Integrating Data- and Model-based Methods to Enable Improved Heart Surgery Planning

该奖项通过调查数据驱动的计算机模型和3D打印技术的使用来改进主动脉狭窄患者的手术计划，从而为国家的健康、繁荣和福利做出贡献。主动脉瓣狭窄是最常见和最严重的心脏瓣膜疾病之一，通常通过经导管主动脉瓣置换术(TAVR)来解决。尽管TAVR手术总体上对生存有好处，但可能会导致不良后果，如瓣膜旁渗漏，这与再次住院、充血性心力衰竭和增加一年死亡率有关。患者瓣膜的计算机模型可能有助于改善手术结果。本项目的目标是建立一个心脏瓣膜的功能模型，以便用基于3D打印瓣膜的实验数据来扩充真实患者的泄漏数据，从而为手术计划开发更可靠的预测模型。这项研究有可能在实践中显著改善心脏病治疗和干预的手术结果，并有助于减少此类手术的不良结果。该奖项支持一名研究生，他的研究方向是数据科学和添加剂制造。该项目探索在3D打印聚合物中嵌入微观结构的想法，以定制其机械性能并模拟真实的人体组织。由于优化需要在无限维的功能域中执行，因此寻找光纤插入物的最佳配置是一个极其困难的问题。该项目利用先进的统计建模技术，使这个优化问题变得容易处理。具体地说，将开发一个新的泛函输入-泛函输出协同克立格法模型，该模型将应力-应变曲线与光纤插入件形状联系起来。这种建模技术的一个关键特征是一个新的光谱距离相关函数，它使模型对平移不变，这是组织模拟问题的理想特性。此外，实验设计和贝叶斯模型校准方面的先进技术被用来获得3D打印的阀门数据，并将它们与现有的历史数据合并。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This award contributes to the national health, prosperity, and welfare by investigating the use of data-driven computer models and 3D printing technology to improve surgery planning for patients suffering from Aortic Stenosis. Aortic Stenosis is among the most common and severe valvular heart disease conditions and is generally addressed surgically through Transcatheter Aortic Valve Replacement (TAVR). Despite the overall survival benefit, the TAVR surgery may result in adverse outcomes such as paravalvular leakage, which is associated with hospital readmission, congestive heart failure, and increased one-year mortality. Computer models of a patient's valve may help improve surgical outcomes. The goal of this project is to produce a functional model of a heart valve in order to augment leakage data from real patients with experimental data based on 3D printed valves so as to develop more reliable predictive models for surgical planning. The research has the potential to significantly improve surgical outcomes in practice for cardiologic treatment and intervention and help reduce adverse outcomes of such surgeries. The award supports a graduate student whose research lies at the intersection of data science and additive manufacturing.This project explores the idea of embedding microstructure in the 3D printed polymer to tailor its mechanical properties and mimic real human tissue. Finding the optimal configuration of the fiber insert is an extremely difficult problem because the optimization needs to be executed in an infinite dimensional functional domain. The project utilizes advanced statistical modeling techniques to make this optimization problem tractable. Specifically, a new functional input-functional output co-kriging model will be developed that relates the stress-strain curves to the fiber insert shapes. A key feature of this modeling technique is a new spectral distance correlation function that makes the model invariant to translations, which is a desirable property for the tissue mimicking problem. Furthermore, advanced techniques in experimental design and Bayesian model calibration are employed to obtain 3D printed valve data and merge them with available historical data.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.