A printed Yagi–Uda antenna for application in magnetic

resonance thermometry guided microwave hyperthermia

applicators
Citation for published version (APA):
Paulides, M. M., Mestrom, R. M. C., Salim, G., Adela, B. B., Numan, W. C. M., Drizdal, T., Yeo, D. T. B., &
Smolders, A. B. (2017). A printed Yagi–Uda antenna for application in magnetic resonance thermometry guided
microwave hyperthermia applicators. Physics in Medicine and Biology, 62(5), 1831-1847. Article 1831.
https://doi.org/10.1088/1361-6560/aa56b3

Document license:
TAVERNE

DOI:
10.1088/1361-6560/aa56b3

Document status and date:

Published: 08/02/2017

Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be
important differences between the submitted version and the official published version of record. People
interested in the research are advised to contact the author for the final version of the publication, or visit the
DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page
numbers.
Link to publication

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please
follow below link for the End User Agreement:
www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at:
[email protected]
providing details and we will investigate your claim.

Download date: ٠٩. ‫ناريزح‬. ٢٠٢٥

Home Search Collections Journals About Contact us My IOPscience

A printed Yagi–Uda antenna for application in magnetic resonance thermometry guided

microwave hyperthermia applicators

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2017 Phys. Med. Biol. 62 1831

(http://iopscience.iop.org/0031-9155/62/5/1831)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 131.155.151.8
This content was downloaded on 13/02/2017 at 06:56

Please note that terms and conditions apply.

You may also be interested in:

Laboratory prototype for experimental validation of MR-guided radiofrequency head and neck
hyperthermia
M M Paulides, J F Bakker, L W Hofstetter et al.

Clinical feasibility of deep hyperthermia treatment in the head and neck

M M Paulides, J F Bakker, M Linthorst et al.

Temperature-based feedback control system for electromagnetic phased-array hyperthermia

M E Kowalski and J-M Jin

Tumor bed brachytherapy for locally advanced laryngeal cancer: a feasibility assessment of
combination with ferromagnetic hyperthermia
Paul R Stauffer, Ilya L Vasilchenko, Aleksey M Osintsev et al.

Single-band radiometric temperature monitoring

Svein Jacobsen and Paul R Stauffer

Association of acute adverse effects with high local SAR induced in the brain from prolonged RF
head and neck hyperthermia
F Adibzadeh, R F Verhaart, G M Verduijn et al.

Hyperthermia treatment planning

J J W Lagendijk

A heterogeneous human tissue mimicking phantom for RF heating and MRI thermal monitoring
verification
Yu Yuan, Cory Wyatt, Paolo Maccarini et al.
 Institute of Physics and Engineering in Medicine Physics in Medicine & Biology

Phys. Med. Biol. 62 (2017) 1831–1847 doi:10.1088/1361-6560/aa56b3

A printed Yagi–Uda antenna for application

in magnetic resonance thermometry
guided microwave hyperthermia
applicators
M M Paulides1, R M C Mestrom2, G Salim3, B B Adela2,
W C M Numan1, T Drizdal1, D T B Yeo4 and A B Smolders2
1
Department of Radiation Oncology, Erasmus University Medical Center, Cancer
Institute, Rotterdam, Netherlands
2
Department of Electrical Engineering, Eindhoven University of Technology,
Eindhoven, Netherlands
3
Department of Radiology and Nuclear Imaging, Erasmus University Medical
Center, Rotterdam, Netherlands
4
General Electric Global Research, Niskayuna, NY, United States of America

E-mail: [email protected]

Received 28 September 2016, revised 1 December 2016

Accepted for publication 4 January 2017
Published 8 February 2017

Abstract
Biological studies and clinical trials show that addition of hyperthermia
stimulates conventional cancer treatment modalities and significantly
improves treatment outcome. This supra-additive stimulation can be
optimized by adaptive hyperthermia to counteract strong and dynamic
thermoregulation. The only clinically proven method for the 3D non-
invasive temperature monitoring required is by magnetic resonance (MR)
temperature imaging, but the currently available set of MR compatible
hyperthermia applicators lack the degree of heat control required. In this
work, we present the design and validation of a high-frequency (433 MHz
ISM band) printed circuit board antenna with a very low MR-footprint.
This design is ideally suited for use in a range of hyperthermia applicator
configurations. Experiments emulating the clinical situation show excellent
matching properties of the antenna over a 7.2% bandwidth (S11  <  −15
dB). Its strongly directional radiation properties minimize inter-element
coupling for typical array configurations (S21  <  −23 dB). MR imaging
distortion by the antenna was found negligible and MR temperature imaging
in a homogeneous muscle phantom was highly correlated with gold-
standard probe measurements (root mean square error: RMSE  =  0.51 °C

1361-6560/17/051831+17$33.00 © 2017 Institute of Physics and Engineering in Medicine Printed in the UK 1831
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

and R2  =  0.99). This work paves the way for tailored MR imaging guided
hyperthermia devices ranging from single antenna or incoherent antenna-
arrays, to real-time adaptive hyperthermia with phased-arrays.

Keywords: hyperthermia, printed antenna, phased-array, MRI, MR

thermometry, radiofrequency, microwave

(Some figures may appear in colour only in the online journal)

Introduction

Extensive biologic research has shown that hyperthermia is one of the most potent modi-
fiers of radiation known today (Overgaard 2013) and that it can also be used to selectively
enhance the effects of several chemotherapies (van der Zee 2002, Kampinga 2006). During
hyperthermia treatments, tissue is heated to 39–44 °C for 60–90 min (Datta et al 2015).
Added to radiotherapy and/or chemotherapy, the impact of hyperthermia on the local con-
trol (Datta et al 2015) and even survival (Issels 2010, Issels 2015, Paulides et al 2016) was
demonstrated through many clinical trials in various tumor sites. Importantly, in these stud-
ies, no increase in late toxicity by hyperthermia was found. Hyperthermia in these trials was
generally induced by using external electromagnetic incoherent (for superficial tumors) or
phased-array (for tumors extending to deeper than 4 cm from the skin) applicators contain-
ing up to 12 independently controlled channels, i.e. antennas or antenna pairs. The location,
shape and size of the heating pattern is controlled by setting the amplitude and phase of the
individual radiofrequency (RF) signals (Fenn et al 1996, Nadobny et al 2002, Franckena
et al 2010). Although existing deep RF hyperthermia devices, particularly those operating at
60–100 MHz, have demonstrated their suitability in positive clinical trials, high temperatures
induced in normal tissues still restrict achieving the optimum temperature range in the tumor
(van Rhoon 2016). Analysis of clinical results showed that achieving higher temperatures
would indeed convert to improved clinical outcomes (Jones et al 2006, Franckena et al 2009,
van Rhoon 2016).
Since the focus diameter generated with a cylindrical phased array is minimum one-third of
the wavelength in tissue (Paulides et al 2005b), ‘microwave’ frequencies (>300 MHz) would
allow for more selective heating. Simulation studies showed that a combination of increasing
the number of independent channels (Henke et al 2001), smart array design (Togni et al 2013)
and advanced treatment planning (Paulides et al 2013) can effectively counteract the reduced
penetration depth at these frequencies. We demonstrated this concept by our HYPERcollar3D
(figure 1), which can heat deeply located tumors in the entire head and neck region (Rijnen
et al 2015, Paulides et al 2016). Selective heating requires more advanced monitoring to pre-
cisely heat the desired region and to prevent unwanted over-heating in normal tissues. In the
clinic, temperatures are generally measured by temperature probes, which can cause morbid-
ity and provide a very poor spatial sampling of the temperature distribution (van der Zee et al
1998, Wust et al 2006). Magnetic resonance (MR) temperature imaging (MRTI), provides a
more complete 3D temperature image without the risk of toxicity (van Rhoon and Wust 2005).
Since MRTI is very demanding in terms of field homogeneity and signal-to-noise (SNR) ratio,
high constraints exist for the MR compatibility of the antennas. Also vice versa, MRTI should
not induce parasitic currents in the antennas, thereby confounding RF heating performance
and affecting the MR scanner’s RF transmit field (B1+) and gradient performance. In this
study, we investigated a novel antenna design that is low-cost, reproducible, and inherently
MR compatible.
1832
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

Figure 1. 3D Model of the Hypercollar 3D applicator (Rijnen et al 2015), showing

three rings of patch antennas in a waterbolus, surrounding a patient. Each antenna
can be phase and amplitude controlled, such that their combined output creates an
electromagnetic interference pattern that heats a target area.

Although many techniques for non-invasive thermometry (NIT) during RF hyperthermia

exist, only MRTI has shown sufficiently accurate to enter the clinic (Gellermann et al 2005,
van Rhoon and Wust 2005). Clinical MR-RF hyperthermia devices have been developed for
the pelvic region (100 MHz) (Gellermann et al 2005) and extremity tumors (140 MHz) (Cheng
et al 2007, Stakhursky et al 2009). Recently, also heating using the imaging coils was demon-
strated in a (Yeo et al 2011, Winter et al 2013, 2015), but this approach restricts the operation
frequency choice to the Larmor frequency (64 or 128 MHz). All previous approaches use RF
waves (<300 MHz) for heating, leading to poor heat focusing and focus-shaping possibili-
ties. For the head and neck region, we experimentally showed the feasibility of MR guided
hyperthermia using a microwave frequency, i.e. 433.92 MHz. Hereto, we have developed an
MR compatible laboratory prototype of the HYPERcollar and demonstrated MRTI accuracy
in a neck-mimicking phantom (Paulides et al 2014). Still, although MRTI during heating is
feasible, accurate measurements require that image distortions by the applicator are reduced to
the absolute minimum (Gellermann et al 2005, Gellermann et al 2008). In this work, we focus
on creating a novel antenna as building block for applicators with maximum MR transparency
aimed at MR measurements with the highest signal-to-noise-ratio (SNR) possible.
The MR thermometry technique used in this paper and generally in hyperthermia, is the proton
resonance frequency shift (PRFS) method (Rieke and Butts Pauly 2008). Accurate MRTI in vivo
however is challenge due to the tissue transitions, air cavities and prominence of motion (respira-
tion, cardiac), as well as magnetic field (B0) distortions by the drift of the scanner, temperature
changes, and the presence of a hyperthermia applicator (Rieke and Butts Pauly 2008, Ludemann
et al 2010, Winter et al 2016). In this work, we used a new PRFS-based method, i.e. 2MT-PRFS
(Salim et al 2015), which corrects for T2* effects, water-fat ratio, and phase at TE  =  0 using
off-resonance estimation, and applies fat-correction (Hofstetter et al 2011). Experiments in a
phantom showed that this technique outperforms a conventional PRFS method (Salim et al 2015).

1833
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

In this paper, we present the design and experimental validation of a printed circuit board
(PCB) Yagi–Uda antenna, to be used in MR compatible hyperthermia applicators operating
in the 433 MHz ISM band5. We specify the requirements, discuss our design approach and
present the antenna design. A coherent set of measurements and simulations was conducted to
investigate the performance of the antenna in hyperthermia applications and analyze the poten-
tial for MR guided single-antenna and multi-antenna phased-array applicators. If the antenna
fulfills all requirements, it will bring improvements in antenna durability and reproducibility,
at reduced production cost, compared to current antenna designs used in hyperthermia.

Materials and methods

Requirements and design process

An overview of the design requirements based on clinical experience from existing applicators
(Paulides et al 2007b, Rijnen et al 2015) and the validated MR compatible prototype (Paulides
et al 2014, Tarasek 2014), is given in table 1. Although most requirements concern single
antenna metrics, these have been formulated such that successful operation of the antenna in
array configurations is guaranteed.
Earlier, we found that target-selective heating requires the use of microwave frequencies,
such as 433.92 MHz (Paulides et al 2005a), resulting in a full-width-half-maximum of the
ellipsoidal heating region’s short axis of 3.5 cm. Further, achieving the required thermal distri-
bution in the tumor requires the antenna elements to handle up to 150 W per antenna (Bakker
et al 2010, Rijnen et al 2013). The antenna should be able to handle a varying environment
because, to increase energy coupling from the antenna to the patient, the antenna is placed
inside a ‘water bolus’ (van der Zee 2002) bag filled with de-ionized (DI) water (figure 1) of var-
ying temperature and thickness. This water is controlled at a low temperature (25 °C  ±  2 °C)
to simultaneously cool the patient’s skin and stabilize antenna performance. In Paulides et al
(2007a), we found that a 5% frequency bandwidth provides sufficient robustness against vari-
ations in the water environment. To avoid excessive heating of subcutaneous tissue and to
reduce high power absorption at fat-muscle transitions, the tangential component(s) of the
electric field inside the patient should be much larger than components normal to the patient
(Paulides et al 2007a). A dominant tangential electric field also results in optimal constructive
interference of the electromagnetic waves (Paulides et al 2007a). To reduce power losses in
the water bolus, the antennas should be positioned at an optimal distance from the patient skin.
This optimal distance is as close as possible to the patient, but far enough to avoid excessive
influence of the patient on the radiation characteristics (so-called loading) of the antennas.
Typically, a distance of half a wavelength in water (h  =  4 cm at 433.92 MHz) is used (Paulides
et al 2007a). To avoid additional matching circuits between the power amplifier and antenna,
a pre-matched antenna design is desired. To reduce performance deterioration in an array
arrangement, the mutual coupling between the antenna elements should be low, preferably
less than  −20 dB (Wust et al 2001). Note that this requirement can be separated into the need
for a directive antenna and the impact of its location and orientation in the antenna array. In
this paper, we focus on directivity to enable using the antenna in a range of array configura-
tions, like planar and circular arrays.
MR compatibility of any device entails the minimization of the device’s interaction with the
scanner’s main static magnetic field, switching gradient fields, transmit RF fields, and the use
of materials that may yield spurious proton signals. Static field interaction and spurious proton

5
ISM band: frequency range allocated for Industry, Science and Medicine.

1834
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

Table 1. The requirements and aims for the antenna design.

Parameter Requirement/aim
Operating frequency 433.92 MHz
Frequency bandwidth (relative) >5% (>21.7 MHz)
Power-handling level 150 W
Input reflection coefficient ⩽−15 dB
Radiation medium De-ionized water
Antenna interface Coaxial cable
Ratio of the tangential component versus >0.8 at the skin surface for
the total electric field the antenna aperture
Operating temperature 25  ±  2 °C
Mutual coupling ⩽−20 dB
Main magnetic field distortion ⩽1 ppm field distortiona
Gradient field distortion No conducting loopsa
MR RF distortion Small cross-section in the B1+  and
B1−  directions, absence of resonating
structures at 64 MHza
a
These requirements can also be met by accurate MRTI measurements in a relevant region-of-interest (ROI).

signal can be generally avoided by material choices, i.e. by avoiding ferromagnetic materials to
minimize any adverse effect on the static magnetic field (B0). Gradient switching-induced eddy
currents may heat the device and dampen the gradient field changes that can be applied. Their
occurrence can be reduced by limiting the size of current paths, generally aiming for small con-
tinuous conducting surfaces, by avoiding conductive loop structures, and/or the use of thin con-
ductive materials. The continuous shared copper ground plane of the HYPERcollar3D antenna
array (Rijnen et al 2015) would for example shield the patient from MR imaging. In earlier
work, we showed that a reduced size and optimized orientation of the conductive elements of a
coaxial-fed patch antenna effectively improved MR measurement performance (Paulides et al
2014). However, further reductions would potentially allow faster MR sequences that use fast
switching gradients, e.g. echo-planar-imaging. The RF properties of the device should also be
aimed towards presenting a high impedance to transmit or receive coils in the MR frequency
band. In previous work, we showed that the base layout of a hyperthermia device already pro-
vides a suitable orientation and position (Paulides et al 2014). In addition, even without reduc-
ing common mode interference by cable-traps, no distortion of the B1  +  RF field at the Larmor
frequency (64 MHz at 1.5 T) was observed (Paulides et al 2014).
Although the system operates at a fixed frequency of 433.92 MHz, a minimum antenna
bandwidth of 5% (21.7 MHz) is required as a surrogate for robustness against variations in
material properties, production tolerances, and varying environmental conditions such as a
changing water temperature.

Antenna design and optimization using electromagnetic modelling

In this study, we adapted and analyzed the well-known Yagi–Uda antenna concept for our
application (Balanis 2005). This antenna consists of a dipole antenna that is made more direc-
tional using two passive elements: a reflector (backside) and a director (frontside). Earlier, we
demonstrated the feasibility of this concept (Adela et al 2013). Figure 2 shows the geometry
optimized for operation at 433.92 MHz. The optimal design was found by manual tuning
after an initial theoretical estimate for the antenna, its director and reflector (Balanis 2005).

1835
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

All simulations used to optimize the design were done with the Finite Integration Technique
solver in CST Microwave Studio (CST AG, Darmstadt, Germany). The metal parts of the
antenna were modeled as perfect electric conductors (PEC). Air was modelled using vac-
uum properties. The PCB-board material (FR4 by Eurocircuits) was modeled with a relative
permittivity of εr  =  4.3 and loss tangent of tanδ  =  0.025. DI water properties were taken at
T  =  25 °C: εr  =  77, and σ  =  0.04 S m−1 (DI water is still somewhat lossy at 433.92 MHz).
Absorbing boundary conditions (absorption  ⩾  99.9%) were used at the sides of the simula-
tion domain to avoid reflections. Figure 2 also shows the ‘clinical environment’ for which the
antenna is optimized. As described earlier, the antenna is embedded in a waterbolus, filled
with DI water that is used for an efficient transfer of electromagnetic energy into the patient
and for cooling the skin and the antennas. The waterbolus was modelled by an 82 mm thick
water volume of de-ionized water, with absorbing boundaries at the four remaining sides to
model an infinite water slab. Note that DI water at 433.92 MHz is still somewhat lossy (0.04
S m−1). As in Paulides, Bakker, Chavannes and Van Rhoon (Paulides et al 2007a), the patient
was modelled by a muscle region with its interface at 40 mm from the antenna. Taking these
environmental parameters into account, the dimensions of the antenna (in water: figure 2) and
the feedlines (crossing the water–air interface: figure 2), were tuned to match to the required
input impedance at the connector in air of 50 Ω.

Impedance and radiation characteristics

For low-power characterization, we used the experimental setup of figure 4, which resembles
the configuration for the simulations in figure 2 (top). It consists of a Styrofoam container,
with inner dimensions of 17.5 cm (x-direction) and 20 cm (y-direction), in which the antenna
is partly submerged into the DI water from the bottom of the setup. A floating muscle phantom
was prepared following a recipe of Ito, Furuya, Okano and Hamada (Ito et al 2001). To study
the robustness of the antenna input reflection coefficient to varying conditions in the clinic, we
varied the phantom-antenna distance (d) between 20 mm, 40 mm and 80 mm and the temper­
ature of the de-ionized water (TDIwater) in steps of 5 °C in the range of 20–35 °C, like in Paulides,
Bakker, Neufeld, van der Zee, Jansen, Levendag and van Rhoon (Paulides et al 2007b). The
ratio between the tangential (Ey) and total electric field for this antenna design was quantified
using post-processing of the simulation results in Matlab (Mathworks, Natick MA, USA).

Cross-coupling in array configurations

In antenna array configurations, the mutual coupling between antennas is specifically of inter-
est. Strong coupling affects the impedance matching of the antennas and can affect the heating
and heating steering possibilities of the applicator. The configuration for measuring the mutual
coupling in an array set-up is shown in figure 7. Because the patient acts as a very effective
RF load in hyperthermia applicators at higher frequencies, we only analyzed directly adjacent
antennas since these exhibit the strongest interactions. The experimental set-up was similar
to the single element arrangement (figure 4), but was modified to allow placement of two
antennas in four different arrangements. Distances (figure 7) were chosen to match those of a
clinically used applicator (Paulides et al 2007b):
• d1  =  broadside parallel array distance, 40 mm or 80 mm;
• d2  =  staggered parallel distance, 20 mm or 60 mm.
• a  =  15 mm for the staggered case

1836
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

Figure 2. Schematic drawings of the setup (nominal situation: h  =  40 mm) used for
optimization of the printed antenna mimicking the clinical environment (top) and the
final printed antenna design and its optimized dimensions (bottom).

High power handling

During hyperthermia treatments, the antennas have to withstand high input powers up to 150 W.
To validate the power handling capability of the antenna, we used the single element arrangement
(figure 4). An RF power amplifier, operating at 433.92 MHz was attached, which is capable of
generating a maximum power of 150 W (Medlogix, Medical Solutions, Italy). To monitor the
power generated by the power amplifier, we used a digital power meter (EMP-442A, Agilent,
USA) connected to the antenna through a  −27 dB directional coupler (3020A, Narda, USA) and
a 6 dB attenuator (R412720000, Radial, USA). This setup was also used to measure the reflected
power from the antenna. In total, the power measured by the power meter was 33 dB less than the
input power to the antenna. Starting from an input power of 0 W and incrementing the power by
10 W every 5 min, we recorded the input reflection coefficient, the temperature of the water and
the temperature at the SMA connector. Once the power reached 180 W, we stepwise decreased the
power down to 10 W. Note that, in contradiction to clinical operation, the water was not circulated
to assess the high-power performance of the antenna under worst-case conditions.
1837
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

MR compatibility and MRTI accuracy

In these experiments, we analyzed the MR-compatibility of the antenna, its heating profile
and the feasibility and accuracy of MRTI. The setup of figure 4 was placed inside a 1.5 T
MR scanner (Optima MR450w, GE Healthcare, Waukesha, WI, USA) and the scanner’s
body-coil was used for transmit and receive of the RF signals. Four closed-tip catheters
were inserted into the muscle-equivalent phantom to allow placement of temperature probes
­during the experiment. In addition, three sunflower oil containers were installed in the setup
as fat-references for correcting B0 heterogeneities in MRTI post-processing. Prior to the
experiments, pre-calibrated Bowman Thermistor probes, supplied by BSD Medical, were
inserted into the catheters and used for ground-truth temperature measurements. MRTI
was performed using the 2MT-PRFS method, since this method was shown more accurate
than conventional corrected PRFS in a similar phantom (Salim et al 2015). The 2MT-PRFS
requires multi-echo gradient-echo (ME-GRE) images, so we applied ME-GRE scanning
with a GE Optima MR450w 1.5 T MRI scanner and the following settings: TR  =  300 ms,
TE  =  3.7, 6.8, 10.2, 13.7, 17.1, 20.6, 24.0 and 27.5 ms, FA  =  29°, bandwidth  =  50 kHz,
matrix  =  256  x  256, FOV  =  40  ×  40 cm, 5 slices, slice thickness  =  0.5 cm. To verify MR
transparency, we created magnitude images using only one echo (TE  =  26.2 ms) of the
ME-GRE scan. Before heating, we made a baseline ME-GRE scan. We then applied five
cycles of approximately 3 min of heating at 150 W, followed by a period without power dur-
ing which a ME-GRE scan was acquired. Temperature increase (ΔT) maps were calculated
using the 2MT-PRFS method.
The first cycle was also used to assess the SAR pattern of the antenna for various antenna—
phantom distances. A short, 3 min, pulse at high power was used to reduce the impact of thermal
diffusion on the SAR measurement (Hand et al 1989). SAR(t1, t2) was calculated using cv * ΔT(t1,
t2)/Δt, where ΔT(t1, t2) ~ Φ(t2)  −  Φ(t1) (Rieke and Butts Pauly 2008), Δt  =  t2  −  t1  =  tpower-
off  −  tpower-on as extracted from the time-temperature curves of the probes and assuming a cv of
3630 J/kg/K in the muscle phantom. Note that the actual cv is unimportant since we will present
only normalized SAR distributions in this paper. Further note that Φ(t) are the phase maps at
time t and that the dependency of SAR and ΔT on the spatial coordinates x, y, z are omitted for
clarity.
After a total of 23 min for the five heating cycles, the setup was allowed to cool down for
another 30 min with regular MRTI measurements. The position of each temperature probe
was reconstructed using the catheter tracks measured with a combination of the MR magni-
tude images and the measured insertion depth. Averaging of the MRTI scans in an in-plane
region-of-interest (ROI) of 5  ×  5 pixels around each probe was applied to reduce noise in this
comparison. The agreement between probes and MRTI was quantified by calculating the root
mean squared error (RMSE) and the correlation coefficient (R2).

Results

The geometry of the printed Yagi–Uda antenna as optimized for operation at 433.92 MHz in the
clinic mimicking environment is shown in figure 2. The antenna is small in size (62 mm) and
can be easily inserted into the water bolus. In addition, the antenna does not require a ground
plane parallel to the RF coil of the MR scanner, enhancing the transparency to the scanner’s
gradient, but has a ground plane perpendicular to the RF coil of the MR scanner as a reflector
in the Yagi–Uda antenna. The base of the antenna contains a microstrip transmission line con-
nected to the coaxial feed point and is surrounded by air or mounting material. The antenna’s

1838
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

top part (the top 41.9 mm) is surrounded by DI water and contains part of the ground-plane/
reflector, the dipole and the director. The parallel strip-fed dipole arms are printed on opposite
sides of the dielectric substrate (FR4). The size of the overlap region (see figure 2) between the
two arms was used as a geometrical parameter to match the antenna impedance. The directing
metal strip at the top of the substrate and the ground of the microstrip line both increase the
directivity of the antenna. The antenna is fed by a 50 Ω microstrip line that transforms into a
twin-lead transmission line. Figure 3 shows the prototype that was manufactured on a 1.6 mm
FR4 substrate according to the layout of figure 2. The metallic parts (copper) were laminated
by a thin layer of anti-corrosion material.

Impedance and radiation characteristics

The simulated and measured input reflection coefficient of the antenna for the nominal situ-
ation (TDIwater  =  25 °C, h  =  40 mm) are depicted in figure 4. The measured response shows
that the antenna is matched at 433.92 MHz, with a  −15 dB impedance bandwidth of 31 MHz
(7.2%, from 416 to 447 MHz). Excellent agreement between simulation and experiment is
observed, i.e. the resonance frequency prediction error is 1 MHz and the value of the reflection
at resonance deviates only 4 dB.
Figure 5 depicts the simulated ratio between Ey and the total electric field (Etot) for the
E-plane of the antenna. The tangential component (Ey) clearly dominates the electric-field at
the water-muscle transition. The ratio |Ey|/|Etot| is  >  80% at this transition for y in the range
of  −5 to 5 cm, which is even wider than the antenna itself and resembles our proven patch
antenna design (Paulides et al 2007a).
Figure 6(a) shows the measured reflection coefficients for variation in h and TDIwater. All dis-
tances h do not impede antenna operation. At a distance h of 20 mm, the presence of the mus-
cle phantom reduces the  −15 dB impedance bandwidth to 4.9% due to loading of the antenna.
Figure 6(b) shows that the antenna behavior is quite robust for temperature variations. Only at
a high water temperatures (35 °C), the antenna is detuned too much. Hence, especially when
the water temperature is controlled to 25  ±  2 °C, the antenna is sufficiently robust against
temperature changes of its water embedding.

Cross-coupling in array configurations

Figure 7 shows the measured mutual coupling for broadside (d1  =  40 mm and 80 mm) and
staggered parallel (d2  =  20 mm and 60 mm) arrangements. Note that these values are mean-
ingful only within the impedance bandwidth of the antenna: 416–447 MHz. As expected
from Paulides, Bakker, Chavannes and Van Rhoon (Paulides et al 2007a), the worst cou-
pling at 433 MHz was found for a broadside arrangement (−23 dB) and the lowest coupling
in the staggered arrangement (<−30 dB). Temperature variations of 25  ±  2 °C had a neg-
ligible influence on the mutual coupling (data not shown). In summary, the antenna met the
specification in table 1.

High power handling

Although increases in power also led to an increased temperature of the water embedding
and the SMA connector, the worst case reflection coefficient remained below  −15 dB at the
operating frequency, even without applying water circulation.

1839
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

Figure 3. The manufactured printed Yagi–Uda antenna, front (left) and back (right).

Figure 4. Simulated and measured reflection coefficient for the nominal situation
(TDIwater  =  25 °C, h  =  40 mm) and the experimental setup used in the measurements
(inlay).

MR compatibility and MRTI accuracy

Figure 8 shows the magnitude (t  =  1 min), power absorption, SAR (t  =  1, 6 min), and temper­
ature increase maps, i.e. ΔT (t  =  1, 24 min), of the setups with varying antenna phantom
distances. In contradiction to common practice in MRTI, the maps are deliberately not spa-
tially averaged to enable a more direct analysis of the measurement results. Note that there
was a delay of 2 s between the MR measurements before application of power so Δt was
calculated used the heating time deduced from the temperature probe measurements. The
magnitude scans show that the impact of the antenna and connector are negligible: apart from
a dark line due to the PCB, only a slight dark region at the bottom of the antenna is noticed.
Additional MR measurements demonstrated that these dark regions were caused by slightly

1840
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

Figure 5. The ratio of tangential (Ey) to the total electric field (Etot) in the E-plane of
the antenna. The dotted line indicates the outer contour of the antenna and the dash-dot
lines represent the transitions from air to DI water (H2O) and from H2O to the muscle
phantom. This setup closely resembles the experimental setup of figure 4, except that
absorbing boundary conditions are applied in x and y-direction.

ferro-magnetic SMA connectors. In the 2MT-PRFS based ΔT map, all disturbances are sub-
stantially reduced. Both magnitude and MRTI results show the excellent MR transparency of
the antenna.
The SAR maps of figure 8 demonstrate that the antenna indeed acts as expected. For a
close antenna distance (h  =  20 mm), the SAR is confined in the y-direction, which increases
for a 40 mm distance. These SAR distributions, particularly the most realistic cases (h  =  20
and 40 mm), show the applicability of the antenna for MR guided superficial hyperthermia.
Note that the blue probe, which was located 3.5 cm from the water-muscle interface and just
within the 25% iso-SAR zone, still measured a local SAR as high as 60 W kg−1, at the 150
W applied power.
Figure 9 visualizes the transient temperature measured in the homogeneous muscle equiva-
lent phantom using temperature probes and MRTI. This figure clearly shows the temperature
increase cycles applied and the moments at which MRTI measurements were taken. Note
that the measurements of both systems were slightly mismatched in time due to a system
clock synchronization issue. This figure demonstrates that the antenna does not impede
highly accurate MRTI. The temperatures measured in the MRTI regions of interest (ROIs)
closely match the temperature curves measured by the high resistive probes. This agreement
can be quantified by the low measurement error (RMSE  =  0.51 °C) and the high correlation
coefficient (R2  =  0.99).

Discussion

A printed Yagi–Uda antenna was developed since this implementation provides a low-cost, repro-
ducible, more directional and MR transparent alternative to currently existing antenna designs.
Analysis against predefined specifications reveals excellent performance for the quantitative

1841
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

Figure 6. Measured robustness of the reflection coefficient of the antenna against

variations in (a) water-muscle phantom distance (h) and (b) the temperature of the water
embedding (TDIwater).

measures. The MR images and highly accurate MRTI results further show the antenna’s MR
transparency and the feasibility of MR thermometry during application of hyperthermia.
In this paper, only single antenna operation was demonstrated. However, MR guided opera-
tion has most potential for phased array applicators since these provide more possibilities
to adapt the heating based on MRTI feedback. Earlier, we demonstrated phased array per-
formance using an array of patch antennas that have similar characteristics, except for an
inherently worse directivity, poorer reproducibility and higher manufacturing costs (Paulides
et al 2014). So, although phased array operation still has to be validated, the antenna’s small
size, high directivity, low cross-coupling, and dominant tangential field component, make the
antenna particularly promising for use in such phased-array hyperthermia applicators.
Antenna—MR system interactions were tested on a 1.5 T MRI scanner and found to be
negligible in terms of impacting image quality and systems performance. However, since
the antennas may be configured for use on higher field MRI systems in the future to obtain a

1842
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

Figure 7. Measured mutual coupling coefficients for the four different arrangements
at the nominal water temperature (TDIwater  =  25 °C). The inlay shows a scheme of the
arrangements and dimensions.

Figure 8. GRE magnitude ((a)–(c)), SAR (1 min, 6 min) ((d)–(f)) and ΔT (1 min,
25 min) ((g)–(i)) distributions measured using for the setup of figure 4 and h  =  2 cm
((a), (d) and (g)), 4 cm ((b), (e) and (h)) or 6 cm ((c), (f ) and (i)). Cross-sections of the
cylindrical oil reservoirs, the muscle phantom and the water volume are indicated by
dotted lines and the locations of the catheters (grey lines in (c)), reconstructed probes
(colored dots) and the very localized distortion by the antenna (black lines in (a)–(c))
are indicated as well. Note that the distortions in de-ionized water (H2O) region present
in the magnitude image are smaller in the ΔT images.

1843
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

Figure 9. Temporal temperatures measured by probes and the 2MT-PRFS technique

for h  =  20 mm (see figure 8). 2MT-PRFS MRTI results are the average of a 5  ×  5 pixel
ROI around the reconstructed probe locations, of which location indications are shown
by colored dots in figure 8. Note that MRTI measures temperature increase and that the
MRTI temperatures at the start of the experiment were subtracted from the temperature
probe measurements.

higher SNR for enabling higher image resolution and/or faster scans, it is advantageous to dis-
cuss the potential level of system interaction issues that could be observed at higher B0 fields
as well. The first question is the printed antenna’s interoperability with typical high magnetic
field systems, such as those operating at 3.0 T (128 MHz) and 7.0 T (298 MHz), which is well
outside of the antenna impedance bandwidth presented in figure 4. Although higher order
modes must still be analyzed, in general, reciprocity dictates that a low antenna sensitivity at
these frequencies would imply low risks of the antennas radiating spurious B1+  fields that
may distort the MR scanner’s target B1+  fields. Distortion of image reception due to coil
sensitivity near 433.92 MHz is dependent on the receive coil design, and interaction can be
further mitigated by refraining from RF transmission via the antennas and properly terminat-
ing the antennas when the scanner is in receive mode. The head and neck area’s small cross
sectional area mitigates the impact of image shading arising from adverse dielectric effects
(Schick 2005). The antenna’s impact on systems interactions and image quality at higher B0
fields would then only be a result of material composition and layout, and an extension of
effects observed for the 1.5 T case. Since very little impact on image quality was observed at
1.5 T, there is no real reason to believe they will be detrimental at 3.0 T or 7.0 T. However,
verification through simulations or experimentation at these field strengths remains necessary.
In contradiction to common practice, we did not apply spatial averaging of MRTI results for
generating the SAR distributions. In this way, we could more clearly visualize the increasing
impact of noise for the larger phantom-antenna distance. The noise in these images, raises the
question if the ESHO quality assurance guidelines, which dictate a heating time of 1 min for
SAR assessments (Hand et al 1989), are also applicable when using MRTI. Even when apply-
ing a heating time of 3 min, we observed considerable noise that, in some cases, exceeded the
heating pattern. In this case, the trade-off between sufficient SNR and avoidance of heat con-
duction for a power-pulse measurement may advocate a shift towards a longer heating time.

1844
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

Conclusions

A printed Yagi–Uda antenna for a hyperthermia applicator was presented. Using a clinical
experimental setup, the performance of this antenna was studied as a single element, in the
vicinity of neighbor antennas and in an MR environment. The measured impedance band-
width of 7.2% is sufficient for the intended application. Simulations show that the dominant
E-field component is generated in the desired direction (patient axis). Changing the distance
between the antenna and the muscle-layer from its nominal value of 40 mm to 20 mm resulted
in an acceptable reduction of the impedance bandwidth to 4.9%. The investigation of the effect
of temperature showed that only at temperatures exceeding 35 °C the antenna is detuned too
much to be used in the applicator. Furthermore, the worst-case measured mutual coupling
level was  −23 dB at 433.92 MHz, which is well below the specification. In addition, power
handling measurements showed a negligible effect of heating up on the antenna character-
istics. MR-compatibility was demonstrated by MR imaging of the antenna in the clinical
experimental setup. MR imaging showed that the image distortion by the antenna itself is very
low and accurate MRTI proved possible. In summary, the performance assessments show that
the antenna has high potential for application in a variety of MR-compatible hyperthermia
applicators.

Acknowledgments

This work was financially supported by the Dutch Cancer Society, grant EMCR2012-5472.
Dirk Poot and Stefan Klein are acknowledged for their advice and Dirk Poot for providing his
FitMRI toolbox (Poot and Klein 2015) (http://fitmri.bigr.nl), which is used in the 2MT-PRFS
method to estimate off-resonance maps.

Conflict of interests

MPS has financial interest in Sensius BV.

References

Adela B, Mestrom R, Paulides M and Smolders A 2013 An MR-compatible Printed Yagi–Uda antenna
for a phased array hyperthemia applicator Proc. of the 7th EuCAP (Gothenburg, Sweden)
Bakker J F, Paulides M M, Westra A H, Schippers H and Van Rhoon G C 2010 Design and test of a
434 MHz multi-channel amplifier system for targeted hyperthermia applicators Int. J. Hyperth.
26 158–70
Balanis C 2005 Antenna Theory: Analysis and Design (New York: Wiley Interscience)
Cheng K S, Stakhursky V, Stauffer P, Dewhirst M and Das S K 2007 Online feedback focusing algorithm
for hyperthermia cancer treatment Int. J. Hyperth. 23 539–54
Datta N R, Ordonez S G, Gaipl U S, Paulides M M, Crezee H, Gellermann J, Marder D, Puric E and
Bodis S 2015 Local hyperthermia combined with radiotherapy and-/or chemotherapy: recent
advances and promises for the future Cancer Treat. Rev. 41 742–53
Fenn A J, Sathiaseelan V, King G A and Stauffer P R 1996 Improved localization of energy deposition
in adaptive phased array hyperthermia treatment of cancer Linc. Lab. J. 9 187–96
Franckena M, Canters R, Termorshuizen F, Van Der Zee J and Van Rhoon G 2010 Clinical implementation
of hyperthermia treatment planning guided steering: a cross over trial to assess its current
contribution to treatment quality Int. J. Hyperth. 26 145–57

1845
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

Franckena M, Fatehi D, de Bruijne M, Canters R A, van Norden Y, Mens J W, van Rhoon G C and van
der Zee J 2009 Hyperthermia dose-effect relationship in 420 patients with cervical cancer treated
with combined radiotherapy and hyperthermia Eur. J. Cancer 45 1969–78
Gellermann J, Faehling H, Mielec M, Cho C H, Budach V and Wust P 2008 Image artifacts during MRT
hybrid hyperthermia—causes and elimination Int. J. Hyperth. 24 327–35
Gellermann J, Wlodarczyk W, Feussner A, Fahling H, Nadobny J, Hildebrandt B, Felix R and Wust P
2005 Methods and potentials of magnetic resonance imaging for monitoring radiofrequency
hyperthermia in a hybrid system Int. J. Hyperth. 21 497–513
Hand J W, Lagendijk J J, Bach Andersen J and Bolomey J C 1989 Quality assurance guidelines for
ESHO protocols Int. J. Hyperth. 5 421–8
Henke F M, Joines W T and Samulski T V 2001 Variations of focal regions versus numbers and positions
of sources in two-dimensional media Int. J. Hyperth. 17 382–400
Hofstetter L, Yeo D, Dixon W T, Davis C and Foo T K 2011 Fat-referenced MR thermometry using
3-echo phase-based fat water separation method 19th Annual Meeting of the Int. Society for
Magnetic Resonance in Medicine (ISMRM) (Montreal, Quebec, Canada)
Issels R D et al 2010 Neo-adjuvant chemotherapy alone or with regional hyperthermia for localised
high-risk soft-tissue sarcoma: a randomised phase 3 multicentre study Lancet Oncol. 11 561–70
Issels R D et al 2015 13LBA Improved overall survival by adding regional hyperthermia to
neo-adjuvant chemotherapy in patients with localized high-risk soft tissue sarcoma (HR-STS):
long-term outcomes of the EORTC 62961/ESHO randomized phase III study 18th ECCO—40th
ESMO European Cancer Congress (Vienna, Austria)
Ito K, Furuya K, Okano Y and Hamada L 2001 Development and characteristics of a biological
tissue-equivalent phantom for microwaves Electron. Commun. Japan I 84 67–77
Jones E, Thrall D, Dewhirst M W and Vujaskovic Z 2006 Prospective thermal dosimetry: the key to
hyperthermia’s future Int. J. Hyperth. 22 247–53
Kampinga H H 2006 Cell biological effects of hyperthermia alone or combined with radiation or drugs:
a short introduction to newcomers in the field Int. J. Hyperth. 22 191–6
Ludemann L, Wlodarczyk W, Nadobny J, Weihrauch M, Gellermann J and Wust P 2010 Non-invasive
magnetic resonance thermography during regional hyperthermia Int. J. Hyperth. 26 273–82
Nadobny J, Fahling H, Hagmann M J, Turner P F, Wlodarczyk W, Gellermann J M, Deuflhard P
and Wust P 2002 Experimental and numerical investigation of feed-point parameters in a 3D
hyperthermia applicator using different FDTD models of feed networks IEEE Trans. Biomed. Eng.
49 1348–59
Overgaard J 2013 The heat is (still) on—the past and future of hyperthermic radiation oncology
Radiother. Oncol. 109 185–7
Paulides M M, Bakker J F, Chavannes N and Van Rhoon G C 2007a A patch antenna design for application
in a phased-array head and neck hyperthermia applicator IEEE Trans. Biomed. Eng. 54 2057–63
Paulides M M, Bakker J F, Neufeld E, van der Zee J, Jansen P P, Levendag P C and van Rhoon G C
2007b The HYPERcollar: a novel applicator for hyperthermia in the head and neck Int. J. Hyperth.
23 567–76
Paulides M M, Bakker J F, Hofstetter L W, Numan W C, Pellicer R, Fiveland E W, Tarasek M,
Houston G C, van Rhoon G C, Yeo D T and Kotek G 2014 Laboratory prototype for experimental
validation of MR-guided radiofrequency head and neck hyperthermia Phys. Med. Biol. 59 2139–54
Paulides M M, Stauffer P R, Neufeld E, Maccarini P F, Kyriakou A, Canters R A, Diederich C J,
Bakker J F and Van Rhoon G C 2013 Simulation techniques in hyperthermia treatment planning
Int. J. Hyperth. 29 346–57
Paulides M M, Verduijn G M and Van Holthe N 2016 Status quo and directions in deep head and neck
hyperthermia Radiat. Oncol. 11 21
Paulides M M, Vossen S H, Zwamborn A P and van Rhoon G C 2005a Theoretical investigation into
the feasibility to deposit RF energy centrally in the head-and-neck region Int. J. Radiat. Oncol.
Biol. Phys. 63 634–42
Paulides M M, Wielheesen D H, Van der Zee J and Van Rhoon G C 2005b Assessment of the local SAR
distortion by major anatomical structures in a cylindrical neck phantom Int. J. Hyperth. 21 125–40
Poot D H and Klein S 2015 Detecting statistically significant differences in quantitative MRI experiments,
applied to diffusion tensor imaging IEEE Trans. Med. Imaging 34 1164–76
Rieke V and Butts Pauly K 2008 MR thermometry J. Magn. Reson. Imaging 27 376–90
Rijnen Z, Bakker J F, Canters R A, Togni P, Verduijn G M, Levendag P C, Van Rhoon G C and
Paulides M M 2013 Clinical integration of software tool VEDO for adaptive and quantitative
application of phased array hyperthermia in the head and neck Int. J. Hyperth. 29 181–93
1846
Phys. Med. Biol. 62 (2017) 1831 M M Paulides et al

Rijnen Z, Togni P, Roskam R, van de Geer S G, Goossens R H and Paulides M M 2015 Quality and
comfort in head and neck hyperthermia: a redesign according to clinical experience and simulation
studies Int. J. Hyperth. 31 823–30
Salim G, Poot D, Numan W, Vogel M, Drizdal T, Paulides M and Klein S 2015 Multipeak multiecho
modeling for improved PRF-based thermometry 32nd Annual Scientific Meeting of the ESMRMB
(Edinburgh, UK)
Schick F 2005 Whole-body MRI at high field: technical limits and clinical potential Eur. Radiol.
15 946–59
Stakhursky V L, Arabe O, Cheng K S, Macfall J, Maccarini P, Craciunescu O, Dewhirst M, Stauffer P
and Das S K 2009 Real-time MRI-guided hyperthermia treatment using a fast adaptive algorithm
Phys. Med. Biol. 54 2131–45
Tarasek M R et al 2014 Validation of MR thermometry: method for temperature probe sensor registration
accuracy in head and neck phantoms Int. J. Hyperth. 30 142–9
Togni P, Rijnen Z, Numan W C, Verhaart R F, Bakker J F, van Rhoon G C and Paulides M M 2013
Electromagnetic redesign of the HYPERcollar applicator: toward improved deep local head-and-
neck hyperthermia Phys. Med. Biol. 58 5997–6009
van der Zee J 2002 Heating the patient: a promising approach? Ann. Oncol. 13 1173–84
van der Zee J, Peer-Valstar J N, Rietveld P J, de Graaf-Strukowska L and van Rhoon G C 1998 Practical
limitations of interstitial thermometry during deep hyperthermia Int. J. Radiat. Oncol. Biol. Phys.
40 1205–12
van Rhoon G C 2016 Is CEM43 still a relevant thermal dose parameter for hyperthermia treatment
monitoring? Int. J. Hyperth. 32 50–62
van Rhoon G C and Wust P 2005 Introduction: non-invasive thermometry for thermotherapy
Int. J. Hyperth. 21 489–95
Winter L, Oberacker E, Paul K, Ji Y, Oezerdem C, Ghadjar P, Thieme A, Budach V, Wust P and
Niendorf T 2016 Magnetic resonance thermometry: methodology, pitfalls and practical solutions
Int. J. Hyperth. 32 63–75
Winter L, Ozerdem C, Hoffmann W, Santoro D, Muller A, Waiczies H, Seemann R, Graessl A, Wust P
and Niendorf T 2013 Design and evaluation of a hybrid radiofrequency applicator for magnetic
resonance imaging and RF induced hyperthermia: electromagnetic field simulations up to 14.0
Tesla and proof-of-concept at 7.0 Tesla PLoS One 8 e61661
Winter L, Oezerdem C, Hoffmann W, van de Lindt T, Periquito J, Ji Y, Ghadjar P, Budach V, Wust P and
Niendorf T 2015 Thermal magnetic resonance: physics considerations and electromagnetic field
simulations up to 23.5 Tesla (1GHz) Radiat. Oncol. 10 201
Wust P, Cho C H, Hildebrandt B and Gellermann J 2006 Thermal monitoring: invasive, minimal-invasive
and non-invasive approaches Int. J. Hyperth. 22 255–62
Wust P, Fahling H, Wlodarczyk W, Seebass M, Gellermann J, Deuflhard P and Nadobny J 2001
Antenna arrays in the SIGMA-eye applicator: interactions and transforming networks Med. Phys.
28 1793–805
Yeo D, Yang X, Wu J, Hofstetter L W, Piel J E, Fiveland E W, Park K J and Foo T K 2011 Investigation
of a dual-function applicator for rf hyperthermia and MRI 19th Annual Meeting of the ISMRM
(Montreal, Quebec, Canada) p 1811

1847