SCH: EXP: Connecting surgical training software solutions on portable clients to interactive dynamics engines on the cloud

整形手术的实践建立在科学和艺术的双重基础之上。在计划创伤性或先天性畸形的重建手术时，外科医生需要考虑美学、解剖功能和术后健康作为决定性因素。整形外科医生可以使用的最重要的工具是使用拓扑变化作为医疗补救措施；事实上，大多数重要的修复都是精心设计的，包括切割、缝合、组织旋转和转位。掌握这种复杂的手术操作通常是一个有成就的外科医生和经验有限的外科住院医生之间最显著的特征。任何训练中的外科医生可以利用在手术室之外发展这种认知技能的机会，在减少错误发生率、减少术后并发症和提高患者生活质量方面都有很大的价值。今天，现代多处理器系统的计算能力表明，计算机模拟在这种训练练习中提供帮助的可能性很大。此外，代表视觉计算社区对我们物理世界的虚拟组件的功能建模和仿真的既得利益提供了丰富的数值和算法技术，可以用来实现这一目标。然而，尤其是在整形外科手术方面，在整形外科医生的认知训练像虚拟飞行模拟器上的飞行员训练一样电脑化之前，仍然存在许多挑战：数学、计算机工程和临床科学的贡献团体需要发展一种密切合作和交流的机制，这种合作发展的成果必须为广泛的临床医生所接受，而不是孤立的研究小组使用外来的计算资源。这个跨学科项目旨在生成一个可扩展的、易于部署的虚拟整形手术模拟测试平台，以促进计算机科学家和临床从业者之间的密切合作，并为训练整形外科医生的微妙认知技能提供基础。该项目的活动结合了专业的、具有成本效益的并行计算平台的优势，以及易于使用的轻量级前端客户端（如平板电脑和移动平台），这些客户端可以部署在私人诊所或教学医院的培训设施中。该项目的主要目标是：(i)定义和构建一系列算法技术和系统组件，通过这些技术和系统组件，以高分辨率模拟可变形生物组织力学的能力将作为网络服务提供给轻量级客户端平台；（ii）细化非线性组织的分析控制规律，并在基于虚拟计算机的模拟中利用它们，以非侵入性方式评估；手术设计的可行性，并为手术结果提供初步预测，（iii）验证建模假设，并通过在外科住院医生和经验丰富的专家的实际用户群体中进行认知训练器的试验部署，提高对实际算法需求的理解，（iv）为连续体力学应用提供计算动力学作为云服务的交付蓝图。通过调查网络延迟和带宽限制的作用，并开发算法补救措施来平衡软件解决方案的需求，以促进远程交付。这一活动的预计产品将为面向保健的机构的教学课程提供额外的好处，因为它使学生在学习的早期阶段能够方便和廉价地使用虚拟模拟器，而不要求专门使用或在当地部署专门的并行计算资源。

The practice of plastic surgery is built on a dual foundation of science and art. In planning reconstructive procedures for either trauma-induced or congenital deformities, the surgeon needs to consider aesthetics, anatomical function and post-operative health as deciding factors. The most important tool at the disposal of a plastic surgeon is the use of topology change as a medical remedy; in fact, most nontrivial repairs are designed as elaborate sequences of cutting, suturing, tissue rotation and transposition. Mastery of this intricate choreography of surgical manipulations is often the most distinguishing feature between an accomplished surgeon and a surgical resident with limited experience. Any opportunities that a surgeon in-training can leverage to develop this cognitive skill away from the operating room have great value in reducing the incidence of errors, minimizing post-operative complications and improving the quality-of-life for patients. Today, the computational capacity of modern multiprocessor systems suggests the strong possibility of computer simulation aiding in this training exercise. In addition, the vested interest on behalf of the visual computing community on functional modeling and simulation of virtual components of our physical world provides a wealth of numerical and algorithmic techniques that can be leveraged towards this goal. However, especially for plastic surgery, a number of challenges remain before cognitive training of plastic surgeons would become as computerized as pilot training on virtual flight simulators: the contributing communities of mathematics, computer engineering and clinical science need to develop a mechanism of intimate collaboration and communication, and the fruit of such collaborative development must become accessible to a broad base of clinicians as opposed to instances of segregated research groups with access to exotic computing resources. This interdisciplinary project seeks to generate an extensible, easy-to-deploy testbed of virtual plastic surgery simulation, tailored to facilitate intimate collaboration between computer scientists and clinical practitioners and provide a foundation for training plastic surgeons in delicate cognitive skills. The activities in this project combine the benefits of specialized, cost-effective parallel computing platforms with the ease of use of lightweight front-end clients, like tablets and mobile platforms, that can be deployed in private practices or training facilities of teaching hospitals.The main objectives of this project are: (i) Defining and constructing a pipeline of algorithmic techniques and system components through which the capability of simulating the mechanics of deformable biological tissues at high resolution will be delivered as a network service to lightweight client platforms, (ii) Refining the analytic governing laws of nonlinear tissue, and leveraging them in a virtual computer-based simulation to assess, in a non-invasive manner, the viability of surgical designs and provide preliminary predictions for the outcome of procedures, (iii) Validate modeling assumptions, and improve the understanding of practical algorithmic needs by performing trial deployments of cognitive trainers for craniofacial repairs on actual users groups of surgical residents and seasoned experts, (iv) Provide a blueprint for delivery of computational dynamics as a cloud service for continuum mechanics applications, by investigating the role of network latency and bandwidth limitations and developing algorithmic remedies for balancing the needs of the software solution to facilitate remote delivery. The projected products of this activity will provide additional benefits to the curriculum of teaching health-oriented institutions, by providing easy and inexpensive access to virtual simulators for students in their early phases of study, without mandating exclusive use or local deployment of specialized parallel computing resources.