Open Software Platform for Data-Driven Image-Guided Robotic Interventions

项目总结/摘要 全世界每年进行超过2.32亿例手术和干预， 我们尽管手术技术和技巧有所改进，但据报道并发症发生率高达20%。 影像引导是提高手术效果和减少并发症的关键因素。虽然成像是 常规用于手术前的诊断和分期，数百万手术和介入治疗中的大多数 病例仍然在没有与术前图像直接联系的情况下进行，而是依赖于手术医生的 基于经验的决定。正在研究图像引导机器人干预（IGRI）， 这些挑战。潜在的假设是，结合图像引导和机器人辅助 增强医生的能力，以最小的侵入性看到和物理接近病变，从而更好地 临床结果。然而，IGRI系统中的图像信息的使用仅仅遵循当前常规的非常规方法。 机器人程序;这些机器人控制简单的基础上获得的几何信息， 这导致计划与体内实际物理空间之间的差异，这是由于各种 不确定性，包括患者运动、异质性组织特性和 系统组件（例如，传感器、机器人硬件和软件）。因此，IGRI没有显示出明确的 在临床结果方面优于传统的非机器人手术。越来越多的IGRI 应用程序和现实世界的挑战的认识导致了研究兴趣从硬件- 从追求机械精度的以人为中心的方法，到“数据驱动”的方法，在这种方法中， 基于从成像扫描仪、传感器和医疗记录获得的数据进行建模，以进行个性化规划 和控制以获得更好的临床结果。数据驱动的方法需要大量的工程资源 因为涉及的软件组件范围很广。特别是软件组件的集成 在不同的领域开发，即，机器人和医学图像计算，是具有挑战性的，由于缺乏鲁棒的 与所使用的每个组件兼容的软件平台。为了应对这一挑战，我们将 流行的医学图像计算和机器人软件包，即机器人操作系统版本2 （ROS 2）和3D Slicer整合到一个平台上。这个名为SlicerROS的新平台将作为插件分发， 在3D Slicer。SlicerROS将允许整合最先进的机器人技术和医学图像计算工具， 在机器人技术和医学图像计算社区中开发和验证的IGRI 系统我们将致力于以下目标：（目标1）扩展3D Slicer/ROS集成，实现无缝 为数据驱动的IGRI整合医学图像计算和机器人工具;（目标2）传播SlicerROS 在医学图像计算和机器人社区建立一个活跃的用户/开发人员社区， 可持续开源软件开发和维护。

Project Summary/Abstract More than 232 million surgeries and interventions are performed each year worldwide, and 51.4 million in the US. Despite improvements in surgical technology and techniques, complication rates of up to 20% are reported. Image guidance is a key factor in improving surgical outcomes and reducing complications. While imaging is routinely used for diagnosis and staging prior to surgery, the majority of the millions of surgical and interventional cases are still performed without a direct link to pre-operative images, instead relying on the operating physician’s decisions based on experience. Image-guided robotic interventions (IGRIs) are being investigated to address those challenges. The underlying hypothesis is that the combined image-guidance and robot-assistance enhance the physician’s ability to see and physically access the lesion with minimal invasion resulting in a better clinical outcome. However, usage of image information in IGRI systems merely follow current conventional non- robotic procedures; those robots are controlled simply based on the geometric information obtained from the image, which leads to a discrepancy between the plan and the actual physical space in vivo due to various uncertainties, including patient motion, heterogenous tissue properties, and communication latencies between system components (e.g., sensors, robot hardware, and software). As a result, IGRI has not shown a clear advantage over conventional non-robotic procedures in terms of clinical outcome. The growing number of IGRI applications and the recognition of the real-world challenge led to a shift of research interest from a hardware- centric approach, which pursues mechanical precision, to a “data-driven” approach, where the treatment is modeled based on data obtained from imaging scanners, sensors, and medical records for personalized planning and control for better clinical outcomes. The data-driven approach requires sizable engineering resources because of the wide range of software components involved. Particularly, integration of software components developed in different fields, i.e., robotics and medical image computing, is challenging due to the lack of robust software platforms that are compatible with every component used. To address this challenge, we will integrate popular medical image computing and robotics software packages, namely Robot Operating System version 2 (ROS2) and 3D Slicer into a single platform. This new platform, called SlicerROS, will be distributed as a plug- in for 3D Slicer. SlicerROS will allow incorporating state-of-the-art robotics and medical image computing tools, which are developed and validated in the robotics and medical image computing communities, into a single IGRI system. We will work on the following aims: (Aim 1) Extend 3D Slicer/ROS integration to achieve seamless integration of medial image computing and robotics tools for data-driven IGRI; (Aim 2) Disseminate SlicerROS in the medical image computing and robotics communities to build an active user/developer community for sustainable open-source software development and maintenance.