Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Alperin, N. J.
Right arrow Articles by Lichtor, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Alperin, N. J.
Right arrow Articles by Lichtor, T.
(Radiology. 2000;217:877-885.)
© RSNA, 2000

Neuroradiology

MR-Intracranial Pressure (ICP): A Method to Measure Intracranial Elastance and Pressure Noninvasively by Means of MR Imaging: Baboon and Human Study1

Noam J. Alperin, PhD, Sang H. Lee, BA, Francis Loth, PhD, Patricia B. Raksin, MD and Terry Lichtor, MD, PhD

1 From the Departments of Radiology (N.J.A., S.H.L.), Bioengineering (N.J.A., S.H.L.), and Mechanical Engineering (F.L.), University of Illinois at Chicago, 1740 W Taylor, Chicago, IL 60612; the Department of Neurosurgery, Rush-Presbyterian-St Luke’s Medical Center, Chicago, Ill (P.B.R.); and the Department of Neurosurgery, Cook County Hospital, Chicago, Ill (T.L.). From the 1999 RSNA scientific assembly. Received February 25, 2000; revision requested April 3; revision received May 3; accepted May 5. N.J.A. and F.L. supported in part by National Institutes of Health grant RR14242-01. N.J.A. supported in part by a seed grant from the Department of Radiology at the University of Illinois at Chicago. Address correspondence to N.J.A. (e-mail: alperin@uic.edu).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 APPENDIX
 REFERENCES
 
PURPOSE: To develop a noninvasive method for intracranial elastance and intracranial pressure (ICP) measurement.

MATERIALS AND METHODS: Intracranial volume and pressure changes were calculated from magnetic resonance (MR) imaging measurements of cerebrospinal fluid (CSF) and blood flow. The volume change was calculated from the net transcranial CSF and blood volumetric flow rates. The change in pressure was derived from the change in the CSF pressure gradient calculated from CSF velocity. An elastance index was derived from the ratio of pressure to volume change. The reproducibility of the elastance index measurement was established from four to five measurements in five healthy volunteers. The elastance index was measured and compared with invasive ICP measurements in five patients with an intraventricular catheter at MR imaging. False-positive and false-negative rates were established by using 25 measurements in eight healthy volunteers and six in four patients with chronically elevated ICP.

RESULTS: The mean of the fractional SD of the elastance index in humans was 19.6%. The elastance index in the five patients with intraventricular catheters correlated well with the invasively measured ICP (R2 = 0.965; P < .005). MR imaging–derived ICPs in the eight healthy volunteers were 4.2–12.4 mm Hg, all within normal range. Measurements in three of the four patients with chronically elevated ICP were 20.5–34.0 mm Hg, substantially higher than the normal limit.

CONCLUSION: MR imaging–derived elastance index correlates with ICP over a wide range of ICP values. The sensitivity of the technique allows differentiation between normal and elevated ICP.

Index terms: Animals • Brain, volume, 10.368, 10.436, 10.82 • Cerebrospinal fluid, flow dynamics, 10.368, 10.436, 10.82 • Cerebrospinal fluid, MR, 10.12144 • Magnetic resonance (MR), experimental studies, 10.12144 • Magnetic resonance (MR), phase imaging, 10.12144 • Magnetic resonance (MR), volume measurement, 10.12144


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
 APPENDIX
 REFERENCES
 
Intracranial pressure (ICP) is measured for the diagnosis and the management of chronic disorders such as hydrocephalus (1) and pseudotumor cerebri (2). More often, ICP is measured and monitored in acute events such as head injuries and intracranial bleeding (3,4). Although ICP measurement is an invasive technique and, as such, is associated with risk and morbidity (5), measurements of ICP improve the outcome in patients with closed head injuries (6). Because ICP measurement is of clinical importance, there is considerable interest in finding a noninvasive way to estimate ICP.

Hassler et al (7) found changes in arterial waveforms measured by means of transcranial Doppler, or TCD, ultrasonography (US) in 71 patients with known intracranial hypertension and subsequent brain death. Transcranial Doppler US has also been used to demonstrate a correlation between increased ICP and a rise in the Goesling pulsatility index, although great variability exists in the pulsatility index for healthy people (8). Schmidt et al (9) measured middle cerebral arterial flow velocity and used a systems-based approach to predict relative increases in ICP waveforms noninvasively. Schoser et al (10) demonstrated, over a certain range, a linear relationship between mean ICP and maximum venous blood flow velocity in the straight sinus or basal vein of Rosenthal. Shakhnovich et al (11) measured flow velocity in the straight sinus during a body tilt test to differentiate normal from increased ICP states. Systolic flow velocity and the amplitude of pulsation were usually higher in patients than in healthy volunteers in the horizontal position. A nontranscranial Doppler US approach was proposed by Ueno et al (12). Skull movement was detected from measurements of distances across the skull. They postulated that the detected movements occur in conjunction with altered ICP.

The purpose of our study was to develop a method to measure ICP noninvasively. The method uses the well-established relationship between ICP and volume (Fig 1). Because this relationship is monoexponential, the derivative of the pressure with respect to the volume is related linearly to absolute ICP (Appendix). This derivative is estimated from intracranial volume and pressure changes that occur naturally during the cardiac cycle. Volume and pressure changes are measured by using phase-contrast magnetic resonance (MR) imaging studies of blood and cerebrospinal fluid (CSF) flow (13–15).



View larger version (35K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Graph shows the ICP-volume relationship. The change in pressure dP due to change in unit volume dV increases linearly with increased ICP.

 

   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 APPENDIX
 REFERENCES
 
Data Acquisition
Two baboons, eight healthy adult human volunteers (seven men, one woman), and nine adult human patients (four men, five women) underwent imaging with a 1.5-T imager (Signa; GE Medical Systems, Milwaukee, Wis). Measurements from the first baboon were used to establish the relationship between ICP and CSF pressure gradients in the cervical CSF space and from the second baboon to evaluate the effect of pressure manipulation on the MR imaging–derived measurements. Five of the nine patients had an intraventricular catheter in place for management of intracranial hemorrhage at the time of the MR imaging study. Measurements in these five patients allowed us to establish the correlation between MR imaging–derived ICP measurements and invasive ICP measurements.

Five of the eight healthy volunteers underwent repeated MR imaging studies (four to five times each for a total of 22 measurements) to evaluate reproducibility of the MR imaging-derived ICP measurements. The false-positive rate of the noninvasive MR imaging method was established from these measurements and from three single measurements in the remaining three healthy volunteers, for a total of 25 measurements in eight healthy volunteers. Only volunteers with no known neurologic problems or previous head trauma were included. Normal ICP values were assumed for these volunteers. The false-negative rate of the method was established from six measurements in the remaining four patients. Chronically elevated ICP was diagnosed in these patients.

The animal protocol was approved by the institutional animal care committee. The MR imaging protocol used for the human subjects was approved by the institutional review board. Signed consent forms were obtained from each human subject after the nature of the study was fully explained.

MR images were obtained by using a velocity-encoded cine phase-contrast pulse sequence with peripheral gating (16). To obtain ICP and volume changes, both transcranial blood and CSF volumetric flow rate were measured. Two images were obtained in a transverse or oblique section just below the foramen magnum. The first image was used to measure CSF and cord oscillatory pulsation, and the second image was used to measure blood flow. In each case, a section 4–6 mm thick was imaged with a 12–16-cm field of view, a 256 x 128 or 256 x 160 matrix, and two signals acquired. Velocity was encoded along the superior-to-inferior axis with a gradient strength chosen for velocities of interest to be just below the aliasing velocity (velocity corresponding to 180° phase value). CSF flow was measured with a velocity-encoding value of 3–12 cm/sec, and blood flow was measured with a velocity-encoding value of 60–90 cm/sec. The shortest available repetition time, 19–27 msec, was used to optimize spatial resolution and to minimize low-pass filtering of the temporal waveform owing to data interpolation (17). The echo time and flip angle were 8–11 msec and 20°–25°, respectively. The data were interpolated to the maximum number of time points available per cardiac cycle (n = 32) to minimize errors due to secondary resampling.

Measurements of Intracranial Volume Change during the Cardiac Cycle
Time-varying intracranial volume change was computed from the net transcranial volumetric flow. Since brain tissue, blood, and CSF are all incompressible, the volume change can be calculated directly from the net volumetric flow rates into and out of the cranium as described in Equation (1) and from the condition described in Equation (2):


where ICVC is the intracranial volume change, QA(t) is the total arterial volumetric flow rate, QV(t) is the total venous volumetric outflow rate, and QCSF(t) and QC(t) are the rates of CSF and cord volumetric outflows through the foramen magnum, respectively. The t is a time point within the cardiac cycle, {Delta}t is the interval between two time points, and T is the period of one cardiac cycle. Equation (2) represents the Monro-Kellie doctrine (18), which states that the total intracranial volume is constant. However, since intracranial volume fluctuates during the cardiac cycle, the Monro-Kellie doctrine holds only for the mean volume (ie, mean volume change is zero). This condition is used to minimize error in intracranial volume change owing to possible biases (baseline shifts) in the individual volumetric flow measurements because of phase shifts and venous outflow in unobserved channels. The total venous outflow is obtained by scaling the jugular flow to satisfy Equation (2).

Volumetric blood flow rates were measured in each of the following six blood vessels: both internal carotid arteries, both vertebral arteries, and both internal jugular veins. A region of interest was drawn manually by two of the authors (S.H.L., N.J.A.) around each vessel and a nearby static background region. The mean volumetric flow rate was calculated by multiplying the mean velocity by the lumen area.

Measurements of ICP Changes
ICP changes during the cardiac cycle are derived from changes in the CSF pressure gradient (13). The relationship between time-varying change in pressure and in pressure gradient was measured directly in a baboon. A baboon model was chosen because important fluid dynamic parameters such as the size of the spinal canal and heart rate are similar to those of humans. The MR images of the cervical spine of baboons and of humans demonstrate this similarity (Fig 2). Pressures in the cranium (brain parenchyma) and in the spinal canal were measured invasively by using tip transducer catheters (Model V420; Camino Laboratories, [city, state/country]). Pressure and CSF flow velocities were measured at three different ICP values—baseline, elevated, and reduced pressure—achieved by either addition or withdrawal of fluid from the CSF space.



View larger version (116K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2a. Sagittal two-dimensional spin-echo T1-weighted MR images (450/14) of the cervical spine region of (a) a baboon and (b) a human. A baboon model was used, since the cervical spine is similar in shape to that of a human. Although cranial size is considerably different in the craniocervical junction, the region where CSF flow is evaluated is similar. The image obtained in this human volunteer is used only for the purpose of comparison; this subject was excluded due to the accidental finding of Arnold-Chiari malformation.

 


View larger version (114K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2b. Sagittal two-dimensional spin-echo T1-weighted MR images (450/14) of the cervical spine region of (a) a baboon and (b) a human. A baboon model was used, since the cervical spine is similar in shape to that of a human. Although cranial size is considerably different in the craniocervical junction, the region where CSF flow is evaluated is similar. The image obtained in this human volunteer is used only for the purpose of comparison; this subject was excluded due to the accidental finding of Arnold-Chiari malformation.

 
CSF pressure gradient waveforms were derived from the CSF velocity phase-contrast images by using the Navier-Stokes relationship (Eq [3]) between pressure gradient and temporal and spatial derivatives of the fluid velocity for incompressible fluid in a rigid tube (19)

where {rho} is the fluid density, µ is the fluid viscosity, and V and P are the velocity vector and pressure, respectively. This equation is essentially a momentum balance, which states that a small volume-element moving with the fluid is accelerated because of the forces acting on it. In other words, this is a statement of Newton’s second law (ie, mass x acceleration = sum of forces) (20). The first and the second terms on the right-hand side represent momentum changes due to transient (pulsatile) and convective accelerations, respectively. Transient accelerations are those related to velocity changes during a period of time (ie, blood accelerates during the systolic phase of the cardiac cycle). Convective accelerations are those related to velocity changes with a change in position (ie, fluid accelerates as area decreases in a closed cavity, with constant flow rate). The third term is the viscous term, which is the momentum loss due to viscous effects. For CSF, {rho} is 1.0007 g/cm3, and µ is 1.1 cP.

The CSF pressure gradient waveform is calculated from the 32 velocity-encoded MR images of the CSF pulsatile flow. The inertial component of the pressure gradient is approximated by means of a first-order central-difference template of the time series images. The viscous component is derived from a pair of second-order central-difference operators (21). A mean of the viscous and inertial components in the region of interest, which includes only the CSF pixels, was added to obtain the pressure gradient value at each phase of the cardiac cycle.

The relationship between pressure and pressure gradient was further studied by means of numeric simulation of CSF oscillatory flow in a circular annular model of the spinal canal. The spinal canal geometry was represented with two concentric cylinders, with the diameters of the outer (spinal canal) and inner (spinal cord) cylinders being 15 and 6 mm, respectively. Calculations were based on the finite volume method to solve the Navier-Stokes equations by using a commercial software package (STAR-CD 3.0; Adapco, Melville, NY). The finite elements representing the simplified spinal canal geometry had 22,000 nodes. The walls were assumed to be rigid, and the fluid was assumed to behave as a newtonian fluid with a viscosity of 1.1 cP. The velocity boundary condition at the inlet was uniform velocity, which oscillated over time as a sine wave. The peak-to-peak pressure and pressure difference (gradient) between two points were calculated for two different boundary conditions, high and low flow rate, which correspond to high and low peak-to-peak pressure values.

A second set of simulations was performed to determine the relationship between the CSF cross-sectional area and the amplitude of the pressure gradient. Since the spinal canal is a long and relatively straight annular duct, parallel pulsatile flow was proposed as the main flow feature inside the spinal cavity. The lateral walls of the spinal cavity were assumed to be rigid, and the pulsatile flow was accommodated with the zero-resistance expansion of the spinal cavity end such that the flow field could be described by using linearized Navier-Stokes equations at each cross section (ie, convective acceleration terms were neglected).

Simulations of the flow field within the spinal cavity were conducted by using a one-dimensional unsteady model for a circular anulus. A CSF volumetric flow waveform measured from a healthy volunteer was used for the simulation for different CSF cross-sectional flow areas. Further details of the computation can be found in the article by Loth et al (22). The radius of the inner anulus was varied from 1 to 9 mm, while the radius of the outer anulus was kept at 10 mm. Since a flow waveform was imposed as an integral constraint in the present problem, an iterative solution was required at each time step to obtain the correct pressure gradient waveform. The peak-to-peak pressure gradient was calculated for each CSF cross-sectional area. MR imaging–derived pressure gradients were normalized by using this relationship.

Reproducibility of Intracranial Volume Changes, Pressure Gradient Changes, and Elastance Index Measurements
The reproducibility of the MR imaging–based intracranial volume change and pressure gradient change measurements was estimated from four to five repetitive pairs of MR images of the CSF and blood flow obtained in four of the healthy subjects. An estimation of the elastance was obtained from the ratio of the pressure and volume changes. This ratio was defined as the elastance index. The fractional SD of the elastance index measurement represents the reproducibility of the MR imaging–derived pressure measurements. The fractional SD of a product of two variables is the square root of the sum of the square of individual fractional SDs (ie, the fractional SD of the volume and the pressure change measurements).

Comparison between Invasive and MR Imaging–derived Measurements
A second baboon experiment was performed to evaluate the effect of ICP manipulation on the MR imaging–derived measurements of peak-to-peak pressure gradient, peak-to-peak volume change, and their ratio (elastance index) at baseline, elevated, and lowered ICP states. The ICP was elevated by applying pressure over the neck region to restrict jugular venous outflow and lowered by withdrawal of CSF.

MR imaging–derived elastance indices and invasive ICP measurements were obtained in the five patients who had an intraventricular catheter at the time of the MR examination. In two patients, the invasive measurements were obtained within half an hour after the time of the MR imaging study. In the other three patients, pressure readings were obtained immediately before and after MR imaging. In these patients, the mean of the two measurements was used as the invasive ICP reference. The MR imaging–derived elastance indices were correlated with the invasive ICP measurements to obtain the elastance constant coefficient. A linear regression analysis performed by using a least squares method was used to estimate the degree of correlation and its statistical significance. Thus, MR imaging–derived ICP was computed from the product of the elastance constant coefficient and the elastance index. A P value less than .05 was considered to indicate a statistically significant difference.

MR imaging–derived ICP values were calculated for the eight healthy volunteers and the four patients with chronically elevated ICP. Since no invasive comparison was available for the healthy subjects and for the patients with chronically elevated ICP, these measurements were used to assess the false-negative and false-positive rates of the method.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 APPENDIX
 REFERENCES
 
Measurements of Intracranial Volume Change
Figures 3–6 show the results in a healthy subject. Figure 3 shows one of the 32 frames from the cine phase-contrast MR imaging study, which was used to measure volumetric blood flow in arteries and veins carrying blood to and from the cranium. Figure 4 shows two frames from the 32 cine phase-contrast MR images, which were used to measure volumetric CSF flow and cord movement between the cranium and the spinal canal. The left frame shows the superior-to-inferior flow during systole, and the right frame demonstrates the reversal flow during diastole. Total arterial and venous volumetric flow rates during one cardiac cycle are shown in Figure 5a. The CSF volumetric flow rate waveform is shown in Figure 5b. Figure 6 shows the calculated intracranial volume change during the cardiac cycle. The mean value of the peak-to-peak intracranial volume change measured in the healthy subjects was 0.34–1.3 mL. The mean of the fractional SD of the volume change measurement was 18%.



View larger version (209K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3. Transverse two-dimensional gradient-echo velocity-encoded (80 cm/sec) phase-contrast MR image (22/8.5; flip angle, 25°) obtained at a level above the carotid bifurcation in a healthy volunteer. Black pixels indicate inferior-to-superior arterial flow. White pixels indicate superior-to-inferior venous flow. Images from this examination were used to compute the arterial and venous transcranial volumetric flow rates.

 


View larger version (174K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4a. Transverse two-dimensional gradient-echo velocity-encoded (7 cm/sec) phase-contrast MR images (26/12.4; flip angle, 20°) obtained at a level below the foramen magnum in a healthy volunteer. (a) Systolic frame: White pixels inside the region of interest depict cranial-to-spinal CSF flow (arrow, cord region; arrowhead, CSF space region). (b) Diastolic frame: black pixels inside the region of interest depict reverse CSF flow. Images from this examination were used to compute the CSF flow and pressure gradient waveforms.

 


View larger version (175K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4b. Transverse two-dimensional gradient-echo velocity-encoded (7 cm/sec) phase-contrast MR images (26/12.4; flip angle, 20°) obtained at a level below the foramen magnum in a healthy volunteer. (a) Systolic frame: White pixels inside the region of interest depict cranial-to-spinal CSF flow (arrow, cord region; arrowhead, CSF space region). (b) Diastolic frame: black pixels inside the region of interest depict reverse CSF flow. Images from this examination were used to compute the CSF flow and pressure gradient waveforms.

 


View larger version (36K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5a. (a) Graph shows total transcranial arterial inflow rates ({bullet}) and venous outflow rates ({circ}) during the cardiac cycle in a healthy volunteer. Arterial inflow is considerably larger than venous outflow mainly during the initial systolic period (220-380 msec). (b) Graph shows cranial-to-spinal CSF volumetric flow rates ({bullet}) during the cardiac cycle in a healthy volunteer. Large CSF outflow occurs during the systolic period. The onset is delayed with respect to the initial increase in arterial inflow.

 


View larger version (28K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5b. (a) Graph shows total transcranial arterial inflow rates ({bullet}) and venous outflow rates ({circ}) during the cardiac cycle in a healthy volunteer. Arterial inflow is considerably larger than venous outflow mainly during the initial systolic period (220-380 msec). (b) Graph shows cranial-to-spinal CSF volumetric flow rates ({bullet}) during the cardiac cycle in a healthy volunteer. Large CSF outflow occurs during the systolic period. The onset is delayed with respect to the initial increase in arterial inflow.

 


View larger version (27K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6. Graph shows intracranial volume change ({bullet}) during the cardiac cycle computed from the arterial, venous, and CSF flow waveforms in a healthy volunteer. The peak-to-peak intracranial volume change is 1.55 mL.

 
Measurements of ICP Changes
Figure 7a shows the CSF pressure gradient waveform in a healthy subject. The contribution of the viscous term to the total pressure gradient is shown. The mean and the SD of the pressure gradient waveform obtained from five repeated measurements are shown in Figure 7b. The mean of the fractional SD of the peak-to-peak CSF pressure gradient measurement was 8%. The mean of the fractional SD of the delivered elastance index was 19.6%.



View larger version (25K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 7a. (a) Graph shows CSF pressure gradient waveform ({bullet}) in a healthy volunteer. The viscous term ({circ}) in the Navier-Stokes equation (Eq [3]) made only a small contribution to the pressure gradient. (b) Graph shows the mean CSF pressure gradient waveform ({bullet}) and the SD, shown by the error bars, obtained from five measurements in the same volunteer.

 


View larger version (22K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 7b. (a) Graph shows CSF pressure gradient waveform ({bullet}) in a healthy volunteer. The viscous term ({circ}) in the Navier-Stokes equation (Eq [3]) made only a small contribution to the pressure gradient. (b) Graph shows the mean CSF pressure gradient waveform ({bullet}) and the SD, shown by the error bars, obtained from five measurements in the same volunteer.

 
In vivo and simulation results showing the relationship between changes in pressure and pressure gradient are reported. Figure 8 shows the linear relationship found between the peak-to-peak CSF pressure gradient measured by means of MR imaging and invasively measured peak-to-peak ICP change in a baboon at the three different values of absolute ICP. A linear relationship between peak-to-peak changes in pulsatile pressure and pressure gradient was also obtained in the simulation. A twofold increase in the amplitude of the oscillating pressure yielded a twofold increase in the amplitude of the pressure gradient. The simulation results are summarized in Table 1.



View larger version (23K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 8. Graph shows the relationship between peak-to-peak (PTP) CSF pressure gradient obtained from the MR imaging flow measurements and peak-to-peak pressure obtained by means of invasive measurements in a baboon. Four measurements ({boxplus}) were obtained at three different pressure states.

 

View this table:
[in this window]
[in a new window]

  		TABLE 1. Computational Fluid Dynamics Simulation
 
Results from the second set of simulations were used to quantify the effect of changes in CSF cross-sectional flow area on the amplitude of the pressure gradient. The pressure gradient waveforms were calculated for a fixed CSF volumetric flow through different cross-sectional areas (Fig 9). The relationship between the amplitude of the pressure gradients and the reciprocal CSF flow area is shown in Figure 10. The amplitude of the pressure gradient increases nearly linearly with the inverse of the cross-sectional flow area over CSF velocities in which flow is dominated by inertia.



View larger version (38K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 9. Graph shows simulated CSF pressure gradient waveforms for a circular anulus with four different cross-sectional areas. The outer radius is 10 mm and the inner radius (Ri) ranges from 1 to 7.5 mm. CSF flow is identical in all four areas. The amplitude increases with decreased flow area, whereas the shape is unchanged.

 


View larger version (20K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 10. Plot of peak-to-peak pressure gradient values ({bullet}) obtained from the simulation versus the reciprocal CSF flow area shows the nearly linear relationship between these parameters.

 
Comparison between Invasive and MR Imaging–derived Measurements
The results from the second baboon experiment are summarized in Table 2. This experiment demonstrated that the MR imaging–derived elastance index value corresponded to the ICP level, as expected, on the basis of the monoexponential pressure-volume relationship. At the elevated ICP, both the peak-to-peak pressure gradient and volume change decreased. However, the peak-to-peak volume change decreased by a larger ratio, from 0.17 to 0.09 mL, than did the peak-to-peak pressure gradient, from 0.029 to 0.023 mm Hg/cm, which resulted in an increased elastance index. The restricted venous outflow resulted in a significant reduction in the intracranial volume change during the cardiac cycle. In contrast, at the lower ICP, the measured volume change was the same as that of the baseline state. Hence, the decrease in peak-to-peak pressure gradient resulted in a smaller elastance index.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Results in the Second Baboon Experiment
 
Pressure traces from invasive recordings obtained in two of the five patients with an intraventricular catheter are shown in Figure 11. The trace in a patient with low ICP is shown on the left and in a patient with elevated ICP is shown on the right. The corresponding CSF pressure gradient waveforms derived from the MR imaging CSF flow measurements are shown at the bottom. The relationship between the invasive ICP measurements and the MR imaging–derived elastance indices are shown in Figure 12. The uncertainty of the invasive measurement shown by the error bar in the horizontal direction was obtained from the difference between the measurements obtained immediately before and after MR imaging. The mean of the fractional SD of the elastance index calculated from repeated imaging in the healthy subjects was assumed to be the uncertainty of the MR imaging measurement. A significant (P < .005) correlation coefficient (R2 = 0.965) between the noninvasive and invasive measurements was obtained.



View larger version (51K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 11. Invasive pressure traces (top left and right grids) obtained with an intraventricular catheter in two patients, one with low (left) and the other with elevated (right) ICP. These individual cardiac cycles were obtained arbitrarily from a paper strip recording of the ICP. One grid square in the horizontal direction corresponds to 200 msec. The vertical pressure scale corresponds to 3 mm Hg. ICP was monitored in these patients to manage intracranial hemorrhage. The corresponding MR imaging-derived CSF pressure gradient waveforms ({bullet}) and the contribution of the viscosity ({lozenge}) are shown at the bottom (left and right graphs).

 


View larger version (26K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 12. Graph shows the relationship between MR imaging-derived elastance indices and invasive ICP measurements in the five patients with intraventricular catheters. These parameters are well correlated (R2 = 0.965) and significant (P < .005). Vertical error bars indicate SD. Horizontal error bars indicate uncertainty in invasive pressure measurements.

 
The 25 MR imaging–based ICP measurements in the eight healthy subjects were 4.2–12.4 mm Hg, which were all within the normal range (3–15 mm Hg) of ICP values. Five of the six MR imaging–based ICP measurements in the four patients with chronically elevated ICP (pseudotumor cerebri) were 20–34 mm Hg, which is considerably higher than the upper normal limit of 15 mm Hg. The MR imaging–based ICP measurement in the fourth patient was 11 mm Hg. However, 5 days prior to MR imaging, this patient’s ICP was reduced from 24 to 5 mm Hg by withdrawal of CSF. The distribution of these measurements is shown in Figure 13.



View larger version (19K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 13. Bar graph shows the distribution of MR imaging-derived pressure measurements (n = 31) obtained in the eight healthy volunteers (black bars, n = 25) and in the four patients (gray bars, n = 6) with chronically elevated ICP (pseudotumor cerebri). All 25 measurements in the healthy volunteers were within the normal range of ICP values. In three of the four patients, MR imaging measurements were considerably higher than normal ICP values. The fourth patient (*) had a measurement within the normal range; 5 days prior to the MR imaging study, this patient’s ICP was reduced from 24 to 5 mm Hg by withdrawal of CSF.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
 REFERENCES
 
A method to measure intracranial elastance and pressure noninvasively that integrates principles of neurophysiology, fluid mechanics, and MR imaging measurements of blood and cerebrospinal fluid flow has been presented. The pulsatile nature of blood flow into the brain results in a cyclic change of intracranial volume. The volume change is on the order of 0.1% of the total intracranial volume. Therefore, this change is treated as a small perturbation of the intracranial system’s steady state. In steady state, the pressure response to the volume change reflects the pressure state of the intracranial system. It has been shown by many researchers (23–26) that a monoexponential curve accurately represents the ICP-volume relationship over a wide range of normal and elevated pressure values. This relationship implies that the elastance is a linear function of the pressure. Therefore, pressure is derived from the estimate of elastance (derivative of pressure with respect to volume) as the ratio of pressure to volume change during the cardiac cycle. This is a reasonable approximation, since volume changes are small.

Volume and pressure changes are computed from phase-contrast time series MR images of blood and CSF pulsatile flow. The first component, intracranial volume change, is calculated from the difference between blood and CSF inflow and outflow at each point in the cardiac cycle. Calculation of a small quantity (intracranial volume change) from the difference of two large quantities often leads to a large error. This error may be attributed both to inaccurate measurement of mean flow rates into and out of the brain and to unmeasured venous outflow through channels other than the jugular veins. The Monro-Kellie doctrine provides a way to minimize this error by stipulating that the sum of all mean flow rates must equal zero (Eq [2]). Since phase-contrast MR imaging with a short repetition time, in the range of 18–27 msec, can be used to measure the dynamic changes in flow rate reliably (17), accurate measurement of small volume changes are possible.

The variability of the volume change measurement was 18% (mean of the fractional SD). Contributing to this variability is the variability in determining the CSF and vessel lumen borders manually. An automated method in which temporal pulsatility is used as a criterion for lumen segmentation is currently being developed. Initial evaluation of this technique for vessels and CSF lumens provided four to five times more reproducible determination of the lumen area (27).

Another important component is accurate estimation of ICP change during the cardiac cycle. This parameter is derived from the CSF pressure gradient. The CSF pressure gradient is computed by using the Navier-Stokes equation (Eq [3]). This calculation assumes that the walls of the spinal canal are rigid. This assumption seems reasonable for CSF flows in the upper cervical region. The dura mater is constrained by the bony wall of the cervical canal at the level of C2.

Peak-to-peak pressure is estimated from the peak-to-peak pressure gradient. A linear relationship between these two parameters was found experimentally in a baboon and by means of computational fluid dynamics simulations. From principles of fluid mechanics, the time derivative of the pressure waveform is related linearly to the pressure gradient for pulsatile sine wave flow in a rigid pipe (19). The theoretic basis for the relationship between peak-to-peak pressure and pressure gradient of physiologic flow waveforms needs to be investigated further (22).

The second baboon experiment showed correspondence between the manipulated ICP value and the MR imaging–derived elastance index in a baboon. Furthermore, results from this experiment confirm that both volume (derived from blood and CSF flow) and pressure changes (derived from CSF velocities) are required to obtain absolute ICP values. At the elevated pressure state, the trend for increased pulse pressure and pulse pressure gradient was reversed by the reduction in volume change owing to the restricted venous flow. These measurements may explain why in previous studies in which transcranial Doppler US was used to measure parameters related to blood flow or volume change the correlation with ICP values was within a limited range or limited to unchanged hemodynamic conditions (7–10).

MR imaging–derived elastance indices in the five patients for whom we had invasive validation were correlated linearly (P < .005) with invasive ICP measurements. This relationship provided the value for the elastance coefficient constant that relates the MR imaging-derived elastance index to ICP. This result supports that of an earlier study (25) of relatively small variability in the elastance coefficient constant. The elastance constant was used to derive the ICP values not only in healthy subjects but also in patients with chronically elevated ICP. The MR imaging–based ICP measurements in three of the four the patients with proved diagnosis of chronically elevated ICP (pseudotumor cerebri) were considerably higher than the upper normal limit. Although the state of the intracranial system may differ from the normal state, the same elastance constant coefficient provided high ICP values, as expected in these patients. This finding suggests that although the elastance constant may vary between individuals and between different states, the variation may not be substantial with respect to MR imaging–derived ICP measurement. A transfer function analysis is now being applied to attempt to quantify the factors that may affect this elastance constant (28,29).

The MR imaging–based ICP measurements in the healthy subjects were 4–12 mm Hg. These values are well within the normal pressure range of 3–15 mm Hg as measured invasively in a study performed in 1,033 healthy subjects (30). The results in the healthy subjects and the patients are encouraging. There were no false-positive or false-negative results, which suggests that the sensitivity of the technique is well within that needed to differentiate between normal and elevated ICP values. Nevertheless, a larger number of patients with a wide variety of diseases need to be studied to determine how robust this approach is for noninvasive measurements of ICP.

The role of an MR imaging–based ICP measurement may be different from that of the invasive technique. Whereas invasive monitoring provides continuous ICP measurements, the MR imaging study provides a measurement at a single time point. There are several clinical settings in which a "snapshot" of ICP may be beneficial. Management of blunt head trauma may be one of these areas. Placement of an invasive ICP monitor is recommended for severe head injuries, defined as those with a Glasgow Coma scale score of 3–8 (31). However, the necessity of ICP monitoring in patients with intermediate Glasgow Coma scale scores of 9–12—and, in particular, those with a normal computed tomographic scan at presentation—has been the subject of debate. Invasive monitoring techniques are not routinely used in this patient population, yet 20% of these patients will deteriorate neurologically during the first 24–48 hours (6,32–34). Another subset of patients, those with diffuse axonal injury, may demonstrate a Glasgow Coma scale score in the severe injury range but have normal ICP at placement of an invasive monitor. Noninvasive MR imaging–based ICP measurement would provide a means of objective assessment of the need for an invasive monitor without incurring the potential morbidities.

MR imaging–based ICP may have an important role in the diagnosis of several chronic disorders that may be associated with changes in ICP. These include hydrocephalus, pseudotumor cerebri, intracranial masses, Arnold-Chiari malformation, and toxic-metabolic encephalopathy in which a depressed level of consciousness may or may not correspond to increased ICP. MR imaging–based ICP measurement may prevent unnecessary invasive monitoring in an increased-risk setting. A single measurement of ICP may also be useful in patients with ventriculoperitoneal shunt, especially in young children with nonspecific symptoms. Normal ICP measurements may prevent unnecessary shunt revision in these patients.


   APPENDIX
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 APPENDIX
REFERENCES
 
In a closed system such as the cranium, the pressure and the volume are related. This relation has been studied extensively by using invasive techniques in animals and in humans. The change in pressure due to volume change is determined by the overall mechanical elastance of the system. Ryder et al (35) and others studied the pressure-volume relationship by injecting fluid into the CSF space. Measurements of pressure change resulting from a given volume change permitted characterization of the ICP-volume dependence. This relationship is known as the intracranial elastance curve. Work performed in both animals and humans has shown that the elastance curve is well described by means of a monoexponential function in the physiologic range. Marmarou et al (23) proposed the following expression for the pressure-volume curve:

where E1 is a constant elastance coefficient (units of volume-1), P1 is a pressure coefficient related to the exponential shape of the curve, and V is the intracranial volume (CSF + blood + brain tissue). The monoexponential elastance curve, shown in Figure 1, illustrates the dependence of the ratio of pressure to volume change during the cardiac cycle, dP/dV, on the mean pressure. This derivative is defined as elastance and is a linear function of ICP as shown in equation A2:

Sklar and Elashvili (24) and others have validated this relationship with a volume pressure test, whereby a uniform amount of fluid is injected rapidly into the lateral ventricle. The pressure change resulting from the volume increase is termed "volume-pressure response." In human patients and experimental animals, the relationship between volume-pressure response and ICP has been shown to be linear. This linearity confirms that the pressure-volume relationship is monoexponential and that the elastance coefficient (E1) is constant. Szewczykowski et al (25) estimated the elastance coefficient from the linear relationship between ICP change during the cardiac cycle and mean ICP in 10 patients. The ICP was manipulated by injecting fluid into the CSF space. The intracranial volume change during the cardiac cycle was not measured, and the same value was assumed in all patients. A relatively small variability in the measured slopes was found (mean, 0.329 ± 0.084 [SD]). Since variability in intracranial volume change changes among subjects was not taken into account, it is possible that the actual variability in the elastance coefficient constant may be even smaller.


   ACKNOWLEDGMENTS
 
The authors gratefully acknowledge Fady Charbel, MD, Ben-Zion Roitberg, MD, and James Goodwin, MD, from the Departments of Neurosurgery and Neuro-Ophthalmology at the University of Illinois at Chicago for recruiting patients for this study.


   FOOTNOTES
 
Abbreviations: CSF = cerebrospinal fluid, ICP = intracranial pressure

Author contributions: Guarantor of integrity of entire study, N.J.A.; study concepts, N.J.A.; study design, N.J.A.; definition of intellectual content, N.J.A., F.L.; literature research, N.J.A., P.B.R.; clinical studies, N.J.A.; experimental studies, N.J.A., T.L.; data acquisition, N.J.A.; data analysis, N.J.A., S.H.L.; statistical analysis, N.J.A., S.H.L.; manuscript preparation, N.J.A., P.B.R., F.L.; manuscript editing, N.J.A., P.B.R.; manuscript review, N.J.A.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 APPENDIX
 REFERENCES
 

  1. Maira G, Rossi GF, Viganti A. Intracranial pressure and pathogenesis of "normotensive" hydrocephalus. In: Lundberg N, Ponten U, Brock M, eds. Intracranial pressure II. Berlin, Germany: Springer-Verlag, 1975; 128-132.
  2. Gjerris F, Soelberg P, Sorenson Vorstrup S, Paulson OB. Intracranial pressure, conductance to CSF outflow, and cerebral blood flow in patients with benign intracranial hypertension (PTC). Ann Neurol 1985; 17:158-162.[Medline]
  3. Jennett B, Teasdale G, Galbraith S, et al. Severe head injury in three countries. J Neurol Neurosurg Psychiatry 1977; 40:291-295.[Abstract/Free Full Text]
  4. Miller JD, Butterworth JF, Gudeman SK, et al. Further experience in the management of severe head injury. J Neurosurg 1981; 54:289-299.[Medline]
  5. Paramore CG. Relative risks of ventriculostomies. Acta Neurochir 1994; 127:79-84.[Medline]
  6. O’Sullivan MG, Stratham PF, Jones PA, et al. Role of ICP monitoring in severely head injured patients without signs of intracranial hypertension on initial CT. J Neurosurg 1994; 80:46-50.[Medline]
  7. Hassler W, Steinmetz H, Gawlowski J. Transcranial Doppler ultrasonography in raised intracranial pressure and in intracranial circulatory arrest. J Neurosurg 1988; 68:745-751.[Medline]
  8. Nadvi SS, DuTrevou MD, Van Dellen JR, Gouws E. The use of transcranial Doppler ultrasonography as a method of assessing intracranial pressure in hydrocephalic children. Br J Neurosurg 1994; 8:573-577.[Medline]
  9. Schmidt B, Klingelhofer J, Schwarze JJ, Sander D, Wittich I. Noninvasive prediction of intracranial pressure curves using transcranial Doppler ultrasonography and blood pressure curves. Stroke 1997; 28:2465-2472.[Abstract/Free Full Text]
  10. Schoser BGH, Riemenschneider N, Hansen HC. The impact of raised intracranial pressure on cerebral venous hemodynamics: a prospective venous transcranial Doppler ultrasonography study. J Neurosurg 1999; 91:744-749.[Medline]
  11. Shakhnovich AR, Shakhnovich VA, Galushkina AA. Noninvasive assessment of the elastance and reserve capacity of the craniovertebral contents via flow velocity measurements in the straight sinus by TCD during body tilting test. J Neuroimaging 1999; 9:141-149.[Medline]
  12. Ueno T, Ballard RE, Shuer LM, Cantrell JH, Yost WT, Hargens AR. Noninvasive measurement of pulsatile pressure using ultrasound. Acta Neurochir Suppl (Wien) 1998; 71:66-69.[Medline]
  13. Alperin N, Betz W, Lopez M, et al. MR measurements of intracraniospinal CSF pulsatile pressure gradients (abstr). Proceedings of the Fourth Meeting of the International Society for Magnetic Resonance in Medicine. 1 Berkeley, Calif: International Society for Magnetic Resonance in Medicine, 1996; 80.
  14. Alperin N, Betz W, Slavin K, Fortman J. Deriving intracraniospinal elastance and pressure from MRI measurements of transcranial CSF and blood flow (abstr). Proceedings of the Fifth Meeting of the International Society for Magnetic Resonance in Medicine. 1 Berkeley, Calif: International Society for Magnetic Resonance in Medicine, 1997; 117.
  15. Alperin NJ, Loth F, Lee SH, Charbel FT, Lichtor T, Roitberg B. MR imaging based noninvasive method for intracranial pressure measurement (abstr). Radiology 1999; 213(P):144.
  16. Pelc NJ, Herfkens R, Shimakawa A, Enzman DR. Phase contrast cine magnetic resonance imaging. Magn Reson Med Q 1991; 7:229-254.
  17. Frayne R, Rutt B. Frequency response to retrospectively gated phase-contrast MR imaging: effect of interpolation. J Magn Reson Imaging 1993; 3:907-917.[Medline]
  18. Magendie F. Recherches physiologiques et cliniques sur le liquide céphalorachidien ou cérébro-spinal Paris, France: Librairie Médicale de Mequigenon-Marvis Files, 1842; 66.
  19. Nichols WW, O’Rourke MF. McDonald’s blood flow in arteries: theoretical, experimental and clinical principles In: Pulsatile pressure/flow relations. 3rd ed. Philadelphia, Pa: Lea & Febiger, 1990; 122-136.
  20. Bird RB, Stewart WE, Lightfoot EN. Transport phenomena New York, NY: Wiley, 1960; 71-82.
  21. Urchuk SN, Plewes DB. MR measurement of pulsatile pressure gradients. J Magn Reson Imaging 1994; 4:829-836.[Medline]
  22. Loth F, Yardimci MA, Alperin N. Hydrodynamic modeling of cerebrospinal fluid motion in the spinal cavity. J Biomech Eng; (in press).
  23. Marmarou A, Shulman K, La Morgese J. Compartmental analysis of compliance and outflow resistance of the CSF system. J Neurosurg 1975; 43:523-534.[Medline]
  24. Sklar FH, Elashvili I. The pressure volume function of brain elasticity. J Neurosurg 1977; 47:670-679.[Medline]
  25. Szewczykowski J, Sliwka S, Kunicki A, Dyko P, Korsak-Sliwaka J. A fast method of estimating the elastance of the intracranial system. J Neurosurg 1977; 47:19-26.[Medline]
  26. Avezzat CJJ, van Eijndhoven JHM. Cerebrospinal fluid pulse pressure and intracranial volume-pressure relationships. J Neurol Psychiatry 1979; 42:687-700.
  27. Alperin N, Lee S. Time series method for automated segmentation of vessels and CSF lumens (abstr). Proceedings of the Eighth Meeting of the International Society for Magnetic Resonance in Medicine, 2 Berkeley, Calif: International Society for Magnetic Resonance in Medicine, 2000; 1540.
  28. Alperin N, Vikingstad EM, Gomez-Anson B, Levin DN. Hemodynamically independent analysis of cerebrospinal fluid and brain motion observed with dynamic phase contrast MRI. Magn Reson Med 1996; 35:741-754.[Medline]
  29. Alperin N, Stelzig C. Does CSF flow reflect the mechanical state of the craniospinal system? (abstr). Proceedings of the Seventh Meeting of the International Society for Magnetic Resonance in Medicine. 3 Berkeley, Calif: International Society for Magnetic Resonance in Medicine, 1999; 2011.
  30. Merrit HH, Fremont-Smith F. The cerebrospinal fluid Philadelphia, Pa: Saunders, 1937.
  31. Bullock R, Chesnut RM, Clifton G, et al. Guidelines for the management of severe head injury: Brain Trauma Foundation. Eur J Emerg Med 1996; 3:109-127.[Medline]
  32. Lobato RD, Sarabia R, Rivas JJ, et al. Normal CT scans in severe head injury. J Neurosurg 1986; 65:784-789.[Medline]
  33. Eisenberg HM, Gary HE, Jr, Aldrich EF, et al. Initial CT findings in 753 patients with severe head injury: a report from the NIH Traumatic Coma Data Bank. J Neurosurg 1990; 73:688-698.[Medline]
  34. Narayan RK, Kishore PR, Beck DP, et al. Improved confidence of outcome prediction in severe head injury: a comparative analysis of the clinical examination, multimodality evoked potentials, CT scanning and intracranial pressure. J Neurosurg 1981; 54:751-762.[Medline]
  35. Ryder HW, Espey FF, Kimbel FD, Penka EJ, Rosenauer A, Evans JP. The mechanism of change in cerebrospinal fluid pressure following an induced change in volume of the fluid space. J Lab Clin Med 1953; 41:428-435.[Medline]



This article has been cited by other articles:


Home page
 Am. J. Neuroradiol.Home page
J. Kim, N.A. Thacker, P.A. Bromiley, and A. Jackson
Prediction of the Jugular Venous Waveform Using a Model of CSF Dynamics
AJNR Am. J. Neuroradiol., May 1, 2007; 28(5): 983 - 989.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
M. F. Quigley, B. Iskandar, M. A. Quigley, M. Nicosia, and V. Haughton
Cerebrospinal Fluid Flow in Foramen Magnum: Temporal and Spatial Patterns at MR Imaging in Volunteers and in Patients with Chiari I Malformation
Radiology, July 1, 2004; 232(1): 229 - 236.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Alperin, N. J.
Right arrow Articles by Lichtor, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Alperin, N. J.
Right arrow Articles by Lichtor, T.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE