Contour detection-based realistic finite-difference-time-domain models for microwave breast cancer detection

详细信息    查看全文

• 作者：Liang Wang 王 梁 ; Xia Xiao 肖 夏 ; Hang Song 宋 航…
• 关键词：3D breast model ; contour detection ; finite ; difference ; time ; domain(FDTD) ; convolutional perfect matched layer(CPML) ; microwave imaging
• 刊名：Transactions of Tianjin University
• 出版年：2016
• 出版时间：December 2016
• 年：2016
• 卷：22
• 期：6
• 页码：572-582
• 全文大小：
• 刊物类别：Engineering
• 刊物主题：Mechanical Engineering; Science, Humanities and Social Sciences, multidisciplinary;
• 出版者：Tianjin University
• ISSN：1995-8196
• 卷排序：22

文摘

In this paper, a collection of three-dimensional(3D)numerical breast models are developed based on clinical magnetic resonance images(MRIs). A hybrid contour detection method is used to create the contour, and the internal space is filled with different breast tissues, with each corresponding to a specified interval of MRI pixel intensity. The developed models anatomically describe the complex tissue structure and dielectric properties in breasts. Besides, they are compatible with finite-difference-time-domain(FDTD)grid cells. Convolutional perfect matched layer(CPML)is applied in conjunction with FDTD to simulate the open boundary outside the model. In the test phase, microwave breast cancer detection simulations are performed in four models with varying radiographic densities. Then, confocal algorithm is utilized to reconstruct the tumor images. Imaging results show that the tumor voxels can be recognized in every case, with 2 mm location error in two low density cases and 7 mm─8 mm location errors in two high density cases, demonstrating that the MRI-derived models can characterize the individual difference between patients’ breasts.