Boundary Extraction Enhanced Distorted Born Iterative Method for Microwave Mammography
Boundary Extraction Enhanced Distorted Born Iterative Method for Microwave Mammography
复制标题
DOI：
10.1109/lawp.2019.2902351
复制
发表时间：
2019-03
影响因子：
4.2
通讯作者：
Kazuki Noritake;S. Kidera
中科院分区：
文献类型：
--
作者：
Kazuki Noritake;S. Kidera
本文致力于提高基于扭曲玻恩迭代法(DBIM)的反散射算法的精度，在微波乳房X光成像中使用精确的边界提取。众所周知，逆散射算法的精度取决于感兴趣区域(ROI)的初始估计。包络算法是一种很有前途的精确感兴趣区域估计算法。然而，在近场乳房成像场景中，天线和乳房表面之间的耦合效应会给基于包络的边界提取带来不可忽略的误差，这主要是由于参考波形和测量波形之间的不匹配造成的。针对这一问题，本文介绍了基于时域有限差分的波形重建方法，以提高基于DBIM的微波成像的精度。二维数值模拟表明，即使在高度异质性的情况下，我们提出的算法也显著地提高了DBIM的精度。
This letter focuses on improving the accuracy of the distorted Born iterative method (DBIM)-based inverse scattering algorithm, using accurate boundary extraction in microwave mammography. It is well known that the accuracy of inverse scattering algorithms depends on the initial estimate of the region of interest (ROI). The Envelope algorithm is one of the promising algorithms for accurate ROI estimation. However, in the near-field breast imaging scenario, the coupling effect between an antenna and the breast surface incurs a nonnegligible error for Envelope-based boundary extraction, which is mainly caused by the mismatching between the reference and measured waveforms. To address this problem, this letter introduces the finite-difference time-domain-based waveform reconstruction for improving the accuracy of DBIM-based microwave imaging. The two-dimensional numerical simulations, using realistic breast phantoms derived from magnetic resonance images, demonstrate that our proposed algorithm remarkably enhances the accuracy of the DBIM, even in highly heterogeneous cases.