Original Research

Sequence Optimization in MangafodipirTrisodium-Enhanced Liver and Pancreas MRI

Chen Wang, MD,1 Lars Johansson,1 Andre Western, MSc,2 Hans Fagertun, MSc,2

and Hakan Ahlstrom, MD, PhD1

To find an optimal magnetic resonance (MR) sequence formangafodipir trisodium-enhanced liver and pancreas imag-ing, six healthy volunteers were studied using a 1.5 T MRsystem with different T1-weighted abdominal imaging se-quences. These were turbo field (gradient)-echo (TFE), fastfield (gradient)-echo (FFE), and spin-echo sequences beforeand after mangafodipir trisodium administration. Variousparameter combinations were investigated within eachsequence type, and then the best combination was foundand compared with those of the other sequences. Signalintensity (SI) measurements were made in regions of inter-est in the liver, pancreas, and a reference marker with aknown T1 value. Contrast index (CI, SItissue/SImarker) andcontrast-to-noise ratio (CNR, [SItissue/SImarker]/SDbackground)were calculated, and percentage CI increase and CNR in thepostcontrast images were used for the best sequence evalu-ation. Regarding CI, the TFE sequence with a TR/TE/flipangle of 15 msec/4.6 msec/20o and inversion time of 300msec had the largest pre- to postcontrast percentage in-crease. The FFE sequence with a TR/TE/flip angle of 140msec/4.6 msec/90o had the highest postcontrast CNR andis considered to be the optimal sequence for mangafodipirtrisodium-enhanced MR imaging of the liver and pancre-as. J. Magn. Reson. Imaging 1999;9:280–284

r 1999 Wiley-Liss, Inc.

Index terms: contrast media; manganese; mangafodipir triso-dium (MnDPDP); MRI; liver; pancreas

Abbreviations: ANOVA 5 analysis of variance, CI 5 contrast index, CNR 5

contrast-to-noise ratio, FFE 5 fast field gradient-echo, FOV 5 field of view, Gd 5

gadolinium, MnDPDP 5 mangafodipir trisodium, NEX 5 number of acquisitions,ROI 5 region of interest, SE 5 spin echo, SI 5 signal intensity, TFE 5 turbo fieldgradient-echo, TI 5 inversion time.

MANGAFODIPIR TRISODIUM (MnDPDP; Teslascan, Ny-comed Amersham) is a paramagnetic contrast agentoriginally designed for liver imaging. As the agent istaken up by the functioning hepatocytes, increasedcontrast between normal liver parenchyma and differ-ent kinds of hepatic tumors can be achieved (1). Since ahepatocellular lesion and a group of liver metastasesfrom endocrine tumors may show uptake of the agent at

various degrees (2–4), some differential diagnostic infor-mation on tumor origin can also be obtained. As well asin the hepatobiliary system, enhancement in otherabdominal parenchymatous tissues was also observedafter administration of the contrast agent (5,6); this wasmost pronounced in the liver and lasted longest in thepancreas (7).

In clinical trials the same sequence parameters aregenerally used in the pre- and postcontrast Tl-weightedsequences. These parameters are chosen according tothe normal hospital routine, which means that they areoptimized for native imaging without contrast, or withgadolinium (Gd)-based contrast agents. In our experi-ence from several clinical trials with Mangafodipir triso-dium, different sequence types and different param-eters for the same type of sequence may result in a greatvariety of enhancement and tissue contrast in theimages. There is a need for an optimized postcontrastimaging protocol to allow full utilization of the effect ofMangafodipir trisodium.

This study was performed with the aim of finding thesequence parameters in our magnetic resonance (MR)system that can give the highest contrast betweennormal tissue and lesions of the liver and pancreas afteradministration of Mangafodipir trisodium. The resultscould then provide guidance to other users of theproduct, to achieve the most effective examination.

MATERIALS AND METHODS

Subjects and Contrast Agent

Six healthy Caucasian male subjects, with a mean ageof 24.3 years (range 22–28 years), mean height of 183.7cm (range 175–191 cm), and mean weight of 76.7 kg(range 68–83 kg) were recruited for this study. One ofthe six subjects was assigned to a ‘‘pilot study’’ that wasperformed prior to the main study on the other fivesubjects. To obtain more information regarding whichsequences and sequence parameters needed to be inves-tigated in depth, a variety of the possible sequences andparameter combinations were tested pre- and postcon-trast in the pilot subject. Only the sequences that wereused later in the main study were included in the overalldata analysis. The trial protocol was approved by theEthics Committee of Uppsala University Hospital. Writ-

1Department of Diagnostic Radiology, Uppsala University Hospital,S-751 85 Uppsala, Sweden.2Nycomed Imaging AS, N-0401 Oslo, Norway.Address reprint requests to: C.W., Department of Diagnostic Radiology,University Hospital, Uppsala University, S-751 85 Uppsala, Sweden.E-mail: Chen.Wang@radiol.uu.seReceived April 3, 1998; Accepted September 16, 1998.

JOURNAL OF MAGNETIC RESONANCE IMAGING 9:280–284 (1999)

r 1999 Wiley-Liss, Inc. 280

ten informed consent was obtained from each subjectprior to entry into the study.

Mangafodipir trisodium, at a concentration of 0.01mmol/mL (0.01 M; 10 µmol/mL), was supplied in 50mL glass vials as a ready-for-use solution. The agent, ata dose of 5 µmol/kg body weight, was administered tothe subjects as a slow infusion at a rate of 2–3 mL/min,in the same way as described previously (8). The meanvolume infused was 38.3 mL, with a mean duration ofinfusion of 15.2 minutes.

MRI

For the MR examinations a Philips ACS/NT systemoperating at 1.5 T was used. Before imaging, a referencetube (marker) containing a Gd complex (Gd-polymer,Nycomed Amersham) with a known T1 relaxation timeof 1000 msec, simulating the T1 of a tumor, wasattached to the volunteer’s side such that it came intothe field of view. The marker was at room temperaturebefore it was placed on the subject. Abdominal MRimaging was performed with a transaxial slice thicknessof 6 mm. There was no requirement that any abdominalorgans should be covered completely. The followingtypes of pulse sequence were investigated: a) T1-weighted turbo field (gradient)-echo (TFE); b) T1-weighted fast field (gradient)-echo (FFE), including bothfat-water in-phase and out-of-phase sequences; and c)T1-weighted spin-echo (SE). See Table 1 for details ofthe parameters.

Postcontrast imaging was started approximately 15minutes after the end of contrast administration. TheMR sequences were repeated using the same sequenceparameters as had been applied before contrast. Toavoid time-related differences in contrast enhance-ment, a 3 3 3 Latin square design that varied thesequence order of TFE, FFE, and SE was used (Table 2).

Measurements

Signal intensity (SI) measurements were made in re-gions of interest (ROIs), in a slice where the targettissues were present in the slices above and below toavoid partial volume effects. The ROIs were defined andmeasured in the liver parenchyma, the pancreatic paren-chyma, and the reference marker. The image back-ground signal and its standard deviation from the sameslice were also measured in a corner of the imagewithout any noticeable motion artifact.

The main efficacy variables included the contrastindex (CI) and the contrast-to-noise ratio (CNR). CI wasan index ratio defined as the SI of the tissue of interest,ie, the liver orpancreas,dividedby theaverageSImeasuredin the reference marker (CI 5 SItissue/SImarker). This indexwas calculated both before and after administration of thecontrast agent, and contrast enhancement was calculatedas the percentage increase in the postcontrast index overthe precontrast index (% index enhancement 5 100 3(CIpostcontrast-CIprecontrast)/CIprecontrast). The CNR was de-rived from the equation: (Sitissue - SImarker)/SDbackground

and was calculated for postcontrast images only.For each sequence type, the maximum CI enhance-

ment and the highest postcontrast CNR were found withdifferent MR parameters, ie, TR, TE, TI (inversion time),and flip angle. The best sequence in each sequence typewas then compared with those of the other sequencetypes, to find the best sequence of all.

Statistics

For testing differences among the sequences, a mixedanalysis of variance (ANOVA) model was used. Sequencesand time period were fixed effects, while subject was arandom effect, and mean squared errors were made usingthese assumptions. A two-sample t-test was appliedwhen testing pairwise between two sequences. All testswere two-sided with a significant level of 5% (P , 0.05).

RESULTS

Contrast Index

The mean CI values in the pancreas and liver for thedifferent sequences applied are shown in Tables 3–5.

In the TFE sequences, CI increased with decreasing TIin both pancreas and liver imaging. In all six subjects, aTI of 300 msec gave the highest percent contrast en-

Table 2Order of Sequence in the Six Subjects

SubjectPeriod

1 2 3

1 TFE FFE SE2 FFE SE TFE3 SE TFE FFE4 TFE SE FFE5 FFE TFE SE6 SE FFE TFE

Table 3Percentage Increase in Contrast Index for TFE Images[Mean 6 (SD)]*

TI(msec)

Pancreas Liver

TR/TE/flip angle, 15msec/4.6 msec/20°

300500700

110.0 (28.9)61.5 (15.7)47.0 (15.4)

6 111.6 (18.2)71.4 (10.7)52.8 (7.0)

6

900 38.8 (11.2) 43.4 (7.5)

*Numbers in bold 5 numbers of subjects in whom the highestcontrast index was found with the given TI.

Table 1MRI Pulse Sequence Parameters

Sequenceparameter

FFETFE SE

In-phase Out-of-phase

TR (msec) 100/140/180 100/140/180 15 200/400/600TE (msec) 4.6 6.9 4.6 15Flip angle 50°/70°/90° 90° 20° 90°TI (msec) — — 300/500/700/900 —Slice (mm) 6 6 6 6FOV (mm) 375 375 375 375Matrix 128 3 256 128 3 256 179 3 256 154 3 256NEX 2 2 4 2

Sequence Optimization in Liver and Pancreas MRI 281

hancement (Table 3). In the FFE sequences, the largestpercent increase in CI in the pancreas and liver wasobtained by using the parameter combination TR 100msec and flip angle 90o. The in-phase FFE with theseparameters was the best sequence for pancreas imagingin 2/6 subjects, and for liver imaging in 4/6 subjects.The out-of-phase FFE was the best sequence in 3/6subjects for the pancreas, but only in one subject for theliver (Table 4). In the SE sequences, CI increased withdecreasing TR value. The sequence with a TR of 200msec was the best in 4/6 subjects for the pancreas, andin all six subjects for the liver (Table 5).

A statistically significant difference between thesesequences was found on tests for overall differences intheir CI values (P 5 0.0013 for pancreas and 0.0019 forliver). When testing pairwise between the sequences, asignificantly greater percent increase in CI was foundfor the TFE sequence compared with the other twosequences (TFE vs. FFE: P 5 0.0006 for pancreas and0.0019 for liver; TFE vs. SE: P 5 0.0016 for pancreasand 0.0011 for liver). There was no difference betweenthe FFE and SE sequences (P 5 0.471 for pancreas and0.675 for liver).

Contrast-to-Noise Ratio

The CNR values for the pancreas and liver are shown inTables 6–8.

In the TFE sequences, a TI of 500 msec yielded thebest result. This was the best sequence in 4/6 subjectsfor both organs (Table 6). In the FFE sequences, CNR

increased with increasing flip angle and had the highestvalue at 90°. With different TR values but the same flipangle, the highest increase was found with a TR of 140msec. The in-phase FFE was seen as the best sequencein 5/6 subjects for both liver and pancreas (Table 7). Inthe SE sequences CNR increased with decreasing TRvalues. Regarding the pancreas, a TR of 200 msec gavethe highest CNR in all six subjects, and this was the bestsequence for the liver in 5/6 subjects (Table 8).

Statistically significant differences were found be-tween the best sequences of each sequence type whentests were made for overall differences in the postcon-trast CNR for the liver, but not for the pancreas (P 50.0001 and P 5 0.145, respectively). Pairwise testingbetween the sequences showed a significantly greaterpostcontrast CNR in the FFE sequence than in the othertwo sequences for the liver, but not for the pancreas(FFE vs. TFE: P 5 0.0001 for liver and 0.083 forpancreas; FFE vs. SE: P 5 0.0001 for liver and 0.098 forpancreas). In addition, for the liver the TFE sequenceshowed a significantly greater CNR compared with theSE sequence (P 5 0.016), but this was not found for thepancreas (P 5 0.917).

DISCUSSION

Various sequences have been used in previous studieswith Mangafodipir trisodium, and usually T1-weightedgradient-echo sequences have been considered to be

Table 4Percentage Increase in Contrast Index in the Pancreas and Liverfor FFE Images [Mean 6 (SD)]*

TR(ms)

Flip angle

50° 70° 90°

a. PancreasIn-phase FFE, TE

4.6 msec100140180

33.2 (9.5)28.1 (11.9)16.4 (6.2)

47.4 (7.2)34.3 (8.9)31.7 (8.2)

51.6 (11.3)50.1 (7.6)41.5 (7.3)

2

Out-of-phase FFE,TE 6.9 msec

100140180

———

———

50.8 (11.6)48.8 (10.9)39.6 (8.7)

31

b. LiverIn-phase FFE, TE

4.6 msec100140180

39.8 (4.2)34.0 (3.1)25.6 (2.6)

61.6 (3.6)47.0 (6.0)42.1 (3.5)

73.9 (6.8)62.2 (6.6)53.1 (5.4)

41

Out-of-phase FFE,TE 6.9 msec

100140180

———

———

67.4 (6.4)60.3 (8.5)50.6 (4.5)

1

*Numbers in bold 5 numbers of subjects in whom the highestcontrast index was found with the given TR and flip angle.

Table 5Percentage Increase in Contrast Index for SE Images[Mean 6 (SD)]*

TR(msec)

Pancreas Liver

SE sequence, TE 15msec

200400600

59.8 (25.3)53.2 (18.5)36.6 (9.8)

42

70.3 (13.5)56.7 (9.7)39.2 (10.0)

6

*Numbers in bold 5 numbers of subjects in whom the highestcontrast index was found with the given TR.

Table 6Postcontrast Contrast-to-Noise Ratio for TFE Images[Mean 6 (SD)]*

TI(msec)

Pancreas Liver

TR/TE/flip angle, 15msec/4.6 msec/20°

300500700

8.4 (1.8)8.4 (2.2)7.4 (2.4)

24

9.2 (0.8)9.9 (1.4)8.3 (1.2)

24

900 5.9 (1.9) 6.6 (1.2)

*Numbers in bold 5 numbers of subjects in whom the highestcontrast index was found with the given TI.

Table 7Postcontrast Contrast-to-Noise Ratio in the Pancreas and Liverfor FFE Images [Mean 6 (SD)]*

TR(msec)

Flip angle

50° 70° 90°

a. PancreasIn-phase FFE, TE

4.6 msec100140180

4.3 (4.0)1.6 (5.7)

23.8 (5.7)

8.1 (2.9)5.9 (4.5)3.6 (3.5)

8.4 (3.8)10.7 (4.2)

6.6 (2.5)5

Out-of-phase FFE,TE 6.9 msec

100140180

———

———

5.4 (4.6)8.3 (6.1)7.1 (4.0)

1

b. LiverIn-phase FFE, TE

4.6 msec100140180

5.4 (2.9)3.2 (3.0)

21.5 (4.0)

11.7 (2.9)8.1 (2.3)6.7 (3.0)

12.8 (2.3)13.9 (2.8)10.0 (2.6)

15

Out-of-phase FFE,TE 6.9 msec

100140180

———

———

7.0 (7.5)9.3 (4.9)9.3 (1.7)

*Numbers in bold 5 numbers of subjects in whom the highestcontrast index was found with the given TR and flip angle.

282 Wang et al.

better in showing differences between lesion and liverthan T1-weighted spin-echo sequences (8). However,many parameter combinations are possible, such asdifferent flip angles and TR values, in an ordinaryT1-weighted gradient-echo sequence. With the moderndevelopment of MR technology there are also somesequence variations of the conventional gradient-echosequence, eg, FFE and TFE with or without fat suppres-sion, which utilize considerably different MR propertiesof the tissues of interest for image contrast. It is notfeasible to investigate every possible sequence variablethat produces T1-weighted tissue contrast. The spec-trum of the sequences tested in this study, however,covers most of the commonly used T1-weighted se-quences, including T1-SE, T1-FFE, and T1-TFE. In eachsequence type, different parameter combinations werechosen to cover a wider range than that of the clinicalroutine, for example, TR of 100–180 msec with a flipangle of 50–90° in the FFE sequence settings.

Generally, the tissue contrast in an FFE image isdependent on both the T1 and T2* of the tissues. As withSE, a short TR is needed for T1-weighted FFE images(TRs of 100–300 msec are normally used). Use of evenshorter TRs (,100 msec) is possible for producingheavier T1-weighted contrasts, at the cost, however, ofmarked drop in the signal-to-noise ratio. Large flipangles are usually required for more T1 weighting. AnFFE sequence with a TR , 100 msec was tested on thepilot subject, but was abandoned because of the grainyappearance of the images.

The TFE technique allows gradient-echo imaging withvery short TR and TE, the aim of which is high imagequality within a short scan time. T1-weighted tissuecontrast can be obtained by using pre-pulses, eg, a 180o

inversion pulse as used in the inversion recovery tech-nique. Depending on the T1 relaxation time, the tissuecontrast in a TFE sequence can be manipulated byselecting TI—the timing of the pre-pulse. The shorterthe TI, the stronger the T1 weighting. If too short a TI isused (,300 msec), the net magnetism and the MRsignal of some abdominal organs, such as the spleenand kidneys, will be nullified on the way of recovery fromminus to equilibrium. As these organs appear dark, theoverall image SI is reduced, resulting in a loss ofsignal-to-noise ratio. This is the reason why a minimumTI of 300 msec was applied in this study and it is provedthat, due to increased background noise, image qualityhas started to deteriorate at TI of 500 msec (Fig. 1).

When analyzing the CI, the TFE sequence with TR/TE/flip angle of 15 msec/4.6 msec/20o and a TI of 300 msecgave the highest enhancement rate after administrationof Mangafodipir trisodium. CI is used as a measure ofabsolute contrast enhancement in tissues without con-

sideration of image quality and contrast between differ-ent tissues. In a clinical situation the interest is focusedon the largest difference between normal tissue andpathological lesions, and the purpose of using a con-trast medium is to increase this difference so that thelesions can be easily detected. Hence, this TFE se-quence is not necessarily the most superior even thoughit yielded the highest percent contrast enhancement.The CNR should therefore be more clinically relevant.

The reference marker with a T1 relaxation time of1000 msec was used both for normalizing the SI mea-surements and for mimicking a solid non-enhancingtumorous lesion, which normally has a lower signalintensity than normal liver and pancreas in T1-weightedimages. The contrast between normal pancreas or liverparenchyma and the marker is expressed as the CNR inour study, which may not be valid for contrast-enhancing tumors. The results showed that the in-phase FFE sequence with TR of 140 msec, TE of 4.6msec, and flip angle of 90° had the best result regardingthe CNR in both liver and pancreas, with the highestvalues in all postcontrast images. Using CNR as theprimary criteria for efficacy assessment, this sequenceis superior to the best SE and best TFE sequences. Thereason why the FFE images surpass the TFE imagescould be that the latter are noisier, since this type ofsequence employs very short TR and a very small flipangle, resulting in a poor signal-to-noise ratio with an

Table 8Postcontrast Contrast-to-Noise Ratio for SE Images [Mean 6 (SD)]*

TR(msec)

Pancreas Liver

SE sequence,TE 15 msec

200400600

8.5 (4.1)3.8 (2.2)

22.2 (3.7)

6 8.1 (1.8)4.1 (3.7)

22.3 (3.2)

51

*Numbers in bold 5 numbers of subjects in whom the highestcontrast index was found with the given TR.

Figure 1. Mangafodipir trisodium-enhanced T1-TFE (TR/TE/flip angle 15 msec/4.6 msec/20o) images. A higher enhance-ment rate was observed when using a 300 msec inversion time(a), at a cost of increased background noise and reduced imagequality, compared with the 500 msec inversion time image (b).

Sequence Optimization in Liver and Pancreas MRI 283

ordinary two-dimensional imaging method. If a three-dimensional acquisition is used in conjunction with theTFE technique, the signal-to-noise ratio can improvesignificantly as a thicker tissue volume is excited. Thismay, however, require a longer imaging time than ispermitted for a breath-hold scan, and respiratory trig-gering may have to be used.

A fat-saturated sequence was originally planned, butnot carried out because of a machine hardware prob-lem. Instead, a fat-water out-of-phase T1-FFE sequencewas conducted. In principle the out-of-phase imagingaccentuates the fat-water chemical shift and can there-fore suppress the signal of fat tissue in fat-containingorgans and improve the delineation of the organs sur-rounded by fat. Fat-suppression techniques are oftensuggested for imaging of the pancreas (9–11), as theycan suppress the signal from the retroperitoneal fataround the pancreas and provide a better delineation ofthe organ. However, since normal pancreatic tissue alsocontains fat (less in the liver), a decreased signal of thepancreas itself in the out-of-phase images will be antici-pated compared with the fat-water in-phase images.Normally, both pancreas and liver tumors have a lowerSI than the normal parenchyma of the organs onT1-weighted images, and they do not contain fat. Hence,the tumor-parenchymal contrast will be lower in theout-of-phase images than in the in-phase images. Thesame theory regarding the out-of-phase images alsoholds for the fat-saturated images. The out-of-phaseFFE sequences in our study always gave a lower CNRthan the in-phase FFE sequence with the same TR andflip angle, and for both liver and pancreas. In ourexperience, although the pancreas may appear‘‘brighter’’ in an out-of-phase (or fat-suppressed) image,the tissue contrast between normal tissue and lesionsmay actually decrease as the overall signal of the organis reduced by the fat signal subtraction, while the signalof the lesions will remain unchanged. The bright appear-ance of the pancreas is simply due to the scaling of theoverall image signal (Fig. 2).

A group of young subjects was included in our studyto minimize possible variation of MR signal with aging.Patients with liver or pancreatic tumors are usuallyolder, having more fat deposit in these organs. Relax-ation time and contrast uptake of the liver and pancre-atic tissues in such patients may differ from that ofyoung healthy people. Further studies are needed totest whether the data acquired from the healthy volun-teers are valid for patients. A drawback of our study isthat fat-saturated imaging was not performed and in-stead, a fat-water out-of-phase imaging was conducted.Although the two methods are similar in some aspects,they have different mechanisms for fat suppression, andMangafodipir trisodium-enhanced MRI with fat satura-tion imaging should be investigated in future studies.

In summary, the in-phase FFE sequence with TR of140 msec, TE of 4.6 msec, and a flip angle of 90° yieldedthe best result with respect to CNR and is considered tobe the optimal sequence for mangafodipir trisodium-enhanced MR imaging of the liver and pancreas. Whenthe pancreas is examined, an out-of-phase, or fat-saturation sequence may also be applied, to improveorgan delineation. Since these results are based on

findings in healthy volunteers, minor modification mightbe needed for the optimal clinical imaging protocol.

REFERENCES1. Hamm B, Vogl TJ, Branding G, et al. Focal liver lesions: MR imaging

with Mn-DPDP—initial clinical results in 40 patients. Radiology1992;182:167–174.

2. Rofsky NM, Weinreb JC, Bernardino ME, Young SW, Lee JK, NozME. Hepatocellular tumors: characterization with Mn-DPDP-enhanced MR imaging. Radiology 1993;188:53–59.

3. Vogl TJ, Hamm B, Schnell B, et al. Mn-DPDP enhancement patternsof hepatocellular lesions on MR images. J Magn Reson Imaging1993;3:51–58.

4. Liou J, Lee JK, Borrello JA, Brown JJ. Differentiation of hepatomasfrom nonhepatomatous masses: use of MnDPDP-enhanced MRimages. Magn Reson Imaging 1994;12:71–79.

5. Gehl HB, Vorwerk D, Klose KC, Gunther RW. Pancreatic enhancementafter low-dose infusion of Mn-DPDP. Radiology 1991;180:337–339.

6. Sidhu MK, Muller HH, Rubin DL, Youny SW. Gastric, pancreatic,and renal enhancement with the MR imaging contrast agent Mn-DPDP. J Magn Reson Imaging 1992;2:103–104.

7. Wang C, Gordon PB, Hustvedt SO, et al. MR imaging properties andpharmacokinetics of MnDPDP in healthy volunteers. Acta Radiol1997;38:665–676.

8. Wang C, Ahlstrom H, Ekholm S, et al. Diagnostic efficacy ofMnDPDP in MR imaging of the liver. Acta Radiol 1997;38:643–649.

9. Semelka RC, Ascher SM. MR imaging of the pancreas. Radiology1993;188:593–602.

10. Brown ED, Semelka RC. Magnetic resonance imaging of the spleenand pancreas. Top Magn Reson Imaging 1995;7:82–89.

11. Mitchell DG, Winston CB, Outwater EK, Ehrlich SM. Delineation ofpancreas with MR imaging: multiobserver comparison of five pulsesequences. J Magn Reson Imaging 1995;5:193–199.

Figure 2. Mangafodipir trisodium-enhanced fat-water in-phase FFE (TR/TE/flip angle 140 msec/4.6 msec/90o) image(a), and fat-water out-of-phase FFE (TR/TE/flip angle 140msec/6.9 msec/90o) image (b). The liver and pancreas appearbrighter and better outlined on the out-of-phase image be-cause of fat signal suppression.

284 Wang et al.