Optimization and High-Order Fast Algorithms Applied to Microwave Imaging

应用于微波成像的优化和高阶快速算法

• 批准号: RGPIN-2014-04142
• 负责人: Jeffrey, Ian
• 金额: $ 1.82万
• 依托单位: University of Manitoba
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2017
• 资助国家: 加拿大
• 起止时间: 2017-01-01 至 2018-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=637451
• 关键词: Optimization; Order; Fast; Algorithms; Applied

The research program proposed herein aims to develop a set of high-performance computational tools that can be applied to optimize the design of microwave imaging (MWI) systems and to use these tools to discover and design state-of-the-art next-generation MWI systems for biomedical and agricultural applications. The motivation for this research is based on recent work that suggests proper system design and modelling will improve MWI resolution, making MWI more amenable to imaging applications. The proposed computational methods will reduce both the computing power necessary to perform MWI and the complexity of the systems, resulting in lower costs for industrial installations.MWI has been the subject of significant research in the areas of biomedical imaging, security measures and non-destructive quality assurance. We have recently extended its application to the area of monitoring the quality of stored grain crops where there is a need for a tool that is sensitive to the entire contents of a storage container. MWI is attractive because it is both safe (non-ionizing) and inexpensive. The goal of MWI is to non-invasively reconstruct a model of the electrical properties of an irradiated target from a sampling of the electromagnetic fields external to the target. Knowledge of the target properties has practical uses such as tumour detection in biomedical applications, and early detection of rot conditions during grain storage. The adoption of MWI for many applications has been hindered by the relatively low resolution obtained from standard MWI systems, despite the fact that there is no known resolution limit except the signal-to-noise ratio in the measured data, and by the computational cost associated with generating images.Historically, research on MWI has been focused on inversion algorithms. More recently, attempts have been made to quantify the amount of retrievable target information contained in the data as a function of noise. To complement this work, effort will be devoted to improving forward solver accuracy and adjusting controllable system parameters (transmitter/receiver position/type, profile/boundaries of the external medium) in an attempt to improve and/or optimize MWI resolution capabilities and minimize modelling error. To accomplish these goals we will develop a novel parallel, high-order, frequency-domain, software package for solving Maxwell’s equations and then apply this numerical tool to optimization procedures for determining the best transmitter/receiver configurations and external electromagnetic property profiles that balance the cost of system implementation with the accuracy of the images that are produced. I will use the software tools to design, implement and test innovative MWI systems for breast cancer detection and grain storage monitoring. The resulting MWI systems will provide enhanced resolution and faster image generation times at a lower cost. These tools will also enable Canadian research groups to improve the imaging accuracy of their own MWI systems for breast cancer detection, with the goal of a robust, affordable, mass-screening tool to permit early diagnosis of one of the leading causes of premature death amongst Canadian women. Grain-storage MWI systems are innovative, novel, and important to Canada for securing our grain stores for both domestic consumption and export. This work will be undertaken at the University of Manitoba, an institution with an established record for research in both MWI, at the Electromagnetic Imaging Lab, and grain storage monitoring, at the Centre for Grain Storage Research. The outcomes of the proposed research will strengthen Canada’s role in developing emerging technologies that will benefit both Canadians and the global population.

本文提出的研究计划旨在开发一套可用于优化微波成像(MWI)系统设计的高性能计算工具，并使用这些工具来发现和设计用于生物医学和农业应用的最先进的下一代MWI系统。这项研究的动机是基于最近的工作，该工作表明适当的系统设计和建模将提高MWI的分辨率，使MWI更适合成像应用。所提出的计算方法将降低执行MW I所需的计算能力和系统的复杂性，从而降低工业安装的成本。MW I一直是生物医学成像、安全措施和无损质量保证领域的重要研究主题。我们最近将其应用扩展到监测储存粮食作物质量的领域，在那里需要一种对储存容器的全部内容物敏感的工具。MWI很有吸引力，因为它既安全(非电离)又便宜。MWI的目标是通过对目标外部电磁场的采样，非侵入性地重建受照射目标的电特性模型。对靶标特性的了解具有实际用途，如生物医学应用中的肿瘤检测，以及谷物储存期间腐烂条件的早期检测。尽管除了测量数据的信噪比之外没有已知的分辨率限制，以及与生成图像相关的计算成本，但标准MWI系统获得的分辨率相对较低，这阻碍了MWI在许多应用中的采用。历史上，MWI的研究一直集中在反演算法上。最近，已经尝试将包含在数据中的可检索目标信息量量化为噪声的函数。为了补充这项工作，将致力于提高正演解算器的精度和调整可控制的系统参数(发射机/接收机位置/类型、外部介质的轮廓/边界)，以试图提高和/或优化MWI分辨率能力，并将建模误差降至最低。为了实现这些目标，我们将开发一种新颖的并行、高阶频域软件包，用于求解麦克斯韦方程，然后将该数值工具应用于优化程序，以确定最佳发射机/接收机配置和外部电磁特性曲线，以平衡系统实施成本和所产生图像的精度。我将使用这些软件工具来设计、实施和测试用于乳腺癌检测和粮食储存监测的创新MWI系统。由此产生的MWI系统将以更低的成本提供更高的分辨率和更快的图像生成时间。这些工具还将使加拿大研究小组能够提高他们自己的乳腺癌检测MWI系统的成像准确性，目标是实现一种强大、负担得起的大规模筛查工具，使其能够对加拿大妇女过早死亡的主要原因之一进行早期诊断。粮食储存MWI系统是创新的、新颖的，对于加拿大确保国内消费和出口的粮食存储非常重要。这项工作将在马尼托巴大学进行，该机构在电磁成像实验室的MWI和谷物储存研究中心的谷物储存监测方面都有既定的研究记录。拟议研究的结果将加强加拿大在开发有利于加拿大人和全球人口的新兴技术方面的作用。