Department of Electrical and Computer Engineering, Schulich School of Engineering, University of Calgary, 2500 University Dr. NW, Calgary, AB, T2N 1N4, Canada.
Med Biol Eng Comput. 2018 Jun;56(6):1027-1040. doi: 10.1007/s11517-017-1740-9. Epub 2017 Nov 13.
Anatomically realistic numerical breast models are essential tools for microwave breast imaging, supporting feasibility analysis, performance verification, and design improvements. Patient-specific models also assist in interpreting the results of the patient studies conducted on microwave imaging prototype systems. The proposed method employs automated and robust 3D processing techniques to construct flexible and reconfigurable breast models. These techniques include noise and artifact suppression with a principal component analysis (PCA) approach, and oversampling of the magnetic resonance imaging (MRI) data to enhance the intensity continuity. The k-means clustering segmentation identifies fatty and fibroglandular tissues and further segments these regions into a selected number of tissues, providing reconfigurable models. A peak Gaussian fitting technique maps the model clusters to the dielectric properties. The robustness of the proposed method is verified by applying it to both 1.5- and 3-T MRI scans as well as to scans of varying breast densities.
解剖学逼真的数值乳房模型是微波乳房成像的重要工具，支持可行性分析、性能验证和设计改进。针对特定患者的模型还有助于解释在微波成像原型系统上进行的患者研究的结果。所提出的方法采用自动化和稳健的 3D 处理技术来构建灵活且可重构的乳房模型。这些技术包括使用主成分分析 (PCA) 方法抑制噪声和伪影，以及对磁共振成像 (MRI) 数据进行过采样以增强强度连续性。k-均值聚类分割识别脂肪和纤维腺体组织，并进一步将这些区域分割成选定数量的组织，提供可重构的模型。峰值高斯拟合技术将模型聚类映射到介电特性上。通过将该方法应用于 1.5T 和 3T MRI 扫描以及不同乳房密度的扫描，验证了该方法的稳健性。
