Article, Neurology

The quantitative evaluation of intracranial pressure by optic nerve sheath diameter/eye diameter CT measurement

a b s t r a c t

Background: The changes of the optic nerve sheath diameter (ONSD) have been used to assess changes of the in- tracranial pressure for 20 years. The aim of this research was to further quantify the technique of measuring the ONSD for this purpose.

Methods: Retrospective study of computed tomographic (CT) data of 1766 adult patients with intracranial hypo- tension (n = 134) or hypertension (n = 1632) were analyzed. The eyeball transverse diameter (ETD) and ONSD were obtained bilaterally, and the ONSD/ETD ratio was calculated. The ratio was used to calculate the normal ONSD for patients and to estimate the intracranial pressure of the patients before and after the onset of the pa- thology. Correlation analysis was performed with invasively measured intracranial pressure, the presence or ab- sence of papilledema, sex, and age.

Results: In hypotension cases, the ONSD by CT was 3.4 +- 0.7 mm (P = .03 against normative 4.4 +- 0.8 mm). In cases with hypertension, the diameter was 6.9 +- 1.3 (P = .02, with a cutoff value 5.5 mm). The ONSD/ETD ratio was 0.29 +- 0.04 against 0.19 +- 0.02 in Healthy adults (P = .01).

Conclusion: The ONSD and the ONSD/ETD ratio can indicate low intracranial pressure, but quantification is impos- sible at intracranial pressure less than 13 mm Hg. In Elevated intracranial pressure, the ONSD and the ratio pro- vide readings that correspond to readings in millimeters of mercury. The ONSD method, reinforced with additional calculations, may help to indicate a raised intracranial pressure, evaluate its severity quantitatively, and establish quantitative goals for treatment of Intracranial hypertension, but the limitations of the method are to be taken into account.

(C) 2016

Introduction

Invasive methods of measuring intracranial pressure such as External ventricular drainage and the use of microtransducer devices provide a practitioner with quantitative reading of ICP expressed in millimeters of mercury. noninvasive methods of measuring are either predominantly qualitative (such as tympanic membrane displacement, optic nerve sheath diameter [ONSD], fundoscopy, papilledema, and assessment of magnetic resonance imaging [MRI] and computed

? This research received no specific grant from any funding agency in the public, com- mercial or not-for-profit sectors.

?? Authors’ contribution: Michael Vaiman, research concept design, analysis of the data, manuscript draft and editing, and final version approval; Tal Sigal, Anna Ben Ely, and Inessa Bekerman, radiologic data collection, measurements, data analysis, and final version ap-

proval; Itzhak Kimiagar, neurologic data collection, measurements, assessment and data analysis, and final version approval.

* Corresponding author at: 33 Shapiro Street, Bat Yam 59561, Israel. Tel.: +972 3 553

6139; fax: +972 3 553 6137.

E-mail address: [email protected] (M. Vaiman).

tomographic [CT] scans to distinguish between normal and increased ICP) or provide imprecise readings (such as transcranial Doppler ultra- sonography) [1-3].

An assessment of ICP by measuring changes in ONSD became practi- cal in the 1990s and is based on the fact that the presence of the en- larged optic nerve sheaths suggests that raised ICP is transmitted intraorbitally [4-6]. Since then, ONSD has been used to assess ICP in pa- tients with traumatic and nontraumatic intracerebral hemorrhage, traumatic brain injury, idiopathic intracerebral hypertension, meningi- tis, cardiac arrest, and various other disorders that may elevate ICP [1,6-10]. Since 1996, 173 articles on the ONSD assessment of ICP have been published in the PubMed. In these studies, ICP was measured by CT, ultrasonography, and rarely by MRI. In general, the authors were sat- isfied with the method and indicated correlation between ONSD and invasively measured ICP (ICP?, ONSD?), but some uncertainty about the accuracy of the method was also expressed. The ONSD method has been used for 20 years; multiple protocols and thresholds have been proposed, but no generally accepted protocol or standardization was se- lected. This methodology is not the same as the criterion standard of measuring ICP using the lumbar puncture or other direct method, and

http://dx.doi.org/10.1016/j.ajem.2016.08.045

0735-6757/(C) 2016

thus, the reliability, validity, reproducibility, and generalizability of the results require additional clarifications.

Because no quantitative connection between ICP expressed in milli- meters of mercury and ONSD expressed in millimeters was established beyond the assessment of ICP greater than 20 mm Hg which is consid- ered pathologic, a normal/abnormal cutoff value of ONSD set by differ- ent researchers varied from 4.8 to 7.3 mm [6-12]. Another obstacle is that both normal and abnormal readings of ONSDs reported in the same studies have wide standard deviations which frequently over- lapped. To improve this situation, the use of either optic nerve diameter to ONSD ratio for ultrasonography or ONSD to the eyeball transverse di- ameter (ETD) ratio for CT investigations has been suggested as possible indexes [13,14]. These indexes have insignificant SD; for example, the ONSD/ETD index for healthy adults is 0.19 +- 0.02 if measured in the middle of the optic nerve intraorbital path [13]. To reduce the SD of ONSD readings, it has also been suggested that measurement of the ONSD farther from the globe would minimize the impact of voluntary and involuntary eyeball movements [13,15-17]. Although these studies presented normative databases, the rationale of measuring the ONSD at a point where the ophthalmic artery crosses the optic nerve was further explained for cases with the traumatic brain injury and pseudotumor cerebri [18,19]. The implementation of the ONSD/ETD index into clinical practice was described for cases of the traumatic brain injury with hem- orrhage [18].

The aim of the current study was to summarize previously obtained

data and to investigate possible ways of establishing quantitative results for the ONSD method. Ideally, we wanted to find a way to establish a correspondence between ONSD readings in millimeters and ICP readings in millimeters of mercury. If this proves impossible, the subsid- iary objective was to set rules of usage for the ONSD/ETD index in clinical practice.

Materials and methods

Background and the methodological considerations

In various sources, the normal ICP is presented as 7 to 15, 5 to 15, or 0 to 15 mm Hg for a supine adult. The ONSD consists of the diameter of the optic nerve itself (2-3 mm), the thickness of dura mater (0.5 mm

x 2 = 1 mm approximately), and a space for cerebrospinal fluid (CSF) between the nerve and the dura that can increase or decrease according to ICP changes. This perineural subarachnoid space is small, holding approximately 0.1 mL of CSF. The first methodological problem was to find the lowest value of ONSD decrease (despite the ONSD/ICP correlation, if the ICP drops to 0 mm Hg the ONSD would not drop to 0 mm). For this purpose, it was necessary to analyze patients with in- tracranial hypotension.

The second methodological problem was to find a quantitative cor- relation between millimeters of mercury of ICP and millimeters of ONSD. For that purpose, a group of patients with simultaneous invasive and noninvasive ONSD measurements of ICP was needed to calculate the sensitivity and specificity of the method. The data were compared with the normative database.

Patients, setting, and Inclusion/exclusion criteria

In a retrospective study, we collected and analyzed the CT data of 1766 adult patients (18+) who had been seen to the department of ra- diology from January 2009 to February 2016. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki (amended 2007) as reflected a priori after approval by the Helsinki committee of our medical center.

For the current study, we included the patients who were admitted to the emergency department (ED), were referred for CT investigation which included the head and neck region, and appeared to have various pathologies usually associated with elevation of ICP ([a] traumatic and

nontraumatic intracerebral or subarachnoid hemorrhage, [b] combined type of hemorrhage associated with trauma, [c] hemorrhagic stroke or Vascular abnormalities, [d] brain tumor, [e] traumatic brain injury pre- senting Brain swelling, [f] pseudotumor cerebri, and [g] meningitis) or with the intracranial hypotension ([h] fresh open head injury with a cra- nial wound open to atmospheric pressure and documented leaks of the CSF and blood, [i] atraumatic CSF leak secondary to structural dural weakness, and [j] postsurgical CSF leaks). Subsequent investigation of these patients was performed after they were hospitalized in the inten- sive care unit, the neurologic department, for head and neck surgery, or other units of the hospital.

Cases with traumas/fractures that could affect the sphenoid bone and the orbit were excluded from the study. Patients with known oph- thalmological or neuroophthalmological disorders were also excluded. We also excluded charts with incomplete data.

The patients were divided into 2 groups for 2 separate investiga- tions. A total of 134 patients were selected with open traumatic brain in- jury and with documented external hemorrhage and leakage of CSF via open cranial wound, atraumatic CSF leak secondary to structural dural weakness, and postsurgical CSF leakage that might suggest intracranial hypotension (group 1). Only CT scans of cases with open traumatic brain injury admitted to the ED no later than 2 hours after the Traumatic event were examined because evolving posttraumatic edema of the brain can change the ICP readings. The selection of such cases was lim- ited to those with clear indication of the state of Brain edema in the charts. The charts also indicated that initial CT scans were performed be- fore any dural aperture (grafting/closure). We assumed that a direct connection between the brain and the outside and CSF hypovolemia would equalize ICP to atmospheric pressure before edema occurs. Con- sequently, these patients should have ICP close or equal to 0 mm Hg (ICP is the pressure inside the cranial cavity relative to atmospheric pres- sure). A subgroup 1A consisted of 42 patients with invasively measured ICP and subsequent CT scans (at least 3 scans) during recovery after wound closure and restoral of the integrity of the cranial cavity. Com- puted tomographic investigation was carried out for all 134 patients upon arrival at the ED, and ETD and ONSD were measured bilaterally using the same initial CT scan. An exclusion criterion for this group was traumatically affected orbit and/or sphenoid bone.

The group 2 consisted of patients with elevated ICP because of closed

head trauma with or without intracerebral hemorrhage, hemorrhagic stroke, pseudotumor cerebri, meningitis, and brain tumor. All 1632 pa- tients in this group also underwent initial CT investigation at the ED as well as subsequent CT investigations, to exclude delayed posttraumatic hemorrhage or to monitor their condition during hospitalization. For them, ETD and ONSD were also measured using the same initial CT scan and subsequent scans. A subgroup 2A consisted of 592 patients with invasively measured and monitored ICP and subsequent CT scans (>=3 scans) during recovery which can be used for subsequent measurements of ETD and ONSD. Two inclusion criteria were applied:

(1) CT scans of the patients were required to indicate a pathology that could lead to intracranial hypertension (focal and/or diffuse hyperdense areas for acute epidural or subdural hematomas, subarach- noid, intraparenchymal, or intraventricular hemorrhage; hypodense areas for brain swelling; abnormally small ventricles, etc), and (2) the charts should include the conclusion of an ophthalmologist on presence or absence of papilledema. Exclusion criteria for this group eliminated from the study all cases with penetrating brain injury and also patients with traumas/fractures that could affect the sphenoid bone and the orbit. Patients with known or suspected ophthalmological or neuroophthalmological disorders were also excluded. The patients’ flow diagram of the study is presented in Fig. 1.

Data sources and measurements

We analyzed axial (transverse) cuts of the CT scans obtained by the 256-slice CT scanner (Brilliance iCT; Philips Healthcare) with an initial

Assessed for eligibility (n = 2013)

Selected (n = 1766)

Analysed (n = 1632)

Excluded from analysis (n = 0)

Analysed (n = 134)

Excluded from analysis (n = 0)

Analysis

Invasive ICP monitoring

With invasive ICP monitoring (n = 42) With invasive ICP monitoring (n = 592)

Without invasive ICP monitoring (n = 92) Without invasive ICP monitoring (n = 1040)

Allocated to HYPERTENSION group (n = 1632) Closed TBI with hemorrhage (n = 335) Closed TBI without hemorrhage (n = 706)

Stroke (n = 471); Other (n = 120)

Allocated to HYPOTENSION group (n = 134) TBI with CSF leakage (n = 99)

Atraumatic CSF leakage (n = 17)

Post-surgical CSF leakage (n = 18)

Allocation

Excluded (n = 247)

Not meeting inclusion criteria (n = 211) Incomplete data (n = 36)

Fig. 1. The patients’ flow diagram of the study. Abbreviations: TBI, traumatic brain injury; CSF, cerebrospinal fluid.

single-slice section of 3 mm and the area of interest slice section of 0.6

Enrollment

mm. The left and right ETD (retina to retina) and the ONSD were mea- sured by the computer program at the same CT scan (Fig. 2A and B) and normal ONSD for a given patient was calculated (ETD x 0.19 = ONSD). The point where the ophthalmic artery crosses the optic nerve was chosen as an anatomical landmark to measure ONSD. Usually, this cross is located 8 to 12 mm from the globe, and this point is not affected by tremor, gaze deviations, and involuntary movements of the eyes after trauma or stroke.

Window parameters used for CT scans were spine window, middle third; WW 60, WL 360 (sometimes abbreviated as C:60,0. W:360,0 spine), accuracy 1 pixel. All analyzed measurements were made using the same window, contrast, and brightness. The error margin was expressed by means of the technical error of measurement (TEM) to cal- culate the interevaluator variability and intraevaluator variability among 3 evaluators, whereas each of them served as an independent control for the remaining 2. The same equipment and methodological procedures for measurements were adopted by all evaluators.

During 2009 to 2016, ICP was measured by lumbar puncture, opening pressure manometer, external ventricular drainage, and microtransducers. For both groups, the radiologists who retrospectively measured the ONSD were not informed about invasive ICP readings. The radiologic data were collected independently and subsequently compared to inva- sive ICP readings.

Analysis

A within-group repeated-measurement experimental statistical analysis was used to evaluate the variables. Normal probability plots

and basic descriptive statistics (mean, SD, minimum, and maximum) were calculated for every variable. The correlation analysis was per- formed with gender and age groups (group I, 18-30; group II, 30-65; group III, 65+) and between the ONSD and invasively measured ICP and ONSD/ETD index and invasively measured ICP. Previously reported normal values for ETD, ONSD, and ONSD/ETD index served as a norma- tive database and were used for comparison purposes [13,15-27]. The cutoff value for pathologically enlarged ONSD was set at greater than

5.5 mm because 5 mm (as indicated by several authors) is within the normative range [13,15,16] and the cutoff value for pathologically en- larged ONSD/ETD index was set at 0.22 (assuming that normal index is 0.19 +- 0.02). In regression analysis, the obtained data were fitted by a method of successive approximations. The data were statistically eval- uated by 3-dimensional analysis of variance; SPSS, standard version

17.0 (2007; SPSS, Chicago, IL); and ?2 criterion using 95% confidence in- terval. The level of significance for all analyses was set at P b .05.

Results

Of 1766 patients (990 males and 776 females; mean age, 38 +- 17 years), there were 335 patients with closed traumatic brain injury with hemorrhage, 99 patients with open traumatic brain injury with external hemorrhage from intracranial vessels and CSF leakage, 17 pa- tients with atraumatic CSF leakage secondary to structural dural weak- ness, 18 patients with postsurgical CSF leakage, 706 patients with closed traumatic brain injury without hemorrhage presenting brain swelling/ edema, 471 patients with hemorrhagic stroke, 36 patients with brain tumor (our hospital has no special unit for these patients which is why the number was small), 45 patients with pseudotumor cerebri,

Fig. 2. A, Axial noncontrast CT scan is used to measure the ETD retina to retina (24 mm). In many cases, the largest ETD and the ONSD are located in different horizontal planes and are measured using different CT slices. B, The measurement of the ONSD of the same patient indicates enlarged diameter of 7.1 mm. The ONSD/ETD ratio is 0.29 against normative 0.19. The white arrow shows the ophthalmic artery crossing the optic nerve (anatomical landmark).

and 39 patients with meningitis. Mortality rate was 14.6% (n = 258), median admission Glasgow Coma Scale score of 8 (interquartile range, 5-10) (except for patients with brain tumor and pseudotumor cerebri to whom the scale was not applied).

The average number of CT scans per patient was 2.5 (range, 2-6), and altogether, we examined 4365 CT scans. We measured 3532 ETDs and 3532 ONSDs (bilaterally). Eyeball transverse diameters were measured once, and ONSDs were measured for all subsequent CT scans. For the TEM calculation, 2 measurements were obtained by 2 evaluators from each variable and the third evaluator served as a judge. The difference between the first and second measurements, for both the interevaluator variability and the intraevaluator variability, was then determined, and the relative TEM (expressed in percentages) was calculated. For intraevaluator TEM, it was 2.88 for the ETD, acceptable, and 3.56 for the optic nerve sheath, acceptable. For interevaluator TEM, it ranged from 3.8 to 3.95, respectively, for evaluator to evaluator and evaluator to judge comparisons ( 5%, acceptable).

Group 1: suspected intracranial hypotension

Table 1 presents the results of the measurements in this group. The difference in measurements between left and right eyes was statistically insignificant (P = .46) and is not presented in the table. The ONSD in cases with intracranial hypotension was significantly lower than normal and showed sensitivity of 98% and specificity of 77%. Optic nerve sheath diameter/ETD index correlated well with ICP both in initial readings (ICP?, ONSD?) and during recovery (ICP?, ONSD?). The sensitivity of the index was 99%, but specificity was only 72% because the index can- not decrease below 0.12 (usually, it stops at 0.13 or 0.14), and correla- tion with ICP less than 5 mm Hg is not possible. (This is because most of CSF leaves the perineural subarachnoid space at ICP reading between 4 and 5 mm Hg and no further reduction of ONSD is possible, while ICP can drop to 0 mm Hg.) During recovery, the maximally reduced ONSD remained unchanged until ICP reaches 4 and 5 mm Hg. After that, an initial enlargement followed and it again remained unchanged or insignificantly increases. (ONSD at 5 mm Hg ICP was almost the same as ONSD at 10 mm Hg in most of the monitored cases. For most of these patients, further increase of ONSD could be traced when ICP reached 13 and 15 mm Hg ICP provided significant further enlargement of the diameter.)

Group 2: intracranial hypertension

Table 2 presents the results of the measurements in this group. No correlation was found between the ONSD/ETD ratio and the sex of the patients or their age; however, comparison between age group I (18- 30 years) and age group III (65+ years) showed some differences. In older patients, an increase of ONSD was less prominent when compared with young patients but remained statistically insignificant (P = .07). In both groups, ONSD in cases with intracranial hypertension was signifi- cantly higher than normal and showed sensitivity of 90% and specificity of 86%. Optic nerve sheath diameter/ETD index correlated well with ICP both during the development of the pathology (ICP?, ONSD?) and dur- ing recovery (ICP?, ONSD?). The sensitivity of the index was 96%, and specificity was 100% for measurements in a sector greater than 17 mm Hg. In a sector between 20 and 30 mm Hg, the increase and decrease of both ONSD and ONSD/ETD index showed linear regression. At 30 to 32 mm Hg and more, the regression became nonlinear with constant error variance. For ICP readings higher than 30 mm Hg, the ICP rises more prominently than the ONSD or the index.

Table 1

Optic nerve sheath and eyeball CT measurements (in millimeters) and the nerve sheath/ eye index (ONSD/ETD ratio) in the patients with suspected intracranial hypotension (group 1) and correlations with ICP (in millimeters of mercury)

Variables Measurements

Mean +- SD Max Min P r

Normative database

ONSDa 4.4 +- 0.8 5.6 3.3

Cases with suspected intracerebral hypotension (n = 134)

ONSD upon admission 3.4 +- 0.7 4.3 3.0 .03 vs normal

Subgroup with invasive ICP readings and monitoring (n = 42)

ONSD upon admission 3.5 +- 0.9 4.4 3.0 .04 vs normal

ICP after wound closure 4 +- 2 7 2

ONSD upon discharge 4.5 +- 0.9 5.8 3.2 .88 vs normal

ICP upon discharge 10 +- 6 17 3 .03 vs initial

ETD 22.7 +- 1.5 25.2 20.0

ONSD/ETD index

Normative 0.19 +- 0.02 0.26 0.15

Intracerebral hypotension 0.14 +- 0.03 0.19 0.12 .05 vs normative

Index to ICP correlation 0.86

a Eight to 12 mm behind the globe at a point where the ophthalmic artery crosses the optic nerve.

Table 2

Optic nerve sheath and eyeball CT measurements (in millimeters) and the nerve sheath/ eye index (ONSD/ETD ratio) in the patients with intracranial hypertension (group 2) and correlations with ICP (in millimeters of mercury)

Variables Measurements

Mean +- SD Max Min P r

Normative database

ONSD 4.4 +- 0.8 5.6 3.3

Cases with intracerebral hypertension (n = 1632)

ONSD upon admission 6.9 +- 1.3 9.4 4.0 .02 vs normal Subgroup with invasive ICP readings and monitoring (n = 592)

ONSD upon admission

6.7 +- 0.9

9.4

4.0

.03 vs normal

ICP upon admission

30 +- 6

37

22

ONSD upon dischargea

4.7 +- 0.9

5.7

4.0

.83 vs normal

ICP upon dischargea

17 +- 6

20

13

.02 vs initial

ETD

22.7 +- 1.5

25.2

20.0

ONSD/ETD index

Normative

0.19 +- 0.02

0.26

0.15

Intracerebral hypertension

0.29 +- 0.04

0.37

0.22

0.01 vs. normative

Index to ICP correlation

0.88

a Excluding cases with mortality or transfer to another institution.

In this group of cases, only 844 (52%) of 1632 patients had docu- mented papilledema upon arrival at the ED, with almost complete ab- sence of this sign in patients who were brought to the hospital an hour or 2 after the traumatic brain injury accident or hemorrhagic stroke onset.

Further analysis of the results

Computed tomographic scans provided readings of pathologically enlarged ONSDs of patients. With the help of the ONSD/ETD index, the ONSD a given patient most probably had when in a healthy condition was calculated. To confirm these theoretical calculations, we obtained CT scans of the patients being investigated before the pathologic cases arose. Of 1766 investigated cases, 298 patients had previous admissions to the medical center because of various health problems. They were se- lected from our hospital database. Eighty-nine of them had recent full- body or cranial CT scans performed during the same 2009-2016 period. Of these, in 58 cases, no cerebral pathology was detected. We compared our calculated normal ONSDs with actual ONSDs using these 58 scans and observed an agreement of 93% (54/58).

The dedicated, thin slide, fat-suppressed 3 T orbital MRI images were available for only 31 patients. The agreement between CT and MRI in measuring the ONSD was 93.5% (29/31), but these results are statistical- ly doubtful because of a small sample.

Discussion

We have confirmed previous reports that the ONSD method of ICP evaluation is an effective diagnostic tool in cases with brain injury, intra- cranial and intracerebral hemorrhage, brain swelling, and intracranial space-occupied lesions that might elevate ICP [6-12,21,22]. We have confirmed previous reports that ONSD is accurately measured by CT [10,11,13,15,22,23]. We have also confirmed the recent results obtained by Sekhon et al [23] indicating that a linear regression model shows a strong correlation between ICP and ONSD. Our goal in the current re- search was to add more quantitative meaning to the method. Once the results were obtained, we might discuss how to use the ONSD and ONSD/ETD index in clinical setting quantitatively. We believe that the ONSD method reinforced by additional calculations can achieve 4 basic objectives as follows:

to establish or to confirm the existence of raised ICP,
  • to evaluate its severity quantitatively,
  • to establish quantitative goals for treatment of intracranial hyper- tension, and
  • to provide prognostic information.
  • In the current article, we wanted to show that more information can be obtained from the initial CT scan (Fig. 3A and B). We do not propose a separate additional investigation; we believe that the investigation that was already performed (like obligatory head CT at the ED) can be more informative. Additional data can reinforce other results obtained about the ICP status, thus speeding up decision-making process for treatment strategy.

    To the basic objectives of this study, we may add the benefits of sav- ing time and effort and minimizing complications of invasive ICP mon- itoring. Infections, hemorrhage, improper handling, and decalibration in long-term observations are known complications of invasive ICP moni- toring [24-26]. Risk of these complications increases with the time dur- ing which the monitoring is applied and invasive ICP evaluation can be contraindicated for some patients. There is no need for special CT inves- tigation to obtain the ONSD and ETD readings because they can be

    Fig. 3. A, Axial noncontrast head CT demonstrates a large convex extra-axial collection along the left cerebral convexity. The collection crosses the sutures indicative of a subdural nature. There is mass effect on the subjacent brain parenchyma that is also causing a Midline shift. There is also compression of the left ventricle. B, The measurement of the ONSD of the same patient indicates enlarged diameters from both sides as 6.6 mm. The ETDs of the patient were 22 mm (r) and 22.1 mm (l). The right and left ONSD/ETD ratio is 0.3 against normative 0.19 (P b .05). The ONSD for this patient was 22 mm x 0.19 = 4.2 mm before hemorrhage occurred.

    obtained from the initial and subsequent scans usually performed for patients with head trauma, hemorrhagic stroke, brain tumor, and others. Twenty-four-hour monitoring can be used instead of long- term ICP monitoring, after which further assessment can be performed by the ONSD method using subsequent CT scans and/or ultrasonogra- phy if CT is not indicated for a given patient. In some cases, it is possible to obtain 2 separate invasive ICP readings taken close to the time of CT investigations, thus avoiding invasive monitoring completely.

    An individual approach in applying the ONSD method is recom- mended. There is no universal correlation between ONSD and ICP which can be expressed in a chart. The ONSD/ICP readings correlate within a single patient, but they vary significantly from patient to pa- tient. One patient may present with 6.6 mm ONSD and 35 mm Hg ICP and another patient with 7 mm ONSD and 30 mm Hg ICP. Optic nerve sheath diameter readings depend upon the initial diameter of the sheath and the elasticity, extensibility, and thickness of the dura mater, which vary from individual to individual [27,28]. Therefore, ICP assessment via ONSD should be performed on an individual basis only, by obtaining initial and subsequent ONSD and ICP readings and making calculations specifically for a given patient.

    Limitations of the ONSD method in general

    The method can indicate intracranial hypotension (the ONSD/ETD index 0.17), but no ONSD readings are possible corresponding to the range 0 to 5 mm Hg. Within the range 5 to 13 mm Hg, it is possible to obtain readings, but they are not informative. Furthermore, for intracra- nial hypotension, patients with skull defects are likely to be affected by positioning such that ICP would vary based on whether the breach is faced laterally, down, or up depending on location and head position. Although we believe that the ONSD method could be used for the detec- tion of the intracranial hypotension as well as for the intracranial hyper- tension, additional research is needed for this group of cases, and our results are considered to be preliminary.

    The most accurate readings are obtained for the group with ICP read- ings in the range 15 to 30 mm Hg with strong ONSD/ICP correlation and linear regression. Greater than 30 to 32 mm Hg, the regression become nonlinear most probably because the elasticity and extensibility of the dura mater introduce limitations that slow down the further enlarge- ment of the ONSD. We suggest using a coefficient of 1.3 for 30 to 35 mm Hg and a coefficient 1.6 for 36 to 40 mm Hg. A major limitation of the method is that within these ranges, although measurement of the ONSD at the same time as performing clinical CT offers considerable convenience, the information obtained offers no assurance that these measures are dynamic enough to afford real-time appraisal of Clinical worsening or of Treatment response. In severe cases with rapidly chang- ing ICP, the invasive monitoring remains the method of choice at least until the patient is stabilized.

    Thus, although the ONSD method, especially in combination with other calculations, might eventually prove useful to indicate a raised ICP, this remains to be seen by addressing the above mentioned limitations.

    Reproducibility issues

    Current efforts in standardization of CT nomenclature and protocols for various CT scanner manufacturers (GE, Hitachi, Philips, Toshiba, Sie- mens) make the research fully reproducible if the same CT window pa- rameters will be used (spine window, middle third; WW 60, WL 360). Our data can be reproduced by all modern fourth-generation CT scan- ners produced by the aforementioned manufacturers. Fat-suppressed 3 T orbital MRI images provide identical results in most of the cases. An application of ultrasonography, however, might require an addition- al investigation because a report has been published indicating that CT and ultrasonographic readings of ONSD do not match exactly [29]. We limited the current research to CT images, but our own experience

    indicates that the ultrasonography has a tendency to overestimate the diameter of the sheath.

    Generalizability issues

    The technical error of measurement was less than 5% in the current study. Therefore, we assess the study as internally valid. The rationale for not including only known ICP cases in the study was to avoid selec- tion bias. For example, the invasive ICP measurement was performed mainly in severe cases that do not represent all the patients being ad- mitted to the hospital. Therefore, all consecutive cases with the ICP is- sues were included with no sampling beyond the aforementioned exclusion criteria. This helped to achieve representativeness via the rep- resentative sample of the adult population (both sexes; age range, 18- 92 years). Therefore, we assess the study as externally valid. Generaliz- ability related to the statistical power was achieved because CT data were collected on a large number of patients (n = 1766).

    Limitations of the current research

    Intracranial pressure measurement and monitoring via the ONSD can supply prognostic information [11,22]. Although our aim was to as- sess the ONSD method in general, we did not estimate the prognostic value of the method because there were patients with various patholo- gies in both our hypotensive and hypertensive ICP groups.

    Conclusion

    The ONSD and the ONSD/ETD ratio can indicate low ICP, but for read- ings of ICP less than 13 mm Hg, quantification is not possible. In cases with elevated ICP, the ONSD and the ONSD/ETD ratio provide readings which correspond to ICP readings in millimeters of mercury. The most accurate correlation can be obtained for an ICP range of 15 to 30 mm Hg. The ONSD method reinforced with additional calculations may help to indicate raised ICP, to evaluate its severity quantitatively, and to establish quantitative goal for treatment of intracranial hypertension, but the limitations of the method are to be taken into account.

    Acknowledgments

    The authors would like to thank most sincerely the physicians who assisted us in our research: Zina Evy Almer, MD, neuroophthalmologist, Department of Ophthalmology, and Evelina Shevtsov, MD, Department of Neurology.

    The authors would like to thank most sincerely Christopher Brook (UK) for copyediting of the manuscript.

    References

    1. Raboel PH, Bartek Jr J, Andresen M, Bellander BM, Romner B. Intracranial Pressure Monitoring: Invasive versus Non-invasive methods-A Review. Crit Care Res Pract 2012;2012:950393 [Epub 2012 Jun 8].
    2. Shen Q, Stuart J, Venkatesh B, Wallace J, Lipman J. Inter observer variability of the transcranial Doppler Ultrasound technique: impact of lack of practice on the accura- cy of measurement. J Clin Monit Comput 1999;15:179-84.
    3. McMahon CJ, McDermott P, Horsfall D, Selvarajah JR, King AT, Vail A. The reproduc- ibility of transcranial Doppler Middle cerebral artery velocity measurements: impli- cations for clinical practice. Br J Neurosurg 2007;21:21-7.
    4. Hansen HC, Helmke K. The subarachnoid space surrounding the optic nerves. An ul- trasound study of the optic nerve sheath. Surg Radiol Anat 1996;18:323-8.
    5. Helmke K, Hansen HC. Fundamentals of transorbital sonographic evaluation of optic nerve sheath expansion under intracranial hypertension. I. Experimental study. Pediatr Radiol 1996;26:701-5.
    6. Kimberly HH, Shah S, Marill K, Noble V. Correlation of optic nerve sheath diam- eter with direct measurement of intracranial pressure. Acad Emerg Med 2008; 15:201-4.
    7. Singleton J, Dagan A, Edlow JA, Hoffmann B. Real-time optic nerve sheath diameter reduction measured with bedside ultrasound after therapeutic lumbar puncture in a patient with idiopathic intracranial hypertension. Am J Emerg Med 2015;33: 860.e5-7. http://dx.doi.org/10.1016/j.ajem.2014.12.030.
    8. Caffery TS, Perret JN, Musso MW, Jones GN. Optic nerve sheath diameter and lumbar puncture opening pressure in nontrauma patients suspected of elevated intracranial pressure. Am J Emerg Med 2014;32:1513-5. http://dx.doi.org/10.1016/j.ajem.2014. 09.014.
    9. Lochner P, Mader C, Nardone R, Tezzon F, Zedde ML, Malferrari G, et al. Sonography of the optic nerve sheath beyond the hyperAcute stage of intracerebral hemorrhage. J Ultrasound 2014;17:225-8. http://dx.doi.org/10.1007/s40477-014-0069-6.
    10. Zaidi SJ, Yamamoto LG. Optic nerve sheath diameter measurements by CT scan in ventriculoperitoneal shunt obstruction. Hawaii J Med Public Health 2014;73:251-5.
    11. Legrand A, Jeanjean P, Delanghe F, Peltier J, Lecat B, Dupont H. Estimation of optic nerve sheath diameter on an initial brain computed tomography scan can contribute prognostic information in traumatic brain injury patients. Crit Care 2013;17:R61. http://dx.doi.org/10.1186/cc12589.
    12. Geeraerts T, Launey Y, Martin L, Pottecher J, Vigue B, Duranteau J, et al. Ultrasonog- raphy of the optic nerve sheath may be useful for detecting raised intracranial pres- sure after severe brain injury. Intensive Care Med 2007;33:1704-11.
    13. Vaiman M, Gottlieb P, Bekerman I. Quantitative relations between the eyeball, the optic nerve, and the optic canal important for intracranial pressure monitoring. Head Face Med 2014;10:32. http://dx.doi.org/10.1186/1746-160X-10-32.
    14. Chen H, Ding GS, Zhao YC, Yu RG, Zhou JX. Ultrasound measurement of optic nerve diameter and optic nerve sheath diameter in healthy Chinese adults. BMC Neurol 2015;15:106. http://dx.doi.org/10.1186/s12883-015-0361-x.
    15. Vaiman M, Abuita R, Bekerman I. Optic nerve sheath’ diameters in healthy adults measured by computer tomography. Int J Ophthalmol 2015;8(6):1240-4. http:// dx.doi.org/10.3980/j.issn.2222-3959.2015.06.30.
    16. Ozgen A, Ariyurek M. Normative measurements of orbital structures using CT. AJR Am J Roentgenol 1998;170(4):1093-6.
    17. Bekerman I, Kimiagar I, Sigal T, Vaiman M. Monitoring of Intracranial Pressure by CT- Defined Optic Nerve Sheath Diameter. J Neuroimaging 2015. http://dx.doi.org/10. 1111/jon.12322 [Epub ahead of print].
    18. Vaiman M, Sigal T, Kimiagar I, Bekerman I. Intracranial pressure assessment in trau- matic head injury with hemorrhage via optic nerve sheath diameter. J Neurotrauma 2016;5 [Epub ahead of print].
    19. Bekerman I, Sigal T, Kimiagar I, Evy Almer Z, Vaiman M. Diagnostic value of the optic nerve sheath diameter in cases of pseudotumor cerebri. J Clin Neurosci 2016;30: 106-9. http://dx.doi.org/10.1016/j.jocn.2016.01.018.
    20. Bekerman I, Gottlieb P, Vaiman M. Variations in eyeball diameters of the healthy adults. J Ophthalmol 2014;2014:503645. http://dx.doi.org/10.1155/2014/503645.
    21. Launey Y, Nesseler N, Le Maguet P, Malledant Y, Seguin P. Effect of osmotherapy on optic nerve sheath diameter in patients with increased intracranial pressure. J Neurotrauma 2014;31(10):984-8. http://dx.doi.org/10.1089/neu.2012.2829.
    22. Masquere P, Bonneville F, Geeraerts T. Optic nerve sheath diameter on initial brain CT, raised intracranial pressure and mortality after severe TBI: an interesting link needing confirmation. Crit Care 2013;17:151. http://dx.doi.org/10.1186/cc12728.
    23. Sekhon MS, Griesdale DE, Robba C, McGlashan N, Needham E, Walland K, et al. Optic nerve sheath diameter on computed tomography is correlated with simultaneously measured intracranial pressure in patients with severe traumatic brain injury. Inten- sive Care Med 2014;40(9):1267-74. http://dx.doi.org/10.1007/s00134-014-3392-7.
    24. Lavinio A, Menon DK. Intracranial pressure: why we monitor it, how to monitor it, what to do with the number and what’s the future? Curr Opin Anaesthesiol 2011; 24:117-23. http://dx.doi.org/10.1097/ACO.0b013e32834458c5.
    25. Smith M. Monitoring intracranial pressure in traumatic brain injury. Anesth Analg 2008;106:240-8. http://dx.doi.org/10.1213/01.ane.0000297296.52006.8e.
    26. Steiner LA, Andrews PJD. Monitoring the injured brain: ICP and CBF. Br J Anaesth 2006;97:26-38. http://dx.doi.org/10.1093/bja/ael110.
    27. van Noort R, Martin TR, Black MM, Barker AT, Montero CG. The mechanical proper- ties of human dura mater and the effects of storage media. Clin Phys Physiol Meas 1981;2(3):197-203.
    28. Chauvet D, Carpentier A, Allain JM, Polivka M, Crepin J, George B. Histological and biomechanical study of dura mater applied to the technique of dura splitting decom- pression in Chiari type I malformation. Neurosurg Rev 2010;33:287-94. http://dx. doi.org/10.1007/s10143-010-0261-x.
    29. Giger-Tobler C, Eisenack J, Holzmann D, Pangalu A, Sturm V, Killer HE, et al. Mea- surement of Optic Nerve Sheath Diameter: Differences between Methods? A Pilot Study. Klin Monatsbl Augenheilkd 2015;232:467-70. http://dx.doi.org/10.1055/s- 0035-1545711.