Identification of brown adipose tissue in mice with fat–water IDEAL‐MRI
ABSTRACT Purpose:To investigate the feasibility of using IDEAL (Iterative Decomposition with Echo Asymmetry and Least squares estimation) fat–water imaging and the resultant fat fraction metric in detecting brown adipose tissue (BAT) in mice, and in differentiating BAT from white adipose tissue (WAT).Materials and Methods:Excised WAT and BAT samples and whole-mice carcasses were imaged with a rapid three-dimensional fat–water IDEAL-SPGR sequence on a 3 Tesla scanner using a single-channel wrist coil. An isotropic voxel size of 0.6 mm was used. Excised samples were also scanned with single-voxel proton spectroscopy. Fat fraction images from IDEAL were reconstructed online using research software, and regions of WAT and BAT were quantified.Results:A broad fat fraction range for BAT was observed (40–80%), in comparison to a tighter and higher WAT range of 90–93%, in both excised tissue samples and in situ. Using the fat fraction metric, the interscapular BAT depot in each carcass could be clearly identified, as well as peri-renal and inguinal depots that exhibited a mixed BAT and WAT phenotype appearance.Conclusion:Due to BAT's multi-locular fat distribution and extensive mitochondrial, cytoplasm, and vascular supply, its fat content is significantly less than that of WAT. We have demonstrated that the fat fraction metric from IDEAL-MRI is a sensitive and quantitative approach to noninvasively characterize BAT. J. Magn. Reson. Imaging 2010;31:1195–1202. © 2010 Wiley-Liss, Inc.
- [Show abstract] [Hide abstract]
ABSTRACT: The relevance of functional brown adipose tissue (BAT) depots in human adults was undisputedly proven approximately seven years ago. Here we give an overview of all dedicated studies that were published on cold-induced BAT activity in adult humans that appeared since then. Different cooling protocols and imaging techniques to determine BAT activity are reviewed. BAT activation can be achieved by means of air- or water-cooling protocols. The most promising approach is individualized cooling, during which subjects are studied at the lowest temperature for nonshivering condition, probably revealing maximal nonshivering thermogenesis. The highest BAT prevalence (i.e. close to 100%) is observed using the individualized cooling protocol. Currently, the most widely used technique to study the metabolic activity of BAT is [(18)F]FDG-PET/CT-imaging. Dynamic imaging provides quantitative information about glucose uptake rates, while static imaging reflects overall BAT glucose uptake, localization and distribution. In general, standardized uptake values (SUV) are used to quantify BAT activity. An accurate determination of total BAT volume is hampered by the limited spatial resolution of the PET-image, leading to spill over. Different research groups use different SUV threshold values, which make it difficult to directly compare BAT activity levels between studies. Another issue is the comparison of [(18)F]FDG uptake in BAT with respect to other tissues or upon with baseline values. This comparison can be performed by using the 'fixed volume' methodology. Finally, the potential use of other relatively noninvasive methods to quantify BAT, like MRI or thermography, is discussed.American journal of physiology. Regulatory, integrative and comparative physiology. 05/2014;
- [Show abstract] [Hide abstract]
ABSTRACT: In rodents, brown adipose tissue (BAT) regulates cold- (CIT) and diet-induced thermogenesis (DIT). Whether BAT recruitment is reversible and how it impacts on energy metabolism has not been investigated in humans. We examined the effects of temperature acclimation on BAT, energy balance and substrate metabolism in a prospective crossover study of 4-month duration, consisting of 4 consecutive blocks of 1-month overnight temperature acclimation [24°C (month 1) → 19°C (month 2) → 24°C (month 3) → 27°C (month 4)] of five healthy men in a temperature-controlled research facility. Sequential monthly acclimation modulated BAT reversibly, boosting and suppressing its abundance and activity in mild cold and warm conditions (p<0.05), respectively, independent of seasonal fluctuations (p<0.01). BAT-acclimation did not alter CIT but was accompanied by DIT (p<0.05) and post-prandial insulin sensitivity enhancement (p<0.05), evident only after cold-acclimation. Circulating and adipose tissue, but not skeletal muscle, expression levels of leptin and adiponectin displayed reciprocal changes concordant with cold-acclimated insulin sensitization. These results suggest regulatory links between BAT thermal plasticity and glucose metabolism in humans, opening avenues to harnessing BAT for metabolic benefits.Diabetes 06/2014; · 8.47 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: The recent discovery of active brown fat in human adults has led to renewed interest in the role of this key metabolic tissue. This is particularly true for neurodegenerative conditions like Huntington disease (HD), an adult-onset heritable disorder with a prominent energy deficit phenotype. Current methods for imaging brown adipose tissue (BAT) are in limited use because they are equipment-wise demanding and often prohibitively expensive. This prompted us to explore how a standard MRI set-up can be modified to visualize BAT in situ by taking advantage of its characteristic fat/water content ratio to differentiate it from surrounding white fat. We present a modified MRI protocol for use on an 11.7 T small animal MRI scanner to visualize and quantify BAT in wild-type and disease model laboratory mice. In this application study using the R6/2 transgenic mouse model of HD we demonstrate a significantly reduced BAT volume in HD mice vs. matched controls (n = 5 per group). This finding provides a plausible structural explanation for the previously described temperature phenotype of HD mice and underscores the significance of peripheral tissue pathology for the HD phenotype. On a more general level, the results demonstrate the feasibility of MR-based BAT imaging in rodents and open the path towards transferring this imaging approach to human patients. Future studies are needed to determine if this method can be used to track disease progression in HD and other disease entities associated with BAT abnormalities, including metabolic conditions such as obesity, cachexia, and diabetes.PLoS ONE 09/2014; 9(8):e105556. · 3.53 Impact Factor
Identification of Brown Adipose Tissue in Mice
With Fat–Water IDEAL-MRI
Houchun H. Hu, PhD,1*Daniel L. Smith, Jr, PhD,2Krishna S. Nayak, PhD,1
Michael I. Goran, PhD,3and Tim R. Nagy, PhD2
Purpose: To investigate the feasibility of using IDEAL
(Iterative Decomposition with Echo Asymmetry and Least
squares estimation) fat–water imaging and the resultant
fat fraction metric in detecting brown adipose tissue (BAT)
in mice, and in differentiating BAT from white adipose tis-
Materials and Methods: Excised WAT and BAT samples
and whole-mice carcasses were imaged with a rapid
three-dimensional fat–water IDEAL-SPGR sequence on a
3 Tesla scanner using a single-channel wrist coil. An iso-
tropic voxel size of 0.6 mm was used. Excised samples
were also scanned with single-voxel proton spectroscopy.
Fat fraction images from IDEAL were reconstructed online
using research software, and regions of WAT and BAT
Results: A broad fat fraction range for BAT was observed
(40–80%), in comparison to a tighter and higher WAT
range of 90–93%, in both excised tissue samples and in
situ. Using the fat fraction metric, the interscapular BAT
depot in each carcass could be clearly identified, as well
as peri-renal and inguinal depots that exhibited a mixed
BAT and WAT phenotype appearance.
Conclusion: Due to BAT’s multi-locular fat distribution
and extensive mitochondrial, cytoplasm, and vascular
supply, its fat content is significantly less than that of
WAT. We have demonstrated that the fat fraction metric
from IDEAL-MRI is a sensitive and quantitative approach
to noninvasively characterize BAT.
Key Words: brown adipose tissue; white adipose tissue;
fat fraction; IDEAL
J. Magn. Reson. Imaging 2010;31:1195–1202.
C 2010 Wiley-Liss, Inc.
BROWN ADIPOSE TISSUE (BAT) remains a topic of
great interest in obesity and metabolism research (1–
3). Whereas white adipose tissue (WAT) is a lipid res-
ervoir and white adipocytes are composed of a single
large intracellular lipid droplet, BAT in contrast is
involved in energy expenditure and non–shivering-
thermogenesis. Brown adipocytes are also highly vas-
cularized and active in lipid metabolism (1). Further-
more, brown adipocytes are characterized by multiple
smaller intracellular lipid droplets and an abundance
of iron-rich mitochondria (1,4). Currently, BAT meas-
urements in animals require killing and necropsy, or
rely on indirect measures of energy expenditure in
response to BAT stimulation (5). BAT studies in
humans have been limited, partly due to the lack of
reliable and noninvasive detection techniques to study
small quantities of diffusely distributed BAT depots in
vivo (6,7). Incidental findings of BAT with positron
emission and computed tomography (PET/CT) have
been reported (8–10) in humans. These findings have
humans, using PET/CT in addition to biochemical
and histological assays (11–14). However, PET/CT is
expensive and requires radionuclide uptake by meta-
bolically active BAT, as well as ionizing radiation
exposure. Consequently, PET/CT is not broadly appli-
cable as a minimal-risk procedure for studying BAT
in large subject populations.
Investigators in the past have used chemical-shift-
selective MRI and spectroscopy (MRS) techniques to
detect and quantify interscapular BAT in rats (15,16).
Lunati et al. used paired two-dimensional (2D) single-
images of fat and water protons at 4.7 Tesla (T). The
180?radiofrequency (RF) refocusing pulse of the spin
echo sequence was manipulated such that it was
selectively tuned to either the proton resonant fre-
quencies of fat (primarily ACH2A methyelene protons)
or water to generate separate images of the two
1Ming Hsieh Department of Electrical Engineering, Viterbi School of
Engineering, University of Southern California, Los Angeles,
2Department of Nutrition Sciences, University of Alabama at
Birmingham, Birmingham, Alabama, USA.
3Departments of Preventive Medicine, Physiology & Biophysics, and
Pediatrics, Keck School of Medicine, University of Southern California,
Los Angeles, California, USA.
Drs. Hu and Smith contributed equally to this work.
Contract grant sponsor: NCI Centers for Transdisciplinary Research
on Energeticsand Cancer(TREC);
U54CA116848; Contract grant sponsor: NIDDK; Contract grant
*Address reprint requests to: H.H.H., University of Southern Califor-
nia, Electrical Engineering, 3740 McClintock Avenue, EEB 408, Los
Angeles, CA 90089-2564. E-mail: email@example.com
Received November 5, 2009; Accepted February 9, 2010.
Published online in Wiley InterScience (www.interscience.wiley.com).
JOURNAL OF MAGNETIC RESONANCE IMAGING 31:1195–1202 (2010)
C 2010 Wiley-Liss, Inc.
moieties. To account for T1and T2relaxation bias on
the measured fat and water signals, Lunati et al
acquired multiple series of fat and water images with
varying repetition time (TR) and echo time (TE) values.
From the multiple TR and TE images, the true proton
density of fat and water was extrapolated - curve-fit-
ting, and subsequently a map of the fat–water proton
ratio (fat fraction) was formed. Due to the need for
two separate spin echo acquisitions per TR/TE combi-
nation, the imaging time spent to acquire a single
slice across the interscapular BAT depot was over 20
Recent advances in rapid chemical-shift MRI have
led to the development time-efficient, computationally
robust and accurate fat–water separation techniques.
One example is the IDEAL (Iterative Decomposition
with Echo Asymmetry and Least squares estimation)
algorithm (17,18). A methodological advancement of
IDEAL is an improvement in the accuracy of comput-
ing the proton density fat fraction. IDEAL uses a
multi-echo 3D data acquisition strategy to capture fat
and water signals, and through iterative post-process-
ing, yields an estimate of the fat fraction by systemati-
cally accounting for T1and T*2relaxation effects (19–
21), magnetic field nonuniformity, and the fact that
the fat spectrum is characterized by multiple proton
resonance peaks (22). The resultant fat fraction map
can be used to analyze the relative amounts of fat on
a voxel-wise basis in local regions of interest. IDEAL
(23,24) and similar variants of the technique (25) have
been recently used to study hepatic fat fractions. One
major benefit of IDEAL is its ability to resolve the full
fat fraction range of 0 to 100% without ambiguity.
In this work, we use 3D IDEAL to identify and dif-
ferentiate BAT from WAT based on the unique fat frac-
tions of the two tissues. Due to intracellular charac-
thermogenesis, we hypothesize that its fat fractions
will be significantly lower than in lipid-rich WAT. We
present corroborating results from excised tissues
and in situ whole-mice experiments. To our knowl-
edge, this is the first work to demonstrate the feasibil-
ity of chemical-shift IDEAL-MRI in characterizing
METHODS AND PROCEDURES
Excised tissue samples of interscapular BAT and go-
nadal WAT and 28 postmortem mice carcasses were
prepared and imaged between September 2008 and
November 2009. All animal research was conducted
after approval by the local Institutional Animal Care
and Use Committee. This work involved researchers
from two separate institutions. At the Small Animal
Phenotyping Laboratory of the University of Alabama
at Birmingham, the animals were housed at 22 6 2?C
and were provided ad libitum access to food before
killing. The carcasses were then immediately shipped
by overnight courier service to investigators at the
University of Southern California, where MRI experi-
ments were conducted on a 3T whole-body human
MRI scanner (Signa HDx, GE Healthcare, Waukesha,
WI, USA) using a single-element wrist receiver coil
(BC-10, Mayo Clinic, Rochester, MN, USA). All sam-
ples were imaged within 24 h after killing. At the time
of imaging, postmortem carcasses ranged in age from
2 to 19 weeks. All samples were set aside to equili-
brate to room temperature before imaging.
A 3D six-echo IDEAL-T*2-SPGR pulse sequence was
used in this work (22). The sequence was an investi-
gational research version of the IDEAL software and
was provided by GE Healthcare. Imaging parameters
for the excised tissue samples were TR ¼ 10 ms, TE
spacing ¼ 0.8 ms, first TE ¼ 1.5–1.8 ms, flip angle ¼
5?, receiver bandwidth ¼ 6 125 kHz, 0.6 mm true
(non–zero-interpolated) isotropic spatial resolution,
field-of-view (FOV) ¼ 15 cm, fractional phase FOV ¼
0.75, and an echo train length (ETL, number of ech-
oes per TR) of one. For each TE, the number of signal
averages (NSA) was set to six to further enhance
image signal-to-noise ratio. All excised tissue samples
were imaged together in a Petri dish, using one IDEAL
data acquisition and 16 slices. Imaging time for the
excised specimens was 12 min. For the postmortem
animal carcasses, a 3D imaging volume was pre-
scribed to encompass the entire carcass, requiring 34
to 56 slices. Similar FOV, spatial resolution, and
imaging parameters were used, with the exception of
TR ¼ 20 ms, ETL ¼ 2 with fly-back unipolar readout
gradients, and NSA ¼ 3. The wrist coil was large
enough to allow four animals to be imaged simultane-
ously per IDEAL data acquisition. Imaging time for
each set of four animals was a little over 20 min. The
multipeak T*2-IDEAL algorithm (22) was implemented
on the scanner console using precalibrated amplitude
(22,26,27). Co-registered water, fat, in-phase (water þ
fat), out-of-phase (water ? fat), T*2, and noise-bias-
corrected fat fraction data sets (20) were generated
online by the IDEAL research software.
In addition to IDEAL, single-voxel proton MRS was
performed on two excised BAT and WAT samples to
clearly demonstrate the differences in fat fraction
between the two tissues. Resulting spectra were ana-
lyzed with the Java-based MRUI software (28). The
areas under the fat and water spectral peaks were
computed to generate a similar fat fraction metric. Pa-
rameters for MRS were as follows: TR ¼ 1.5 s, TE ¼
14 ms, 20 ? 20 ? 20 mm3voxel, 2048 spectral data
points, 2.5 kHz bandwidth, no water suppression,
and eight signal averages. The MRS experiment was
carried out for proof-of-concept purposes only and
thus involved only one TE acquisition. Thus, T*2cor-
rection was not performed on the MRS spectra.
For illustration and anatomical reference, cryosec-
tion was performed on a separate adult mouse that
was not part of the imaged cohort. For cryosection,
the carcass was skinned from the base of the skull to
the bottom of the rib cage, and fixed in a 10% buf-
fered formalin solution overnight at 4?C. The speci-
men was washed with formalin and immersed in a
30% sucrose solution (0.1 M phosphate buffer, pH
?7.4) at 4?C to cryoprotect the soft tissues. The speci-
men was subsequently imbedded in Tissue-TekV
1196Hu et al.
frozen in liquid nitrogen, and stored at ?80?C until
sectioning. All sections were prepared using a Leica
CM1900 Cryostat and photographed with a 7.1 mega-
pixel Canon SD1000 camera.
excised BAT and WAT samples. The photo in Figure
1a shows that the two tissues are visually distinctive
in terms of color. The brown-reddish tone of BAT is
due to its high vascularity and iron-rich content
within the large number of intracellular mitochondria.
IDEAL water and fat images are shown in Figures 1b
and 1c, respectively. Note that the two BAT samples
have markedly lower fat signal intensity (arrows) than
lipid-rich WAT, consistent with lower fat content.
Conversely, BAT exhibited greater water content in
the reconstructed water image. The quantitative dif-
ference between BAT and WAT is evident in the fat
fraction map. The average (6 standard deviation) fat
fractions were as follows (left to right): 92% 6 2%
(WAT), 51% 6 8% (BAT), 93% 6 2% (WAT), and 62%
6 7% (BAT). The BAT sample on the left is from 4-
week-old juvenile mice; the BAT sample on the right
is from 12-week-old adults. Note the uniform appear-
ance of the WAT samples in the fat fraction map, in
contrast to the more heterogeneous nature of BAT.
The extent of variation in fat fraction for BAT was
broader than WAT, possibly due to the fact that the
BAT-labeled samples contained admixtures of both
tissue types. The lowest fat fraction was coincident
with the darkest (most reddish) BAT sample (second
Figure 2 illustrates MRS results from two represen-
tative BAT and WAT samples. Raw spectra are shown.
Note that the WAT sample is predominantly composed
of lipids, exhibiting a strong methylene proton peak at
1.3 parts-per-million (ppm). In contrast, the BAT sam-
ple also contained a significant water signal at 4.7
ppm. Both spectra exhibit other minor fat resonance
peaks (arrows) in addition to the methylene peak.
These include methyl (0.9 ppm), olefinic (5.2–5.5
ppm), diallylic (2.8 ppm), and a-carboxyl (2.1–2.2
1 summarizes IDEALimage results from
ppm) groups. The fat fractions of these WAT and BAT
samples were approximately 92% and 58%, respec-
tively. To facilitate plotting and visualization, the
spectral data have been scaled and normalized. Due
to the small size of the excised tissues and their
placement in small vials that also contained air, sig-
nificant magnetic field inhomogeneity may have been
present. This would have consequently led to the
broader line widths, which can be seen in the WAT
spectrum. Although T*2was not performed, it remains
conceptually evident that the BAT fat fraction is sig-
nificantly lower than that of WAT.
Figure 3a illustrates in situ results from four post-
mortem mice specimens. The same fat fraction color
map is used from Figure 1. A representative coronal
Figure 1. IDEAL results from excised tissues. a: Brown and white adipose tissue (BAT, WAT) are visually distinct. b: Water
image. c: Fat image. d: Fat fraction map. Reconstructed fat image illustrates lower fat content in BAT (arrows) than WAT.
Mean fat fraction for the four samples were (left to right) 92%, 51%, 93%, and 62%. The sample vials were placed in a water-
filled Petri dish. Thus, note that water in the dish is correctly reconstructed to the water image. WAT, white adipose tissue;
BAT, brown adipose tissue.
Figure 2. Single-voxel proton MRS results of two representa-
tive tissue samples. WAT (dashed black) is composed nearly
entirely of lipids, with a dominant methylene (ACH2A)
located at 1.3 ppm. In contrast, BAT (solid gray) exhibits an
appreciable water signal, at approximately 4.7 ppm. Arrows
point to the multiple minor fat resonant peaks that are visi-
ble on either spectrum, including the methyl group (ACH3)
at 0.9 ppm, olefinic group (AHC¼ ¼HCA) at 5.2–5.5 ppm, dia-
llylic group (¼ ¼CHACH2AHC¼ ¼) at 2.8 ppm, and ACH2A a-to-
carboxyl group at 2.1–2.2 ppm. Ppm, parts per million; a.u.,
Fat-Water MRI of Brown Adipose Tissue1197
slice highlights the prominent interscapular BAT de-
pot (red contours in fat images). The top two ani-
mals were 12-week-old fully grown adults. The bot-
tom two animals were 4-week-old juveniles. The
average interscapular BAT fat fractions of the adults
were 64% and 52%, 10–15% greater than that of the
juveniles (42% and 39%). In the fat fraction map,
note the visually distinctive color representation of
BAT (green and yellow shades) versus that of lipid-
rich WAT (red) and lean tissue (dark purple). Figure
3b further showcases the peri-renal and inguinal
adipose tissue depots from one adult (top) and one
juvenile (bottom) mouse surrounding the kidneys.
The colorized fat fraction map has been overlaid
with the grayscale water image. Note the apparent
difference in fat fraction in these depots between the
two animals. The depots in the juvenile exhibit a
mixed appearance of BAT and WAT phenotypes,
whereas that of the adult is nearly entirely composed
Axial fat fraction reformats of the dorsal interscapu-
lar BAT depot (arrows) are shown in Figure 4a for one
adult (top) and one juvenile (bottom) mouse, illustrat-
ing the 0.6-mm high-resolution isotropic nature of the
3D data sets. The distinctive trapezoidal, bi-lobed
shape of theinterscapular
delineated. In the adult specimen, note the evident
layer of WAT above the BAT depot. This WAT layer is
less prominent in the juvenile. For anatomical refer-
ence, Figure 4b shows an axial slice prepared from
cryosection, with arrows identifying the visually dis-
tinct interscapular BAT and WAT depots.
Figure 5a summarizes the average fat fraction
within the interscapular BAT depot as a function of
age for carcasses imaged in this work where age and
body weight data were available. There is anecdotal
evidence suggesting that the interscapular BAT fat
fraction is lower in young mice and increases progres-
sively with age. Figure 5b summarizes the average
interscapular BAT fat fraction across the imaged
specimens, showing a similar trend of BAT with
higher fat fraction in heavier animals. In contrast,
WAT fat fractions in all mice were greater than 90%
(data points not shown). Additionally, Figure 6 illus-
trates the interscapular BAT depot volume computed
from 3D IDEAL data set segmentations as a function
of age and weight, both showing positive correlations.
Note that, in each mouse, there is less than 0.5 ml of
BAT within the interscapular depot.
Figure 3. In situ identification of BAT in mice. a: Reconstructed water and fat images, and fat fraction map, with interscapu-
lar BAT depot (red contours in fat image). b: Fat fraction map overlaid on water image, illustrating the peri-renal and inguinal
adipose tissue depot in one adult (top) and one juvenile (bottom) mouse. Significant differences in the fat fraction maps are
evident. Note particularly that the juvenile depot contains a mixture of WAT and BAT, whereas the adult depot is almost
entirely WAT. The same color map is used from Figure 1.
Figure 4. a: Axial reformats highlighting the interscapular
BAT (gray arrows) depot along the dorsal aspects of an adult
(top) and an juvenile (bottom) mouse. b: For anatomical ref-
erence, a representative photo from axial cryosection (of a
separate animal, same gender, similar age and weight) is
shown. The same color map is used from Figure 1.
1198Hu et al.
This work has demonstrated the feasibility of using
the 3D IDEAL fat fraction metric as a sensitive index
for distinguishing BAT and WAT. Collectively, results
from Figures 1–4 corroborate the hypothesis of this
work—BAT fat fraction is lower than WAT fat fraction.
Due to its involvement in thermogensis, BAT appears
to be uniquely characterized by fat fractions that are
distinct from lipid-rich WAT and lean tissues. Addi-
tional results from Figure 5a suggested lower BAT fat
fractions in young juvenile mice and higher BAT fat
fractions in adult mice. We speculate that this is
potentially indicative of the juveniles’ higher thermo-
genic potential and requirement. Due to their greater
body surface-to-volume ratio, juveniles likely experi-
ence more body heat loss than adults. Consequently,
their BAT is regularly engaged in thermogenesis,
which requires increased lipid metabolism. This heat
demand is gradually reduced in larger adult mice due
to their smaller surface-to-volume ratio. As a result,
adult mice exhibit fattier BAT, which can differentiate
and become more similar in appearance to WAT (1).
Plots from Figure 5b also suggested lower BAT fat
fractions in animals with lower body weights, consist-
ent with the notion that active BAT can metabolize
excess lipids to produce heat. The heat is then dissi-
pated by the animal into the surroundings. Thus, in
past literature, BAT has been suggested as a mecha-
nism of protecting the animal against weight gain and
obesity (3,29,30). Results from Figure 6 showed that,
in our cohort of normal mice, interscapular BAT depot
volume positively correlated with both age and weight.
Furthermore, the small size of the interscapular BAT
depot in each animal emphasizes the tissue’s thermo-
There are several limitations to the 2D spin echo
method used by previously investigators for MRI of
BAT (16). First, two separate chemical-shift-selective
acquisitions were needed to obtain separate fat-only
and water-only images, which is time consuming.
Second, the performance of the spin echo approach
will be less robust at lower magnetic field strengths.
The 180?refocusing pulse used by Lunati et al had a
bandwidth of 500 Hz. This bandwidth would be too
broad and inadequate for use at 1.5 and 3T, where
methylene fat and hydroxyl water resonances are sep-
arated by approximately 210 and 420 Hz, respec-
tively. Consequently, use of a broad pulse would
cause signal leakage between fat and water. Another
disadvantage of using selective fat saturation methods
to quantify fat is its inability to completely saturate
the minor fat peaks that exist near the water peak,
such as the olefinic groups (31). In contrast, the
IDEAL method with multipeak fat spectral modeling is
much more time-efficient; it intrinsically decomposes
Figure 5. a: Plot of average inter-
scapular BAT fat fraction versus
age in mice, suggesting a moder-
ate trend of leaner (lower fat con-
tent) BAT in young mice and
fattier BAT in adult mice. b: Plot
of average interscapular BAT fat
fraction versus body weight. Gray
circles denote animals 2–4 weeks
of age. Black circles denote ani-
mals 9–12 weeks of age.
Figure 6. a,b: Plot of interscapu-
lar BAT depot volume versus age
(a) and weight (b). Both plots
show similar positive correlation
denote animals 2–4 weeks of age.
Black circles denote animals 9–
12 weeks of age.
Fat-Water MRI of Brown Adipose Tissue1199
fat and water signals without requiring spectrally
selective pulses; and it provides a full dynamic range
of 0 to 100% fat fraction that is crucial in differentiat-
ing BAT and WAT.
The most significant advantage of IDEAL is that the
algorithm more accurately models the fat spectrum
with multiple peaks, in contrast to the chemical-shift-
selective spin echo approach which models the fat
spectrum with only a single methylene resonance.
Previous works have shown that the fatty acid chains
of BAT and WAT can be significantly unsaturated with
olefinic protons (27,32). With the single-peak model,
signals from the olefinic protons are often erroneously
assigned to the water component. If unaccounted for,
a considerable underestimation of the true fat fraction
can occur (22,33). In the current implementation of
IDEAL, we used a precalibrated multi-peak spectral
profile based on WAT from human data. While this
gave accurate measures of WAT fat fraction, the corre-
sponding estimates of BAT fat fraction could have
been improved slightly had we adopted a profile
specific to BAT. However, the underlying chemical
composition and ratio of saturated and unsaturated
triglycerides between BAT and WAT is not believed to
be significantly different (27).
The application of MRI to study BAT in humans is
under investigation. Reports using PET/CT and his-
tology have confirmed the presence of BAT in adult
humans (11–14). With PET/CT, BAT activity was iden-
tified by observing false-positive symmetrical uptake
of radionuclide fluorodeoxyglucose with cold tempera-
ture stimulation. One advantage of using MRI will be
its ability to identify BAT based on the tissue’s fat
fraction, regardless of its metabolic activity due to
environmental conditions. MRI is also safer by not
requiring the use and uptake of exogenous radionu-
clides. Another benefit IDEAL-MRI is its relatively
high-spatial-resolution in comparison to other imag-
ing modalities. This allows an investigator to poten-
tially isolate BAT in very small and localized regions.
While studies have demonstrated the presence of BAT
in humans across several anatomic sites and age
ranges (6,34), it has been shown that they may be
buried in highly localized and small islets within
major WAT depots (4). As demonstrated in Figures 3
and 4, IDEAL fat fraction measures can discriminate
and depict heterogeneous adipose tissue mixtures of
interscapular and peri-renal BAT from surrounding
WAT in mice.
Although an isotropic resolution of 0.6 mm was
used in this work, a fundamental limitation of identi-
fying BAT by fat fraction MRI in very small depots,
especially in the presence of WAT, is partial volume
effects. In circumstances where fat has infiltrated
organs and skeletal muscles such as in hepatic stea-
tosis, the regional fat fraction may be similar to the
range occupied by BAT. Furthermore, voxels residing
at the interface of WAT and lean tissue will also mimic
fat fractions that are similar to that of BAT. To mini-
mize false-identification of BAT, assistance from a pri-
ori knowledge of common sites in animals and
humans will be critical (5–7). Therefore, for large BAT
depots such as the interscapular and peri-renal
regions, fat fraction IDEAL-MRI should be sufficient
for BAT characterization. However, for smaller depots
or mixtures of WAT and BAT that is predominantly
WAT, other MRI methods that provide either greater
spatial resolution or an alternative contrast mecha-
nism between BAT and WAT need to be pursued.
Nonetheless, further validation of quantifying BAT in
mass or volume units is needed against gold-standard
references such as histological staining, necropsy,
and chemical assay.
Another limitation of this work was the imaging of
carcass versuslive animals.
involved two groups of investigators at two different
research institutions. One group has expertise in
small animal work, while the other has expertise in
MRI. Investigators who performed the MRI experi-
ments did not have access to a laboratory facility at
their institution for the handling of live animals, and
also did not have sedation equipment for live animal
MRI. Investigators at both institutions are currently
pursuing live animal BAT MRI and human protocols.
The implications of imaging carcasses versus live
specimens include differences in body temperature,
respiratory and cardiac motion, and the presence of
vascular flow. With respect to temperature, the chemi-
cal shift between water and fat may change as a func-
tion of temperature, thereby impacting the resultant
fat fraction measurement. With our investigational
IDEAL software, we did perform several separate
IDEAL reconstructions where
between water and fat was varied between ?410 and
?440 Hz (at 3 Tesla), based on spectral measure-
ments from prescan calibration. In the tested carcass
data sets, we observed less than 2% change in the
resulting BAT fat fraction as a result of the chemical
shift variation. Additionally, the WAT fat fraction
remained greater than 90% in all cases, and the qual-
ity of the separated fat and water images were not sig-
In addition to fat–water imaging, other sensitive
Because BAT is highly vascularized due to the need
to efficiently transport the generated heat throughout
an animal’s body, techniques that measure blood
flow can be developed to exploit the differences in
local blood tissue perfusion rates between BAT and
WAT (35). Again, the use of postmortem carcasses
was a limitation of the present work where perfusion
approaches could not be tested. It is also plausible
that the absence of blood flow in carcasses could
have altered the apparent fat fraction of BAT in com-
parison to in vivo conditions (16). With blood flow,
the fat fraction of BAT is expected to be lower. Yet
another contrast mechanism that can potentially be
weighted imaging. It is speculated that BAT T*2 will
be lower than that of WAT, due to the presence of
iron in the mitochondria of BAT and other cellular
differences. In the current work, we additionally
measured T*2 values in BAT and surrounding WAT
from the carcasses, using T*2 maps that were gener-
ated from the six-echo IDEAL algorithm. For WAT, a
range of 9 to 20 ms T*2 values was measured.
the chemical shift
1200Hu et al.
However, for interscapular BAT, we did not observe
lower T*2 values, but instead measured a range
between 11 and 50 ms. Of interest, when compared
against corresponding fat fractions of BAT, a modest
positive correlation between T*2 and fat fraction was
obtained (r ¼ 0.6). This appears consistent with the
notion that, as BAT becomes more active in lipid me-
tabolism (lower fat fraction), increased local blood
flow and intracellular iron and mitochondria con-
tents would lead to shortened T*2values. The feasibil-
ity of T*2imaging of BAT and WAT should be further
explored, particularly in live animal experiments.
In conclusion, the present work has demonstrated
the feasibility of identifying brown adipose tissue
using rapid IDEAL-MRI. It provides a framework for
using MRI to noninvasively characterize BAT in vivo.
The fat fraction index could potentially enable accu-
rate quantification of BAT volumes in vivo, as well as
the monitoring of cellular differentiation between BAT
and WAT depots as a function of age. Furthermore,
BAT relations to muscle development can be explored,
as recent literature has suggested that brown adipo-
cytes and myoblasts originate from the same stem
cell line (36–38). IDEAL-MRI will be an attractive
method to longitudinally investigate BAT function
and physiology as well, and in the assessment of
The authors gratefully thank Dr. Huanzhou Yu and
Ann Shimakawa from General Electric Healthcare for
providing technical assistance with the IDEAL soft-
ware. H.H.H., M.I.G., and K.S.N. were funded by the
NCI Centers for Transdisciplinary Research on Ener-
getics and Cancer (TREC) and all authors were funded
1. Cinti S. The role of brown adipose tissue in human obesity. Nutr
Metab Cardiovasc Dis 2006;16:569–574.
2. Nedergaard J, Bengtsson T, Cannon B. Unexpected evidence for
active brown adipose tissue in adult humans. Am J Physiol
Endocrinol Metab 2007;293:E444–E452.
3. Himms-Hagen J. Obesity may be due to a malfunctioning of
brown fat. Can Med Assoc J 1979;121:1361–1364.
4. Zingaretti MC, Crosta F, Vitali A, et al. The presence of UCP1
demonstrates that metabolically active adipose tissue in the neck
of adult humans truly represents brown adipose tissue. FASEB J
5. Cannon B, Nedergaard J. Brown adipose tissue: function and
physiological significance. Physiol Rev 2004;84:277–359.
6. Heaton JM. The distribution of brown adipose tissue in the
human. J Anat 1972;112:35–39.
7. Aherne W, Hull D. The site of heat production in the newborn
infant. Proc R Soc Med 1964;57:1172–1173.
8. Hany TF, Gharehpapagh E, Kamel EM, Buck A, Himms-Hagen J,
von Schulthess GK. Brown adipose tissue: a factor to consider in
symmetrical tracer uptake in the neck and upper chest region.
Eur J Nucl Med Mol Imaging 2002;29:1393–1398.
9. Cohade C, Osman M, Pannu HK, Wahl RL. Uptake in supracla-
vicular area fat (‘‘USA-Fat’’): description on 18F-FDG PET/CT.
J Nucl Med 2003;44:170–176.
10. Yeung HW, Grewal RK, Gonen M, Schoder H, Larson SM. Pat-
terns of (18)F-FDG uptake in adipose tissue and muscle: a poten-
tial source of false-positives for PET. J Nucl Med 2003;44:
11. van Marken Lichtenbelt WD, Vanhommerig JW, Smulders NM,
et al. Cold-activated brown adipose tissue in healthy men. N Engl
J Med 2009;360:1500–1508.
12. Virtanen KA, Lidell ME, Orava J, et al. Functional brown adipose
tissue in healthy adults. N Engl J Med 2009;360:1518– 1525.
13. Cypess AM, Lehman S, Williams G, et al. Identification and im-
portance of brown adipose tissue in adult humans. N Engl J Med
14. Saito M, Okamatsu-Ogura Y, Matsushita M, et al. High incidence
of metabolically active brown adipose tissue in healthy adult
humans: effects of cold exposure and adiposity. Diabetes 2009;
15. Sbarbati A, Guerrini U, Marzola P, Asperio R, Osculati F. Chemi-
cal shift imaging at 4.7 tesla of brown adipose tissue. J Lipid Res
16. Lunati E, Marzola P, Nicolato E, Fedrigo M, Villa M, Sbarbati A.
In vivo quantitative lipidic map of brown adipose tissue by chemi-
cal shift imaging at 4.7 Tesla. J Lipid Res 1999;40:1395–1400.
17. Reeder SB, Wen Z, Yu H, et al. Multicoil Dixon chemical species
separation with an iterative least-squares estimation method.
Magn Reson Med 2004;51:35–45.
18. Reeder SB, Pineda AR, Wen Z, et al. Iterative decomposition of
water and fat with echo asymmetry and least-squares estimation
(IDEAL): application with fast spin-echo imaging. Magn Reson
19. Yu H, McKenzie CA, Shimakawa A, et al. Multiecho reconstruc-
tion for simultaneous water-fat decomposition and T2* estima-
tion. J Magn Reson Imaging 2007;26:1153–1161.
20. Liu CY, McKenzie CA, Yu H, Brittain JH, Reeder SB. Fat quantifi-
cation with IDEAL gradient echo imaging: correction of bias from
T1 and noise. Magn Reson Med 2007;58:354–364.
21. Bydder M, Yokoo T, Hamilton G, et al. Relaxation effects in the
quantification of fat using gradient echo imaging. Magn Reson
22. Yu H, Shimakawa A, McKenzie CA, Brodsky E, Brittain JH,
Reeder SB. Multiecho water-fat separation and simultaneous R2*
estimation with multifrequency fat spectrum modeling. Magn
Reson Med 2008;60:1122–1134.
23. Kim H, Taksali SE, Dufour S, et al. Comparative MR study of he-
patic fat quantification using single-voxel proton spectroscopy,
two-point dixon and three-point IDEAL. Magn Reson Med 2008;
24. Reeder SB, Robson P, Yu H, et al. Quantification of hepatic stea-
tosis with MRI: the effects of accurate fat spectral modeling.
J Magn Reson Imaging 2009;29:1332–1339.
25. Yokoo T, Bydder M, Hamilton G, et al. Nonalcoholic fatty liver dis-
ease: diagnostic and fat-grading accuracy of low-flip-angle multi-
echo gradient-recalled-echo MR imaging at 1.5 T. Radiology
26. Ren J, Dimitrov I, Sherry AD, Malloy CR. Composition of adipose
tissue and marrow fat in humans by 1H NMR at 7 Tesla. J Lipid
27. Zancanaro C, Nano R, Marchioro C, Sbarbati A, Boicelli A, Oscu-
lati F. Magnetic resonance spectroscopy investigations of brown
adipose tissue and isolated brown adipocytes. J Lipid Res 1994;
28. Naressi A, Couturier C, Castang I, de Beer R, Graveron-Demilly
D. Java-based graphical user interface for MRUI, a software
package for quantitation of in vivo/medical magnetic resonance
spectroscopy signals. Comput Biol Med 2001;31:269–286.
29. Himms-Hagen J. Impaired thermogenesis and brown fat in obe-
sity. Can J Surg 1984;27:125.
30. Himms-Hagen J. Thermogenesis in brown adipose tissue as an
energy buffer. Implications for obesity. N Engl J Med 1984;311:
31. Bley TA, Wieben O, Francois CJ, Brittain JH, Reeder SB. Fat and
water magnetic resonance imaging. J Magn Reson Imaging 2010;
32. Lunati E, Farace P, Nicolato E, et al. Polyunsaturated fatty acids
mapping by (1)H MR-chemical shift imaging. Magn Reson Med
33. Hines CD, Yu H, Shimakawa A, McKenzie CA, Brittain JH, Reeder
SB. T1 independent, T2* corrected MRI with accurate spectral
Fat-Water MRI of Brown Adipose Tissue 1201
modeling for quantification of fat: validation in a fat-water-SPIO
phantom. J Magn Reson Imaging 2009;30:1215–1222.
34. Tanuma Y, Tamamoto M, Ito T, Yokochi C. The occurrence of
brown adipose tissue in perirenal fat in Japanese. Arch Histol
35. Heim T, Hull D. The blood flow and oxygen consumption of brown
adipose tissue in the new-born rabbit. J Physiol 1966;186:42–55.
36. Seale P, Bjork B, Yang W, et al. PRDM16 controls a brown fat/
skeletal muscle switch. Nature 2008;454:961–967.
37. Crisan M, Casteilla L, Lehr L, et al. A reservoir of brown adipocyte
progenitors in human skeletal muscle. Stem Cells 2008;26:
38. Farmer SR. Brown fat and skeletal muscle: unlikely cousins? Cell
1202Hu et al.