REVIEW PAPER

Noninvasive methods of detecting increased intracranial pressure

Wen Xu1& Patrick Gerety1 & Tomas Aleman2,3

& Jordan Swanson1& Jesse Taylor1

Received: 7 April 2016 /Accepted: 5 June 2016 /Published online: 28 June 2016# Springer-Verlag Berlin Heidelberg 2016

Abstract The detection of elevated intracranial pressure (ICP)is of paramount importance in the diagnosis and management ofa number of neurologic pathologies. The current gold standard isthe use of intraventricular or intraparenchymal catheters; howev-er, this is invasive, expensive, and requires anesthesia. On theother hand, diagnosing intracranial hypertension based on clini-cal symptoms such as headaches, vomiting, and visual changeslacks sensitivity. As such, there exists a need for a noninvasiveyet accurate and reliable method for detecting elevated ICP. Inthis review, we aim to cover both structural modalities such ascomputed tomography (CT), magnetic resonance imaging(MRI), ocular ultrasound, fundoscopy, and optical coherence to-mography (OCT) as well as functional modalities such as trans-cranial Doppler ultrasound (TCD), visual evoked potentials(VEPs), and near-infrared spectroscopy (NIRS).

Keywords Noninvasive intracranial pressure . Opticalcoherence tomography . Optic nerve sheath diameter .

Papilledema . Transcranial Doppler ultrasound .

Visual evoked potentials

Introduction

Elevated intracranial pressure (ICP) is a sign of numerous neu-rologic diseases and can have devastating consequences if di-agnosis and treatmentaredelayed [1].TheonsetofelevatedICPmay be quite rapid such as with traumatic brain injury (TBI) orepidural hematoma. A more difficult diagnostic situation iswhen elevated ICP occurs insidiously over a long time intervalsuch as may be the case in craniosynostosis, hydrocephalus,tumor, or idiopathic intracranial hypertension. Currently, evi-dence for an association between nonsymptomatic elevatedICP and neurocognitive delays in craniosynostosis is lacking[2]. However, intracranial hypertension associatedwith hydro-cephalus or papilledemamost likely requires surgical interven-tion. Thus, although the role of ICP in pediatric neurocognitiveoutcomes is actively debated, knowledge of ICP values andtrends may still be useful for long-term monitoring and man-agement of patients with a variety of neurological pathologies.

Certain symptoms such as headaches, vomiting, decreasedconsciousness, and visual changes are suggestive of increasedICP but sensitivity and specificity are low, and assessment inyoung children is difficult [3, 4]. The current gold standard foraccurate ICP measurement is invasive placement of an externalventricular drain (EVD), which has the ability to both measureICP aswell as drain cerebral spinal fluid in an effort to lower ICP.However, EVDs have been associated with high rates of infec-tion (up to 27%) [5–7], malposition (8.8%) [8], and hemorrhage(up to 18 %) [8, 9]. An alternative to EVDs is theintraparenchymal ICP monitor, which measures the ICP withinthe brain parenchyma itself and is associated with reduced risksof hemorrhage and infection [8]. However, both EVDs andintraparenchymal ICP monitors involve an invasive procedurethat requires hospitalization and typically 24–48 h of monitoringin the intensive care unit. Finally, placement of a measurementcathetermay not always be possible due to the patient’s conditionor lack of access to an experienced surgeon and appropriatefacilities [1, 10].

Electronic supplementary material The online version of this article(doi:10.1007/s00381-016-3143-x) contains supplementary material,which is available to authorized users.

* Jesse Taylortaylorj5@email.chop.edu

1 Craniofacial Surgery Center and Division of Plastic Surgery, TheChildren’s Hospital of Philadelphia and the University ofPennsylvania, Colket Translational Research Building, 3401 CivicCenter Blvd., Philadelphia, PA 19104, USA

2 Scheie Eye Institute and the Department of Ophthalmology,Perelman Center for Advanced Medicine and the University ofPennsylvania, Philadelphia, PA, USA

3 Division of Ophthalmology, The Children’s Hospital of Philadelphia,Philadelphia, PA, USA

Childs Nerv Syst (2016) 32:1371–1386DOI 10.1007/s00381-016-3143-x

The unreliability of clinical indicators and the invasive natureof intraventricular and intraparenchymal approaches haveprompted a search for alternative noninvasive methods for mea-suring ICP. Modalities include computed tomography (CT),magnetic resonance imaging (MRI), transcranial Doppler ultra-sound (TCD), and near-infrared spectroscopy (NIRS). The eye asan extension of the central nervous system (CNS) has been tra-ditionally used as a monitor of ICP through the exam of the opticnerve head during dilated fundus examinations. The subjectivenature of this assessment leads to variable or low reliability as apredictor of ICP [11–13]. These noninvasive modalities assesseither the three-dimensional structure of the brain, skull, ventri-cles, optic disk, or optic nerve, or the function of cerebral bloodflow or nerve conduction, each of which can be related to theICP. Quantitative measures of visual function such as visualevoked potentials (VEPs) or retinal and optic nerve structure,such as ocular ultrasound and optical coherence tomography(OCT), have thus been used as alternative biomarkers for ICP.The need for such noninvasive, objective measures of increasedICP is particularly acute in the pediatric setting where symptomsand clinical signs of increased ICP may be more difficult tointerpret. The purpose of this work is to review the current liter-ature on these emerging modalities and to compare their efficacyas noninvasive measures of ICP.

Structural modalities

Computed tomography

Head CT has become universally available as a diagnostic tool.Several CT findings that may correlate with increased ICP in-clude thumb-printing, ventricle effacement, midline shift, andobliteration of the basal cisterns (Fig. 1). In a study of adult TBIpatients by Toutant et al., 74 % of patients with absent cisternson CT had ICP values greater than 30 mmHg [14]. Eisenberg

et al. found that midline shift, compression or obliteration of themesencephalic cisterns, and the presence of subarachnoidblood were most significantly associated with abnormal ICP(measured as percent of time spent above 20 mmHg) and deathin 753 adult TBI patients [10]. Similarly, using multiple regres-sion analysis, Mizutani et al. looked at 39 CT parameters andfound that the following contributed to the prediction of abnor-mal ICP: the appearance of cisterns, the size of a subduralhematoma, ventricular size, status of subarachnoid hem-orrhage, status of cerebral contusion, magnitude of mid-line shift, and ventricular index [15]; they subsequentlyderived an equation to estimate ICP, which in 80 % ofcases was able to produce an estimated ICP value within10 mmHg of the true ICP. Another study used an auto-mated midline shift and blood and texture analysis algo-rithm to detect increased ICP in TBI patients with a sen-sitivity of 65 % and specificity of 73 % [16].

However, despite the results of these studies, there also existsliterature that questions the sensitivity of using CT to detect in-creased ICP. In a study conducted with adult trauma patients,Miller et al. examined ventricle size, sulci size, and gray/whitematter differentiation. They found that these three parameterstrended with ICP but were not able to reliably predict increasedICP. Using a logistic regression, they demonstrated a 4.9 % in-crease in the prediction of high ICP, which allowed them topredict twomore patients with high ICP than if they had not usedthe headCT [17]. Compared to the adult population, the evidencefor using CT to detect abnormal ICP in children is less compel-ling. One of the CT findings thought to predict acute elevationsin ICP is compressed or obliterated basal cisterns. However,Kouvarellis et al. found that the presence of open basal cisternswas only 58 % specific in predicting ICPs less than 20 mmHgand thereby concluded that open cisterns should not provideassurance of normal ICPs in children [18]. In a study of 68children with moderate to severe TBI, nine patients with normalCTs underwent invasive ICP monitoring due to clinical

Fig. 1 Sagittal (a) and axial (b)head CT scans demonstratingthumb-printing, or increasedconvolutional markings of theinner table

1372 Childs Nerv Syst (2016) 32:1371–1386

symptoms, and seven of these nine children had increased ICP[19]. Similarly, in a study with 34 children in nontraumatic co-mas, five out of seven patients with normal CTs developed ICPsgreater than 20mmHgwithin 36 h [20]; however, the correlationbetween CT findings and ICP in this study is difficult to state dueto the fact that they were not measured simultaneously. Rather,patients were scanned, received treatment, and then began con-tinuous ICP monitoring within 12 h of the CT scan.

Another use of head CT is in the evaluation of the opticnerve sheath diameter (ONSD). The optic nerve is an exten-sion of the CNS so its investing sheath is continuous with thedura mater that surrounds the brain parenchyma. As such, anyincreased pressure in the subarachnoid space of the brain istransferred to the fluid in the subarachnoid space surroundingthe optic nerve [21]. Due to this anatomical and physiologicalrelationship, elevated ICP has been shown to cause distentionof the optic nerve sheath and thereby, increase the optic nervesheath diameter (ONSD). In a group of 57 adult TBI patientswho had CT scans evaluating ONSD as well as invasivelychecked ICP values, there was a strong correlation betweenICP and ONSD (r = 0.74) with AUC = 0.83 for detectingICP > 20 mmHg. A 6.0 mm cutoff for the ONSD was used,which yielded a sensitivity of 97 %, specificity of 42 %, apositive predictive value (PPV) of 67 %, and negative predic-tive value (NPV) of 92 % [22].

HeadCT is a quick,widely available, operator independent,and easily obtained exam that may be able to predict elevatedICP (Table 1). The literature demonstrates that it is most suitedto identifying signs of elevated ICP in the most acute presenta-tions (Table 2). However, its disadvantages include variablesensitivity and the exposure of the patient to ionizing radiation,the effect of which is of particular concern in the pediatric pop-ulation because children are more radiosensitive and have alonger life expectancy compared to adults [23].

Magnetic resonance imaging (MRI)

MRI is another imaging modality that has shown promise inthe search for a noninvasive method of detecting elevated ICP.Several MRI-derived parameters have been proposed, includ-ing the elastance index, cerebral blood flow, and cerebrospinalfluid (CSF) velocity through the aqueduct. In addition, MRIhas been used to measure the ONSD.

MRI data can be used to calculate changes in intracranialvolume and pressure that occur with the cardiac cycle. Thechange in volume is derived from net transcranial CSF andblood volumetric flow rates, while the change in pressure isderived from the change in CSF pressure gradient calculatedfrom CSF velocity. By definition, elastance (E) is the ratio of

pressure change to volume change E ¼ ΔPΔV

� �, so using the

MRI-derived measurements, one can calculate an elastanceindex for the brain parenchyma. The exponential pressure-T

able1

Barriersto

useof

noninvasiveICPmeasurementm

odalities

Cost

Examination

time

Technician

required

Portability

Levelof

patient

compliance

required

Radiatio

nexposure

Use

inthe

ED

Needforsedation

inyoungchild

ren

Dataavailable

forpediatric

populatio

n

CT

Moderate

Short

Yes

No

Minim

alYes

Yes

No

Yes

MRI

High

Long

Yes

No

High

No

No

Yes

No

Ocularultrasound

Low

-moderate

Short

No

Yes

Moderate

No

Yes

Yes

aYes

Fundoscopy

Low

Short

No

Yes

Moderate

No

Yes

Yes

aYes

OCT

Low

Short

No

Yes

Moderate

No

Yes

Yes

Yes

TCD

High

Short

Yes

Yes

Moderate

No

Yes

Yes

aMinim

al

VEP

Moderate

Long

Yes

Yes

High

No

No

No

Yes

NIRS

Unknown

Long

No

Yes

Minim

alNo

No

No

Minim

al

aDenotes

thepossibility

ofperformingassessmentw

ithoutsedationin

cooperativeyoungchild

ren

Childs Nerv Syst (2016) 32:1371–1386 1373

volume curve for the brain introduced by Marmarous et al.states that at low ICP values, a small change in volume willcause a small change in pressure, whereas at high ICP values,a small change in volume will cause a large change in pres-sure; this relationship speaks to the high compliance (lowelastance) of the brain at low ICPs and low compliance (highelastance) of the brain at high ICPs. Knowing this relationshipbetween elastance and ICP, one can use the elastance indexcalculated from the MRI parameters to predict ICP [24, 25].Alperin et al. found that the elastance index and invasivelymeasured ICP values were highly correlated in five adult pa-tients (r2 = 0.965) [25]. Similarly, another study found that inchildren with hydrocephalus, shunt valve opening pressuresand predictions of ICP based on MRI-derived elastance werepositively correlated (Spearman ρ = 0.64) [26].

Another use for MRI to predict ICP involves the study ofcerebral blood flow (CBF) as measured by MR angiography.In a study of infants with hydrocephalus, CBF was measuredwith MRA at the level of the internal carotid arteries andbasilar artery both pre-shunt and post-shunt. The correlationcoefficient between CBF and invasively measured ICP wasr = −0.55 [27]. Furthermore, MRI has been used to look atCSF flow velocity through the cerebral aqueduct in patientswith suspected normal pressure hydrocephalus [28]. The de-gree of agreement between ICPmonitoring andMR dynamicswas 82 %; sensitivity and specificity of CSF velocity for ab-normal ICP were 90 and 50 %, respectively [29]. It has beenproposed that both CBF and CSF flow velocities are de-creased in settings of increased ICP due to decreased intracra-nial compliance [27, 28].

Like head CT, MRI can also be used to measure the opticnerve sheath diameter (ONSD). Geeraerts et al. demonstrateda positive correlation between MRI-derived ONSD andinvasively measured ICP (r = 0.71), and using a 5.82 mmcutoff, they determined that the ability of ONSD to detect

ICP > 20 mmHg was excellent with AUC = 0.94 and NPVof 92 % [30, 31]. Interestingly, a comparison of CT to MRIconcluded that the two modalities produce comparable mea-surements of ONSD [32].

MRI has the potential to noninvasively measure ICP basedon parameters such as flow velocities, intracranial elastance,and ONSD. However, disadvantages of this method includeits relatively higher cost, lack of portability, and long exami-nation time requiring high patient compliance and sedation inyoung children.

Ocular ultrasound

Although CTand MRI have been used to measure ONSD, themost widely studied modality for measuring ONSD is ocularultrasound. Greater resolution and detail are the obvious ad-vantages of this modality compared to CT and MRI. Severalstudies have correlated ultrasound-derived ONSD values withCT or MRI signs suggestive of increased ICP (Fig. 2).Shirodkar et al. found ONSD cutoffs (4.6 mm for females;4.8 mm for males) that had a sensitivity between 84.6 %(females) and 75 % (males) and a specificity of 100 % (bothgenders) for the detection of MRI signs of elevated ICP, suchas midline shift, edema, and effacement [33]. Similarly, astudy of 24 adult patients compared ONSD by ocular ultra-sound to CT findings of midline shift, sulcal effacement withsignificant edema, third ventricle collapse, and hydrocephalus,and using a cutoff of 5.0 mm, they achieved a sensitivity of100 %, specificity of 75 %, PPVof 95.4 %, and NPVof 100 %[34]. There have also been a number of studies that comparedocular ultrasound-derived ONSD with ICP measured by LPopening pressure or invasive monitoring. In the adult popula-tion, the literature shows evidence for cutoff values rangingfrom 4.1 to 5.7 mm, which are associated with 74.1 to 100 %sensitivity and 74 to 100 % specificity [35, 36]. Interestingly,

Table 2 Detection of acute vs.chronic ICP elevation bynoninvasive modalities

Acute findings of elevated ICP Chronic findings of elevated ICP

CT ▪ Obliteration or compression of the basalcisterns

▪ Ventricle effacement

▪ Midline shift

▪ Thumb-printing

▪ Ventricle effacement

MRI ▪ Increased elastance index ▪ Increased elastance index

Ocular ultrasound ▪ Increase in optic nerve sheath diameter ▪ Increase in optic nerve sheath diameter

Fundoscopy ▪ Likely normal ▪ Optic disk swelling

▪ Blurred disk margin

OCT ▪ Unknown ▪ Increase in RNFL thickness

▪ Increase in TRT

▪ Increase in ONHVTCD ▪ Increase in PI ▪ Increase in PIVEP ▪ Prolongation of peak latencies ▪ Prolongation of peak latencies

NIRS ▪ Decrease in cerebral oxygenation ▪ Decrease in cerebral oxygenation

1374 Childs Nerv Syst (2016) 32:1371–1386

one study demonstrated that changes in ONSD appear quiterapidly after the onset of elevated ICP. In that study of 18 adultTBI patients, Maissan et al. investigated the effects of acuteICP changes on ONSD by measuring ONSD before, during,and after tracheal manipulation, which is known to increaseICP [37]. They found that ONSD increased with tracheal ma-nipulation and immediately returned to baseline once manip-ulation was stopped. On the other hand, one study found thatalthough there was a correlation between ONSD and ICP, nosingle cutoff value could be used as a suitable diagnostic toolto predict ICP [38]. Another study of 10 adult trauma patientsusing a cutoff of 6.0 mm demonstrated 36 % sensitivity and38 % specificity, and thereby concluded that ONSD measuredby US is not a useful tool for predicting ICP [39]. Table 3enumerates a list of ONSD studies, modality of measurement,cutoff values, and relevant statistics.

A challenge associated with this modality is the pres-ence of inter-individual variability. A Chinese study with279 adults found an optimal cutoff of 4.1 mm, which is atleast 0.5 mm lower than any other study. Furthermore,interpreting ONSD measurements in children provides anadditional challenge in that ONSD tends to increase withage [40]. One normative study claims that ONSD valuesgreater than 4.0 mm in infants less than 1 year and 4.5 mmin older children should be regarded as abnormal [41];similarly, a study of 483 children determined the upperlimit of normal to be 4.5 mm based on neurological exam-ination, imaging, and/or clinical follow-up [42]. Thesestudies demonstrate that sophisticated normative data thatallows for differences in age, gender, race, and comorbid-ities do not exist.

ONSD measured by ocular ultrasound has the potentialto be a powerful tool in the noninvasive detection of ele-vated ICP. A common concern regarding ultrasound tech-niques is operator variability. A study by Ballantyne et al.measured both intra- and inter-observer variation whenmeasuring ONSD in 67 healthy adults [43]; they found thatwith three independent observers, the median intra-

observer variation was ±0.1 mm with 5th–95th centilevalues of ±0–0.4 mm, and the median inter-observer vari-ation was ±0.2–0.3 mm with 5th–95th centile values of±0–0.7 mm. Therefore, they concluded that with standard-ization of technique, operator variability can be minimized.Overall, ONSD as measured by ocular ultrasound involveslittle risk to the patient, is relatively inexpensive, quick touse, portable, and generally provides values that correlatewell with ICP.

Fundoscopy

Papilledema, defined as optic disk edema due to increasedICP, diagnosed on fundoscopic examination is widely usedas a clinical indicator of increased ICP. It is a subjective mea-sure that can be divided into six grades as described by Frisén[44], with grade 0 denoting a normal optic disk and grade 5denoting a severe degree of optic disk edema (Fig. 3). In theliterature, the sensitivity of papilledema for increased ICPranges from 14 to 40 % [11, 12]. Steffen et al. showed thatin 37 patients with acute elevated ICP due to hemorrhage ortrauma, papilledema was an uncommon occurrence [45]; this,however, may be related to the pathophysiology ofpapilledema, and perhaps the duration of elevated ICP wasinsufficient to cause clinically apparent papilledema.

Papilledema has also been tested in clinical situations inwhich ICP is more chronically elevated. In a group of adultswith idiopathic intracranial hypertension (IIH; formerlyknown as pseudotumor cerebri), 5.7 % did not havepapilledema [46]. In another study of 85 adult patients withchronic headache, 15 % had ICP > 18.4 mmHg and no evi-dence of papilledema [47]. The results for the pediatric popu-lation are more variable. Two studies of children with IIHshowed that 17.8–48 % of patients did not demonstratepapilledema on exam [48, 49].

Fundoscopic exam to detect papilledema has long been amainstay for diagnosing elevated ICP noninvasively and canbe performed relatively quickly and easily. Best diagnostic

Fig. 2 Measurement of the opticnerve sheath diameter (ONSD) bycomputer tomography (a) andocular ultrasound (b). Adaptedfrom Kalantari et al. [32] andBlaivas et al. [105]

Childs Nerv Syst (2016) 32:1371–1386 1375

Tab

le3

Summaryof

ONSDstudies

Study

Imagingmodality

Patient

type

Studysize

(n)

ONSD

cutoff

Invasive

ICPmeasurementa

Relevantstatistics

Sekhon

etal.2014[22]

CT

AdultTBIpatients

576.0mm

Yes

r=0.74,p

<0.001

AUC=0.83

Sensitivity

97%

Specificity

42%

Geeraertsetal.2008[30]

MRI

AdultTBIpatients

385.82

mm

Yes

r=0.71,p

<0.0001

AUC0.94

NPV

92%

Shirodkaretal.2014[33]

Ultrasound

Adults

with

signsof

increasedICP

604.6mm

(fem

ale)

4.8mm

(male)

No.comparedto

MRIfindings

suggestiv

eof

increasedICP

Sensitivity:8

4.6%

females,75%

males

Specificity

:100

%both

genders

Cam

marataetal.2011[108]

Ultrasound

Adulttraumapatients

11N/A

Yes

r=0.74,p

<0.001

Qayyum

etal.2013[34]

Ultrasound

Adults

with

signsof

increasedICP

245.0mm

No.comparedto

CTfindings

suggestiv

eof

increasedICP

Sensitivity

100%

Specificity

75%

PPV95.4%

NPV

100%

Nabetaetal.2014[38]

Ultrasound

HIV

+adultswith

suspected

cryptococcalmeningitis

815.0mm

Yes

Relativerisk

of2.39

forICP>200mmH2O

Wangetal.2015[35]

Ultrasound

Adults

with

signsof

increasedICP

279

4.1mm

Yes

Sensitivity

95%

Specificity

92%

Soldatos

etal.2008[109]

Ultrasound

Adults

with

braininjury

505.7mm

Yes

r=0.68,p

=0.002

Sensitivity

74.1%

Specificity

100%

Maissan

etal.2015[37]

Ultrasound

AdultTBIpatients

185.0mm

Yes

r2=0.8

Sensitivity

94%

Specificity

98%

AUC0.99

Aminietal.2013

[36]

Ultrasound

Adultnontraum

apatientsatrisk

forincreasedICP

505.5mm

Yes

Sensitivity

100%

Specificity

100%

Strumwasseretal.2011[39]

Ultrasound

Adulttraumapatients

106.0mm

Yes

AUC0.36

Sensitivity

36%

Specificity

38%

Korberetal.2005[42]

Ultrasound

Children(m

eanage7.5years)with

suspectedincreasedICP

483

4.5mm

No

NormalICPmeandiam

eter:

3.4mm

±0.7mm

IncreasedICPmeandiam

eter:

5.6mm

±0.9mm

Ballantyneetal.1999[41]

Ultrasound

Childrenwith

noconcernfor

increasedICP

102

4.0mm

(<1year)

4.5mm

(>1year)

No

MeanONSD

:3.08mm

±0.36

mm

r2=0.48

forONSD

andagerelationship

aInvasive

ICPmeasurementd

enotes

intraventricular

catheter,intraparenchymaldevice,orlumbarpuncture

1376 Childs Nerv Syst (2016) 32:1371–1386

yield is obtained by binocular biomicroscopic examination ofthe optic nerve with the use of a slit lamp which limits its use tocooperative and often only ambulatory patients. The literatureshows that this exam is much less sensitive than other tests.

Optical coherence tomography (OCT)

Optical coherence tomography (OCT) uses interferometry toobtain micron-resolution images of the optic disk based on thevariation the different reflectivities of each of the retinallayers. There are two main types of OCT, time domain OCT(TD-OCT) and spectral domain OCT (SD-OCT); the majordifference between the two lies in the way in which informa-tion is processed as it is reflected and returned to the detector[50]. While TD-OCT uses a photodetector in the detector arm,the newer SD-OCT uses a spectrometer which allows formuch faster acquisition of images with increased lateral reso-lution as well as for online imaging processing resulting in aremarkable increase in resolution [50]. A typical SD-OCTexam comprises of directing the light beam onto the optic diskand capturing an image that can then be analyzed manually orautomatically by built-in software typically embedded within

the OCT instruments (Fig. 4). By providing a detailed exam-ination of the retina and optic nerve head, equivalent in manyways to an in vivo, noninvasive biopsy, OCT has been shownto be a valuable tool in the diagnosis and monitoring of anumber of pathologies, including glaucoma [50, 51], maculardegeneration [52], papilledema [53], and optic neuropathy inolder children with craniosynostosis [54].

There are a number of parameters that can be studied on theOCT-acquired image of the retina, including peripapillary ret-inal nerve fiber layer (RNFL) thickness, total retinal thickness(TRT), macular thickness, and optic nerve head height andvolume. Normative data for some of these parameters havebeen obtained in adults and children. Gabriele et al. describedRNFL thickness at different points on the retina of healthyadults [55], while Varma et al. looked specifically at adultLatinos and concluded that peripapillary RNFL and macularthickness decrease with age but do not vary with gender [56].Normative data for children have also been partiallyestablished. A Turkish study looked at children ages 6–16 years and determined that peripapillary RNFL, macularthickness, and macular volume were not correlated with age,spherical equivalence, or axial length (cornea-retina distance)[57]. A similar study in Hong Kong children ages 6–17 yearsshowed that while age did not significantly affect OCT mea-surements, axial length demonstrated a negative correlationwith RNFL thickness [58]. Another normative study ofNorth American children ages 5–15 years found that macularthickness varied with age; meanwhile, RNFL thickness didnot correlate with age but in general, tended to be greater inthe pediatric population [59]. Importantly, the NorthAmerican study reported a 95 % success rate in obtaininghigh-quality OCT images in children ages 5 and above with-out sedation, which is significant given that the successful useof OCT requires reasonable patient compliance. In compari-son to the other normative studies, El-Dairi et al. found that in286 healthy children ages 3–17 years, OCT measurementsactually varied with three different factors: race, axial length,and age [60]. Of note, normative OCT values for infants andyoung children less than 3 years old have not been extensively

Fig. 3 Fundoscopic image of an optic disk with grade 3 papilledema.Adapted from Wall et al. [106]

Fig. 4 Optical coherencetomography (OCT) image at thelevel of the optic nerve head

Childs Nerv Syst (2016) 32:1371–1386 1377

studied. Determining universal normative values for childrenis an ongoing challenge, but interestingly, a study by Sasapinet al. found that OCT parameters did not change significantlyin healthy children followed over a 4-year time span [61],suggesting that OCT may be useful for long term monitoringregardless of generalizable normative data.

A complex interplay exists between OCT parameters,papilledema, and ICP. The correlation between OCT andFrisen grading for papilledema has been demonstrated in sev-eral studies [62–64]. One study found a positive correlationbetween Frisen grading and RNFL thickness with r = 0.7952[63]. Taking it one step further, Vartin et al. looked at bothRNFL thickness and peripapillary total retinal thickness(TRT) and concluded that TRT is a more sensitive indicatorfor mild papilledema [62]. Meanwhile, Scott et al. found thatin 36 patients with papilledema, both RNFL thickness andTRT correlated well with Frisen scores with r = 0.85 andr = 0.87, respectively. However, they determined that withhigher-grade abnormalities, TRTmay showmore proportionalchange per change in degree of disk edema, which wouldmake TRT a more favorable parameter to study when dealingwith more extreme pathology. They hypothesized that Bthisdisproportional increase of TRT above that of the RNFL rep-resents fluid from neurosensory retinal detachment in theperipapillary retina^ [64].

There have also been a few studies looking at the relation-ship of OCT to ICP. Kaufhold et al. found that the OCT pa-rameter optic nerve head volume (ONHV) was actually a bet-ter measure of ICP compared to RNFL thickness withAUC = 0.835 and AUC = 0.464, respectively [65]. The majorflaw with their study, however, is that the ICP measurementswere acquired from lumbar puncture (LP) opening pressuresthat were performed anytime within 0 to 24 months of theOCT scan, which in some cases, resulted in a large time gapbetween ICP and OCT measurements. Another study of pa-tients with newly diagnosed IIH compared RNFL thicknessand TRT with LP opening pressure and found that includingthese parameters in multiple regression models to detectICP > 25 cmH2O (∼18.4 mmHg) improved the AUC from77.1 to 86.9 [66].

Finally, a couple of studies have studied the relationship ofOCT parameters with both papilledema and ICP. The OCTSub-Study Committee for the NORDIC IdiopathicIntracranial Hypertension Study Group found that OCT pa-rameters (RNFL thickness, TRT, ONHV) were significantlycorrelated with both Frisen grade and ICP measured by LPopening pressure, but that the correlation was stronger withthe former (r > 0.76) and weaker with the latter (r > 0.24) [67].In another study, OCT and papilledema were compared headto head in a group of 20 patients with confirmed elevated ICPvia LP opening pressure. Direct ophthalmoscopy and fundusphotography detected 80 and 70 %, respectively, while theOCT parameters of peripapillary average retinal thickness

and peripapillary RNFL thickness detected 90 and 85 %, re-spectively [68], demonstrating that perhaps OCT may be amore sensitive indicator of elevated ICP compared topapilledema.

In general, OCT is a promising technology that may beuseful in the noninvasive detection of elevated ICP. Furtherinquiry should be performed to determine the best parametersto measure as well as to increase our knowledge of normativevalues. By allowing a detailed quantitative examination of theoptic disk, it is a technological leap from fundoscopy. Currentevidence suggests that OCT parameters may be a more sensi-tive indicator of increased ICP compared to fundoscopy andtherefore may be able to detect changes in the eyes related toICP before they become apparent on routine ophthalmologicalexam [68]. Additionally, OCT machines are reasonablypriced, portable, can provide quantitative measurements, anddo not involve ionizing radiation, and therefore, would be safeto be used as a long-term monitoring and screening tool inboth the adult and pediatric populations.

Functional modalities

Transcranial Doppler ultrasonography (TCD)

TCD is an ultrasound technology that looks at blood vesselflow velocities in order to predict ICP. The relationship be-tween ICP and TCD flow variables was first described byKlingelhofer in 1988 and relies heavily on cerebral hemody-namics [69]. Physiologically, as ICP increases, diastolic flowvelocity is reduced to a greater degree compared to systolicflow velocity, which results in an increased pulse peak be-tween diastole and systole. The Gosling pulsatility index

(PI) is defined as PI ¼ FV peak systole−FVend diastole

FV peak systole[70]. Therefore, a

rise in ICP leads to increases of both pulse peak and PI. Oneimportant detail to note is that although flow velocities areaffected by the angle of insonation, PI is a ratio and thus notaffected, making it a more robust predictor of ICP [71]. Astandard TCD exam consists of finding an adequate acousticwindow in the temporal bone with the ultrasound probe andusing Doppler technology to measure flow velocities (Fig. 5).Current clinical pathologies for which TCD can be used, asdetermined by the American Academy of Neurology (AAN),include sickle cell stroke, angiographic vasospasm, intracrani-al steno-occlusive disease, and cerebral circulatory arrest.TCD has also been used for monitoring during cerebral throm-bolysis, coronary artery bypass grafting, and carotid endarter-ectomy [72].

Various studies have compared TCD measurements to ICPmeasured invasively or with LP opening pressure. Ragauskaset al. found that TCD was a superior indicator ofICP > 14.7 mmHg compared to ultrasound-measured

1378 Childs Nerv Syst (2016) 32:1371–1386

ONSD; ONSD with a cutoff value of 5.0 mm yielded a sen-sitivity of 37.0 %, specificity of 58.5 %, and AUC = 0.57,while TCD demonstrated a sensitivity of 68.0 %, specificityof 84.3 %, and AUC = 0.87 [73]. A number of studies havefound that PI and ICP are highly correlated with correlationcoefficients ranging from 0.89 to 0.938 [74, 75]; similarly,Hunter et al. found that the change in ICP was correlated withthe change in PI before and after CSF withdrawal in patientswith suspected idiopathic intracranial hypertension(p = 0.004) [76]. Using PI cutoff values ranging from 1.26to 1.335, several studies determined that PI could predict ele-vated ICP values with 80–88.5 % sensitivity and 90–97 %specificity [75, 77, 78].

Although it seems that PI is a good predictor of increasedICP with high sensitivity and specificity, there is opposingevidence as well. One study stated that although PI seems tocorrelate with ICP trends (e.g., higher PI in hydrocephaluspatients vs. normal controls), an exact noninvasive measure-ment of ICP by TCD is unlikely [79]. Likewise, Zweifel et al.found that the diagnostic value of PI to assess ICP has anoverall AUC = 0.62 for ICP > 15 mmHg and AUC = 0.74for ICP > 35mmHg [80]; from these data, they concluded thatthe relationship between PI and ICP is strong for highly ele-vated ICP but weak for marginally elevated ICP, thereby mak-ing PI an unreliable diagnostic tool in the range of ICP forwhich a noninvasive method would be most essential.Another limitation of TCD is that physiologic variations canhave profound effects on the ICP-PI relationship. Behrenset al. found that variations in vessel compliance, autoregula-tion, and arterial pressure can all confound the ICP prediction[81]. In terms of using TCD for long-term monitoring, a studyof infants with hydrocephalus showed that PI correlated wellwith ICP immediately before and after shunt procedures, butthe ICP-PI relationship was not maintained at long-term fol-low-up [82]. This may be due to changes in brain complianceassociated with chronically abnormal intracranial pressures.One other consideration when using TCD is that the

technology relies on the presence of an adequate temporalbone acoustic window, which is not present in all patients.In the Behrens study, two out of ten patients were unable tocomplete testing due the absence of an acoustic windowthrough which TCD measurements could be made [81].

Generally, TCD-derived parameters such as PI seem to becorrelated with invasively measured ICP values. Possible dis-advantages of TCD include the confounding effects of variousphysiologic differences (e.g., arterial pressure) and pathologicdifferences (steno-occlusive vessel disease) on TCD measure-ments as well as the need for an adequate acoustic windowand interpretation by experienced sonographers [83].

Visual evoked potentials (VEPs)

VEPs are the summed electrical signals produced in the oc-cipital lobe in response to visual stimuli. There are many com-ponents to VEP waveforms, including number of positive ornegative waves, latency of these waves from time of stimulus,and shape of the waveform. The peaks of the VEP are namedP or N, indicating positive or negative deflection, with a num-ber subscript, indicating the average peak latency followingthe time of stimulus [84]. VEPs provide an overarching pic-ture of the status of the visual system from optic nerve toprimary visual cortex. Obtaining a VEP tracing involves plac-ing standard EEG electrodes on the patient’s scalp and record-ing the summed signals as various types of visual stimuli arepresented to the patient. Current applications of VEPs includethe management of optic nerve pathologies and the monitor-ing of brain activity during select surgeries [85].

Studies of children with hydrocephalus have demonstratedthat progression of their intracranial pathology is associatedwith concurrent changes in the shapes and latencies of theirflash VEP (F-VEP) waveforms, and treatment with shuntplacement leads to F-VEP normalization [86, 87].Specifically, Sjöström et al. found that the P′ (P prime) latencywas disproportionately prolonged just prior to surgical

Fig. 5 Transcranial Doppler ultrasound waveforms. a Patient withbacterial meningitis with CSF pressure of 37 cmH2O and PI = 1.46(upper) on day 1 of treatment and CSF pressure of 16c mH2O and

PI = 1.12 (lower) on day 5 of treatment. b Transcranial Doppler spectrapossibilities at varying pulsatility indices (PI). Adapted from Alvarez-Fernández et al. [107] and Wakerley et al. [78]

Childs Nerv Syst (2016) 32:1371–1386 1379

intervention and normalized following the surgery [88]. As anexplanation for these changes, Fichsel postulated that the al-terations in VEP were produced by the effects of enlargingventricles on the visual radiation pathways in the brain [87].

In addition to being correlated with high ICP disease pro-cesses, VEPs and specifically the N2 latency have also beencompared directly to invasively measured ICP values. Severalstudies have reported a positive correlation between ICP andF-VEP N2 wave latency with correlation coefficients rangingfrom 0.83 to 0.84 [89, 90]. Similarly, Gumerlock et al. foundthat a F-VEP N2 latency >80 ms linearly corresponded to anICP > 20 cmH2O [91].

The potential variability in VEPs within and between indi-vidual patients is an important source of debate. A longitudi-nal study that recorded monthly F-VEPs for up to 4 months inchildren has demonstrated that although there seems to besignificant inter-individual variability in N2 latency, it is sim-ilar within individuals across time; these data suggest that N2latencymay be useful for the long-termmonitoring of childrenat risk of elevated ICP [92]. On the other hand, Anderssonet al. showed that in healthy adults tested on three differentoccasions with F-VEP, there was significant intra- and inter-individual variability in N2 and P2 latencies, amplitudes, andoverall waveform shape [93].

In terms of emerging technologies, multifocal VEPs, whichinvolve using hundreds of stimulations presented close in time,may provide a more detailed look at the visual pathways [85,94]. However, the same questions regarding inter- and intra-individual variability will undoubtedly arise. As for the current-ly available literature on VEP and its use in ICP measurement,the majority of the studies have employed F-VEP rather thanpatterned VEP. It has been shown that patterned VEP displaysless inter-individual variability and thus, may be preferable fordetermining ICP [95]. However, compared to F-VEP whichcan be performed easily on most patients, patterned VEP mea-surement necessitates an awake and attentive patient who canfixate on the stimulus as instructed and therefore would bedifficult to do with younger children and patients with alteredmental status [85]. Another consideration is that VEPs are af-fected by anesthesia since they are a measure of brain electricalactivity; therefore, their use in trending ICP would be some-what limited in sedated and intubated patients [93].

Near-infrared spectroscopy (NIRS)

First described by Franz Jobsis in 1977 [96], NIRS is a noninva-sive optical technique that utilizes light in the near infrared spec-trum. In between source and detector, the light can be scatteredand absorbed by chromophores, leading to a certain amount ofattenuation that can then be used to derive the ratio of oxyhemo-globin to deoxyhemoglobin and therefore calculate the tissue ox-ygen saturation in the local area [97]. The majority of NIRScerebral oximeters currently on the market comprise of a

combination light source and sensor pad that is applied to thepatient’s head and connected to a monitor that displays regionalsaturation values calculated from proprietary algorithms. Mostcurrent clinical use of NIRS is for monitoring during cardiacand vascular procedures aswell as in the care of TBI patients [97].

In order to understand the use of NIRS in studying ICP andbrain injury, one must first have knowledge of the conceptsbehind cerebrovascular reactivity and specifically the pressurereactivity index (PRx). PRx is a correlation coefficient be-tween simultaneous measures of ICP and arterial blood pres-sure (ABP) and is therefore reflective of the reactivity of ves-sels to changes in ABP [98]. Several studies have shown that apositive value for PRx is correlated with higher ICP andpoorer long-term outcomes. Czosnyka et al. studied 82 pa-tients with brain injuries and found that a positive PRx corre-lated with high ICP (r = 0.366; p < 0.001), low admissionGCS score (r = 0.29; p < 0.01), and poor outcome at 6 monthsafter injury (r = 0.48, p < 0.00001), and therefore concludedthat although they were unable to use PRx to predict an exactICP, it does have some correlative and prognostic implications[98]. Similarly, another study found that a value of PRx > 0.35was associated with over 50 % mortality rate at 6 monthsfollowing TBI [99]. An interesting use of PRx was demon-strated by Steiner et al., who studied 114 TBI patients who hadcontinuous mean arterial pressure and ICP monitoring. Withthese values, they were able to determine an optimal cerebralperfusion pressure (CPP) for each patient, which was definedas the CPP at which PRx reached its minimal value. At6 months post-injury, Glasgow outcome scale results showedthat patients managed at the CPP closest to the calculatedoptimal CPP had the best outcomes [100].

Although PRx seems to correlate with ICP and can beprognostically significant, its calculation requires invasiveICP measurement, so by definition, it cannot be a noninvasiveindicator of elevated ICP. There is a potential to use NIRSparameters combined with ABP measurements to noninva-sively estimate ICP by creating a new index that is similar toPRx. Lee et al. postulated that the same blood changes thatcause ICP shifts would also induce changes in the relative totalhemoglobin. They defined hemoglobin volume index (HVx)as the correlation coefficient between simultaneous measuresof relative total hemoglobin (rTHb), obtained by NIRS, andarterial blood pressure (ABP) [101]. By measuring rTHb,ABP, and ICP simultaneously in a piglet experimental model,they determined that HVx correlated with PRx (r = 0.73)[101]. In a similar study, Zweifel et al. defined total hemoglo-bin reactivity (THx) as a moving correlation coefficient be-tween total hemoglobin index, obtained by NIRS, and ABP[102]. PRx and THx were significantly correlated within in-dividuals (r = 0.49, p < 0.0001) and between individuals(r = 0.56, p = 0.0002). They also went one step further andused THx to calculate optimal CPP, which showed significantagreement with the optimal CPP determined using PRx [102].

1380 Childs Nerv Syst (2016) 32:1371–1386

In addition to being tested against PRx, NIRS-derivedparameters have also been directly compared to variousvalues of ICP and changes in ICP. One study used NIRSto calculate regional cerebral oxygenation (rSO2) inpatients with ICP > 25 mmHg and patients withICP < 25 mmHg [103]. They found that rSO2 values weresignificantly lower in patients with elevated ICP and con-cluded that NIRS may noninvasively detect impaired cere-bral microcirculation in patients with increased ICP. Asecond study looked at six comatose children in whomphenylephrine injections were used to temporarily manipu-late BP and ICP. Continuous MAP, ICP, and NIRS moni-toring showed a significant correlation between cerebralhemoglobin saturation changes measured by NIRS andICP changes (r = 0.82, p < 0.001) [104].

In summary, several studies have shown that NIRS-derivedindices of cerebrovascular reactivity seem to be correlatedwith ICP and long-term patient outcomes after TBI[102–104]. Currently, NIRS has not been used to predict exactICP measurements but further research into this modality isrequired in order to make any relevant conclusions regardingits use as a noninvasive detector of increased ICP.

Conclusion

In this review, we have outlined the current state of bothstructural (CT, MRI, fundoscopy, OCT, and ONSD) andfunctional (TCD, VEP, NIRS) noninvasive methods of

detecting elevated ICP (Tables 4 and 5). Some modalitiessuch as NIRS have been shown to just trend with ICP whilemodalities like ocular ultrasound and TCD have demon-strated high sensitivity and specificity for predicting clini-cally important elevations in ICP. Additionally, while CT,ocular ultrasound, TCD, VEP, and NIRS may be able todetect acute ICP changes, fundoscopy and OCT may bebetter suited for examining ICP trends and detecting insi-dious rises in ICP. In terms of the pediatric population,there is some data available for CT, ocular ultrasound, fun-doscopy, OCT, and VEP, but factors such as patient coope-ration and need for sedation should be considered.

At our high-volume craniofacial referral center, weregularly utilize noninvasive methods of ICP measure-ment. Almost all craniosynostosis patients undergo rou-tine fundoscopic examination by an experienced pediatricophthalmologist both preoperatively and during routinefollow-up. Additionally, children who experience anyclinical symptoms of elevated ICP undergo fundoscopicexamination and often go on to require 24–48 h ofintraparenchymal ICP monitoring in the intensive careunit. However, due to the low sensitivity of papilledemaand the invasiveness of intraparenchymal monitoring, adilemma exists in the management of a patient who hasclinical symptoms but no papilledema. Should the patientbe observed, undergo invasive ICP monitoring, orundergo surgical cranial expansion? As a result of theproblems posed by relying on only fundoscopy andintraparenchymal monitoring, we have decided to explore

Table 4 Comparative studies between modalities

CT

MR

I

Fu

nd

osc

opy

OC

T

TC

D

VE

P

NIR

S

CT

MRI xOcular

ultrasound xFundoscopy

OCT xTCD xVEP

NIRS

Ocu

lar

ult

raso

un

d

Childs Nerv Syst (2016) 32:1371–1386 1381

the use of OCT in detecting elevated ICP. As a noninva-sive modality, OCT can be used preoperatively in theclinic for older children and during sedation for CT orMRI for younger children and infants. Additionally, ifvalidated, OCT could be used to observe children in theimmediate postoperative time period as well as duringtheir yearly follow-up visit. Due to its noninvasivenessand higher sensitivity, OCT may be able to bridge thegap between fundoscopy and intraparenchymal monito-ring in the care of patients with craniofacial anomalies.As we work on validating the OCT parameters, we hopeto further incorporate OCT as a noninvasive measure ofICP into our craniofacial practice.

Overall, challenges of almost all techniques discussed in-clude obtaining a reliable set of normative data for a popula-tion diverse in race, gender, and age, as well as reducing intra-and inter-individual variability, which may not be universallypossible. In addition, within each modality, one must deter-mine the most indicative parameter to study as well as estab-lish the best cutoff points for the continuous parameters. Otherconsiderations include examination time, ease of use, cost ofequipment, and cost of personnel and training (OnlineResource 1 for cost estimates). At present, no single methodhas prevailed, so further research is required in order to findand optimize a noninvasive method of measuring ICP that isreliable, easy to use, and cost-effective.

Table 5 Summary of comparative study findings

Modalities Study Findings and conclusions

OCT vs. fundoscopy Vartin et al. [62] ▪ No comparison to invasively measured ICP▪ OCT parameter peripapillary total retinal thickness (PTR) is a sensitive indicator

of mild papilledema

Ahuja et al. [63] ▪ No comparison to invasively measured ICP▪ Positive correlation between Frisen grading of papilledema and OCT parameter

retinal nerve fiber layer (RNFL) thickness with r = 0.80

Scott et al. [64] ▪ No comparison to invasively measured ICP▪ Positive correlation between modified Frisen stage of papilledema and OCT

parameter RNFL thickness with r = 0.85▪ Positive correlation between modified Frisen stage of papilledema and OCT

parameter total retinal thickness (TRT) with r = 0.87

OCT Sub-Study CommitteeGroup [67]

▪ Comparison to ICP measured via lumbar puncture (LP)▪ Strong correlation between OCT parameters and Frisen grade with r > 0.76▪ Mild correlation between OCT parameters and LP opening pressure with r > 0.24

Skau et al. [68] ▪ Comparison to ICP measured via lumbar puncture (LP)

Parameter/modality Sensitivity

Fundus photography 70 %

Direct ophthalmoscopy 80 %

OCT RNFL thickness 85 %

OCT peripapillary average retinal thickness 90 %

CT vs. MRI forONSDMeasurement

Kalantari et al. [32] ▪ No comparison to invasively measured ICP▪ ONSD measured by CT and ONSD measured by MRI correlate strongly with

r = 0.8941

Ocular ultrasoundvs. CT

Ohle et al. [108] ▪ No comparison to invasively measured ICP▪ Meta-analysis with 12 studies▪ ONSD determined by ocular ultrasound is 95.6 % sensitive and 92.3 % specific for

CT findings suggestive of increased ICP (ex. shift, collapse of ventricles, intracranialbleeding of greater than 2 mm)

▪ ONSD cutoff values:Age <1 year: 4 mmAge 1–17 years: 4.5 mmAge > 17 years: 5 mm

Ocular ultrasoundvs. TCD

Ragauskas et al. [73] ▪ Comparison to ICP measured via lumbar puncture (LP)▪ Elevated ICP defined as ICP > 14.7 mmHg

Modality Sensitivity Specificity AUC

Ultrasound ONSD 37.0 % 58.5 % 0.57

TCD 68.0 % 84.3 % 0.87

1382 Childs Nerv Syst (2016) 32:1371–1386

Acknowledgments The authors acknowledge Dr. Steve Siegel for re-view writing guidance. The first author of this publication was supportedby the National Center for Advancing Translational Sciences of theNational Institutes of Health under award number TL1TR000138. Thecontent is solely the responsibility of the authors and does not necessarilyrepresent the official views of the National Institutes of Health.

Compliance with ethical standards

Funding The first author of this publication was supported by theNational Center for Advancing Translational Sciences of the NationalInstitutes of Health under award number TL1TR000138. The content issolely the responsibility of the authors and does not necessarily representthe official views of the National Institutes of Health.

Conflict of interest The authors declare that they have no conflict ofinterest.

Research involving human participants and/or animals This articledoes not contain any studies with human participants or animals per-formed by any of the authors.

Informed consent This article does not require any form of informedconsent.

References

1. Czosnyka M (2004) Monitoring and interpretation of intracranialpressure. J Neurol Neurosurg Psychiatry 75:813–821. doi:10.1136/jnnp.2003.033126

2. Hayward R, Britto J, Dunaway D, Jeelani O (2016) Connectingraised intracranial pressure and cognitive delay in craniosynosto-sis: many assumptions, little evidence. J Neurosurg Pediatr 13:1–9. doi:10.3171/2015.6.PEDS15144

3. Medina LS, Pinter JD, Zurakowski D, et al. (1997) Children withheadache: clinical predictors of surgical space-occupying lesionsand the role of neuroimaging. Radiology 202:819–824. doi:10.1148/radiology.202.3.9051039

4. de Jong T, Bannink N, Bredero-Boelhouwer HH, et al. (2010)Long-term functional outcome in 167 patients with syndromiccraniosynostosis; defining a syndrome-specific risk profile. JPlast Reconstr Aesthet Surg 63:1635–1641. doi:10.1016/j.bjps.2009.10.029

5. Koskinen L-OD, Grayson D, Olivecrona M (2013) The compli-cations and the position of the Codman MicroSensor™ ICP de-vice: an analysis of 549 patients and 650 sensors. Acta Neurochir155:2141–2148 . doi:10.1007/s00701-013-1856-0discussion2148

6. Hoefnagel D, Dammers R, Ter Laak-Poort MP, Avezaat CJJ(2008) Risk factors for infections related to external ventriculardrainage. Acta Neurochir 150:209–214. doi:10.1007/s00701-007-1458-9

7. Dasic D, Hanna SJ, Bojanic S, Kerr RSC (2006) External ventric-ular drain infection: the effect of a strict protocol on infection ratesand a review of the literature. Br J Neurosurg 20:296–300. doi:10.1080/02688690600999901

8. Anderson RCE, Kan P, Klimo P, et al. (2004) Complications ofintracranial pressure monitoring in children with head trauma. JNeurosurg 101:53–58. doi:10.3171/ped.2004.101.2.0053

9. Binz DD, Toussaint LG, Friedman JA (2009) Hemorrhagic com-plications of ventriculostomy placement: a meta-analysis.Neurocrit Care 10:253–256. doi:10.1007/s12028-009-9193-0

10. Eisenberg HM, Gary HE, Aldrich EF, et al. (1990) Initial CTfindings in 753 patients with severe head injury. A report from

the NIH traumatic coma data Bank. J Neurosurg 73:688–698. doi:10.3171/jns.1990.73.5.0688

11. Nazir S, O’Brien M, Qureshi NH, et al. (2009) Sensitivity ofpapilledema as a sign of shunt failure in children. J AAPOS 13:63–66. doi:10.1016/j.jaapos.2008.08.003

12. Rangwala LM, Liu GT (2007) Pediatric idiopathic intracranialhypertension. Surv Ophthalmol 52:597–617

13. Tuite G, ChongWK, Evanson J, et al. (1996) The effectiveness ofpapilledema as an indicator of raised intracranial pressure in chil-dren with craniosynostosis. Neurosurgery 38:272–278

14. Toutant SM, Klauber MR, Marshall LF, et al. (1984) Absent orcompressed basal cisterns on first CT scan: ominous predictors ofoutcome in severe head injury. J Neurosurg 61:691–694. doi:10.3171/jns.1984.61.4.0691

15. Mizutani T, Manaka S, Tsutsumi H (1990) Estimation of intracra-nial pressure using computed tomography scan findings in patientswith severe head injury. Surg Neurol 33:178–184

16. Chen W, Belle A, Cockrell C, et al. (2013) Automated midlineshift and intracranial pressure estimation based on brain CT im-ages. J Vis Exp. doi:10.3791/3871

17. Miller MT, Pasquale M, Kurek S, et al. (2004) Initial head com-puted tomographic scan characteristics have a linear relationshipwith initial intracranial pressure after trauma. J Trauma 56:967–972 discussion 972–973

18. Kouvarellis AJ, Rohlwink UK, Sood V, et al. (2011) The relation-ship between basal cisterns on CT and time-linked intracranialpressure in paediatric head injury. Childs Nerv Syst 27:1139–1144. doi:10.1007/s00381-011-1464-3

19. Bailey BM, Liesemer K, Statler KD, et al. (2012) Monitoring andprediction of intracranial hypertension in pediatric traumatic braininjury: clinical factors and initial head computed tomography. JTrauma Acute Care Surg 72:263–270. doi:10.1097/TA.0b013e31822a9512

20. Tasker RC, Matthew DJ, Kendall B (1990) Computed tomogra-phy in the assessment of raised intracranial pressure in non-traumatic coma. Neuropediatrics 21:91–94. doi:10.1055/s-2008-1071469

21. Killer HE, Laeng HR, Flammer J, Groscurth P (2003) Architectureof arachnoid trabeculae, pillars, and septa in the subarachnoidspace of the human optic nerve: anatomy and clinical consider-ations. Br J Ophthalmol 87:777–781

22. Sekhon MS, Griesdale DE, Robba C, et al. (2014) Optic nervesheath diameter on computed tomography is correlated with si-multaneously measured intracranial pressure in patients with se-vere traumatic brain injury. Intensive Care Med 40:1267–1274.doi:10.1007/s00134-014-3392-7

23. Pauwels EKJ, Bourguignon MH (2012) Radiation dose featuresand solid cancer induction in pediatric computed tomography.Med Princ Pract 21:508–515. doi:10.1159/000337404

24. Raksin PB, Alperin N, Sivaramakrishnan A, et al. (2003)Noninvasive intracranial compliance and pressure based on dy-namic magnetic resonance imaging of blood flow and cerebrospi-nal fluid flow: review of principles, implementation, and othernoninvasive approaches. Neurosurg Focus 14:e4

25. Alperin NJ, Lee SH, Loth F, et al. (2000) MR-intracranial pressure(ICP): a method to measure intracranial elastance and pressurenoninvasively by means of MR imaging: baboon and humanstudy. Radiology 217:877–885. doi:10.1148/radiology.217.3.r00dc42877

26. Muehlmann M, Koerte IK, Laubender RP, et al. (2013) Magneticresonance-based estimation of intracranial pressure correlates withventriculoperitoneal shunt valve opening pressure setting in chil-dren with hydrocephalus. Investig Radiol 48:543–547. doi:10.1097/RLI.0b013e31828ad504

Childs Nerv Syst (2016) 32:1371–1386 1383

27. Leliefeld PH, Gooskens RHJM, Vincken KL, et al. (2008)Magnetic resonance imaging for quantitative flow measurementin infants with hydrocephalus: a prospective study. J NeurosurgPediatr 2:163–170. doi:10.3171/PED/2008/2/9/163

28. Mase M, Yamada K, Banno T, et al. (1998) Quantitative analysisof CSF flow dynamics using MRI in normal pressure hydroceph-alus. Acta Neurochir Suppl 71:350–353

29. Poca MA, Sahuquillo J, Busto M, et al. (2002) Agreement be-tween CSF flow dynamics in MRI and ICP monitoring in thediagnosis of normal pressure hydrocephalus. Sensitivity and spec-ificity of CSF dynamics to predict outcome. Acta Neurochir Suppl81:7–10

30. Geeraerts T, Newcombe VFJ, Coles JP, et al. (2008) Use of T2-weighted magnetic resonance imaging of the optic nerve sheath todetect raised intracranial pressure. Crit Care 12:R114. doi:10.1186/cc7006

31. Kimberly HH, Noble VE (2008) Using MRI of the optic nervesheath to detect elevated intracranial pressure. Crit Care 12:181.doi:10.1186/cc7008

32. Kalantari H, Jaiswal R, Bruck I, et al. (2013) Correlation of opticnerve sheath diameter measurements by computed tomographyand magnetic resonance imaging. Am J Emerg Med 31:1595–1597. doi:10.1016/j.ajem.2013.07.028

33. Shirodkar CG, Rao SM, Mutkule DP, et al. (2014) Optic nervesheath diameter as a marker for evaluation and prognostication ofintracranial pressure in Indian patients: an observational study.Indian J Crit Care Med 18:728–734. doi:10.4103/0972-5229.144015

34. Qayyum H, Ramlakhan S (2013) Can ocular ultrasound predictintracranial hypertension? A pilot diagnostic accuracy evaluationin a UK emergency department. Eur J Emerg Med 20:91–97. doi:10.1097/MEJ.0b013e32835105c8

35. Wang L, Feng L, Yao Y, et al. (2015) Optimal optic nerve sheathdiameter threshold for the identification of elevated opening pres-sure on lumbar puncture in a Chinese population. PLoS One 10:e0117939. doi:10.1371/journal.pone.0117939

36. Amini A, Kariman H, Arhami Dolatabadi A, et al. (2013) Use ofthe sonographic diameter of optic nerve sheath to estimate intra-cranial pressure. Am J Emerg Med 31:236–239. doi:10.1016/j.ajem.2012.06.025

37. Maissan IM, Dirven PJAC, Haitsma IK, et al. (2015)Ultrasonographic measured optic nerve sheath diameter as anaccurate and quick monitor for changes in intracranial pressure.J Neurosurg 1–5. doi: 10.3171/2014.10.JNS141197

38. Nabeta HW, Bahr NC, Rhein J, et al. (2014) Accuracy of nonin-vasive intraocular pressure or optic nerve sheath diameter mea-surements for predicting elevated intracranial pressure inCryptococcal meningitis. Open forum Infect Dis 1:1–8. doi:10.1093/o

39. Strumwasser A, Kwan RO, Yeung L, et al. (2011) Sonographicoptic nerve sheath diameter as an estimate of intracranial pressurein adult trauma. J Surg Res 170:265–271. doi:10.1016/j.jss.2011.03.009

40. Shofty B, Ben-Sira L, Constantini S, et al. (2012) Optic nervesheath diameter on MR imaging: establishment of norms andcomparison of pediatric patients with idiopathic intracranial hy-pertension with healthy controls. AJNR Am J Neuroradiol 33:366–369. doi:10.3174/ajnr.A2779

41. Ballantyne J, Hollman AS, Hamilton R, et al. (1999) Transorbitaloptic nerve sheath ultrasonography in normal children. ClinRadiol 54:740–742

42. Körber F, Scharf M, Moritz J, et al. (2005) Sonography of theoptical nerve – experience in 483 children. RöFo 177:229–235.doi:10.1055/s-2004-813936

43. Ballantyne S, O’Neill G, Hamilton R, HollmanA (2002) Observervariation in the sonographic measurement of optic nerve sheath

diameter in normal adults. Eur J Ultrasound 15:145–149. doi:10.1016/S0929-8266(02)00036-8

44. Frisén L (1982) Swelling of the optic nerve head: a stagingscheme. J Neurol Neurosurg Psychiatry 45:13–18

45. Steffen H, Eifert B, Aschoff A, et al. (1996) The diagnostic valueof optic disc evaluation in acute elevated intracranial pressure.Ophthalmol (Rochester, Minn) 103:1229–1232

46. Digre KB, Nakamoto BK, Warner JEA, et al. (2009) A compari-son of idiopathic intracranial hypertension with and withoutpapilledema. Headache 49:185–193. doi:10.1111/j.1526-4610.2008.01324.x

47. Mathew NT, Ravishankar K, Sanin LC (1996) Coexistence ofmigraine and idiopathic intracranial hypertension withoutpapilledema. Neurology 46:1226–1230

48. Aylward SC, Aronowitz C, Roach ES (2015) Intracranial hyper-tension without papilledema in children. J Child Neurol. doi:10.1177/0883073815587029

49. Faz G, Butler IJ, KoenigMK (2010) Incidence of papilledema andobesity in children diagnosed with idiopathic benign^ intracranialhypertension: case series and review. J Child Neurol 25:1389–1392. doi:10.1177/0883073810364853

50. Chen TC, Zeng A, Sun W, et al. (2008) Spectral domain opticalcoherence tomography and glaucoma. Int Ophthalmol Clin 48:29–45

51. Sihota R, Sony P, Gupta V, et al. (2006) Diagnostic capability ofoptical coherence tomography in evaluating the degree ofglaucomatous retinal nerve fiber damage. Invest Ophthalmol VisSci 47:2006–2010. doi:10.1167/iovs.05-1102

52. Hee MR, Baumal CR, Puliafito CA, et al. (1996) Optical coher-ence tomography of age-related macular degeneration and choroi-dal neovascularization. Ophthalmology 103:1260–1270

53. Driessen C, Eveleens J, Bleyen I, et al. (2014) Optical coherencetomography: a quantitative tool to screen for papilledema in cra-niosynostosis. Childs Nerv Syst 30:1067–1073. doi:10.1007/s00381-014-2376-9

54. Dagi LR, Tiedemann LM, Heidary G, et al. (2014) Using spectral-domain optical coherence tomography to detect optic neuropathyin patients with craniosynostosis. J AmAssoc Pediatr OphthalmolStrabismus 18:543–549. doi:10.1016/j.jaapos.2014.07.177

55. Gabriele ML, Ishikawa H, Wollstein G, et al. (2007) Peripapillarynerve fiber layer thickness profile determined with high speed,ultrahigh resolution optical coherence tomography high-densityscanning. Invest Ophthalmol Vis Sci 48:3154–3160. doi:10.1167/iovs.06-1416

56. Varma R, Bazzaz S, Bai M (2003) Optical tomography-measuredretinal nerve fiber layer thickness in normal latinos. InvestOphthalmol Vis Sci 44:3369–3373

57. Turk A, Ceylan OM, Arici C, et al. (2012) Evaluation of the nervefiber layer and macula in the eyes of healthy children usingspectral-domain optical coherence tomography. Am JOphthalmol 153:552–559. doi:10.1016/j.ajo.2011.08.026

58. Leung MMP, Huang RYC, Lam AKC (2010) Retinal nerve fiberlayer thickness in normal Hong Kong chinese children measuredwith optical coherence tomography. J Glaucoma 19:95–99. doi:10.1097/IJG.0b013e3181a98cfa

59. Yanni S, Wang J, Cheng C, et al. (2013) Normative referenceranges for the retinal nerve fiber layer, macula, and retinal layerthicknesses in children. Am J Ophthalmol 155:354–360. doi:10.1097/WAD.0b013e3181aba588.MRI

60. El-DairiMA, Asrani SG, Enyedi LB, Freedman SF (2009) Opticalcoherence tomography in the eyes of normal children. ArchOphthalmol 127:50–58. doi:10.1001/archophthalmol.2008.553

61. Sasapin P, Freedman S, Lokhnygina Y, et al. (2012) Longitudinalreproducibility of optical coherence tomography measurements inchildren. J Am Assoc Pediatr Ophthalmol Strabismus 29:997–1003. doi:10.1016/j.biotechadv.2011.08.021.Secreted

1384 Childs Nerv Syst (2016) 32:1371–1386

62. Vartin CV, Nguyen AM, Balmitgere T, et al. (2012) Detection ofmild papilloedema using spectral domain optical coherence tomog-raphy. Br J Ophthalmol 96:375–379. doi:10.1136/bjo.2010.199562

63. Ahuja S, Anand D, Dutta TK, et al. (2015) Retinal nerve fiberlayer thickness analysis in cases of papilledema using optical co-herence tomography—a case control study. Clin NeurolNeurosurg 136:95–99. doi:10.1016/j.clineuro.2015.05.002

64. Scott CJ, KardonRH, LeeAG, et al. (2010) Diagnosis and gradingof papilledema in patients with raised intracranial pressure usingoptical coherence tomography vs clinical expert assessment usinga clinical staging scale. Arch Ophthalmol 128:705–711. doi:10.1001/archophthalmol.2010.94

65. Kaufhold F, Kadas EM, Schmidt C, et al. (2012) Optic nerve headquantification in idiopathic intracranial hypertension by spectraldomain OCT. PLoS One 7:e36965. doi:10.1371/journal.pone.0036965

66. Skau M, Yri H, Sander B, et al. (2012) Diagnostic value of opticalcoherence tomography for intracranial pressure in idiopathic intra-cranial hypertension. Graefes Arch Clin Exp Ophthalmol:567–574. doi:10.1007/s00417-012-2039-z

67. Group OS-SC for the NIIHS (2014) Baseline OCT measurementsin the idiopathic intracranial hypertension treatment trial, part II:correlations and relationship to clinical features. InvestOphthalmol Vis Sci 55:8173–8179. doi:10.1167/iovs.14-14961

68. Skau M, Milea D, Sander B, et al. (2011) OCT for optic discevaluation in idiopathic intracranial hypertension. Graefes ArchClin Exp Ophthalmol 249:723–730. doi:10.1007/s00417-010-1527-2

69. Klingelhöfer J, Conrad B, Benecke R, et al. (1988) Evaluation ofintracranial pressure from transcranial Doppler studies in cerebraldisease. J Neurol 235:159–162

70. Gosling RG, King DH (1974) Arterial assessment by Doppler-shift ultrasound. Proc R Soc Med 67:447–449

71. Michel E, Zernikow B (1998) Gosling’s Doppler pulsatility indexrevisited. Ultrasound Med Biol 24:597–599

72. (2004) AAN Guideline Summary for Clinicians: Assessment ofTranscranial Doppler Ultrasonography.

73. Ragauskas A, Bartusis L, Piper I, et al. (2014) Improved diagnos-tic value of a TCD-based non-invasive ICP measurement methodcompared with the sonographic ONSD method for detecting ele-vated intracranial pressure. Neurol Res 36:607–614. doi:10.1179/1743132813Y.0000000308

74. Bellner J, Romner B, Reinstrup P, et al. (2004) TranscranialDoppler sonography pulsatility index (PI) reflects intracranialpressure (ICP). Surg Neurol 62:45–51 . doi:10.1016/j.surneu.2003.12.007discussion 51

75. Wang Y, Duan Y-Y, Zhou H-Y, et al. (2014) Middle cerebral arte-rial flow changes on transcranial color and spectral Doppler so-nography in patients with increased intracranial pressure. JUltrasound Med 33:2131–2136. doi:10.7863/ultra.33.12.2131

76. Hunter G, Voll C, Rajput M (2010) Utility of transcranial dopplerin idiopathic intracranial hypertension. Can J Neurol Sci 37:235–239

77. Prunet B, Asencio Y, Lacroix G, et al. (2012) Noninvasive detec-tion of elevated intracranial pressure using a portable ultrasoundsystem. Am J Emerg Med 30:936–941. doi:10.1016/j.ajem.2011.05.005

78. Wakerley BR, Kusuma Y, Yeo LLL, et al. (2014) Usefulness oftranscranial doppler-derived cerebral hemodynamic parameters inthe noninvasive assessment of intracranial pressure. JNeuroimaging 25:111–116. doi:10.1111/jon.12100

79. Rainov NG, Weise JB, Burkert W (2000) Transcranial Dopplersonography in adult hydrocephalic patients. Neurosurg Rev 23:34–38

80. Zweifel C, Czosnyka M, Carrera E, et al. (2012) Reliability of theblood flow velocity pulsatility index for assessment of intracranial

and cerebral perfusion pressures in head-injured patients.Neurosurgery71:853–861.doi:10.1227/NEU.0b013e3182675b42

81. Behrens A, Lenfeldt N, Ambarki K, et al. (2010) TranscranialDoppler pulsatility index: not an accurate method to assess intra-cranial pressure. Neurosurgery 66:1050–1057

82. Hanlo PW, Gooskens RH, Nijhuis IJ, et al. (1995) Value of trans-cranial Doppler indices in predicting raised ICP in infantile hydro-cephalus. A study with review of the literature. Childs Nerv Syst11:595–603

83. Ahmad M, Legrand M, Lukaszewicz A-C, et al. (2013)Transcranial Doppler monitoring may be misleading in predictionof elevated ICP in brain-injured patients. Intensive Care Med 39:1150–1151. doi:10.1007/s00134-013-2885-0

84. Banich M, Compton R (2011) Cognitive neuroscience, 3rd edn.Cengage Learning, Belmont

85. Creel D (2015) Visually Evoked Potentials. In: Webvision. http://webvision.med.utah.edu/book/electrophysiology/visually-evoked-potentials/. Accessed 22 Jul 2015

86. Sklar FH, Ehle AL, Clark WK (1979) Visual evoked potentials: anoninvasive technique to monitor patients with shunted hydro-cephalus. Neurosurgery 4:529–534

87. Fichsel H (1976) Diagnosis of hydrocephalus. Changes in visualevoked potentials in children with progressive hydrocephalusinternus. Fortschr Med 94:1141–1142

88. Sjöström A, Uvebrant P, Roos A (1995) The light-flash-evokedresponse as a possible indicator of increased intracranial pressurein hydrocephalus. Childs Nerv Syst 11:381–387 discussion 387

89. VieiraMADCS, Cavalcanti MDAS, Costa DL, et al. (2015) Visualevoked potentials show strong positive association with intracra-nial pressure in patients with cryptococcal meningitis. ArqNeuropsiquiatr 73:309–313. doi:10.1590/0004-282X20150002

90. York DH, PulliamMW, Rosenfeld JG,Watts C (1981) Relationshipbetween visual evoked potentials and intracranial pressure. JNeurosurg 55:909–916. doi:10.3171/jns.1981.55.6.0909

91. GumerlockMK,YorkD, Durkis D (1994) Visual evoked responsesas a monitor of intracranial pressure during hyperosmolar blood-brain barrier disruption. Acta Neurochir Suppl (Wien) 60:132–135

92. Desch LW (2001) Longitudinal stability of visual evoked poten-tials in children and adolescents with hydrocephalus. Dev MedChild Neurol 43:113–117

93. Andersson L, Sjölund J, Nilsson J (2012) Flash visual evokedpotentials are unreliable as markers of ICP due to high variabilityin normal subjects. Acta Neurochir 154:121–127. doi:10.1007/s00701-011-1152-9

94. Sutter EE (2010) Noninvasive testing methods: multifocal electro-physiology. In: Darlene A. Dartt (ed) Encycl. Eye. pp 142–160

95. (2008) Guideline 9B: Guidelines on Visual Evoked Potentials.96. LaManna JC (2007) In situ measurements of brain tissue hemo-

globin saturation and blood volume by reflectance spectrophotom-etry in the visible spectrum. J Biomed Opt 12:062103. doi:10.1117/1.2804184

97. Ghosh A, Elwell C, SmithM (2012) Review article: cerebral near-infrared spectroscopy in adults: a work in progress. Anesth Analg115:1373–1383. doi:10.1213/ANE.0b013e31826dd6a6

98. Czosnyka M, Smielewski P, Kirkpatrick P, et al. (1997)Continuous assessment of the cerebral vasomotor reactivity inhead injury. Neurosurgery 41:11–17 discussion 17–9

99. Zweifel C, Lavinio A, Steiner LA, et al. (2008) Continuous mon-itoring of cerebrovascular pressure reactivity in patients with headinjury. Neurosurg Focus 25:E2. doi:10.3171/FOC.2008.25.10.E2

100. Steiner LA, Czosnyka M, Piechnik SK, et al. (2002) Continuousmonitoring of cerebrovascular pressure reactivity allows determi-nation of optimal cerebral perfusion pressure in patients with trau-matic brain injury. Crit Care Med 30:733–738

Childs Nerv Syst (2016) 32:1371–1386 1385

101. Lee JK, Kibler KK, Benni PB, et al. (2009) Cerebrovascular reac-tivity measured by near-infrared spectroscopy. Stroke 40:1820–1826. doi:10.1161/STROKEAHA.108.536094

102. Zweifel C, Castellani G, Czosnyka M, et al. (2010) Noninvasivemonitoring of cerebrovascular reactivity with near infrared spec-troscopy in head-injured patients. J Neurotrauma 27:1951–1958.doi:10.1089/neu.2010.1388

103. Kampfl A, Pfausler B, Denchev D, et al. (1997) Near infrared spec-troscopy (NIRS) in patients with severe brain injury and elevatedintracranial pressure. A pilot study. Acta Neurochir Suppl 70:112–114

104. Wagner BP, Pfenninger J (2002) Dynamic cerebral autoregulatoryresponse to blood pressure rise measured by near-infrared spec-troscopy and intracranial pressure. Crit Care Med 30:2014–2021.doi:10.1097/01.CCM.0000025889.96603.B0

105. Blaivas M, Theodoro D, Sierzenski PR (2003) Elevated intracra-nial pressure detected by bedside emergency ultrasonography ofthe optic nerve sheath. Acad Emerg Med 10:376–381. doi:10.1111/j.1553-2712.2003.tb01352.x

106. Wall M (2010) Idiopathic intracranial hypertension. Neurol Clin28:593–617. doi:10.1016/j.ncl.2010.03.003

107. Alvarez-Fernández JA, Pérez-Quintero R (2009) Use of transcra-nial Doppler ultrasound in the management of post-cardiac arrestsyndrome. Resuscitation 80:1321–1322. doi:10.1016/j.resuscitation.2009.07.011

108. Ohle R, McIsaac SM, Woo MY, Perry JJ (2015) Sonography ofthe optic nerve sheath diameter for detection of raised intracranialpressure compared to computed tomography: a systematic reviewand meta-analysis. J Ultrasound Med 34:1285–1294. doi: 10.7863/ultra.34.7.1285

1386 Childs Nerv Syst (2016) 32:1371–1386