AC

ITmcpdtcbhmnbit

*

†

‡§

T

A

0d

dvancing Fetal Brain MRI: Targets for the Futureatherine Limperopoulos, PhD,*,†,‡,§ and Cedric Clouchoux, PhD*,§

Fetal MRI is becoming an increasingly powerful imaging tool for studying brain develop-ment in vivo. Until recently, the application of advanced magnetic resonance imagingtechniques was limited by motion in the nonsedated fetus. Extensive research effortscurrently underway are focusing on the development of dedicated magnetic resonanceimaging sequences and sophisticated postprocessing techniques that are revolutionizingour ability to study the healthy and compromised fetus. The ongoing refinement of thesemagnetic resonance imaging techniques will undoubtedly lead to the development ofcornerstone biomarkers that will provide healthcare caregivers with vital, and currentlylacking, information upon which to counsel parents effectively, and base rational decisionsregarding the timing and type of novel medical and surgical interventions currently on thehorizon.Semin Perinatol 33:289-298 © 2009 Elsevier Inc. All rights reserved.

KEYWORDS fetal, brain, MRI, development

CTvtndttfolt

cutpppatai

npcp

n vivo fetal magnetic resonance imaging (MRI) is revolu-tionizing our ability to study human brain development.

o date, fetal MRI has been used primarily for the qualitativeorphologic evaluation of the fetal brain. However, the re-

ent application of advanced MRI techniques to the fetus hasrovided an unprecedented opportunity to investigate theeveloping brain in vivo. The ability to begin reliable inves-igation of brain growth and development in the healthy andompromised fetus promises a range of new quantitativeiomarkers that can be applied clinically, and that willelp to formulate a better understanding of brain develop-ent and lead to improved management of high-risk preg-ancies. This article explores the advancing role of fetalrain MRI, provides an overview of the current obstacles

n MRI of the living fetus, and highlights future targets inhis rapidly evolving field.

Department of Neurology and Neurosurgery, McGill University, Montreal,Quebec, Canada.

School of Physical and Occupational Therapy, McGill University, Montreal,Quebec, Canada.

Department of Pediatrics, McGill University, Montreal, Quebec, Canada.McConnell Brain Imaging Centre, Neurological Institute, McGill University,

Montreal, Quebec, Canada.his work was partially supported by the Canadian Research Chairs Program

(to C.L.) and the Canadian Institutes of Health Research.ddress reprint requests to Catherine Limperopoulos, PhD, Canada Re-

search Chair in Brain and Development, Montreal Children’s Hospital,Pediatric Neurology, 2300 Tupper St A-334, Montreal, Quebec, Canada,

MH3H 1P3. E-mail: catherine.limperopoulos@mcgill.ca

146-0005/09/$-see front matter © 2009 Elsevier Inc. All rights reserved.oi:10.1053/j.semperi.2009.04.002

hallenges of Fetal MRIhe in vivo study of the developing cerebral parenchyma isery complex and continues to be challenged by several fac-ors. One of the biggest problems encountered in performingonsedated fetal MRI is fetal motion.1,2 Despite the recentevelopment in ultrafast MRI that facilitates image acquisi-ion in milliseconds, advanced fetal MRI acquisitions, such ashree-dimensional (3-D)-acquired T2-weighted sequencesor 3-D brain reconstruction, diffusion tensor imaging (DTI),r magnetic resonance spectroscopic imaging, typically takeonger to acquire and are therefore more susceptible to mo-ion artifact.

Although adult brain motion can be controlled and easilyorrected,3 fetal motion is uncontrollable without sedation, isnpredictable, and occurs in all planes.4,5 In normal condi-ions, motion implies that the resulting slices are not perfectlyarallel to each other, and that motion occurs with overlap-ing signal tissues resulting in image artifact.3 Briefly, theroblem can be summarized as a compromise between shortcquisition time and good image resolution. A short acquisi-ion time is needed to reduce both maternal and fetal motionrtifact as well as subject discomfort; however, improvedmage resolution requires an increased scanning time.

The high water content of the immature, largely unmyeli-ated fetal brain also results in poor contrast in the cerebralarenchyma (ie, low resolution), making it inherently diffi-ult to obtain high-resolution MR images. Consequently,oor tissue contrast can affect the diagnostic accuracy of fetal

RI. Decreased tissue contrast, for example, can hinder the

289

rwmitlfmttnlvcs

rsd

OCMSrqit

tmsltstpg

MAtcfAai3htThitt(gt

290 C. Limperopoulos and C. Clouchoux

eliable identification of diffuse white matter abnormalities,hich is of critical importance for detecting conditions thatay affect the fetal brain, such as hypoxia-ischemia and viral

nfections.6,7 Another important challenge is the clinical in-erpretation of more subtle anatomical anomalies of unclearong-term significance, which are increasingly detected byetal MRI. Their consideration often requires taxing decision-

aking by both care providers and families, and may lead toermination of pregnancy.8,9 Moreover, it has been shownhat in cases in which a viable fetus is delivered, a misdiag-osis and resulting errors in prognostication have serious

ong-term consequences for parents.8,10 It is noteworthy thatery few MRI series of normal fetal brain development areurrently available, resulting in a fundamentally poor under-tanding of the limits of normality and variability.1,8

These ongoing fetal MRI challenges are subject to extensiveesearch, including the development of dedicated fetal MRIequences, as well as sophisticated postprocessing algorithmsescribed later in the text.

vercoming Currenthallenges of Fetal MRIR Image Acquisition

everal strategies have been explored to decrease motion-elated artifact. The development of specific fetal MRI se-uences is one way to attempt to overcome or minimize

mage degradation associated with motion artifact; however,his remains a multifaceted task because of the inherent na-

Figure 1 High resolution image reconstruction of a hT2-weighted images. Each row represents low-resolut

column: resulting high-resolution isotropic volume (1 � 1 �

ure of the immature brain and surrounding uterine environ-ent. Jiang et al4 have proposed the use of repeated dynamic

ingle-shot MRI sequences to sample the region of interest,eading to the acquisition of multiple overlapping slices. Onhe basis of the assumption that every part of the fetal brain isampled, a registration postprocessing technique allows forhe recovery of the original information in its entirety. Thisreliminary work has shown promising results, although on-oing research in fetal MRI acquisition is still needed.

RI Post-Processing Techniquesnother approach to overcome image degradation secondary

o fetal motion relies on the application of advanced postpro-essing techniques. Development in dedicated motion arti-act detection and removal tools is currently under research.lgorithms developed for adults often fail, given that fetalnd maternal motion are often mixed, as well as the intrinsicmmaturity and limited contrast of the fetal brain. Typically,

fetal scans are acquired, 1 in each direction, providing aigh-resolution view for each direction. The challenge then iso re-create a single, isotropic high-resolution volume.4,5

hese techniques rely on the assumption that the entire brainas been sampled in multiple shots, and that the resulting

mages contain the entire information set. Various steps arehen mandatory, and specific algorithms must be developedo recover this information and create high-resolution imagesFig. 1). A complex set of registration and segmentation al-orithms are involved; however, a detailed description ofhese methodologies is beyond the scope of this article.

35 week gestational age fetus. Left column: originalns in 1 direction (axial, sagittal, and coronal). Right

ealthyion sca

1 mm3) after preprocessing and coregistration.

piutatocmf

CFVQQstTwpone

arcfatnhubmtov

F3ri3

Advancing fetal brain MRI 291

The presence of corrupted slices during the acquisitionhase is another motion-related challenge in fetal MRI stud-

es. Although conventional interpolation methods may besed to overcome the problem, an elegant and simple solu-ion has recently been proposed,11 in which corrupted slicesre replaced using a novel in-painting process, taking advan-age of existing data from surrounding slices (Fig. 2). Thengoing development and refinement of dedicated postpro-essing algorithms is mandatory for image contrast enhance-ent and the creation of high-resolution images is necessary

or quantitative fetal brain measurements.

igure 2 MR image reconstruction using 3-D inpainting in a healthy4 week gestational age fetus. Axial T2-weighted image on the leftepresents the original corrupted MRI slice (red arrow), and themage on the right demonstrates the same reconstructed slice using-D inpainting. (Color version of figure is available online.)

Figure 3 Three-dimensional manual segmentation of to

using multiplanar T2-weighted images. (Color version of figur

urrent Status of Advancedetal MR Imaging Techniquesolumetric Brain Growth of the Fetal Brainuantitative 3-D Volumetric MRIuantitative 3-D volumetric MRI has provided major in-

ights into the developmental changes in specific brain struc-ures and tissue subtypes of the immature brain infant.12,13

he ability to make quantitative measurements of gray andhite matter volumes in the preterm infant over the thirdostconceptional trimester has advanced our understandingf the rate and progression of brain development,13 as well asormal and abnormal cerebral cortical development and my-lination.

More recent studies have begun to explore the feasibility ofcquiring 3-D volumetric data in utero, allowing for 3-Deconstruction and volume rendering of fetal brain paren-hyma and extra-axial cerebrospinal fluid.14,15 However, 3-Detal MRI remains difficult because of motion during imagecquisition. Motion corrupts the 3-D position and orienta-ion of individual slices during the acquisition. Studies areow beginning to examine fetal brain growth in healthy andigh-risk fetuses. To date, determination of fetal brain vol-me has been performed manually (Fig. 3) to quantify 3-Drain growth.16,17 A recent study using quantitative 3-D volu-etric MRI demonstrated a strong linear relationship be-

ween total brain volume and gestational age in healthy sec-nd and third trimester fetuses. It also described the first inivo evidence of abnormal brain growth in fetuses with con-

n volume (blue) and intracranial cavity volume (red),

tal brai e is available online.)

gttia

Mmrtahwcbdcbabm

MDDDuccdtbmsnd

gsalc

mftwtbletcocrtobc

dIomtniitws

line.)

292 C. Limperopoulos and C. Clouchoux

enital heart disease as compared with healthy control fe-uses, characterized by progressively smaller third trimesterotal brain volume.18 These preliminary data promise excit-ng opportunities to begin to quantify the timing and type ofberrant brain growth in compromised fetuses in utero.

The various postprocessing steps currently in use in fetalRI are cumbersome and time-consuming, and their auto-ation would represent a major advancement for several

easons. The first advantage is the obvious gain of time, par-icularly when processing large MRI datasets. The seconddvantage of automated processing pipelines is minimizinguman bias. Understandably, extensive validation studiesould be needed to ensure the reliability of these postpro-

essing automated techniques. Innovative automated model-ased techniques for fetal brain parcellation (Fig. 4) are beingeveloped and validated.19 These techniques are already fa-ilitating the quantitative study of third trimester regionalrain growth in healthy fetuses (Fig. 4). Application of theseutomated approaches to characterize temporal and regionalrain growth in fetuses at risk for impaired brain develop-ent is currently underway.

icrostructuralevelopment of the Fetal Brainiffusion-Weighted Imaging and DTIiffusion-weighted imaging (DWI) and DTI are techniq-es used to evaluate maturation-dependent microstructuralhanges associated with brain growth and development oferebral white matter.13,20-23 From a technical perspective,iffusion refers to the migration of water molecules over in-racapillary distances.23 On the basis of this principle, DWI isased on the preferential diffusion of water molecules in aagnetic field. Increased diffusion reduces MR signal in a

pecific direction, whereas lower diffusion results in less sig-al loss and a brighter display in the image. The degree of

Figure 4 Automatic volumetric parcellation of a healthyT2-weighted images. Bottom row represents automatic pbrainstem (blue). (Color version of figure is available on

iffusion weighting corresponds to the strength of diffusion n

radients, characterized by their b-value. Specific acquisitionequences are then used to capture diffusion-weighted datand to generate images that give precise information on mo-ecular displacements over short distances, comparable toell dimensions.24

To date, these advanced MRI techniques have been limitedainly to postnatal MRI studies in high-risk pre-term and

ull-term infants.1 Technical challenges encountered with fe-al diffusion imaging relate primarily to echo-planar imagingith inherently noisy sequences very sensitive to motion ar-

ifacts. Additionally, the high-water content of the immaturerain amplifies this phenomenon of diffusion.24 Spatial reso-

ution is poor, which may in turn lead to a partial volumeffect. Despite these challenges, several studies have exploredhe role of DWI in the fetus and have demonstrated that DWIan be used to investigate normal and abnormal brain devel-pment.24-30 A progressive decrease of the apparent diffusionoefficient after 30 weeks gestation age in the supratentorialegions has been demonstrated, using fetal MRI,86 similar tohat previously reported in premature infants. Normal valuesf fetal brain apparent diffusion coefficient measures haveeen established, which allow detection of pathologichanges, such as hemorrhage and acute ischemia.26,27,31,32

DTI is a technique derived from DWI, determining theirection and magnitude of the water molecules diffusion.33

n particular, DTI allows the visualization and quantificationf white matter fiber direction. This is made possible byeasuring fractional anisotropy, which gives information on

he shape of the diffusion tensor for each voxel (based onormalized variance of eigenvalues). The difference between

sotropic and anisotropic diffusion is then obtained, givingnformation on white matter structure and state. Even thoughhere is an intrinsic low anisotropy in the developing brain,hite matter fiber tract maturational changes have been de-

cribed from birth.34 Using fractional anisotropy in this man-

k fetal brain. First row represents mulitplanar originaltion of the cerebrum (white), cerebellum (yellow), and

33-weearcella

er, 3 broad phases in postnatal white matter development

hsctmf

dltif

iw

MPPcmhdm1srmemi

ioqstIwwacmbps

mcbaw

FP(fitafi

Advancing fetal brain MRI 293

ave been described: that is, rapid changes in the size andhape of white matter tracts over the first 12 months, slowerhanges over the second year, and a relatively stable picturehereafter. These findings have been used to generate a nor-ative database34 that is available online and provides a use-

ul framework for studies of postnatal brain development.Despite its potential importance for the study of fetal brain

evelopment, the clinical application of fetal DTI remainsimited. To date, only 1 study has successfully applied DTractography to the apparently normal fetus35 to describe then vivo microstructural development of white matter (Fig. 5)rom as early as 18 weeks of gestation. This pioneering work

igure 5 In vivo diffusion tensor tractography (courtesy of Drsrayer and Kasprian). The fibers of the bilateral corticospinal tractsgreen and blue fibers), the genu (pink fibers), and splenium (yellowbers) of the corpus callosum have been superimposed to an ana-omical T2-weighted image. Fiber tracking was performed based ondiffusion-tensor sequence with 36 directions. (Color version ofgure is available online.)

Figure 6 Proton MRS (TE 144 ms) acquired in a healthy,cerebral hemisphere. Identifiable peaks on magnetic res

tine (Cr), and choline (Cho). (Color version of figure is availab

s an important landmark for future studies of abnormalhite matter development in the living fetus.

etabolic Development of the Fetal Brainroton Magnetic Resonance Spectroscopyroton magnetic resonance spectroscopy (1H-MRS) is be-oming a powerful noninvasive tool for examining cerebraletabolism in the fetus in vivo. This technique is based onydrogen resonance frequency that is altered by the imme-iate chemical environment,36 and the concentration of theetabolite is directly linked to the resulting signal intensity.

H-MRS allows measurement of specific brain metabolites,uch as the N-acetyl aspartate (NAA), a neuroaxonal markereflecting development of dendrites and synapses, as well asitochondrial metabolism, creatine (Cr) reflecting cellular

nergy metabolism, choline (Cho) a marker of myelination,yoinositol, a glial marker, and lactate that accumulates dur-

ng anaerobic metabolism (Fig. 6).The application of 1H-MRS to the fetus continues to face

nherent scanning challenges, such as the lack of availabilityf dedicated coils, the distance from the fetal brain, and re-uirements of a long acquisition time.1 Nonetheless, severaltudies have described the in vivo metabolic maturation ofhe fetal brain between 22 and 39 weeks’ gestational age.36-43

n the normal fetus, Cr is visible as early as 22 weeks, alongith a small NAA peak. Both NAA and Cr values increaseith increasing gestational age, presumably reflecting syn-

pse and dendrite development, whereas Cho gradually de-reases likely representing variations in substrate required forembrane synthesis and myelination.36-39,43 Although cere-

ral lactate has been identified by neonatal 1H-MRS in stableremature infants, it has not been reported in normal fetu-es.40,44

The availability of normative 1H-MRS data for fetal brainetabolites provides a valuable reference for measurement of

erebral metabolites in pathologic conditions of the fetalrain. Several studies have begun to explore metabolic alter-tions associated with MRI-detected structural changes, asell as in conditions of potential fetal compromise.45-47 In a

ek gestational age fetus with voxel positioned in the lefte (MR) spectra include N-acetyl-aspartate (NAA), crea-

33-weonanc

le online.)

rlfhwsLp

NTFFostatsvbTedtraMtti

FPosspipimssfssws1fdcamp

aalmt

PNgetnitetflp

munts

aatbmmfmfiotaauose

mieiha2tbp(m

294 C. Limperopoulos and C. Clouchoux

ecent study, Limperopoulos et al48 described progressivelyower NAA and/or Cho ratios during the third trimester inetuses with congenital heart disease as compared withealthy fetuses. In addition, the presence of cerebral lactateas detected in 20% of fetuses with congenital heart disease,

uggesting impaired third trimester brain metabolism.48

ong-term follow-up studies are underway to examine theredictive validity of impaired brain metabolism in utero.

ovel Fetal MR Imagingechniques on the Horizonetal Brain Activity by MRIunctional MRI has been used extensively in the adult and anlder child to study the regional activation of the brain topecific stimuli and activities.49-51 This approach is based onhe normal regional neuronal perfusion coupling in the brainnd investigates regional changes in blood oxygenation usinghe blood oxygen level-dependent response, which can beummarized as follows. Specifically, regional neuronal acti-ation responses to specific stimuli trigger a local increase inlood flow, blood volume, and venous blood oxygenation.his in turn changes the oxy- and deoxyhemoglobin differ-nce, thereby increasing the local MRI blood oxygen level-ependent contrast signal.52,53 With regard to applying func-ional MRI to the fetus, important technical challenges persistelated to fetal motion, applying stimuli to the fetus, and thenalysis and interpretation of data.53,54 Therefore, functionalRI in the fetus remains limited to research protocols at this

ime. More details on this technique and its clinical applica-ion are reviewed by Gowland and Fulford52 elsewhere in thisssue.

etal Behavior by MRIreliminary studies are beginning to explore the applicationf MRI for the evaluation of fetal movement to provide anapshot of the functional development of the fetal nervousystem at different gestational ages. This approach has theotential to expand the assessment of normal brain function

n the fetus through assessment of spontaneous movementatterns.2 Fetal behavior comprises spontaneous and reflex-

ve movement55 and increases in complexity with increasingaturity. Fetal movements rely on an intact neuromuscular

ystem and a normal metabolic state of the central nervousystem.56 A range of known movements characterize normaletal behavior. As described by Prayer,2 the different growthtages have specific and characteristic movements. For in-tance, the first fetal movements occur at around the 8theek of gestation and are characterized by flexion and exten-

ion of the vertebral column. It is noteworthy that until the9th week of gestation, these movements do not originaterom cerebral activity, and likely result from spontaneousischarges at the spinal and brainstem levels. Coordinatedomplex movements, involving the arm, leg, neck and trunk,ppear from the 9th week of gestation.57 More organizedovements are observable from the 14th week, while inde-

endent movements of the extremities appear between 26 w

nd 32 weeks’ gestational age.58 This approach could becomeuseful adjunct to other techniques for studying fetal neuro-

ogic development. Deviation of gestational age-appropriateovement repertoires may be used in future to identify dis-

urbed fetal development.

lacental Perfusion by Fetal MRIormal function of the uteroplacental unit is critical for fetalrowth and development. Therefore, reliable techniques forvaluating dynamic placental function are of major impor-ance for the assessment and management of high-risk preg-ancies, such as those complicated by pre-eclampsia and

ntrauterine growth restriction.59-61 To date, the mainstayechnique in the field has been Doppler ultrasound. How-ver, MRI has several potential advantages over Doppler ul-rasound in that it is not dependent on adequate amnioticuid volume or affected by maternal obesity or a posteriorlyositioned placenta.62

Several features of the placenta, including its relative im-obility, high blood volume (about 50% of placental vol-me), and high rate of perfusion, make it particularly ame-able in theory to MRI perfusion studies. Despite the

heoretical ease of placenta imaging, to date, few MR perfu-ion imaging studies have been carried out.

Perfusion imaging techniques depend on the availability ofcontrast agent to track microscopic perfusion exchanges63

nd a detection system with high temporal resolution, as theracer may be visible only briefly. In animal models, it haseen shown that MRI perfusion techniques provide a robusteasure of placental blood flow between the fetus and theother.64 In these animal studies, microcirculatory MRI per-

usion studies have become possible because of the develop-ent of new contrast agents with specific biodistribution and

aster MRI acquisition sequences.64-68 Data from these exper-mental studies have enabled the development of new modelsf placental perfusion and permeability in vivo.64,69,70 Despitehe promising results provided by the use of MRI contrastgents in experimental studies, the extent to which contrastgents cross the placenta and their rate of clearance remainsnclear, thus limiting their application for clinical evaluationf placental perfusion. Consequently, other methods fortudying placental perfusion by MRI techniques have beenxplored.

Arterial spin labeling is an MRI technique widely used toeasure perfusion. Specifically, the flow-sensitive alternat-

ng inversion recovery technique compares the rate of recov-ry of the magnetization following nonselective and selectivenversion pulses. In studies testing this approach in bothealthy and high-risk pregnant women,71 it was found thatlthough average global placental perfusion was similar in thegroups, regions of decreased perfusion were described in

he placentas of growth-restricted fetuses. Another MRI-ased approach that has been applied to measurement oflacental perfusion is the intravoxel incoherent motionIVIM) technique. The IVIM technique was first developed toeasure the perfusion (or blood volume) in capillary net-

orks by measuring signal attenuation caused by phase dis-

pfiwvpmItpstegbfiav

smcws

DVAibeddpt

sitepstbp(2wwaona

vhoaibswtttatvbmu

Advancing fetal brain MRI 295

ersion of randomly moving water protons when a magneticeld gradient is applied.72 Water moving in a capillary net-ork can be modeled as a fraction of the voxel that is diffusedery rapidly. Therefore, the volume of moving blood in thelacenta can be measured as that fraction of water that isoving fast enough to be dephased during acquisition of

VIM sequence.72 Preliminary studies applying the IVIMechnique to measure placental perfusion have providedromising results.73 Specifically, this technique has demon-trated a band of high moving-blood fraction on the side ofhe placenta adjacent to the uterine wall.73 In cases of pre-clampsia or fetal growth retardation, IVIM studies in lateestation have shown reduced perfusion of the placentalasal plate. Conversely, the fraction of moving blood on theetal side of the placenta is increased in fetuses with intrauter-ne growth restriction, changes ascribed by the authors to anlteration of the villous structure and a regional decrease inascular resistance.

Although the application of MR imaging techniques fortudying placental function are promising, substantiallyore longitudinal data with these techniques are needed to

haracterize normal placental perfusion, and to determinehether early changes in placental perfusion are predictive of

ubsequent fetal compromise.

evelopment of an Inivo 3-D Fetal MRI Brain Atlaslthough brain atlases are widely used for image processing

n adult MRI, no such atlas is currently available for the fetalrain. The development of a 3-D MRI fetal brain atlas wouldxpedite our understanding of normal and aberrant in vivoevelopment of the fetal brain. The fundamental steps inevelopment of an MRI brain atlas involve building a tem-late from the statistical analysis of multiple brain studies inhe population of interest.74,75 The basic principles would be

Figure 7 The development of a high-resolution in vivo(mean gestational age of 27 weeks). The bottom row

weeks).

imilar to those used for the creation of adult brain atlases,ncluding brain segmentation, registration, and averagingechniques of a set of fetal brain volumes to a common refer-nce. However, there would be important differences in ap-roach extending beyond the technical considerations de-cribed earlier. Perhaps the most fundamental difference ishe dynamic anatomic changes that characterize normalrain development. In the early second trimester, for exam-le, the surface of the cerebral cortex is normally smoothlissencephalic) with primary sulci emerging between 18 and0 weeks’ gestation and secondary sulci around 30 to 32eeks. Consequently, a meaningful fetal brain MRI atlasould actually consist of multiple gestational age-dependent

tlases that reflect these and multiple other evolving featuresf the developing brain. This is no trivial challenge given theeed for a large-scale normative fetal MRI database that spanscross all gestational ages.

Recent innovations in fetal image reconstruction using ad-anced registration methods to correct for fetal motion5,76,77

ave allowed acquisition of high-resolution 3D-MRI imagesf the fetal brain, and will facilitate the creation of a fetal braintlas. Development of such a 3D-MRI fetal brain atlas extend-ng from 25 to 37 weeks of gestation is in preparation,78

ased on a large normative database of nonsedated fetal MRItudies (Fig. 7). The availability of a 3-D fetal MRI brain atlasill assist in the development of advanced post-processing

echniques for fetal volumetric brain reconstruction and au-omatic volume rendering of fetal brain structures and tissueypes (e.g., cortical grey matter, unmyelinated and myelin-ted white matter, germinal matrix, cerebrospinal fluid) forhe future study of regional tissue-specific in utero brain de-elopment. In addition to volumetric studies of the majorrain structures, this approach may ultimately allow volumeeasurements of tissue subtypes (eg, cortical gray matter,nmyelinated and myelinated white matter, cerebrospinal

RI brain atlas. The top row represents the first atlastes the second fetal atlas (mean gestational age of 33

fetal Millustra

flsvfimrtmfcttov

SFisaimqfottmlfsc

rit

driwwbwma

R

1

1

1

w show

296 C. Limperopoulos and C. Clouchoux

uid) in specific parcellated brain regions, facilitating thetudy of regional tissue-specific brain development. The de-elopmental schedule for gyral formation in the developingetus has been well defined through both pathology and qual-tative MRI studies.78-80 Quantitative techniques for measure-

ent of cortical folding previously described in adults, haveecently been applied to preterm infants ex utero.81 An ex-ension of this approach to the study of fetal sulcal develop-ent is currently underway, using a novel methodological

ramework for fetal cortical surface rendering (Fig. 8) (Clou-houx et al, unpublished data). This approach is based on aechnique previously validated in adults using a 2-D mesh ofhe cortical surface brains.82-85 Other potential future devel-pments might include cortical thickness measurement andolume-based morphometry.

ummary and Future Directionsetal MRI is rapidly becoming a powerful tool for investigat-

ng in vivo fetal brain development in both clinical and re-earch studies. To date, our understanding of the dynamicnd highly intricate evolution of the fetal brain has beenmpeded by a lack of standardized and reliable MRI bench-

arks. The recent development and ongoing refinement ofuantitative 3-D volumetric MRI, DTI, MR spectroscopy, andunctional and perfusion imaging now provide us with thepportunity to evaluate the fetal brain from a complemen-ary, integrated perspective to include volume, microstruc-ure, metabolism, and function, respectively. These timelyethodological developments now provide us with an un-

imited potential to study in unprecedented detail normaletal brain development as well as the mechanisms and con-equences that underlie impaired brain development in theompromised fetus.

Advancing the role and accuracy of MRI in the fetus willequire ongoing methodological developments in the areas ofmage acquisition and advanced postprocessing techniques

Figure 8 Extracted cortical surface from a 28-week gehemispheres, viewed from 3 different angles. Bottom ro

o overcome current challenges of fetal motion and image 1

egradation. Moreover, the creation of a large-scale high-esolution 3-D in vivo fetal brain atlas will permit greaternsight into neuronal and axonal pathway development, asell as sulcal and gyral formation, and provide a uniqueindow on the timing of insults that might disrupt normalrain development. Advances expected to result from thisork will greatly improve our ability to reliably detect andonitor the high-risk fetus, improve parental counseling,

nd evaluate the potential benefits of prenatal interventions.

eferences1. Garel C: Fetal MRI: What is the future? Ultrasound Obstet Gynecol

31:123-128, 20082. Prayer D: Investigation of the normal organ development with fetal

MRI. Eur Radiol 17:2458-2471, 20063. Jenkinson M, Smith S: Improved optimization for the robust an accu-

rate linear registration and motion correction of brain images. Neuro-image 17:825-841, 2002

4. Jiang S, Xue H, Counsell S, et al: In-utero three dimension high reso-lution fetal brain diffusion tensor imaging. MICCAI 10:18-26, 2007

5. Rousseau F, Glenn OA, Iordanova B, et al: Registration-based approachfor reconstruction of high-resolution in utero fetal MR brain images.Acad Radiol 13:1072-1081, 2006

6. Garel C, Delezoide: MRI of the Fetal Brain: Normal Development andCerebral Pathologies. New York, NY, Springer, 2004

7. Girard N, Gire C, Sigaudy S, et al: MR imaging of acquired fetal braindisorders. Childs Nerv Syst 19:490-500, 2003

8. Limperopoulos C, Robertson RL, Estroff JA, et al: Diagnosis of inferiorvermian hypoplasia by fetal magnetic resonance imaging: Potential pit-falls and neurodevelopmental outcome. Am J Obstet Gynecol 194:1070-1076, 2006

9. Limperopoulos C, Robertson RL Jr, Khwaja OS, et al: How accuratelydoes current fetal imaging identify posterior fossa anomalies? Am JRoentgenol 190:1637-1643, 2008

0. Sjogren B: Psychological indications for prenatal diagnosis. Prenat Di-agn 16:449-454, 1996

1. Manjon JV, Guizard N, Robles M, et al: Reconstruction using 3D in-painting. Hum Brain Mapp (in press)

2. Limperopoulos C, Soul JS, Gauvreau K, et al: Late gestation cerebellargrowth is rapid and impeded by premature birth. Pediatrics 115:688-95, 2005

al age fetus. Top row shows right and left cerebrals the left hemisphere (external and internal surfaces).

station

3. Hüppi PS, Warfield S, Kikinis R: Quantitative magnetic resonance im-

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4

4

4

4

4

5

5

5

5

5

5

5

5

5

5

6

6

6

6

6

6

Advancing fetal brain MRI 297

aging of brain development in premature and mature newborns. AnnNeurol 43:224-235, 1998

4. Kubik-Huch RA, Wildermuth S, Cettuzzi L, et al: Fetus and Uteropla-cental unit: Fast MR imaging with three-dimensional reconstructionand volumetry—Feasibility study. Radiology 219:567-73, 2001

5. Schierlitz L, Dumanli H, Robinson JN, et al: In-utero three-dimensionalmagnetic resonance imaging of fetal brains. Lancet 357:1177-1178,2001

6. Grossman R, Hoffman C, Mardor Y, et al: Quantitative MRI measure-ments of human fetal brain development in utero. Neuroimage 33:463-470, 2006

7. Kazan-Tannus JF, Dialani V, Kataoka ML, et al: MR volumetry of brainand CSF in fetuses referred for ventriculomegaly. Am J Roentgenol189:145-151, 2007

8. Limperopoulos C, Tworetzky W, Newburger JW, et al: Third-trimestervolumetric brain growth is impaired in fetuses with congenital heartdisease. Circulation 118:S908, 2008

9. Guizard N, Evans AC, Lepage C, et al: Automatic model-based fetalbrain parcellation to quantify in vivo fetal brain development. HumBrain Mapp (in press)

0. Mukherjee P, Miller JH, Shimony JS, et al: Diffusion-tensor MR imagingof gray matter and white matter development during normal humanmaturation. Am J Neuroradiol 23:1445-1456, 2002

1. Anjari M, Srinivasan L, Allsop JM, et al: Diffusion tensor imaging withtract-based spatial statistics reveals local white matter abnormalities inpreterm infants. Neuroimage 35:1021-1027, 2007

2. Dubois J, Dehaene-Lambertz G, Soarès C, et al: Microstructural corre-lates of infant functional development: Example of the visual pathways.J Neurosci 28:1943-1948, 2008

3. Stadnik T, Luypaert R, Jager T, et al: Diffusion imaging: From basicphysics to practical imaging. RSNA EJ Radiog 3, 1999

4. Prayer D, Prayer L: Diffusion-weighted magnetic resonance imaging ofcerebral white matter development. Eur J Radiol 45:235-243, 2003

5. Garel C: New advances in fetal MR neuroimaging. Pediatr Radiol 36:621-625, 2006

6. Baldoli C, Righini A, Parazzini C, et al: Demonstration of acute ischemiclesions in the fetal brain by diffusion magnetic resonance imaging. AnnNeurol 52:243-246, 2002

7. Righini A, Bianchini E, Parazzini C, et al: Apparent diffusion coefficientdetermination in normal fetal brain: A prenatal MR imaging study. Am JNeuroradiol 24:799-804, 2003

8. Agid R, Lieberman S, Nadjari M, et al: Prenatal MR diffusion-weightedimaging in a fetus with hemimegalencephaly. Pediatr Radiol 36:138-140, 2006

9. Bui T, Daire JL, Chalard F, et al: Microstructural development of humanbrain assessed in utero by diffusion tensor imaging. Pediatri Radiol36:1133-1140, 2006

0. Kim DH, Chung S, Vigneron DB, et al: Diffusion-weighted imaging ofthe fetal brain in vivo. Magn Reson Med 59:216-220, 2008

1. Brunel H, Girard N, Confort-Gouny S, et al: Fetal brain injury. J Neu-roradiol 31:123-137, 2004

2. Erdem G, Celik O, Hascalik S, et al: Diffusion-weighted imaging eval-uation of subtle cerebral microstructural changes in intrauterine fetalhydrocephalus. Magn Reson Imaging 25:1417-1422, 2007

3. Le Bihan D: Molecular diffusion nuclear magnetic resonance imaging.Magn Reson Q 7:1-30, 1991

4. Hermoye L, Saint Martin C, Cosnard G, et al: Pediatric diffusion tensorimaging: Normal database and observation of the white matter matu-ration in early childhood. Neuroimage 29:493-504, 2005

5. Kasprian G, Brugger PC, Weber M, et al: In utero tractography of fetalwhite matter development. Neuroimage 43 213-24, 2008

6. Kok RD, van der Bergh AJ, Heerschap A, et al: Metabolic informationfrom the human fetal brain obtained with proton magnetic resonancespectroscopy. Am J Obstet Gynecol 185:1011-1015, 2001

7. Kok RD, van der Berg PP, van der Bergh AJ, et al: Maturation of thehuman fetal brain as observed by 1H MR spectroscopy. Magn ResonMed 48:611-616, 2002

8. Heerschap A, Kok RD, van der berg PP: Antenatal proton MR spectros-

copy of the human brain in vivo. Childs Nerv Syst 19:418-421

9. Kreis R: Brain metabolite composition during early human brain devel-opment as measured by quantitative in vivo 1H magnetic resonancespectroscopy. Magn Reson Med 48:949-958, 2002

0. Girard N, Gouny SC, Viola A, et al: Assessment of normal fetal brainmaturation in utero by proton magnetic resonance spectroscopy. MagnReson Med 56:768-775, 2006

1. Roelant-Van Rijn AM, Groenendaal F, Stoutenbeek P, et al: Lactate inthe fetal brain: Detection and implications. Acta Pædiatr 93:937-940,2004

2. Van der Voorn JP, Pouwels PJ, Hart AA, et al: Childhood white matterdisorders: Quantitative MR imaging and spectroscopy. Radiology 241:510-517, 2006

3. Fenton BW, Lin CS, Macedonia C, et al: The fetus at term: In uterovolume-selected proton MR spectroscopy with breath-hold tech-nique—A feasibility study. Radiology 219:563-566, 2001

4. Girard N, Fogliarini C, Viola A, et al: MRS of normal and impaired fetalbrain development. Eur J Radiol 57:217-225, 2006

5. Borowska-Matwiejczuk K, Lemancewicz A, Tarasow E, et al: Assess-ment of fetal distress based on magnetic resonance examinations: Pre-liminary report. Acad Radiol 10:1274-1282, 2003

6. Wolfberg AJ, Robinson JN, Mulkern R, et al: Identification of fetalcerebral lactate using magnetic resonance spectroscopy. Am J ObstetGynecol 196:e9-e11, 2007

7. Azpurua H, Alvarado A, Mayobre F, et al: Metabolic assessment of thebrain using proton magnetic resonance spectroscopy in a growth-re-stricted human fetus: Case report. Am J Perinatol 25:305-309, 2008

8. Limperopoulos C, Tworetzky W, Robertson RL, et al: Impaired brainmetabolism in fetuses with congenital heart disease. Circulation 118:S651-S-652, 2008

9. Hirsch J, Kim KHS, Souweidane MM, et al: fMRI reveals a developinglanguage system in a 15-month old sedated infant (Abstract). Soc Neu-rosci 23: 222, 1997

0. Huettel SA, Song AW, McCarthy G: Functional Magnetic ResonanceImaging. Sunderland, MA, Sinauer Associates, 2004

1. Iannetti GD, Wise RG: BOLD functional MRI in disease and pharma-cological studies: Room for improvement? Magn Reson Imaging 25:978-988, 2007

2. Gowland P, Fulford J: Initial experiences of performing fetal fMRI. ExpNeurol 190 (suppl 1):S22-S27, 2004

3. Fulford J, Vadeyar SH, Dodampahala SH, et al: Fetal brain activity andhemodynamic response to a vibroacoustic stimulus. Hum Brain Mapp22:116-121, 2004

4. Fulford J, Vadeyar SH, Dodampahala SH, et al: Fetal brain activity inresponse to a visual stimulus. Hum Brain Mapp 20:239-245, 2003

5. Sparling JW: Concepts in Fetal Movement Research. New York, Ha-worth Press, 1993

6. Olesen AG, Svare JA: Decreased fetal movements: Background, assess-ment, and clinical management. Acta Obstet Gynecol Scand 83:818-826, 2004

7. Prechtl HF, Einspieler C: Is neurological assessment of the fetus possi-ble? Eur J Obstet Gynecol Reprod Biol 75:81-84, 1997

8. Sparling JW, Van Tol J, Chescheir NC: Fetal and neonatal hand move-ment. Phys Ther 79:24-39, 1999

9. Gagnon R: Placental insufficiency and its consequences. Eur J ObstetGynecol Reprod Biol 110 (suppl 1):99-107, 2003

0. Mihaly GW, Morgan DJ: Placental drug transfer: Effects of gestationalage and species. Pharmacol Ther 23:253-266, 1983

1. Polliotti BM, Holmes R, Cornish JD, et al: Long-term dual perfusion ofisolated human placental lobules with improved oxygenation for infec-tious diseases research. Placenta 17:57-68, 1996

2. Duncan KR: Fetal and placental volumetric and functional analysisusing echo-planar imaging. Top in Magn Res Imaging 12:52-66, 2001

3. Le Bars E, Gondry-Jouet C, Deramond H, et al: MR diffusion andperfusion imaging in clinical practice. J Neurorad 57:56-76, 2000

4. Salomon LJ, Siauve N, Balvay D, et al: Placental perfusion MR imagingwith contrast agents in a mouse model. Radiology 235:73-80, 2005

5. Zhu XP, Li KL, Kamaly-Asl ID, et al: Quantification of endothelial

permeability, leakage space, and blood volume in brain tumors using

6

6

6

6

7

7

7

7

7

7

7

7

7

7

8

8

8

8

8

8

8

298 C. Limperopoulos and C. Clouchoux

combined T1 and T2* contrast-enhanced dynamic MR imaging. J MagnReson Imaging 11:575-585, 2000

6. Brasch R, Turetschek K: MRI characterization of tumors and gradingangiogenesis using macromolecular contrast media: Status report. EurJ Radiol 34:148-155, 2000

7. Vallee JP, Sostman HD, MacFall JR, et al: Quantification of myocardialperfusion with MRI and exogenous contrast agents. Cardiology 88:90-105, 1997

8. Pradel C, Siauve N, Bruneteau G, et al: Reduced capillary perfusion andpermeability in human tumour xenografts treated with the VEGF sig-naling inhibitor ZD4190: An in vivo assessment using dynamic MRimaging and macromolecular contrast media. Magn Reson Imaging21:845-851, 2003

9. Roberts JM, Lain KY: Recent insights into the pathogenesis of pre-eclampsia. Placenta 23:359-372, 2002

0. Zygmunt M, Herr F, Munstedt K, et al: Angiogenesis and vasculogen-esis in pregnancy. Eur J Obstet Gynecol Reprod Biol 110 (suppl 1):10-18, 2003

1. Gowland PA, Francis ST, Duncan KR, et al: In vivo perfusion measure-ments in the human placenta using echo planar imaging at 0.5T. MagnReson Med 40:467-473, 1998

2. Gowland PA: Placental MRI. Seminars in fetal and neonatal. Medicine10:485-490, 2005

3. Moore RJ, Ong SS, Tyler DJ, et al: Spiral artery blood volume in normalpregnancies and those compromised by pre-eclampsia. NMR Biomed21:376-380, 2008

4. Evans A, Collins D, Mills S, et al: 3rd statistical neuranatomical modelsfrom 305 MRI volumes. Proc. IEEE Nuclear Science Symposium and Med-

ical Imaging Conference 1813-1817, 1993

5. Kochunov P, Lancaster J, Thompson P, et al: An optimised individualtarget brain in the Talairach coordinates system. Neuroimage 17:922-927, 2002

6. Rousseau F, Glenn O, Iordanova B, et al: A novel approach to highresolution fetal brain MR imaging. MICCAI 8:548-555, 2005

7. Guizrad N, Lepage C, Fonov V, et al: Development of fetus brain atlasfrom multi-axial MR acquisitions. ISMRM. (in press)

8. Rados M, Judas M, Kostovic I: In vitro MRI of brain development. EurJ Radiol 57:187-198, 2006

9. Perkins L, Hughes E, Srinivasan L, et al: Exploring cortical subplateevolution using magnetic resonance imaging of the fetal brain. DevNeurosci 30:211-220, 2008

0. Salomon LJ, Garel C: Magnetic resonance imaging examination of thefetal brain. Ultrasound Obstet Gynecol 30:1019-1032, 2007

1. Dubois J, Benders M, Cachia A, et al: Mapping the early cortical foldingprocess in the preterm newborn brain. Cereb Cor Advanc 18:1444-1454, 2007

2. Van Essen D, Drury H, Joshi S, et al: Functional and structural mappingof human cerebral cortex: Solution are in the surfaces. Proc Natl AcadSci USA 95:788-795, 1998

3. Fischl B, Sereno M, Tootell R, et al: Cortical surface-based analysis. II.Segmentation and surface reconstruction. Neuroimage 9:179-194,1999

4. Toro R, Burnod Y: Geometric atlas: Modeling the cortex as an organizedsurface. Neuroimage 20:1468-1484, 2003

5. Clouchoux C, Coulon O, Rivière D, et al, LNCS: Anatomically con-strained surface parameterization for cortical localization. MICCAI3750:344-35, 2005

6. Schneider JF, Confort-Gouny S, Le Fur Y, et al: Diffusion-weighted

imaging in normal fetal maturation. Eur Radiol 17:2422-2429, 2007