| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Technical Developments |
1 From the Department of Radiology, University of Bonn, Sigmund-Freud-Strasse 25, D-53105 Bonn, Germany (P.M., S.F., F.T., H.H.S.); and Philips Medical Systems, Best, the Netherlands (J.S.B., J.G.). Received June 27, 2001; revision requested August 16; revision received October 24; accepted December 10. P.M. supported by a BONFOR grant from the Medical Department of the University of Bonn. Address correspondence to P.M. (e-mail: muertz@uni-bonn.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMaterials and Methods Results DiscussionREFERENCES |
|---|
© RSNA, 2002
Index terms: Abdomen, MR, 70.12144, 70.121416 • Magnetic resonance (MR), diffusion study, 70.12144 • Magnetic resonance (MR), echo planar, 70.121416 • Magnetic resonance (MR), technology, 70.121416
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods Results DiscussionREFERENCES |
|---|
Ideal pulse sequences for diffusion-weighted MR imaging of the abdomen are still matters for research (2–7). Breath holding is one requirement to reduce respiratory motion at diffusion-weighted MR imaging of the abdomen. However, such an approach limits the acquisition time and, consequently, both SNR and the spatial resolution are compromised. Until now, the best image quality was obtained with single-shot echo-planar MR imaging sequences in breath-hold conditions (8–14). To date, fixed repetition times were used in these pulse sequences. Consequently, diffusion-weighted MR images with varying b values are acquired in different phases of the cardiac cycle. The b value incorporates the gradient strength and timing parameters that affect the diffusion sensitivity. The signal attenuation due to diffusion is given by ln(SIb/SI0) = -ADC x b, where SIb is the signal intensity on the diffusion-weighted MR image, SI0 is the signal intensity on the MR image obtained without diffusion gradients, and ADC is the apparent diffusion coefficient (1). Thus, the diffusion-weighted MR images of one series may be affected differently by pulsatile motion, which could limit the accuracy of the ADC measurement.
As indicated by findings in preliminary experiments (7), diffusion-weighted MR imaging of the abdomen can be improved if data acquisition is triggered to the cardiac cycle. Thus, the purpose of our study was to investigate the extent to which quantitative diffusion measurements of abdominal organs can be improved by using pulse-triggered single-shot sequences.
|
Materials and Methods |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods Results DiscussionREFERENCES |
|---|
MR Imaging
All MR imaging examinations were performed at 1.5 T (Gyroscan Intera [maximum amplitude of the gradients, 30 mT/m; minimum slope, 0.2 msec; maximum slew rate, 150 T/m/sec]; Philips Medical Systems, Best, the Netherlands) with a four-element surface coil for optimum SNR.
A finger pulse–triggered diffusion-weighted single-shot spin-echo echo-planar MR imaging sequence (repetition time of four heartbeats, echo time of 83 msec, field of view of 400 x 280 mm, matrix of 128 x 90 interpolated to 256 x 256, echo-planar imaging factor of 61, half-Fourier factor of 0.672, diffusion gradient duration of 26.7 msec, diffusion gradient distance of 41.6 msec, bandwidth of 1,560 Hz/pixel, one signal acquired) was used to acquire diffusion-weighted MR images. The images were acquired at different phases of the cardiac cycle with the following b values: 50, 300, 700, 1,000, and 1,300 sec/mm2. The sequence was triggered to the increasing slope of the pulse wave, and acquisition was started after a trigger delay, which was varied in steps of 100 msec between 45 msec and the maximal adjustable value (500–600 msec, depending on the heart rate).
The diffusion-weighted gradients were applied in three orthogonal directions called M', P', and S', which are defined as M' = (x, -0.5y, z), P' = (-0.5x, y, z), S' = (x, y, -0.5z), with x pointing from floor to ceiling, y from left to right when standing in front of the magnet, and z from feet to head. Diffusion-weighted MR images in each gradient direction (five b values) were acquired during one breath hold in expiration (acquisition time, about 17 seconds for 70 heartbeats per minute). Sequence calibration cycles were performed during a separate breath hold. Four regional presaturation slabs were used, two parallel to the sections and two orthogonal, to suppress motion influences and fat signal from the abdominal wall. Fat suppression (spatial presaturation with inversion recovery) was used to suppress fat signals around the kidneys. Four transverse MR images with 8-mm section thickness and 18-mm gap were acquired during each breath hold to allow the liver, spleen, and kidneys to be imaged in at least three sections (Fig 1).
|
Data Processing and Analysis
Optimum trigger delay and its effect on accuracy of ADC measurements.—Diffusion-weighted MR images obtained at different phases of the cardiac cycle were first analyzed on the basis of motion artifacts on the images and then on the basis of influences on the signal intensities. The motion artifacts on the diffusion-weighted MR images were scored with agreement (P.M., S.F.) on a three-point scale: 1, no obvious motion artifacts; 2, moderate artifacts; 3, strong motion artifacts. Influences on the signal intensities were investigated by comparing mean signal intensities in regions of interest placed in liver (20–50 pixels with larger vessels excluded), spleen (20–40 pixels), and renal cortex (15–20 pixels) with the reference signal intensity of the water bottle. The trigger delay that resulted in maximum signal intensities of the abdominal organs compared with the reference signal intensity, for which the signal attenuation due to macroscopic motion influences was at minimum, was defined as the best trigger delay. The SNR on images obtained with a b value of 50 sec/mm2 was determined as the mean signal intensity in the regions of interest divided by the SD of the background signal intensity.
The influence of pulse triggering on the accuracy of ADC measurements was then investigated for selected regions of interest placed in liver, spleen, and kidney by using diffusion gradients in the S' direction. The diffusion-weighted MR series were analyzed by investigating the linearity of the natural logarithm of the mean signal intensities versus b value. The accuracy of ADC measurements was determined on the basis of the fitting error of the linear regression. Optimally triggered data were compared with nontriggered data.
Effect of pulse triggering, diffusion gradient direction, and section positioning on measurements of ADC of liver, spleen, and renal cortex and their reproducibility.—ADC parameter maps were generated by means of pixelwise linear regression of the natural logarithm of the signal intensities versus b values for optimally triggered and nontriggered data. The quality of the ADC maps was scored with agreement (P.M., S.F.) on a two-point scale: 1, organs appeared homogeneous; 2, organs partially or completely disappeared or appeared heterogeneous. With these parameter maps, ADCs of the abdominal organs were determined by positioning the regions of interest on the MR images obtained with a b value of 50 sec/mm2 and copying them to the ADC maps. The ADC maps obtained with repeated measurements were analyzed in identical anatomic locations. For each volunteer, ADCs for liver, spleen, and renal cortex were determined on three images for three diffusion gradient directions for both triggered and nontriggered data, which yielded 1,944 measurements.
Reproducibility of single ADC measurements for each anatomic location was analyzed by determining the averaged ADCs ± SDs of repeated measurements (n = 3). These averaged ADCs and corresponding SDs derived for each volunteer were then grouped and averaged separately for each organ and each diffusion gradient direction (n = 36). For each group, the effect of triggering on the ADCs and their reproducibility was investigated separately by comparing triggered and nontriggered data.
The influence of the diffusion gradient orientation on the ADCs and their reproducibility was assessed separately for each organ by comparing grouped data for different diffusion gradient orientations. Multiple comparisons were performed of data for the three gradient directions. Triggered and nontriggered data were investigated separately.
Similarly, the influence of section position on the ADCs and their reproducibility was assessed for each organ. Therefore, the data were grouped according to the section position, and each group contained all data for different volunteers and different diffusion gradient directions (n = 36). Multiple comparisons were performed of data for the three sections. Again, triggered and nontriggered data were investigated separately.
Physiologic range of ADCs for liver, spleen, and renal cortex.—The range of ADCs was determined by analyzing the averaged ADCs for three repeated measurements to reduce random fluctuations of single measurements. Furthermore, mean averaged ADCs ± SDs, which represent the interindividual variations, were determined separately for each organ and each diffusion gradient direction. The results for acquisitions with and those without triggering and for different diffusion gradient directions were compared, and those with the smallest SDs and the best reproducibility were summarized.
Statistical Analysis
For statistical comparison between triggered and nontriggered data sets, a Student t test for matched pairs was used. A two-tailed P < .05 was considered statistically significant. For multiple comparisons, as performed with the data for diffusion gradient directions, a Bonferroni correction for three comparisons was applied, and the level of significance was set to P = .017.
|
Results |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods Results DiscussionREFERENCES |
|---|
|
|
|
|
|
Table 2 shows the results of diffusion measurements in different sections. Mean ADCs differed much more for different sections in acquisitions without triggering: as much as 16% for liver, 28% for spleen, and 5% for renal cortex. Mean ADCs and mean SDs for the liver and, particularly, the spleen were larger for section 4, which is closest to the heart and diaphragm. With triggering, much smaller and nonsignificant differences were found between the mean ADCs and mean SDs for the different sections. The differences between mean ADCs were less than 2% for all organs.
|
The following normal ADCs (mean ± SD) were determined: (a) liver, P' direction, 96 x 10-5 mm2/sec ± 9 (range, 80–120 x 10-5 mm2/sec), (b) spleen, P' direction, 60 x 10-5 mm2/sec ± 4 (range, 47–67 x 10-5 mm2/sec), (c) renal cortex, S' direction, 164 x 10-5 mm2/sec ± 9 (range, 147–184 x 10-5 mm2/sec).
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods Results DiscussionREFERENCES |
|---|
To date, applications in the liver are limited to the differentiation of lesions with large differences in ADC, such as abscesses and necrotic tumor (15). Further differentiation between various solid lesions is difficult due to large interindividual SDs of the ADCs (10,12,14,17). Similarly, only a weak correlation was found between ADCs and serum creatinine levels in patients with renal failure or renal artery stenosis (8,13,18). One reason for these unsatisfactory results in previous studies might be the use of fixed repetition times in the single-shot spin-echo echo-planar MR imaging sequences. With a fixed repetition time, images in a diffusion series are acquired in different phases of the cardiac cycle. Therefore, diffusion-weighted MR images may be affected differently by pulsatile motion, which could limit the accuracy of ADC determination. As indicated by preliminary results (7), synchronization of the acquisition to the heart phase may substantially influence the ADC determination. Therefore, we investigated the extent to which quantitative diffusion measurements in abdominal organs can be improved by using a pulse-triggered acquisition scheme.
We conducted our study with 12 volunteers and two acquisition schemes (one with and one without triggering) by applying diffusion gradients in three directions, investigating several sections per organ, and repeating each measurement three times. The use of b values between 50 and 1,300 sec/mm2 helped reduce perfusion influences (10,12,14). The use of five b values guaranteed precise measurement of a wide range of ADCs at the same time, which is important because the ADCs of spleen, liver, and kidney are different. Moreover, in pathologic conditions, ADCs that are much higher than those measured in this study may appear (eg, for malignant lesions, which contain a considerable amount of free water).
Optimum SNR was obtained by using an array of surface coils. Echo times were kept short to allow sufficient SNR in the liver with even the highest b values. Additionally, all diffusion-weighted MR images in one series were acquired during a single breath hold. In earlier studies, different b values were most often acquired during different breath holds. Thus, acquisition of the same pixel location was not guaranteed with different b values, and section misregistration influenced ADC map calculation (12). By keeping these optimized imaging parameters constant for acquisitions with and those without triggering, we could demonstrate that pulse triggering on the diastolic heart phase enhances the quality of diffusion-weighted images, as well as the accuracy and reproducibility of the ADCs and their interindividual SDs.
ADCs obtained with triggering were smaller than those obtained with fixed repetition times, which indicates that ADCs acquired without triggering are artificially enlarged by motion influences. Without triggering, ADCs varied much more with respect to diffusion gradient orientation. Without triggering, the ADCs of liver and spleen, as well as their interindividual SDs, tended to be greater in the M' direction than in the P' and S' directions. These findings may be explained by the fact that cardiac motion that is transferred directly through the diaphragm has a larger part in the M' direction than in the P' and S' directions. With triggering, differences between ADCs for different diffusion gradient directions were less than 5% for all organs.
For triggered acquisitions, we noticed only residual influences on ADC measurements in the spleen in the S' diffusion gradient direction. These influences may be caused by pulsatile motion of the aorta or cardiac motion that is transferred through the diaphragm with a combined preferred direction in the S' diffusion gradient direction. These results also show that if influences on residual motion are sufficiently reduced with triggering, there is no remarkable physiologic diffusion anisotropy for the organs. Similarly, differences between ADCs for different sections are remarkably smaller with triggering than with fixed repetition times. Without triggering, ADCs for liver and spleen were remarkably greater in the section placed nearest to the heart owing to the influence of cardiac motion. With triggering, differences between ADCs for different sections were less than 2% and were not significant. Thus, for reliable diffusion measurements independent of section and diffusion gradient orientation, triggering is a requirement.
In previous studies in which high b values were used, mean ADCs for liver, spleen, and kidney, respectively, of 87 x 10-5 mm2/sec ± 26, 88 x 10-5 mm2/sec ± 33, and 155 x 10-5 mm2/sec ± 27 were given by Yamada et al (14); 69 x 10-5 mm2/sec ± 31, 78 x 10-5 mm2/sec ± 35, and 163 x 10-5 mm2/sec ± 34 by Namimoto et al (12); and 105 x 10-5 mm2/sec ± 30, 93 x 10-5 mm2/sec ± 30, and 179 x 10-5 mm2/sec ± 27 by Kim et al (10). Compared with these values, the normal ADCs determined with triggering in the present study tended to be smaller, whereas values determined without triggering in sections placed around the center of the organs were in the range of the ADCs reported in earlier studies. This fact proves the hypothesis that ADCs in previous studies were affected by motion influences. The more than threefold smaller interindividual SDs of the normal ADCs in our study shows the extent of motion influences without triggering.
The technical improvement in abdominal diffusion-weighted MR imaging presented in this study has several potential clinical applications. Whether improved reproducibility of ADC measurements will allow more reliable differentiation of larger focal liver lesions is still questionable, but it will allow better monitoring of patients. Potential applications include follow-up of patients with renal failure or of those with transplanted kidneys. Whether changes in ADC measurements can help detection of pathologic changes at an earlier stage will have to be studied. The pulse sequence proposed in this study is suitable for clinical applications. Use of a repetition time of four heartbeats allows acquisition of four sections at the same trigger delay and guarantees unchanged ADC measurements if some oximetry pulses are missed with the triggering unit. Even then, the acquisition is short enough to allow use of breath-hold conditions.
Diffusion-weighted MR imaging of the abdomen as presented in this study has several limitations. The reduced in-plane resolution of about 3 mm and the section thickness of 8 mm are compromises between spatial resolution and SNR. Thus, the sequence is suitable for investigation of diffuse diseases and larger lesions but not of small lesions. Moreover, separation of the sequence calibration cycles and the image acquisition pulses into two breath holds is a special feature that may not be available with other MR imaging systems. Thus, the acquisition duration would be prolonged to about 25 seconds for 70 heartbeats per minute, which would exceed a tolerable breath-hold duration for most patients. For clinical applications in which a smaller range of ADCs is expected, however, the acquisition time can be reduced by using fewer b values.
For investigations of renal diseases, application of three diffusion gradients in orthogonal directions in a single breath hold and analysis of the anisotropic diffusion in the renal medulla would be more relevant than the application of more than two b values. In contrast to findings in previous studies (10,12,15), no major distortion artifacts were observed on the diffusion-weighted MR images in this study. This finding leads to the question of whether the large artifacts in the lateral segment and subdiaphragmatic region of the liver (10) might be caused by motion influences rather than by distortion effects. Furthermore, the short echo times used in our study require high gradient performance. Moreover, the use of whole-body gradients with amplitudes higher than 30 mT/m, as in this study, would improve the SNR, which can be traded off against spatial resolution.
In conclusion, the application of pulse triggering with single-shot sequences improves diffusion-weighted MR imaging of the abdomen and permits a more reliable determination of the ADCs.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, P.M., H.H.S.; study concepts, P.M., F.T., J.G.; study design, P.M., J.S.v.d.B.; literature research, P.M.; clinical studies, P.M., J.S.v.d.B.; data acquisition, P.M., S.F.; data analysis/interpretation, P.M., S.F., F.T.; statistical analysis, P.M., S.F.; manuscript preparation and definition of intellectual content, P.M.; manuscript editing, S.F.; manuscript revision/review, F.T., J.G., J.S.v.d.B.; manuscript final version approval, H.H.S.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods Results DiscussionREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  N. Inan, A. Arslan, G. Akansel, Y. Anik, and A. Demirci Diffusion-Weighted Imaging in the Differential Diagnosis of Cystic Lesions of the Pancreas Am. J. Roentgenol., October 1, 2008; 191(4): 1115 - 1121. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Zhang, Y. Mazaheri Tehrani, L. Wang, N. M. Ishill, L. H. Schwartz, and H. Hricak Renal Masses: Characterization with Diffusion-weighted MR Imaging--A Preliminary Experience Radiology, May 1, 2008; 247(2): 458 - 464. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Nomori, T. Mori, K. Ikeda, K. Kawanaka, S. Shiraishi, K. Katahira, and Y. Yamashita Diffusion-weighted magnetic resonance imaging can be used in place of positron emission tomography for N staging of non-small cell lung cancer with fewer false-positive results. J. Thorac. Cardiovasc. Surg., April 1, 2008; 135(4): 816 - 822. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Inan, A. Arslan, G. Akansel, Y. Anik, H. T. Sarisoy, E. Ciftci, and A. Demirci Diffusion-Weighted Imaging in the Differential Diagnosis of Simple and Hydatid Cysts of the Liver Am. J. Roentgenol., November 1, 2007; 189(5): 1031 - 1036. [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]
|
|
|
|
|
|
D.-M. Koh and D. J. Collins Diffusion-Weighted MRI in the Body: Applications and Challenges in Oncology Am. J. Roentgenol., June 1, 2007; 188(6): 1622 - 1635. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Yoshikawa, H. Kawamitsu, D. G. Mitchell, Y. Ohno, Y. Ku, Y. Seo, M. Fujii, and K. Sugimura ADC measurement of abdominal organs and lesions using parallel imaging technique. Am. J. Roentgenol., December 1, 2006; 187(6): 1521 - 1530. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. C. Thoeny, F. De Keyzer, R. H. Oyen, and R. R. Peeters Diffusion-weighted MR Imaging of Kidneys in Healthy Volunteers and Patients with Parenchymal Diseases: Initial Experience Radiology, June 1, 2005; 235(3): 911 - 917. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Taouli, A. J. Martin, A. Qayyum, R. B. Merriman, D. Vigneron, B. M. Yeh, and F. V. Coakley Parallel Imaging and Diffusion Tensor Imaging for Diffusion-Weighted MRI of the Liver: Preliminary Experience in Healthy Volunteers Am. J. Roentgenol., September 1, 2004; 183(3): 677 - 680. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
) and SDs (error bars) are depicted for eight repeated measurements for the same volunteer, with diffusion gradients in the S' direction. The relative signal intensity—SIk/SIw, where SIk is the signal intensity of the kidney and SIw is the signal intensity of water—is scaled to the value with a trigger delay of 600 msec.
