Hajime
Sakuma,
Tsuyoshi
MD
Nakagawa,
Yoshiyuki #{149} Yoichi
#{149}
MD
Nomura, MD #{149}Kan Takeda, MD #{149}Tomoyasu Tamagawa, MS #{149}Yasushi Ishii, MD #{149}Tetsuji
Adult and Neonatal Diffusional Anisotropy with Diffusion-weighted Diffusional anisotropy of human brain was investigated clinically in six adult volunteers, eight premature neonates, and three infants aged 5-10 months. Diffusion-weighted magnetic resonance imaging was performed with gradient b of 450 sec/ 2 The direction of diffusionsensitive
gradients
was
changed
among x, y, and z axes according the orientation of neurofibers in white matter. Ti- and T2-weighted images
also
were
obtained
to
for evalua-
tion
of myelination. Diffusional anisotropy was demonstrated in white matter in the adults. Extensive signal attenuation was observed when gradients were parallel to white matter fibers. Conversely, in neonates, diffusional anisotropy of white matter, in which no myelination was shown on Ti- and T2-weighted images, was weak. Diffusional anisotropy was more distinct after brain maturation, as observed
in adult
white
matter.
Detection of dilfusional anisotropy is useful in evaluating neonatal brain development and various white matter disorders. Index
terms:
Brain,
anatomy,
10.92
M
of self-diffusion of water with magnetic resonance (MR) imaging has gained increasing interest recently because it enables evaluation of the microenvironment of in vivo tissue water. Moseley et a! (1) demonstrated anisotropic water diffusion in white matter of cat brain with a small-bore MR system with extensive gradient strength. They proved that signal attenuation EASUREMENT
by water diffusion in white matter greatly depends on the relationship between the orientation of neurofi-
bers and the direction of the diffusion-sensitive gradients. Greater signal attenuation (faster diffusion) was observed when their direction was parallel, compared with that obtained with perpendicular alignment. Recently,
man by
diffusional
white
anisotropy
matter
Chenevert
was
in hu-
demonstrated
et a! (2).
explanation that random cu!es
was
One
suggested
of the anisotropy motion of water restricted
by
the
myelin
of neoevaluated MR images
(3,4).
are
images
gener-
water diffusion man brain with
we
attempted
MR
From the Department of Radiology, Mie University School of Medicine, 2-174 Edobashi, Tsu, Mie 514,Japan (H.S., Y.N., K.T., T. Tagami, TN.); the Department of Radiology, Fukui Medical School, Fukui, Japan (Y.T., Y.I.); and
Yokogawa
Medical
Systems,
Hino,
Tokyo
(T.
Tsukamoto). From the 1990 RSNA scientific assembly. Received November 28, 1990; revision requested January 8, 1991; revision received February 13; accepted February 25. Address reprint requests to H.S. 0 RSNA, 1991
to detect
machine
anisotropic
in white a widely
and
matter used
of huclinical
to investigate
effect of myelination anisotropy in white paring adult, infant, brain.
at 35-40
full-term
the
on diffusional matter by comand neonatal
reflexes
cies
recorded
were
ical Systems,
A pulse
Six
and
volunteers
healthy
neonates
were
cardiac
were
neurologic neonates
aged
23-28
abnormality. included eight
or
with
for
diffusion-weighted
single-echo
were
used
spin-
as a standard
image
(repetition
cycles
time
in adults
with
car-
[TR] of two
and
four
cardiac
in neonates,
300-msec delay after cardiac trigger, 120-msec echo time [TEl, one excitation, 128 phase-encoding steps, 5- or 10-mm thickness). Diffusionweighted SE images were obtained by adding diffusion-sensitive gradient pulses cycles
of 44 msec,
gradient
separation
of 10.8 msec, gradient amplitude of 0.9 G/cm) on either side of a 180#{176} radio-frequency pulse (Fig 1). The direction of diffusion-gradient pulses could be changed between x (readout), y (phase-encoding), and z (section-selection) axes to detect diffusional anisotropy. Moreover, a Ti-
weighted SE sequence (TR of 400 msec, of 20 msec, one excitation, 192 phase-encoding steps, 7-mm and a T2-weighted 2,000
msec,
thickness, SE sequence
TE of 80 msec)
3-mm gap) (TR of
were
obtained
for evaluation of myelination. Diffusional anisotropy was visually evaluated by comparing two or three fusion-weighted
obtained
brain
with
images
different
sion-sensitive
analysis was performed ing a region of interest (RO!) The
to the
neurofiber
apparent
for each =
RO!
coefficient
calculated
following equation (6): -ln (S1/SO)/b, where 0
orientation
relative
quan-
by studyon the
diffusion
was
dif-
were
of diffu-
In addition,
titative images.
that
directions
gradients.
TE
accord-
is the
to the diffu-
and 11 infants
studied
between
Octo-
ber 1989 and July 1990. The adult teers
equipped
non-diffusion-weighted
gating
infants
originally developed with language for the detection water diffusion. Axial or
(SE) images
diac
of the
coils. A standard head coil was used.
single-section
echo
No
deficien-
images were obSigna system (GE Med-
sequence
ADC(O) METHODS
in any
gradient bird cage
imaging was pulse program of anisotropic
three
months.
Milwaukee)
self-shielded quadrature
and
5-10
or neurologic
neonates. Diffusion-weighted tamed with a 1.5-T
(ADC)
AND
gestation
aged
abnormal
ing
MATERIALS
weeks
infants
(duration
was mole-
sheath. White matter maturation nates and infants has been with Ti- and T2-weighted Ti-weighted
nates
coronal
Radiology
180:229-233
MS
Brain: Myelination MR Imaging’
ally used in monitoring brain development in the first 6 months of life; T2-weighted images are more valuable after 6 months (5). In this study,
1991;
MD
Tsukamoto,
Human and
Brain, growth and development, 10.92 #{149}Brain, MR studies, 10.1214 #{149} Brain, white matter Magnetic resonance (MR), diffusion study Magnetic resonance (MR), technology #{149} Myelm, 10.92 #{149}
Tagami,
years
and
volunhad
The infants premature
no
and neo-
Abbreviations: ADC efficient, ROI = region echo, TE = echo time,
apparent diffusion coof interest, SE = spin TR = repetition time. =
229
sion-gradient direction (in degrees), Si is signal intensity with diffusion-sensitive gradients, SO is signal intensity without diffusion-sensitive gradients, and b is a gradient factor. Gradient factor b in the y axis is calculated as b = -y2(DG)2(213 D + I), where -y is the gyromagnetic ratio, D and
G are the duration spectively,
and I
and the amplitude,
of diffusion-sensitive is the
sensitive
interval
factor
The
the
U
re-
diffusion-
resultant
in the y direction
Slice (Z)
p
gradients,
between
gradients.
RF and Echo
is 450
Phase ‘)
gradient As
sec/mm2.
n____ ________
for the gradient b factors in the x and z directions, the effects of cross terms between the diffusion and imaging gradient pulses must be considered (7). Therefore,
Read (X)
the gradient factors in the x and z direclions are slightly larger than the value estimated in the y direction (493 sec/mm2 in the x direction and 475 seo’mm2 in the z
Diffusion sensi. tive gradient
direction).
To
test
the
accuracy
of the
mea-
surement, diffusion coefficients with water phantoms and acetone phantoms were measured. The values of ADC(O) were determined for 0 = O (parallel orientation) and 90#{176} (perpendicular
frontal
orientation).
white
matter
radiation
were
The
of diffusional
ratio
calculated
Four
ROIs
.
________
I
1#{149}
axis)
(X,YorZ
Figure 1. Diffusion-weighted live gradients was applied
separate
I____
I
experiments
SE pulse to x (readout),
to detect
sequence used y (phase-encoding),
diffusional
anisotropy.
in this RF
=
study. A pair of diffusion-sensiand z (section-selection) axes radio frequency.
in
in
and four in the optic
obtained
in each
subject.
anisotropy
was
as ADC(90#{176})/ADC(O#{176}).
The statistical significance of the difference between the ratio of anisotropy in adults and that in neonates was determined with a Student t test.
RESULTS Diffusion coefficients of water and acetone at 21#{176}C are shown in Table 1. The measured diffusion coefficients (1.85 x iO- mm2/sec ± 0.02 x mean ± standard deviation, for water and 4.12 x iO- mm2/sec ± 0.03 x iO for acetone in x direction) were in agreement
(2.07 water
0.05
x iOand x i0-
with reported values mm2/sec ± 0.05 x 1O 3.88 x i0 mm2/sec ± for acetone)
coefficients
were
for
(2). Diffusion
fairly
constant
when
the direction of diffusion-sensitive gradients was changed among x, y, and z axes. White matter myelination of brain in adults appeared normal on Ti- and
T2-weighted anisotropy in white
MR images. was matter
images. When of diffusion-sensi-
the
attenuation by water diffusion was greater than when the direction of the gradients was perpendicular to the
(Fig 2). Diffusional
anisotropy
was most obvious in white where the fiber was aligned to either
pus
callosum,
ternal
230
the capsule.
Radiology
#{149}
x, y, or z axis,
optic
radiation,
Directional
matter parallel such
and
in the
peared
normal
radiation
for
aged white
(Table
2).
5-10 months, matter ap-
their
ages.
High
imradia-
tion and the centrum semiovale. The development of white matter myelination was normal. On diffusion-
weighted images, diffusional anisotropy of white matter was well demonstrated,
in-
optic infants of the
white mat± 0.16 x
signal intensity on Ti-weighted ages was observed in the optic
brain
as cor-
difference
± 0.14 x iO in frontal ter and 0.48 x iO mm2/sec In three maturation
tive gradients was parallel to the onentation of white matter fibers, signal
fibers
sec
io
Diffusional
clearly demonstrated of all adults on the
diffusion-weighted relative direction
of diffusion was confirmed by calculating the ADCs in various areas in white matter of the adult brain. ADCs of white matter varied considerably, depending on the relative alignment of the direction of the diffusion-sensitive gradients and the orientation of the fibers. In six adult brains, the mean ADC(O#{176})was 1.24 x iO sec ± 0.16 x i0in frontal lobe white matter and 1.25 x iO mm2/sec ± 0.20 x i#{248}-in the optic radiation. The mean ADC(90#{176}) was 0.48 x iO mm2,
as shown
(Fig 3). Areas
diffusional
anisotropy
in adult
showing on
distinct diffusion-
weighted MR images seemed to be larger than areas with increased sig-
nal intensity on Ti-weighted images. The directional difference of ADC in ROl-image analysis was great. In three infant brains, the mean ADC(O#{176})was 1.37 x iO mm2/sec ± 0.11 x iO- in frontal white matter and 1.47 x iO- mm2/sec ± 0.17 x iO in the optic radiation. The mean ADC(90#{176})was 0.61 x iO mm2/sec ± 0.14 x iO-3 in frontal white matter and 0.67 x iO- mm2/sec ± 0.20 x iO in the optic radiation. These findings were similar to those in adult white matter. In all premature neonates, signal intensity on Ti-weighted images was slightly increased in the brain stem and the posterior limb of the internal capsule. No increased signal intensity on Ti-weighted images was observed in deep and cortical white matter, including the optic radiation. On diffusion-weighted images, diffusional anisotropy was weak in deep and cortical white matter in neonates compared with myelinated adult white matter (Fig 4). Diffusional anisotropy
seemed
relatively
the optic radiation other areas of deep ROI-image analysis,
obvious
in
compared with white matter. In ADCs showed July
1991
b.
Figure gradient. (arrows)
2.
MR images in brain (b, c) Diffusion-weighted was changed between
the direction
of the diffusion-sensitive
greatest. For example, plied perpendicular
C.
of a healthy 25-year-old subject. (a) Axial SE image (TR = 1,800 msec, TE = images (TR = 1,800 msec, TE = 120 msec, b = 450 sec/mm2). The direction readout axis (anterior-to-posterior direction in b) and phase-encoding axis
gradient
optic radiation to the fibers (c).
has
low
is parallel signal
to the orientation
intensity
of white
(b, arrows).
In contrast,
matter less
fibers, diffusion
signal
120 msec) without diffusion-sensitive of the diffusion-sensitive gradient (left-to-right direction in c). When
attenuation
by water
is hyperintense
when
(6,12,13).
However,
weighted
images
by eddy
the
diffusion
is
gradient
is ap-
diffusionwere
current,
deteriorated
especially
when
the
diffusion-sensitive gradient was applied on phase-encoding and sectionselection axes. Recently, detection of anisotropy of in vivo water diffusion
has become possible ager, and excellent mance was achieved shielded
gradient
Moseley lent
coils.
et a! (1,14)
with
differences. the mean
In
significantly nates (P
lower < .01).
ADC(0#{176}) was 1.36 x iO mm2/sec ± 0.22 x iOin frontal white matter and 1.25 x iO- mm2/sec ± 0.30 x iO-
in the optic radiation. The mean ADC(90#{176})was 1.19 x iO mm2/sec ± 0.24 x iO- in frontal white matter and 0.84 x iO- mm2/sec ± 0.25 x i03 in the optic di.ffusional
radiation. anisotropy-that
The
ratios of is, aver-
aged ADC(90#{176})/ADC(O#{176}) in neonates, infants aged 5-10 months, and adults-are summarized in Figure In neonates,
the
ADC(O#{176})was
0.88
white
optic white mean
matter
0.10 in frontal
white
matter,
which
was significantly lower than that in neonates (P < .01). The mean ADC(90#{176})/ADC(O#{176}) of the optic radiation in adults (0.39 ± 0.13) was also Volume
gradient
radiation. The ratio decreased as matter matured. In adults, the ADC(90#{176})/ADC(O#{176}) was 0.38 ±
180
#{149} Number
1
in neo-
(8). was
In addition,
diffusion performed
measurement
with
orientation of the They demonstrated efficient measured
Calif) that with self-
early
we
weighted images of human brain with a clinical MR system equipped with ordinary self-shielded gradient coils. Because the gradient b factor in this study was limited (450 seVmm2), our diffusion-weighted
images
were
not
so heavily weighted as those shown by Moseley et al (1,14) (b factor 1,413 sec/mm2). The images obtained, =
a pulsed in the
history of nuclear MR diffusion measurements (9,10). Cleveland et al (ii) studied water diffusion of rat muscle by changing the relative direction of diffusion-sensitive
Fremont, equipped
of cat
(GE
fusion weighted. For this report, attempted to obtain diffusion-
Several factors exist that affect the signal intensity of nuclear MR. One factor is water-proton diffusion, which has been investigated for more than 25 years with the nuclear MR of anisotropic
mean ADC(90#{176})/ ± 0.11 in frontal and 0.67 ± 0.11 in the
that
MR unit
shielded gradient coils (Acustar; GE Medical Systems). The images obtamed in that study were heavily dif-
DISCUSSION
method
5.
than
excel-
images
a CSI 2.0-T
Medical Systems, had a small bore
slight directional neonatal brains,
showed
diffusion-weighted
brain
only eight
with an MR imgradient perforby means of self-
however, diffusional
matter
successfully anisotropy
of human
Evaluation
with
demonstrated in the white
adult MR
brain. imaging
of
normal and abnormal neonatal brain gradients to the development has been performed images (3muscle fiber 0#{176}-90#{176}.with Ti- and T2-weighted that diffusion at 0#{176} is 1.39-fold
co-
larger than that measured at 90#{176}. In the past 5 years, spatial mapping of in vivo water diffusion with MR imaging techniques has been developed
5). In studies strength MR
with a 1.5-T high-fieldsystem, the increase
in
signal intensity on Ti-weighted images preceded the decrease in signal intensity on the T2-weighted images. It was hypothesized that Ti shortenRadiology
231
#{149}
a. Figure
C.
b.
3.
along
Diffusion-weighted images in brain readout axis (a, anterior-to-posterior
the
matter
also was observed
creased
signal
intensity
in adults.
The
on TI-weighted
extent images
of a healthy 5-month-old direction) and phase-encoding
infant.
of white
anisotropic
4.
to-right
Diffusion-weighted
(b) directions.
relatively
obvious
age (c), signal
images
Diffusional
anisotropy
intensity
in neonatal
anisotropy
diffusion
seemed
is observed
is increased
only
water content (5). The changes
of mature in signal
matter
was
that
was
in adults.
Radiology
#{149}
significantly
Ti-
and
higher
T2-
than
areas
with
in-
gradient
matter
of neonates
directional
differences
limb of the internal
(arrows)
was
compared
applied
with
that
of ADC are smaller
in anterior-to-posterior
in white
(a) and
matter
than in adults.
of adults.
left-
Although
On TI-weighted
im-
capsule.
.
ADC(90) A.DC(O)
El
Adult
Infant Neonatal
brain brain
(n=6) (>5
brain
months,
n=3)
(n=8)
1.0
0.8 0.6 0.4
ob-
scure compared with that in adults. In the quantitative analysis, the ADC in neonatal white matter, measured with the gradients parallel to the fibers, did not differ significantly from that seen in adults. In contrast, the ADC in neonates measured with the perpendicular gradients
in white
in the optic radiation, in the posterior
white matter intensity on
in neonates
Diffusion-sensitive
weak
Ti- and T2-weighted images reflect the concentrations of various biochemical compounds and water. In our study, diffusional anisotropy of white
to be larger
(arrows) was of white
C.
brain.
seems
ing of myelinated white matter was caused by the increase of various proteins, cholesterol, or glycolipids in the developing myelin sheath. Decreased signal intensity on T2-weighted images was explained by the decrease in
232
water
b.
Figure
than
showing
gradient anisotropy
(c).
a.
the
matter
The direction of the diffusion-sensitive axis (b, left-to-right direction). Diffusional
0.2 0
Frontal
white matter
Optic radiation
Figure 5. Comparison of average ratio of diffusional anisotropy neonates, infants, and adults. Mean values in myelinated white tic radiation in the adults were significantly less than in neonates
(ADC[90#{176}VADC[0#{176}]) among matter of frontal lobe and op(*). SD = standard devia-
lion. July 1991
weighted images obtained concurrently showed no evidence of development of myelination in the corresponding suggested
that
area. These diffusional
findings anisotropy
was closely related to neonatal brain maturation, especially with development of myelination. In addition, as shown in the optic radiation, diffusional anisotropy could be observed earlier than the increase of signal intensity on Ti-weighted images. This finding led to the possibility of earlier detection of neonatal brain development with diffusion-weighted MR imaging. Diffusion-weighted imaging provided new information about the microscopic structure of white matter, which could not be assessed with ordinary
Ti-
and
T2-weighted
images.
Several technical limitations have been discussed in diffusion-weighted imaging; the first is fragility to the slightest motion. Profound sleep and careful fixation sary to perform
of the head successful
there
In conclusion,
weighted were
is a limitation
tiple-section data acquisition. planar imaging is the solution these motion artifact problems
in mu!-
Echofor (15).
excellent
images obtained
uation
of neonatal
#{149} Number
1
Le Bihan D, Breton E, Lallemand D, et al. MR imaging of intravoxel incoherent molions: application of diffusion and perfusion in neurogenic disorders. Radiology
7.
Le Bihan D. Magnetic resonance imaging of perfusion. Magn Reson Med 1990; 14: 283-292. Stejeskal EO, TannerJE. Spin diffusion measurements: spin-echoes in the presence of a time-dependent field gradient. J Chem Phys 1965; 42:288-292. Stejeskal EO. Use of spin echo in pulsed magnetic-field gradient to study anisotropic, restricted diffusion and flow. Chem Phys 1965; 43:3597-3603. Cooper RL, Chang DB, Young AC, et al. Restricted diffusion in biophysical systems: experiment. Biophys J 1974; 14:161-177. Cleveland GG, Chang DC, Hazelwood CF. Rorshach HE. Nuclear magnetic resonance measurements of skeletal muscle anisotropy of the diffusion coefficient of the intracellular water. Biophys J 1976; 16:
1986;
8.
b value
development
of various
9.
10.
11.
white
#{149}
References 1.
2.
3.
et ah.
Developmental
features
12.
Radiology Dietrich
13.
14.
ah.
MR evaluation
patterns
delayed 5.
Barkovich man D.
WG,
of early
in normal
15. Zaragoza
EJ, et
myehination
and
developmentally infants. AJR 1988; 150:889-896. AJ, Kjos BO,Jackson DE, NorNormal maturation of the neona-
brain:
1988;
MR imaging
CB, Lee
SV, Nalcioglu
0, Cho
ZH.
1987; 14:43-48. Merboldt KD, Hanicke W, Frahm J. Selfdiffusion NMR imaging using stimulated echoes. J Magn Reson 1985; 64:479-486. Moseley ME, Cohen Y, Mintorovitch J, et
al.
of the neo-
1987; 162:223-229. RB, Bradley
Ahn
The effect of random directional distributed flow in MMR imaging. Med Phys
natal brain: MR imaging. I. Gray-white matter differentiation and myelination. 4.
161:401-407.
1043-1053.
Moseley ME, Cohen Y, KucharczykJ, et al. Diffusion-weighted MR imaging of anisotropic water diffusion in cat central nervous system. Radiology 1990; 176:439-445. Chenevert TL, BrunbergJA, Pipe JG. Anisotropic diffusion in human white matter: demonstration with MR techniques in vivo. Radiology 1990; 177:401-405. McArdle CB, Richardson CJ, Nicholas DA,
Radiology
180
a gradient
brain
and the diagnosis matter diseases.
6.
brain
sec/mm2 by using a clinical i.5-T MR system. Diffusional anisotropy was closely related to the development of white matter in neonatal brain. Detection of diffusional anisotropy has clinical potential in the eva!-
tal and infant
Volume
no exgra-
diffusion-
of human with
im-
of 450
are necesdiffusion-
weighted imaging of infants. Cardiac gating is indispensable, and delay after an R wave of cardiac trigger must be chosen carefully to minimize motion artifact from brain pulsation. Therefore,
Eddy current artifact is another portant problem in diffusionweighted imaging. In our study, substantial eddy current problem isted because of the self-shielded dient coils.
Early detection
of regional
cerebral
ischemia in cats: comparison of diffusionand T2-weighted MRI and spectroscopy. Magn Reson Med 1990; 14:330-346. Turner R, Le Bihan D, MaierJ, Vavrek R, Hedges LK, PekarJ. Echo-planar imaging of intravoxel incoherent motion. Radiology 1990;
177:407-414.
at 1.5 T.
166:173-180.
Radiology
#{149} 233