Breast Cancer Detection Using an Ultrashort-Pulse Radar System in Synthetic Breast Phantom Model

Dan Zhang and Atsushi Mase

1.Introduction

As reported in a review paper, the application of microwaves to medical imaging, and specifically to breast cancer imaging, has attracted the interest of a number of research groups around the world[1].There are two main approaches to microwave breast imaging: microwave tomography[2]and radar-based imaging[3].

Microwave tomography is to reconstruct the electrical profile of the breast, by solving a nonlinear and ill-posed inverse scattering problem.The tomographic systems presented so far operate as a narrowband devices in a lower gigahertz regime.

Radar-based imaging has the specific use of a short-pulse, wide frequency band, confocal microwave imaging (CMI) system that employs backscattering to locate breast tumors, makes technology similar to that of ground-penetrating radar (GPR)[3].In contrast to X-ray mammography, the CMI technique exploits the translucent nature of the breast and obtains a large dielectric contrast of the tissues according to their water content.Moreover,the CMI avoids complex image reconstruction algorithms.As the illuminating signal is wideband, a simple time shifting and summing of the signals are enough to detect the malignant tissues.

Several wideband antennas specially designed to transmit short transient pulses have been proposed[4]-[9]for the increasing demand of the CMI technique to detect early breast tumors and the noninvasive GPR method for subsurface exploration, characterization, and monitoring.The antenna should efficiently focus the microwave signal towards the target and collect the back-scattered energy.We require a wideband antenna with an unidirectional radiation pattern for these applications.Moreover, the antenna should be compact enough for easy installation,integration with other electronic circuits, and the ease in fabrication of the arrays.Some microstrip antennas are compact, but they exhibit narrow frequency bandwidth,which makes them unsuitable for time-domain applications.Reference [10]reported an efficient wideband coplanar stripline-fed antenna with improved bandwidth, low cross polarization, and reduced back radiation.

In our previous works, we have studied the simulation and the imaging of a metallic ball in a breast model[11],[12].In this paper, we will perform the preliminary experiments of breast cancer detection by using ultrashort-pulse radar(USPR) with a type of ultra-wideband (UWB) antenna such as compact vivaldi antennas, and utilize an impulse with picosecond (ps) pulse width as a source and measures reflected waves from two types metallic balls located in the model.

In the experiments, breast models for the preliminary experiments are synthetic breast phantom.We use the reflection data to present image reconstructions of the breast models.

2.Processing Method

First, the distances between the focal point and each antenna position are determined and converted to the time delays.The reflected waves obtained by each antenna are summed up, and the square of this sum is assigned to the pixel value at the focal point to make the image.The reconstruction method of the image proposed to the present breast cancer detection named as CMI bases on synthetic aperture radar (SAR)[7].The intensity U of the pixel is written as:

where Anis the backscattered waveform at the n-th antenna located at rn, τn(r) is the time delay from the nth antenna to the synthetic focal point at r, and Δt is the sampling time.

As we know, the problems of the propagation,scattering, and reflection in electromagnetic wave can be solved by the Maxwell’s equations.The numerical scheme is based on the 2D-finite-difference time-domain (FDTD)model, and the simulations are performed in two-dimensional 30 cm×35 cm space of which the grid size is 0.5 mm with Higdon’s second-order absorbing boundary conditions[10].The electrical properties of a material are generally characterized by the complex permittivity.A foreign object having a permittivity different from an object of inspection causes wave reflection and scattering.Breast has comparatively simple structure and lower dielectric constant and loss tangent than the other part of human body.In addition, the dielectric properties of breast tumors are much higher than the normal breast tissue.

When we perform the numerical simulation, the number of antenna positions is corresponding to 19.The calculation process is repeated at each antenna position.Although the incident wave and the reflected wave from the breast skin are much larger than that from the tumor, it is not difficult to distinguish those waves due to the difference of the time-of-flight.Therefore, the spurious signals can be removed.

Fig.1 Simulation breast model with tumor inside and antenna location.

The numerical simulation model is combined by a semicircle with radius of 10 cm and a pillar with length of 5 cm is utilized as a breast model shown in Fig.1.Two breast tumors with diameter of 7 mm and 5 mm have been located at 7 cm and 8 cm depth from the top breast surface but left 3 cm to center, respectively.We assume that the electric property of normal breast tissue is εr=9, σ=0.4 S/m and that of the malignant breast tissue is εr=50, σ=4 S/m[6],where εris the dielectric constant and σ is the conductivity.A Gaussian-shape pulse with a 65 ps full-width half-maximum (FWHM) irradiates to the breast through an antenna located at the position of 10 cm apart from the breast surface.The antenna is rotated every 5°around the breast from 0°to 180°.The reflected waves from the skin and the tumor are received by the identical antenna.Here,the propagation velocity inside the breast is calculated by using the assumed dielectric constant of the breast tissue,εr=9.Fig.2 shows the reconstructed image using the CMI method.White line indicates the breast skin.Note that the two different size tumors located at about 7 cm and 8 cm depth from the skin all can be detected by the method,although the signal of the 5 mm tumor is much lower than 7 mm tumor.

3.Experiment

A synthetic breast phantom simulated for a real breast biopsy made by CRIS 51 with a relative dielectric constant close to 20 and attenuation of 19 db/cm at 8 GHz has been utilized for the model experiment.As shown in Fig.3, the phantom is with 55 mm higher, 110 mm in diameter, lying about 20 mm below from the antenna elements.This distance between antennas and breast provides a reasonable coverage of a breast by an antenna radiation pattern.

Two metallic balls with 9 mm and 6 mm in diameter utilized as tumors are located inside the breast phantom model.The antennas attached to a rotational stage with a stepping motor are moved every 5°around the breast model from 0°to 180°through the top.The reflected waves from the breast model are received at each antenna location, thus 37 reflected waves are provided.

Fig.2 Image reconstruction of a breast tumor: (a) 7 mm tumor located in 7 cm from surface and (b) 5 mm tumor located in 8 cm from surface but left 3 cm to center.

Fig.3.Phantom simulated for a real breast biopsy made by CRIS 51.

The impulse fed to a planar-type antenna as a probe beam of the USPR is an 8 V, 65 ps FWHM pulse.The vivaldi antenna whose dielectric constant is 2.7 and thickness is 1.57 mm has been selected since it provides UWB characteristic acceptable performance as the present frequency domain.The reflected wave detected by an identical antenna is recorded by a high-speed sampling oscilloscope: Agilent 86100B and model 86117B.We set the two antennas in 30 cm distance for measurement of S parameters and the antenna gain in the frequency-domain.The results are shown in Fig.4.We can see that the frequency band is from 1 GHz to 10 GHz which can be suitable for our experiment.Compared to the antenna used in the previous work[11], the antenna has covered a little lower frequency domain.This may lead to have a little lower attenuation in breast tumor detection.

When we apply the image reconstruction, the tumor response must be extracted from measured data at the receiving antenna.The measured data contains the tumor response, and additional undesired signals such as antenna coupling, reflections from the skin.We can subtract all the unwanted signals by the measured signal from the model without the tumor[10].

Fig.4.Compact vivaldi-antenna and its S parameters and gain.

By this method, the unnecessary components in received signals, such as transmitting wave which gets into receiving antenna directly, and the reflected waves of the skin are removed.First we use 9 mm diameter ball as tumor,when the depth from the top surface of the model is selected to 10 mm, the reflected waves from the tumor can be shown in Fig.5 (a), the right is the enlarge figure with the time delay window is selected from 0.5 ns to 0.8 ns.Because from delay time (t=2depth/c0(εμ)0.5), we can know the delay time is about 0.29 ns for 10 mm depth in the phantom.The reflected waves from the surface are located at 0.5 ns and the reflected waves from the tumor are located at about 0.79 ns for depth 10 mm, 1.08 ns for depth 20 mm, and 1.37 ns for depth 30 mm in our experiment,respectively.The image reconstruction is performed after removing the unwanted components from this remaining signal and the reconstructed image has been shown in Fig.5 (b).The location of the tumor can be conjectured by the maximum intensity at the boundary of the dielectric constant changed from the surface of the model.Using the same method, the reflected waves and the image reconstructions have been obtained when the depths have been selected at 20 mm and 30 mm as shown in Fig.6 and Fig.7 with the same data processing method, respectively.We can roughly calculate the maximum S/N from the figures, for example, the max value is about 3.7×10-3V in Fig.5 (a) and the max value is 1.9×10-3V in the enlarge figure, so the maximum S/N is about 0.51(1.9×10-3V/3.7×10-3V).It can be found that reflected waves become lower and more difficult to detect when the tumor has been located deeper from the top surface of the model.

Finally, we want to know effects in results according to the size of tumor, we select the ball’s diameter to 6 mm.Using the same method discussed above, the ball located at 5 mm, 10 mm, and 20 mm from the top surface of the model, the reflected waves and the image reconstructions have been shown from Fig.8 to Fig.10, respectively.Compared to the 9 mm diameter ball, although reflected waves become much lower, we also can detect the ball, and we find the maximum S/N is about 0.42 to detect the tumor in our experiments.

Fig.5.9 mm tumor’s depth is 10 mm: (a) the reflected waves of the tumor and (b) image reconstruction of the tumor.

Fig.6.9 mm tumor’s depth is 20 mm: (a) the reflected waves of the tumor and (b) image reconstruction of the tumor.

Fig.7.9 mm tumor’s depth is 30 mm: (a) the reflected waves of the tumor and (b) image reconstruction of the tumor.

Fig.8.6 mm tumor’s depth is 5 mm: (a) the reflected waves of the tumor and (b) image reconstruction of the tumor.

Fig.9.6 mm tumor’s depth is 10 mm: (a) the reflected waves of the tumor and (b) image reconstruction of the tumor.

Fig.10.6 mm tumor’s depth is 20 mm: (a) reflected waves of the tumor and (b) image reconstruction of the tumor.

4.Conclusions

In the ultrashort-pulse radar system, we have confirmed that the breast cancer detection is possible by using compact vivaldi-antennas.The image reconstruction of the breast models with two different size tumors have been completed by a technique named CMI.We have performed the experiments by using synthetic breast phantom which is approximate to the actual breast tissue.In the future, the research is focused on the application of pulse compression technique to reflected waves for clear image formation.

Acknowledgment

The authors would like to thank Dr.Padhi for his valuable discussion.This work is partly supported by the Grant-in-Aid for Scientific Research, the Ministry of Education, Science, Sports and Culture (No.20360186), and Grant for Practical Application of University R&D Results under the Matching Fund Method,NEDO.

[1]E.C.Fear, P.M.Meaney, and M.A.Stuchly, “Microwaves for breast cancer detection?” ⅠEEE Potentials, vol.22, no.1,pp.12-18, 2003.

[2]P.M.Meaney, K.D.Paulsen, A.Hartov, and R.K.Crane,“Microwave imaging for tissue assessment: initial evaluation in multitarget tissue equivalent phantoms,” ⅠEEE Trans.on Biomedical Engineering, vol.43, no.9, pp.878-890, 1996.

[3]S.C.Hagness, A.Taflove, and J.E.Bridges,“Two-dimensional FDTD analysis of a pulsed microwave confocal system for breast cancer detection: fixed-focus and antenna-array sensors,” ⅠEEE Trans.on Biomedical Engineering, vol.45, pp.1470-1479, Dec.1998.

[4]S.C.Hagness, A.Taflove, and J.E.Bridges, “Wideband ultra-low reverberation antenna for biological sensing,” ⅠEE Electronics Letters, vol.33, pp.1594-1595, Sep.1997.

[5]G.Zheng, A.Z.Elsherbeni, and C.E.Smith, “A coplanar waveguide bow-tie aperture antenna,” in Proc.of ⅠEEE Ⅰnt.Symp.Antennas Propagate, San Antonio, 2002, pp.564-566.

[6]E.C.Fear, S.C.Hagness, P.M.Meaney, Michal Okoniewski, and M.A.Stuchly, “Enhancing breast tumor detection with near-field imaging,” ⅠEEE Microwave Magazine, vol.3, pp.48-56, Mar.2002.

[7]X.Li and S.C.Hagness, “Confocal microwave imaging algorithm for breast cancer detection,” ⅠEEE Microwave Theory and Wireless Component Letters, vol.11, pp.130-132, Mar.2001.

[8]E.C.Fear, Xu Li, and S.C.Hagness, “Confocal microwave imaging for breast cancer detection: localization of tumors in three dimensions,” ⅠEEE Trans.on Biomedical Engineering, vol.49, pp.812-822, Aug.2002.

[9]E.J.Bond, X.Li, and S.C.Hagness, “Microwave imaging via space-time beamforming for early detection of breast cancer,” ⅠEEE Trans.on Antennas and Propagation, vol.51,pp.1690-1705, Aug.2003.

[10]M.Lazebnik, D.Popovic, L.McCartney, C.B.Watkins, M.J.Lindstrom, J.Harter, S.Sewall, T.Ogilvie, A.Magliocco,T.M.Breslin, W.Temple, D.Mew, J.H.Booske, M.Okoniewski, and S.C.Hagness, “A large-scale study of the ultrawideband microwave dielectric properties of normal,benign and malignant breast tissue obtained from cancer surgeries,” Physics in Medicine and Biology, vol.52, pp.6093-6115, Oct.2007.

[11]S.Takaichi, A.Mase, Y.Kogi, et al., “Simulation study and experiment of breast cancer using an ultrashort-pulse radar,”in Proc.of Asia-Pacific Microwave Conf.2008, Hong Kong,2008, pp.1-4.

[12]D.Zhang and A.Mase, “Phantom-model experiment of breast cancer detection using ultrashort-pulse radar with compact vivaldi antennas,” in Proc.of Asia-Pacific Microwave Conf., Yokohama, 2010, pp.1356-1359.