Multi-modal Tracking of In Vivo Skeletal Structures and Implants

摘要 该R 01应用程序的目标是开发最先进的开源软件，用于基于图像的 骨骼运动学全球有超过2.5亿人患有肌肉骨骼疾病，包括 关节炎、创伤、骨质疏松症和脊柱病变，预计随着人口的增加， 年龄对正常关节功能的深入了解以及与衰老、损伤和 疾病需要定量测量骨骼运动学的能力。目前最先进的 量化骨骼运动学，特别是关节表面的复杂运动，称为关节运动学， 使用来自双平面视频放射成像（BVR）的数据集执行的基于图像的对象跟踪，以及静态和 动态计算机断层扫描（分别为3DCT和4DCT）。不管是哪种成像方式，图像- 基于骨骼的跟踪涉及图像分割和骨骼模型生成，骨骼图像配准， 坐标系选择和数据显示。软件和计算基础设施对于 准确性和效率。缺乏“行业标准”软件或工作流程模板是主要障碍 在这个领域取得进展。实验室使用他们自己的商业，公共领域和定制的组合- 写代码。当前的个性化实施模式效率低下、重复劳动、阻碍发展 合作，更重要的是分享软件和技术进步。最近的重点研讨会和 调查表明，人们显然希望找到更好的解决办法。因此，基于我们长期的专业知识， 基于图像的跟踪，我们将开发一个开源程序，用于基于图像的骨骼运动跟踪能力 接受所有常用成像模式（视频放射成像、3DCT和4DCT）作为输入。我们 长期目标是建立一个由合作者和贡献者组成的全球用户群，以促进创新， 在肌肉骨骼研究的调查。在我们的第一个目标中，我们将与Kitware，Inc.合作。一个有经验和 成功的开源软件开发公司，以完善和增强Autoscoper，并将其集成到 3D Slicer平台生成SlicerAutoscoperM（SAM）。Autoscoper是一个现有的BVR软件程序 由布朗大学开发，用于半自动将骨骼结构（骨骼和植入物）与X射线对齐 视频.将根据项目共同调查人员的投入和已建立的核心用户基础，完善SAM。在 目标2我们将通过将SAM的输出与使用 传统方法，使用四个独立实验室进行的现有研究的数据。最后，在Aim 3中 我们将使用合成模型来评估SAM在四个实验室的循环测试中的准确性（Brown， Cleveland Clinic、马约诊所和皇后大学）使用来自3DCT、4DCT和BVR的图像数据。工作 该提案中概述的将产生一个最先进的开源软件解决方案，该解决方案将接受来自 多种成像模式。SAM将简化和改进基于图像的骨骼跟踪，促进共享 新的分析算法，方法和数据，并加快转化为临床实施。

Abstract The goal of this R01 application is to develop state-of-the-art, open-source software for image-based analysis of skeletal kinematics. Worldwide, over 250 million people are affected by musculoskeletal disorders, including arthritis, trauma, osteoporosis, and spine pathology, a number that is projected to increase as the population ages. The in-depth understanding of normal joint function and the changes associated with aging, injury and disease requires the ability to quantitatively measure skeletal kinematics. The current state-of-the art for quantifying skeletal kinematics – especially the complex motion at the joint surface, called arthrokinematics – is image-based object tracking performed with datasets from biplane videoradiography (BVR), and static and dynamic computed tomography (3DCT and 4DCT, respectively). Regardless of the imaging modality, image- based skeletal tracking involves image segmentation and bone model generation, bone image registration, coordinate system selection, and data presentation. Software and computing infrastructure are critical for accuracy and efficiency. The lack of “industry-standard” software or templates for workflow are major obstacles to progress in the field. Laboratories use their own combination of commercial, public-domain, and custom- written code. The current individualized implementation model is inefficient, duplicates effort, and impedes collaboration, and, importantly, the sharing of software and technical advances. Recent focus workshops and surveys demonstrate clear interest in better solutions. Accordingly, based on our longstanding expertise in image-based tracking, we will develop an open source program for image-based skeletal motion tracking capable of accepting as input all of the commonly used imaging modalities (videoradiography, 3DCT, and 4DCT). Our long-term objective is to build a world-wide user base of collaborators and contributors to foster innovation and inquiry in musculoskeletal research. In our first Aim we will partner with Kitware, Inc. an experienced and successful open-source software development company, to refine and enhance Autoscoper, and integrate it into the 3D Slicer platform to yield SlicerAutoscoperM (SAM). Autoscoper is an existing BVR software program developed at Brown University to semi-automatically align skeletal structures (bones and implants) to x-ray videos. SAM will be refined with input from the project’s co-investigators and an established core user base. In Aim 2 we will determine the agreement and accuracy of SAM by comparing its outputs to those of obtained using legacy methods, using data from existing studies performed in four independent laboratories. Finally, in Aim 3 we will use a synthetic model to evaluate the accuracy of SAM in round-robin testing in four labs (Brown, Cleveland Clinic, Mayo Clinic, and Queens Universiyt) using image data from 3DCT, 4DCT and BVR. The work outlined in this proposal will yield a state-of-the-art, open-source software solution that will accept datasets from multiple imaging modalities. SAM will simplify and improve image-based skeletal tracking, facilitate the sharing of novel analysis algorithms, methodologies, and data, and hasten the translation to clinical implementation.