Advanced image reconstruction for accurate and high-resolution breast ultrasound tomography

摘要 该项目的总体目标是开发和改进先进的断层图像重建。 用于超声计算机断层扫描(UST)的方法，称为波形反转方法，其将 允许高分辨率和定量的乳房成像。这些方法将产生对 声速(SOS)和声学衰减(AA)在乳房内的分布。SOS和AA 表示可以揭示组织几何和弹性特性差异的生物参数。是这样的 信息可以极大地促进乳腺癌与正常组织或良性疾病的区分。 因此，科技大学在改善乳腺癌的检测和管理方面具有巨大的潜力，因为它 利用有效的内源性组织对比，不受辐射和乳房压迫，并且相对 便宜。由我们团队成员开发的SoftVue全乳房UST系统荣获 FDA 510(K)诊断应用程序的许可。 大多数已报道的乳房超声扫描方法都是基于射线的，并且没有考虑声学衍射 效果；这会导致图像的空间分辨率和准确性相对较差。这是非常不受欢迎的 乳房成像应用，其中分辨细微特征的能力对于区分健康很重要 从患病的组织中。用于UST图像重建的波形反演方法都是基于全息图 声波方程，可以绕过基于射线的方法的限制，从而允许较高的 分辨率定量UST乳腺成像。然而，波形反演方法在乳房中的应用 使用环形换能器阵列的UST迄今使用了2D重建方法来估计截面 UST图像。因为3D波的传播物理和换能器的聚焦特性 在这种2D方法中，图像可能包含显著的伪影和降低的空间分辨率。 在这个项目中，我们将开发和优化三维UST波形反演方法来重建SOS 以及质量空前的AA级乳房图像。这些方法将利用在以下位置测量的声学数据 环形换能器阵列的一个或多个位置，并将补偿3D波物理和聚焦 换能器的属性。在这种方法中，将重建薄(高度)体积，而不是 单个2D切片。然后，可以通过合并薄的重建体积来完成整个乳房成像 对应于不同的位置，而不是像在现有2D方法中所做的那样堆叠质量较低的2D切片。 开发的方法将通过使用体模和临床数据进行评估和改进。的具体目标 该项目包括：(1)开发高分辨率SOS成像的波形反演方法；(2)开发 用于高分辨率AA成像的波形反演方法；(4)重建方法的改进 乳房体模研究；(5)利用临床数据评估和改进重建方法。

ABSTRACT The broad objective of this project is to develop and refine advanced tomographic image reconstruction methods for ultrasound computed tomography (UST), referred to as waveform inversion methods, which will permit high resolution and quantitative breast imaging. These methods will yield volumetric estimates of the speed of sound (SOS) and acoustic attenuation (AA) distributions within the breast. The SOS and AA represent bio-parameters that can reveal differences in the geometric and elastic properties of tissue. Such information can greatly facilitate the differentiation of breast cancer from normal tissue or benign disease. Accordingly, UST holds great potential for improving the detection and management of breast cancer since it exploits effective endogenous tissue contrasts, is radiation- and breast-compression-free, and is relatively inexpensive. The SoftVue whole breast UST system developed by members of our team has been awarded FDA 510(k) clearance for diagnostic applications. Most reported methods for breast UST are ray-based and do not take into account acoustic diffraction effects; this results in images of relatively poor spatial resolution and accuracy. This is highly undesirable for breast imaging applications, in which the ability to resolve fine features is important for distinguishing healthy from diseased tissues. Waveform inversion methods for UST image reconstruction are based on the full acoustic wave equation and can circumvent the limitations of ray-based methods, thereby permitting high- resolution quantitative UST breast imaging. However, the application of waveform inversion methods to breast UST employing ring-transducer arrays has to-date employed 2D reconstruction methods to estimate sectional UST images. Because 3D wave propagation physics and the focusing properties of the transducers are not accounted for in this 2D approach, the images can contain significant artifacts and degraded spatial resolution. In this project, we will develop and optimize 3D UST waveform inversion methods for reconstructing SOS and AA images of the breast of unprecedented quality. These methods will utilize acoustic data measured at one or more locations of the ring-transducer array and will compensate for 3D wave physics and the focusing properties of the transducers. In this approach, a thin (in height) volume will be reconstructed instead of a single 2D slice. Whole breast imaging can then be accomplished by merging the thin reconstructed volumes corresponding to different locations instead of stacking lower quality 2D slices as done in existing 2D methods. The developed methods will be evaluated and refined by use of phantom and clinical data. The specific aims of this project are: (1) Develop waveform inversion methods for high resolution SOS imaging; (2) Develop waveform inversion methods for high resolution AA imaging; (4) Refinement of reconstruction methods via breast phantom studies; (5) Assessment and refinement of reconstruction methods using clinical data.