Int. J.

Hyperthermia, May 2010; 26(3): 273282

Non-invasive magnetic resonance thermography during regional

hyperthermia
DEMANN, WALDEMAR WLODARCZYK, JACEK NADOBNY,
LUTZ LU
MIRKO WEIHRAUCH, JOHANNA GELLERMANN, & PETER WUST

Int J Hyperthermia Downloaded from informahealthcare.com by HINARI on 10/20/14

For personal use only.

Department of Radiation Therapy, CVK, Charite Universititatsmedizin Berlin, Berlin, Germany

(Received 16 October 2009; Revised 5 January 2010; Accepted 5 January 2010)

Abstract
Regional hyperthermia is a non-invasive technique in which cancer tissue is exposed to moderately high temperatures
of approximately 4345 C. The clinical delivery of hyperthermia requires control of the temperatures applied. This is
typically done using catheters with temperature probes, which is an interventional procedure. Additionally, a catheter allows
temperature monitoring only at discrete positions. These limitations can be overcome by magnetic resonance (MR)
thermometry, which allows non-invasive mapping of the entire treatment area during hyperthermia application.
Various temperature-sensitive MRI parameters exist and can be exploited for MR temperature mapping. The most popular
parameters are proton resonance frequency shift (PRFS) (D corresponding to a frequency shift of 0.011 ppm, i.e. 0.7 Hz per

C at 1.5 Tesla), diffusion coefficient D (DD/D 23 % per  C), longitudinal relaxation time T1 (DT1 =T1  1% per  C), and
equilibrium magnetisation M0 (DM 0 =M 0:3% per  C). Additionally, MRI temperature mapping based on temperaturesensitive contrast media is applied. The different techniques of MRI thermometry were developed to serve different
purposes.
The PRFS method is the most sensitive proton imaging technique. A sensitivity of 0.5 C is possible in vivo but use
of PRFS imaging remains challenging because of a high sensitivity to susceptibility effects, especially when field homogeneity
is poor, e.g. on interventional MR scanners or because of distortions caused by an inserted applicator. Diffusion-based MR
temperature mapping has an excellent correlation with actual temperatures in tissues. Correct MR temperature
measurement without rescaling is achieved using the T1 method, if the scaling factor is known. MR temperature imaging
methods using exogenous temperature indicators are chemical shift and 3D phase sensitive imaging. TmDOTMA appears
to be the most promising lanthanide complex because it showed a temperature imaging accuracy of <0.3 C.
Keywords: regional hyperthermia (RHT), magnetic resonance imaging (MRI), magnetic resonance thermography, water proton
resonance frequency shift (PRFS)

Introduction
The clinical delivery of hyperthermia treatment
requires control of the temperatures applied. This
is typically done using catheters with temperature
probes, which record temperatures along the catheter
tracks. Temperature monitoring using these probes
is limited to the catheter implantation sites. Magnetic
resonance (MR) thermometry offers the opportunity
to non-invasively collect temperature information
throughout the body including regions where no
catheters are implanted.

MR thermometry has been used for temperature

monitoring since the 1990s [1, 2]. Thermotherapy
interventions can be classified into two categories:
ablative (60 C) and non-ablative. Ablative techniques include radiofrequency ablation (RFA), laserinduced thermotherapy (LITT), focused ultrasound
(FUS), and, in a sense, also cryosurgery (CS).
These techniques are employed to destroy small
tumours by heat (coagulation) or cold. The techniques using high or low temperatures for ablative
treatment are highly focused ultrasound [35], LITT
[69], RFA [10], and CS [11].

Correspondence: Dr. Lutz Ludemann, Klinik fur Strahlentherapie, Campus Virchow-Klinikum, Charite Universitatsmedizin Berlin, Augustenburger Platz 1,
13353 Berlin, Germany. Tel: 49-30-450-557188. Fax: 49-30-450-557979. E-mail: lutz.luedemann@charite.de
ISSN 02656736 print/ISSN 14645157 online 2010 Informa UK Ltd.
DOI: 10.3109/02656731003596242

Int J Hyperthermia Downloaded from informahealthcare.com by HINARI on 10/20/14

For personal use only.

274

L. Ludemann et al.

Hyperthermia (HT) is a non-ablative technique

in which cancer tissue is exposed to moderately
high temperatures of approximately 4345 C to
induce various biochemical, physiological, and
immunological processes. Therapeutic hyperthermia
is currently mainly used to supplement established
treatment modalities such as radiotherapy [12],
chemotherapy, and surgery [13]. Finally, there
are novel therapeutic approaches that are guided or
triggered by temperature manipulations such as
heat-induced gene therapy [14] or targeting methods
like heat-sensitive liposomes [15].
Various temperature-sensitive MRI parameters
exist and can be exploited for magnetic resonance
temperature mapping. MR thermometry indicators
can be endogenous or exogenous. Endogenous
temperature indicators that can be measured by
MRI are (in order of temperature sensitivity): proton
resonance frequency shift (PRFS) (D corresponding
to a frequency shift of 0.011 ppm, i.e., 0.7 Hz, per  C
at 1.5 Tesla), diffusion coefficient D (DD/D 23%
per  C), longitudinal relaxation time T1 (DT1 =T1 
1% per  C), and equilibrium magnetisation M0
(DM0 =M 0:3% per  C) [1, 2, 1618]. Although
the temperature sensitivities given in brackets are
derived from ex vivo data (phantoms and tissue
specimens), it can be assumed that the ranking
of sensitivities also applies to in vivo temperature
mapping. However, the absolute values will vary with
the specific situations in which MRI mapping is
performed (mode of heat delivery, magnetic
field strength, MR pulse sequence used, spatial
and temporal resolution, volume coverage, etc.).
MRI temperature mapping based on exogenous
temperature indicators uses temperature-sensitive
contrast medium.
The different techniques of MRI thermometry
were developed to serve different purposes. When
used to monitor fast ablation techniques (RFA,
LITT), high temporal and spatial resolution is the
most important feature. Qualitative monitoring of
the thermal effects is considered sufficient for these
applications; more accurate temperature measurement would be of interest only at the margin of the
necrosis induced. Highly focused ultrasound ablation is usually not an option for lesions in moving
organs or requires tracking by fast MRI temperature
monitoring to follow organ motion. While interstitially placed RFA or LITT applicators need not be
readjusted to the target region during treatment,
it is nevertheless necessary to register the dynamically
acquired MR images to reference baseline images
in order to yield quantitative MRI temperature maps.
In the monitoring of regional hyperthermia (RHT)
treatment, a high temperature resolution (<1 C) is

the most desirable feature of temperature mapping.

This is impaired by susceptibility artefacts caused
by the applicators and by motion artefacts [18].

MRI methods using exogenous indicators

MR temperature imaging methods using exogenous
temperature indicators exploit various temperaturedependent mechanisms to measure the temperaturedependent effect of an injected contrast agent.
The relation between a thermosensitive MRI parameter and temperature may be linear or non-linear.
Substances with a linear temperature dependence are
so-called shift reagents, which contain a molecular
group that induces a linear chemical shift of spectral
lines [1921]. For spatial encoding of the temperature information classical 2D chemical shift imaging
(CSI) and 3D phase sensitive imaging were used.
The complex of praseodymium-2-methoxyethylDO3A (Pr-MOE-DO3A) has been investigated
as an exogenous temperature indicator in 1H MR
spectroscopy (MRS) [22, 20]. Like other lanthanides, Pr strongly influences the 1H MR spectrum
when exposed to varying temperatures. Compared
with gadolinium as a central atom of MR contrast
media, Pr is only weakly paramagnetic and, therefore, more suitable as a basis for synthesis of
temperature indicators. The temperature-dependent
functional group of the Pr probe is the methoxy
group (OCH3), which has a thermal sensitivity of
0.120.13 ppm  C1.
The thulium-based complex TmDOTP5 is also
potentially useful as a MR thermometric probe but
cannot be used in human studies because it produces
a slight drop in blood pressure [23, 24]. Zuo and
colleagues demonstrated that TmDOTP5 has more
than a 100-fold temperature sensitivity as compared
to water for both proton and phosphorus signals.
This complex has six groups of four equivalent
protons and a group of four equivalent phosphorous.
Among all the proton signals, the H6 signal from the
side chains is the narrowest and most suitable for
MR thermometry [19]. It has a thermal sensitivity of
0.87 ppm  C1 and is used in animal studies [21].
The thulium-based complex TmDOTA was
evaluated from MR thermometry by Hekmatyar
et al. [25], Zuo et al. [26], and James et al.[27].
TmDOTA is expected to be non-toxic in-vivo
because the gadolinium complex of DOTA4 is a
widely used MRI contrast agent clinically. Similar to
TmDOTP5, the chemical shifts of the proton signal
from TmDOTA are hundreds of ppm away from
water, and the thermal sensitivity is TmDOTA is
60 times higher than water proton sensitivity.
The H5 signal from the side chains has a temperature

Int J Hyperthermia Downloaded from informahealthcare.com by HINARI on 10/20/14

For personal use only.

Non-invasive magnetic resonance thermography during regional hyperthermia

coefficient of 0.91 ppm  C1 [25, 26]. All the proton
signals from TmDOTA show extremely short
relaxation times. The short T2 s are a drawback
because they broaden the line width, which reduces
the accuracy of measuring chemical shifts. In
addition, the short T2 s require short echo times
in imaging experiments and, hence, put more
demand on the gradient hardware.
Lanthanide complexes of the methyl-substituted
analogue of DOTA4, DOTMA4 provide a more
intense proton signal compared to the other macrocyclic lanthanide complexes proposed for MR
thermometry because DOTMA4 contains twelve
magnetically equivalent protons on the four methyl
groups [21]. In addition to the higher signal
intensity, the methyl 1H signals have significantly
narrower line width and longer T2 compared to
the corresponding proton signals from DOTA4
complexes. The combination of higher signal intensity and narrower line width results in a four-fold
increase of signal-to-noise of DOTMA4 complexes
with respect to DOTA4 complexes [28]. Aime et al.
demonstrated the use of YbDOTMA for MR
thermometry [29] but the temperature coefficient
for the methyl resonance of the complex is only
0.04 ppm  C1, which is only four times more than
the water proton signal.
Hekmatyar et al. evaluated the temperature sensitivity of Pr3, Yb3, Tm3, Tb3, and Dy3
complexes of DOTMA4 [28]. The proton chemical
shift of the methyl signal from TmDOTMA is
approximately 100 ppm away from the water signal,
thus, selective images of the complexes can be easily
obtained with a frequency selctive RF excitation
pulse. The feasibility of monitoring absolute temperature with TmDOTMA has been demonstrated
in vivo in mice [28] and rats [27]. Among the
lanthanide complexes proposed for MR thermometry, TmDOTMA appears to be the most promising
[21]. A comparison of regional temperatures measured with TmDOTMA showed that the accuracy
of temperature is <0.3 C [27].
The substances with non-linear temperature
dependence include conventional contrast agents
entrapped in liposomal carriers. These agents do
not interact with tissue water until they are released
from the carriers at a given temperature. Once
released, the agent significantly alters the T1 relaxation rate of the tissue [30]. Another class of
exogenous non-linear temperature-sensitive indicators is constituted by molecular complexes with a
covalent bond that are bi-stable between diamagnetic
and paramagnetic states in a temperature-dependent
manner [31]. An in vivo study is presented by
Lindner et al. [30]. A qualitative summary of the

275

principles and performance of the methods that

have been proposed is given by Quesson et al. [2]
and especially for paramagnetic lanthanide complexes by Hekmatyar et al. [21].

Diffusion imaging
In general, a linear relationship between diffusion
coefficient D and temperature (via thermal Brownian
motion) is given by the StokesEinstein relation.
In the physiological temperature range the change
in diffusion is approximately 2.4%/ C [32]. Nearly
all types of MR pulse sequences can be sensitised
for diffusion by applying a pair of magnetic field
gradient pulses [33], which cause a phase off-set
in spins randomly moving along these gradients.
The attenuation of the echo signal, S, can then be
expressed by
1
S S0 ebD
where S0 is the signal in the absence of diffusion
gradients and b is a factor for diffusion sensitisation. For rectangle diffusion gradients, b can be
calculated by
2
b G2 D  =3
where G is the strength,  the duration, and D the
separation of the diffusion gradient pulses. An
accurate multipoint fit of D requires several acquisitions of the same pulse sequence with several b
values. In tissue, the relationship between signal
S and b is often non-exponential [34]. Unfortunately,
diffusion measurement using different b-values
excessively prolongs the total acquisition time.
Therefore, echo planar read out is preferred for
diffusion imaging which is highly sensitive for
susceptibility artefacts.
In anisotropic media, molecular mobility may
depend on the direction of diffusional motion, and
therefore diffusion has to be measured in at least
three major directions. Usually, a conventional echo
planar image (EPI) without (b 0) and diffusionweighted EPI with diffusion gradients (b  500 in the
body and b  1000 in the brain) with at minimum
three directions are used. The b-value is given by the
strength and duration of the diffusion gradients,
see Equation 2. Often, diffusion tensor imaging
(DTI) is applied acquiring diffusion-weighted images
in 20 or more directions to obtain a more accurate
picture of diffusion anisotropy. DTI allows calculating the full diffusion tensor, which describes the
diffusion anisotropy, but this method requires
more acquisition time than conventional diffusion
measurement in three directions. At least six directions and an image without diffusion-weighting
(b 0) are necessary for DTI analysis. Since most

Int J Hyperthermia Downloaded from informahealthcare.com by HINARI on 10/20/14

For personal use only.

276

L. Ludemann et al.

tumours do not show diffusion anisotropy, DTI

offers little advantage for temperature mapping.
Use of diffusion-weighted imaging for assessment
of temperature requires the measurement of relative
changes in diffusion values with respect to a
reference image. The reference image has to be
acquired before switching on the power for heat
delivery. Therefore, motion is a potential source
of error since there is often an interval of up to 1 h
between switching the power on for heat delivery
and measuring temperatures. Additionally, the diffusion signal may also be affected by perfusion [16].
Accurate and fast diffusion-based thermography
can be performed when using EPI [35], but this
requires fast gradient hardware [36, 37]. EPI enables
the acquisition of a number of slices with diffusion
gradients at an acceptable spatial resolution during
a breath-hold. Unfortunately, EPI is very sensitive
to susceptibility artefacts and is preferably used for
brain imaging. Typical artefacts arise from susceptibility changes and distortions at the boundary of the
field of view (FOV), resulting in distorted temperature images in the body [35]. For body applications,
segmented EPI is preferred, which is less sensitive to
susceptibility artefacts because it uses a shorter echo
train length. Segmented diffusion-weighted EPI is
available for body MRI on state-of-the-art MRI
scanners,
but
increases
acquisition
time.
Additionally, diffusion-weighted imaging should be
combined with fat saturation to eliminate ghost
artefacts resulting from a chemical shift yielding to
ghost images several cm distant from the expected
position. Additionally, lipid suppression is necessary
because fat has a different change in diffusion
coefficient with temperature.

Spin-lattice relaxation time T1

Spin-lattice relaxation time T1 is another parameter
that can be explored for temperature mapping. The
relationship between T1 and small temperature
variations was found to be approximately linear
with sensitivities of 0.8%/ C to 2%/ C [3841].
The quality of T1 -based MR thermography depends
on the accuracy with which T1 itself can be
determined. Additionally, the sensitivity depends
on the tissue. It has also to be taken into account
that the spin-spin relaxation time T2 is also a function
of the temperature. The temperature dependence,
however, is relatively small [42] and usually
neglected.
Conventional methods, which determine T1 by
fitting a recovery curve to a series of measurements,
such as the multipoint inversion recovery (IR) or the
repeated saturation recovery (SR) methods based on
spin echo sequences, are the most accurate because

prepared sequences explore the largest dynamic

signal range. Since multiple acquisitions are required
these methods are also the most time-consuming
approaches [43].
Methods based on only a few measurements, such
as spin echo (SE) with two different repetition times,
TR, or gradient echo (GE) with different flip angles,
produce inaccurate T1 estimates [44]. The recently
reported rapid inversion prepared gradient echo T1
mapping techniques, known as snapshot or
TurboFLASH, significantly speed up the acquisition process, but they either suffer from a large
uncertainty of measured T1 [45] or need excessive
post-processing [46]. Improving the accuracy of T1
for these methods again requires a greater number
of measurements [47].
Also, pulse sequences with a strongly T1 -weighted
gradient echo have occasionally been applied for fast
temperature monitoring, especially in MR-guided
thermoablative interventions with high temperature
increases [48]. These methods are neither sensitive
nor accurate enough for monitoring small temperature changes during hyperthermia treatment [49].
An imaging version of the spectroscopic Look
and Locker technique, which has the potential to
combine high accuracy with moderate acquisition
times [50], was used as a T1 -based MR thermography method by Wlodarczyk et al. [16]. The
method known as TOMROP (T one by multiple
read out pulses) applies an inversion pulse followed
by a train of evenly spaced interrogation pulses of
small flip angle [51]. The total acquisition time of the
TOMROP method is approximately 4 min [16].
T1 measurement using EPI requires fast gradient
hardware and is inherently prone to susceptibility
artefacts especially in the body. This limitation might
also be overcome by segmented EPI acquisition.
Combining the EPI technique with T1 inversion
allows very fast T1 -weighted imaging [52]. Using an
acquisition scheme with optimised rearrangement of
the slices acquired with each inversion preparation,
a stack of 60 slices can be acquired in only about
3 min [53].
Dynamic contrast-enhanced MRI (DCE-MRI)
explores a fast method for dynamic T1 mapping.
Li et al. proposed a method based on a baseline scan
consisting of multiple unprepared 3D gradient echo
sequences acquired with different flip angles and a
dynamic scan acquired with a single large flip angle
[54]. Brix et al. describe a similar approach but use
an optimised saturation recovery TurboFLASH
sequence with different saturation times for the
baseline scan [55]. These methods might be appropriate to monitor the temperature immediately after
switching the power on. However, these methods
have a lower sensitivity because they explore a larger
dynamic T1 relaxation range and therefore might not

Int J Hyperthermia Downloaded from informahealthcare.com by HINARI on 10/20/14

For personal use only.

Non-invasive magnetic resonance thermography during regional hyperthermia

be appropriate to capture the small changes in T1
relaxation rate encountered during RHT.
Accurate T1 measurement is difficult and may be
impaired by inhomogeneities of the B1 field and/or
imperfections of excitation pulses, especially as their
frequency blurs in the presence of the slice selection
gradients or drifts during heating. Quantitative
thermometry based on the measurement of T1
relaxation is further limited by the fact that it is
subject to hysteresis and the relationship is linear
only over a small temperature range of approximately
4345 C [56]. Deviations are especially pronounced
in the presence of tissue alterations (e.g., coagulation) [57]. In these cases, quantitative T1 measurements are used to monitor temperature development
starting at the margin of the necrosis generated [58].
A threshold temperature of 4345 C is suitable to
monitor HT. Some difficulties may arise from the
fact that the relationship between T1 relaxation and
temperature strongly depends on the tissue in which
the measurement is performed.
Changes in T1 relaxation generated in vivo depend
not only on the temperature but also on a thermoregulatory redistribution of oxygenation [59]. Using
T1 relaxation to monitor HT delivery on low-field
MR scanners, a strikingly higher inaccuracy of T1
calibration was found in muscle compared with fatty
tissue in vivo, suggesting more marked redistribution
of oxygenation [60].
T1 and the T1 change per  C increase with
increasing field strength [61], but T1 contrast
diminishes [62]. Thus, apart from SNR advantages
at higher field, T1 -based temperature mapping using
T1 -weighted images appears more sensitive at low
field [2]. T1 weighted images can be acquired with
relatively motion-insensitive pulse sequences, but
image registration between successive images has to
be ensured. Thus qualitative change in T1 relaxation
evaluation is often used to guide thermoablative
interventions; however, the temperature dependence
of equilibrium magnetisation, M0 , is rarely considered, although it accounts for as much as one third
of the temperature sensitivity of T1 relaxation [63].
In general, the methods measuring T1 relaxation are
not sensitive enough to monitor regional HT.

Water proton resonance frequency shift

The most common MR imaging method used to
measure temperature changes is the proton resonance
frequency shift (PRFS) method. With increasing
temperature the screening effect of bounded electrons
increases, resulting in a lower local magnetic field,
and consequently in a negative shift of water
resonance frequency. The average electron screening
constant of pure water varies approximately linearly

277

with temperature over a wide range in temperatures

from 15 C to 100 C [64], including the temperature range of interest for interventional procedures.
The resulting shift in the proton resonance frequency
has been reported to be 0.0107 ppm C1 in pure
water [64, 65] and in the range between 0.007 and
0.009 ppm C1 in muscle and other organ tissues
with a linear relationship up to 50 C [66, 1] and
can be used to calculate the temperature change.
The temperature-induced shift in water PRF is
usually extracted from the phase difference of two
gradient echoes [67]. Phase images are acquired
before and after heating and subtracted to obtain
phase difference images, which are proportional to the
temperature-dependent PRFS. The temperaturedependent phase difference can be written as
D
B0 T TE
DT

where /(2) 42.58 MHz/T is the gyromagnetic

ratio of hydrogen, B0 is the field strength of the main
magnetic field, (T) is the temperature-dependent
PRFS in ppm/ C, and TE is the echo time for image
acquisition between excitation and resonance.
The PRFS method provides the highest sensitivity
to temperature changes of the proton imaging
techniques. Except for fatty tissue, the type of
tissue has only little effect on PRFS. In water, the
dependence of the PRFS on temperature is attributed to changes in the hydrogen bonds, which are
absent in fat [64]. The resulting temperature
sensitivity of fat is some orders of magnitude smaller
[68], indicating that thermometry inside fatty tissue
is difficult. The fact that the lipid resonance
frequency is almost temperature-independent poses
an important problem for temperature measurement
using PRFS because many biological tissues are
composed of both water and fat.
Therefore, two problems arise when PRFS imaging is used in fatty tissue. First, gradient echo
sequences are hampered by the typical artefacts
arising from the spatial misregistration of fatty
areas, which is due to chemical shifts between fat
and water molecules. Second, temperature maps
are degraded by partial volume effects because fat
molecules modify the PRFS and thus lead to
temperature errors. Partial volume effects primarily
occur in mixed tissue (fat/parenchyma) and therefore
will degrade PRF imaging when used to map
temperatures. These partial volume effects can be
markedly reduced by applying frequency-selective
slice excitation [69] or by additionally inducing an
intrasequential phase difference between gradient
echo and spin echo in a combined gradient echo
spin echo pulse sequence [70]. Another approach to
reduce the error due to small amounts of fat (<20%)
is the combination of temperature maps acquired at

Int J Hyperthermia Downloaded from informahealthcare.com by HINARI on 10/20/14

For personal use only.

278

L. Ludemann et al.

three different echo times [71]. In the presence of fat

the measured temperature maps become dependent
on the echo time. Overestimation and underestimation of the temperature changes approximately every
, suggesting that a combination of different echo
times with adjusted weighting reduces the error
in the temperature measurement.
The additionally disturbing effects are multi-fold
because of a high sensitivity for susceptibility effects
[18]. As for the other MRI temperature measuring
techniques, motion is a problem since a reference
scan must be obtained before power is switched on
for heat delivery. De Senneville proposed techniques
for correction of periodic motion [17].
A major difficulty is the drift or slow change of the
B0 field over time during a series of temperature
measurements, which may span one to two hours
when used to monitor hyperthermia therapy, since
the phase shift is determined with respect to a
baseline measurement performed before switching
the power on. The maximum magnetic field drift
quoted for a modern commercial MRI system
magnet is 0.1 ppm /h, meaning that a temperature
error of up to 10 C can occur during one hour
of hyperthermia treatment due to B0 drift alone,
if not corrected, which is larger than the maximum
temperature change induced by hyperthermia
treatment.
Since it is impossible to predict or extrapolate the
B0 shift [72], many attempts have been made to
correct for the drift by using reference probes.
Water gelatine [73], a saline-filled hose [74], body
fat [7577], external water [7577], and external oil
[72] were used as a temperature reference. A major
advantage of oil and fat references for drift correction
is that their MR resonance frequency change with
temperature is nearly zero in comparison to that of
water [78]. While using body fat has some potential
error due to the presence of a small percentage of
water in adipose tissue, use of an external oil probe
inside the hyperthermia applicator minimises any
resonance shift in the reference [79] and thus allows
temperature assessment with an error 1 C [72].
The use of reference probes is an option to correct
for the B0 drift. Unfortunately, the use of a single
reference probe is insufficient and up to eight
probes are required for optimal correction [72].
Alternatively, the water bolus surrounding the
patient serves as a reference [35]. The use of multiple
references requires interpolation between the reference probes in order to yield information inside
the patients body. A multi-parametric polynomial
fit is used including linear, quadratic [72], and
probably cubic terms [75]. However, the PRFS
method is limited by the fact that the results obtained
may represent actual temperature changes and the
superimposed effects of perfusion changes due to

thermoregulation and vascular pulsation. Such interference with the PRFS technique has even been
observed in the extremities [59, 80].
Changes in susceptibility due to the so-called
BOLD effect (blood oxygenation level dependent)
may also affect PRFS imaging. The oxygen level
of a tissue responds to heating for several reasons
including a global increase in perfusion resulting
from an increase in cardiac output. A flow increase
in the aorta of 50100% during HT is demonstrated
in Figure 1 and this also increases perfusion in the
target region. The increased blood flow is associated
with a higher oxy-haemoglobin level. As a result,
diamagnetic susceptibility is increased, reducing
the local magnetic field and, like the temperature
rise, will cause a negative PRFS. In addition,
local thermoregulation will lead to a redistribution
of perfusion with a corresponding redistribution
of
diamagnetic
oxy-haemoglobin
(oxy-Hb).
Additionally, vasodilation may result in an increase
in the blood volume, and the resulting change in the
absolute amount of deoxy-Hb can in turn affect
phase. The ratio of oxy-Hb to deoxy-Hb may also
vary with temperature-related changes in oxygen
binding to haemoglobin (shift of the dissociation
curve) [81].
More marginal interference may result from
changes in the electrical conductivity of a tissue
exposed to heat. A method to correct for this
interference was proposed by Peters et al. [82]. The
latter is the reason for relevant temperature errors
in homogeneous phantoms filled with large amounts
of media high electrical conductivity [35]. The
temperature measurement accuracy can be further
improved by use of a double gradient echo pulse
sequence [35], which will reduce conductivity
artefacts. This is accomplished by subtracting
the phases of both echo times (45 ms, 20 ms) so
that the position-dependent phase differences
cancel. Therefore, the double echo method
yields the best temperature resolution (0.5 C),
see Figures 2 and 3.
PRFs imaging is limited by the need for a baseline
phase map to serve as a reference for the temperature
measurements performed during heat application. In
the clinical setting of hyperthermia treatment, up to
1 h may elapse between acquisition of the reference
measurement and treatment monitoring. This is why
the spectroscopic version of PRFS measurement is so
attractive, which uses a temperature-independent
spectral line, e.g., of lipid, as an internal reference
[83, 84]. The shift measured between the water peak
and a reference peak that remains constant with
temperature, such as lipids [78] or N-acety-aspartate
in the brain [85]. The internal reference makes the
spectroscopic PRFS relatively insensitive to field

Int J Hyperthermia Downloaded from informahealthcare.com by HINARI on 10/20/14

For personal use only.

Non-invasive magnetic resonance thermography during regional hyperthermia

279

Figure 1. Systolic blood flow velocities measured by phase-contrast MRI in the large abdominopelvic arteries before
and 30 min after RHT of the cervix. Note that flow velocity in the large pelvic arteries (iliac and femoral arteries), which give
off smaller branches to the pelvic muscles, increases by 100% after hyperthermia. There is only little change in flow velocity
in the inferior mesenteric artery, the superior rectal artery, and the two sigmoid arteries, which supply highly perfused organs
(large intestine and rectum).

Figure 2. HT treatment in a patient with sarcoma. Morphologic image (A) and temperature map with superimposed 42 C
isotherms. Also shown is the catheter for insertion of a temperature sensor (B).

Figure 3. Comparison of the temperatures obtained using the PRFS technique with the temperatures measured by a sensor
along the catheter tract in a patient with soft tissue sarcoma of the calf.

Int J Hyperthermia Downloaded from informahealthcare.com by HINARI on 10/20/14

For personal use only.

280

L. Ludemann et al.

drifts and interscan motion and allows for absolute

temperature measurement.
Mapping of the spatial distribution of the chemical
shift of protons induced by temperature changes
is accomplished by extending the spectroscopic
approach to spectroscopic imaging. Proton chemical
shift imaging (CSI) sequences have been shown to
measure absolute temperature distributions within
1 min or less on a spatial resolution of 34 mm [84].
Different acquisition methods have been proposed
for spectroscopic temperature measurement, such
as single voxel spectroscopy, magnetic resonance
spectroscopic imaging (MRSI), echo planar
spectroscopic imaging (EPSI), and line scan echo
planar spectroscopic imaging (LSPESI) [83, 86].
As mentioned earlier, echo planar imaging is very
sensitive to susceptibility artefacts. A drawback of
this method, as with any spectroscopic technique,
is the poor signal-to-noise ratio on conventional MR
scanners, which is most deleterious in combination
with the inhomogeneities occurring in body trunk
imaging or hybrid MRI-hyperthermia systems.

Conclusions
Among the lanthanide complexes proposed for MR
thermometry, TmDOTMA appears the most promising. Chemical shift imaging of the methyl signal
from TmDOTMA allows temperature mapping
with a resolution <0.3 C and the complex can be
imaged at physiologically acceptable doses (0.5
1.0 mmol kg1) in vivo in only a few minutes.
Employing proton imaging techniques, the temperature sensitivity of PRFS imaging can be increased far
above that of other MR proton imaging temperature
mapping methods by using slightly longer echo
times, which are feasible when monitoring slow
interventions such as RFA or LITT and even more
so in conjunction with HT treatment. A sensitivity
of 0.5 C is therefore also possible in vivo [35, 72]
but use of PRFS imaging remains challenging
because of a high sensitivity to susceptibility effects.
Diffusion-based MR temperature mapping has an
excellent correlation with actual temperatures in
tissues. The coefficient of determination and temperature resolution are nearly as good as those of the
PRFS method (0.6 C) but image noise is higher
[35]. Correct MR temperature measurement without
rescaling is achieved using the T1 method, if the
scaling factor is known. When the field homogeneity
is poor, e.g. on interventional MR scanners or
because of distortions caused by an inserted applicator, the PRFS method may not be as appropriate as
diffusion or T1 relaxation, which can be acquired
with spin echo methods.

Acknowledgement
The authors are very thankful for the support of the
Berliner Sparkassenstiftung Medizin.
Declaration of interest: The authors report no
conflicts of interest. The authors alone are responsible for the content and writing of the paper.

References
1. Lewa CJ, de Certaines JD. Body temperature mapping by
magnetic resonance imaging. Spectroscoy Letters 1994;27:
13691419.
2. Quesson B, de Zwart JA, Moonen CT. Magnetic resonance
temperature imaging for guidance of thermotherapy. J Magn
Reson Imaging 2000;12:525533.
3. Salomir R, Palussie`re J, Vimeux FC, de Zwart JA, Quesson B,
Gauchet M, Lelong P, Pergrale J, Grenier N, Moonen CT.
Local hyperthermia with MR-guided focused ultrasound:
Spiral trajectory of the focal point optimized for temperature
uniformity in the target region. J Magn Reson Imaging
2000;12:571583.
4. Hynynen K, McDannold N, Mulkern RV, Jolesz FA.
Temperature monitoring in fat with MRI. Magn Reson Med
2000;43:901904.
5. McDannold N, King RL, Hynynen K. MRI monitoring of
heating produced by ultrasound absorption in the skull:
In vivo study in pigs. Magn Reson Med 2004;51:10611065.
6. Vogl TJ, Mack MG, Muller P, Phillip C, Bottcher H,
Roggan A, Juergens M, Deimling M, Knobber D, Wust P.
Recurrent nasopharyngeal tumors: Preliminary clinical results
with interventional MR imaging-controlled laser-induced
thermotherapy. Radiology 1995;196:725733.
7. Kettenbach J, Silverman SG, Hata N, Kuroda K,
Saiviroonporn P, Zientara GP, Morrison PR, Hushek SG,
Black PM, Kikinis R, et al. Monitoring and visualization
techniques for MR-guided laser ablations in an open MR
system. J Magn Reson Imaging 1998;8:933943.
8. Weidensteiner C, Quesson B, Caire-Gana B, Kerioui N,
Rullier A, Trillaud H, Moonen CT. Real-time MR temperature mapping of rabbit liver in vivo during thermal ablation.
Magn Reson Med 2003;50:322330.
9. Rieke V, Vigen KK, Sommer G, Daniel BL, Pauly JM,
Butts K. Referenceless PRF shift thermometry. Magn Reson
Med 2004;51:12231231.
10. Chen JC, Moriarty JA, Derbyshire JA, Peters RD,
Trachtenberg J, Bell SD, Doyle J, Arrelano R, Wright GA,
Henkelman RM, et al. Prostate cancer: MR imaging and
thermometry during microwave thermal ablation-initial
experience. Radiology 2000;214:290297.
11. Wansapura JP, Daniel BL, Pauly J, Butts K. Temperature
mapping of frozen tissue using eddy current compensated half
excitation RF pulses. Magn Reson Med 2001;46:985992.
12. van der Zee J, Gonzalez Gonzalez D, van Rhoon GC,
van Dijk JD, van Putten WL, Hart AA. Comparison of
radiotherapy alone with radiotherapy plus hyperthermia in
locally advanced pelvic tumours: A prospective, randomised,
multicentre trial. Dutch Deep Hyperthermia Group. Lancet
2000;355:11191125.
13. Rau B, Wust P, Tilly W, Gellermann J, Harder C, Riess H,
Budach V, Felix R, Schlag PM. Preoperative radiochemotherapy in locally advanced or recurrent rectal cancer:
Regional radiofrequency hyperthermia correlates with clinical
parameters. Int J Radiat Oncol Biol Phys 2000;48:381391.

Int J Hyperthermia Downloaded from informahealthcare.com by HINARI on 10/20/14

For personal use only.

Non-invasive magnetic resonance thermography during regional hyperthermia

14. Lohr F, Hu K, Huang Q, Zhang L, Samulski TV,
Dewhirst MW, Li CY. Enhancement of radiotherapy by
hyperthermia-regulated gene therapy. Int J Radiat Oncol Biol
Phys 2000;48:15131518.
15. Needham D, Anyarambhatla G, Kong G, Dewhirst MW.
A new temperature-sensitive liposome for use with mild
hyperthermia: characterization and testing in a human tumor
xenograft model. Cancer Res 2000;60:11971201.
16. Wlodarczyk W, Hentschel M, Wust P, Noeske R, Hosten N,
Rinneberg H, Felix R. Comparison of four magnetic
resonance methods for mapping small temperature changes.
Phys Med Biol 1999;44:607624.
17. Denis de Senneville B, Quesson B, Moonen CT.
Magnetic
resonance
temperature
imaging.
Int
J
Hyperthermia 2005;21:515531.
18. Rieke V, Butts Pauly K. MR thermometry. J Magn Reson
Imaging 2008;27:376390.
19. Zuo CS, Bowers JL, Metz KR, Nosaka T, Sherry AD,
Clouse ME. TmDOTP5: A substance for NMR temperature
measurements in vivo. Magn Reson Med 1996;36:955959.
20. Hentschel M, Wust P, Wlodarczyk W, Frenzel T, Sander B,
Hosten N, Felix R. Non-invasive MR thermometry by 2D
spectroscopic imaging of the Pr[MOE-DO3A] complex.
Int J Hyperthermia 1998;14:479493.
21. Hekmatyar SK, Kerkhoff RM, Pakin SK, Hopewell P,
Bansal N. Noninvasive thermometry using hyperfine-shifted
MR signals from paramagnetic lanthanide complexes. Int J
Hyperthermia 2005;21:561574.
22. Frenzel T, Roth K, Kossler S, Raduchel B, Bauer H, Platzek J,
Weinmann HJ. Noninvasive temperature measurement
in vivo using a temperature-sensitive lanthanide complex
and 1H magnetic resonance spectroscopy. Magn Reson Med
1996;35:364369.
23. Seshan V, Germann MJ, Preisig P, Malloy CR, Sherry AD,
Bansal N. TmDOTP5 as a 23Na shift reagent for the in vivo
rat kidney. Magn Reson Med 1995;34:2531.
24. Winter PM, Bansal N. TmDOTP5 as a (23)Na shift reagent
for the subcutaneously implanted 9L gliosarcoma in rats.
Magn Reson Med 2001;45:436442.
25. Hematyar SK, Poptani H, Babsky A, Leeper DB, Bansal N.
Non-invasive magnetic resonance thermometry using
thulium-1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetate
(TmDOTA). Int J Hyperthermia 2002;18:165179.
26. Zuo CS, Mahmood A, Sherry AD. TmDOTA: A sensitive
probe for MR thermometry in vivo. J Magn Reson 2001;
151:101106.
27. James JR, Gao Y, Miller MA, Babsky A, Bansal N. Absolute
temperature MR imaging with thulium 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetramethyl-1,4,7,10-tetraacetic
acid
(TmDOTMA). Magn Reson Med 2009;62:550556.
28. Hekmatyar SK, Hopewell P, Pakin SK, Babsky A, Bansal N.
Noninvasive MR thermometry using paramagnetic lanthanide complexes of 1,4,7,10-tetraazacyclodoecane-alpha,
alpha0 ,alpha00 ,alpha000 -tetramethyl-1,4,7,10-tetraacetic
acid
(DOTMA4). Magn Reson Med 2005;53:294303.
29. Aime S, Botta M, Fasano M, Terreno E, Kinchesh P, Calabi L,
Paleari L. A new ytterbium chelate as contrast agent
in chemical shift imaging and temperature sensitive probe
for MR spectroscopy. Magn Reson Med 1996;35:648651.
30. Lindner LH, Reinl HM, Schlemmer M, Stahl R,
Peller M. Paramagnetic thermosensitive liposomes for
MR-thermometry. Int J Hyperthermia 2005;21:575588.
31. Henrotte V, Muller RN, Bartholet A, Elst LV. The presence
of halide salts influences the non-covalent interaction of MRI
contrast agents and human serum albumin. Contrast Media
Mol Imaging 2007;2:258261.
32. Zhang Y, Samulski TV, Joines WT, Mattiello J, Levin RL,
LeBihan D. On the accuracy of noninvasive thermometry

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.
46.
47.

48.

49.

50.
51.

281

using molecular diffusion magnetic resonance imaging.

Int J Hyperthermia 1992;8:263274.
Stejskal EO, Tanner JE. Spin diffusion measurements:
Spin echoes in the presence of a time-dependent field
gradient. J Chem Phys 1965;42:288292.
van der Toorn A, Sykova E, Dijkhuizen RM, Vorsek I,
Vargova L, Skobisova E, van Lookeren Campagne M,
Reese T, Nicolay K. Dynamic changes in water ADC,
energy metabolism, extracellular space volume, and tortuosity
in neonatal rat brain during global ischemia. Magn Reson
Med 1996;36:5260.
Gellermann J, Wlodarczyk W, Feussner A, Fahling H,
Nadobny J, Hildebrandt B, Felix R, Wust P. Methods and
potentials of magnetic resonance imaging for monitoring
radiofrequency hyperthermia in a hybrid system. Int J
Hyperthermia 2005;21:497513.
MacFall J, Prescott DM, Fullar E, Samulski TV. Temperature
dependence of canine brain tissue diffusion coefficient
measured in vivo with magnetic resonance echo-planar
imaging. Int J Hyperthermia 1995;11:7386.
Ilyasov KA, Hennig J. Single-shot diffusion-weighted RARE
sequence: Application for temperature monitoring during
hyperthermia session. J Magn Reson Imaging 1998;8:
12961305.
Parker DL, Smith V, Sheldon P, Crooks LE, Fussell L.
Temperature distribution measurements in two-dimensional
NMR imaging. Med Phys 1983;10:321325.
Parker DL. Applications of NMR imaging in hyperthermia:
an evaluation of the potential for localized tissue heating
and noninvasive temperature monitoring. IEEE Trans
Biomed Eng 1984;31:161167.
Dickinson RJ, Hall AS, Hind AJ, Young IR. Measurement
of changes in tissue temperature using MR imaging. J Comput
Assist Tomogr 1986;10:468472.
Hall AS, Prior MV, Hand JW, Young IR, Dickinson RJ.
Observation by MR imaging of in vivo temperature changes
induced by radio frequency hyperthermia. J Comput Assist
Tomogr 1990;14:430436.
Nelson TR, Tung SM. Temperature dependence of proton
relaxation times in vitro. Magn Reson Imaging 1987;5:
189199.
Crawley AP, Henkelman RM. A comparison of one-shot
and recovery methods in T1 imaging. Magn Reson Med
1988;7:2334.
Prato FS, Drost DJ, Keys T, Laxon P, Comissiong B,
Sestini E. Optimization of signal-to-noise ratio in calculated
T1 images derived from two spin-echo images. Magn Reson
Med 1986;3:6375.
Haase A. Snapshot FLASH MRI. Applications to T1, T2, and
chemical-shift imaging. Magn Reson Med 1990;13:7789.
Tong CY, Prato FS. A novel fast T1-mapping method.
J Magn Reson Imaging 1994;4:701708.
Bluml S, Schad LR, Stepanow B, Lorenz WJ. Spin-lattice
relaxation time measurement by means of a TurboFLASH
technique. Magn Reson Med 1993;30:289295.
Matsumoto R, Mulkern RV, Hushek SG, Jolesz FA. Tissue
temperature monitoring for thermal interventional therapy:
comparison of T1-weighted MR sequences. J Magn Reson
Imaging 1994;4:6570.
Wlodarczyk W, Boroschewski R, Hentschel M, Wust P,
Monich G, Felix R. Three-dimensional monitoring of small
temperature changes for therapeutic hyperthermia using MR.
J Magn Reson Imaging 1998;8:165174.
Look DC, Locker DR. Time saving in measurements of NMR
and EPR relaxation times. Rev Sci Instrum 1970;41:250251.
Brix G, Schad LR, Deimling M, Lorenz WJ. Fast and precise
T1 imaging using a TOMROP sequence. Magn Reson
Imaging 1990;8:351356.

Int J Hyperthermia Downloaded from informahealthcare.com by HINARI on 10/20/14

For personal use only.

282

L. Ludemann et al.

52. Gowland P, Mansfield P. Accurate measurement of T1 in vivo

in less than three seconds using echo-planar imaging.
Magn Reson Med 1993;30:351354.
53. Clare S, Jezzard P. Rapid T(1) mapping using multislice echo
planar imaging. Magn Reson Med 2001;45:630634.
54. Li KL, Zhu XP, Waterton J, Jackson A. Improved 3D
quantitative mapping of blood volume and endothelial
permeability in brain tumors. J Magn Reson Imaging
2000;12:347357.
55. Brix G, Kiessling F, Lucht R, Darai S, Wasser K, Delorme S,
Griebel J. Microcirculation and microvasculature in breast
tumors: pharmacokinetic analysis of dynamic MR image
series. Magn Reson Med 2004;52:420429.
56. Ong JT, dArcy JA, Collins DJ, Rivens IH, ter Haar GR,
Leach MO. Sliding window dual gradient echo (SW-dGRE):
T1 and proton resonance frequency (PRF) calibration for
temperature imaging in polyacrylamide gel. Phys Med Biol
2003;48:19171931.
57. Graham SJ, Bronskill MJ, Henkelman RM. Time and
temperature dependence of MR parameters during thermal
coagulation of ex vivo rabbit muscle. Magn Reson Med
1998;39:198203.
58. Brieger J, Schmidt D, Pereira PL, Schick F. Local temperature distribution in HF thermotherapy in an ex-vivo liver
model. Biomed Tech (Berl) 2002;47:S714S716.
59. Young IR, Hand JW, Oatridge A, Prior MV. Modeling
and observation of temperature changes in vivo using MRI.
Magn Reson Med 1994;32:358369.
60. Peller M, Kurze V, Loeffler R, Pahernik S, Dellian M,
Goetz AE, Issels R, Reiser M. Hyperthermia induces T1
relaxation and blood flow changes in tumors. A MRI
thermometry study in vivo. Magn Reson Imaging 2003;
21:545551.
61. Bottomley PA, Foster TH, Argersinger RE, Pfeifer LM.
A review of normal tissue hydrogen NMR relaxation times
and relaxation mechanisms from 1-100 MHz: Dependence on
tissue type, NMR frequency, temperature, species, excision,
and age. Med Phys 1984;11:425448.
62. Fung BM, Durham DL, Wassil DA. The state of water
in biological systems as studied by proton and deuterium
relaxation. Biochim Biophys Acta 1975;399:191202.
63. Germain D, Chevallier P, Laurent A, Savart M, Wassef M,
Saint-Jalmes H. MR monitoring of laser-induced lesions of
the liver in vivo in a low-field open magnet: Temperature
mapping and lesion size prediction. J Magn Reson Imaging
2001;13:4249.
64. Hindman JC. Proton resonance shift of water in the gas and
liquid states. J Chem Phys 1966;44:45824592.
65. Lutz NW, Kuesel AC, Hull WE. A 1H-NMR method for
determining temperature in cell culture perfusion systems.
Magn Reson Med 1993;29:113118.
66. Kuroda K, Abe K, Tsutsumi S, Ishihara Y, Suzuki Y, Sato K.
1993Water proton magnetic resonance spectroscopic imaging.
Biomed Thermol 1993;13:4362.
67. Ishihara Y, Calderon A, Watanabe H, Okamoto K, Suzuki Y,
Kuroda K, Suzuki Y. A precise and fast temperature mapping
using water proton chemical shift. Magn Reson Med
1995;34:814823.
68. Kuroda K, Oshio K, Mulkern RV, Jolesz FA. Optimization
of chemical shift selective suppression of fat. Magn Reson
Med 1998;40:505510.
69. de Zwart JA, Vimeux FC, Delalande C, Canioni P,
Moonen CT. Fast lipid-suppressed MR temperature mapping
with echo-shifted gradient-echo imaging and spectral-spatial
excitation. Magn Reson Med 1999;42:5359.
70. Wu T, Felmlee J, Grimm R, Rydberg J. Temperature
mapping using a spin-echo gradient-echo sequence.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

In: Tenth Scientific Meeting and Exhibition, Magnetic

Resonance in Medicine IS, conference proceedings, May
1824, Honolulu, Hawaii, USA. ISMRM 2002:2210.
Rieke V, Butts Pauly K. Echo combination to reduce proton
resonance frequency (PRF) thermometry errors from fat.
J Magn Reson Imaging 2008;27:673677.
Wyatt C, Soher B, Maccarini P, Charles HC, Stauffer P,
Macfall J. Hyperthermia MRI temperature measurement:
Evaluation of measurement stabilisation strategies for extremity and breast tumours. Int J Hyperthermia 2009;25:
422433.
De Poorter J, De Wagter C, De Deene Y, Thomsen C,
Stahlberg F, Achten E. Noninvasive MRI thermometry with
the proton resonance frequency (PRF) method: In vivo results
in human muscle. Magn Reson Med 1995;33:7481.
El-Sharkawy AM, Schar M, Bottomley PA, Atalar E.
Monitoring and correcting spatio-temporal variations of the
MR scanners static magnetic field. MAGMA 2006;
19:223236.
Gellermann J, Wlodarczyk W, Ganter H, Nadobny J,
Fahling H, Seebass M, Felix R, Wust P. A practical approach
to thermography in a hyperthermia/magnetic resonance hybrid
system: Validation in a heterogeneous phantom. Int J Radiat
Oncol Biol Phys 2005;61:267277.
Wust P, Cho CH, Hildebrandt B, Gellermann J. Thermal
monitoring: Invasive, minimal-invasive and non-invasive
approaches. Int J Hyperthermia 2006;22:255262.
Gellermann J, Hildebrandt B, Issels R, Ganter H,
Wlodarczyk W, Budach V, Felix R, Tunn PU, Reichardt P,
Wust P. Noninvasive magnetic resonance thermography of
soft tissue sarcomas during regional hyperthermia: Correlation
with response and direct thermometry. Cancer 2006;
107:13731382.
Kuroda K, Oshio K, Chung AH, Hynynen K, Jolesz FA.
Temperature mapping using the water proton chemical shift:
A chemical shift selective phase mapping method. Magn
Reson Med 1997;38:845851.
Gellermann J, Weihrauch M, Cho CH, Wlodarczyk W,
Fahling H, Felix R, Budach V, Weiser M, Nadobny J,
Wust P. Comparison of MR-thermography and planning
calculations in phantoms. Med Phys 2006;33:39123920.
Young IR, Hajnal JV, Roberts IG, Ling JX,
Hill-Cottingham RJ, Oatridge A, Wilson JA. An evaluation
of the effects of susceptibility changes on the water chemical
shift method of temperature measurement in human
peripheral muscle. Magn Reson Med 1996;36:366374.
Kunihara T, Sasaki S, Shiiya N, Murashita T, Matsui Y,
Yasuda K. Near infrared spectrophotometry reflects cerebral
metabolism during hypothermic circulatory arrest in adults.
ASAIO J 2001;47:417421.
Peters RD, Henkelman RM. Proton-resonance frequency shift
MR thermometry is affected by changes in the electrical
conductivity of tissue. Magn Reson Med 2000;43:6271.
Kuroda K, Takei N, Mulkern RV, Oshio K, Nakai T,
Okada T, Matsumura A, Yanaka K, Hynynen K, Jolesz FA.
Feasibility of internally referenced brain temperature imaging
with a metabolite signal. Magn Reson Med Sci 2003;2:1722.
Kuroda K, Mulkern RV, Oshio K, Panych LP, Nakai T,
Moriya T, Okuda S, Hynynen K, Jolesz FA, Joles FA.
Temperature mapping using the water proton chemical
shift: Self-referenced method with echo-planar spectroscopic
imaging. Magn Reson Med 2000;43:220225.
Cady EB, DSouza PC, Penrice J, Lorek A. The estimation of
local brain temperature by in vivo 1H magnetic resonance
spectroscopy. Magn Reson Med 1995;33:862867.
Kuroda K. Non-invasive MR thermography using the water
proton chemical shift. Int J Hyperthermia 2005;21:547560.