SI2:SSE-Collaborative Research: Advanced Software Infrastructure for Biomechanical Inverse Problems

生物力学成像是指对原位和体内组织力学特性的远程测量。因此，可以通过可视化力学性能分布来创建组织的图像。该技术依赖于在组织被一系列外力变形时对其进行成像。通过图像处理，推断出感兴趣区域各处的位移场。然后，根据测量的位移场、组织本构方程的假设形式和动量守恒定律，解决了相关力学性能的反问题。重建参数的图像在疾病的检测、诊断和治疗监测以及为外科训练和计划设计患者特定模型中得到应用。在过去的十年中，我们开发了一个软件包（NLACE）来在不同的应用领域中有效地解决这个逆问题。通过这份合同，我们将对NLACE进行改进，使其更容易使用和修改，并将其用户基础扩展到更广泛的社区。加强NLACE的具体任务可分为两类：(a)从NLACE的工作原型过渡到社区软件资源的步骤。这些包括为NLACE建立一个输入/输出标准，为输入数据和监控解决方案的进度创建GUI，托管NLACE发行版，以及为其发布创建一组测试数据和文档。(b)加强国家环境监测系统功能的任务。其中包括创建用户自定义的超弹性材料模型模块，以解决大量组织和材料类型，NLACE在分布式内存、共享内存和GPU平台上的并行化，以及量化重构参数空间分布的不确定性。我们将通过在年度验证测试中获得的用户反馈来衡量我们的进展，该测试将由一个专注的用户组进行，该用户组将测试拟议研究的所有方面。对NLACE的改进将进一步促进其在疾病的检测、诊断和治疗监测方面的应用，为手术计划和图像引导应用生成针对患者的模型，以及在生物力学和力学生物学方面的研究。为NLACE开发的并行和不确定性量化策略可以应用于具有PDE约束的广泛类型的逆问题，包括声学和电磁散射、地震反演、漫射光学层析成像、含水层渗透率和温度测量。我们的外展计划确保通过Simtk NIH中心网站向我们的焦点小组和更广泛的社区传播NLACE。最后，两名研究生将在计算科学、数学和生物力学领域接受培训，研究结果将在计算和生物医学科学与工程会议上发表。

Biomechanical imaging refers to the remote measurement of the mechanical properties of tissues, in-situ and in-vivo. Images of the tissue can be thus created by visualizing the mechanical property distributions. This technique relies on imaging tissue while it is deformed by a set of externally applied forces. Through image processing, the displacement field everywhere in the region of interest is inferred. An inverse problem for the relevant mechanical properties is then solved, given the measured displacement fields, an assumed form of the tissue's constitutive equation, and the law of conservation of momentum. Images of reconstructed parameters find applications in the detection, diagnosis and treatment monitoring of disease, and in designing patient specific models for surgical training and planning. Over the last decade we have developed a software package (NLACE) to solve this inverse problem eciently in different application domains. Through this award we will make enhancements to NLACE that will make it easier to utilize and modify, and extend its user base to a wider community.The specific tasks for enhancing NLACE can be divided into two categories: (a) Steps to transition from a working prototype of NLACE to a software resource for the community. These include establishing an Input/Output standard for NLACE, creation of a GUI for input data and for monitoring the progress of the solution, hosting NLACE distributions, and creating sets of test data and documentation for its release. (b) Tasks that would enhance the functional capability of NLACE. These include the creation of a user-defined hyperelastic material model module to address a large class of tissue and material types, parallelization of NLACE on distributed memory, shared memory and GPU platforms, and quantifying uncertainty in the spatial distribution of the reconstructed parameters. We will measure our progress through user-feedback obtained during annual validation tests performed by a committed focus user-group that will test all aspects of the proposed research.The proposed improvements to NLACE will further its application in the detection, diagnosis and treatment monitoring of diseases, generation of patient-specic models for surgical planning and image-guidance applications, and studies in biomechanics and mechanobiology. The parallel and uncertainty quantification strategies developed for NLACE can be applied to a broad class of inverse problem with PDE constraints, including acoustic and electromagnetic scattering, seismic inversion, diffuse optical tomography, aquifer permeability and thermometry. Our outreach plans ensure the dissemination of NLACE to our focus group and to a broader community through hosting on the Simtk NIH center website. Finally, two graduate students will be trained in the elds of computational science and mathematics and biomechanics, and results from the proposed research will be presented at conferences on computational and biomedical science and engineering.