Department of Mechanical Engineering, Centre for Bioengineering, University of Canterbury, Christchurch 8041, New Zealand.
Med Phys. 2013 Nov;40(11):113503. doi: 10.1118/1.4826168.
It is estimated that every year, 1 × 10(6) women are diagnosed with breast cancer, and more than 410,000 die annually worldwide. Digital Image Elasto Tomography (DIET) is a new noninvasive breast cancer screening modality that induces mechanical vibrations in the breast and images its surface motion with digital cameras to detect changes in stiffness. This research develops a new automated approach for diagnosing breast cancer using DIET based on a modal analysis model.
The first and second natural frequency of silicone phantom breasts is analyzed. Separate modal analysis is performed for each region of the phantom to estimate the modal parameters using imaged motion data over several input frequencies. Statistical methods are used to assess the likelihood of a frequency shift, which can indicate tumor location. Phantoms with 5, 10, and 20 mm stiff inclusions are tested, as well as a homogeneous (healthy) phantom. Inclusions are located at four locations with different depth.
The second natural frequency proves to be a reliable metric with the potential to clearly distinguish lesion like inclusions of different stiffness, as well as providing an approximate location for the tumor like inclusions. The 10 and 20 mm inclusions are always detected regardless of depth. The 5 mm inclusions are only detected near the surface. The homogeneous phantom always yields a negative result, as expected.
Detection is based on a statistical likelihood analysis to determine the presence of significantly different frequency response over the phantom, which is a novel approach to this problem. The overall results show promise and justify proof of concept trials with human subjects.
据估计，每年有 1×10(6)女性被诊断患有乳腺癌，全球每年有超过 41 万人因此死亡。数字图像弹性断层成像术（DIET）是一种新的无创乳腺癌筛查方式，它在乳房中产生机械振动，并使用数字相机对其表面运动进行成像，以检测硬度的变化。本研究开发了一种使用 DIET 基于模态分析模型诊断乳腺癌的新自动化方法。
分析硅树脂模型乳房的第一和第二固有频率。对模型的每个区域分别进行模态分析，使用在多个输入频率下拍摄的运动数据来估计模态参数。使用统计方法评估频率偏移的可能性，这可能表明肿瘤的位置。测试了带有 5、10 和 20mm 硬度夹杂物的模型，以及一个均匀（健康）的模型。夹杂物位于四个不同深度的位置。
第二固有频率被证明是一种可靠的指标，具有清晰区分不同硬度的病变夹杂物的潜力，同时还能提供肿瘤样夹杂物的大致位置。10 和 20mm 的夹杂物无论深度如何都能被检测到。5mm 的夹杂物仅在靠近表面处被检测到。均匀模型总是产生阴性结果，这是预期的。
检测基于统计可能性分析来确定模型上存在明显不同的频率响应，这是解决此问题的一种新方法。整体结果显示出有希望的前景，并证明了对人体进行概念验证试验的合理性。
