Matthias Gordian

Taupitz, Branding,

MD MD

Hamm, Karl-Jurgen

#{149} Bernd #{149}

Multisection MR Imaging

terms:

plasms, nance

Liver, MR, 761.1214 I Liver neo761.31, 761.32, 761.33 #{149} Magnetic reso(MR), pulse sequences I Magnetic reso-

nance

(MR), rapid

Radiology

I

1992;

imaging

Steglitz,

of Radiology,

Klini-

Freie Universit#{228}t Berlin, Hinden30, 1000 Berlin 45, Germany (MT.,

burgdamm B.H., AS., GB., K.J.W.) and Siemens AG, Medical Engineering Group, Erlangen, Germany (M.D.). Received September 9, 1991; revision requested October 3; revision received November 14; accepted November 18. MT. supported by the

Institut

Address C

RSNA,

f#{252}r Diagnostikforschung, requests to MT. 1992

reprint

Speidel,

MD

Michael

#{149}

Deimling,

PhD

number of different techniques have been developed to overcome the problems associated with motion artifacts and long examination

multisection pulse risk of overlooking

times

inspiratory In this

in magnetic

resonance

repeat

(MR)

imaging of the liver. In Ti-weighted MR imaging of the abdomen at intermediate field strengths (about 0.5 T), motion artifacts can be reduced efficiently with multiple data averaging (1-3). However, at high field strengths (about

1.5 T), the

(TR)

multiple-acquisition

does

not

yield

Ti-weighted men (4-6). A variety have been conventional sate

for

short

repetition

time

method

satisfactory

results

MR imaging

in

of the abdo-

motion-in-

duced artifacts. These strategies indude respiratory ordered phase encoding (7), respiratory gating (8), gradient moment nulling (9), fat suppression, sion-recovery troscopy

short

inversion sequences (li,i2). Though

techniques artifacts, contours

reduce

time inver(10), or specthese

motion-induced

motion-induced blurring remains a problem, and

of re-

duction of peristaltic artifacts is insufficient. In addition, the long acquisition times continue to be a disadvantage of conventional techniques. Conversely,

fast

imaging

tech-

niques with suspended respiration not only prevent motion-induced

Berlin.

arti-

echo (i3-i9), spin-echo (SE) (20), or echo-planar imaging (21,22) have been described for MR imaging of the liver. One major disadvantage of fast imaging with single-section or limited

Abbreviations: difference-to-noise echo time, TR

FLASH =

ratio, repetition

=

SE

fast =

low-angle spin echo,

because

staggered

depth. study, we evaluated

section breath-hold fast shot (FLASH) technique

the acquisition ages

ac-

of differences

in

a multi-

low-angle that enables

of i9 Ti-weighted

of the

entire

liver

within

ima single

breath hold. Initially, we tested this technique in volunteers for contrast optimization. Thereafter, we applied the optimized technique in a group of ho patients with various liver lesions to investigate its ability to render high-quality images with good livertumor contrast. The study protocol

of specific approaches suggested for use with sequences to cornpen-

or to suppress

sequences is the liver lesions in

craniocaudally

quisitions

facts and blurring but also dramatically reduce imaging time. Different procedures of suspended respiration used in combination with gradient-

183:73-79

From the Department

kum

#{149} Andrea MD

FLASH: Method for Breath-Hold ofthe Entire Liver’

One hundred ten patients with various focal liver lesions were imaged with a multisection fast low-angle shot (FLASH) gradient-echo sequence with an echo time of 4.6 msec. This sequence enabled the acquisilion of 19 Ti-weighted magnetic resonance (MR) images of the liver within a single 26-second breath hold. Patients were also examined with standard Ti- and T2-weighted spin-echo (SE) sequences. The multisection FLASH sequence provided significantly higher (P < .01) liverspleen contrast, liver-spleen signaldifference-to-noise ratio (SD/N), liver-tumor contrast, and liver-tumor SD/N than the Ti-weighted SE sequence but lower values than the T2-weighted SE sequence. Motion artifacts were reduced with the multisection FLASH sequence compared with both SE sequences (P < .01). The overall image quality of the multisection FLASH images was similar to that of the Ti-weighted SE images and superior to that of T2-weighted SE images. The most important characteristics of the multisection FLASH technique in MR imaging of the liver are the high Ti contrast, the prevention of motion artifacts, and a dramatic reduction in imaging time. Index

MD Wolf,

included section ventional sequences

the comparison of the multiFLASH sequence with conTi- and T2-weighted SE in all patients.

SUBJECTS

AND

METHODS

Subjects Seven healthy volunteers (four three women; mean age, 28 years

were

examined

for pulse

men, 3)

±

sequence

optimi-

zation. Then, ii0 consecutive patients who had undergone MR imaging of the abdomen for suspected focal liver lesions

were examined according to the protocol described below. Focal liver lesions were detected in i02 of these patients. Forty-two patients had benign liver tumors (hemangiomas in 25, focal nodular hyperplasias in five, and cysts in 12). The diagnoses were confirmed at histologic examination in four cases each of heman-

giomas

and focal nodular

hyperplasias,

and in all other cases by means of the typical appearance with other imaging procedures (hemangiomas had high attenuation fill-in at dynamic contrast material-enhanced computed tomography [CT] and

shot, RASE = rapid SI = signal intensity,

acquisition spin echo, S/N = signal-to-noise

SD/N ratio,

=

TE

signal=

time.

73

positive

blood-pool

imaging; early

focal

scans

nodular

arterial

biliary by

hyperplasias

enhancement

contrast-enhanced

Sixty

imaging)

patients

had

malignant

in 49 patients,

cellular

carcinomas

in eight,

noses

were

stration or detection presence tumors

and

liver

0

0

0.2

zL

cholan-

.v

0.1

The diag-

4

#{149} TR

L

o

6

z 0 V)

=

162

2

94msec

TR=

at histologic

in 41 patients

of tumor

S

hepato-

in three).

confirmed

examination

Liri[L rAKLL

0.3

at fol-

(metastases

carcinomas

and

8

Li-U

hepato-

findings

tumors

giocellular

0.4

positive

at radionuclide of unchanged

low-up.

had

at dynamic

CT and

scans

means

0.5

at radionuclide

and

of multiple of histologically

C

by demon-

progression

20

40

60

at follow-up

lesions proved

in the primary

80

flip

angle

100

120

20

60

80

flip

a.

angle

100

120

[deg]

b.

Figure

in 19.

40

[deg]

I.

Graphs

technique 4.6 msec).

show

pulse

sequence

performance

in seven healthy volunteers deg = degree. (a) Liver-spleen

of the multisection

as a function contrast.

of flip angle (b) Liver-spleen

for

breath-hold

two SD/N.

different

FLASH TRs

(TE

was

MR Imaging MR imaging mercially

was

performed

available

1.5-T

with

a com-

Table 1 Results of Quantitative

superconducting

magnet (Magnetom GBS II; Siemens, Erlangen, Germany). All images were acquired in the transaxia! plane with a section thickness of 8 mm and an intersection gap

of 1.6 mm.

ing

was

Rapid

multisection

performed

with

an

Liver

was

long

compared

TR

values

T2 of the liver

despite

Pulse gate

the

sequence

the

use

of large

provided

with

FLASH technique TRs of 94 msec

100.

between We

a TE

of 4.6

and

a 40

angle

x 40-cm

could

was

used

with was not

ware limitations. Mean signal spleen

(L,ver)’ (Background)

artifacts

vascular standard

nature)

were

Hemangioma

( 51Liscr 74

Spleen)”S’Background

Radiology

#{149}

0.21

±

0.09

0.07

±

0.13

0.16

±

0.09

0.14

±

0.12

6.5

±

3.9

1.9

±

0.1

-7.9

5.2

±

3.5

3.3

±

2.7

1.3

±

0.4

2.9

±

1.2

-6.5 2.6

plus or minus standard

than

field of view in the

fre-

35)

=

(n

Focal nodular

6.4

-0.39

±

0.17

-0.32

±

0.17

±

4.6

± ±

4.9 1.0

deviations.

Contrast

and SD/N

FLASH

Ti-weighted

T2-weighted

=

Hepatocellular (n = 10) Note-Values

and

presented

0.20

±

0.08

±

0.30 0.16

±

0.15

-0.21

±

±

0.04 0.10 0.08

±

0.25 0.15 0.15

±

0.12

-0.16

±

0.18

4.3

±

3.0

-10.7

±

6.3

1.7 5.1 3.2

±

1.1

-3.6

±

4.2

±

2.8

-8.6

±

6.1

±

2.1

-5.9

±

3.9

±

2.6

-2.4

±

2.6

±

±

0.03 0.13

0.06 0.27

±

±

0.07

0.14

±

0.12

±

3.9

2.5

±

9.8

±

4.7

±

1.3 4.2 2.4

0.5

±

3.9

7.7

-0.04

185)

±

-0.46 -0.30

±

carcinoma -0.8

are only given for lesion types that were examined

as means

-0.44

0.19

-0.01

35)

=

0.11

0.07

hyperplasia

(n = 5) Cyst (n = 18) Metastasis (n

including

determined

of Liver-Tumor

Multisection

(n

SD/N Hemangioma

data acof hard-

respiratory

Evaluation

Focal nodular hyperplasia (n = 5) Cyst (n = 18) Metastasis (n = 185) Hepatocellular carcinoma (n = 10)

flip

plus

or minus

standard

in at least five patients.

Data

are

deviations.

with

measurements

in three different sections from each set of images (upper, middle, and lower parts of spleen and liver). Contrast and signal-difference-to-noise ratio (SD/N) between liver and spleen were calculated to determine the soft-tissue calculated as (SILVer SIsP1,,,j, and SD/N

±

15.4

Contrast

(SIs) of liver

region-of-interest

12.3

Parameter a

to more

the high-speed possible because

of both

±

Sequence

Oversamp!ing direction in

direction,

37.1

as means

Table 2 Results of Quantitative

(SISPI=n)’ and background (outside of the abdomen in the

phase-encoding ghosting

The

a!iasing

intensities

are presented

of

excitation, resolution),

A smaller

to avoid

9.8

multi-

radio-frequency

quency-encoding direction. in the frequency-encoding combination quisition

the

of view.

of excessive

deposition.

not

Note-Data

a band-

not be increased

120#{176} because

energy

field

±

NSYdNSTAT

msec (19 time) were ifip angle was

width of 390 Hz/pixel, one 256 x 128 matrix (anisotropic

T2-weighted

22.1

Liver-spleen Liver-tumor (alllesions)

162

msec,

Ti-weighted

angles. investi-

30#{176} and 120#{176} in steps

used

S/N

FLASH

SD/N

in the seven vol(ii sections, 16-

second imaging time) and sections, 26- second imaging used. At both settings, the

varied

flip

optimization-To

contrast

section unteers,

Multisection

Contrast Liver-spleen Liver-tumor (alllesions)

at 1.5 T ( 58 msec [Taupitz M, Hamm B, Speidel A, et al, unpublished data, 1989]) are used with the multisection FLASH sequence. Thus, complete decay of transverse coherence could be assumed at the end of each cycle,

with

only

Parameter

unspoiled

with an echo spoiler gradient

because

Sequence

imag-

FLASH-type (23) sequence time (TE) of 4.6 msec. The omitted

Evaluation

contrast.

Contrast

was

+

S’SpleenW(Liver

was

calculated

as

In four volunteers, sections were acquired in three different orders to determine the effect of possible intersection cross-excitation on the liver signal-to-noise ratio (S/N) (ie, excitation of the 1st, 2nd, 3rd, . . . 19th sections vs that of the 1st, 3rd, .

.

.

16th

2nd,

4th,

sections).

.

. . 18th

and

1st, 5th,

9th,

...

Imaging tients

protocol

underwent

(a) multisection and TE bandwidth

in patients-All the

FLASH

following

110 pasequences:

(TR of 162 msec

of 4.6 msec [162/4.6], of 390 Hz/pixel,

80#{176} flip angle, one signal av-

erage, 19 sections per breath hold, 128 matrix [anisotropic resolution], 40-cm field of view, and a 26-second

256 x 40 x imag-

April

1992

ing time); bandwidth tions,

(b) Ti-weighted

SE (500/15, four excita-

of 130 Hz/pixel,

19 sections,

256

x 256 matrix,

50 x

50-cm field of view, and an 8.35-minute imaging time [spatial presaturation was not used to keep the multisection capacity high at a TR of 500 msec]); and (c) doubleecho SE with proton density-weighted first echo and T2-weighted second echo (2,000/15, 90, bandwidth of 130 Hz/pixel and reduced bandwidth of 65 Hz/pixel for sampling S/N

of the second two excitations,

[24],

256 x 128 matrix 50 x 50-cm field

tion [25,26] artifacts,

echo

to increase

17 sections,

[anisotropic of view,

to reduce

10.2-minute

resolution],

spatial

presatura-

vascular

pulsation

imaging

time,

and

first-order gradient moment nulling [9] for the second echo). For the SE sequences, the field of view was set to 50 x 50 cm to obtain maxima! S/N. An abdominal

belt

was

used

branches

or

evaluation-Lesions

motion

artifacts.

SD/N

is used

as a

measure for imaging system performance and correlates with the diagnostic value different MR imaging techniques (4,2729). In addition,

the ratio

between

of

system-

atic and statistical noise (Nsys/NSTAT) was calculated as a measure for the intensity of motion artifacts. Contrast, SD/N, and SI of motion artifor the various

compared

pulse

by means

matched-pairs

sequences

were

test

evaluation-All

183

visi-

presented sentative

with 10 cases serving examples for different

as repreratings.

ous criteria were ing the k*2fields

tested for differences usX2 test (30), with a signif-

icance

Multisection FLASH: method for breath-hold MR imaging of the entire liver.

One hundred ten patients with various focal liver lesions were imaged with a multisection fast low-angle shot (FLASH) gradient-echo sequence with an e...
2MB Sizes 0 Downloads 0 Views