usg

DOI: 10.1542/neo.2-1-e3 2001;2;3 NeoReviews

William D. Rhine and Francis G. Blankenberg Cranial Ultrasonography

http://neoreviews.aappublications.org/cgi/content/full/neoreviews;2/1/e3located on the World Wide Web at:

The online version of this article, along with updated information and services, is

Online ISSN: 1526-9906. Illinois, 60007. Copyright 2001 by the American Academy of Pediatrics. All rights reserved. by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village,it has been published continuously since 2000. NeoReviews is owned, published, and trademarked NeoReviews is the official journal of the American Academy of Pediatrics. A monthly publication,

by JoDee Anderson on January 21, 2008 http://neoreviews.aappublications.orgDownloaded from

Cranial UltrasonographyWilliam D. Rhine, MD,*

and Francis G.

Blankenberg, MD

Objectives After completing this article, readers should be able to:1. Describe the difference in relative reflectance of various structures in the brain.2. Describe common imaging approaches for cranial ultrasonography.3. Explain uses of the resistive index and pulsatility index in Doppler flow studies.4. Describe the findings on cranial ultrasonography of intracranial hemorrhage,

periventricular leukomalacia, and hydrocephalus.

IntroductionA relatively unique modality of neuroimaging is available to newborns in the form ofcranial ultrasonography (CUS). Technologic advances and ever-increasing experience withobtaining and interpreting CUS images have led to its widespread acceptance, yieldingvaluable insight into the anatomy of the preterm and term newborn brain.

Physics and Methods of CUSThe fontanelles in the newborn provide unique windows for ultrasonographic examinationof the neonatal brain. These gaps between the calvarial bones serve as acoustic windowsthrough which sound waves of the ultrasound probe can be transmitted and received. Thepresence of air or bony structures in the path of the sound beam will degrade any echosignal emanating from these regions. CUS employs frequencies between 8 and 15 MHz toyield echoreflective images.

Following application of acoustic gel over the fontanelle of interest, a small (1 to 2 cm)solid-state transceiver probe is placed to obtain an adequate impedance match with thescalp. The resultant signal is processed by a computer to yield a real-time wedge orfan-shaped image that has a working focusing depth of 2 and 12 cm for a standard 8 MHzsector probe with a maximal resolution of approximately 1 to 2 mm. For higher resolution,

increased sensitivity to flow, and better near-field focusing(several millimeters to 8 cm), linear phased-array probesdesigned with higher frequency ranges (8 to 15 MHz) and alarger number of detectors per unit area can be used. Incontrast to sector probes, phase-array probes generaterectangular-shaped images of the underlying brain. How-ever, because of their length (5 to 6 cm), which is longer thanthe largest fontanelle, views are obtained of only the midlineor slightly off-midline structures. Further, there is a relativelack of tissue penetration at higher frequencies (ie, the higherthe frequency, the less penetration of the sound beam).

Tissue ReflectanceThe relative reflectance of tissue in the region of interestprovides the imaging contrast to distinguish various intrace-rebral structures (eg, the cerebral and cerebellar sulci), mid-line structures (eg, the cavum septum pellucidum), third andfourth ventricles, and the posterior fossa. In addition, thereflectance allows detection of pathology such as hemor-

*Associate Professor of Pediatrics.Assistant Professor of Radiology, Stanford University, Palo Alto, CA.

Abbreciations

CBF: cerebral blood flowCNS: central nervous systemCSF: cerebrospinal fluidCT: computed tomographyCUS: cranial ultrasonographyHIE: hypoxic-ischemic encephalopathyICH: intracranial hemorrhageIVH: intraventricular hemorrhageMRI: magnetic resonance imagingPHH: posthemorrhagic hydrocephalusPI: pulsatility indexPVHI: periventricular hemorrhagic venous infarctionPVL: periventricular leukomalaciaRI: resistive index

Article radiology

NeoReviews Vol.2 No.1 January 2001 3 by JoDee Anderson on January 21, 2008 http://neoreviews.aappublications.orgDownloaded from

rhage, ischemia, or hydrocephalus. Cerebrospinal fluid(CSF) without any hemorrhage is normally anechoic,comprising the darkest component of a CUS. Themost echogenic normal structures in the brain are thechoroid plexus and the germinal matrix, a structurenormally found in preterm infants. The choroid plexus,located posteriorly in the lateral ventricles and in the roofof the third ventricle, and the germinal matrix, located inthe caudothalamic groove of preterm neonates, are bothhighly reflective because of the millions of tiny interfaceswithin these dense tangles of capillaries. These interfacesare approximately the same size as the wavelength of thesound beam, which causes the beam to scatter in multipledirections, including back toward the transducer. Nor-mal gray and white matter are of medium echogenicity,with brightness somewhere between that of CSF and thechoroid plexus and germinal matrix.

Blood within CSF or the brain contains millions ofminiature mirrors in the form of clotted red bloodcells, which strongly reflect sound waves and, therefore,appear sonographically brighter than either CSF or nor-mal gray or white matter. Curiously, regions of ischemiain either the gray or white matter are also stronglyreflective, generating bright echoes. The reason for thisapparent anomaly is that regions of ischemia also containmicroscopic regions of red blood cells that have extrava-sated into the interstium.

Imaging ApproachesThe most commonly described method for CUS is theso-called anterior fontanelle approach in which theprobe is placed on the anterior fontanelle. The sonogra-pher slowly angles the probe from side to side and frontto back to obtain a sweep of the entire brain in at leasttwo different planes (ie, coronal and sagittal). It is impor-tant to angle the fan of the sector probe serially as farlateral, anterior, and posterior as possible to ensure thatthe lateral convexities of the cerebrum and the frontaland occipital-parietal regions are visualized adequately.Typically, images are viewed real-time during CUS andprints are made of sagittal and coronal views (Fig. 1).CUS images sometimes are displayed or printed withreverse black-white contrast (ie, nonechogenic CSF inthe ventricles is the whitest part of the image).

Other approaches are employed routinely to supple-ment the anterior fontanelle view, including the typicallysmall posterior fontanelle view and the so-called mas-toid view that is obtained through the smaller posterior-lateral fontanelles located immediately posterior to theear. The posterior fontanelle view provides excellentdetail of the occipital horns for assessing intraventricular

hemorrhage that may be layering posteriorly and difficultto distinguish from the echogenic choroid plexus in theanterior fontanelle view. The posterior fontanelle viewalso is helpful in assessing the echogenicity of theperiventricular white matter, which can be increasedartifactually in the anterior fontanelle view because ofmultiple crossings of the normal tiny medullary veins andarteries within the deep white matter exactly perpendic-ular to the sound beam. The mastoid view yields excel-lent images of the cerebellum, fourth ventricle, and theremainder of the posterior fossa, structures that may bedifficult to see from the more distant anterior fontanelle.

Figure 1. Sagittal and coronal views of a normal preterm brainusing the anterior fontanelle approach. Views were obtainedwith an 8 MHz solid-state sector probe. A. Sagittal notch viewshowing the normal smooth tapering of the usually thinnonbulbous echogenic germinal matrix as it courses anteriorlywithin the caudothalamic groove. B. Coronal view through theregion of the caudothalamic groove and third ventricle,showing the cavum septum pellucidum that contains anechoicCSF just below the corpus callosum (arrow). Below the cavumis the third ventricle and the foramen of Monro enteringbilaterally. The thin echogenic germinal matrix is seen withinthe caudothalamic groove bilaterally.

radiology cranial ultrasonography

4 NeoReviews Vol.2 No.1 January 2001 by JoDee Anderson on January 21, 2008 http://neoreviews.aappublications.orgDownloaded from

Doppler Flow StudiesDoppler measurements during CUS provide informationabout relative cerebral blood flow (CBF) of newborns.The Doppler signal is obtained by calculating the fre-quency shift of the returning sound beam reflected by thered blood cells flowing in a vessel of interest comparedwith that of the original pulse. Blood flowing away fromthe transducer has a negative shift less than the originalfrequency; blood flowing toward the transducer has apositive shift that is greater than the original sound beam.Doppler signals can be displayed as a continuous wave ofblood velocities on a scale obtained from a graphicallydetermined volume of interest that is electronically su-perimposed on the CUS images (ie, pulsed-wave Dopp-ler). The signals also can be displayed on a color scale(typically blue for blood flowing away from the trans-ducer and red for blood flowing toward the transducer)and electronically superimposed on the CUS images.Another display of Doppler signal is the power Dopplermode in which all flow, regardless of direction, issummed to provide a single color image of the magni-tude of blood flow. Although precise quantitation ofCBF currently is not possible, other calculated Dopplerflow indices have been correlated to neonatal brainpathology and neurodevelopmental outcome. Relativechanges in the measurements of flow velocities within theanterior or middle cerebral arteries correlate proportion-ately to changes in global and regional CBF.

One measure of cerebral vascular dynamics is theresistive index (RI or Pourcelot index), which is:

(peak systolic velocity end diastolic velocity)/peak systolic velocity

RI measurements from the anterior cerebral artery arenormally slightly higher in preterm infants than terminfants, averaging 0.80 (range, 0.5 to 1.0) and 0.71(range, 0.6 to 0.8), respectively. Increases in intracranialpressure caused by hypoxic-ischemic encephalopathy,brain edema, or hydrocephalus are accompanied by ele-vations in RI. When intracranial pressure exceeds that ofthe arterial perfusion pressure, causing brain death, thediastolic flow is reversed, and eventually peak systolicflow is dampened. These Doppler flow findings havebeen used to support the difficult diagnosis of braindeath in infants.

To improve specificity for the diagnosis of hydroceph-alus, some institutions recommend serial measurementsof RI with the ultrasound probe compressing the ante-rior fontanelle (,5 sec pressure). The percent change inthe RI with compression over time correlates with directCSF pressure measurements within the ventricles. Othermarkers of CBF include measurements of area under the

velocity curve and velocity amplitudes (systolic, diastolic,mean velocity). The variation of blood flow velocitiesduring the cardiac cycle can be summarized by the cal-culation of the pulsatility index (PI):

(peak systolic velocity 2 end diastolic velocity)/mean velocity

Increased PI has been associated with the presence ofa patent ductus arteriosis and may increase the risk fordevelopment of intraventricular hemorrhage (IVH).

Another pathophysiologic phenomenon that can bedetected by serial Doppler measurements in the first fewdays of life is impaired cerebrovascular autoregulationfollowing hypoxic-ischemic injury. Preterm infants whootherwise have the transient ability to autoregulate re-gional vasomotor tone and vascular cross-sectional areaslose these automonic functions, particularly in watershedregions such as the periventricular white matter. The lossof autoregulation allows for pressure-passive flow inwhich the vascular bed is unable to constrict appropri-ately in response to increases in mean systemic arterialpressure or dilate in response to falls in mean arterialpressure. Pressure-passive flow in preterm infants whohave normal mean arterial pressures causes a significantrise in peak velocity within the thalamostriate vessels (thesurrogate markers of the perfusion of the periventricularwhite matter in preterm infants) and the vascular bedcross-sectional area. Both of these findings are associatedwith an increased risk for the development of intracranialhemorrhage (ICH) and periventricular leukomalacia(PVL).

Doppler CUS studies also have been used to examinethe effects of therapies on neonatal CBF. One studysuggested that sampling from high-lying umbilical arte-rial catheters had a more adverse effect on CBF velocitythan sampling from low-lying catheters. However, nosignificant differences in the rate of IVH were found in alarge prospective, randomized trial examining umbilicalarterial catheter positioning.

Observable PathologyIntracranial Hemorrhage

The most common use of CUS in the neonatal intensivecare unit is to detect the presence and evolution of ICH,especially in the preterm newborn. The incidence of ICHvaries most significantly with gestational age and birth-weight. Extremely low-birthweight newborns (gesta-tional age ,27 wk, birthweight ,750 g) are estimated tohave ICH at a rate of 10 to 30 per 100 survivingnewborns. The incidence of ICH decreases with increas-ing gestational age and is quite rare in newborns of



greater than 32 weeks gestational age or 1,500 gbirthweight.

Although the precise mechanisms leading to the evo-lution of ICH remain obscure, it is generally felt that thepreterm brain is at highest risk for ICH for a variety ofreasons. These include: relatively late development of theend-arterial circulation during the last trimester; rela-tively decreased mesenchymal support for this vascularbed, particularly in the region of the germinal matrix(which normally does not involute completely until afterthe 33rd week of gestation); and decreased autoregula-tion of CBF, with the extremely preterm infant havingvirtual pressure-passive CBF. The rare term newbornwho has ICH usually has some predisposing factor, suchas a history of asphyxia or coagulation abnormalities thatmay be intrinsic or due to anticoagulation therapy such asthat used during extracorporeal membrane oxygenation.

The most widely used classification system for ICH isthat originally described by Papile and associates, whichgrades from 1 to 4 with increasing severity.

Grade 1 occasionally is referred to as a germinal matrixor subependymal hemorrhage. This subset of ICH isseen on CUS as an abnormally increased number ofechoes in the caudothalamic groove (ie, notch) in theexpected location of the germinal matrix. Normally thegerminal matrix echoes taper smoothly as they courseanteriorally in the caudothalamic groove. Because thegerminal matrix never is located anterior to the foramenof Monro, echoes seen anterior to the foramen of Monrowithout tapering as they course anteriorally in the cau-dothalamic groove indicate hemorrhage.

Grade 2 (Fig. 2) describes extension of a germinalmatrix/subependymal hemorrhage into the ventricleswithout any ventricular enlargement. Detection of anIVH that is echogenic in the first several days after ahemorrhage can be confused easily with normal choroidplexus (which normally can be amorphous and has mul-tiple echogenic lumps). The observation of echogenicmaterial within the occipital horn (posterior to the calcaravis) is diagnostic of IVH because the choroid plexusdoes not extend into the occipital lobe. The observationof an echogenic blood-CSF fluid level with the ventricu-lar system is also diagnostic of IVH. The rare primarychoroid plexus hemorrhage, 95% of which occur in termneonates, is considered as Grade 2 for purposes ofprognostication.

Grade 3 (Fig. 3) has blood extending into the ventri-cles and causing ventriculomegaly at the time of theinitial observation of IVH. IVH without accompanyinghydrocephalus within the first 24 hours after detection is

Figure 2. Grade 2 germinal matrix hemorrhage in a preterminfant on the third day of life. Views obtained with an 8 MHzsolid-state sector probe through the anterior fontanelle. A.The sagittal view demonstrates the echogenic bulbous collec-tion of blood that bears no resemblance to the normalgerminal matrix that tapers as it courses anteriorally in thecaudothalamic groove and also never is seen anterior to theforamen of Monro. B. Coronal view, showing a bulbousechogenic collection of blood in the left caudothalamicgroove. C. A sagittal view through the anterior fontanelle thatis angled slightly more posteriorly shows an echogenic clotfilling the occipital horn posterior to the calcar avis. Thechoroid plexus never is seen in the occipital horn.



considered as a Grade 2 hemorrhage even if the infantdevelops posthemorrhagic hydrocephalus (PHH) at alater date.

Grade 4 (Fig. 4) describes a germinal matrix hemor-rhage that dissects and extends into the adjacent brainparenchyma, irrespective of the presence or absence ofIVH. It is also referred to as an intraparenchymal hem-orrhage (IPH) when found elsewhere in the paren-chyma. Bleeding extending into the periventricular whitematter in association with an ipsilateral IPH has beenclassified as periventricular hemorrhagic venous infarc-tion (PHVI).

It is useful to note that 70% of all germinal matrixhemorrhages in preterm infants occur by days 3 and 4 oflife; 90% occur within the first week of life. Once diag-nosed, 90% of germinal matrix hemorrhages do not

progress in severity after 24 hours. Hemorrhages occur-ring 1 week or more after birth almost invariably areGrades 1 or 2 and have minor to no clinical significance.

The final reading of a CUS study for ICH shouldinclude a narrative commentary beyond a specific grad-ing that includes a description of both hemispheres and acomparison to previous studies. For Grade 2 IVH, itmay be helpful to know the volume of the ventriclesthat is occupied by blood. Similarly, Grade 3 ICH mayhave minimal to massive ventriculomegaly, which is im-portant to know when deciding about possible surgicalintervention.

The severity of ICH, as reflected by the Papile or othergrading systems, generally correlates with increasing riskof long-term neurodevelopmental abnormalities. Pre-term infants who suffer Grade 1 and 2 germinal matrix

Figure 3. Grade 3 germinal matrix hemorrhage 3 and 10 days after birth. Views are obtained with an 8 MHz solid-state sector probethrough the anterior fontanelle. A. On day 3 of life, the coronal view demonstrates massive bilateral intraventricular and germinalmatrix hemorrhage with ventricular dilation. B. The sagittal view confirms the presence of massive intraventricular and germinalmatrix hemorrhage. On day 10 of life, progressive posthemorrhagic hydrocephalus is evident on the coronal (C) and sagittal (D)views. The development of ependymitis is represented as increased echogenicity of the wall of the lateral ventricles. This transientphenomenon is seen several days after a hemorrhage and is an ependymal reaction to intraventricular heme that persists for at least2 to 3 weeks.



hemorrhages are at a minimally increased risk of neuro-developmental abnormalities compared with pretermneonates of the same size and gestation (a risk that still ishigher than that of term newborns). Grade 3 ICH has agreater risk for central nervous system (CNS) sequelaewhen hydrocephalus progresses to the point of requiringsurgical drainage or shunting. Grade 4 ICH that is severeand bilateral has the most guarded prognosis for normalneurodevelopmental outcome.

Recommendations for timing of CUS in high-riskvery low-birthweight newborns vary and depend greatlyupon the potential use of the information. ICH usuallybegins within the first 24 to 72 hours of life. Someauthors suggest that the initial examination be made at3 to 5 days of life; others recommend deferring it until7 to 10 days. If results of the examination are negative forICH, it is unlikely to find a hemorrhage thereafter.Although there are no definitive data about the extensionof ICH that theoretically could be caused by the effectsof indomethacin on platelets, some use early CUS to aidin the decision between medical and surgical interventionfor patent ductus arteriosus. Obtaining a final CUS closer todischarge (eg, 34 to 36 weeks gestation) also has beenrecommended to look for both ICH and PVL. However,recent studies suggest that magnetic resonance imaging(MRI) may be a more sensitive and specific measure of PVLand white matter injury in preterm infants.

Clinical circumstances may play an important role in

the timing of CUS. In the very unstable preterm infant,especially one who is experiencing an unexplainedhematocrit drop, acidosis, or a change in neurologicstatus, an earlier CUS may be indicated. The presence ofsevere ICH then may be incorporated in decisions aboutthe prolongation or escalation of intensive care support.If an initial CUS detects ICH, serial examinations may beperformed every 1 to 2 weeks until the ICH is believed tobe stable.

Periventricular Leukomalacia (PVL)PVL describes a characteristic pattern of white matterinjury found predominantly in preterm newborns, appar-ently as a response to hypoxic-ischemic insults. PVL isassociated with ICH, an association that increases withworsening ICH severity. However, PVL can arise with-out ICH and vice versa. On CUS, PVL initially presentsas numerous foci of increased periventricular echogenic-

Figure 4. Grade 4 germinal matrix hemorrhage on day 2 oflife. The view was obtained with an 8 MHz solid-state sectorprobe through the anterior fontanelle. Coronal view demon-strates a large germinal matrix hemorrhage extending directlyinto the brain parenchyma. Grade 4 hemorrhages graduallyinvolute over several months, resulting in porencephalic cystthat communicates with the ventricular system.

Figure 5. PVL in weeks 1 and 4 of life. Views were obtainedwith an 8 MHz solid-state sector probe through the anteriorfontanelle. A. Coronal view of the frontal lobe region dem-onstrates abnormally increased periventricular echogenicitybilaterally at week 1. B. Follow-up coronal view at week 4demonstrates cystic degeneration, involution of the periven-tricular white matter, and mild ventricular dilation.



ity adjacent to the lateral ventricles that become evidentwithin the first several days after the instigating insult(Fig. 5). Classically these periventricular regions undergocystic degeneration over the next 2 to 3 weeks, resultingin a swiss cheese pattern of white matter loss that canbe detected readily with CUS. The evolution of PVL alsocan be noncystic, manifested primarily by the irregularloss of the periventricular white matter and irregularventricular dilation. This form of PVL is more difficult todetect with CUS. After 2 to 3 months, both cystic andnoncystic PVL are characterized by variable degrees ofventricular dilation and irregular scalloping of the ven-tricular walls. At 2 to 4 months, MRI is the modality ofchoice for detection of PVL because it can discern notonly the morphologic changes due to PVL but abnormaltissue signals within the periventricular white matter,information that is not available from CUS.

PVL must be distinguished from PHVI, which mayhave a similar geographic distribution but subtle differ-ences in the evolution of echogenic changes. PVL usuallyhas echodensities that evolve into multiple smaller cyststhat do not communicate with the lateral ventricle. Bassand associates used these characteristics to classify 77% ofwhite matter lesions into either PVL or PVHI, with only11% having overlapping features and an equally smallsubgroup having echodensities without evolving cysticchanges. Neurodevelopmental deficits were observed inall groups, but patients who had PVHI had a normalmean developmental quotient. One large series foundPVL in 3.2% of infants whose birthweights were less than1,500 g, with most affected patients having benign clin-ical courses without identified risk factors for CNS insult.

Hypoxic-ischemic Encephalopathy (HIE)HIE in both preterm and term neonates may cause awide range of CNS injuries that may not be visible byCUS. In the term newborn, severe HIE can lead initiallyto generalized cerebral edema, including small, slit-likeventricles and poor gray-white signal differentiation onCUS. However, HIE may lead to severe neurologicinjury without any neuroimaging findings of edema.HIE also has been associated with focal, intraparenchy-mal hemorrhage, which also may be seen by CUS. Atleast one recent study of more than 100 asphyxiatednewborns was unable to correlate neonatal CUS findingswith the infants outcomes at 1 year of age.

A recent prospective study found that both computedtomography (CT) and MRI were vastly superior to CUSfor the detection of hypoxic-ischemic injury, particularlyin the cortical and subcortical gray matter of the lateralcerebral convexities, the most difficult regions to exam-

ine adequately by ultrasonography, especially via theanterior fontanelle approach.

InfectionCongenital or perinatal infections can lead to anatomicchanges that are detectable by CUS. CUS can detectsignificant infarction from meningitis and more subtlefindings, such as ventriculitis, which is seen as diffuselyabnormally echogenic ependyma within the walls of thelateral ventricles. CT and MRI with and without intrave-nous contrast are both vastly superior to CUS for thediagnosis of CNS infections and their complications.

Mineralizing lenticulostriate vasculopathy and paren-chymal and periventricular calcifications that may accom-pany congenital viral or toxoplasmosis infections oftenare sufficiently echogenic to be seen by CUS. However,in the asymptomatic newborn, isolated mineralizing len-ticulostriate vasculopathy (seen as highly echogenic ves-sels in the basal ganglia) without associated brain tissuecalcifications is much more likely to be an incidentalfinding not associated with congenital infection.

Hydrocephalus and Abnormal BrainDevelopment

Congenital and posthemorrhage hydrocephalus are seeneasily by CUS. Furthermore, CUS can be used to quan-titate the progression of hydrocephalus, which can aid inthe decision about neurosurgical intervention. One suchvolumetric analysis for following PHH has been de-scribed by Brann and coworkers. PHH tends to bemaximal at 6 to 8 weeks after birth. One third of thesecases resolve spontaneously over the next several weeks,one third stabilize, and one third progress and requireserial ventricular punctures or shunting.

Congenital hydrocephalus often is detected by fetalultrasonography, which may influence maternal deliverymanagement (eg, whether to transport to a tertiary cen-ter or whether a cesarean section is indicated). The mostcommon cause of congenital hydrocephalus is Arnold-Chiari malformation, which almost invariably is associ-ated with myelomeningocele and may have CUS findingsof distortion of the lateral ventricles, inferior displace-ment of the fourth ventricle, and obliteration of thecisterna magna due to a small posterior fossa.

CUS can detect other malformations of the develop-ing brain such as holoprosencephaly, lissencephaly, andschizencephaly. Narrowing or agenesis of the corpuscallosum can be suggested by CUS findings, but usuallyis confirmed by CT or MRI. CUS will demonstrate thelarger arteriovenous malformations, which exhibit re-markably abnormal flow studies by Doppler evaluation.



Cerebellar malformations, including vermian and cere-bellar hypoplasia, also can be seen by CUS. Arterio-venous malformations are detected most easily by CUSwhen accompanied by vascular abnormalities that can bedetected by Doppler flow studies. CUS also may find therare neonatal brain tumor or tuberous sclerosis, the latterhaving periventricular calcifications or echogenic hamar-tomas. However, the modality of choice for definitiveimaging evaluation of these and other central nervoussystem abnormalities is MRI.

LimitationsDespite being noninvasive and portable, certain technicalconsiderations limit clinical correlations with CUS find-ings and may require alternative neuroimaging modali-ties. More superficial structures beneath the skull, partic-ularly adjacent to the lateral cerebral convexities, havelimited resolution. For example, subarachnoid hemor-rhage routinely is missed on CUS, although massivesubarachnoid hemorrhage may appear as prominent,thickened echogenicity within the normally anechoicsubarachnoid space. Deeper structures of the brain, suchas the basal ganglia and brainstem, are not visualized wellby CUS, although supplemental views through the pos-terior and posterior-lateral fontanelles may alleviate thisdifficulty partially. More subtle developmental problems(eg, abnormal opercular development), especially ofmore lateral structures; neuronal migrational abnormal-ities; and gyral malformations are unlikely to be diag-nosed by CUS. These problems require cross-sectionalimaging with MRI or CT.

CUS generally has poorer tissue characterizationcapabilities than CT or MRI. For example, CUS willshow an echogenic lesion with mass effect for both IPHand severe focal infarction. In contrast, CT or MRI easilycan distinguish blood from the focal cerebral edemaassociated with ischemia or infarction. CUS is also sur-prisingly insensitive to global and brainstem ischemia andfocal cerebral infarction in the first 24 to 48 hours after anhypoxic-ischemic insult. CT and MRI are vastly superiorto CUS for visualizing ischemia/infarction in this timeperiod. These modalities also are superior in thefollow-up of complications of severe ischemia, such astranstentorial herniation (and other types of shifts due tomassive edema) and white and gray matter loss.

Several authors have tried to determine the relativeaccuracy of CUS findings in terms of neurodevelopmen-tal outcome or neuropathology. CUS shows a baselinerange of anatomic variants in 10% to 20% of healthy termnewborns. The more common variants associated with

normal neurologic outcome include choroid plexus, sub-ependymal cysts, cavum septum pellucidum, and mildventricular enlargement. CUS findings that include sig-nificant bilateral periventricular changes have the greatestcorrelation with adverse neurologic outcome (ie, cere-bral palsy). In one study comparing CUS results withpostmortem neuropathology, CUS correctly identifiedthe primary CNS insult in only 59% of patients, some-times due to imaging timing. However, in nearly 25% ofthe patients, CUS did not find the primary injury. Notevery CUS abnormality represents significant neurologicinjury, and technical limitations may lead to a significantfalse-negative predictive value for certain pathologies(eg, subarachnoid or subdural hemorrhage).

Summary and Future AdvancesAlthough CUS is used widely in the neonatal intensivecare unit, further advances may expand its utility to theclinician. Its portability and relative ease of use have madeit the standard of care for initial screening for ICH inpreterm infants. Clinicians in intensive care nurseriesneed to establish protocols for the appropriate and judi-cious use of CUS, recognizing that other neuroimagingtechniques may be needed to complement CUS findings.Besides preterm infants, neonates most likely to haveabnormal CUS findings include those who have knowncoagulopathy, seizures, macrocephaly, or other majordysmorphic findings. Improved CUS technology, withhigher frequencies and improved postacquisition pro-cessing, may provide increased resolution, especiallywhen combined with Doppler flow studies of the smallercerebral vasculature. Three-dimensional ultrasonogra-phy is being developed to evaluate both the fetus andnewborn, again improving the resolution with which wemay view the developing or injured brain. Further studiesneed to determine the correlation of CUS findings withclinical outcomes and interventions.

Suggested ReadingBarr LL. Neonatal cranial ultrasound. Radiol Clin North Am.

1999;37:11271146Bass WT, Jones MA, White LE, et al. Ultrasonographic differential

diagnosis and neurodevelopmental outcome of cerebral white mat-ter lesions in premature infants. J Perinatol. 1999;19:330336

Blankenberg FG, Loh N, Bracci P, et al. Sonography, CT, and MRimaging: a prospective comparison of neonates with suspectedintracranial ischemia and hemorrhage. Am J Neuroradiol. 2000;21:213218

Boo NY, Chandran V, Zulfiquar MA, et al. Early cranial ultrasound



changes as predictors of outcome during the first year of life interm infants with perinatal asphyxia. J Pediatr Child Health.2000;36:363369

Kuban K, Sanocka U, Leviton A, et al. White matter disorders ofprematurity: association with intraventricular hemorrhage andventriculomegaly. The Developmental Epidemiology Network.J Pediatr. 1999;134:539546

Perlman JM, Rollins N. Surveillance protocol for the detection ofintracranial abnormalities in premature neonates. Arch PediatrAdolesc Med. 2000;154:822836

Weiss HH, Goldtein RB, Piecuch RE. A critical review of cranialultrasounds: is there a closer association between intraventricu-lar blood, white matter abnormalities or cysts, and cerebralpalsy? Clin Pediatr. 1999;38:319323

NEOREVIEWS QUIZ

1. Cranial ultrasonography is performed through the anterior fontanelle using either a solid-state transceiverprobe or a linear phased-array probe. Of the following, the linear-phased array probe in comparison to thesolid-state transceiver probe is most likely to provide:

A. Decreased sensitivity to flow.B. Lower resolution.C. Poorer near-field focusing.D. Views of only midline structures.E. Wedge-shaped images.

2. A very low-birthweight neonate has an unexplained drop in hematocrit within 72 hours after birth. Cranialultrasonography reveals a germinal matrix hemorrhage. Of the following, the echnogenicity associated witha germinal matrix hemorrhage is most likely to be located:

A. In the caudothalamic groove.B. In the roof of the third ventricle.C. Posterior to the foramen of Monroe.D. Posteriorly in the lateral ventricle.E. Within the frontal horn.

3. Cranial ultrasonography, computed tomography, and magnetic resonance imaging are modalities used in thediagnosis of intracranial pathology in newborns. Of the following, the diagnosis most likely to be confirmedby cranial ultrasonography is:

A. Brainstem abnormality.B. Central nervous system infection.C. Intracranial arteriovenous malformation.D. Hypoxic-ischemic encephalopathy.E. Subarachnoid hemorrhage.



DOI: 10.1542/neo.2-1-e3 2001;2;3 NeoReviews

William D. Rhine and Francis G. Blankenberg Cranial Ultrasonography

& ServicesUpdated Information

s;2/1/e3http://neoreviews.aappublications.org/cgi/content/full/neoreviewincluding high-resolution figures, can be found at:

Supplementary Material

s;2/1/e3/DC1http://neoreviews.aappublications.org/cgi/content/full/neoreviewSupplementary material can be found at:

Permissions & Licensing

http://neoreviews.aappublications.org/misc/Permissions.shtmltables) or in its entirety can be found online at: Information about reproducing this article in parts (figures,

Reprints http://neoreviews.aappublications.org/misc/reprints.shtml

Information about ordering reprints can be found online:

by JoDee Anderson on January 21, 2008 http://neoreviews.aappublications.orgDownloaded from

usg

Documents