基于超宽带微波成像的乳腺肿瘤影像技术机理研究

Breast cancer is becoming one of the primary malignant cancers which are harmful to women's health. Early detection is very significant for the survival probability of the breast cancer patients. This project will carry out the research of early breast cancer detection by ultra-wideband microwave imaging, which is very attractive in the international academic community of interdiscipline of medical information and electronic information engineering. This technoloy shows obvious advantages such as accurate, sensitive, radiationless, painless,low cost, etc.. The reasonable investigating models will established very carefully in our research according to the complex internal structure of the breast and their corresponding electromagnetic dispersion characteristics. The dispersion model will applied to approach the real freatures of the breast organisms. Time-frequency analysis, all phase spectrum analysis and singularities expansion method for extracing characteristic of poles are employed to identify the effect of microwave transmition from the impact of breast tissues. These measures will also be used to extract tumor information from the detected backscattering signals by studying the amplitude, frequency, loss, phase and other properties of the signals. Inversion imaging algorithm will be developped and optimized to obtain the reconstructed breast image. Through the investigation, the developed technology shall obtain the clear image for detecting the tumor with the small size till 3 mm in diameter and located within the scope range of 50 mm below the skin. This study will promote the technological progress of ultra-wideband microwave detection, antenna design and miniaturization, signal processing and imaging algorithms. The research results will be very essential for the application of the UWB technology in the general medical check of the early breast cancer detection.

乳腺肿瘤是女性发病率最高的恶性肿瘤疾病，早期诊断无疑对治愈率和成活率都具有决定性意义。本项目将对国际学术界近期提出的乳腺肿瘤超宽带微波诊测方法从机理上进行深入研究。此技术具有定位准确、图像清晰、免于X射线辐照、无疼痛、费用低等突出优点。本项目将根据乳房内部复杂结构和电磁色散特性进行合理建模，研究色散条件下的乳腺肿瘤超宽带微波探测模型。引入时频分析、全相位谱分析、奇点展开法提取目标回波特征极点等手段，从幅度、频率、损耗、相位等参量变化特点来明确各乳腺组织对微波传输的影响，提取回波中的肿瘤信息。发展和优化反演成像算法，获得乳房重构图像能够对皮下50 mm深度范围内，尺寸小到3 mm的早期肿瘤进行正确而直观的显示。研究成果将促进超宽带微波目标检测、天线设计与小型化、信号处理与成像算法等方面的技术进步，亦将推动此技术在医疗界对早期乳腺肿瘤诊查的应用。

安全准确的乳腺癌早期诊断对于肿瘤早期发现和延长患者的生命至关重要。新兴的超宽带微波成像技术具有定位精准、图像清晰、对人体安全、无疼痛、成像分辨率高、便宜以及能够具有较深的探测深度等诸多优势。本课题重点研究了超宽带微波技术应用于早期乳腺癌检测的成像关键问题。. 本项目完成了基于磁共振（MRI）医学影像导出了适用于时域有限差分（FDTD）仿真的二维、三维乳房模型的建模。研究模型考虑了皮肤、乳腺组织、腺体以及色散特性对电磁波传播的影响，采用时域有限差分方法模拟了电磁波在乳房组织中的传播特性。提出了一种通过对天线阵列进行设计的方法提取无肿瘤模型对应的参考基准信号，从而获得肿瘤特征信号的方法。开发了一种基于集成经验模态分解方法（EEMD）对肿瘤特征信号直接提取的方法。并与日本广岛大学合作，对该直接提取肿瘤响应信号进行了成功的实验验证。该方法大大节省了检测时间以及检测成本，使得超宽带微波检测早期乳腺肿瘤方法更加适合于临床应用。采用三维共焦成像算法实现了乳腺肿瘤的检测。分别采用了加权向量范数约束波束形成算法以及双约束波束形成算法用于乳腺肿瘤的成像，提高成像算法的鲁棒性。可以实现对皮肤下50mm深度范围内，肿瘤尺寸小到2mm的正确定位检测和直观的图像显示。采用矩阵束方法实现不同大小不同介电特性的乳腺肿瘤特征极点的提取。通过提取的肿瘤特征极点完成了特征极点原子库的建立，采用匹配追踪算法完成了基于特征原子库的肿瘤散射波形的重构，通过最大似然估计判别法实现了不同类别的乳腺肿瘤的鉴别。设计制作了一款超宽带小型天线用于肿瘤检测系统。该天线工作频带范围为2.5 GHz到9 GHz，具有良好的超宽带特性。. 总体上，本项目顺利完成了研究内容，实现了研究目标，获得了预期的科研成果。完成了青年研究骨干的培养计划，促进了国内外科研合作与交流。研究成果促进了超宽带微波目标检测方面的技术进步，推动了此技术在医疗界对早期乳腺肿瘤诊查的应用。