Clinical Microwave Breast Imaging – 2D Results and th

Clinical Microwave Breast Imaging – 2D Results and the

Evolution to 3D

P.M. Meaney 1, M.W. Fanning 2, T. Zhou 3, A. Golnabi 4, S.D. Geimer 5, K.D. Paulsen 6 1 Thayer School of Engineering, Dartmouth College, Hanover, NH 03755, USA,

e-mail: paul.meaney@9d87456748d7c1c708a145e0 , tel.: 001 603 646-3939, fax: 001 603 646-3856.

2 Thayer School of Engineering, Dartmouth College, Hanover, NH 03755, USA,

e-mail: margaret.fanning@9d87456748d7c1c708a145e0 , tel.: 001 603 646-9107, fax: 001 603 646-3856.

3 Thayer School of Engineering, Dartmouth College, Hanover, NH 03755, USA,

e-mail: tian.zhou@9d87456748d7c1c708a145e0 , tel.: 001 603 667-3356, fax: 001 603 646-3856.

4 Thayer School of Engineering, Dartmouth College, Hanover, NH 03755, USA,

e-mail: amir.golnabi@9d87456748d7c1c708a145e0 , tel.: 001 603 646-3939, fax: 001 603 646-3856.

5 Thayer School of Engineering, Dartmouth College, Hanover, NH 03755, USA,

e-mail: shireen.geimer@9d87456748d7c1c708a145e0 , tel.: 001 603 646-6518, fax: 001 603 646-3856.

6 Thayer School of Engineering, Dartmouth College, Hanover, NH 03755, USA,

e-mail: keith.paulsen@9d87456748d7c1c708a145e0 , tel.: 001 603 646-2695, fax: 001 603 646-3856.

Abstract The incidence of breast cancer and associated deaths are a recognized world health problem. Conventional screening approaches such as x-ray mammography, ultrasound and increasingly contrast-enhanced MRI are life saving to many women. However, the sensitivity and specificity of these modalities are still limited and there is clearly room for alternatives. Microwave imaging is a potentially important approach in this area because the tissue dielectric properties present important functional information which can be exploited to improve overall sensitivity and specificity in breast imaging. At Dartmouth College, our microwave breast imaging system is currently being used in pilot clinical studies for both tumor diagnosis and for monitoring treatment response during neoadjuvant chemotherapy. These

early 2D imaging studies have demonstrated significance with respect to distinguishing tumors from normal tissue in the diagnostic mode and an ability to predict treatment response at a relatively early stage of treatment. The associated innovations and clinical results set a solid foundation as we advance towards full 3D imaging. Keywords: microwave imaging, breast, dielectric constant, conductivity, tomography, 2D, 3D 1 INTRODUCTION Microwave imaging is an up and coming new strategy for breast imaging based largely on the functional information the dielectric properties present about the breast tissue composition [1]. It has become an important new research topic – especially within academic circles, but because of limited clinical success, has had muted interest from commercial ventures due to its limited clinical success. Several research teams have made compelling arguments for advancing to full 3D imaging efforts due to perceived limitations in the 2D imaging scenarios [2]. On the whole, these claims are made by

groups whose actual 2D efforts were deemed unsatisfactory or who studied the problem primarily in simulation. There is clearly merit to these arguments given that physical measurements are 3D vector phenomena while the various 2D approaches make important assumptions about overall wave behavior. However, given the current clinical successes of our group utilizing 2D reconstruction algorithms, weaknesses of other approaches cannot be fully attributed to the neglect of 3D problem aspects [3,4]. 2D imaging has allowed us to refine various strategies and techniques which will be directly applicable to our 3D effort. Key among these is the implementation of our Gauss-Newton approach with a log transformation [5]. Recently this has been shown to have mathematically similar, and important, characteristics to the multiplicative regularization strategy promoted by A bubakar et al [6]. However, because our technique also tracks the phase terms of the measured and computed fields over neighboring Riemann sheets, it is capable of accurately recovering

the properties of complex, large scatterers without convergence to local minima and the need for a priori information [7]. A dditionally, within the context of

our illumination chamber configuration, we have demonstrated accurate image reconstructions for

substantially under-determined inverse problems which have direct implications to the amount of measurement data required to reconstruct accurate images [8]. These characteristics of robustness and the reduction of the measurement data requirements have driven the realization of our 2D system and are guiding motivations in our 3D development. In the remainder of this paper we describe early 2D imaging results for the area of treatment monitoring of breast cancer while patients undergo chemotherapy. This is an important, underserved application because several of the conventional 978-1-4244-3386-5/09/$25.00 ?2009 IEEE

881

modalities (especially mammography and ultrasound) have been deemed ineffective with only more expensive and invasive approaches such as MR and PET currently accepted for this use. Finally, we will discuss the current hardware and software efforts being made as part of the 3D imaging effort. Figure 1. Photograph of the microwave breast imaging system in the clinic – illumination chamber aperture for the breast is to the right. 2 CLINICAL RESULTS Figure 1 shows our clinical imaging system. The patient lies prone on the table where one breast at a time is pendant into a glycerin:water coupling bath. An array of 16 monopole antennas travels up from the base of the tank and surrounds the breast for illumination. One antenna transmits at a time while the rest act as receivers. This is repeated for all antennas acting as the transmitter, for a range of frequencies and for multiple imaging planes (7 planes starting close to the chestwall and translating vertically in 1 cm increments). The images are reconstructed off-line for a range of operating frequencies and for all

vertical array positions. Each 2D image can be

reconstructed in 2-3 minutes using a single processor

of a Dell Blade system.

Figure 2 shows a series of sagittal view MR

images for the right breast of a 36 year old patient

having a large tumor in the lower portion of the breast

nominally spanning from the nipple area to the

chestwall. The top row of images was for the

beginning of therapy and the lower set was for the 21st

day of Cycle 4 of the drug treatment. For the original

set, the full extent of the tumor can be readily seen in

both the contrast-enhanced and subtraction images. In addition, significant skin thickening can be seen in the

contrast image along with a layer of edema under the skin layer in the T2 image. For the second set of

images, there are still small pockets of enhancement

visible in the subtraction image dispersed from the nipple to the chestwall, but the amount of

enhancement has substantially decreased. In addition, the skin thickening has significantly decreased along with the subcutaneous edema in the T2 image. While this was not an image set at the end of treatment, it is clearly indicative of a full response. The final pathology assessment after mastectomy confirmed that

'#

$ % $ %

'" (* " '& )" $ % '# ! $ % ! $ % $ % " 882

this was a complete response. Figure 3 shows the 1300 MHz microwave images at Plane 5. For the initial date, there is a substantially enhanced arc about the breast perimeter in the lower left quadrant for both the permittivity ('r )and conductivity (() images with a localized elevated spot just inside the arc. The former feature is most likely due to the skin thickening effect since its properties would be much greater than that for the normal breast and would also be increased by the presence of the edema. The image set for Day 1, Cycle 3 still shows a diminished ring enhancement in the ( image, but it has been eliminated in the 'r image with only a small, slightly elevated zone at the site of the tumor. Finally, at the end of treatment, the ring enhancement is completely removed for both images and there remains only a slightly elevated zone in the 'r image at the original tumor site. This elevation remains most likely because the remaining tissue is comprised of scar tissue which would probably be elevated compared to the more dominant adipose tissue. The outline of the breast is also more readily discernible at this latter stage than the first two which is also indicative of a good response. 3 3D EFFORTS

While the clinical efforts described above continue, we are also working towards implementing a full 3D imaging system. This is happening on both the hardware and algorithm fronts. Figure 4 shows a view

of the new imaging tank with the computer controlled arrays mounted below the tank. Opposing sets of motors move two sets of 8 antennas independently such that inpidual antennas can transmit signals at

each possible vertical height while receiving antennas

Figure 4. Photograph of the new illumination tank with interdigitated monopole antenna array antennas controlled from underneath by opposing pairs of motors.

can be positioned in the same plane (similar to the original 2D arrangement) or at other heights to provide

a range of cross-plane propagation possibilities. Initial 3D imaging studies have demonstrated that the cross-plane data is highly effective in reducing image

% & " % & + $ * % & "

+! *+)) % & % & , # % & " % & + * % & " !

883

artifacts we suspect come from the mismatch of 3D data to our 2D algorithm (primarily those outside of the breast region) [9]. This hybrid of a multi-channel system with a minor degree of antenna mechanical motion acts to substantially reduce mutual coupling effects and reduces costs by limiting the number of expensive receiver channels. This minimal dependence on mechanical motion also limits the overall data acquisition time which is essential in any clinical setting. A s mentioned before, all of the 2D algorithmic innovations are being translated to the 3D effort. In addition, because the image reconstruction time is highly dependent on the time to compute the electric field forward solutions at each iteration, we have adopted a finite difference time domain (FDTD) approach for these computations to maintain our forward solution accuracy but at a much reduced computation time.

Acknowledgments

This work was supported by NIH/NCI Grant # PO1-CA080139.

References

[1] E.C. Fear, S.C. Hagness, P.M. Meaney, M. Okoniewski, M.A. Stuchly, “Breast tu m or detection with near-field i m aging,” IEEE Microwave Magazine , vol. 3, pp. 48-56, 2002.

[2] S.Y. Semenov, R.H. Svenson, A.E. Boulyshev,

A .E. Souvorov, V.Y. Borisov, Y. Sizov, A .N. Starostin, K.R. Dezern, G.P. Tatsis, V. Baranov, “Microwave tomography: two-dimensional system for biological imaging,” IEEE Trans. Biomed. Eng., vol. 43, pp. 869-877, 1996.

[3] S.P. Poplack, K.D. Paulsen, A . Hartov, P.M. Meaney, B.W. Pogue, T. Tosteson, M. Grove, S. Soho, W. Wells, “Electromagnetic breast imaging: pilot results in women with abnormal mammography,” Radiology , vol. 243, pp. 350-359, 2007.

[4] P.M. Meaney, M.W. Fanning, T. Raynolds, C.J. Fox, Q. Fang, C.A. Kogel, S.P. Poplack, Paulsen KD, “Initial clinical experience with microwave breast imaging in women with normal mammography,” A cademic Radiology, vol. 14, pp. 207-218, 2007.

[5] P.M. Meaney, K.D. Paulsen, B.W. Pogue, M.I. Miga, "Microwave image reconstruction utilizing log-magnitude and unwrapped phase to improve high-contrast object recovery," IEEE Trans. Medical Imaging , vol. 20, pp. 104-116, 2001.

[6] A . A bubakar, T.M. Habashy, V.L. Druskin, L. Knizhnerman, D. Alumbaugh, “2.5D forward and

inverse modeling for interpreting low frequency electromagnetic measurements,” Geophysics, vol. 73, pp. F165-F177, 2008. [7] P.M. Meaney, Q. Fang, T. Rubaek, E. Demidenko, K.D. Paulsen, “Log transformation benefits parameter estimation in microwave tomographic imaging,” Medical Physics, vol. 34, pp. 2014-2023, 2007. [8] Q. Fang, P.M. Meaney, K.D. Paulsen, “Singular value analysis of the Jacobian matrix in

microwave image reconstructions,” IEEE

Transactions on A ntennas and Propagation, vol. 54, pp. 2371-2380, 2006. [9] Q. Fang, P.M. Meaney, K.D. Paulsen, “Viable three-dimensional microwave imaging: theory and experiments,” IEEE Transactions on A ntennas and Propagation, 2009 (submitted). 884

本文来源：https://www.bwwdw.com/article/zt8q.html