Adaptive Information Refinement Modeling of Nonlinear Shear Wave Propagation in Biological Tissue

该项目将结合高分辨率实验和数学信息理论的最新进展来描述剪切波在人体组织和模拟物中的传播，为开发新型医疗诊断设备和更好地理解脑震荡等损伤奠定基础。波在生物组织中的传播是一个基本的物理过程，是许多诊断和治疗技术的基础。成熟的诊断技术包括基于高频声波的超声成像或基于无线电波的磁共振成像。研究生物组织的一个新窗口是使用机械剪切波，它在硬组织中传播得更快，在软组织中传播得更慢。通过使用剪切波，可以有效地进行内部器官的触诊，否则只能通过手术进行。然而，剪切波在生物组织中的传播还不是很清楚，因为波速在不同组织类型中的变化很大，比声波的变化大得多。这可能会导致违反直觉的影响，例如由于头盖骨反射的剪切波聚焦，在远离头部打击的直接区域形成脑震荡。相反，横波速度的巨大变化可以被利用来进行其他诊断技术能力之外的生物组织的详细调查。为了进行这样的诊断，需要对组织结构和横波传播之间的关系有一个基本的了解，正如本项目所研究的那样。项目活动包括指导研究生和博士后参与者，将研究成果整合到本科生的数据驱动建模课程中，以及通过莫尔黑德天文馆和北卡罗来纳大学科学日的外展活动向公众介绍信息理论。应变变硬，可能是由于细胞和细胞外基质中纤维结构的微观重组，导致低至1%应变激发振幅诱导的剪切激波的形成。随后的波前变陡很容易超过7%的应变，导致组织损伤。剪切模量在不同的组织类型中可以变化5个数量级，在诸如大脑等器官中表现出随机位置的充满液体的间质中接近于零。具有离散谱的体积均质粘弹性模型在其校准域之外不适用，并且不通过物理原理与特定组织联系起来。本研究采用实验-理论相结合的方法来开发高度异质生物组织中剪切波传播的尺度相关均质模型。高分辨率，10khz 1024通道测量和1微米分辨率用于构建剪切波通过生物组织和幻影传播的详细图像。利用互信息等信息论概念将不同的仿真层次联系起来，构建了多尺度计算模型，为传统的自适应网格和模型细化提供了新的视角和发展。这项研究的结果将普遍适用于非线性，多尺度随机系统，不表现出尺度分离。实验结果和计算代码将通过识字编程和可重复的研究实践。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This project will combine high-resolution experiments with recent advances in mathematical information theory to describe the propagation of shear waves in human tissue and simulants, laying the foundation for development of novel medical diagnostic equipment and a better understanding of injuries such as concussion. Wave propagation through biological tissue is a fundamental physical process upon which numerous diagnostic and therapeutic techniques are based. Well-established diagnostic techniques include ultrasound imaging based upon high-frequency sound waves or magnetic resonance imaging based upon radio waves. A new window to investigate biological tissue is the use of mechanical shear waves that travel faster through stiff tissue and slower through soft tissue. By use of shear waves it is possible to effectively carry out palpation of internal organs otherwise accessible only by surgery. However, the propagation of shear waves in biological tissue is not well understood due to the large variability in wave velocity through different tissue types, much larger than that of sound waves. This can lead to counterintuitive effects such as formation of a concussion far from the immediate area of a head blow due to focusing of shear waves reflected by the cranium. Conversely, the large variability in shear wave velocity can be leveraged to carry out detailed investigations of biological tissue outside the capabilities of other diagnostic techniques. To enable such diagnostics there is a need to obtain a fundamental understanding of the relationship between tissue structure and shear wave propagation, as investigated in this project. Project activities include mentoring of graduate student and postdoctoral participants, integration of research results into data-driven modeling courses for undergraduates, as well as presentation of information theory to the general public through outreach activities at the Morehead Planetarium and University of North Carolina Science Days.Strain-stiffening, probably due to microscopic reorganization of filamentary structures in cells and the extracellular matrix, leads to the formation of shear shock waves induced by excitation amplitudes as low as 1% strain. The subsequent wavefront steepening can readily surpass 7% strain leading to tissue injury. The shear modulus can vary by five orders of magnitude in different tissue types and is close to zero in liquid-filled interstitia that exhibit random placement in organs such as the brain. Bulk-homogenized viscoelastic models with a discrete spectrum are inapplicable outside their calibration domain and are not linked by physical principles to the specific tissue. This study pursues a combined experimental-theoretical approach to development of scale-dependent homogenized models of shear wave propagation in highly heterogeneous biological tissue. High-resolution, 1024 channel measurements at 10 kHz and 1 micrometer resolution are used to construct detailed images of shear wave propagation through biological tissue and phantoms. A multiple-scale computational model is constructed that uses information-theoretic concepts such as mutual information to link the different simulation levels, offering a new perspective on and development of traditional adaptive mesh and model refinement. Results from this investigation will be generally applicable to nonlinear, multiscale stochastic systems that do not exhibit scale separation. Both experimental results and computational code will be available through literate programming and reproducible research practices.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.