Read Alperin-MS1.pdf text version

Current Medical Imaging Reviews, 2006, 2, 117-129


From Cerebrospinal Fluid Pulsation to Noninvasive Intracranial Compliance and Pressure Measured by MRI Flow Studies

Noam Alperin*,1, Monica Mazda1,Terry Lichtor2 and Sang H. Lee1


Physiological Imaging and Modeling Lab, Department of Radiology (M/C 711), The University of Illinois at Chicago, 830 S. Wood St., Chicago, IL 60612, 2Department of Neurosergery, Rush University Medical Center, Chicago, IL, USA

Abstract: The pulsation of the cerebrospinal fluid (CSF) has fascinated investigators of the intracranial physiology since it was first documented by invasive CSF pressure measurements. Advances in dynamic Magnetic Resonance Imaging (MRI) now enable visualization of and quantitation the CSF flow dynamics and has contributed to our understanding of the origin of CSF pulsation and its relation to the pulsatile blood flow. This, in turn, has led to the development of a noninvasive method for measurement of intracranial compliance and pressure by MRI. This article reviews the neurophysiologic and hydrodynamic principles that are the basis of the method, it describes the implementation of the method and validation studies to date with a non-human primate animal model, computer simulations, healthy human subjects and patients. The article further reviews the application of this method to study the effect of body posture on the cerebral physiology in humans through the relationships between blood and CSF flow dynamics. Finally, recent results from the application of the method in Chiari Malformations (CM) are briefly presented as an example of a potential clinical application of this methodology. The application to CM provided, for the first time, evidence of the important role intracranial compliance plays in the pathophysiology of this poorly understood disorder. The potential diagnostic value of an MRI-based measurement of ICP for other neurological problems is discussed.

Keywords: Cerebrospinal fluid pulsation, MRI flow studies, intracranial compliance, intracranial pressure, cerebral blood flow. EARLY INVESTIGATIONS DYNAMICS OF CSF FLOW et al. [3, 4] suggested that the CSF pulsation arises entirely from the pulsation of the choroid plexus. They further noticed that CSF pulse pressure in the lumbar subarachnoid space is attenuated compared with the CSF pulse pressure in the ventricles and suggested that it is due to the elasticity of the dural envelope. Hamit et al. [5] suggested that the CSF pulse pressure follows a venous rather than an arterial pattern. Dunbar et al. [6] found similarity between CSF pulse pressure waveforms measured in the ventricles, cisterna magna and lumbar subarachnoid space, and the arterial pulse pressure wave measured in the aorta in dogs. Dardenne et al. [7] examined the effect of mean CSF pressure on the similarity between the waveforms of blood pressure in the carotid artery and the CSF pressure. They noted that the CSF pulse pressure becomes more similar to the carotid pulsations as the mean CSF pressure rises. They also noted that with reduced cardiac output CSF pulsations resembled venous pressure. The influence of systemic arterial pressure on the CSF pulsation was further studied by Dereymaeker et al. [8]. They demonstrated a tendency of the cisternal CSF pulsations to resemble arterial pulsations when the arterial pressure is low and attain a more venous like morphology at high arterial blood pressure. They argued that at high blood pressure, venous pulsation dominates the CSF pulse wave due to the reduced ability of the arterial wall to pulsate. In 1977, Hamer et al. [9] categorized the intracranial pressure pulse wave under normal and pathophysiological conditions. In normal conditions, they found that the systolic portion of the intracranial CSF pulse wave is primarily arterial in origin and the descending diastolic portion is similar to the superior

©2006 Bentham Science Publishers Ltd.

The Cerebrospinal Fluid (CSF) flow is influenced by two separate processes; the circulation of the CSF from its formation sites to its absorption sites (i.e., bulk flow), and an oscillatory (back and forth) flow during the cardiac cycle (pulsatile flow). The first process governs the overall volume of CSF in the craniospinal space and thereby influences intracranial pressure (ICP). The second process, the oscillatory movement of the CSF within the craniospinal compartments, is caused by the pulsatile blood flow entering and leaving the intracranial compartment during the cardiac cycle. These two processes occur over different time scales. The circulation and replenishing of CSF in the craniospinal system occurs over minutes while the time scale of the pulsatile CSF flow is milliseconds. The pulsatile nature of the CSF flow has been the subject of many investigations since 1943, when the pulsatility of CSF pressure was first linked to vascular factors [1]. Simultaneous dynamic measurements of blood and CSF pressures revealed that the two pressures are synchronized and therefore related. However, the specific relationship and the origin of CSF pulsation have been debated ever since. First, it was suggested that CSF pulsations arise primarily from the large arteries at the base of the brain [2]. Bering

*Address correspondence to this author at the Physiological Imaging and Modeling Lab, Department of Radiology (M/C 711), The University of Illinois at Chicago, 830 S. Wood St., Chicago, IL 60612, USA; Tel: 312996-4620; Fax: 312-996-6619, USA; E-mail: [email protected]

1573-4056/06 $50.00+.00


Current Medical Imaging Reviews, 2006, Vol. 2, No. 1

Alperin et al.

vena cava pulse pressure. Under increased ICP, however, the descending part of the CSF pulse wave also resembled the arterial pulse. During increased central venous pressure the pulsations became more venous in shape. Based on these observations they concluded that the morphology of the CSF pulsations could be regarded as an index of the state of the intracranial elastance or compliance. Following the development of cardiac-synchronized motion-sensitive MRI techniques, in the mid 1980s, MR imaging became an important noninvasive means for detailed investigation of the CSF flow dynamics and brain motion [10-12]. Greitz et al. [13] used velocity imaging by phased contrast MR (PCMR) to document the movement of various brain regions during the cardiac cycle to elucidate the role of the brain movement on CSF flow. They postulated that arterial expansion causes brain movement and concluded that expansion of the central and lower parts of the brain acts as an expanding and retracting piston, which causes the CSF pulsation at the foramen magnum. The piston mechanism for the CSF pulsation concurs with the much earlier but similar explanation by O'Connel et al. [1]. Based on observations that CSF pulse pressure varies with cardiac and respiratory cycles O'Connel et al. concluded that CSF pulsation is due to changes in brain blood volume caused by arterial expansion [1]. MRI measurements of volumetric flow rates provide information about the volumes of blood and CSF that are entering and leaving the intracranial space during the cardiac cycle [14-16]. In 1993, Enzmann and Pelc [14] used PCMR to study the temporal relationship between net blood volume entering the brain (arteriovenous difference) and CSF craniospinal flow volume. The arterial inflow was measured by adding together the volumetric flow rate (VFR) through the carotid and vertebral arteries. The venous outflow was estimated from the volumetric flow through the internal jugular veins. They reported that peak arteriovenous difference waveform occurred simultaneously with the peak of the CSF VFR waveform. The arteriovenous blood volume difference was considered to be both a measure of brain expansion and the driving force behind the cranio-to-spinal CSF flow. The notion that net transcranial blood flow is driving the oscillatory CSF flow between the cranium and the spinal canal was further explored by Alperin et al. [15] who applied modulation transfer function analysis to characterize the biomechanical properties of the intracranial compartment. The analysis demonstrated that the normal intracranial biomechanical state can be approximated by an exponential impulse response function (IRF) with a time constant proportional to the product of the (intracranial) compliance and the resistance to cranio-spinal CSF flow. This implies that the CSF flow dynamics are modulated by intracranial compliance. Since intracranial compliance is inversely proportional to ICP, it also explains the dependency of the relationship between arterial, venous, and CSF flows and pulse pressures on intracranial pressure. A finite intracranial compliance dictates that peak CSF flow occurs after the peak arterial inflow. A more recent work by Baledent et al. [16] has indeed confirmed a small time delay between these two peaks.

MEASUREMENT OF INTRACRANIAL COMPLIANCE AND PRESSURE BY MRI Although MRI CSF flow studies provide quantitative measures of the CSF flow dynamics, clinical utilization of these studies has been limited due to the difficulty in interpreting these measurements and due to large interindividual variability, even among healthy subjects. Establishing a relationship between CSF pulsation and important clinical physiological parameters such as intracranial compliance and pressure may therefore enhance the clinical role of CSF flow studies. A noninvasive method for characterization of the intracranial physiology through measurements of blood and CSF flow has been recently described [17-19]. This method provides simultaneous measurements of total cerebral blood flow (TCBF), intracranial compliance, and intracranial pressure (ICP). These physiological parameters characterize the hemodynamic and hydrodynamic states of the craniospinal system, i.e., the craniospinal biomechanical state. The craniospinal biomechanical state changes with normal activities (e.g., change in body posture), with aging, and with head trauma and diseases (e.g., strokes, intracranial hemorrhages, hydrocephalous, Chiari Malformations, etc.). A noninvasive assessment of cerebral blood flow and intracranial pressure would make these important parameters more accessible and therefore it may improve our understanding of the normal intracranial physiology and of the pathophysiology in related diseases. 1. The Neurophysiology Basis of the MR-ICP Method The MRI-based method integrates human neurophysiology and fluid dynamic principles with dynamic MRI techniques to measure intracranial elastance (inverse of compliance) and ICP. The method utilizes the small fluctuations in intracranial volume and pressure that occur with each heartbeat. The intracranial elastance, i.e., a change in pressure due to a small change in volume, is derived from the ratio of these changes. A mean ICP value is then derived from the linear relationship between intracranial pressure and elastance. In a closed system such as the cranium, the pressure and volume are related. Ryder et al. [20] and others [21-23] studied the pressure-volume relationship by injecting fluid into the CSF space. The degree of the change in pressure resulting from an injection of a known volume permitted characterization of the intracranial pressure-volume dependence. Marmarou et al. [21] proposed the following expression for the pressure-volume curve P = P1 eE1V (1)

where E1 is a constant elastance coefficient (unitsvolume-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 mono-exponential elastance curve, shown in Fig. (1) illustrates the dependence of the ratio of pressure and volume changes, dP/dV, on the mean pressure. The change in pressure with each heartbeat

A Review of Wavelet Denoising in MRI and Ultrasound Brain Imaging

Current Medical Imaging Reviews, 2006, Vol. 2, No. 1 119

Similarly, MR-ICP measures the elastance from the ratio of pressure and volume changes that occur naturally with each cardiac cycle. Intracranial pressure is then derived through the linear relationship between elastance and pressure. The small volume change that occurs with each cardiac cycle is analogous to the injected volume used in the volume-pressure response test. Volume and pressure changes occur because of the pulsatile nature of blood flow. During systole, more volume flows into the cranial vault (arterial inflow exceeds venous and CSF outflows) while during diastole, the outflow is larger. The MRI method measures the arterial, venous, and CSF flows into and out of the cranial vault. A small volume change, on the order of one milliliter, is derived from the differences between inflow and outflow at each time points in the cardiac cycle. The pressure change is derived from the CSF velocities using fluid dynamics principles. The derivation of these quantities with MRI is described in following sections. 2. Intracranial Volume Change During the Cardiac Cycle

Fig. (1). The monoexponential curve depicting the relation between the intracranial pressure and volume. Note that at low pressures the ratio of dP/dV (elastance) is small while at a high pressure the ratio is large.

(i.e., the amplitude of the pulse pressure) due to a change in the intracranial volume is determined by the elastance of the system. At a low pressure, this ratio is small, while at a high pressure, this ratio is large. Since the pressure-volume curve is mono-exponential intracranial elastance is a linear function of ICP. The pressure-volume relationship was not determined from a direct measurement of the volume of the intracranial space; there was no in vivo means of measuring that volume. Instead, it was inferred from the linear relationship between elastance and pressure [21-23]. The elastance was measured with a volume-pressure response test. In this test, shown in Fig. (2) (from Marmarou et al., 1975), the total intracranial volume is rapidly increased by injecting a known amount of fluid into the ventricles. The elastance is then derived from the ratio of the resultant pressure change (Pp-P0) and the amount of injected volume, V.

In a compartment with inlets and outlets, a change in the volume of that compartment can be determined from the instantaneous difference between volumetric flows entering and leaving the compartment as long as the flowing fluid and the content inside are incompressible. The Monro-Kellie doctrine makes this observation [24]. This doctrine also assumes that the volume of the intracranial space is constant. Today, we know that a small periodic change in the intracranial volume occurs with each cardiac cycle. The mean intracranial volume, however, is constant. The change in the intracranial volume during the cardiac cycle is calculated from measurements of volumes of blood and CSF that are entering and leaving the cranial vault, as described in equation 2 and from the condition described by equation 3, ICV(t) = [QA(t) - Qv(t) - QCSF(t)] t 0<t<T ICV(T) = [QA(t) - Qv(t) - QCSF(t) ] t = 0 card. cycle (2) (3)

where QA(t) is the total arterial volumetric flow rate, QV(t) is the total venous volumetric outflow rate, QCSF(t) is the rate of CSF volumetric outflows through the foramen magnum, and T is the time period of the cardiac cycle. Equation 3 is equivalent to the Monro-Kellie doctrine assumption of constant intracranial volume averaged over the cardiac cycle. The MRI method measures the small intracranial volume change (ICV) during the cardiac cycle using a previously described craniospinal flow-volume model [15]. The model, shown in Fig. (3), includes the intracranial and spinal canal compartments, as well as the following inputs and outputs: arterial inflow, venous outflow, and CSF oscillatory flow between the cranium and the spinal canal. During the systolic phase of the cardiac cycle, arterial blood inflow is greater than venous outflow; thus, the intracranial blood volume increases. This produces a rise in pressure that, in turn, forces CSF to flow from the cranium into the spinal canal. During the diastolic phase, blood volume decreases and the CSF flow is reversed. Motion-sensitive MRI

Fig. (2). (From Marmarou et al. 1975). The volume-pressure response test for invasive measurement of compliance). The pressure trace showing the increase in pressure following injection of a volume of fluid, V. The elastance is then derived from the ratio of the resultant pressure change (Pp-P0) to the amount of injected volume, DV.


Current Medical Imaging Reviews, 2006, Vol. 2, No. 1

Alperin et al.

Fig. (3). The craniospinal flow-volume-pressure model proposed by Alperin et al. [15]. This model shows the arterial inflow, the venous outflow and the CSF flow that oscillates between the cranium and spinal canal. During systole arterial blood inflow is greater than the venous outflow which is accommodated by the CSF movement to the spinal canal. The CSF moves to the cranium during diastole as the venous outflow becomes greater.

techniques allow for direct imaging and quantitation of these flow dynamics. 3. Measurements of CSF Flow, Total Cerebral Blood Flow, and Volume Change During the Cardiac Cycle by MRI Flow measurements with dynamic velocity-encoded phase contrast can be reliably and accurately made [25, 26]. The reported accuracy of the MR velocity and volumetric flow measurements in non-steady (pulsatile) flow are within 7.5% and 2.8%, respectively [25]. The cine PCMR technique provides a series of images with values of picture elements proportional to the velocity at that location (velocity maps). The phase contrast MRI technique is based on the principle

that the precession frequency of the protons is proportional to the magnetic field strength. Therefore, velocity can be phased-encoded by application of magnetic field gradients. Fig. (4) demonstrates the principle of the velocity encoding phase contrast technique. When a gradient field is applied along an axis for a short time, the proton's phase will change based on its location along that axis. When a bipolar (positive and then negative) gradient field is applied, the phase of the stationary protons (located at +x) will increase during the positive portion (lobe) of the bipolar gradient and then will decrease during the negative lobe. If the lobes were of equal area, no net phase change would occur. However, moving protons, such as those in the blood or CSF, will experience different field strength during each lobe due to

Fig. (4). A schematic representation of the effect a magnetic field gradient has on stationary protons. Left: Protons at different locations along the x-axis accumulate a different phase when a magnetic field gradient is applied. Right: Time evolution of the phase of a stationary proton located at +x that is experiencing a bipolar field gradient. The positive phase shift caused by the positive lobe is canceled after the application of the negative lobe resulting with no phase shift for a stationary proton. A moving proton will accumulate a phase shift that is proportional to its velocity.

A Review of Wavelet Denoising in MRI and Ultrasound Brain Imaging

Current Medical Imaging Reviews, 2006, Vol. 2, No. 1 121

their change in position; this will result in a net phase change proportional to the proton velocity. Examples of MRI phase contrast images of CSF and blood flow are shown in Fig. (5) and Fig. (6), respectively. The oscillatory CSF flow between the cranial and the spinal compartments is visualized in images taken in a transverse anatomical orientation through the upper cervical spinal canal. The location of this plane is indicated on a midsagittal T1 weighted MR image shown in Fig. (5a). Fig. (5b) depicts outflow (white pixels) during the systolic phase and Fig. (5c) depicts inflow (black pixels) during the diastolic phase. Fig (5d) depicts a 3D plot of the velocities in a region of interest containing the CSF space and an epidural vein. The CSF flow is imaged with a low velocity encoding and the faster blood flow through the neck arteries and veins is imaged using high velocity encoding. The location of the imaging plane used for blood flow measurement is shown in Fig. (6a) and a velocity encoded image of blood flow is shown in Fig. (6b). Fig (6c) depicts a 3D plot of the

velocities in a region of interest containing the internal carotid and vertebral arteries and the jugular vein. Volumetric flow rates are obtained by integration of the velocities throughout a lumen cross-sectional area. An automated method for delineation of the lumen boundaries, Pulsatility Based Segmentation [26], has been applied to improve the reliability of the CSF and blood flow volumetric flow rate measurements. Velocity waveforms inside a lumen are similar to each other while those outside the lumen region are different. Pixels with velocity waveform similar to a reference velocity waveform selected from inside a lumen of interest are segmented and identified as the lumen region. The degree of similarity between waveforms is utilized to differentiate between lumen and background pixels. The method incorporates computation of an unbiased threshold to increase measurement reproducibility. The total volumetric arterial flow rate -that is, total cerebral blood flow (TCBF) - is calculated directly from the sum of the volumetric flow through the four vessels carrying





Fig. (5). A. Anatomical mid-sagittal T1 weighted MR image showing the location of the axial plane used for CSF flow measurement (dark line). B and C are Phase-contrast MRI images of CSF flow in the spinal canal. B. CSF flow during systole. C. CSF flow during diastole. The pixel values in these images are proportional to velocities in a direction perpendicular to the image plane. Gray- static tissue, white- outward flow (caudal direction), and black- inward flow (cranial direction). D. a 3D plot of the CSF velocities during systole.


Current Medical Imaging Reviews, 2006, Vol. 2, No. 1

Alperin et al.



C Fig. (6). a. A blood vessel MRI scout image showing the location of the axial plane for blood flow measurement (dash line). B. A Phase Contrast MRI image of blood flow through that location. Black pixels indicate arterial inflow and white are venous outflow. C. A 3D plot of the blood flow velocities. In the right carotid artery, vertebral artery and jugular vein.

blood to the brain (internal carotid and vertebral arteries) over one cardiac cycle. The venous blood outflow is derived by summation of the flow through the jugular veins, and the epidural, vertebral and deep cervical veins when present. The later veins are parts of the cervical venous plexus, which is a secondary venous outflow pathway. Flows through the jugular veins do not always represent the entire venous outflow. Infact, we found that the measured total venous flow can be lower than the total measured arterial inflow. This may be due to some venous outflow through additional secondary channels (e.g., cervical plexus, ophthalmic veins). This unmeasured venous flow is accounted for by the constraint of no mean net change in the intracranial volume over one cardiac cycle as formulated in equation 3. An example of the volumetric flow waveforms for CSF, arterial inflow, and venous outflow measured in a healthy volunteer is shown in Fig. (7).

The rate of the time-varying intracranial volume change (net transcranial volumetric flow rate) is obtained by subtracting outflow rates from inflow rates at each time point. The intracranial volume change (delta of volume from a given reference point) is obtained by integrating that waveform with respect to time. The net transcranial volumetric flow rate and the ICV waveforms, obtained from the CSF and blood flow waveforms shown in Fig. (7), are shown in Fig. (8a) and Fig. (8b), respectively. The maximum volume change in this example is close to 0.5 mL. 4. MRI Measurements of the ICP Change During the Cardiac Cycle The change in intracranial pressure during the cardiac cycle is derived from the change in the CSF pressure gradient. A method to measure pressure gradient of pulsatile flow in tubes with MRI was reported by Urchuk and Plewes

A Review of Wavelet Denoising in MRI and Ultrasound Brain Imaging

Current Medical Imaging Reviews, 2006, Vol. 2, No. 1 123

Convective accelerations are those related to velocity changes with a change in position (i.e. fluid with constant flow rate accelerates as the conduit's cross-sectional area decreases). The third term is momentum loss due to viscous effects. Our lab has adapted this method to calculate CSF pulsatile pressure gradients in the upper cervical spine [17]. The assumptions of a rigid conduit and incompressible fluid are justified in the case of CSF flow in the cervical spinal canal. The CSF pressure gradient waveform is calculated from the velocity-encoded MRI images of the CSF pulsatile flow. The inertia component of the pressure gradient is approximated by the first order central difference templates of the time series images. The shear (viscous) component is derived by using a pair of second-order central difference operators. The shear and inertia components are summed to obtain a pressure gradient value for each time point of the cardiac cycle. Fig. (9) shows the MRI-derived CSF pressure gradient waveform from a healthy subject. The open circle curve shows the contribution of the viscous term to the total pressure gradient. It is important to note that the overall contribution of the viscous term is negligible. Several pathologies, particularly the inflammatory process, are associated with a change in the viscosity of the CSF. Because of the negligible contribution of the viscous term, a pathologic change in CSF viscosity is not expected to affect the CSF pressure gradient waveform. VALIDATION STUDIES 1. Validation Accuracy and Reproducibility of the ICV Measurement The volume of the entire intracranial space is on the order of 1500mL, the change in that volume during the cardiac cycle is on the order of 1 mL, less than 0.1%. Therefore, reproducible and accurate ICV measurement poses a challenge. The validity of the ICV measurement was assessed from studies in patients with pathologies that affect the ICV in a predictable manner, and the inherent accuracy was assessed with a specially designed craniospinal flow phantom [29]. These studies were necessary since currently there is no alternative method that could be used as a reference.

Fig. (7). The volumetric flows into and out of the cranial vault during the cardiac cycle derived from the MRI phase contrast scans. Total arterial inflow (filled circles), venous outflow (open) and the cranial-to-spinal CSF volumetric flow rate (diamonds) during the cardiac cycle. Note that arterial inflow is greater than venous outflow during systole.

[27]. Pulsatile pressure gradients are derived from the MRI velocity encoded phase contrast images using the NavierStokes relationship between pressure gradient and temporal and spatial derivatives of the fluid velocity for incompressible fluid in a rigid tube [28]. This relationship is shown in equation 4 V +V V +µ 2 V P = ( ) t



where 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 upon it. The first and 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 (i.e. blood accelerates during systolic phase of the cardiac cycle).

Fig. (8). Left: The MRI derived Net transcranial volumetric flow rate waveform. Right: The Intra Cranial Volume Change during the cardiac cycle derived by integrating the net transcranial volumetric flow waveform on the left. Note that the maximal volume change in this subject is 0.5 mL.


Current Medical Imaging Reviews, 2006, Vol. 2, No. 1

Alperin et al.

be determined independently of the MRI-based method. A flow phantom composed of a nearly rigid container and tubes leading to and from the container were constructed to simulate the intracranial space and the arterial and venous flow channels [29]. The average, ICV, and oscillatory CSF volume (the volume that moves back and forth between the cranium and the spinal canal) were calculated and compared with independent measurements of these quantities. The intracranial volume change was measured independently of the MRI measurement by blocking the arterial and venous channels and deflecting the syringe to produce a pressure value equal to the pulse pressure recorded while the channels are open. Results from these measurements are summarized in Table 1. The average maximum ICV value measured by MRI was within 5% of the ICV value measured independently. The high reliability by which ICV is measured by MRI is attributed to the excellent temporal response of non-steady flow measurement with the cine phase contrast technique. The phantom experiments demonstrated high intrinsic accuracy and reproducibility of the method. In contrast to the phantom studies, early evaluation of measurement reproducibility in human subjects demonstrated a much larger measurement variability of 18% [17]. This rather large variability needed to be improved before the method could become clinically viable. The source of this variability was found to be a change in heart rate that often occurred between the two separate cine phase contrast scans used for imaging of the fast blood flow and the slow CSF flow. Two separate scans were used because scanning parameters needed to be adjusted for the measured velocity range. This problem was solved by the development of a new MRI data acquisition scheme, the dual velocity encoding scheme, which allows for simultaneous acquisition of fast and slow flow in an interleaved fashion [30]. With this technique, the CSF and the blood flow measurements are always synchronized-- regardless of a change in the patient's heart

Fig. (9). A. The MRI-derived CSF pressure gradient waveform from a healthy volunteer (filled). The open circles show the small contribution of the viscous term in the Navier-Stokes equation (eq. 1).

Measurements in patients with a skull defect before and after corrective surgical procedure (cranioplasty) offered an opportunity to test the validity of the ICV measurement by MRI as a larger volume change is expected during the "open skull" state. In all 3 patients studied, a larger maximum ICV was measured before cranioplasty, as expected. The ICV waveforms measured before and after cranioplasty in one of these patients are shown in Fig. (10). The systolic ICV before and after surgery were 1.01 mL and 0.52 mL, respectively. Measurement results obtained in the cranioplasty patients supported the validity of the method but can not be used for assessment of the method accuracy since the true ICV value at the two states is unknown. Instead, the inherent accuracy of the method was determined using a specially built craniospinal flow phantom where the ICV value could

Fig. (10). Intracranial Volume change waveforms in a patient with a skull defect that was later corrected. The maximum volume change at each state is indicated on the graph. Left: CT rendered image of the open skull state. Middle: ICV during that state. Right: ICV after skull closure. As expected the maximum volume change is reduced after cranioplasty (skull closure) surgery.

A Review of Wavelet Denoising in MRI and Ultrasound Brain Imaging

Current Medical Imaging Reviews, 2006, Vol. 2, No. 1 125

Table 1.

Results from the Flow Phantom Study for the Validation of the IC Volume Change Measurement Accuracy and Reproducibility

Mean Value (mL/cardiac cycle) Intracranial Volume Change Oscillatory CSF Volume 1.94 1.41

Std. Dev. mL 0.09 0.11

% SD (5%) (8%)

Independent Measurement 2.0 +/- 0.1 1.35 +/- 0.1

rate. A study comparing the ICV measurement variability in human subjects with the new and the previous MRI data acquisition techniques demonstrated overall improvement of 2.7 times in the reproducibility of the ICV measurement with the dual VENC MRI acquisition technique [30]. The current ICV measurement reproducibility in humans. 2. Validation Accuracy and Reproducibility of the ICP Change Measurement The intracranial pressure change during the cardiac cycle is derived from the change in the CSF pressure gradient. The relationship between time varying change in pressure and in pressure gradient was evaluated experimentally with a nonhuman primate [17] and theoretically with computational fluid dynamics (CFD) [31]. The experimental validation required the use of a large nonhuman primate (baboon) because hydrodynamics is scale-dependent and important fluid dynamic parameters such as the size of the spinal canal and heart rate in baboons are similar to those of humans. A MRI image of the cervical spine anatomy of a baboon demonstrating this similarity is shown in Fig. (11).

invasively measured pulse pressure at the three different values of absolute ICP was found to be close to a linear relationship. The relationship between the pulse pressure and the amplitude of the pressure gradient was also studied with CFD simulation [31]. In these simulations the spinal canal geometry was represented with two concentric cylinders and the calculations were based on the finite volume method to solve the Navier-Stokes equations. A nearly linear relationship was demonstrated between the pulse pressure and pulse pressure gradient calculated for two different boundary conditions, high and low flow rate, which correspond to high and low pulse pressure values. The reproducibility of the pressure change measurement by MRI is therefore determined by the reproducibility of the peak-to-peak amplitude of the CSF pressure gradient measurement. The CSF pressure gradient measurement reproducibility was assessed from repeated MRI scans of healthy subjects. A measurement variability of 8% was found in these studies [17]. The MRI-derived intracranial compliance and pressure (MR-ICP) are derived from the ratio of the pressure and volume changes. Therefore the overall MR-ICP measurement reproducibility is the square root of the sum of the square of individual fractional standard deviations, i.e., the fractional SD of the volume and the pressure change measurements, which are currently approximately 8% each. Therefore the overall MR-ICP measurement variability is approximately 10%. 3. Early Assessment of the MR-ICP Method Performance in Humans Direct comparison between MRI derived elastance index (the ratio of the pressure change to volume change) and mean ICP value measured invasively at the time of the MRI study in 5 patients demonstrated a linear relationship, as expected, with high degree of correlation (R2 = 0.96) [17]. This linear relationship allows for expression of the MRIderived elastance index in units of absolute ICP value. The false positive rate of the method was determined by measurements in healthy subjects. A total of 71 simultaneous measurements of total CBF and ICP were performed in twenty three young adults (20 males, range 20 to 39, mean age 25 +/- 5 years) with no known neurological problems. The distribution of the MR-derived ICP values (MR-ICP) and TCBF measurements are shown in Fig. (12). ICP values range from 3.5 to 17.1 mmHg; most measurements were between 7 and 9 mmHg. Wide distribution of ICP values measured invasively in normal human subjects has been previously reported: from 2 to 16 mmHg in one study [32] and up to 18 mmHg in another [33]. Based on the current view that an ICP value of 20 mmHg is a critical threshold for

Fig. (11). Anatomical sagittal 2D T1-weighted MR images of the cervical spine region of a baboon (left) and a human (right). The similarity between the cervical spinal anatomy of the baboon and the human is demonstrated.

In the baboon validation study, pressures and CSF flow velocities were measured at three different ICP values (baseline, elevated, and reduced pressure) achieved by either addition or by withdrawal of fluid from the CSF space. The relationship between the amplitudes of the CSF pressure gradient measured by MRI, and the amplitude of the


Current Medical Imaging Reviews, 2006, Vol. 2, No. 1

Alperin et al.

Fig. (12). The distribution of the MRI-derived ICP (MR-ICP) measurements (left) and the total CBF (right). The mean value and the relative standard deviation were and 9.6 ±31%, and 700 mL/min ±13%, respectively. Note that the ICP values are distributed over a wider range relative to the CBF values.

elevated ICP [34], no false positives were measured with MRI (i.e., a zero-false positive rate). The distribution of MRI-derived TCBF measurements in the same subjects ranged from 514 to 956 mL/min. It is interesting to note that the relative range of TCBF values span a much more narrow range than ICP. The percentage SD of the ICP and TCBF measurements are 31% and 13% respectively. It is known that the blood flow to the brain is tightly regulated; a more narrow distribution of TCBF may reflect that degree of regulation. STUDYING THE EFFECT OF BODY POSTURE ON INTRACRANIAL PRESSURE AND CBF WITH MRI The cerebral biomechanical state is different in the upright and supine postures. The application of the MR-ICP methodology to study healthy subjects in supine and upright postures provides an insight into the coupling that exists between CSF and blood flow and the way by which intracranial compliance and pressure affect CSF and blood flow dynamics [35]. Ten healthy volunteers with no history of neurological problems or head trauma were imaged in both upright and supine positions in a vertical gap MRI scanner (Signa SP/i 0.5T, GE Healthcare, Milwaukee). The vertical gap allows the subject to be positioned seated

upright. The MR-ICP scanning protocol was applied in both postures. An MR angiography technique (2D Time of Flight) was added to the MR-ICP protocol to visualize the cerebral venous drainage at the two postures. Representative 3D MR images of the neck veins at the two postures are shown in Fig. (13). These images demonstrate a different pattern of venous drainage between the supine and sitting postures. In the supine posture, the flow is primarily through the internal jugular veins (IJV) while in the upright posture, the IJV partially or fully collapse and the main pathway for venous outflow is the cervical venous plexus. Plots of the total arterial inflow and venous outflow waveforms from one of the subjects are shown in Fig. (14). Arterial inflow (open circles) and venous outflow (filled circles) during the cardiac cycle in the supine position are shown on the left and in the sitting position on the right. While the arterial volumetric flow rates are slightly lower in the sitting posture the overall shape (dynamics) and amplitude (the difference between the maximum and the minimum) in the two postures are similar. In contrast, the shape and the amplitude of the venous flow were different between the two postures; the venous flow is much less pulsatile in the upright position.

Fig. (13). The 2DTOF MR images of the Veins (MRV) in the neck from one of the healthy subjects obtained in the upright (right) and supine (left) postures. Both jugular veins are visualized in the supine position while in the upright posture these veins are collapsed. The redistribution of the venous drainage in the upright posture to the epidural and vertebral veins is clearly demonstrated.

A Review of Wavelet Denoising in MRI and Ultrasound Brain Imaging

Current Medical Imaging Reviews, 2006, Vol. 2, No. 1 127

Fig. (14). MRI-based measurements of the arterial inflow (open circles) and venous outflow (filled circles) volumetric flow waveforms during the cardiac cycle in a supine (left) and sitting (right) postures. Note the slightly lower arterial flow rates in the sitting position. While the amplitude of the arterial waveform at the two postures are similar, the amplitude of the venous flow in significantly smaller (less pulsatile flow) at the upright posture, suggestive of higher cerebral vascular compliance.

Plots of the measured CSF flow waveforms at the two postures are shown in Fig. (15). The CSF flow waveforms are plotted together with the net transcranial blood flow (arterial - venous) to demonstrate how the net blood flow drives the CSF flow. Note that the in the supine position, the CSF flow waveform follows the A-V flow waveform "more closely"- an indication of a lower intracranial compliance. The mean and SD of the derived measurements in the two postures are summarized in (Table 2). The differences in the measured parameters between the two postures are statistically significant. They include a 2-fold increase in the ICV, a 2-fold reduction in the CSF volume that moves in and out of the cranium, and a large increase (almost 3-fold) in the intracranial compliance with a corresponding decrease in pressure. All these changes are consistent with a more compliant intracranial compartment caused by a smaller mean volume of blood and CSF residing in the intracranial space in the sitting position.

Another parameter strongly affected by posture is the net transcranial blood flow (A-V). The amplitude of the net transcranial blood flow is much larger in the sitting position than in the supine position. This is the result of the reduced pulsatility of venous outflow together with relatively unchanged arterial inflow dynamics. A larger net transcranial blood flow results in a larger volume of blood entering the cranium during systole. However, because the overall mean intracranial volume is smaller in the sitting position, the blood entering the cranium can be easily accommodated even with less CSF being displaced into the spinal canal. The combination of smaller amount of CSF that leaves the cranium with each cardiac cycle, and larger net transcranial inflow explains the larger ICV measured in the sitting position.

Fig. (15). Graphs depicting the A-V and CSF flow waveforms in the supine (Left) and upright (Right) postures. Note that the CSF follows the A-V less closely in the upright posture compared to the supine posture.


Current Medical Imaging Reviews, 2006, Vol. 2, No. 1

Alperin et al.

Table 2. Mean and SD of Main MRI-Derived Physiologic Parameters Measured in Supine and Upright Postures

OTHER POTENTIAL CLINICAL ROLES FOR THE MR-ICP METHOD Physiological parameters measured by the MRI-based method, e.g., total cerebral blood flow, intracranial compliance, and pressure, are potentially important for many neurological problems. They are especially important in traumatic brain injury (TBI) because it has been shown that TCBF and ICP are important markers for the progression of head trauma to secondary brain injury [40, 41]. The MRI method provides these parameters noninvasively. The role of MRI-based ICP measurement would be different from that of the invasive technique. While invasive monitoring provides continuous ICP measurements, the MRI study provides a single time point measurement and therefore it is a diagnostic test. There are several clinical settings in which a "snapshot" of ICP may be beneficial. Complementing neurological and CT exams in intermediate severity of head trauma may be one. Placement of an invasive monitor to track ICP continuously is recommended for patients with severe head injuries, defined as Glasgow Coma Scale (GCS) score = 3-8. However, the necessity of ICP monitoring for patients with intermediate GCS scores (9-12) ­ and in particular, those with a normal CT scan on presentation ­ has been the subject of much debate [42, 43]. Ten to thirty percent of these patients will develop increased ICP over the first few days after injury [44-46]. Invasive monitoring techniques are not routinely employed in this population. Yet, at the same time, it is not clear which of these patients will progress to an increased pressure state. Likewise, patients with diffuse axonal injury (DAI) may demonstrate a GCS score in the severe injury range but be found to have normal ICP upon placement of an invasive monitor. In these settings, noninvasive MR-ICP measurement would provide a means of objective assessment of the need for an invasive monitor without incurring the potential morbidities. MR-ICP may also play an important role in diagnosis and serial evaluation of several chronic disorders that may be associated with changes in ICP. Such processes include hydrocephalus, pseudotumor cerebri, intracranial mass lesions, and toxic-metabolic encephalopathy (where a depressed level of consciousness may or may not correspond to an increased ICP). Single time measurement of ICP may be helpful in the evaluation of patients with possible ventriculoperitoneal shunt malfunction ­ particularly, young children who present with nonspecific complaints and/ or who are unable to communicate their symptoms. ACKNOWLEDGEMENTS Supported in part by the NIH/NIRR (RR14242-01) and the Dana Foundation's Clinical Hypotheses Program in Imaging. REFERENCES

[1] [2] [3] [4] O'Connel JEA. Vascular factor in intracranial pressure and maintenance of cerebrospinal fluid circulation. Brain 1943; 66: 204228. Antoni N. Pressure curves from the cerebrospinal fluid. Acta Med Scand 1946; suppl 170: 431-462. Bering EA, Ingraham FD. Arterial pulsation of the cerebrospinal fluid. Trans Amer Neurol Assoc 1953; 49-54. Bering EA. Choroid plexus and arterial pulsations of cerebrospinal fluid. Arch Neurol Psychiat 1955; 73: 165-172.

Supine (Mean +/- SD) TCBF (mL/min) Venous Flow in IJV (mL/min) Osc. CSF Volume (mL) Max. ICV (mL) Compliance Index ICP (mmHg) 825 +/- 166 614 +/- 143 0.55 +/- .12 0.48 +/- .15 7.3 +/- 2.6 10.6 +/- 3.6

Upright (Mean +/- SD) 724 +/- 127 304 +/- 261 0.23 +/- .11 0.89 +/- .44 20.2 +/- 10.7 4.5 +/- 1.82

* All differences are statically significant with a P value of 0.001 or smaller

A commonly accepted explanation for a lower ICP in an upright posture argues that venous flow is increased in that position and this avoids compression of the jugular veins [36]. This explanation is clearly inaccurate since venous flow in the jugular veins is considerably lower in an upright posture [37]. The lower ICP is therefore, most likely, the result of the increased intracranial compliance due to a lower mean volume of the intracranial compartment caused by the reduced volume of CSF and blood. The shift of blood and CSF volume out of the cranium occurs during the transition from the supine to the upright posture. APPLICATION OF MR-ICP TO STUDY THE EFFECT OF DECOMPRESSION SURGERY IN CHIARI MALFORMATIONS Chiari I malformation is characterized by displacement of the cerebellar tonsils more than 5 mm caudally through the foramen magnum [38]. However, since no direct correlation has been observed between symptoms and the anatomical severity of the herniation the pathophysiology of the disease cannot be explained by the abnormal anatomy of the craniospinal junction alone. Even though the disease process is not well understood, decompression surgery is now commonly offered as a treatment of choice for patients with Chiari malformation. In order to better understand the effect of the decompression surgery on the intracranial physiology the MR method for measurement of intracranial compliance was applied to study 12 Chiari Malformation patients before and after decompression surgeries. Changes in CSF flow and cord motion in the upper cervical spine and in the overall intracranial compliance were measured and were evaluated for their significance [39]. Following surgery, changes were found, both, in the local CSF flow and cord motion and in intracranial compliance (ICC). However, only the change in ICC, an average increase of more than 60%, was statistically significant. An increased ICC, associated with improved outcome, was measured in 10 of the 12 patients, no significant change was found in one patient, and decreased ICC was measured in one patient whose symptoms persisted following surgery. Intracranial compliance was the most sensitive measure for the effect of the surgery and was the only parameter that reached statistical significance. The study provided evidence that intracranial compliance plays an important role in the pathophysiology of this disorder.

A Review of Wavelet Denoising in MRI and Ultrasound Brain Imaging [5] [6] [7] [8] Hamit HF, Beall Jr. AC, DeBakey ME. Hemodynamic influences upon brain and cerebrospinal fluid pulsations and pressures. J Trauma 1965; 5: 174-184. Dunbar HS, Guthrie TC, Karpell B. A study of the cerebrospinal fluid pulse wave. Arch Neurol 1966; 14: 624-630. Dardenne G, Dereymaeker A, Lacheron JM. Cerebrospinal fluid pressure and pulsatility. Europ Neurol 1969; 2: 193-216. Dereymaeker A, Stevens A, Rombouts JJ, Lacheron JM, Pierquin A. Study on the influence of the arterial pressure upon the morphology of cisternal CSF pulsations. Europ Neurol 1971; 5: 107-114. Hamer J, Alberti E, Hoyer S, Wiedemann K. Influence of systemic and cerebral vascular factors on the cerebrospinal fluid pulse waves. J Neurosurg 1977; 46: 36-45. Bergstand G, Bergstrom M, Nordell F, Stahlberg A Ercsson A, et al. Cardiac gating MRI imaging of the cerebrospinal fluid flow. J. Comput. Assist. Tomogr 1985; 9: 1003-1006. Bradley WG, Kortman KE, Burgonye B. Flowing cerebrospinal fluid in normal and hydrocephalic states: appearance in MR image. Radiology 1986; 159: 611-616 Finberg DA, Mark SA. Human brain motion and cerebrospinal fluid circulation demonstrated with MR velocity imaging. Radiology 1987; 60: 423-427. Greitz D, Wirestam R, Franck A, Nordell B, Thomsen C, Stahlberg F. Pulsatile brain movement and associated hydrodynamics studied by magnetic resonance phase imaging. Neuroradiology 1992; 34: 370-380. Enzmann D, Pelc N. Cerebrospinal fluid flow measured by Phasecontrast Cine MR. AJNR 1993; 14: 1301-1307. Alperin N, Vikingstad EM, Gomez-Anson B, Levin DN. Hemodynamically independent analysis of cerebrospinal fluid and brain motion observed with dynamic phase contrast MRI. Magnetic Resonance in Medicine 1996; 35: 741-754. Balédent O, Henry-Feugeas MC, and Idy-Peretti I, Cerebrospinal fluid dynamics and relation with blood flow: a magnetic resonance study with semiautomated cerebrospinal fluid segmentation. Invest Radiol 2001; 36: 368-377. Alperin NJ, Lee SH, Loth F, Raksin PB, Lichtor T. MR-Intracranial Pressure (ICP): A method to measure intracranial elastance and pressure noninvasively by means of MR Imaging: Baboon and human study. Radiology 2000; 217: 877-885. Loth FM, Yardimici MA, and Alperin N. Hydrodynamic modeling of Cerebrospinal Fluid Motion within the spinal cavity. Jour. of Biomechanical Engineering. 2001, Vol. 123, pp. 71-79. Raksin P, Alperin N, Surapaneni S, Lichtor T. Noninvasive Intracranial Compliance and Pressure from Dynamic MR Imaging of Blood and CSF Flows: Review of Principles, Implementation, and Other Noninvasive Approaches. Neurosurg. Focus. 2003; 14(4);1-8 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. Marmarou A, Shulman K, La Morgese J. Compartmental analysis of compliance and outflow resistance of the CSF system. J of Neurosurgery 1975; 43: 523-534. Sklar FH, Elashvili I. The pressure volume function of brain elasticity. J of Neurosurgery 1977; 47: 670-679. Szewczykowski J, Sliwka S, Kunicki A, Dyko P, Korsak-Sliwaka J. A fast method of estimating the elastance of intracranial system. J of Neurosurgery 1977; 47: 19-26. Magendie F. Recherches physiologiques et cliniques sur le liquide céphalorachidien ou cérébro-spinal. Paris: Librairie Medicale de Mequigenon-Marvis Files, 1842; p66. [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36]

Current Medical Imaging Reviews, 2006, Vol. 2, No. 1 129 Frayne R, Steinman DA, Ethier CR, Rutt BK. Accuracy of MR phase contrast velocity measurements for unsteady flow. J Magn Reson Imag 1995; 5(4): 428-431. Alperin N, Lee SH. PUBS: Pulsatility based segmentation of lumens conducting nonsteady flow. Magnetic Resonance in Medicine 2003; 49: 934­944. Urchuk SN, Plewes DB. MR measurement of pulsatile pressure gradients. JMRI 1994; 4: 829-836. Bird RB, Stewart WE, Lightfoot EN. (1960). "Transport Phenomena." John Wiley and Sons, New York. Alperin N, Kadkhodayan Y, Loth F, Yedavalli R. MRI measurements of intracranial volume change: A phantom study. Proc. Intl. Soc. Mag. Reson. Med. 9 2001; 3: 1981. Dhoondia H, Alperin N. Improved MR-Intracranial Pressure (MRICP) Measurement using a new Data Acquisition Technique. Intl. Soc. Mag. Reson. Med. 1 2003. Loth F, Yardimci MA, Alperin N. Hydrodynamic modeling of cerebrospinal fluid motion within the spinal cavity. J of Biomechanical Engineering 2001; 123: 71-79. Merritt HH, Fremont ­Smith F. The cerebrospinal fluid. Philadelphia, WB Saunders Co. 1937. Ayer JB. Cerebrospinal fluid pressure from the clinical point of view. Assn. Res. Nerv. Mental Dis 1924; 4: 159-171. Marmarou A, Eisenberg HM, Foulkes MA. Impact of ICP instability and hypertension on outcome in patients with severe head trauma. J. Neurosrg. 1991; 75: s59-s66. Alperin N, Lee S, Sivaramakrishnan A, Hushek S. Quantifying the effect of posture on intracranial physiology in humans by MRI flow studies. J Mag Res Imag 2005; 22(5): 591-596. Nathan BR. In: Goetz CG, Pappert EJ Eds, Textbook of Clinical Neurology, Cerebrospinal Fluid and Intracranial Pressure; Treatment of Increased Intracranial Pressure. Philadelphia, WB Saunders Co. 1999; 489. Valdueza JM, Munster TV, Hoffman O, Schreiber S, Einhaupl KM. Postural dependency of the cerebral venous outflow. The Lancet 2000; 355: 200-201. Elster AD, Chen MYM: Chiari I Malformations: clinical and radiologic reappraisal. Radiology 1992; 183: 347-353 Sivaramakrishnan A, Surapaneni S, Lichtor T, Alperin N. Evaluating the Effect of Decompression Surgery on CSF Flow and Intracranial Compliance in Patients with Chiari Malformation Using MRI Flow Studies. Neurosurgery 2004; 55: 1344-1351. Gennarelli TA. In: Cooper PR Ed, Head Injury. Cerebral concussion and diffuse brain injuries. New York, Williams and Wilkins. 1993; 137-158. Siegal JH, Gens DR, Marmentov T, et al. Effect of associated injuries and blood volume replacement on death, rehabilitation needs, and disability in blunt traumatic brain injury. Crit Care Med. 1991; 19: 1252-1265. Schmidt B, Klingelhöfer 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. 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. Eisenberg HM, Gary Jr HE, 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. Lobato RD, Sarabia R, Rivas JJ, et al. Normal CT scans in severe head injury. J Neurosurg 1986; 65: 784-789. 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: 4650.

[9] [10] [11] [12] [13]

[14] [15]


[37] [38] [39]


[18] [19]

[40] [41]



[21] [22] [23] [24]


[44] [45] [46]

Received: February 10, 2005

Revised: March 07, 2005

Accepted: March 11, 2005


13 pages

Find more like this

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate


You might also be interested in

Short-answer Questions and MCQs in Anaesthesia and Intensive Care
basic sciences mcqs