same CNR is the potential of fasterscanning and thus capturing any fetal motion on a finer scale,resulting in a lesser degree of image registration error, althoughthere have been some advances in addressing the interaction offetal movement and tissue sampling.68

DIFFUSION TENSOR IMAGINGDiffusion tensor imaging69,70 has been helpful in the un-

derstanding of the structure of the fetal brain.71,72 Certainly,the increased SNR at 3.0 T compared with 1.5 T can be used toimprove imaging times to help mitigate the issue of fetal move-ment during data collection, although offline methods do existto improve data quality.73,74 An aspect of fetal brain develop-ment and DTI is picking the appropriate b value. The typicaldiffusion weight image will have a signal scaled from the non-diffusion weighted image by a factor of

ejbD

where D is the diffusion coefficient in the tissue of interest,and b is the b value determined by the imaging parameters on thescanner. Typical b values for adult brain are on the order ofb = 700 to 1000 s/mm2.75 What is important to note is thatthe average diffusion coefficient D in the developing fetus is

FIGURE 15. Comparison of (adult) abdominal imaging with single-shot-fast-spin-echo (SSFSE) at 3.0 T. Left: Achieva 3.0T. Right:Ingenia 3.0T. Data courtesy of Dr Thomas Chenevert, Department of Radiology, University of Michigan.

FIGURE 16. Comparison of (adult) abdominal imaging with SSFSE at 3.0 T. Left: Achieva 3.0T. Right: Ingenia 3.0T. Data courtesy ofDr Thomas Chenevert, Department of Radiology, University of Michigan.




much higher76Y78 than that of the adult brain, and thus if ab value that is typically used for adult imaging is specified forfetal brain, the diffusion-weighted image (DWI) will suffer inrelative signal as compared with the adult brain DWI. Assum-ing a monoexponential decrease in signal intensity with increas-ing b value (for low b values) and DFetus ; 1.7 Km2/s79 andDAdult ; 0.8 Km/s,80 we can estimate the appropriate b value touse in fetus with the emphasis to achieve the same relative sig-nal depression. This is illustrated in Figure 14 and shows that ab value more on the order of b = 400 to 500 s/mm2 is appropri-ate. The lower b value then of course can result in a shorter echotime, allowing for an even more rapid image acquisition, al-though this is independent of the main field strength.

ACCELERATION METHODS ANDMULTITRANSMIT

Improvements in image acquisition in both the design ofpulse sequences and hardware higher quality imaging can takeplace. Awell-known issue with imaging the abdominal region isRF penetration.81 A means to mitigate this is to have multipletransmit coils that can result in more homogeneous.82 Recently,the University of Michigan and the University of Vienna haveacquired Philips Ingenia 3.0T (Philips Achieva, Philips Medi-cal Systems, Best, The Netherlands) systems with multitransmit

(parallel transmit) capability. With multitransmit, a flatter effec-tive B1 field can be generated. This is even more important inthe situation of pregnancy as the amniotic fluid is conductiveand can further affect the RF field penetration. In Figure 15, dataare shown from the University of Michigan for a healthy adultin comparison between an Achieva 3.0T system and the newerhardware present in the Ingenia 3.0T system. Clearly, there isvisual evidence of a more homogeneous field of view.

A rough estimate was also made on the improvement inSNR in various regions of the abdomen. In general, in the im-ages in Figure 16, an SNR gain was observed in a range of 40%to 60%, with the exception of the kidney only having a mar-ginal improvement of approximately 9%. For image acquisitionspeed up, parallel imaging is available. In addition to multi-transmit, multireceive coil systems are quite routine in clinicalsettings with 8 to 16 receive coils. It is envisioned though thata time when possibly 64- and 128-channel receive systems may1 day be used in fetal imaging.83

In addition to parallel imaging methods, such as SENSE,faster imaging sequences have been developed. For neurosci-ence applications, Feinberg et al84 reported in 2010 a methodby which to acquire a full gradient-echo echo-planar-image ofthe adult brain in times as fast as 400 milliseconds. Of course,to be balanced with this technique is ensuring that there is not

FIGURE 17. Comparison between 1.5 T (left) and 3.0 T (right) of 2 fetuses at 25-week gestational age. Left image: TE = 140 milliseconds,TR = 23.4 seconds, flip angle = 90 degrees, resolution = 0.90 mm, thickness = 3.0 mm. Right image: TE = 140 milliseconds,TR = 4.1 seconds, flip angle = 90 degrees, resolution = 0.72 mm, thickness = 3.0 mm. Data from Medical University of Vienna.

FIGURE 18. Comparison between 1.5 T (in utero) and 3.0 T (postmortem) at 25 gestational week and 27 gestational week, respectively.Left image: TE = 140 milliseconds, TR = 18.7 seconds, flip angle = 90 degrees, resolution = 0.90 mm, thickness = 3.0 mm. Rightimage: TE = 140 milliseconds, TR = 3 seconds, flip angle = 90 degrees, resolution = 0.40 mm, thickness = 3.0 mm. Data from MedicalUniversity of Vienna.




an appreciable increase in energy deposition for the pregnantwoman and fetus.

HIGH-RESOLUTION IMAGINGFetal brain development research has a long and illustrious

publication record.85Y99 Assessment of microstructures suchas synapses is still the realm of the wet laboratory,99 but MRIdoes have a role to play in the understanding of cortical de-velopment in utero/in vivo).87,91 Indeed, with the emergence ofadvanced offline techniques to achieve high-resolution imagesof the developing fetal brain,100Y103 groups around the worldhave started to develop templates and atlases of the normal fetalbrain.86,100,104Y107 Through empirical MRI studies of the fetalbrain using 3.0-T systems can a direct assessment be made tothe use of a stronger magnetic field compared with 1.5 T. Initialdata from Wayne State University and from the University ofVienna indicate better structure delineation at the higher strength.Examples from Vienna are in Figure 17. In the unfortunatesituation of fetal demise, an MRI gold standard of resolutionand contrast can be determined through postmortem MRI. InFigure 18, we show data of the same fetus at 25 GW (in utero) and27 GW (postmortem, complications of diaphragmatic herniaand reduced lung volume). Certainly, the fine nuances of post-mortem high-resolution imaging are yet to be achieved in utero,but with the combined increased SNR at 3.0 T and the advancedpostprocessing techniques, we are confident that improved struc-tural in utero imaging is possible.

CONCLUSIONSWe are just starting to see some of the advantages of doing

fetal imaging at 3.0 T. A paramount message of course is thesafety of both the mother and fetus and that this safety is not tobe compromised by imaging at a higher field. The recent devel-opments in SAR calculations certainly give the direction to followin patient safety with direct impact on fetal imaging. With theassumption that more exact RF and thermal models can be in-formative to the formation of guidelines, the horizon of com-monplace fetal imaging is not far away. Presently, 3.0-T imagingof the fetus is the realm of university and research facilities.With the advent of more sophisticated pulse sequences that canmitigate fetal motion and offline data processing methods thatcan also produce high-quality data void of motion, these 3.0-Tresearch facilities will pave the way and demonstrate the trueuse of fetal imaging at the higher field strength.

REFERENCES

1. Badash L. Marie Curie: In the laboratory and on the battlefield.Phys Today. 2003;9:37Y43.

2. Smith FW, Adam AH, Phillips WD. NMR imaging in pregnancy.Lancet. 1983;1:61Y62.

3. Girard N, Gouny SC, Viola A, et al. Assessment of normal fetalbrain maturation in utero by proton magnetic resonance spectroscopy.Magn Reson Med. 2006;56:768Y775.

4. Prayer D. Fetal MRI. Berlin, Germany: Springer-Verlag; 2011.

5. Levine D, Barnes PD, Edelman RR. Obstetric MR imaging. Radiology.1999;211:609Y617.

6. Zaretsky MV, Reichel TF, McIntire DD, et al. Comparison ofmagnetic resonance imaging to ultrasound in the estimation of birthweight at term. Am J Obstet Gynecol. 2003;189:1017Y1020.

7. Brugger PC, Stuhr F, Lindner C, et al. Methods of fetal MR: BeyondT2-weighted imaging. Eur J Radiol. 2006;57:172Y181.

8. Bulas D. Fetal magnetic resonance imaging as a complement to fetalultrasonography. Ultrasound Q. 2007;23:3Y22.

9. Reiser MF, Semmler W, Hricak H. Magnetic Resonance Tomography.Berlin, Germany: Springer-Verlag; 2007.

10. Jackson HA, Panigrahy A. Fetal magnetic resonance imaging:the basics. Pediatr Ann. 2008;37:388Y393.

11. Chen MM, Coakley FV, Kaimal A, et al. Guidelines for computedtomography and magnetic resonance imaging use during pregnancyand lactation. Obstet Gynecol. 2008;112:333Y340.

12. Jokhi RP, Whitby EH. Magnetic resonance imaging of the fetus.Dev Med Child Neurol. 2010;53:18Y28.

13. Brugger PC, Prayer D. Actual imaging time in fetal MRI. Eur J Radiol.2011;81:1Y3.

14. Bottomley PA, Foster TH, Argersinger RE, et al. A review of normaltissue hydrogen NMR relaxation times and relaxation mechanisms from1 to 100 MHz: dependence on tissue type, NMR frequency,temperature, species, excision, and age. Med Phys. 1984;11:425.

15. Prayer D, Kasprian G, Krampl E, et al. MRI of normal fetal braindevelopment. Eur J Radiol. 2006;57:199Y216.

16. Abd Almajeed A, Adamsbaum C, Langevin F. Myelin characterizationof fetal brain with mono-point estimated T1-maps. Magn Reson

Imaging. 2004;22:565Y572.

17. Williams LA, Gelman N, Picot PA, et al. Neonatal Brain: regionalvariability of in vivo MR imaging relaxation rates at 3.0 TVinitialexperience. Radiology. 2005;235:595Y603.

18. Duyn JH, Tan CX, van Gelderen P, et al. High-sensitivity single-shotperfusion-weighted fMRI. Magn Reson Med. 2001;46:88Y94.

19. Yongbi MN, Fera F, Mattay VS, et al. Simultaneous BOLD/perfusionmeasurement using dual-echo FAIR and UNFAIR: sequencecomparison at 1.5T and 3.0T. Magn Reson Imaging. 2001;19:1159Y1165.

20. Yongbi MN, Fera F, Yang Y, et al. Pulsed arterial spin labeling:comparison of multisection baseline and functional MR imagingperfusion signal at 1.5 and 3.0 T: initial results in six subjects.Radiology. 2002;222:569Y575.

21. Fera F, Yongbi MN, van Gelderen P, et al. EPI-BOLD fMRI ofhuman motor cortex at 1.5 T and 3.0 T: sensitivity dependence onecho time and acquisition bandwidth. J Magn Reson Imaging.2004;19:19Y26.

22. Hussain SM, Wielopolski PA, Martin DR. Abdominal magneticresonance imaging at 3.0 T: problem or a promise for the future?Top Magn Reson Imaging. 2005;16:325Y335.

23. Stanisz GJ, Odrobina EE, Pun J, et al. T1, T2 relaxation andmagnetization transfer in tissue at 3T. Magn Reson Med. 2005;54:507Y512.

24. Rohrer M, Bauer H, Mintorovitch J, et al. Comparison of magneticproperties of MRI contrast media solutions at different magneticfield strengths. Invest Radiol. 2005;40:715Y724.

25. Merkle EM. Abdominal MRI at 3.0 T: the basics revisited. AJR Am

J Roentgenol. 2006;186:1524Y1532.

26. Lauzon ML. Time-efficient breath-hold abdominal MRI at 3.0 T.AJR Am J Roentgenol. 2006;187:649Y657.

27. Schmidt GP, Wintersperger B, Graser A, et al. High-resolutionwhole-body magnetic resonance imaging applications at 1.5 and3 Tesla: a comparative study. Invest Radiol. 2007;42:449Y459.

28. Cornfeld D, Weinreb J. Simple changes to 1.5-T MRI abdomen andpelvis protocols to optimize results at 3 T. AJR Am J Roentgenol.2008;190:W140YW150.

29. Homann H, Bornert P, Eggers H, et al. Toward individualized SARmodels and in vivo validation. Magn Reson Med. 2011;66:1767Y1776.

30. Helle M, Jerosch-Herold M, Voges I, et al. Improved MRI of theneonatal heart: feasibility study using a knee coil. Pediatr Radiol.2011;41:1429Y1432.




31. Murtz P, Kaschner M, Traber F, et al. Diffusion-weighted whole-bodyMRI with background body signal suppression: technicalimprovements at 3.0 T. J Magn Reson Imaging. 2011;35:456Y461.

32. Nehrke K, Bornert P. DREAMVa novel approach for robust, ultrafast,multislice B(1) mapping. Magn Reson Med. 2012;DOI10.1002/mrm.24158. [Epub ahead of print].

33. Bernstein MA, King KF, Zhou XJ. Handbook of MRI PulseSequences. San Diego, CA: Elsevier Academic Press; 2004.

34. Kanal E, Barkovich AJ, Bell C, et al. ACR guidance document forsafe MR practices: 2007. AJR Am J Roentgenol. 2007;188:1447Y1474.

35. Triulzi F, Manganaro L, Volpe P. Fetal magnetic resonance imaging:indications, study protocols and safety. Radiol Med. 2011;116:337Y350.

36. Levine D, Zuo C, Faro CB, et al. Potential heating effect in thegravid uterus during MR HASTE imaging. J Magn Reson Imaging.2001;13:856Y861.

37. Hand JW, Li Y, Thomas EL, et al. Prediction of specific absorptionrate in mother and fetus associated with MRI examinations duringpregnancy. Magn Reson Med. 2006;55:883Y893.

38. US Food and Drug Administration. Criteria for Significant RiskInvestigations of Magnetic Resonance Diagnostic Devices.Washington, DC: US Department of Health and Human Services;2003;1Y5.

39. International Commission on Non-Ionizing Radiation Protection.Medical magnetic resonance (MR) procedures: protection ofpatients. Health Phys. 2004;87:197Y216.

40. Gowland PA, De Wilde J. Temperature increase in the fetus due toradio frequency exposure during magnetic resonance scanning.Phys Med Biol. 2008;53:L15YL18.

41. Hand JW, Li Y, Hajnal JV. Numerical study of RF exposure andthe resulting temperature rise in the foetus during a magnetic resonanceprocedure. Phys Med Biol. 2010;55:913Y930.

42. Kikuchi S, Saito K, Takahashi M, et al. Temperature elevation in thefetus from electromagnetic exposure during magnetic resonanceimaging. Phys Med Biol. 2010;55:2411Y2426.

43. Neufeld E, Gosselin M-C, Murbach M, et al. Analysis of the localworst-case SAR exposure caused by an MRI multi-transmit bodycoil in anatomical models of the human body. Phys Med Biol.2011;56:4649Y4659.

44. Story L, Damodaram MS, Allsop JM, et al. Proton magneticresonance spectroscopy in the fetus. Eur J Obstet Gynecol Reprod Biol.2011;158:3Y8.

45. Fenton BW, Lin CS, Seydel F, et al. Lecithin can be detected byvolume-selected proton MR spectroscopy using a 1.5 Twhole-bodyscanner: a potentially non-invasive method for the prenatal assessmentof fetal lung maturity. Prenat Diagn. 1998;18:1263Y1266.

46. Cetin I, Barberis B, Brusati V, et al. Lactate detection in the brain ofgrowth-restricted fetuses with magnetic resonance spectroscopy.Am J Obstet Gynecol. 2011;205:350.e1Y350.e7.

47. Rijn AMR-V, Groenendaal F, Stoutenbeek P, et al. Lactate in thefoetal brain: detection and implications. Acta Paediatrica. 2007;93:937Y940.

48. Bhattacharyya PK, Lowe MJ, Phillips MD. Spectral quality controlin motion-corrupted single-voxel J-difference editing scans: aninterleaved navigator approach. Magn Reson Med. 2007;58:808Y812.

49. Keating B, Deng W, Roddey JC, et al. Prospective motion correctionfor single-voxel 1H MR spectroscopy. Magn Reson Med. 2010;64:672Y679.

50. Keating B, Ernst T. Real-time dynamic frequency and shim correctionfor single-voxel magnetic resonance spectroscopy. Magn Reson Med.2012;DOI10.1002/mrm.24129. [Epub ahead of print].

51. Germuska M, Tunariu N, Leach MO, et al. An evaluation ofmotion compensation strategies and repeatability for abdominal(1) H MR spectroscopy measurements in volunteer studies and clinicaltrials. NMR Biomed. 2012;25:859Y865.

52. Hess AT, Dylan Tisdall M, Andronesi OC, et al. Real-time motionand B0 corrected single voxel spectroscopy using volumetricnavigators. Magn Reson Med. 2011;66:314Y323.

53. Lange T, Maclaren J, Buechert M, et al. Spectroscopic imagingwith prospective motion correction and retrospective phase correction.Magn Reson Med. 2012;67:1506Y1514.

54. Ogawa S. Brain magnetic resonance imaging with contrastdependent on blood oxygenation. Proc Natl Acad Sci U S A.1990;87:9868Y9872.

55. Ogawa S, Tank DW, Menon R, et al. Intrinsic signal changesaccompanying sensory stimulation: functional brain mapping withmagnetic resonance imaging. Proc Natl Acad Sci U S A.1992;89:5951Y5955.

56. Hykin J, Moore R, Duncan K, et al. Fetal brain activity demonstratedby functional magnetic resonance imaging. Lancet. 1999;354:645Y646.

57. Moore RJ, Vadeyar SH, Fulford J, et al. Antenatal determination offetal brain activity in response to an acoustic stimulus using functionalmagnetic resonance imaging. Hum Brain Mapp. 2001;12:94Y99.

58. Fulford J, Vadeyar SH, Dodampahala SH, et al. Fetal brain activityin response to a visual stimulus. Hum Brain Mapp. 2003;20:239Y245.

59. Fulford J, Vadeyar SH, Dodampahala SH, et al. Fetal brain activityand hemodynamic response to a vibroacoustic stimulus.Hum Brain Mapp. 2004;22:116Y121.

60. Schopf V, Kasprian G, Brugger PC, et al. Watching the fetal brainat ‘‘rest.’’ Int J Dev Neurosci. 2012;30:11Y17.

61. Haacke EM. Magnetic Resonance Imaging: Physical Principles and

Sequence Design. New York, NY: J. Wiley & Sons; 1999.

62. Schenck JF. The role of magnetic susceptibility in magneticresonance imaging: MRI magnetic compatibility of the first andsecond kinds. Med Phys. 1996;23:815Y850.

63. Kruger G, Kastrup A, Glover GH. Neuroimaging at 1.5 T and 3.0 T:comparison of oxygenation-sensitive magnetic resonance imaging.Magn Reson Med. 2001;45:595Y604.

64. Wedegartner U, Popovych S, Yamamura J, et al. R2* in fetal sheepbrains during hypoxia: MR imaging at 3.0 T versus that at 1.5 T.Radiology. 2009;252:394Y400.

65. Westlye LT, Walhovd KB, Dale AM, et al. Differentiating maturationaland aging-related changes of the cerebral cortex by use of thicknessand signal intensity. NeuroImage. 2010;52:172Y185.

66. Rivkin MJ, Wolraich D, Als H, et al. Prolonged T*2 values in newbornversus adult brain: implications for fMRI studies of newborns.Magn Reson Med. 2004;51:1287Y1291.

67. Gonzalez-Castillo J, Roopchansingh V, Bandettini PA, et al.Physiological noise effects on the flip angle selection in BOLD fMRI.NeuroImage. 2011;54:2764Y2778.

68. Kim K, Habas PA, Rousseau F, et al. Intersection based motioncorrection of multislice MRI for 3-D in utero fetal brain imageformation. IEEE Trans Med Imaging. 2010;29:146Y158.

69. Basser PJ, Pierpaoli C. Microstructural and physiological featuresof tissues elucidated by quantitative-diffusion-tensor MRI.J Magn Reson B. 1996;111:209Y219.

70. Basser PJ, Pajevic S, Pierpaoli C, et al. In vivo fiber tractographyusing DT-MRI data. Magn Reson Med. 2000;44:625Y632.

71. Kasprian G, Brugger PC, Weber M, et al. In utero tractography offetal white matter development. NeuroImage. 2008;43:213Y224.




72. Kasprian G, Del Rıo M, Prayer D. Fetal diffusion imaging: pearlsand solutions. Top Magn Reson Imaging. 2010;21:387Y394.

73. Jiang S, Xue H, Counsell S, et al. Diffusion tensor imaging (DTI)of the brain in moving subjects: application to in-utero fetal andex-utero studies. Magn Reson Med. 2009;62:645Y655.

74. Oubel E, Koob M, Studholme C, et al. Reconstruction of scattereddata in fetal diffusion MRI. Med Image Anal. 2012;16:28Y37.

75. Basser P, Jones D. Diffusion-tensor MRI: theory, experimental designand data analysisVa technical review. NMR Biomed. 2002;15:456Y467.

76. Righini A, Bianchini E, Parazzini C, et al. Apparent diffusioncoefficient determination in normal fetal brain: a prenatal MR imagingstudy. AJNR Am J Neuroradiol. 2003;24:799Y804.

77. Kim D-H, Chung S, Vigneron DB, et al. Diffusion-weightedimaging of the fetal brain in vivo. Magn Reson Med.2007;59:216Y220.

78. Schneider JF, Confort-Gouny S, Fur Y, et al. Diffusion-weightedimaging in normal fetal brain maturation. Eur Radiol. 2007;17:2422Y2429.

79. Cartry C, Viallon V, Hornoy P, et al. [Diffusion-weighted MRimaging of the normal fetal brain: marker of fetal brain maturation].J Radiol. 2010;91:561Y566.

80. Dong Q, Welsh RC, Chenevert TL, et al. Clinical applications ofdiffusion tensor imaging. J Magn Reson Imaging. 2003;19:6Y18.

81. Merkle EM, Dale BM, Paulson EK. Abdominal MR imaging at 3T.Magn Reson Imaging Clin N Am. 2006;14:17Y26.

82. Murtz P, Kaschner M, Traber F, et al. Evaluation of dual-sourceparallel RF excitation for diffusion-weighted whole-body MR imagingwith background body signal suppression at 3.0 T. Eur J Radiol.2011;1Y10.

83. Grant PE. Nonconventional MRI for fetal imaging. Proceedsingsof the 19th Annual Meeting of the International Society of Magnetic

Resonance in Medicine. Montreal, Canada, 2011:1Y1.

84. Feinberg DA, Moeller S, Smith SM, et al. Multiplexed echo planarimaging for subsecond whole-brain fMRI and fast diffusionimaging. PLoS ONE. 2010;5:e15710.

85. Clouchoux C, Guizard N, Evans AC, et al. Normative fetal braingrowth by quantitative in vivo magnetic resonance imaging.Am J Obstet Gynecol. 2011;206:1Y8.

86. Yang Y, Roussotte F, Kan E, et al. Abnormal cortical thicknessalterations in fetal alcohol spectrum disorders and their relationshipswith facial dysmorphology. Cereb Cortex. 2012;22:1170Y1179.

87. Huppi PS. Cortical development in the fetus and the newborn: advancedMR techniques. Top Magn Reson Imaging. 2011;22:33Y38.

88. Kostovi( I, Judas M. The development of the subplate andthalamocortical connections in the human foetal brain.Acta Paediatrica. 2010;99:1119Y1127.

89. Tau GZ, Peterson BS. Normal development of brain circuits.Neuropsychopharmacology. 2009;35:147Y168.

90. Glenn OA. Normal development of the fetal brain by MRI.Semin Perinatol. 2009;33:208Y219.

91. Kostovic I, Vasung L. Insights from in vitro fetal magneticresonance imaging of cerebral development. Semin Perinatol.2009;33:220Y233.

92. Limperopoulos C, Clouchoux C. Advancing fetal brain MRI: targetsfor the future. Semin Perinatol. 2009;33:289Y298.

93. Marsh R, Gerber AJ, Peterson BS. Neuroimaging studies ofnormal brain development and their relevance for understandingchildhood neuropsychiatric disorders. J Am Acad ChildAdolesc Psychiatry. 2008;47:1233Y1251.

94. Rados M, Judas M, Kostovi( I. In vitro MRI of brain development.Eur J Radiol. 2006;57:187Y198.

95. Fogliarini C, Chaumoitre K, Chapon F, et al. Assessment of corticalmaturation with prenatal MRI. Eur Radiol. 2005;15:1781Y1789.

96. Huisman TAGM, Martin E, Kubik-Huch R, et al. Fetal magneticresonance imaging of the brain: technical considerations andnormal brain development. Eur Radiol. 2002;12:1941Y1951.

97. Kostovi( I, Judas M, Rados M, et al. Laminar organization of thehuman fetal cerebrum revealed by histochemical markers and magneticresonance imaging. Cereb Cortex. 2002;12:536Y544.

98. Chong BW, Babcook CJ, Pang D, et al. A magnetic resonance templatefor normal cerebellar development in the human fetus. Neurosurgery.1997;41:924Y928 [discussion 928Y9].

99. Molliver ME, Kostovic I, van der Loos H. The development ofsynapses in cerebral cortex of the human fetus. Brain Res.1973;50:403Y407.

100. Gholipour A, Estroff JA, Warfield SK. Robust super-resolutionvolume reconstruction from slice acquisitions: application to fetal brainMRI. IEEE Trans Med Imaging. 2010;29:1739Y1758.

101. Neustadter DM, Chiel HJ. Imaging Freely Moving Subjects UsingContinuous Interleaved Orthogonal Magnetic Resonance Imaging.Magn Reson Imaging. 2004;22:329Y343.

102. Rousseau F, Glenn O, Iordanova B, et al. A Novel Approach to HighResolution Fetal Brain MR Imaging. Med Image Comput ComputAssist Interv. 2005;8:548Y555.

103. Rousseau F, Kim K, Studholme C, et al. On Super-Resolution forFetal Brain MRI. Med Image Comput Comput Assist Interv.2010;13:355Y362.

104. Habas PA, Scott JA, Roosta A, et al. Early folding patterns andasymmetries of the normal human brain detected from in utero MRI.Cereb Cortex. 2012;22:13Y25.

105. Scott JA, Hamzelou KS, Rajagopalan V, et al. 3D morphometricanalysis of human fetal cerebellar development. Cerebellum.2011;1Y10.

106. Habas PA, Kim K, Rousseau F, et al. Atlas-based segmentation ofdeveloping tissues in the human brain with quantitative validationin young fetuses. Hum Brain Mapp. 2010;31:1348Y1358.

107. Chapman T, Matesan M, Weinberger E, et al. Digital atlas of fetalbrain MRI. Pediatr Radiol. 2010;40:153Y162.




Fetal Magnetic Resonance Imaging at 3.0 T - pedrad.org MRI/Fetal MRI 3T 2011.pdf · Fetal Magnetic...

Documents

Transcript of Fetal Magnetic Resonance Imaging at 3.0 T - pedrad.org MRI/Fetal MRI 3T 2011.pdf · Fetal Magnetic...