Liver Imaging 21

C. J. Zech, MD

Department of Clinical Radiology, University Hospitals – Gross hadern, Ludwig Maximilian University of Munich, Marchioni nistr. 15, 81377 Munich, Germany

C O N T E N T S

21.1 Introduction

21.2 3D Sequences

21.3 Navigator Triggering

21.4 Parallel Imaging

21.4.1 Current Value of Parallel Imaging for

Liver MRI

21.4.2 Higher Acceleration Factors with Multi-Channel Systems

21.4.3 Combination of Specifi c Sequence Types with Parallel Imaging 225

21.4.4 Application of Parallel Imaging at High-Field Systems

21.5 Summary

References

Liver Imaging 21

Christoph J. Zech

21.1

Introduction

Magnetic resonance imaging has been used since the early 1990s as a standard procedure in abdominal imaging. Since then, various technical developments have taken place that have helped to further increase the diagnostic value of the method and to further establish image quality. Important milestones towards a fast and robust utilization of MRI in abdominal imaging have been the development of gradient-echo (GRE) sequences, single-shot techniques and phased- array coils. However, it is only now that abdominal

imaging has reached a quality in which respiratory or motion artefacts are negligible and spatial reso- lution is considered suffi cient. Among many other infl uences, the development of 3D sequences, navi- gator triggering and parallel-imaging strategies has had important infl uence in that fi eld. The introduc- tion of high-fi eld systems might allow an additional increase in spatial resolution and image quality. This chapter provides an overview about the above-men- tioned important technical developments for MRI of the liver. It gives background knowledge about these techniques and discusses the clinical value of each innovation derived from the recent literature as well as from our own experience.

21.2

3D Sequences

The evolution from 2D to 3D sequences has been of special value for typical applications requiring a very high spatial resolution and dedicated post-process- ing as, for example, MR angiography. However, it has also gained importance in liver imaging (Lee et al.

2000) . The two basic applications of 3D sequences in abdominal imaging are dynamic studies after bolus injection of contrast agents with T1-weighted 3D gradient-echo sequences and depiction of the biliary tree with MR cholangiography (MRC) with T2-weighted 3D turbo-spin-echo (TSE) sequences, which will be discussed in Chap. 22. The advantage of all 3D techniques is that an isotropic 3D dataset can be acquired, for which various means of post-pro- cessing are possible (Lee et al. 2000; Schaible et al.

2001; Hennig et al. 2005). With 3D sequences a high

spatial resolution can be acquired. The resulting loss

of signal-to-noise ratio (SNR) can be compensated

with the substantial shortening of the T1 relaxation

time by gadolinium chelates for dynamic examina-

tions on the one hand and the long T2 relaxation time

partitions (in contrast to the single slices in 2D tech- niques). 3D techniques have the capacity to examine the liver with high spatial resolution in-plane and along the z-axis, without gaps and with fat satura- tion within one breath-hold (Rofsky et al. 1999).

Although more phase-encoding steps are needed, in contrast to 2D techniques, the overall acquisition time is shortened in 3D sequences by applying low fl ip angles of only 15° to 20° and a repetition time of less than 5 ms (Nitz et al. 2003).

A dedicated 3D sequence for abdominal imaging is the so-called VIBE sequence (volumetric inter- polated breath-hold examination), which is further accelerated by zero-fi lling of the k-space in the slice direction (Rofsky et al. 1999). With this sequence, a slab of about 20-cm thickness can be acquired at 1.5- T in 15 to 20 s with 4-mm slice thickness and a matrix size of 256×192. The VIBE sequence has to be consid- ered superior to a T1-weighted 2D GRE sequence in the early dynamic phases after the injection of gado- linium contrast agents due to a high contrast between gadolinium-enhancing lesions and fat-saturated liver parenchyma in combination with high spatial resolu- tion (Rofsky et al. 1999; Dobritz et al. 2002).

21.3

Navigator Triggering

Artefacts from respiratory motion are one of the most important phenomena disturbing imaging quality in MR examinations of the liver. A breath-hold time of 20 s or less is crucial, therefore, for an acceptable image quality even in patients with a good general state of health. However, this restriction makes an increase in spatial resolution almost impossible, since other sequence parameters with infl uence on the acquisition time like the repetition time or the

hold strategy, respiratory-triggered T2-weighted sequences have been the focus of various studies (Katayama et al. 2001; Augui et al. 2002; Zech et al. 2004). According to Katayama et al. (2001), one advantage of respiratory triggering is a better T2 contrast with an improved signal-to-noise ratio of the liver parenchyma and improved liver-to-lesion contrast in comparison to breath-hold T2-weighted TSE and single-shot sequences. Our own experience shows also superiority of respiratory triggering over breath-hold strategies.

Respiratory triggering can be realized with a respi- ratory belt or with a navigator sequence. For triggering with the belt, a fl exible, air-fi lled tube is used, which is extended when the patient breathes in and the cir- cumference of the abdomen is enlarged. The exten- sion of the air-fi lled tube causes a negative pressure in the tube, which is transmitted to the MR system by a connecting system. The variation of air pressure in the tube represents the movement of the abdominal wall and can be used as an indicator of the respira- tory position of the patient. Instead of the belt, also an air-fi lled cushion can be used, which is squeezed during inspiration resulting in a positive pressure.

Triggering with navigator sequences has been devel- oped mainly for cardiac imaging. The basic principle of the navigator-based technique is the direct visual- ization of the diaphragm movement with sequences that have typically a low spatial, but high temporal resolution. This can be achieved by means of a gradi- ent-echo sequence with a very low fl ip angle (<10°), with a small fi eld of view (and accordingly only a few phase-encoding steps) and with a low matrix size.

Moreover, the low fl ip angle ensures that magnetiza- tion is not saturated and dark lines in the anatomical images therefore are avoided.

For both ways of triggering, the acquisition of the

image data typically takes place in the longer expira-

tion phase. The acquisition begins when the respira-

tory level is below a predefi ned point (usually around

20% of the maximal level of inspiration). This point is called the triggering threshold. Acquisition is trig- gered when the respiratory curve falls below this value. Since during the inspiration phase the tissue has time to rebuild longitudinal magnetization, the effective repetition time is always longer than the minimal repetition time and is equal to the respira- tory cycle time. The time of a single respiratory cycle usually is between 3 and 5 s; therefore, the effective repetition time is ideal for a good T2 weighting.

This makes all free-breathing techniques interest- ing, especially for T2-weighted sequences, whereas the development of robust and effective T1-weighted gradient-echo sequences with respiratory triggering is still in progress.

One of the most important drawbacks of all free- breathing techniques has been their long time of acquisition, which led sometimes to sequences last- ing around 10 min with gross movement artefacts.

This problem has been addressed successfully with the help of parallel imaging strategies, as will be pointed out in Sect. 21.4.1.

21.4

Parallel Imaging

The balance between spatial resolution, signal-to- noise ratio and acquisition time is crucial for liver MRI. Parallel imaging as a universal tool for acceler- ated acquisition is therefore of high interest for liver imaging.

21.4.1

Current Value of Parallel Imaging for Liver MRI

With the introduction of parallel-imaging techniques, the image acquisition can be accelerated. In a recently published evaluation, the acquisition time for high- resolution (5-mm slice thickness, full 320 matrix) T2-weighted TSE sequences with fat-saturation was markedly reduced for breath-hold sequences from 19 to 13 s per nine slices and for respiratory-trig- gered sequences from a mean of 7:02 min to 4:00 min when parallel imaging with an acceleration factor of R=2 and the GRAPPA algorithm (see Chap. 2) were utilized (Zech et al. 2004). This study also showed that this reduction in acquisition time is possible without negative impact on image quality. Neither the

breath-hold nor the respiratory-triggered sequences with parallel imaging were rated inferior in compari- son to the corresponding non-accelerated sequences.

Similar results were also found for a T1-weighted 3D GRE sequence, which can be used as a dynamic sequence after injection of gadolinium (McKenzie et al. 2004). Although image quality in this evaluation was rated inferior for some categories, the authors concluded that overall utilization of parallel imaging is benefi cial also in T1-weighted 3D GRE sequences, especially in patients with limited breath-hold capac- ity (McKenzie et al. 2004). With the use of parallel imaging in T2-weighted single-shot sequences, as for example half-Fourier acquired single-shot turbo spin echo (HASTE) sequences, the decreased number of phase-encoding steps enables shortened echo trains.

This leads to an improved clarity of edges and reduces blurring; cf. Chap. 10.

One problem of parallel acquisition methods is the inferior SNR of these sequences (Griswold et al. 2002) . However, a high SNR and contrast-to- noise ratio (CNR) are crucial in liver imaging due to intrinsic low contrast between certain liver lesions and adjacent liver parenchyma. The GRAPPA algo- rithm shows an improved SNR ratio in comparison to previous AUTO-SMASH techniques (Griswold et al. 2002). Compared with mSENSE, aliasing artefacts in the central parts of the fi eld of view are reduced with GRAPPA; however, image noise is increased in the central parts of the fi eld of view, which is less dis- turbing compared to real aliasing in our experience (Fig. 21.1). Therefore, it might be favorable to use this algorithm for liver MRI, although comparative stud- ies for image quality or lesion detection are not pres- ent.

With the GRAPPA algorithm, image noise is con- centrated in the center of the image. This has some drawbacks. Firstly, in the center of a transversal plane of the upper abdomen, there are always important anatomical structure and liver parenchyma, which might be obscured by these artefacts, although increased image noise is less disturbing than central aliasing, as discussed above. Secondly, as mentioned beforehand in Chap. 4, the calculation of SNR and CNR is not correct with the standard method (using the standard deviation of the background noise).

This can cause problems not only for scientifi c evalu-

ations, but might also be problematic in the daily rou-

tine since, e.g., it might affect the accurate measure-

ment of signal loss or increase after contrast agent

injection for the characterization of focal liver lesions

when liver-specifi c contrast agents are used. However,

if the signal intensity is measured at the same posi- tion pre- and post-contrast and if the parameter set- tings (especially the acceleration factor) and the coil geometry are not changed, this systematic mistake can be negligible. Altogether the overall image qual- ity in an intra-individual comparison was not rated inferior for the parallel-imaging sequences despite these drawbacks (Zech et al. 2004). This indicates that the SNR loss for parallel imaging is within a limit that does not visually disturb the images (Fig. 21.2).

In our experience, we can fully recommend the use of parallel imaging with an acceleration factor of R=2 in liver MRI. For breath-hold techniques, the saved time with parallel imaging is on the one

hand important to minimize respiratory artefacts, and on the other hand the saved time can be used for, e.g., T2-contrast optimization employing a longer TR and a shorter echo train length or to increase the spatial resolution. For respiratory- triggered sequences the work-flow and patient acceptance can be increased by shorter times of acquisition. Moreover, the spatial resolution can be optimized for the combination of respiratory trig- gering with parallel imaging even up to a full 384 matrices without running into gross artefacts from patient movement, which becomes problematic in respiratory triggered sequences exceeding 5 min time of acquisition (Fig. 21.3).

Fig. 21.1. Appearance and spatial dis- tribution of artefacts arising from the different parallel-imaging algorithms.

For both sequences a too-small fi eld

of view was chosen to generate super-

aliasing. With the mSENSE algorithm

(upper row) this results in severe cen-

tral aliasing, which obscures central

parts of the section shown. With the

GRAPPA algorithm image quality is

also slightly reduced by centralized

image noise; however, in direct com-

parison to the mSENSE image there

are fewer artefacts

Fig. 21.2. Differences in image quality for breath-hold sequences without (upper left) and with parallel imaging (GRAPPA algo- rithm, acceleration factor R=2) (upper right) as well as for respiratory-triggered sequences without (lower left) and with parallel imaging (GRAPPA algorithm, acceleration factor R=2) (lower right) in a T2-weighted TSE sequence in a healthy volunteer. Note the signifi cantly improved image quality with parallel imaging in the breath-hold examination (upper row) due to the reduced breathing artefacts. The image quality in the respiratory-triggered sequences shows no signifi cant difference (lower row); however, with utilization of parallel imaging, it was possible to reduce the time of acquisition from 6:24 min to 3:01 min in this volunteer

21.4.2

Higher Acceleration Factors with Multi-Channel Systems

The introduction of higher acceleration factors is still problematic. One major problem is the loss of the signal-to-noise ratio, which reaches a critical level with acceleration factors of R=4 (Fig. 21.4).

The second diffi cult point is that the phase-encod- ing direction in most clinically applied abdominal sequences is anterior-posterior for the transversal orientation, which is still the most important ori- entation in the abdomen. The anterior-posterior phase-encoding direction is chosen due to the rect-

angular fi eld of view, which helps to save acquisi- tion time by a reduced number of phase-encoding steps.

To achieve optimally differing coil sensitivity pro-

fi les in the phase-encoding direction, the elements

of a phased-array coil have to be arranged parallel

to the direction of the phase-encoding steps. This

is anatomically diffi cult for more than two coil ele-

ments in transversal slices with phase encoding in the

anterior-posterior direction because of the typically

larger left-right diameter of the abdomen compared

to its anterior-posterior diameter. Thus, optimal coil

geometry for parallel imaging with higher accelera-

tion factors in a transversal orientation is not pro-

vided by typically used coil systems, in contrast to the coronal orientation with phase encoding in the left-right direction. However, with dedicated coil arrangements, imaging artefacts in axial slices can be decreased. In case of multi-channel MR systems, more than one conventional phased-array coil (with, e.g., six coil elements each) can be wrapped around the abdomen as shown in Fig. 21.5. Therefore, higher factors of acceleration are theoretically feasible in the anterior-posterior direction; however, they still suffer from increased artefacts.

For the coronal orientation, the multi-channel phased-array coils placed on the abdomen have an ideal position to acquire higher acceleration factors in the left-right direction. This is helpful for the acquisi- tion of MRCP examinations, as discussed in Chap. 22.

Fig. 21.3. Female patient suffering from breast cancer; MRI to rule out liver metastases. Although the patient has been in a good general state of health, she was not able to hold her breath properly, which is indicated by severe breathing artefacts in the T2-weighted TSE sequence with fat saturation.

These artefacts also obscure the focal liver lesion in segment 7. In the cor- responding slice of the respiratory- triggered T2-weighted TSE sequence, the focal lesion with only slight hy- perintensity (arrow), suspicious of a metastasis, is excellently depicted. Note also the tiny liver cyst adjacent to the metastasis (small arrow). The time of acquisition was three times a 16-s breath-hold (effective time of acquisi- tion with intervals between the indi- vidual breath-holds was around 2 min) versus 2:51 min for the respiratory- triggered sequence. Both sequences are acquired with parallel imaging (accel- eration factor R=2)

Innovative coil designs may enable higher accel- erations in the future; however, meanwhile accelera- tion factors of usually up to R=2 are feasible for the transversal orientation, whereas factors up to R=4 are feasible for the coronal orientation.

21.4.3

Combination of

Specifi c Sequence Types with Parallel Imaging

The utilization of parallel imaging offers advanta-

geous effects in several sequence types, especially

in the different single-shot techniques. Single-shot

fast spin echo pulse sequences, such as the HASTE

sequence, often suffer from image artefacts related

Fig. 21.4. Limitations for the utilization of higher acceleration factors in transversal studies of the upper abdomen. The same T1-weighted gradient-echo sequence with an acceleration factor of R=2 (left), R=3 (middle) and R=4 (right) is presented. Im- aging artefacts due to parallel imaging cannot be delineated with an acceleration factor of R=2, whereas with a factor of R=3 substantial and a factor of R=4 severe artefacts in the center of the fi eld of view can be delineated. These artefacts are also present with multi-channel MRI systems due to the coil geometry for the transversal orientation with the typical coil sensitivity profi les (explanation in the text)

to their long echo trains with apparent blurring due to the T2 signal decay during the readout of the echo train (cf. Chap. 10). This problem can be reduced by applying parallel imaging. With an acceleration factor of 2, approximately only half of the phase-encoding steps have to be measured. The thereby shortened length of the echo train results in more signal, since late echoes with low signal are no longer acquired.

This leads to an increased image quality with reduced blurring artefacts without the loss of spatial resolu- tion (Fig. 21.6).

The shortened echo train is of particular value for echo-planar imaging (EPI) sequences since they suffer from severe distortions due to eddy-current and off-resonance phenomena. Because of their high sensitivity for inhomogeneities in the magnetic fi eld, these sequences at present do not play a major role in daily clinical practice for abdominal imaging. Par- allel imaging together with optimized gradient sys- tems can help to establish a robust image quality for abdominal EPI sequences. The black-blood effect, which can be achieved by utilizing a slight diffusion gradient (e.g., with a b-value of 50 s/mm

), helps to establish a contrast situation which is optimal for the detection of focal liver lesions: since the liver as well as the hepatic vasculature is dark, focal lesions are depicted as bright spots within a dark background.

Especially the problem of small peripheral liver veins mimicking or masking small focal liver lesions is markedly reduced (Fig. 21.7).

Preliminary results in comparison to a standard T2-weighted TSE sequence have shown promising results with regard to image quality, robustness, and

liver lesion detection (Zech et al. 2005). However, susceptibility artefacts, especially at the liver mar- gins, are still present, which can cause false-posi- tive fi ndings. Therefore, the value of these sequences might be the fast perception of lesions, which has to be validated on the additionally performed contrast- enhanced examination.

21.4.4

Application of Parallel Imaging at High-Field Systems

With 1.5-T state-of-the-art scanners, a very high image quality can be reached with parallel imaging in combination with navigator triggering, as men- tioned earlier. However, signal-to-noise constraints still exist at 1.5 T: with parallel imaging the SNR is at least reduced by the square root of the acceleration factor R. Thus, with higher acceleration factors, e.g, of three, the SNR is reduced to at least 58% of the original value. This causes a visible increase of image noise in the images at 1.5 T. Moreover, an increase in spatial resolution in breath-hold examinations at 1.5 T is nearly impossible, since a breath-hold time of about 20 s cannot be exceeded.

Due to the higher SNR at 3 T, image noise at com-

parable acceleration factors is barely visible in the

images, even when a very high spatial resolution is

acquired (see also Fig. 21.6). Especially T1-weighted

gradient-echo sequences, which are unsuitable for

respiratory triggering, can be acquired in a very high

spatial resolution with excellent image quality within

Fig. 21.5. Two different coil arrangements for a T1-weighted 2D gradient-echo sequence. With multi-channel systems, as shown here with the Total Imaging Matrix (TIM) system (Siemens Medical Solutions, Erlangen, Germany), more than one phased-array coil can be positioned on the patient. The phased-array coils consist of two clusters with up to three matrix coil elements each, which can be used as separate receivers in the so-called triple mode (see Chap. 13). The upper

repositioning. Together with the three spine elements integrated into the patient table, six coil elements are available for signal reception at each anatomic level. The lower row shows a coil arrangement for dedicated MR examinations of the upper abdomen (e.g., dedicated MRI of the liver), where two fl exible coil arrays are wrapped around the patient.

With the help of this arrangement, more sensitivity profi les from a total of 18 different coil elements (2×2×3 elements

in the phased-array coils + 1×2×3 elements in the spine matrix) help to decrease the typical parallel imaging artefacts

(centralized increase in noise), which can be delineated in the images of the upper row. Both sequences have the same

imaging parameters and use parallel imaging (GRAPPA algorithm) with an acceleration factor of R=2.

an acceptable time of acquisition, which is not pos- sible with 1.5 T systems (Figs. 21.8 and 21.9).

However, the application of abdominal studies to high-fi eld systems is still considered problematic due to arising new problems. One of these problems is the specifi c absorption rate (SAR), which increases qua- dratically with the fi eld strength, resulting in a major limitation, particularly for turbo-spin-echo sequences

(Kangarlu et al. 1999). Among other innovative tech- niques, such as, for example, the hyperecho technique (Scheffl er et al. 2003), which cannot be explained in detail here, parallel imaging helps to decrease the number of phase-encoding steps, thereby reducing the SAR effectively by 1/R with respect to the long- term averaging of SAR contributions. Other problems like dielectric resonance effects can be addressed by

Fig. 21.6. Infl uence of parallel imaging on blurring artefacts in single-shot sequences for the example of a coronal HASTE sequence of the liver. The sequences were acquired with a dedicated 32-channel coil (Rapid Biomedical, Rimpar, Germany).

Without parallel imaging severe blurring artefacts obscure the margins of the liver (upper left), with increasing acceleration

factors of R=4 (upper right), R=6 (lower left) and R=8 (lower right) these blurring artefacts disappear, because the echo train

length can be reduced effectively from 1,756 ms without parallel imaging to 220 ms with an acceleration factor of R=8. However,

due to increased image noise in the center of the fi eld of view – which is a typical artefact from parallel imaging – the optimal

image quality is reached at a factor of R=6. The high image quality despite such a high acceleration factor is due to the fi eld

strength of 3 Tesla in this example (courtesy of Mark Griswold PhD, RAPID Biomedical, Rimpar, Germany and University of

Wuerzburg)

external dielectric gel pads, placed on the anterior body surface, which increase the resistance for focal eddy currents. Overall, the combination of higher fi eld strength with parallel imaging might offer new horizons for abdominal MRI.

21.5 Summary

In summary, the above-mentioned innovations help to establish a very high quality in MRI examinations of the liver. Each innovation alone, but particularly their combination, is of great value. Whereas the application

Fig. 21.7. Comparison of a diffusion-weighted EPI sequence (left) with a standard T2-weigthed TSE sequence (right). Due to the black-blood effect of the diffusion-weighted sequence, multiple liver metastases in this patient suffering from colorectal carcinoma can be delineated easily in contrast to the T2-weighted sequence, where high signal from liver vessels makes the evaluation more diffi cult in the upper row. The lower row shows a patient with liver cirrhosis and ascites suffering from a he- patocellular carcinoma, which can be delineated only in the diffusion-weighted sequence (arrow) due to fewer artefacts and superior tumor-to-liver contrast

of 3D sequences and navigator triggering is still con- fi ned to some sequences, parallel imaging is a universal technique to speed up the acquisition in all sequence types used for liver imaging at present. Moreover, the utilization of parallel imaging also allows increasing the spectrum of suitable sequences, as shown above with the example of diffusion-weighted EPI sequences.

This outstanding and universal role of parallel imag- ing holds also true for the combination of parallel imaging with high-fi eld MR systems.

Future developments have to aim at new coil

geometries and more refi ned algorithms, which

might allow even higher acceleration factors also for

transversally orientated acquisitions without the loss

of image quality. This is of importance as this will

remain the orientation of choice for liver imaging.

Fig. 21.8. T1-weighted 2D gradient-echo sequence from a 3-T MRI system (Mag- netom Tim Trio, Siemens Medical Solu- tions, Erlangen) at three different levels.

Throughout the whole liver, an excellent

image quality can be seen. Note also the

brilliant delineation of the right (mid-

sequence with a 384 matrix and 5-mm

slice thickness can be acquired at 3 T

using a parallel imaging acceleration of

total of 39 slices the whole liver can be

imaged without limitations in anatomic

coverage)

Fig. 21.9. MR examination of a 49-year-old female suffering from a Klatskin tumor with status post right-sided hemihepatec- tomy. T1-weighted 3D gradient-echo sequence (VIBE) in the liver-specifi c phase after injection of Gadolinium-EOB-DTPA (Primovist, Schering Deutschland GmbH, Berlin) at a 1.5-T MRI system (Magnetom Avanto, Siemens Medical Solutions, Er- langen) in the upper row and a 3-T MRI system (Magnetom Tim Trio, Siemens Medical Solutions, Erlangen) in the lower row.

Note the excellent visualization of only 2 mm measuring tiny metastasis in the periphery of liver segment 2, which is enlarged after the hemihepatectomy; the depiction of the subtle fi nding is sharper in the sequence acquired on the 3-T MRI system due to the increased spatial resolution (3-mm slices; 256×180 in-plane matrix at 1.5 T; 2-mm slices; 320×160 in-plane matrix at 3 T;

parallel imaging acceleration of R=2 in both acquisitions)

References

Augui J, Vignaux O, Argaud C et al (2002) Liver: T2-weighted MR imaging with breath-hold fast-recovery optimized fast spin-echo compared with breath-hold half-Fourier and non-breath-hold respiratory-triggered fast spin-echo pulse sequences. Radiology 223:853–859

Dobritz M, Radkow T, Nittka M, Bautz W, Fellner FA (2002) VIBE with parallel acquisition technique - a novel approach to dynamic contrast-enhanced MR imaging of the liver.

Röfo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 174:738–741

Griswold MA, Jakob PM, Heidemann RM et al (2002) Gen- eralized auto-calibrating partially parallel acquisitions (GRAPPA). Magn Reson Med 47:1202–1210

Hennig J, Weigel M, Scheffl er K (2003) Multiecho sequences with variable refocusing fl ip angles: optimization of signal behavior using smooth transitions between pseudo steady states (TRAPS). Magn Reson Med 49:527–535

Kangarlu A, Baertlein BA, Lee R, et al (1999) Dielectric reso- nance phenomena in ultra high fi eld MRI. J Comput Assist Tomogr 23:821–831

Katayama M, Masui T, Kobayashi S et al (2001) Fat-sup- pressed T2-weighted MRI of the liver: comparison of respiratory-triggered fast spin-echo, breath-hold single- shot fast spin-echo, and breath-hold fast-recovery fast spin-echo sequences. J Magn Reson Imaging 14:439–

449

McKenzie CA, Lim D, Ransil BJ et al (2004) Shortening MR image acquisition time for volumetric interpolated breath- hold examination with a recently developed parallel imag-

ing reconstruction technique: clinical feasibility. Radiology 230:589–594

Lee VS, Lavelle MT, Rofsky NM et al (2000) Hepatic MR imag- ing with a dynamic contrast-enhanced isotropic volumet- ric interpolated breath-hold examination: feasibility, repro- ducibility, and technical quality. Radiology 215:365–372 Nitz WR (2003) Magnetic resonance imaging. Sequence acro-

nyms and other abbreviations in MR imaging. Radiologe 43:745–763

Rofsky NM, Lee VS, Laub G et al. (1999) Abdominal MR imag- ing with a volumetric interpolated breath-hold examina- tion. Radiology 212:876–884

Schaible R, Textor J, Kreft B, Neubrand M, Schild H (2001) Value of selective MIP reconstructions in respiratory trig- gered 3D TSE MR-cholangiography on a workstation in comparison with MIP standard projections and single- shot MRCP. Röfo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 173:416–423

Zech CJ, Herrmann KA, Huber A, Dietrich O, Stemmer A, Herzog P, Reiser MF, Schoenberg SO (2004) High-resolu- tion MR-imaging of the liver with T2-weighted sequences using integrated parallel imaging: Comparison of prospec- tive motion correction and respiratory triggering. J Magn Reson Imaging 20:443-450

Zech CJ, Schoenberg SO, Menzel MI, Herrmann KA, Dietrich

O, Reiser MF (2005) Black-blood-EPI sequences with paral-

lel imaging for detection of focal liver lesions - compari-

son to a standard T2-weighted sequence. Proc Intl Soc Mag

Reson Med 13:2070