The PSTD Method with the 4th-Order Time Integration for 3D TAT Reconstruction of a Breast Model

Abstract

The thermoacoustic tomography (TAT) is a novel noninvasive and nonionizing medical imaging modality for breast cancer detection. In the TAT, a short pulse of microwave is irradiated to the breast tissue. The tissue absorbs the microwave energy and is heated up momentarily, thus it generates acoustic waves due to the thermoelastic expansion. If the pulse width of the microwave radiation is around one microsecond, the generated acoustic waves are ultrasonic and are in the MHz range. Wide-band ultrasonic transducers are employed to acquire the time-resolved ultrasound signals, which carry information about the microwave absorption properties (mainly related to conductivities) of different tissues. An image showing such properties can then be reconstructed from the time-resolved ultrasound signals. Most existing TAT reconstruction methods are based on the assumption that the tissue under study is acoustically homogeneous. In practice, however, most biological tissues are inhomogeneous. For example, the speed of sound has about 10% variation in breast tissue. The acoustic heterogeneity will cause phase distortion of the pressure field, which will in turn cause blurring in the reconstructed image, thus limiting the ability to resolve small objects. In this work, a 3D inhomogeneous reconstruction method based on pseudo-spectral time-domain (PSTD) is presented to overcome this problem. The method includes two steps. The first step is a homogeneous reconstruction process, from which an initial image is obtained. Since the inhomogeneity itself is usually an acoustic source, the shape and location of the inhomogeneity can be estimated. Then, the acoustic properties of the inhomogeneities (available from the literatures for known tissue types) are assigned to the classified regions, and the other reconstruction based on the updated acoustic property map is conducted. With this process, the phase distortion can be effectively corrected. So it can improve the ability to image small objects. A 3D breast phantom is used to study the proposed method. The breast phantom was generated based on the data set of the Visible Human Project. Regions of different tissue types have been classified and acoustic and electric properties are assigned to such regions. Small phantom tumors placed in the breast phantom have been reconstructed successfully with the inhomogeneous reconstruction method. Improved resolution has been achieved compared to that obtained by homogeneous method.