Beruflich Dokumente
Kultur Dokumente
Abstract—An H-shape ultrawideband (UWB) dielectric res- short pulses with low distortions and high efficiency in order to
onator (DR) mounted on a vertical ground plane edge with obtain reliable data from tumors. Although many sensors were
broadside radiation patterns is evaluated for breast tumor detec- proposed before, it is not clear what would be the most appro-
tion. In order to scan the whole breast, the proposed DR sensor
can be attached to the skin without a need for matching medium. priate sensor design for breast tumor detection.
The sensor has a wide half power beamwidth, therefore path First criterion is related to the sensor bandwidth requirement.
loss is reduced. In addition, the DR sensor provides very good It expresses the tradeoff relation between resolution and tissue
pulse-preserving performance such as low distortion and constant penetration. Therefore, we need to design sensor, which has
gain characteristics with high efficiency. The small footprints appropriate bandwidth for breast tumor detection. Second
of the sensor over the skin allows for more sensors nearby or
in contact with the breast skin. Suitability of the compact DR
criterion is related to the dispersion/distortion. In order to
sensor for breast tumor detection can be seen from frequency reduce signal dispersion/distortion, the constant gain and group
domain analyses as well as time domain analyses. Accurate tumor delay are important characteristic for the pulse-preserving
response is obtained due to the compact size and very good sensor performance. Third criterion is the sensor efficiency. This is
characteristics. important to compensate low gain characteristics of UWB
Index Terms—Breast tumor detection, dielectric resonator an- antennas. However, the sensor efficiency within the UWB
tennas (DRAs), ultrawideband (UWB). operating frequencies is from 60% to 90% [17]–[19]. Recently
proposed UWB dielectric resonator antennas (DRAs) mounted
on a vertical ground plane edge [20], [21] achieve higher than
I. INTRODUCTION 95% antenna efficiency within the operating UWB band, be-
cause one of the attractive of the DRAs characteristics is their
B REAST cancer is one of the most common cancers for
women [1]. Early diagnosis is the key to survive from
breast cancer. The radar-based UWB microwave imaging
high radiation efficiency. Fourth criterion is the compactness
of the sensor size. All of the imaging formulation techniques
techniques have attracted great attention since Hagness et assume that the sensor is a point source, which originally comes
al. proposed the possibility of breast tumor detection using from the phase center concept. However, no real antenna is
UWB signal at 1998 [2], [3]. This technique uses a broadband small enough to be assumed a point source, so the radiation
microwave pulse and reconstructs the backscattered energy must appear to emanate from a larger area. Even though the
inside the breast based on the significant contrast in dielectric point source assumption is common for Global Positioning
properties between normal breast tissue and malignant tumors. System (GPS) radar imaging, sensors for Global Network
Radar-based UWB microwave imaging technology research Satellite System (GNSS) applications still suffer due to the
can be divided into two parts. One is the imaging formulation phase center variation (PCV) to obtain more suitable position
techniques, such as the DAS, MIST, RCB, APES, ATAPES information [22], [23]. The reason is that the phase center is
and DCRCB methods [4]–[9]. The other is UWB sensor design not a fixed position. This implies that the point source has a
[10]–[15]. varied position too. Compared with GNSS applications, it may
Optimized ideal UWB antennas have sufficient impedance be more difficult to apply the point source assumption in the
matching bandwidth to transmit short impulse signals, high ra- breast cancer detection application because the sensor size is
diation efficiency, constant gain, constant group delay, and con- quite large compared with the scanned area for tumor detection
sistent uniform radiation pattern to avoid undesirable distortions and the distance between the sensor and the target is small.
of the radiated and received pulse [16]. Unfortunately, no UWB Therefore, a very compact sensor is more desirable for breast
antenna can achieve all of the ideal UWB performance. While tumor detection and the small aperture of the compact sensor
the UWB impedance matching bandwidth based on FCC reg- can provide more nodes or footprints near or in contact with
ulations is mainly concerned with communication systems, a the breast skin.
high quality UWB antenna is required to transmit and receive The main difference between proposed sensor and previous
sensors for the breast tumor detection is not only the compact
size and the high efficiency, but also the ability to use it without
Manuscript received December 15, 2010; revised March 04, 2011; accepted
March 07, 2011. Date of publication August 08, 2011; date of current version
a need for matching medium, because of the DR characteristics.
October 05, 2011. Most of the metallic-type sensors are required to be immersed
The authors are with the Department of Electrical Engineering, University of into a matching medium with permittivity, which is varied
Mississippi, University, MI 38677 USA (e-mail: kennyryu77@yahoo.com).
Color versions of one or more of the figures in this paper are available online
from 3 to 36 for reducing the reflection between free space and
at http://ieeexplore.ieee.org. breast [24]. A lot of additional work is required to implement
Digital Object Identifier 10.1109/TAP.2011.2163790 the matching medium to surround the patient’s breast, and it
0018-926X/$26.00 © 2011 IEEE
RYU AND KISHK: EVALUATION OF DIELECTRIC RESONATOR SENSOR FOR NEAR-FIELD BREAST TUMOR DETECTION 3739
Fig. 1. Geometry of the DR sensor (a) front view (b) isometric view.
Fig. 2. Measured and simulated (a) reflection coefficient (b) group delay (c)
(1) gain.
TABLE I
SENSOR SIZES FOR BREAST TUMOR DETECTION
TABLE II
DIELECTRIC PROPERTIES OF THE BREAST MODEL
[7], so that we obtain only the late time ringing signal from the
tumor. This is usually called the calibrated backscattered signal.
Fig. 6. Tumor responses of three different depths (20 mm, 30 mm, and 40 mm). Fig. 7. Tumor responses of different skin thickness (1 mm and 2 mm).
errors is from the different mesh structure between the models Usually, the peak location of the received signal is considered
with and without the presence of the tumor. In such case, the as the center of the tumor position in the breast tumor detection
response from the numerical error is bigger than the response and tumor size is quite a small compared with the sensor size.
from the breast tumor. To avoid this numerical error, we should Therefore, we use geometric techniques for position estimation
use the same mesh structure whether the tumor exists or not. In to determine the position of the source point from the known
addition, the use of a dense mesh structure is also one of the fac- target positions. Therefore, in the presence of simulations (
tors to reduce the numerical error, because a dense mesh usually mm and mm of yz plane and xz plane with 20, 30, and 40
provides very good convergence rate. mm depths), the position of the point source variation can be
From the above preliminary results, we have obtained useful calculated by the triangulation method, as shown in Fig. 10.
information. First, we obtain received signal strength (RSS). As The calculated point source variation on the z axis is only
the energy of a signal changes with distance, the RSS at a target 0.095 mm, on the y axis is 0.53 mm, and on the x axis is 1.94
conveys information about the distance between that target and mm. The biggest point source variation is on the x axis because
the sensor that has transmitted the signal. Two factors that affect the E plane pattern of this sensor is slightly asymmetric and the
the signal energy are the path-loss, which refers to the reduction geometric radiating element of the x axis is longer than the y
of signal energy as it propagates through space, and shadowing axis.
effects, which represent signal energy variations due to the ob-
stacles in the environment. Second, we obtain the time of arrival IV. IMAGE OF BREAST TUMOR
(TOA) of a signal traveling from sensor to target, and the time
difference of arrival (TDOA) information between the target and To process the signals coherently, we shift the time by a
two known reference positions. In our breast tumor detection, a number of samples to align the returns from the focal point.
combination of position parameters can be utilized in order to After that, we apply a time-window to the time-shifted signals.
obtain more information about the position of the tumor loca- The window is given by
tion. The above hybrid method using a combination of position
parameters specifies the range between a sensor and a target,
which defines a circle for the possible target positions. (2)
RYU AND KISHK: EVALUATION OF DIELECTRIC RESONATOR SENSOR FOR NEAR-FIELD BREAST TUMOR DETECTION 3743
Fig. 9. Shift tumor position responses at 010 mm, 0 mm, and 10 mm.
Fig. 8. Comparison of the tumor response, glandular tissue response, and tumor
in glandular tissue response.