Automated data curation to ensure model credibility in the Vascular Model Repository

自动数据管理以确保血管模型存储库中模型的可信度

• 批准号: 10175029
• 负责人: Alison L Marsden
• 金额: $ 33.03万
• 依托单位: STANFORD UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-09-12 至 2023-05-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10175029
• 关键词: 3-Dimensional; Adoption; Anatomic Models; Anatomy; Basic Science; Blood Vessels; Blood flow; Cardiac; Cardiovascular Diseases; Cardiovascular Models; Cardiovascular system; Clinical; Clinical Data; Clinical Sciences; Collaborations; Communities; Complex; Computer software; Data; Data Science; Data Set; Databases; Diagnosis; Disease; Electrophysiology (science); Ensure; Feedback; Fostering; Funding; Goals; High Performance Computing; Image; Image Analysis; Incentives; Intake; Intervention; Joints; Laws; Machine Learning; Manuals; Maps; Mechanics; Medical Device Designs; Medical Imaging; Methods; Modeling; Musculoskeletal; Operative Surgical Procedures; Patient risk; Patients; Physics; Physiological; Process; Publications; Radiology Specialty; Recording of previous events; Reproducibility; Research; Resolution; Resources; Risk Assessment; Running; Science; Software Engineering; Source Code; Supervision; System; Techniques; Time; Triage; Uncertainty; United States National Institutes of Health; automated analysis; base; clinical application; clinical care; computing resources; data curation; data format; data repository; data resource; deep learning; experience; gigabyte; imaging Segmentation; innovation; large scale data; models and simulation; novel; online repository; open data; open source; repository; respiratory; shape analysis; simulation; software development; stem; success; supercomputer; supervised learning; three-dimensional modeling; treatment planning; trustworthiness; unsupervised learning; web portal

Three-dimensional anatomic modeling and simulation (3D M&S) in cardiovascular (CV) disease have become a crucial component of treatment planning, medical device design, diagnosis, and FDA approval. Comprehensive, curated 3-D M&S databases are critical to enable grand challenges, and to advance model reduction, shape analysis, and deep learning for clinical application. However, large-scale open data curation involving 3-D M&S present unique challenges; simulations are data intensive, physics-based models are increasingly complex and highly resolved, heterogeneous solvers and data formats are employed by the community, and simulations require significant high-performance computing resources. Manually curating a large open-data repository, while ensuring the contents are verified and credible, is therefore intractable. We aim to overcome these challenges by developing broadly applicable automated curation data science to ensure model credibility and accuracy in 3-D M&S, leveraging our team’s expertise in CV simulation, uncertainty quantification, imaging science, and our existing open data and open source projects. Our team has extensive experience developing and curating open data and software resources. In 2013, we launched the Vascular Model Repository (VMR), providing 120 publicly-available datasets, including medical image data, anatomic vascular models, and blood flow simulation results, spanning numerous vascular anatomies and diseases. The VMR is compatible with SimVascular, the only fully open source platform providing state-of-the-art image-based blood flow modeling and analysis capability to the CV simulation community. We propose that novel curation science will enable the VMR to rapidly intake new data while automatically assessing model credibility, creating a unique resource to foster rigor and reproducibility in the CV disease community with broad application in 3D M&S. To accomplish these goals, we propose three specific aims: 1) Develop and validate automated curation methods to assess credibility of anatomic patient-specific models built from medical image data, 2) Develop and validate automated curation methods to assess credibility of 3D blood flow simulation results, 3) Disseminate the data curation suite and expanded VMR. The proposed research is significant and innovative because it will 1) enable rapid expansion of the repository by limiting curator intervention during data intake, leveraging compatibility with SimVascular, 2) increase model credibility in the CV simulation community, 3) apply novel supervised and unsupervised approaches to evaluate anatomic model fidelity, 4) leverage reduced order models for rapid assessment of complex 3D data. This project assembles a unique team of experts in cardiovascular simulation, the developers of SimVascular and creator of the VMR, a professional software engineer, and radiology technologists. We will build upon our successful track record of launching and supporting open source and open data resources to ensure success. Data curation science for 3D M&S will have direct and broad impacts in other physiologic systems and to ultimately impact clinical care in cardiovascular disease.

心血管 (CV) 疾病的三维解剖建模与模拟 (3D M&S) 已成为一种 治疗计划、医疗设备设计、诊断和 FDA 批准的重要组成部分。综合的， 精心策划的 3D M&S 数据库对于应对重大挑战以及推进模型简化、形状塑造至关重要 分析和深度学习的临床应用。然而，涉及 3D M&S 的大规模开放数据管理 提出独特的挑战；模拟是数据密集型的，基于物理的模型越来越复杂， 社区采用高度解析的异构求解器和数据格式，并进行模拟 需要大量的高性能计算资源。手动管理大型开放数据存储库，同时 因此，确保内容经过验证和可信是很困难的。我们的目标是克服这些挑战 通过开发广泛适用的自动化管理数据科学来确保模型的可信度和 3-D M&S 的准确性，利用我们团队在 CV 模拟、不确定性量化方面的专业知识， 成像科学，以及我们现有的开放数据和开源项目。我们的团队拥有丰富的经验 开发和管理开放数据和软件资源。 2013年，我们推出了血管模型库 (VMR)，提供 120 个公开数据集，包括医学图像数据、解剖血管模型和 血流模拟结果，涵盖众多血管解剖结构和疾病。 VMR兼容 SimVasular 是唯一一个完全开源的平台，提供最先进的基于图像的血流建模 CV 仿真界的分析能力。我们建议新颖的策展科学将使 VMR 可快速获取新数据，同时自动评估模型可信度，从而创建独特的资源 培养 CV 疾病领域的严谨性和可重复性，并在 3D M&S 中广泛应用。为了完成 为了实现这些目标，我们提出了三个具体目标：1）开发和验证自动管理方法来评估 根据医学图像数据构建的患者特定解剖模型的可信度，2) 开发和验证自动化 评估 3D 血流模拟结果可信度的管理方法，3) 传播数据管理套件 并扩展了VMR。拟议的研究具有重要意义和创新性，因为它将 1) 实现快速 通过限制管理者在数据获取期间的干预来扩展存储库，利用与 SimVasular，2）提高模型在 CV 模拟社区中的可信度，3）应用新颖的监督和 无监督方法评估解剖模型保真度，4）利用降阶模型快速 评估复杂的 3D 数据。该项目组建了一支独特的心血管模拟专家团队， SimVascular 的开发者和 VMR 的创建者、专业软件工程师和放射学 技术人员。我们将在推出和支持开源和开放的成功记录的基础上再接再厉。 数据资源是成功的保证。 3D M&S 的数据管理科学将对其他领域产生直接和广泛的影响 生理系统并最终影响心血管疾病的临床护理。

项目成果

svMorph: Interactive Geometry-Editing Tools for Virtual Patient-Specific Vascular Anatomies

• DOI: 10.1115/1.4056055
• 发表时间: 2023-03-01
• 期刊: JOURNAL OF BIOMECHANICAL ENGINEERING-TRANSACTIONS OF THE ASME
• 影响因子: 1.7
• 作者: Pham，Jonathan;Wyetzner，Sofia;Marsden，Alison L. L.
• 通讯作者: Marsden，Alison L. L.

Automated generation of 0D and 1D reduced-order models of patient-specific blood flow.

• DOI: 10.1002/cnm.3639
• 发表时间: 2022-10
• 期刊: INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN BIOMEDICAL ENGINEERING
• 影响因子: 2.1
• 作者: Pfaller, Martin R.;Pham, Jonathan;Verma, Aekaansh;Pegolotti, Luca;Wilson, Nathan M.;Parker, David W.;Yang, Weiguang;Marsden, Alison L.
• 通讯作者: Marsden, Alison L.

Branched Latent Neural Maps

分支潜在神经图

• DOI: 10.1016/j.cma.2023.116499
• 发表时间: 2024
• 期刊: Computer Methods in Applied Mechanics and Engineering
• 影响因子: 7.2
• 作者: Salvador, Matteo;Marsden, Alison Lesley
• 通讯作者: Marsden, Alison Lesley

Computational simulation-derived hemodynamic and biomechanical properties of the pulmonary arterial tree early in the course of ventricular septal defects.

• DOI: 10.1007/s10237-021-01519-4
• 发表时间: 2021-12
• 期刊: BIOMECHANICS AND MODELING IN MECHANOBIOLOGY
• 影响因子: 3.5
• 作者: Dong, Melody L.;Lan, Ingrid S.;Yang, Weiguang;Rabinovitch, Marlene;Feinstein, Jeffrey A.;Marsden, Alison L.
• 通讯作者: Marsden, Alison L.

Hemodynamic effects of entry and exit tear size in aortic dissection evaluated with in vitro magnetic resonance imaging and fluid-structure interaction simulation.

• DOI: 10.1038/s41598-023-49942-0
• 发表时间: 2023-12-18
• 期刊: Scientific reports
• 影响因子: 4.6