Toward a practical ultrasound tomography algorithm for improved breast imaging

描述（由申请人提供）：超声断层扫描是一种新兴的乳房成像方式。然而，目前的超声断层成像算法受到计算机有限的内存和处理器速度的限制，基于射线理论并假设均匀背景，这对于复杂的非均匀区域是不准确的，并且当病变大小与所用超声波长大致相同或小于所用超声波长时失败。因此，超声成像算法必须运用波动理论，正确处理乳腺组织的异质性和衍射效应，才能准确成像小病变。此外，充分利用图形处理单元（graphics Processing Units, gpu）的计算架构，可以大大减轻波形层析成像的繁重计算负担。本研究的长期目标是开发新颖实用的临床乳腺成像方法，能够在病变转移前定期检测病变。第一阶段研究的目标和实现这一目标的第一步是使波形断层扫描在临床环境中实用。我们的中心假设是，利用Delphinus Medical Technologies SoftVue采集几何和计算架构实现的波形断层扫描算法，将能够在临床相关的时间尺度上产生高精度和高分辨率的乳房图像。提出这项研究的基本原理是，最近技术发展的汇合使新方法能够显著提高超声断层扫描的性能，这将导致乳腺癌检测和诊断的重大改进。预期的结果是验证所提出的技术，支持我们的长期目标是一个实用的，低成本的设备，乳腺癌检测和诊断。我们将通过以下具体目标来检验我们的假设和追求我们的目标。目标1：使波形超声层析成像算法适应于SoftVue计算架构。目的2：使用数值乳房模型评估目的1中算法的准确性和分辨率。目的3：利用数值乳腺模型研究目的1中算法在实际临床应用中的计算速度。

DESCRIPTION (provided by applicant): Ultrasound tomography is an emerging modality for breast imaging. However, current ultrasonic tomography's imaging algorithms, hindered by the limited memory and processor speed of computers, are based on ray theory and assume a homogeneous background, which is inaccurate for complex heterogeneous regions and fails when the size of lesions is about the same size or smaller than the wavelength of ultrasound used. Therefore, wave theory must be used in ultrasonic imaging algorithms to properly handle the heterogeneous nature of breast tissue and the diffraction effects in order to accurately image small lesions. Moreover, by taking full advantage of Graphic Processing Units (GPUs) computational architecture, the intensive computation burden of waveform tomography can be greatly alleviated. The long term goal of this research is to develop novel and practical clinical breast imaging methods capable of regularly detecting lesions before they become metastatic. The objective of this Phase I study and the first step in the pursuit of that goal is to make waveform tomography practical in a clinical setting. Our central hypothesis is that the waveform tomography algorithms, implemented by utilizing the Delphinus Medical Technologies SoftVue acquisition geometry and computation architecture, will be able to produce high-accuracy and high-resolution breast images on clinically relevant time scales. The rationale for the proposed study is that a confluence of recent technical developments has enabled new approaches for dramatically improving the performance of ultrasound tomography, which would lead to major improvements in breast cancer detection and diagnosis. The expected outcome is a validation of the proposed technique that supports our long term goal of a practical, low-cost device for breast cancer detection and diagnosis. We will test our hypothesis and pursue our goals through the following specific aims. Aim 1: Adapt waveform ultrasound tomography algorithm to the SoftVue computational architecture. Aim 2: Assess the accuracy and resolution of the algorithm in Aim 1 using numerical breast models. Aim 3: Investigate the computational speed of the algorithm in Aim 1 for practical clinical applications using numerical breast models. PUBLIC HEALTH RELEVANCE: The proposed research explores novel waveform ultrasound tomography algorithms integrated with the computational architecture of our commercial blade-server reconstruction engine with embedded GPUs (SoftVue system) for rapid and optimal multi-parameter breast imaging. Successful outcome of the proposed technique will lead to major improvements in breast cancer imaging and support our long term goal of a practical, low-cost device for breast cancer detection and diagnosis.