CompBio: Reality-based Data-driven Computer Models for Surgical Simulation

CompBio：用于手术模拟的基于现实的数据驱动计算机模型

• 批准号: 0621999
• 负责人: Jaydev Desai
• 金额: $ 27.5万
• 依托单位: Drexel University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2006
• 资助国家: 美国
• 起止时间: 2006-09-01 至 2007-02-28
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0621999&HistoricalAwards=false
• 关键词: CompBio; Reality; based; Data; driven

Current computational models for deformable tissue simulation in interactive surgical training applications make enormous sacrifices in physical accuracy to achieve real-time performance. This is particularly true of challenging real-time, haptic force-feedback rendering of tool-tissue interactions, where simulations must be ideally performed at near kilohertz rates. Furthermore, even if real-time demands are met, such as by faster processors in coming decades, we must still address our fundamental lack of accurate computational models for (a) force response of realistic, variable, nonlinear, soft tissues, and (b) many nontrivial physical processes associated with realistic tool-tissue interaction that are central to surgical intervention, such as, needle insertion, for example. Consequently, the scientific community can help surgical simulation practitioners improve patient safety by providing both accurate and real-time computer models of soft tissues interactions.INTELLECTUAL MERITS: At the very fundamental level, this project seeks to bridge the apparent gap between efficient computer simulation algorithms, and the realistic surgical tissues and interactions they seek to mimic. Specifically, we propose a multi-disciplinary research program on reality-based measurement and computer simulation that leverages our unique strengths to address three key areas:1. Reality-based modeling: Accurate, realistic mathematical models that describe nonlinear soft-tissue response and calibrated interaction models for performing needle insertion. We will develop a range of experimental apparatuses for measuring soft-tissue responses during needle insertion. All measurements will be accompanied by a rigorous empirical modeling process to arrive at accurate parametric tissue and interaction models.2. Data-driven, real-time, simulation algorithms: Given accurate mathematical models of needle-tissue interaction, very efficient simulation algorithms will be devised for real-time haptics and graphics display. Novel computer models based on data-driven, pre-computed, reduced-coordinate, deformation algorithms will be used to accelerate accurate nonlinear deformation, and needle-tissue contact interactions. To support adaptation, and allow runtime modification, we will combine hierarchical domain decomposition using fast reduced domain models, with direct and data-driven simulation models.3. Closing the loop: Quantitative evaluation of experimental and computer simulation models will help refine and validate our reality-based modeling processes throughout the project. In our final year, we will also develop a simulator-based training system for sample needle insertion tasks.BROADER IMPACTS: The proposed research will make fundamental advances in our ability to simulate and reason about soft-tissue interactions accurately, and will lead to several exciting scientific and clinical possibilities. Scientifically, we will be able to develop accurate and reality-based soft-tissue models based on actual experimental trials, which have wide application in medicine. Optimized numerical algorithms can then build on these accurate nonlinear material and contact interaction models for real-time graphical and haptic force-feedback display of soft tissues. Clinically, this will allow a more widespread use of surgical simulators for resident training (for both minimally invasive and direct procedures), whereby residents will be able to experience more realistic soft-tissue interaction response in surgical tasks. Advancement in this area will also open avenues for modeling any other organ or soft-tissue for which training is desired, after the core reality-based simulation issues are resolved.

在交互式外科训练应用中，当前用于可变形组织模拟的计算模型在物理精度上做出了巨大的牺牲，以实现实时性能。对于工具-组织交互的具有挑战性的实时、触觉反馈力反馈渲染尤其如此，其中模拟必须以接近千赫兹的速率理想地执行。此外，即使实时需求得到满足，比如未来几十年更快的处理器，我们仍然必须解决我们基本缺乏准确的计算模型，以用于(A)现实的、可变的、非线性的软组织的力响应，以及(B)与现实的工具-组织相互作用相关的许多非平凡的物理过程，这些过程对外科干预至关重要，例如，针头插入。因此，科学界可以通过提供准确和实时的软组织相互作用的计算机模型来帮助手术模拟实践者提高患者的安全性。INTELLECTUAL的优点：在最基本的层面上，该项目寻求弥合有效的计算机模拟算法与他们寻求模拟的真实手术组织和相互作用之间的明显差距。具体地说，我们提出了一个多学科的基于现实的测量和计算机模拟的研究计划，利用我们独特的优势来解决三个关键领域：1.基于现实的建模：描述非线性软组织反应的准确、真实的数学模型，以及用于执行针插入的校准交互模型。我们将开发一系列实验仪器，用于测量针插入过程中的软组织反应。所有测量都将伴随着严格的经验建模过程，以得出准确的参数组织和相互作用模型。数据驱动的实时模拟算法：给定针-组织相互作用的精确数学模型，将为实时触觉和图形显示设计非常高效的模拟算法。基于数据驱动、预计算、简化坐标、变形算法的新型计算机模型将用于加速精确的非线性变形和针-组织接触交互。为了支持自适应，并允许运行时修改，我们将使用快速约简域模型的层次化区域分解与直接和数据驱动的仿真模型相结合。闭合循环：实验和计算机模拟模型的定量评估将有助于改进和验证我们在整个项目中基于现实的建模过程。在我们的最后一年，我们还将为样本针插入任务开发一个基于模拟器的培训系统。BROADER影响：拟议的研究将在我们准确模拟和推理软组织相互作用的能力方面取得根本性进展，并将带来几种令人兴奋的科学和临床可能性。科学上，我们将能够在实际实验试验的基础上开发出准确和现实的软组织模型，这些模型在医学上有广泛的应用。然后，优化的数值算法可以建立在这些精确的非线性材料和接触交互模型的基础上，用于软组织的实时图形和触觉反馈力显示。在临床上，这将允许更广泛地使用外科模拟器进行住院医师培训(用于微创和直接手术)，从而住院医师将能够在外科任务中体验更真实的软组织交互反应。这一领域的进展还将为在基于现实的核心模拟问题得到解决后，为需要训练的任何其他器官或软组织建模开辟道路。