Computational Framework to Enhance Antenna-based Electromagnetic Imaging

点击翻译按钮获取中文摘要

Project Summary Electromagnetic (EM) imaging has demonstrated significant potential to become a powerful medical imaging technology. Developments in microwave, millimeter-wave, and radio-frequency technology over the last few decades have demonstrated applications in fields such as brain stroke identification and stroke-type differentiation, breast cancer detection, bladder state tracking, and osteoporosis monitoring, among others. EM imaging’s key advantages include deep tissue imaging, non-ionizing radiation, and cost-effective/portable form factors. Unfortunately, EM imaging suffers from two key disadvantages that limit its clinical utility: 1) traditional reconstruction models cannot efficiently account for microwave scattering in complex and heterogeneous biological tissues; and 2) current state-of-the-art antenna arrays are limited by the physical size of individual antenna elements, which do not allow measurements to be densely or optimally captured around an object. This drastically reduces imaging resolution, and can prevent accurate visualization of lesion size and shape. This proposal develops a new EM imaging paradigm where measurements can be collected from a set of optimized antenna locations to drastically enhance 3D EM imaging capabilities. Our proposed project will include two major components: 1) computational frameworks will be formulated to reconstruct 3D permittivity from noninvasive microwave scattering measurements. These frameworks will leverage recent advances in big-data computing, and will utilize optimization-based and machine-learning tools to model microwave scattering through biological tissue; and 2) specialized antenna arrays will be developed to collect microwave scattering measurements, with individual antenna elements positioned at either ultra-high densities or optimized non-regular spacings. The antenna spacings in the non-regular spaced array will be computed based on identifying antenna-positions within the ultra-high-density array that have a greater effect than others on 3D permittivity reconstruction. Identifying these positions will enable 3D permittivity to be accurately reconstructed with fewer measurements and less reconstruction time. The utility of this new approach will be demonstrated through application-oriented testing in the context of both imaging of the head to differentiate stroke-type and of the breast to detect and monitor cancer. In this proposal, the test-case is on using microwaves to reconstruct 3D dielectric permittivity (which is an indicator for tissue water content, and is modified by disease state), but this paradigm can readily be extended to other regions of the electromagnetic spectrum. Overall, this project will enable the high resolution imaging with accurate localization of lesion size and shape that is pivotal for successful clinical translation of EM imaging. The outcomes of this project will be applicable to the diverse clinical applications of EM imaging, spanning from cancer detection, personalized treatment progress monitoring, quantification of inflammation, to real-time tracking of thermal therapies.