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来执行自动3D神经元重建。该软件将在装有Microsoft Windows(R)操作系统的标准台式个人电脑上运行，并在从标本载玻片中获取的图像上运行。可以使用多种成像方式，包括共焦显微镜、明场和宽场荧光。该项目的一大创新是AutoNeuron将同时在交互和无人值守模式下运行。使用我们基于边缘检测和管道跟踪的稳健跟踪算法，将可能对许多神经元图像进行无人值守的自动跟踪；然而，一些图像将包含复杂的内容，使得自动跟踪变得困难。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线段-准确地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。在此阶段，我们将开发一个初步的用户界面，其中包含用于测试的基本编辑工具。软件的准确性和效率将与来自各种神经元类型、染色和成像方法的手动跟踪进行比较评估。一旦跟踪引擎的可行性确立，将在第二阶段开发商业产品。神经元重建和跟踪是神经系统许多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，因为目前的手工和计算机辅助重建技术太费力了。自动化重建工作站将提高许多目前得到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.