| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Departments of Radiology (O.R., M.Y., M.A.D., P.J.R., J.L.F., R.L.E.) and Anatomic Pathology (L.J.B.), Mayo Clinic, 200 First Street SW, Rochester, MN 55905. Received April 12, 2005; revision requested June 13; revision received July 15; accepted August 11; final version accepted October 3. Supported by NIH grants EB001981, CA91959, and CA95683. Address correspondence to R.L.E. (e-mail: ehman.richard{at}mayo.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Materials and Methods: This HIPAA-compliant study was approved by an institutional review board, and informed consent was obtained from each subject. First, to determine the feasibility of applying shear waves to the liver, a pneumatic acoustic wave generator was developed and tested by using a tissue-simulating gel phantom with ribs on one side and without ribs on the other. The effect of interposed ribs on stiffness measurements was tested. Then, liver stiffness was measured with MR elastography in 12 healthy volunteers (eight men, four women; mean age, 26.7 years; age range, 19–39 years) by using the subcostal approach and the transcostal approach and in 12 patients with chronic liver disease (six men, six women; mean age, 50.5 years; age range, 36–60 years) by using the transcostal approach. Various statistical analyses were performed to assess all measurements.
Results: Ex vivo, interposed ribs reduced shear wave amplitude but did not hinder stiffness measurements. In volunteers, the transcostal approach surprisingly yielded better shear waves in the liver than did the subcostal approach. The mean liver shear stiffness was significantly lower in volunteers (mean, 2.0 kPa ± 0.3 [standard deviation]) than it was in patients with liver fibrosis (mean, 5.6 kPa ± 5.0; median, 3.7 kPa; range, 2.7–19.2 kPa; P < .001).
Conclusion: MR elastography of the liver is feasible and shows promise as a quantitative method for noninvasive assessment of liver fibrosis.
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
The degree of fibrosis is currently assessed with liver biopsy, and several semiquantitative pathologic scoring systems have been proposed (5–7). Liver biopsy is an invasive procedure that carries a risk of serious complications, including a procedure-related mortality rate of one in 10 000–12 000 (8,9).
Several factors point out the need for a noninvasive method to evaluate liver fibrosis. First, there is no way to predict which patients with chronic hepatitis C will develop substantial fibrosis, and performing repetitive biopsies in all patients is impossible. A noninvasive method for measuring liver fibrosis would help in deciding which patients require antiviral therapy and in predicting the approximate time to the development of cirrhosis (1). Second, liver fibrosis may be reversible when the cause is treated (10–14), and a noninvasive assessment of treatment response would also be helpful.
Ex vivo and intraoperative studies have shown that liver elasticity correlates with the degree of fibrosis found in biopsy specimens and with the results of liver function tests (15–19). More recently, ultrasonography (US)-based techniques have been proposed for noninvasive assessment of tissue elasticity (20–26), and in vivo measurement of liver fibrosis has proved feasible with some of these methods (23,27,28).
Other investigators have proposed a modified phase-contrast magnetic resonance (MR) imaging sequence to image propagating shear waves in tissue (29,30). This technique, called MR elastography, has been applied to quantitatively assess the viscoelastic properties of the breast, brain, and muscle in humans (31–33).
The purpose of our study was to develop a method for measuring liver stiffness with MR elastography and to prospectively test this technique on healthy volunteers and patients with liver fibrosis.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Phantom Experiments
Experimental setup.—For MR elastographic experiments, we used a cylindrical passive pneumatic driver (a drumlike device designed to apply acoustic vibrations to the surface of the object to be imaged) that was 8 cm in diameter and connected with an acoustic waveguide to an active driver, as described elsewhere (34,35). The passive driver was positioned against a tissue-simulating 15% bovine gel (Bovine skin powder; Sigma-Aldrich, St Louis, Mo) phantom that measured 25 cm in diameter and 9 cm in height (Fig 1). The phantom contained parallel porcine ribs embedded 15 mm below the surface on one side, and the interval between ribs was approximately 10 mm.
|
|
A signal generator and power amplifier unit were used to apply an 80-Hz sinusoidal waveform to a 30-cm-diameter audio driver in a rigid container located adjacent to the imager. Vibrations were conducted to the passive pneumatic driver through a 6-m-long waveguide conduit. MR elastographic wave images were collected with a modified phase-contrast gradient-echo sequence (repetition time, 37.5 msec; echo time, 19.5 msec [37.5/19.5]; flip angle, 30°; field of view, 20 cm; matrix, 256 x 64 pixels; section thickness, 5 mm). One pair of 1.76 G/cm (1.76 · 10–2 T/m) cyclic motion-encoding gradients was synchronized to the passive pneumatic driver by using a trigger provided by the imager. This trigger was varied to obtain eight phase-offset images during a cycle. For each imaging plane, the motion-encoding gradients were applied successively in the two directions orthogonal to the vibration direction of the driver.
By convention, we designated Y as the direction parallel to the vibrational motion of the surface of the driver. X and Z were the two directions perpendicular to Y; X was parallel to the imaging plane, and Z was perpendicular to the imaging plane. Y was the phase-encoding direction, and X was the frequency-encoding direction. Images with cyclic motion sensitized in the X direction will be referred to as Xs images, and those in the Z direction will be referred as to Zs images.
The phantom was positioned so that the shear waves created with the driver could propagate without interference of interposed ribs. Images with planes orthogonal to the vibrating surface (0, 10, 20, and 30 mm from the central axis of the driver) and oblique (10° and 20° from the central axis of the driver) to the driver's surface were acquired, as shown in Figure 1.
The phantom was then inverted so that the side with the ribs was positioned against the driver. Orthogonal and oblique images (with respect to the vibrating surface) were acquired first perpendicular and then parallel to the ribs by using the same protocol as was used for the side without ribs. The phantom was imaged with the center of the driver positioned between two ribs and then positioned against a rib.
Image analysis.—Each MR elastographic acquisition provides a phase-difference image that represents the displacement caused by shear wave propagation in the medium (29,30). The wavelength visible on the phase-difference images can be used to calculate the shear modulus by using the formula
 | (1) |
is the wavelength in meters, f is the driving frequency in hertz, and 
is the mass density in kilograms per cubic meter, which for tissue is approximately the same as that for water (1000 kg/m3). This approach relies on a simple Hookean model and assumes that the medium is nonviscous. Because of this assumption, we designate µ as the shear stiffness instead of the shear modulus (32,36). On each phase-difference image, a region of interest (ROI) of 1462 pixels was placed by two authors (O.R., M.Y.) by consensus in an area of adequate wave amplitude. Its size and distance from the surface of the driver were standardized (center of the ROI positioned 35 mm from the vibrating surface) (Fig 2). Custom software was used to calculate the mean displacement and shear stiffness within the ROI by using a previously described direct inversion algorithm (31).
|
In vivo MR elastographic image acquisition.—The same setup (active driver and 8-cm-diameter passive pneumatic driver) that was used for the phantom experiment was used. The volunteers were imaged in the prone position while lying on the driver. First, the driver was positioned below the rib cage on a ramp that aimed the vibrations toward the liver (subcostal approach). Second, the driver was positioned against the rib cage (transcostal approach). We used the same orthogonal (0–30 mm) and oblique (10°–20°) imaging planes as was used in the phantom experiment. For each plane, Xs and Zs images were acquired (12 different MR elastographic acquisitions for each approach).
The same driving voltage amplitude was applied for all volunteers, and, on the basis of preliminary experiments, the operating frequency was set at 90 Hz. The parameters of the MR elastographic gradient-echo sequence were as follows: 33.3/17.8; flip angle, 30°; field of view, 34 (n = 4) or 38 (n = 8) cm; matrix, 256 x 64; section thickness, 10 mm. One pair of 1.76 G/cm (1.76 · 10–4 T/cm) motion-encoding gradients was synchronized to the passive pneumatic driver. Wave images were obtained at four phase offsets. The acquisition of each offset (10 seconds) was obtained during a breath hold.
Finally, a dual-echo breath-hold fast spoiled gradient-echo sequence was performed in the center orthogonal plane with the following parameters: 150/2.3 (out-of-phase) and 150/4.6 (in-phase); flip angle, 75°; field of view, 34 (n = 4) or 38 (n = 8) cm; matrix, 256 x 160; section thickness, 10 mm; imaging time, 25 seconds.
Image analysis.—First, two independent readers (O.R. with 10 years of experience and M.Y. with 6 months of experience with MR imaging of the liver) evaluated the quality of the shear waves on the phase-difference images by using a five-point subjective scale (0, no visible wave; 1, barely visible waves in the liver; 2, visible waves in the liver; 3, clearly visible waves with fewer than two wavelengths penetrating the liver; and 4, clearly visible waves with more than two wavelengths penetrating the liver). An image quality score was calculated by adding the two readers' scores. For each volunteer, a subcostal quality score was obtained by averaging the image quality scores obtained subcostally, and a transcostal quality score was obtained by averaging the image quality scores obtained transcostally.
Next, the two readers obtained by consensus a line profile along the direction of wave propagation from each wave image. The wavelength was obtained by using a direct peak-to-peak measurement. The shear stiffness was estimated according to Equation (1), and the values measured on the four phase-offsets were averaged.
Finally, an ROI of 350 pixels was positioned by consensus on fast spoiled gradient-echo images in the area where the stiffness was measured. A steatosis index was calculated with the equation (37) SI = (Si – So)/2Si, where SI is the steatosis index, Si is liver signal intensity on in-phase images, and So is the liver signal intensity on out-of-phase images.
Patient Study
Twelve patients (six men, six women) with a history of chronic liver disease and who underwent liver biopsy in the previous 2 years agreed to undergo MR elastographic examination of the liver. Their mean age, weight, and height were 50.5 years (range, 36–60 years), 80.6 kg (range, 59–115 kg), and 169.8 cm (range, 150–180 cm). None of the patients had ascites at the time of the examination.
The liver biopsy had been performed because of hepatitis C (n = 8), autoimmune hepatitis (n = 2), nonalcoholic steatohepatitis (n = 1), or liver cirrhosis due to sarcoidosis (n = 1), on average 8.4 months (range, 2–23 months) before undergoing MR elastographic examination.
Biopsy slides were reviewed by a single pathologist (L.J.B.) with 11 years of experience in liver biopsy and who was blinded to MR elastographic findings. The degree of fibrosis was assessed by using the Batts and Ludwig scoring system (6) as follows: stage 0, no fibrosis; stage 1, portal fibrosis; stage 2, periportal fibrosis (rare portal-portal septa allowed); stage 3, septal fibrosis (fibrous septa with architectural distortion); and stage 4, cirrhosis.
MR elastographic image acquisition and image analysis.—Transcostal images were acquired with the same protocol as was used for the healthy volunteers. Subcostal images were not acquired. Image analysis was done the same way it was for healthy volunteers. The two readers were blinded to biopsy findings.
Statistical Analysis
Quantitative variables are reported as mean ± standard deviation.
Because this was our first attempt at quantifying liver stiffness with MR elastography and because we could make no assumptions on the results, no calculation of the optimal sample size was made for the patient and volunteer groups. For this preliminary study, 12 volunteers and 12 patients seemed reasonable numbers to evaluate the feasibility of the technique and its potential to quantify liver fibrosis.
Statistical analysis was made by using statistical software (StatGraphics Plus 5.1; StatPoint, Herndon, Va).
Mean quality scores and mean liver stiffness measured in the same volunteers were compared by using paired t tests after verifying the absence of substantial departure from the normality of the samples of data (standardized kurtosis and standardized skewness <2 and nonsignificant Shapiro-Wilks test for normality). Two-sided P values were calculated because no assumptions could be made on the direction of the results.
Patient liver stiffness distribution showed significant departure from normality (standardized kurtosis and standardized skewness >2 and significant Shapiro-Wilks test for normality, P < .001). Therefore, mean liver stiffness measured on orthogonal and oblique images in the same patients were compared by using the nonparametric Wilcoxon matched-pairs signed-ranks test. For the same reason, the mean liver stiffness measured in volunteers and patients were compared by using a Mann-Whitney U test. P < .05 was considered to indicate a significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
On images with interposed ribs, the mean displacement was five to six times lower than those without interposed ribs because of a distorted distribution of the waves' amplitudes in space. When the driver's center was between two ribs, however, the evaluation of the phantom stiffness was close to that obtained on images without ribs (Table).
|
|
|
|
|
|
|
|
|
Figure 7 illustrates a wave image from one of the patients, and Figure 8 shows the distribution of the MR elastographic liver shear stiffness measurements in the patients and healthy volunteers. The shear stiffness in the single patient with chronic liver disease but no detectable fibrosis at biopsy (stage 0) was 2.7 kPa. For the 11 other patients, the mean liver shear stiffness was 5.6 kPa ± 5.0 (range, 2.7–19.2 kPa; median, 3.7 kPa).
|
|
|
The mean quality score of MR elastographic wave images was 4.3 ± 1.2. The mean stiffness measured on orthogonal and oblique images showed no significant difference (5.3 kPa ± 4.8 vs 5.7 kPa ± 5.2, respectively; P = .36). Only two patients had substantial steatosis (steatosis index = 11% and 34%). For the others, the steatosis index was less than 5%.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Our phantom experiments indicated that interposed ribs might cause a substantial decrease in the amplitude of transmitted shear waves, but in in vivo studies the transcostal approach yielded adequate wave penetration into the liver and flat wave fronts of good amplitude. The mean image quality score of the transcostal approach was even significantly higher than that with the subcostal approach. It seems likely that, in vivo, the driver vibrates the entire rib cage and that ribs behave as independent in-phase actuators that create, according to the Huygen principle, a flat wave front in the liver.
The use of subcostal and transcostal approaches yielded similar estimations of normal liver stiffness (1.9 and 2.1 kPa for a 90-Hz driving frequency, respectively).
We included oblique planes in our analysis because previous studies have revealed that, in stiff media, a surface longitudinal driver can create a diffraction effect in the propagating wave field, which results in obliquely propagating shear wave beams (35). We did not, however, observe this phenomenon in vivo.
For liver MR imaging, we used a 90-Hz driving frequency because the relatively short wavelength allowed several waves to be depicted in the anterior portion of the liver. The optimum frequency, however, is yet to be determined. Lower frequencies are less attenuated and allow stiffness estimation in deeper portions of the liver. If liver shear stiffness is measured at several different frequencies, it would be possible, in principle, to calculate the shear viscosity of liver tissue, which is potentially an independent parameter for tissue characterization (36).
In recent studies, estimates of the shear stiffness of normal liver tissue vary over an extremely wide range of 0.2–1300 kPa (18,19,23,27,28,36,38). This range reflects the technical difficulty of estimating the elastic properties of semisolid tissue and the varying experimental conditions (in vivo or ex vivo, temperature, type of mechanical excitation, frequency, strain amplitude, etc). If the very high estimates are set aside (18,23), however, the others suggest that the shear modulus of normal liver tissue is between 0.2 and 5.1 kPa. Interestingly, two independent series that used US-based transient elastography found a median liver shear modulus of 1.2–1.43 kPa in healthy volunteers or in patients with no histologic fibrosis (27,28). This finding is in good agreement with our findings. Our results are also in agreement with the 1.8–1.93 kPa stiffness obtained with a subcostal MR elastographic approach but with a different pneumatic driver and 50–80-Hz excitation frequencies (34).
All of the patients with chronic liver disease in this study had MR elastographic liver stiffness measurements that were higher than those of healthy volunteers, which suggests that MR elastography should be evaluated as a method to discriminate normal from fibrotic livers. Whether MR elastography will discriminate different stages of fibrosis remains to be determined in a larger population. The preliminary results of this work suggest a quadratic trend of the curve plotting histologic fibrosis scores versus elasticity measurements. This finding is in agreement with the results of other studies performed either ex vivo (19) or in vivo with US-based transient elastography (27,28). Part of the difficulty in distinguishing among stages 1–3 on the basis of stiffness may be due to the limitations of liver biopsy and grading systems (27). Liver biopsy results are subject to sampling errors, biopsy reproducibility is mediocre, and the degree of fibrosis in the biopsy specimens is not always representative of the findings at autopsy (39,40). These limitations are likely to be even more important in subjects with mild degrees of fibrosis.
Besides fibrosis, steatosis could also theoretically influence liver stiffness measurements. Sandrin et al (27) did not find any influence of steatosis on liver elasticity, but no patient in their series had massive steatosis. It is possible with MR imaging to estimate the degree of steatosis by obtaining dual-echo gradient-echo images (37) in the same plane as the MR elastographic images. MR imaging protocols incorporating MR elastography could thus be a convenient single-step examination to estimate noninvasively the degree of steatosis and fibrosis and to take into account, if necessary, the degree of steatosis in the assessment of the liver stiffness.
MR elastography appears to be an alternative to US-based elastographic techniques (23,27,28) for noninvasive measurement of liver stiffness. MR elastography benefits from the intrinsic advantages of MR imaging, such as freely oriented field of view, no "acoustical window" requirement, the ability to quantify steatosis, and the ability to perform conventional liver MR imaging at the same time. It is also possible that MR elastography might be used to detect or characterize focal liver lesions on the basis of stiffness. Such applications would require more uniform penetration of shear waves in the liver, which might be achieved by using multiple drivers.
As with US-based transient elastography (27), with MR elastography it is possible that the presence of ascites may make it difficult to adequately apply shear waves to the liver through the body wall. This limitation is not important in a practical sense, because the diagnosis of cirrhosis is usually obvious in patients with ascites.
In conclusion, our preliminary results demonstrate that patients with liver fibrosis have higher MR elastographic liver stiffness measurements than do healthy volunteers, which suggests that the MR elastographic technique may have a role in the diagnosis and evaluation of diffuse liver fibrosis. Further studies, however, are needed to define its sensitivity and specificity.
|
ADVANCES IN KNOWLEDGE |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
FOOTNOTES |
|---|
Abbreviations: ROI = region of interest
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, O.R., M.Y., R.L.E.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, O.R., M.Y., M.A.D., R.L.E.; clinical studies, O.R., M.Y., R.L.E.; experimental studies, O.R., M.Y., P.J.R., R.L.E.; statistical analysis, O.R., M.Y., R.L.E.; and manuscript editing, O.R., M.Y., M.A.D., R.L.E.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  N. Salameh, B. Larrat, J. Abarca-Quinones, S. Pallu, M. Dorvillius, I. Leclercq, M. Fink, R. Sinkus, and B. E. Van Beers Early Detection of Steatohepatitis in Fatty Rat Liver by Using MR Elastography Radiology, July 8, 2009; (2009) 2523081817. [Abstract] [Full Text]
|
|
|
|
 |
|
 B. Taouli, R. L. Ehman, and S. B. Reeder Advanced MRI Methods for Assessment of Chronic Liver Disease Am. J. Roentgenol., July 1, 2009; 193(1): 14 - 27. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. A. Talwalkar, M. Yin, S. Venkatesh, P. J. Rossman, R. C. Grimm, A. Manduca, A. Romano, P. S. Kamath, and R. L. Ehman Feasibility of In Vivo MR Elastographic Splenic Stiffness Measurements in the Assessment of Portal Hypertension Am. J. Roentgenol., July 1, 2009; 193(1): 122 - 127. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A M BLAMIRE The technology of MRI -- the next 10 years? Br. J. Radiol., August 1, 2008; 81(968): 601 - 617. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. K. Venkatesh, M. Yin, J. F. Glockner, N. Takahashi, P. A. Araoz, J. A. Talwalkar, and R. L. Ehman MR Elastography of Liver Tumors: Preliminary Results Am. J. Roentgenol., June 1, 2008; 190(6): 1534 - 1540. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. G. Lall, A. M. Aisen, N. Bansal, and K. Sandrasegaran Nonalcoholic Fatty Liver Disease Am. J. Roentgenol., April 1, 2008; 190(4): 993 - 1002. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. C. Georges, J.-J. Hui, Z. Gombos, M. E. McCormick, A. Y. Wang, M. Uemura, R. Mick, P. A. Janmey, E. E. Furth, and R. G. Wells Increased stiffness of the rat liver precedes matrix deposition: implications for fibrosis Am J Physiol Gastrointest Liver Physiol, December 1, 2007; 293(6): G1147 - G1154. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Huwart, C. Sempoux, N. Salameh, J. Jamart, L. Annet, R. Sinkus, F. Peeters, L. C. ter Beek, Y. Horsmans, and B. E. Van Beers Liver Fibrosis: Noninvasive Assessment with MR Elastography versus Aspartate Aminotransferase to-Platelet Ratio Index Radiology, November 1, 2007; 245(2): 458 - 466. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Taouli, A. J. Tolia, M. Losada, J. S. Babb, E. S. Chan, M. A. Bannan, and H. Tobias Diffusion-Weighted MRI for Quantification of Liver Fibrosis: Preliminary Experience Am. J. Roentgenol., October 1, 2007; 189(4): 799 - 806. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. L. Ehman, W. R. Hendee, M. J. Welch, N. R. Dunnick, L. B. Bresolin, R. L. Arenson, S. Baum, H. Hricak, and J. H. Thrall Blueprint for Imaging in Biomedical Research Radiology, July 1, 2007; 244(1): 12 - 27. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
