Automated Neuron Tracing and 3D Reconstruction Software

描述（由申请人提供）：我们建议开发一种商业软件产品，AutoNeuron TM，用于自动3D神经元重建。该软件将在带有微软Windows(r)操作系统的标准台式PC上运行，并对从标本幻灯片获取的图像进行处理。多种图像模式可以使用，包括共聚焦显微镜，明场，和宽视场荧光。这个项目的一个主要创新是，自动神经元将在交互和无人值守模式下运行。基于边缘检测和导管跟踪的鲁棒跟踪算法将使许多神经元图像的无人值守自动跟踪成为可能；然而，一些图像将包含复杂性，使自动跟踪变得困难。在这些情况下，科学家在交互模式下标记分支点，而软件自动跟踪标记点之间的段-精确定位中心线并沿着其长度跟踪过程的直径。在第一阶段，我们证明了将一种先进的跟踪算法应用于更大的神经元重建问题的可行性。在这个阶段，我们将开发一个带有基本编辑工具的初步用户界面，用于测试。软件的准确性和效率将与各种神经元类型、染色和成像方法的人工跟踪进行比较评估。一旦追踪引擎的可行性确定，将在第二期开发商业化产品。神经元重建和追踪是许多神经系统形态和功能分析的重要方面。然而，今天许多有价值的研究没有被考虑，因为目前的人工和计算机辅助重建技术实在太费力了。自动化重建工作站将提高许多目前资助的研究以及曾经被认为不切实际的未来研究的效率和生产力。

DESCRIPTION (provided by applicant): We propose development of a commercial software product, AutoNeuron TM to perform automatic 3D neuron reconstruction. This software will run on a standard desktop PC with the Microsoft Windows(r) operating system and is performed on images acquired from specimen slides. Multiple image modalities can be used, including confocal microscopy, brightfield, and widefield fluorescence. A major innovation of this project is that AutoNeuron will operate in both interactive and unattended modes. Unattended automated tracing of many neuron images will be possible with our robust tracing algorithm based on edge detection and conduit tracking; however, some images will contain complexities that make automated tracing difficult. In these cases, the scientist marks branch points in interactive mode, while the software automatically traces the segment between marked points - accurately locating the centerline and tracing the diameter of the process along its length. In Phase I, we prove the feasibility of applying an advanced tracing algorithm to the larger problem of neuron reconstruction. During this phase we will develop a preliminary user interface with basic editing tools to be used in testing. The software accuracy and efficiency will be assessed comparatively with manual tracings from a variety of neuron types, staining, and imaging methods. Once feasibility of the tracing engine is established, a commercial product will be developed in Phase II. Neuron reconstruction and tracing are important aspects of many analyses of form and function in the nervous system. However, today many valuable studies are not considered because current manual and computer assisted reconstruction techniques are simply too laborious. An automated reconstruction workstation will improve the efficiency and productivity of many currently funded studies as well as future studies once considered impractical.