CIUS and the Department of Circulation and Medical Imaging, Norwegian University of Science and Technology, Norway.
Medical Ultrasound Imaging Center, Department of Medical Imaging, Radboud University Medical Center, Nijmegen, The Netherlands.
Comput Methods Programs Biomed. 2023 Aug;238:107604. doi: 10.1016/j.cmpb.2023.107604. Epub 2023 May 16.
Ultrasound based blood velocity estimation is a continuously developing frontier, where the vast number of possible acquisition setups and velocity estimators makes it challenging to assess which combination is better suited for a given imaging application. FLUST, the Flow-Line based Ultrasound Simulation Tool, may be used to address this challenge, providing a common platform for evaluation of velocity estimation schemes on in silico data. However, the FLUST approach had some limitations in its original form, including reduced robustness for phase sensitive setups and the need for manual selection of integrity parameters. In addition, implementation of the technique and therefore also documentation of signal integrity was left to potential users of the approach.
In this work, several improvements to the FLUST technique are proposed and investigated, and a robust, open source simulation framework developed. The software supports several transducer types and acquisition setups, in addition to a range of different flow phantoms. The main goal of this work is to offer a robust, computationally cheap and user-friendly framework to simulate ultrasound data from stationary blood velocity fields and thereby facilitate design and evaluation of estimation schemes, including acquisition design, velocity estimation and other post-processing steps.
The technical improvements proposed in this work resulted in reduced interpolation errors, reduced variability in signal power, and also automatic selection of spatial and temporal discretization parameters. Results are presented illustrating the challenges and the effectiveness of the solutions. The integrity of the improved simulation framework is validated in an extensive study, with results indicating that speckle statistics, spatial and temporal correlation and frequency content all correspond well with theoretical predictions. Finally, an illustrative example shows how FLUST may be used throughout the design and optimization process of a velocity estimator.
The FLUST framework is available as a part of the UltraSound ToolBox (USTB), and the results in this paper demonstrate that it can be used as an efficient and reliable tool for the development and validation of ultrasound-based velocity estimation schemes.
基于超声的血流速度估计是一个不断发展的前沿领域,其中大量可能的采集设置和速度估计方法使得评估哪种组合更适合给定的成像应用变得具有挑战性。FLUST(基于流线的超声模拟工具)可用于解决这一挑战,为在模拟数据上评估速度估计方案提供一个通用平台。然而,FLUST 方法在其原始形式中存在一些限制,包括对相敏设置的稳健性降低以及需要手动选择完整性参数。此外,该技术的实现,因此也包括信号完整性的文档,留给了该方法的潜在用户。
在这项工作中,提出并研究了对 FLUST 技术的几项改进,并开发了一个强大的、开源的模拟框架。该软件支持多种换能器类型和采集设置,以及一系列不同的血流体模。这项工作的主要目标是提供一个强大、计算成本低且用户友好的框架,用于模拟来自静止血流速度场的超声数据,从而促进估计方案的设计和评估,包括采集设计、速度估计和其他后处理步骤。
这项工作中提出的技术改进导致插值误差减小、信号功率变化减小,并且还自动选择空间和时间离散化参数。结果表明了所提出的解决方案的有效性和面临的挑战。改进后的模拟框架的完整性在一项广泛的研究中得到了验证,结果表明,散斑统计、空间和时间相关性以及频率内容都与理论预测非常吻合。最后,一个说明性示例展示了如何在速度估计器的设计和优化过程中使用 FLUST。
FLUST 框架作为 UltraSound ToolBox(USTB)的一部分提供,本文的结果表明,它可以作为开发和验证基于超声的速度估计方案的有效且可靠的工具。