Rainer Erlemann, Ernst Rummeny,

MD MD

#{149} Pierre #{149} Ullnich

Vassallo, MD St#{246}ber #{149} Peter

Musculoskeletal Shot MR Imaging

#{149} George

E. Peters,

Bongartz, MD

Neoplasms: with and

I

MD

#{149} Heribert

Fast without

In 121 patients, image contrast and contrast-to-noise ratios (C/Ns) obtamed with gadolinium diethylenetriaminepentaacetic acid (DTPA)enhanced and nonenhanced fast low-angle shot (FLASH) sequences were compared with those achieved with spin-echo (SE) sequences. Among FLASH sequences, contrast between neoplasms and muscle was sufficient with a flip angle of 90#{176} following administration of GdDTPA but was 61% lower than that with the T2-weighted SE sequence. High contrast levels were obtained between tumors and bone marrow or fat in sclerotic, calcified, and fibrotic lesions with the use of a flip angle of 90#{176} and in lytic lesions with a flip angle of 10#{176}, reaching 66% of the contrast level obtained with the Ti-weighted SE sequence. C/N between tumor and surrounding tissue was always significantly lower with FLASH sequences than with the ideal SE sequence usually used for tumor delineation. Thus, a replacement of the T2-weighted SE sequences by FLASH sequences cannot be recommended. A replacement of the Ti-weighted SE sequences by FLASH sequences seems possible but does not significantly reduce examination time.

magnetic resonance (MR) irnaging, two-dimensional Fourier transformed spin-echo (SE) sequences are used for most clinical applications. One disadvantage of standard SE sequences is the relatively long acquisition times needed, which relate directly to the repetition time (TR) and are therefore particularly long in heavily T2-weighted SE sequences. Shortening of acquisition time can be achieved by an accelerated data acquisition with gradientecho (GRE) sequences, which offer a spatial resolution similar to that of SE sequences (1-3). This technique allows a reduction of the flip angle of the radio-frequency pulse to less than 90#{176} and involves an echo stimulation by inversion of the readout gradient and often of the section-select gradient. Besides shortening acquisition time, GRE sequences reduce the radio-frequency deposit to patients (1,4,5). Reports on the application of GRE sequences for the investigation of the muscuboskeletal system are few and are limited to the joint cartilage and menisci (6-9). The goal of our study was to assess fast low-angle shot (FLASH) sequences as a possible replacement for SE sequences in the investigation of muscuboskeletab neoplasrns. Conse-

Index

quently contrast levels and contrastto-noise ratios (C/Ns) between neoplasms and normal tissue obtained with “Ti-weighted” and “T2-weight-

terms: Bone neoplasms, MR studies, 40.1214, 40.3 #{149} Gadolinium #{149} Magnetic resonance (MR), pulse sequences #{149} Muscles, MR studies, 40.1214 #{149} Muscles, neoplasms, 40.3 Radiology

ed” fied

1990; 176:489-495

the

Department

FLASH sequences and compared

WestfSlische

Wilhelms-Universit#{227}t,

Schweitzer-Strasse 33, 4400 Republic of Germany. From entific assembly. Received vision

April

requested

ii;

accepted

quests to RE. C

RSNA,

1990

March

April

with

were quantithose

achieved with SE sequences large patient group.

of Clinical

January

5; revision

18. Address

AND

in a

METHODS

Radiology, Albert-

M#{252}nster, Federal the 1989 RSNA 16,

1990;

#{149}

Low-Angle Gd-DTPA’

i07

were

in

the

extremities.

Patients

ranged in age from 6 to 80 years, with a median age of 32 years. All malignant tumoms were histologically proved by means of biopsies and surgical nesections. Histologic confirmation was obtained in benign tumors through biopsies and/or curettage in all lesions except two nonossifying fibnomas and three enchondromas. The latter showed very characteristic radiographic features and were therefore followed up clinically and nadiographically for at least 1 year.

Patients scime-

received

reprint

MD

N

PATIENTS From

M#{252}ller-Miny,

re-

One hundred twenty-one patients with bone and soft-tissue tumors were exammed. Thirty-three of the tumors were benign, and 88 were malignant (Table). Fifteen tumors were within the pelvis, and

Abbreviations: DTPA FLASH

view,

= =

GRE

signals time,

C/N = contrast-to-noise diethylenetniaminepentaacetic fast low-angle-shot, FOV =

gradient

averaged, TR

=

repetition

SE

echo,

NSA

spin

echo,

ratio, acid, field of

number TE

of

echo

time.

489

MR

Examination

All studies were performed with a i.5-T superconducting magnet (Magnetom; Siemens, Enlangen, Federal Republic of Gemmany) with an acquisition matrix of 512 pixels in the readout and 256 pixels in the phase-encoding directions, while the images were displayed in a 256 X 256 matnix. Peripheral skeletal tumors were examined with a cylindrical volume surface coil (n = 70); all other tumors were exammed with the body coil. SE and FLASH sequences were penformed in all patients. The following SE sequences were implemented: Ti-weighted SE sequences with 600/ 15 (TR msec/ echo time [TE] msec), number of signals averaged (NSA) 2, and section thickness 4-7 mm; T2-weighted SE sequences with 2,500/70, NSA 1, and section thickness = 4-9 mm; and Ti-weighted SE sequences following intravenous administration of 0.i mmol of gadolinium diethylenetniaminepentaacetic acid (DTPA) pen kilogram of body weight (enhanced Ti-weighted SE sequence) with 600/ 15, NSA = 2, section thickness = 4-7 mm. In SE sequences, the gap between sections was usually 30% of the section thickness. Since tumor staging was also of clinical relevance in these examinations, Ti-weighted SE sequences were usually performed in a longitudinal plane (both nonenhanced and enhanced sequences in identical sagittal on coronal planes), whereas most of the T2-weighted SE sequences were performed in a transaxial plane due to time constraints. Ti-weighted (flip angle, 90#{176}) and T2* weighted (flip angle, 10#{176}) FLASH sequences were selected. FLASH sequences with a short TR setting (40/ 10, NSA 2, section thickness = 7 mm) in single-section technique were used with a flip angle of 10#{176} (FLASH-iO), 90#{176} (FLASH-90), and 90#{176} following intravenous administration of Gd-DTPA (0.1 mmol/kg) (enhanced FLASH-90). These were penformed in the plane (usually longitudinal) showing maximal extent of the tumor in the SE sequences. Additionally, in 20 patients a multisection FLASH sequence with longer TR and TE settings (320/ 14, NSA = 2, section thickness 4-9 mm) and with a flip angle of 10#{176} (FLASH10/320) was performed in a transaxial plane, usually by using a gap of 30% of the section thickness. Each study started with the acquisition of the nonenhanced SE and FLASH sequences. Gd-DTPA was then administened to the patient while he or she was lying inside the MR unit. Four minutes after Gd-DTPA administration the enhanced FLASH-90 sequence was performed, followed by the enhanced Tiweighted SE sequence 2-3 minutes later.

Contrast

Calculation

Representative sections of all sequences displaying the largest portion of the tumon were selected for measuring signal intensities in regions of interest in both

490

#{149} Radiology

intraand extmaosseous tumor components, bone marrow, muscle, and fatty tissue. Since sequences were performed in different planes, inhomogeneities evident only in the periphery of a neoplasm in one plane were excluded from the megion of interest. In addition, the mean intensities of background noise, including systematic

and

suned outside large region The image culated with

statistical

noise,

the examined of interest. contrast and the following

were

body

$,-.

mea-

part

in a Figure

C/N were formulas:

cal-

1.

Acquisition

sequences

ben of sections. quisition

Contrast

(Slsampie

=

Sltissue)/ (Slsampie

+

(1)

Sitissue)

and C/N where sity

(SLmpie

=

represents

S’sample

of the

SItssue)/SInoise,

different

tumor

the

signal

(2)

C/N .

-

(iO/SL)2

(FOVnorm/FOV)2

.

NSA

,

acquired

with

num-

of shorten

FLASH

than

with

acSE se-

quences decreases with increasing number of sections. When i3-17 sections are acquired, there is no obvious reduction in acquisition time compared with that of the Ti-

weighted

SE sequences. both

flip

(Statgraphics, ville, rent

Md). study

bution

(FLASH angles

Time

For all sequences, the time required for the acquisition of one to 20 sections with NSA = i was calculated for each set of sequence parameters. The shortest possible acquisition time was defined as that time depending on TR of the sequence irrespective of the number of sections obtamed. Because of our parameter settings, the maximum number of sections that could be acquired with the T2-weighted SE sequence was 20; we consider 20 sections to be sufficient for delineating tumon from surrounding tissue in most cases.

version The

[40/10]

of 10#{176} and

2.6; STCS,

data

did

evaluated

not

show

(standardized

standardized and quartiles sequences

90#{176}.)

the

a normal

displayed

cur-

distni-

skewness

> 2.0,

kumtosis > 2.0), were estimated. that

Rockin

so medians To compare

both

positive

and negative contrast values, absolute data were used for statistical computation. Statistical significance of differences between contrast values and C/N was evaluated with the Mann-Whitney LI test (significance level of P < .05). Contrast vabues were compared at a patient level by using

(3)

where SL = section thickness. Image contrast is not influenced by these variables because they are common to both numeratom and denominator of Equation (1) and can therefore be eliminated. In lesions with both intra- and extnaosseous components, image contrast and C/N between the intnaosseous component and bone marrow and between extraosseous component and muscle on fatty tissue were calculated. In the instances in which isobated intraon extmaosseous components were present, image contrast and C/N against surrounding normal tissue were determined.

Acquisition

of the different

the

components

and 51tissue represents the signal intensity of muscle, bone marrow, or fatty tissue. With these formulas, contrast may mange from -1 to +1 on from 0 to +1 by using absolute data. On the other hand, absolute data of C/N may theoretically range from 0 to positive infinity. To connect for differing section thickness, field of view (FOV), and NSA, C/N was normalized for all sequences with reference to section thickness of 10 mm, the unmagnified FOV defined by the manufacturer (FOVnorm), and NSA 1 by using the following formula (10,11): C/Nnorm

on

The advantage

time

represents

inten-

time

depending

a ratio

sequences

of the

from

contrast

which

values

both

quartiles of these ratios were This procedure was performed

FLASH

sequences

against

of two

medians

and

computed. for all

all SE se-

quences.

RESULTS Acquisition

Time

FLASH

sequences have significantacquisition times than SE sequences, particularly if only very few sections are required for tumor delineation. With the TR and TE values in our study, acquisition time with FLASH-lO and FLASH-90 was only 1.6% of that of the T2-weighted SE sequence, and 6.8% of that of the Ti-weighted SE sequence for a single

by shorter

section. In contrast, acquisition time with FLASH sequences was 33.0% of that of the T2-weighted and 68.7% of that of the Ti-weighted SE sequence when 20 sections were acquired. If 15-17 sections were required, acquisition time for FLASH sequences was actually slightly longer than that for the Ti-weighted SE sequences (Fig 1).

Statistical

Analysis

Contrast

Statistical commercially

analysis was performed on a available software package

With image

FLASH contrast

sequences, between

highest tumor and

August

1990

I

:::

..e

..a

..a

..e

..7

..7 ‘.

..e S

..

a

..

3

!

I

..S

;

!

:::

..

..a

..a ..

‘.

..S

;

I

..

TI

T2

TIG

I#{149}

F32

9#{149}

ii

3

S.. S. I

I

Ti

9G

T2

TIQ

iS

F32

aS

TI

9G

b.

a.

.

T2

TIG

Pi

P32

PS

#{149}a

C.

Figure

2. Box and whisker plots for image contrast between tumor and muscle (a), tumor and bone marrow (b), and tumor and fatty tissue (c). Boxes represent 50% of the data values between the lower and upper quartiles. Vertical lines extend to the extremes (minimum and maximum values), while central horizontal lines represent the medians. Unusual values more than 1.5 times the interquamtile range away from the box are plotted as separate points. TI Ti weighted, T2 T2 weighted, T1G = enhanced Ti weighted, FlO = FLASH-lO, F32 = FLASH10/320, F90 = FLASH-90, F9OG = enhanced FLASH-90. Absolute contrast values between tumor and muscle differed significantly with SE and FLASH sequences at P < .001 for all parameter combinations except FLASH-10/320, Ti weighted (P < .01); enhanced FLASH-90, enhanced Ti weighted (P < .01); FLASH-90, Ti weighted (P < .05); and FLASH-b, Ti weighted (P > .05). Absolute contrast values between tumom and bone marrow differed significantly with SE and FLASH sequences at P < .001 for all parameter combinations except FLASH-90, T2 weighted (P < .01); enhanced FLASH-90, T2 weighted (P > .05); FLASH-10/320, T2 weighted (P > .05); enhanced FLASH-90, T2 weighted (P > .05); FLASH-90, enhanced Ti weighted (P > .05); FLASH-iO, enhanced Ti weighted (P > .05); FLASH-10/320, enhanced Ti weighted (P> .05). Absolute contrast values between tumor and fatty tissue differed significantly with SE and FLASH sequences at a level of P < .001 for all parameter combinations except FLASH-90, enhanced Ti weighted (P < .01); FLASH-b, enhanced Ti weighted (P < .01); FLASH-iO/320, T2 weighted (P < .05); FLASH-b, T2 weighted (P > .05); enhanced FLASH-90, T2 weighted (P > .05); FLASH-iO/320, enhanced Ti weighted (P > .05).

b.

a.

C.

Figure 3. Comparison of image contrast with FLASH and SE sequences at a patient bevel. The broad columns represent the medians of the ratios (contrast with FLASH/contrast with SE), while the overlaid small black columns represent the lower and upper quartiles. Ratios below 100% indicate higher contrast with SE sequences, and ratios above 100% indicate higher contrast with FLASH sequences. Image contrast between tumor and muscle was always significantly lower with FLASH than with the T2-weighted SE sequence (a). Image contrast between tumor and bone marrow (b) or between tumor and fatty tissue (C) was somewhat lower with nonenhanced FLASH than with nonenhanced Ti-weighted SE sequences.

::

::[

2#{149}S

aS.

a..

ie

te.

is

ia

141

141

121

La.

i:.

i::

:::L I

----

S.,-

ri

4.-

a.-

L!

.L

is

::

54S

La. #{176}

S.

S.

-

4S

4S

-

a. .

TI

T2

TIG

PI#{149}

-#{149}-

F32

P55

FSO

TI

T2

TIG

PiS

-#{149}

P3

P95

PSIG

TI

T2

TIG

i#{149}

P3a

SS

P550

S_Q_a#{149}

a. b. c. Figure 4. Box and whisker plots for C/N between tumor and muscle (a), tumor and bone marrow (b), and tumor and fatty tissue (c). TI = Ti weighted, T2 T2 weighted, T1G enhanced Ti weighted, FlO FLASH-lO, F32 = FLASH-iO/320, F90 = FLASH-90, F9OG = enhanced FLASH-90. Absolute C/N between tumor and muscle differed significantly with SE and FLASH sequences at P < .001 for all parameter combinations except enhanced FLASH-90 Ti weighted (P < .01); FLASH-iO, Ti weighted (P < .05); enhanced FLASH-90, Ti weighted (P > .05); FLASH-10/320, Ti weighted (P > .05). Absolute C/N between tumor and bone marrow differed significantly with SE and FLASH sequences at P < .001 for all parameter combinations except FLASH-90, T2 weighted (P > .05); FLASH-10/320, T2 weighted (P > .05). Absolute C/N between tumor and fatty tissue differed significantly with SE and FLASH sequences at P < .001 for all parameter combinations except FLASH10/320, T2 weighted (P > .05) and FLASH-b, T2 weighted (P > .05). Volume

176

#{149} Number

2

Radiology

#{149} 491

a.

b.

Figure

5.

fibula

Osteosarcoma

in a 15-year-old

girl.

90 (40/10), delineation mass and surrounding possible.

(b)

With

C.

of the proximal (a) With

FLASH-

between soft-tissue muscle is almost im-

enhanced

FLASH-90

(40/

10), the bright soft-tissue mass due to marked Gd-DTPA uptake can be clearly delineated from adjacent muscle. (c) With FLASH-iO (40/10), contrast between soft-tissue mass and muscle is poor. (d) With FLASH-iO/320 (320/14), contrast is higher than

with

FLASH-lO

but

is clearly

inferior

to that with enhanced FLASH-90. (e) With the T2-weighted SE sequence (2,500/70), contrast

is only

the enhanced

slightly

higher

than

with

FLASH-90.

muscle was obtained with the enhanced FLASH-90 and the FLASH10/320 sequence; the medians, however, reached only 0.19 (Fig 2a). Image contrast with FLASH sequences was 37%-i15% higher than with the Ti-weighted SE sequence, 6i%-82% bower than with the T2-weighted SE sequence, and i7%-58% bower than with the enhanced Ti-weighted SE sequence (Fig 3a). C/Ns with FLASH sequences were comparable to those with the nonenhanced Ti-weighted SE sequence but definitely lower than with the enhanced Ti-weighted or T2-weighted SE sequence (Figs 4a, 5). In 56 tumors, image contrast was at least 50% higher with enhanced FLASH-90 than with FLASH-iO. These included malignant (n 52) and aggressive benign (n = 4) tumoms, which showed marked GdDTPA uptake and therefore a higher signal intensity than muscle (Fig 6). In 23 tumors, image contrast was at least 50% higher with FLASH-iO than with enhanced FLASH-90. These tumors included cystic or predominantly necrotic (n 9) and welldifferentiated poorly calcified, chondrogenic obvious sclerotic 492

.

(n = 4) tumors showing no Gd-DTPA uptake and highly (n = 5) or fibrotic (n = 5) tu-

Radiology

d.

e.

moms with minor degrees of enhancement. The signal intensities were higher in the former and lower in the latter group of neoplasms in rebation to muscle with the FLASH-lO Sequence (Fig 7). With all nonenhanced FLASH sequences, image contrast between tumor and bone marrow was higher than that between tumor and muscle, with medians reaching values of up to 0.29. Image contrast was bow with the enhanced FLASH-90 sequence; this occurred because the tumors, which

quences depending on the histologic composition of a tumor. In 27 tumors, image contrast was at least 50% highem with FLASH-90 than with FLASH10 sequences. These tumors included predominantly sclerotic (n = 16), calcified (n = 6), or highly fibrotic (n 5) lesions. With FLASH-90 sequences, all these lesions were displayed as low-signal-intensity lesions within bright yellow marrow (Fig 8). In 20 tumors, image contrast was at least 50% higher with FLASH-iO than with FLASH-90 sequences. These included poorly calcified chondrogenic (n = 2), cystic (n 4), and purely osteobytic tumors with minimal fibrous tissue components (n 14). These lesions were displayed as bright lesions within bowsignal-intensity bone marrow with the FLASH-iO sequence (Fig 9). With FLASH-90, image contrast between tumors and med bone marrow was poor in all cases (n = 15). With FLASH sequences, image contrast between tumor and fatty tissue was as high as that between tumor and bone marrow, and the nonenhanced FLASH sequences achieved higher image contrast than the enhanced FLASH-90, reaching a maxi-

appeared bright due to marked GdDTPA uptake, blended with the high signal intensities of bone marrow (Fig 2b). Image contrast with the nonenhanced FLASH sequences was only 34%-40% lower than with the Ti-weighted SE sequence, 33%-80% higher than with the T2-weighted SE sequence, and 16%-22% higher than with the enhanced Ti-weighted SE sequence (Fig 3b). C/N with FLASH sequences was comparable to that with the T2-weighted SE sequence but definitely bower than with the nonenhanced and enhanced Tiweighted SE sequence (Fig 4b). Optimal image contrast was obtamed with different FLASH se-

August

1990

Figure tibia

6.

Osteosarcoma

in a 21-year

old

of the proximal woman.

(a) With

FLASH-90 (40/10), high contrast between the tumor and bone marrow is evident, while demarcation of the soft-tissue mass from surrounding muscle is poor. (b) With the enhanced FLASH-90 (40/iO), the softtissue mass can be delineated rounding muscle with high

from contrast,

sumwhere-

as contrast is low between the intraosseous tumor mass and adjacent bone marrow. (c) With FLASH-iO (40/iO), contrast between inferior

soft-tissue to that

90. Contrast adjacent

bone

mass and muscle with the enhanced

between

is clearly FLASH-

intraosseous

marrow

is low.

mass (d)

and

Contrast

between the soft-tissue mass and muscle is as high with the enhanced Ti-weighted SE sequence (600/15) as with the enhanced FLASH-90 sequence, and contrast between intraosseous

marrow

tumor

is clearly

mass

and

adjacent

bone

higher.

between fatty tissue and tumor was somewhat higher with FLASH-90 and somewhat bower with FLASH-b, since fatty tissue usually showed higher signal intensity than did bone marrow.

d.

C.

mal averaged value of 0.31 (Fig 2c). Image contrast with nonenhanced FLASH sequences was 36%-62% bowem than that with the Ti-weighted SE sequence, 23%-i09% higher than with the T2-weighted SE sequence, and 22% bower to 25% higher than that with the enhanced Ti-weighted SE sequence (Fig 3c). C/N with FLASH sequences was comparable to

Volume

DISCUSSION

b.

a.

176

#{149} Number

2

that

with

the

T2-weighted

SE se-

quence but definitely bower than with the enhanced and nonenhanced Ti-weighted SE sequences (Fig 4c). In general, the correlation observed between contrast of a tumor against fatty tissue and the histologic composition was similar to that seen when comparing tumor and bone marrow. However, image contrast

The main advantage of GRE sequences is that they can be performed quickly, thus allowing examination time to be shortened in cases in which standard SE sequences can be replaced. FLASH sequences can predominantly display differences in either longitudinal or transverse melaxation times within tissues; howevem, proton density, susceptibility, and chemical shift also have an impact on signal intensity. With flip angles banger than 60#{176} and short TR and TE values, contrast depends mainly on Ti differences; however, with flip angles smaller than 25#{176} and long TR and TE values, contrast is influenced mainly by T2* differences (12). Theoreticab estimation and published data on the investigation of the spinal canab have indicated that contrast values depend mainly on the flip angle and that variation of TR or TE produces only minor to moderate changes in contrast (13,14). However, the limitations of GRE sequences must also be considered. In the absence of a 180#{176} impulse, phase differences caused by gross magnet field inhomogeneity, susceptibility, and chemical shift are not mephased. Thus effective transverse mebaxation times (T2*) are clearly shorter than the inherent transverse relaxation times of samples; also, as TE increases, a significant detenioration in image quality occurs due to artifacts (4,12). Prosthetic devices and metallic clips within or adjacent to

Radiology

#{149} 493

the field of view can therefore limit the diagnostic value of GRE sequences. Vascular pulsations also produce more artifacts with GRE than with SE sequences; these can be markedly reduced through the use of flow-mephasing techniques. In our study, image contrast and C/N with FLASH sequences were usually lower than with SE sequences. Image contrast was diagnosticalby sufficient for the diffementiation of tumor from bone marrow or fatty tissue, but delineation of tumor from muscle posed problems. The image contrast determined by means of expected differences in transverse relaxation time was particularly poor; this was obvious with FLASH-iO and FLASH-10/320, where image contrast between tumor and muscle was clearly bower than that with the T2weighted SE sequence in all cases. These FLASH sequences achieved diagnostically sufficient image contrast only in the few cases of either predominantly cystic or heavily sclerotic tumors; this may be explained by large differences in T2* values or proton density. The poor T2*dependent image contrast may be attributed to the mange of TEs used (10-14 msec). Signab intensity is inversely proportional to the quotients TE/T2* in FLASH sequences and TE/T2 in SE sequences. Thus a short TE with FLASH sequences can have a comparabbe impact on signal intensity as a long TE setting usually chosen in T2weighted SE sequences only in tissues with very short T2* relaxation times. If, for example, T2* values are only a third of T2 values, the influence on signal intensity of a given TE with FLASH would be threefold that with SE sequences. Thus T2-dependent contrast at a given TE would be approximately threefold that with SE sequences, and therefore only a minor T2*dependent contrast would be obtained with a TE of 14 rnsec with FLASH sequences. It is worth noting that increasing T2* weighting by creating long TEs is of less practical value, as it results in a marked increase in susceptibility artifacts with concomitant image deterioration. Enhanced FLASH-90 was superior to FLASH sequences with a flip angle of 10#{176} for the differentiation of tumom from muscle, indicating that im-

age

contrast

resulting

from

differ-

ences times

in longitudinal relaxation was more obvious. Enhanced FLASH-90, however, showed high contrast marked

494

only in those tumors with Gd-DTPA uptake; in all these

#{149} Radiology

a.

b.

d.

C.

Figure

7.

is high

between

Aneurysmal the

bone cyst

and

cyst

of

the

adjacent

markedly higher between the cyst and (b). (c) With FLASH-90 (40/ 10), contrast than with the nonenhanced Ti-weighted

ischium

red

(arrowheads)

in

a 10-year-old

boy.

Contrast

(a) but the T2-weighted SE sequence (2,500/70) between cyst and marrow or muscle is clearly lower SE sequence (600/15) (d) and FLASH-b. marrow

muscle

a.

and

muscle

with

FLASH-lO

(40/10)

with

b.

Figure 8. Nonossifying fibroma of the proximal fibula (arrowheads) in a 13-year-old boy. (a) With FLASH-90 (40/10), the low-signal-intensity lesion can be clearly delineated from adjacent moderately bright bone marrow. (b) With FLASH-b (40/10), both the lesion and the bone marrow show low signal intensity; thus, delineation is impossible.

cases, high contrast was also noted with the enhanced Ti-weighted SE sequence. The selection of a suitable FLASH sequence for the differentiation of tumor from bone marrow depended on the predominating histologic tu-

mom component. FLASH-iO achieved maximal image contrast between bright osteolytic tumors with low fibrous component and the bow-signalintensity red or yellow marrow. The very bow signal intensity of bone marrow with FLASH sequences with

August

1990

b. C. a. Figure 9. Multifocal angiosarcomas of the left and right proximal tibias in a 28-year-old man. (a) With intensity lesions can be delineated from adjacent low-signal-intensity bone marrow with a high level of 10), contrast between the low-signal-intensity lesions and moderately bright bone marrow is lower than weighted SE sequence (600/15), contrast between the lesions and bone marrow is somewhat higher than

small flip relaxation

angles time

indicating short T2* is due to strong sus-

ceptibibity phenomena between cancelbous bone and marrow (9,15). In most cases, signal intensity of bone marrow was lower than that of muscle in spite of an approximately twofold longer T2 relaxation time (16). Thus with small flip angles, lesions with long T2* relaxation times appear bright within bow-signab-intensity bone marrow. On the other hand, heavily scberotic, calcified, and fibrotic tumors that show bow signal intensities due to low proton density with all FLASH sequences were displayed with highest contrast with the FLASH-90 sequence against the bright yellow marrow. Because of the susceptibility phenomena, yellow marrow appears less bright with FLASH-90 than with Ti-weighted SE sequences, and therefore contrast was lower in the former. Image contrast between besions and med marrow was poor with FLASH-90 in all cases, since the red marrow showed low signal intensity. Similar results were noted for the image contrast between neoplasms and fatty tissue, although the latter was somewhat brighter than the formen in all FLASH sequences. The FLASH sequences currently available cannot be recommended for initial investigation of bone and softtissue tumors. They offer either poor image contrast and C/N or an insignificant shortening of examination time compared with SE sequences. For differentiation between tumor and muscle, a FLASH sequence with a large flip angle performed following Gd-DTPA administration may be of value. However, no significant

shortening of examination time is achieved compared with the enhanced Ti-weighted SE sequence (0.82 minute if 10 sections are acquired with NSA 1 and 1.64 mmutes if 20 sections are acquired), and contrast values are somewhat lower. For both sequences in lesions showing no Gd-DTPA uptake, a T2weighted SE sequence must be performed additionally, thus further limiting their value. A correctly selected FLASH sequence allows a reliable differentiation of intmaosseous tumor components from bone manmow and may replace the nonenhanced Ti-weighted SE sequence. However, only minor shortening of examination time can be obtained. A total replacement of the T2-weighted SE sequences by FLASH sequences, which would significantly reduce exarnination time (between 7 and 11 minutes, depending on the number of acquired sections and with NSA 1), is not possible. Current indications for FLASH sequences may indude imaging in additional planes with very few sections when the tumom has been previously displayed with Ti- and T2-weighted SE sequences. U

1.

#{149} Number

2

J, Matthaei

A, Frahm

Merbold NMR 2.

KD.

FLASH

imaging

imaging:

using

pulses.

J Magn

Oppelt

A, Craumann

D, HInicke

low

18. van

Ortendahl

Characteristics

and

gradient

ology KOnig schneller

reversal

ROFO

KOnig

H,

M,

sions sional

1987;

Reson

1986;

RL,

13.

H,

comparison with 166:865-872.

1986;

Meulen C.

P. Croen Fast

field

JP, echo

Tinus imaging:

overview and contrast calculations. Reson Imaging 1988; 6:355-368. Mills TC, Ortendahl DA, Hylton LE,

Carlson

flip angle 162:531-539.

JW, MR

Kaufmann

imaging.

an

Magn

Peters

DR.

Bradley imaging. RR, TJ.

Assist

Rosen

JB.

disk

1988; PA,

Yeakley

NF,

Campagna

lespy

gated H, Slone

Radiology

Radiolo-

spin

spine: angle, gnaII. Spinal

166:473-478. McArdle CB, Wehrli

FW.

using multicomparison

echo.

Magn

Reson

6:517-525. RM,

T, Fitzsimmons

culoskeletal times.

imaging imaging:

gra-

I. General

disease.

sequence.

Pettersson

angle,

sequence.

pulse

1988;

11:7-

spine:

flip

Rubin JB. Cervical with a partial flip

MR echo

imag-

1987;

Cervical

a partial

pulse

cardiac

BR, in rapid

Contrast

cord disease. Radiology Kulkamni MV, Narayana

Imaging

neso-

WC, eds. St Louis:

Tomogn

Rubin

dy-

imaging.

Tl- and T2-weighted

with

JW,

et al.

and

MR

1988; 14. RB, Edelman CL, Brady

Enzman DR. MR imaging

with

PE,

static

1989; 171:767-773. Principles of magnetic

F.

Cervical spine slice gradient

fast

AMC.

MF,

considerations and gy 1988; 166:467-472.

54:15-

Modic imaging

athroscopy.

neoplasms:

dient-refocused

Fischer

M, MR

Gd-DTPA-enhanced

16. Enzman

Ra-

1988;

MR imaging

14.

Skeletal

K, Pathnia

R, Reiser

Mosby, Buxton Wismer

le-

three-dimen-

imaging.

Glucken

MR imaging: ing. J Comput

15.

a new

with

nance. In: Stark DD, Magnetic resonance 12.

R, et al.

three-dimensional

Erlemann

Radiology Wehnli

M.

164:753-758. Enlemann

joint

Musculoskeletal

11.

Vogt

17:465-471.

Fast

namic

M,

in cartilaginous

knee

of the knee: Radiology

67:258-266.

R, Barfuss

C,

gradient-echo

Tynrell

f#{252}r

comparison of spinFLASH sequence MR

resonance

1988;

MT.

10.

Radi-

145:307-310.

Bongantz

of the

et

angle

imaging.

R, Deimling

Radiology

Reisen

diol

TC,

flip

gelenkdiag-

1986;

Sauten

Magnetic

9.

Mills

kemnspintomognaphische

images.

W,

rapid

MR

Cartilage disorders: echo, CHESS, and 8.

DA, of partial

1988; 166:17-26. H, Deimling M. Bedeutung gnadientenechosequenzen

nostik.

7.

16. den

Bruntink

4.

ML,

al.

die

flip-angle

H, Hartl W, Schajor W. FISP: MRI sequence. Electromedica 3.

Winkler

dient-nefocused

Haase

Partial 1987;

176

6.

References

Crooks

Volume

5.

FLASH-iO (40/iO), both high-signalcontrast. (b) With FLASH-90 (40/ with FLASH-iO. (c) With the Tiwith the FLASH sequences.

Spanier

JR.

tumors:

Scott

Ti and 1988;

5, GilKN.

Mus-

T2 relaxation

167:783-785.

NM, L. Radiology

Radiology

#{149} 495