Magnetic Resonance Imaging 32 (2014) 1284–1289
Contents lists available at ScienceDirect
Magnetic Resonance Imaging journal homepage: www.mrijournal.com
Transfer characteristics of arterial pulsatile force in regional intracranial tissue using dynamic diffusion MRI: A phantom study Hirohito Kan a, b, 1, Tosiaki Miyati b,⁎, Harumasa Kasai a, 1, Nobuyuki Arai a, 1, Naoki Ohno b, 2, Mitsuhito Mase c, 1, Yuta Shibamoto a, 1 a b c
Department of Radiology, Nagoya City University Hospital, 1-Kawasumi, Mizuho-cho Mizuho-ku, Nagoya City, Aichi, 4678602, Japan Division of Health Sciences, Graduate School of Medical Science, Kanazawa University, 5-11-80 Kodatsuno, Kanazawa, Ishikawa, 9200942, Japan Department of Neurosurgery, Nagoya City University Hospital, 1-Kawasumi, Mizuho-cho Mizuho-ku, Nagoya City, Aichi, 4678602, Japan
a r t i c l e
i n f o
Article history: Received 19 March 2014 Revised 29 July 2014 Accepted 21 August 2014 Keywords: Magnetic resonance imaging Diffusion weighted imaging Apparent diffusion coefficient Water molecular fluctuation
a b s t r a c t Introduction: To clarify the mechanism underlying apparent diffusion coefficient (ADC) changes in regional intracranial tissue during the cardiac cycle, we investigated relationships among ADC changes, volume loading, and intracranial pressure using a hemodialyzer phantom in magnetic resonance imaging (MRI). Materials and Methods: The hemodialyzer phantom consisted of hollow fibers (HF), the external space of HFs (ES), and a gateway of dialysis solution, filled with syrup solution and air. The high-volume and lowvolume loadings were periodically applied to HFs by a to-and-fro flow pump, and syrup solution was permitted to enter or leave HFs during the volume loading cycles. ADC maps at each volume loading phase were obtained using ECG-triggered single-shot diffusion echo-planar imaging. Dynamic phase contrast MRI was also used to measure volume loading to the phantom. We compared the ADC changes, volume loading, and intracranial pressure in the phantom during the cardiac cycle. Results: ADC changes synchronized significantly with absolute volumetric flow rate change. The maximum ADC change was higher in high-volume loading cycles than in low-volume loading cycles. Results showed that the water molecules in ES fluctuated according to the force transferred from HF to ES. ADC changes were dependent upon the volumetric flow rate during the cardiac cycle. Conclusions: Our original phantom allowed us to clarify the mechanism underlying water fluctuations in intracranial tissues. Measurement of maximum changes in ADC is an effective method to define the transfer characteristics of the arterial pulsatile force in regional intracranial tissue. © 2014 Elsevier Inc. All rights reserved.
1. Introduction In the past years, several studies have attempted to use quantitative imaging techniques to evaluate the biomechanical properties of the brain using magnetic resonance imaging (MRI). MRI-based measurement of the biomechanical properties of the brain may be useful in characterizing and diagnosing of intracranial diseases not easily evaluated using conventional imaging contrasts. Magnetic resonance elastography (MRE), a technique that mea-
⁎ Corresponding author. Tel.: +81 76 265 2500. E-mail addresses: [email protected] (H. Kan), [email protected] (T. Miyati), [email protected] (H. Kasai), [email protected] (N. Arai), [email protected] (N. Ohno), [email protected] (M. Mase), [email protected] (Y. Shibamoto). 1 Tel.: +81 52 851 5511. 2 Tel.: +81 76 265 2500. http://dx.doi.org/10.1016/j.mri.2014.08.026 0730-725X/© 2014 Elsevier Inc. All rights reserved.
sures biomechanical properties by directly applying shear waves to the brain, has been used in the past [1–4]. However, MRE was limited because it requires specialized equipment to generate the shear waves. Another technique was to measure the intracranial compliance (ICC) calculated from blood flow and cerebral spinal fluid (CSF) flow, and movement of the spinal cord during the cardiac cycle using a retrospectively gated phase-contrast cine MRI. Alperin et al. [5] developed an ICC analysis for evaluating the mechanical coupling of brain and CSF oscillations to the driving vascular pulsations. Arterial flow into the cranium induces venous flow, CSF flow, and spinal cord movement during the cardiac cycle. The relationship between these measurements is well established [6–13]. Next, the intracranial volume loading during the cardiac cycle was calculated from the differences between the inflow and outflow volumes of the cranium at each time point in the cardiac cycle [7]. ICC represented the ability of the intracranial space to accommodate an increase in volume without a large increase in intracranial pressure [5,6]. ICC analysis has been used in patients
H. Kan et al. / Magnetic Resonance Imaging 32 (2014) 1284–1289 Fig. 1. (a) Phantom for simulating change ADC changes during the cardiac cycle following volume loading. (b) Schematic illustration of the hemodialyzer phantom. Volume loading was applied to the hemodialyzer, and syrup was permitted to enter or leave HFs through the tube only. (c) Apparent diffusion coefficient image without volume loading. (d) Microscopic image of the imaging plane.
1285
1286
H. Kan et al. / Magnetic Resonance Imaging 32 (2014) 1284–1289
with abnormal intracranial conditions such as Chiari malformation and idiopathic normal pressure hydrocephalus [14,15]. This analysis has been helpful in the diagnosis and the subsequent management of these diseases. However, this method has some limitations; for example, the estimated ICC represented information about the whole brain, not a localized intracranial tissue. In a previous work, we found that the apparent diffusion coefficient (ADC) in the white matter changed during the cardiac cycle, despite minimizing the brain bulk motion effect [16]. The regional changed ADC resulted in volume loading of the cranium during the cardiac cycle. We also reported that the variation in ADC during the cardiac cycle was closely related to the regional biomechanical properties and the measurement of the maximum change in ADC during the cardiac cycle (delta-ADC) may be of help in the diagnosis of idiopathic normal pressure hydrocephalus [17]. However, the more detailed mechanisms and relationships among intracranial pressure, volumetric flow and ADC changes during the cardiac cycle remain unclear. The present study aimed to clarify the mechanisms underlying the ADC changes during the cardiac cycle in intracranial tissue using a hemodialyzer phantom. 2. Materials and methods Fig. 1a illustrates the experimental setup of the phantom. The phantom was designed to simulate water diffusion changes in intracranial tissues by arterial blood volume loading. A schematic drawing of the hemodialyzer is shown in Fig. 1b. The hemodialyzer consists of hollow fibers (HF), external spaces of HF (ES), and the gateway of dialysis solution (Hollow Fiber Dialyzer KF-201, Kawasumi Laboratories Inc., Tokyo, Japan). These components represented the blood capillaries, intracranial tissues, and the intracranial capacity, respectively. The intracranial capacity plays a role in dampening the volume loading caused by pulsatile arterial blood flow. HFs and ES were filled with syrup solution adjusted to 0.6 x10 -3 mm2/s ADC in x-y plane of the hemodialyzer phantom. (Fig. 1c). The syrup solution was a viscous liquid consisting of a solution of sugar in water. The pore size of the HF membrane was sufficiently small and the syrup solution could not pass easily between HF and ES. Fig. 1d shows a microscopic image of the cross section of the hemodialyzer. The gateway of the hemodialyzer was filled with air to dampen the cyclic volume loading. Volume loading was periodically applied to the hemodialyzer via a to-and-fro flow hand-made pump. The flow pump generated a cyclic volumetric flow similar to a sine wave. The hemodialyzer was connected to the pump by a polyvinyl chloride tube filled with the syrup solution (Green bubble tube 2302, Covidien Japan, Shizuoka, Japan). The volume loading applied to the hemodialyzer was permitted to enter or leave HF only through the tube. All measurements were carried out using the same volume loading cycle, i.e., 1 Hz. High-volume and low-volume loadings were applied to the phantom. The pump generated a trigger signal, which was used to synchronize with the MR data acquisition. To minimize a susceptibility effect in the MR data acquisition for the volumetric flow measurement, the tube was penetrated through a plastic case filled with water. On a 1.5-T MRI (Gyroscan Intera, Philips Medical Systems International, Best, the Netherlands), the long axis of the phantom was set in a direction parallel to the magnetic field, i.e., the flow direction in HF corresponded to the z-axis. DWIs for each b value were obtained using ECG-triggered multi-phase single-shot diffusion echo-planar imaging (EPI), which is largely insensitive to motion. However, single-shot EPI alone cannot completely eliminate the bulk motion effect of the brain parenchyma because of the long data-sampling window required for single-shot EPI, and artifacts may remain visible [18]. Therefore, single-shot EPI was combined with parallel imaging, half-scan, and rectangular field-of-view (FOV) techniques to minimize the bulk motion effect [19]. The trigger delays from
the R-peak for DWIs were set at regular intervals (30 ms), and we acquired 30 phases. Then, ADC was calculated by Eq (1); ADC ¼
ln S1 − ln S2 b2 −b1
ð1Þ
where S1 and S2 were signal intensities of DWIs at 0 and 1000 b-factors, respectively, and b1 and b2 were 0 and 1000 b-factors, respectively. The ADC map in a direction perpendicular to HF, i.e., the x–y plane, was calculated from DWIs at each volume loading phase to minimize the flow effect in HFs. Then, the ADC change during the volume loading cycle (delta-ADC) was calculated as maximum ADC minus minimum ADC of all volume loading phase images on a pixel-by-pixel basis. To measure the volume loading change, a retrospective ECG-gated phase contrast cine MRI was then performed at a vertical slice plane against the long axis of the tube. An ROI was set in the tube on velocity-mapped images in each volume loading phase to measure the flow velocities into the phantom. The baseline offset of the flow velocity, due to eddy currents, was then corrected using a subtraction process [20]. We multiplied the cross-sectional area of the tube by the flow velocity to obtain the volumetric flow rate. The volume loading change was derived by the temporal integration of the volumetric flow rate as shown in Eq. (2). Z VLðt Þ ¼
V ðt Þ dt
ð2Þ
where VL(t), V(t) and t are the volume loading change, volumetric flow rate through the tube, and time index, respectively. The relationship between the waveforms of ADC and absolute velocity was confirmed using a cross-calibration coefficient (CCC) in Eq. (3). XN
ADCk − ADC V k −V ffi CCC ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 X XN 2 N ADC −ADC V −V k k k¼1 k¼1 k¼1
ð3Þ
where ADC and V were ADC and absolute volumetric flow rate at each volume loading phase, k was the time index, and N was the total number of images in the time series. We directly measured the internal pressure changes in the gateway of the hemodialyzer by a pressure indicator (Samba 201, Monte System Corporation, Tokyo, Japan). The sampling rate was 100 Hz. The absolute pressure gradient was calculated using a temporal differentiation of the internal pressure during the volume loading cycle. A multiple linear regression analysis was used to determine the relationship between the change in ADC and the absolute pressure gradient during the volume loading cycle. Finally, the compliance in the phantom was directly calculated by a ratio of the maximum change in volume loading and internal pressure using Eq. (4). Compliance ¼
VLp‐p pp‐p
ð4Þ
where VLp-p and Pp-p were the volume loading and the maximum change in internal pressure, respectively. We compared the changed ADC, the volume loading change, the absolute volumetric flow rate, the internal pressure change, the absolute pressure gradient, delta-ADC, and compliance at each volume loading. 2.1. The imaging condition The scanning parameters of ECG-triggered single-shot EPI were set at an echo time of 81 ms, field of view of 256 mm, b factors of 0 and 1000 s/mm 2, imaging matrix of 64 × 62, volume loading phase of 30, flip angle of 90°, and acceleration factor of 2, half-Fourier factor
H. Kan et al. / Magnetic Resonance Imaging 32 (2014) 1284–1289
1287
of 0.61, and the average of two signals was taken. The scanning parameters of a retrospective ECG-gated phase contrast cine MRI were set at an echo time of 5.6 ms, field of view of 140 mm, imaging matrix of 128 × 128, volume loading phase of 32, flip angle of 20°, and velocity encoding of 20 cm/s. 3. Results The results of the changes in ADC during high-volume and lowvolume loading cycles are shown in Fig. 2. The ADC varied in accordance with the volume loading change. The applied high-volume and low-volume loadings were 0.187 and 0.110 mL, respectively (Fig. 3). Both volume loadings were periodically applied to the hemodialyzer phantom. The absolute volumetric flow rates changes in each volume loading are shown in Fig. 4. The ADC and absolute volumetric flow rates waveforms were the analogous fashions (Figs. 2 and 4). The result of CCC analysis between these showed that the ADC changes were in exact agreement with the absolute volumetric flow rates during the volume loading cycles (P = 0.961 and 0.824 in highvolume and low-volume loadings, respectively). The internal pressure change during high-volume loading was larger than that during low-volume loading (Fig. 5). The maximum change in internal pressures during the volume loading cycle was 3.84 and 2.45 mmHg in high- and low-volume loadings, respectively. The internal pressure change had a close resemblance to the volume loading change during the volume loading cycle. There were significant correlations between ADC and absolute pressure gradient at each volume loading phase (Fig. 6) (R 2 = 0.804 and 0.738 in high-volume and low-volume loadings, respectively). The delta-ADCs were 0.400 ± 0.193 and 0.273 ± 0.137 × 10 −3 mm2/s in high-volume and low-volume loadings, respectively (Fig. 7). In contrast, compliances were 0.049 and 0.045 mL/mm Hg in high-volume and low-volume loadings, respectively (Fig. 8).
Fig. 3. Volume loading changes during high-volume and low-volume loading cycles. The applied high-volume loading was 1.70 times higher than the low-volume loading.
In the current study, we investigated the mechanism underlying changes in ADC during the cardiac cycle in intracranial tissue using a hemodialyzer phantom. As indicated in Figs. 2 and 3, the changes in ADC correlated with the changes in volume loading. This result was consistent with the report of Nakamura et al. [16]. The temporal ADC
waveform was disaccorded with the volume loading waveform and therefore the relationship between the ADC waveform and the absolute volumetric flow rate waveform was analyzed using CCC analysis. The important finding was that the changes in ADC significantly synchronized with absolute volumetric flow rate changes. This observation can be explained by the expansion and the compression of HFs, due to the cyclic volumetric flow. The force applied to the wall of HF transferred the syrup solution into ES. As a result, the syrup solution in ES was displaced beyond the intrinsic diffusional mobility of the syrup solution by the force transferred to ES. In other words, the applied force resulted in fluctuations in the syrup solution. When arterial blood flow enters the cranium, arterial vessels are expanded in humans. The expanded vessels result in brain arterial pulsation. The arterial pulsation force is transferred not only to the cerebral ventricle and veins but also to the brain parenchyma during the short phase lag between arterial pulsation and CSF flow [10,21]. The water molecules in the brain parenchyma fluctuate according to the force during the cardiac cycle. Consequently, the maximum change in ADC in the brain parenchyma is also dependent upon the degree of force applied by arterial pulsation. Moreover, the internal pressure in the high-volume loadings was higher than that in the low-volume loadings. In
Fig. 2. Changes in ADC during high-volume and low-volume loading cycles. The change in ADC increase with the volume loading. The maximum change in ADC in high-volume loading was higher than that in low-volume loading. The minimum ADCs were almost the same values in each volume loading.
Fig. 4. Absolute volumetric flow rate during high-volume and low-volume loading cycles. The peak absolute volumetric flow rate in high-volume loading was 1.83 times higher than that in low-volume loading.
4. Discussion
1288
H. Kan et al. / Magnetic Resonance Imaging 32 (2014) 1284–1289
Fig. 5. Change in internal pressure during high-volume and low-volume loading cycles.
Fig. 7. Comparison of delta-ADC in high-volume and low-volume loading cycles. The delta-ADC in high-volume loading was 1.46 times higher than that in low-volume loading. Error bars are standard deviation.
addition, there was a significant correlation between ADC and absolute pressure gradient at each time point for both high-volume and low-volume loadings. These findings were the result of the syrup solution fluctuations during the volume loading cycle. With the increase or decrease in the force applied to the HF wall during the volume loading cycle, the pressure gradient in the phantom also rose or fell, respectively. The force transferred to the syrup solution in ES and the fluctuations resulted in changes in pressure gradient. Although the compliances in the phantom directly calculated from the volume loading cycles were almost the same for both volume loadings, delta-ADC in high-volume loading cycles was higher. Ohno et al. [17] reported that idiopathic normal pressure hydrocephalus had a higher delta-ADC, and this was attributed to a decrease in ICC. The delta-ADC result was reliant not only on ICC but also upon the hemodynamic states in regional intracranial tissue. These relationships implied that delta-ADC represented the transfer characteristics of arterial pulsatile force in regional intracranial tissue. Earlier studies suggested that low ICC was affected by the venous vessel compliance, venous flow, and CSF flow [22–24]. In this study, however, it is still unclear as to whether a direct relationship exists between the fluctuation, venous vessel compliance, and both flows during the cardiac cycle. Future studies should be carried out using several compliances of the phantom, because of these changing factors. Our study had some limitations. First, the demonstrated experimental data used two different volume loadings. The
relationship between fluctuations and volumetric flow rate might not be a linear system. Second, several ADCs of syrup solution are needed. Delta-ADC may be affected by the intrinsic ADC. Transfer characteristics were changed by baseline ADC, i.e., changed diffusional mobility. Despite these limitations, it is still highly possible that the delta-ADC analysis provides a new biomarker for estimating regional intracranial tissues. Furthermore, the advantage of delta-ADC analysis was to acquire information about regional tissue characteristics in comparison with ICC analysis. Our results can contribute to the ongoing development of new models capable of explaining the mechanisms of water fluctuation during the cardiac cycle.
Fig. 6. Relationship between ADC and absolute pressure gradients at each time point. The ADC increases with the absolute pressure gradient.
Fig. 8. Comparison of compliance in high-volume and low-volume loading cycles. The compliances were the similar values in high- and low-volume loadings.
5. Conclusions This study has shown that the delta-ADC analysis is a potentially effective method for estimating the transfer characteristics of arterial pulsatile force in regional intracranial tissue. Our original phantom makes it possible to analyze the changes in ADC during the cardiac cycle and verify the mechanisms of water fluctuation in intracranial tissues. Acknowledgments This work was supported by JSPS KAKENHI (24601009).
H. Kan et al. / Magnetic Resonance Imaging 32 (2014) 1284–1289
References [1] Muthupillai R, Lomas DJ, Rossman PJ, et al. Magnetic resonance elastography by direct visualization of propagating acoustic strain waves. Science 1995;265: 1854–7. [2] Kruse SA, Rose GH, Glaser KJ, et al. Magnetic resonance elastography of the brain. Neuroimage 2008;39(1):231–7. [3] Sack I, Beierbach B, Hamhaber U, et al. Non-invasive measurement of brain viscoelasticity using magnetic resonance elastography. NMR Biomed 2008;21: 265–71. [4] Latta P, Gruwel ML, Debergue P, et al. Convertible pneumatic actuator for magnetic resonance elastography of the brain. Magn Reson Imaging 2011;29: 147–52. [5] Alperin NJ, Lee SH, Loth F, et al. MR-intracranial pressure (ICP): a method to measure intracranial elastance and pressure noninvasively by means of MR imaging: baboon and human study. Radiology 2000;217:877–85. [6] Tain RW, Alperin N. Noninvasive intracranial compliance from MRI-based measurements of transcranial blood and CSF flows: indirect versus direct approach. IEEE Trans Biomed Eng 2009;56:544–51. [7] Baledent O, Fin L, Khuoy L, et al. Brain hydrodynamics study by phase-contrast magnetic resonance imaging and transcranial color Doppler. J Magn Reson Imaging 2006;24:995–1004. [8] Wagshul ME, Eide PK, Madsen JR. The pulsating brain: a review of experimental and clinical studies of intracranial pulsatility. Fluids Barriers CNS 2011;8:5. [9] Enzmann DR, Pelc NJ. Normal flow patterns of intracranial and spinal cerebrospinal fluid defined with phase-contrast cine MR imaging. Radiology 1991;178:467–74. [10] Greitz D, Wirestam R, Franck A, et al. Pulsatile brain movement and associated hydrodynamics studied by magnetic resonance phase imaging. Neuroradiology 1992;34:370–80. [11] Baledent O, Henry-Feuguea MC, Idy-Peretti I. Cerebrospinal fluid dynamics and relation with blood flow. A magnetic resonance study with semiautomated cerebrospinal fluid segmentation. Invest Radiol 2001;36:368–77.
1289
[12] Baledent O, Goundry-Jouet C, Meyer M, et al. Relationship between cerebrospinal fluid and blood dynamics in healthy volunteers and patients with communicating hydrocephalus. Invest Radiol 2004;39:45–55. [13] Wahlin A, Ambarki K, Hauksson J, et al. Phase contrast MRI quantification of pulsatile volumes of brain arteries, veins, and cerebrospinal fluids compartments: repeatability and physiological interactions. J Magn Reson Imaging 2012;35:1055–62. [14] Alpein N, Sivaramakrishnan A, Lichtor T. Magnetic resonance imaging-based measurements of cerebrospinal fluid and blood flow as indicators of intracranial compliance in patients with Chiari malformation. J Neurosurg 2005;103:46–52. [15] Miyati T, Mase M, Kasai H, et al. Noninvasive MRI assessment of intracranial compliance in idiopathic normal pressure hydrocephalus. J Magn Reson Imaging 2007;26:274–8. [16] Nakamura T, Miyati T, Kasai H, et al. Bulk motion-independent analyses of water diffusion changes in the brain during the cardiac cycle. Radiol Phys Technol 2009;2:133–7. [17] Ohno N, Miyati T, Mase M, et al. Idiopathic normal-pressure hydrocephalus: temporal changes in ADC during cardiac cycle. Radiology 2011;261:560–5. [18] Skare S, Andersson JL. On the effects of gating in diffusion imaging of the brain using single shot EPI. Magn Reson Imaging 2001;19:1125–8. [19] Norris DG. Implications of bulk motion for diffusion-weighted imaging experiments: effects, mechanisms, and solutions. J Magn Reson Imaging 2001;13:486–95. [20] Miyati T, Mase M, Banno T, et al. Frequency analyses of CSF flow on cine MRI in normal pressure hydrocephalus. Eur Radiol 2003;13:1019–24. [21] Ambarki K, Baledent O, Kongolo G, et al. A new lumped-parameter model of cerebrospinal hydrodynamics during the cardiac cycle in healthy volunteers. IEEE Trans Biomed Eng 2007;54:483–91. [22] Bateman GA. The pathophysiology of idiopathic normal pressure hydrocephalus: cerebral ischemia or altered venous hemodynamics? AJNR Am J Neuroradiol 2008;29:198–203. [23] Beggs CB. Venous hemodynamics in neurological disorders: an analytical review with hydrodynamic analysis. BMC Med 2013;11:142. [24] Williams H. A unifying hypothesis for hydrocephalus, Chiari malformation, syringomyelia, anencephaly and spina bifida. Cerebrospinal Fluid Res 2008;5:7.
Contents lists available at ScienceDirect
Magnetic Resonance Imaging journal homepage: www.mrijournal.com
Transfer characteristics of arterial pulsatile force in regional intracranial tissue using dynamic diffusion MRI: A phantom study Hirohito Kan a, b, 1, Tosiaki Miyati b,⁎, Harumasa Kasai a, 1, Nobuyuki Arai a, 1, Naoki Ohno b, 2, Mitsuhito Mase c, 1, Yuta Shibamoto a, 1 a b c
Department of Radiology, Nagoya City University Hospital, 1-Kawasumi, Mizuho-cho Mizuho-ku, Nagoya City, Aichi, 4678602, Japan Division of Health Sciences, Graduate School of Medical Science, Kanazawa University, 5-11-80 Kodatsuno, Kanazawa, Ishikawa, 9200942, Japan Department of Neurosurgery, Nagoya City University Hospital, 1-Kawasumi, Mizuho-cho Mizuho-ku, Nagoya City, Aichi, 4678602, Japan
a r t i c l e
i n f o
Article history: Received 19 March 2014 Revised 29 July 2014 Accepted 21 August 2014 Keywords: Magnetic resonance imaging Diffusion weighted imaging Apparent diffusion coefficient Water molecular fluctuation
a b s t r a c t Introduction: To clarify the mechanism underlying apparent diffusion coefficient (ADC) changes in regional intracranial tissue during the cardiac cycle, we investigated relationships among ADC changes, volume loading, and intracranial pressure using a hemodialyzer phantom in magnetic resonance imaging (MRI). Materials and Methods: The hemodialyzer phantom consisted of hollow fibers (HF), the external space of HFs (ES), and a gateway of dialysis solution, filled with syrup solution and air. The high-volume and lowvolume loadings were periodically applied to HFs by a to-and-fro flow pump, and syrup solution was permitted to enter or leave HFs during the volume loading cycles. ADC maps at each volume loading phase were obtained using ECG-triggered single-shot diffusion echo-planar imaging. Dynamic phase contrast MRI was also used to measure volume loading to the phantom. We compared the ADC changes, volume loading, and intracranial pressure in the phantom during the cardiac cycle. Results: ADC changes synchronized significantly with absolute volumetric flow rate change. The maximum ADC change was higher in high-volume loading cycles than in low-volume loading cycles. Results showed that the water molecules in ES fluctuated according to the force transferred from HF to ES. ADC changes were dependent upon the volumetric flow rate during the cardiac cycle. Conclusions: Our original phantom allowed us to clarify the mechanism underlying water fluctuations in intracranial tissues. Measurement of maximum changes in ADC is an effective method to define the transfer characteristics of the arterial pulsatile force in regional intracranial tissue. © 2014 Elsevier Inc. All rights reserved.
1. Introduction In the past years, several studies have attempted to use quantitative imaging techniques to evaluate the biomechanical properties of the brain using magnetic resonance imaging (MRI). MRI-based measurement of the biomechanical properties of the brain may be useful in characterizing and diagnosing of intracranial diseases not easily evaluated using conventional imaging contrasts. Magnetic resonance elastography (MRE), a technique that mea-
⁎ Corresponding author. Tel.: +81 76 265 2500. E-mail addresses: [email protected] (H. Kan), [email protected] (T. Miyati), [email protected] (H. Kasai), [email protected] (N. Arai), [email protected] (N. Ohno), [email protected] (M. Mase), [email protected] (Y. Shibamoto). 1 Tel.: +81 52 851 5511. 2 Tel.: +81 76 265 2500. http://dx.doi.org/10.1016/j.mri.2014.08.026 0730-725X/© 2014 Elsevier Inc. All rights reserved.
sures biomechanical properties by directly applying shear waves to the brain, has been used in the past [1–4]. However, MRE was limited because it requires specialized equipment to generate the shear waves. Another technique was to measure the intracranial compliance (ICC) calculated from blood flow and cerebral spinal fluid (CSF) flow, and movement of the spinal cord during the cardiac cycle using a retrospectively gated phase-contrast cine MRI. Alperin et al. [5] developed an ICC analysis for evaluating the mechanical coupling of brain and CSF oscillations to the driving vascular pulsations. Arterial flow into the cranium induces venous flow, CSF flow, and spinal cord movement during the cardiac cycle. The relationship between these measurements is well established [6–13]. Next, the intracranial volume loading during the cardiac cycle was calculated from the differences between the inflow and outflow volumes of the cranium at each time point in the cardiac cycle [7]. ICC represented the ability of the intracranial space to accommodate an increase in volume without a large increase in intracranial pressure [5,6]. ICC analysis has been used in patients
H. Kan et al. / Magnetic Resonance Imaging 32 (2014) 1284–1289 Fig. 1. (a) Phantom for simulating change ADC changes during the cardiac cycle following volume loading. (b) Schematic illustration of the hemodialyzer phantom. Volume loading was applied to the hemodialyzer, and syrup was permitted to enter or leave HFs through the tube only. (c) Apparent diffusion coefficient image without volume loading. (d) Microscopic image of the imaging plane.
1285
1286
H. Kan et al. / Magnetic Resonance Imaging 32 (2014) 1284–1289
with abnormal intracranial conditions such as Chiari malformation and idiopathic normal pressure hydrocephalus [14,15]. This analysis has been helpful in the diagnosis and the subsequent management of these diseases. However, this method has some limitations; for example, the estimated ICC represented information about the whole brain, not a localized intracranial tissue. In a previous work, we found that the apparent diffusion coefficient (ADC) in the white matter changed during the cardiac cycle, despite minimizing the brain bulk motion effect [16]. The regional changed ADC resulted in volume loading of the cranium during the cardiac cycle. We also reported that the variation in ADC during the cardiac cycle was closely related to the regional biomechanical properties and the measurement of the maximum change in ADC during the cardiac cycle (delta-ADC) may be of help in the diagnosis of idiopathic normal pressure hydrocephalus [17]. However, the more detailed mechanisms and relationships among intracranial pressure, volumetric flow and ADC changes during the cardiac cycle remain unclear. The present study aimed to clarify the mechanisms underlying the ADC changes during the cardiac cycle in intracranial tissue using a hemodialyzer phantom. 2. Materials and methods Fig. 1a illustrates the experimental setup of the phantom. The phantom was designed to simulate water diffusion changes in intracranial tissues by arterial blood volume loading. A schematic drawing of the hemodialyzer is shown in Fig. 1b. The hemodialyzer consists of hollow fibers (HF), external spaces of HF (ES), and the gateway of dialysis solution (Hollow Fiber Dialyzer KF-201, Kawasumi Laboratories Inc., Tokyo, Japan). These components represented the blood capillaries, intracranial tissues, and the intracranial capacity, respectively. The intracranial capacity plays a role in dampening the volume loading caused by pulsatile arterial blood flow. HFs and ES were filled with syrup solution adjusted to 0.6 x10 -3 mm2/s ADC in x-y plane of the hemodialyzer phantom. (Fig. 1c). The syrup solution was a viscous liquid consisting of a solution of sugar in water. The pore size of the HF membrane was sufficiently small and the syrup solution could not pass easily between HF and ES. Fig. 1d shows a microscopic image of the cross section of the hemodialyzer. The gateway of the hemodialyzer was filled with air to dampen the cyclic volume loading. Volume loading was periodically applied to the hemodialyzer via a to-and-fro flow hand-made pump. The flow pump generated a cyclic volumetric flow similar to a sine wave. The hemodialyzer was connected to the pump by a polyvinyl chloride tube filled with the syrup solution (Green bubble tube 2302, Covidien Japan, Shizuoka, Japan). The volume loading applied to the hemodialyzer was permitted to enter or leave HF only through the tube. All measurements were carried out using the same volume loading cycle, i.e., 1 Hz. High-volume and low-volume loadings were applied to the phantom. The pump generated a trigger signal, which was used to synchronize with the MR data acquisition. To minimize a susceptibility effect in the MR data acquisition for the volumetric flow measurement, the tube was penetrated through a plastic case filled with water. On a 1.5-T MRI (Gyroscan Intera, Philips Medical Systems International, Best, the Netherlands), the long axis of the phantom was set in a direction parallel to the magnetic field, i.e., the flow direction in HF corresponded to the z-axis. DWIs for each b value were obtained using ECG-triggered multi-phase single-shot diffusion echo-planar imaging (EPI), which is largely insensitive to motion. However, single-shot EPI alone cannot completely eliminate the bulk motion effect of the brain parenchyma because of the long data-sampling window required for single-shot EPI, and artifacts may remain visible [18]. Therefore, single-shot EPI was combined with parallel imaging, half-scan, and rectangular field-of-view (FOV) techniques to minimize the bulk motion effect [19]. The trigger delays from
the R-peak for DWIs were set at regular intervals (30 ms), and we acquired 30 phases. Then, ADC was calculated by Eq (1); ADC ¼
ln S1 − ln S2 b2 −b1
ð1Þ
where S1 and S2 were signal intensities of DWIs at 0 and 1000 b-factors, respectively, and b1 and b2 were 0 and 1000 b-factors, respectively. The ADC map in a direction perpendicular to HF, i.e., the x–y plane, was calculated from DWIs at each volume loading phase to minimize the flow effect in HFs. Then, the ADC change during the volume loading cycle (delta-ADC) was calculated as maximum ADC minus minimum ADC of all volume loading phase images on a pixel-by-pixel basis. To measure the volume loading change, a retrospective ECG-gated phase contrast cine MRI was then performed at a vertical slice plane against the long axis of the tube. An ROI was set in the tube on velocity-mapped images in each volume loading phase to measure the flow velocities into the phantom. The baseline offset of the flow velocity, due to eddy currents, was then corrected using a subtraction process [20]. We multiplied the cross-sectional area of the tube by the flow velocity to obtain the volumetric flow rate. The volume loading change was derived by the temporal integration of the volumetric flow rate as shown in Eq. (2). Z VLðt Þ ¼
V ðt Þ dt
ð2Þ
where VL(t), V(t) and t are the volume loading change, volumetric flow rate through the tube, and time index, respectively. The relationship between the waveforms of ADC and absolute velocity was confirmed using a cross-calibration coefficient (CCC) in Eq. (3). XN
ADCk − ADC V k −V ffi CCC ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 X XN 2 N ADC −ADC V −V k k k¼1 k¼1 k¼1
ð3Þ
where ADC and V were ADC and absolute volumetric flow rate at each volume loading phase, k was the time index, and N was the total number of images in the time series. We directly measured the internal pressure changes in the gateway of the hemodialyzer by a pressure indicator (Samba 201, Monte System Corporation, Tokyo, Japan). The sampling rate was 100 Hz. The absolute pressure gradient was calculated using a temporal differentiation of the internal pressure during the volume loading cycle. A multiple linear regression analysis was used to determine the relationship between the change in ADC and the absolute pressure gradient during the volume loading cycle. Finally, the compliance in the phantom was directly calculated by a ratio of the maximum change in volume loading and internal pressure using Eq. (4). Compliance ¼
VLp‐p pp‐p
ð4Þ
where VLp-p and Pp-p were the volume loading and the maximum change in internal pressure, respectively. We compared the changed ADC, the volume loading change, the absolute volumetric flow rate, the internal pressure change, the absolute pressure gradient, delta-ADC, and compliance at each volume loading. 2.1. The imaging condition The scanning parameters of ECG-triggered single-shot EPI were set at an echo time of 81 ms, field of view of 256 mm, b factors of 0 and 1000 s/mm 2, imaging matrix of 64 × 62, volume loading phase of 30, flip angle of 90°, and acceleration factor of 2, half-Fourier factor
H. Kan et al. / Magnetic Resonance Imaging 32 (2014) 1284–1289
1287
of 0.61, and the average of two signals was taken. The scanning parameters of a retrospective ECG-gated phase contrast cine MRI were set at an echo time of 5.6 ms, field of view of 140 mm, imaging matrix of 128 × 128, volume loading phase of 32, flip angle of 20°, and velocity encoding of 20 cm/s. 3. Results The results of the changes in ADC during high-volume and lowvolume loading cycles are shown in Fig. 2. The ADC varied in accordance with the volume loading change. The applied high-volume and low-volume loadings were 0.187 and 0.110 mL, respectively (Fig. 3). Both volume loadings were periodically applied to the hemodialyzer phantom. The absolute volumetric flow rates changes in each volume loading are shown in Fig. 4. The ADC and absolute volumetric flow rates waveforms were the analogous fashions (Figs. 2 and 4). The result of CCC analysis between these showed that the ADC changes were in exact agreement with the absolute volumetric flow rates during the volume loading cycles (P = 0.961 and 0.824 in highvolume and low-volume loadings, respectively). The internal pressure change during high-volume loading was larger than that during low-volume loading (Fig. 5). The maximum change in internal pressures during the volume loading cycle was 3.84 and 2.45 mmHg in high- and low-volume loadings, respectively. The internal pressure change had a close resemblance to the volume loading change during the volume loading cycle. There were significant correlations between ADC and absolute pressure gradient at each volume loading phase (Fig. 6) (R 2 = 0.804 and 0.738 in high-volume and low-volume loadings, respectively). The delta-ADCs were 0.400 ± 0.193 and 0.273 ± 0.137 × 10 −3 mm2/s in high-volume and low-volume loadings, respectively (Fig. 7). In contrast, compliances were 0.049 and 0.045 mL/mm Hg in high-volume and low-volume loadings, respectively (Fig. 8).
Fig. 3. Volume loading changes during high-volume and low-volume loading cycles. The applied high-volume loading was 1.70 times higher than the low-volume loading.
In the current study, we investigated the mechanism underlying changes in ADC during the cardiac cycle in intracranial tissue using a hemodialyzer phantom. As indicated in Figs. 2 and 3, the changes in ADC correlated with the changes in volume loading. This result was consistent with the report of Nakamura et al. [16]. The temporal ADC
waveform was disaccorded with the volume loading waveform and therefore the relationship between the ADC waveform and the absolute volumetric flow rate waveform was analyzed using CCC analysis. The important finding was that the changes in ADC significantly synchronized with absolute volumetric flow rate changes. This observation can be explained by the expansion and the compression of HFs, due to the cyclic volumetric flow. The force applied to the wall of HF transferred the syrup solution into ES. As a result, the syrup solution in ES was displaced beyond the intrinsic diffusional mobility of the syrup solution by the force transferred to ES. In other words, the applied force resulted in fluctuations in the syrup solution. When arterial blood flow enters the cranium, arterial vessels are expanded in humans. The expanded vessels result in brain arterial pulsation. The arterial pulsation force is transferred not only to the cerebral ventricle and veins but also to the brain parenchyma during the short phase lag between arterial pulsation and CSF flow [10,21]. The water molecules in the brain parenchyma fluctuate according to the force during the cardiac cycle. Consequently, the maximum change in ADC in the brain parenchyma is also dependent upon the degree of force applied by arterial pulsation. Moreover, the internal pressure in the high-volume loadings was higher than that in the low-volume loadings. In
Fig. 2. Changes in ADC during high-volume and low-volume loading cycles. The change in ADC increase with the volume loading. The maximum change in ADC in high-volume loading was higher than that in low-volume loading. The minimum ADCs were almost the same values in each volume loading.
Fig. 4. Absolute volumetric flow rate during high-volume and low-volume loading cycles. The peak absolute volumetric flow rate in high-volume loading was 1.83 times higher than that in low-volume loading.
4. Discussion
1288
H. Kan et al. / Magnetic Resonance Imaging 32 (2014) 1284–1289
Fig. 5. Change in internal pressure during high-volume and low-volume loading cycles.
Fig. 7. Comparison of delta-ADC in high-volume and low-volume loading cycles. The delta-ADC in high-volume loading was 1.46 times higher than that in low-volume loading. Error bars are standard deviation.
addition, there was a significant correlation between ADC and absolute pressure gradient at each time point for both high-volume and low-volume loadings. These findings were the result of the syrup solution fluctuations during the volume loading cycle. With the increase or decrease in the force applied to the HF wall during the volume loading cycle, the pressure gradient in the phantom also rose or fell, respectively. The force transferred to the syrup solution in ES and the fluctuations resulted in changes in pressure gradient. Although the compliances in the phantom directly calculated from the volume loading cycles were almost the same for both volume loadings, delta-ADC in high-volume loading cycles was higher. Ohno et al. [17] reported that idiopathic normal pressure hydrocephalus had a higher delta-ADC, and this was attributed to a decrease in ICC. The delta-ADC result was reliant not only on ICC but also upon the hemodynamic states in regional intracranial tissue. These relationships implied that delta-ADC represented the transfer characteristics of arterial pulsatile force in regional intracranial tissue. Earlier studies suggested that low ICC was affected by the venous vessel compliance, venous flow, and CSF flow [22–24]. In this study, however, it is still unclear as to whether a direct relationship exists between the fluctuation, venous vessel compliance, and both flows during the cardiac cycle. Future studies should be carried out using several compliances of the phantom, because of these changing factors. Our study had some limitations. First, the demonstrated experimental data used two different volume loadings. The
relationship between fluctuations and volumetric flow rate might not be a linear system. Second, several ADCs of syrup solution are needed. Delta-ADC may be affected by the intrinsic ADC. Transfer characteristics were changed by baseline ADC, i.e., changed diffusional mobility. Despite these limitations, it is still highly possible that the delta-ADC analysis provides a new biomarker for estimating regional intracranial tissues. Furthermore, the advantage of delta-ADC analysis was to acquire information about regional tissue characteristics in comparison with ICC analysis. Our results can contribute to the ongoing development of new models capable of explaining the mechanisms of water fluctuation during the cardiac cycle.
Fig. 6. Relationship between ADC and absolute pressure gradients at each time point. The ADC increases with the absolute pressure gradient.
Fig. 8. Comparison of compliance in high-volume and low-volume loading cycles. The compliances were the similar values in high- and low-volume loadings.
5. Conclusions This study has shown that the delta-ADC analysis is a potentially effective method for estimating the transfer characteristics of arterial pulsatile force in regional intracranial tissue. Our original phantom makes it possible to analyze the changes in ADC during the cardiac cycle and verify the mechanisms of water fluctuation in intracranial tissues. Acknowledgments This work was supported by JSPS KAKENHI (24601009).
H. Kan et al. / Magnetic Resonance Imaging 32 (2014) 1284–1289
References [1] Muthupillai R, Lomas DJ, Rossman PJ, et al. Magnetic resonance elastography by direct visualization of propagating acoustic strain waves. Science 1995;265: 1854–7. [2] Kruse SA, Rose GH, Glaser KJ, et al. Magnetic resonance elastography of the brain. Neuroimage 2008;39(1):231–7. [3] Sack I, Beierbach B, Hamhaber U, et al. Non-invasive measurement of brain viscoelasticity using magnetic resonance elastography. NMR Biomed 2008;21: 265–71. [4] Latta P, Gruwel ML, Debergue P, et al. Convertible pneumatic actuator for magnetic resonance elastography of the brain. Magn Reson Imaging 2011;29: 147–52. [5] Alperin NJ, Lee SH, Loth F, et al. MR-intracranial pressure (ICP): a method to measure intracranial elastance and pressure noninvasively by means of MR imaging: baboon and human study. Radiology 2000;217:877–85. [6] Tain RW, Alperin N. Noninvasive intracranial compliance from MRI-based measurements of transcranial blood and CSF flows: indirect versus direct approach. IEEE Trans Biomed Eng 2009;56:544–51. [7] Baledent O, Fin L, Khuoy L, et al. Brain hydrodynamics study by phase-contrast magnetic resonance imaging and transcranial color Doppler. J Magn Reson Imaging 2006;24:995–1004. [8] Wagshul ME, Eide PK, Madsen JR. The pulsating brain: a review of experimental and clinical studies of intracranial pulsatility. Fluids Barriers CNS 2011;8:5. [9] Enzmann DR, Pelc NJ. Normal flow patterns of intracranial and spinal cerebrospinal fluid defined with phase-contrast cine MR imaging. Radiology 1991;178:467–74. [10] Greitz D, Wirestam R, Franck A, et al. Pulsatile brain movement and associated hydrodynamics studied by magnetic resonance phase imaging. Neuroradiology 1992;34:370–80. [11] Baledent O, Henry-Feuguea MC, Idy-Peretti I. Cerebrospinal fluid dynamics and relation with blood flow. A magnetic resonance study with semiautomated cerebrospinal fluid segmentation. Invest Radiol 2001;36:368–77.
1289
[12] Baledent O, Goundry-Jouet C, Meyer M, et al. Relationship between cerebrospinal fluid and blood dynamics in healthy volunteers and patients with communicating hydrocephalus. Invest Radiol 2004;39:45–55. [13] Wahlin A, Ambarki K, Hauksson J, et al. Phase contrast MRI quantification of pulsatile volumes of brain arteries, veins, and cerebrospinal fluids compartments: repeatability and physiological interactions. J Magn Reson Imaging 2012;35:1055–62. [14] Alpein N, Sivaramakrishnan A, Lichtor T. Magnetic resonance imaging-based measurements of cerebrospinal fluid and blood flow as indicators of intracranial compliance in patients with Chiari malformation. J Neurosurg 2005;103:46–52. [15] Miyati T, Mase M, Kasai H, et al. Noninvasive MRI assessment of intracranial compliance in idiopathic normal pressure hydrocephalus. J Magn Reson Imaging 2007;26:274–8. [16] Nakamura T, Miyati T, Kasai H, et al. Bulk motion-independent analyses of water diffusion changes in the brain during the cardiac cycle. Radiol Phys Technol 2009;2:133–7. [17] Ohno N, Miyati T, Mase M, et al. Idiopathic normal-pressure hydrocephalus: temporal changes in ADC during cardiac cycle. Radiology 2011;261:560–5. [18] Skare S, Andersson JL. On the effects of gating in diffusion imaging of the brain using single shot EPI. Magn Reson Imaging 2001;19:1125–8. [19] Norris DG. Implications of bulk motion for diffusion-weighted imaging experiments: effects, mechanisms, and solutions. J Magn Reson Imaging 2001;13:486–95. [20] Miyati T, Mase M, Banno T, et al. Frequency analyses of CSF flow on cine MRI in normal pressure hydrocephalus. Eur Radiol 2003;13:1019–24. [21] Ambarki K, Baledent O, Kongolo G, et al. A new lumped-parameter model of cerebrospinal hydrodynamics during the cardiac cycle in healthy volunteers. IEEE Trans Biomed Eng 2007;54:483–91. [22] Bateman GA. The pathophysiology of idiopathic normal pressure hydrocephalus: cerebral ischemia or altered venous hemodynamics? AJNR Am J Neuroradiol 2008;29:198–203. [23] Beggs CB. Venous hemodynamics in neurological disorders: an analytical review with hydrodynamic analysis. BMC Med 2013;11:142. [24] Williams H. A unifying hypothesis for hydrocephalus, Chiari malformation, syringomyelia, anencephaly and spina bifida. Cerebrospinal Fluid Res 2008;5:7.
