FluoRender: Rapid Quantitative Analysis and Adaptive Workflows for Fluorescence Microscopy Data in Fundamental Biomedical Research

摘要 FluoRender是一个用于多通道和多维的交互可视化和分析的软件包 荧光显微镜数据。该项目将满足生物学家利用荧光显微镜进行 灵活可靠的数据分析和解决基础生物医学研究中需要快速解决的问题 测量和工作流原型。具体目标1：互动和协作衡量和分析 大型多维显微镜数据。我们将增加专门为三个试点项目设计的快速测量工具 我们在犹他大学的亲密合作者的研究。FluoRender将充分利用最新的图形 处理单元(GPU)计算技术和流处理，以交互速度处理大数据，确保 合作项目的成功。具体目标2：将机器学习应用于用户工作流程和数据分析。 我们将支持FluoRender用户的各种数据分析需求，并使用 机器学习。我们将在人在环中的方法中整合用户交互，以解决以下问题 训练样本不足，提高了机器学习的可解释性。具体目标3：相互之间的互操作性 FluoRender等流行的开源图像分析软件。我们将支持从调用ImageJ/斐济模块 FluoRender用户界面。用户将能够将熟悉的ImageJ/斐济功能与FluoRender交互功能相结合 工具。频繁访问的外部函数将转换为原生的FluoRender实现以改进 效率和准确性。具体目标4：身临其境的体积数据展示。我们将支持增强现实 (Ar)用于身临其境数据分析的耳机和全息显示器。这些新兴的显示技术将拥有更多 比虚拟现实(VR)设备更自然的用户交互，有利于科学分析3D数据 研究。

Summary FluoRender is a software package for interactive visualization and analysis of multichannel and multidimensional fluorescence microscopy data. This project will serve the pressing needs of biologists utilizing fluorescence microscopy for flexible and reliable data analysis and address the problems in fundamental biomedical research that demands rapid measurements and workflow prototyping. Specific Aim 1: Interactive and collaborative measurement and analysis of large multidimensional microscopy data. We will add rapid measurement tools specifically designed for three pilot studies of our close collaborators at the University of Utah. FluoRender will take full advantage of latest graphics processing unit (GPU) computing techniques and streamed processing to handle large data at interactive speed, ensuring the success of the collaborative projects. Specific Aim 2: Applying machine learning to user workflows and data analysis. We will support diverse data analysis needs from FluoRender users and provide automatic workflow assembly using machine learning. We will incorporate user interactions in a human-in-the-loop approach to address the problem of insufficient training examples and enhance interpretability in machine learning. Specific Aim 3: Interoperability between FluoRender and other popular open-source image analysis software. We will support invoking ImageJ/Fiji modules from FluoRender user interface. Users will be able to apply familiar ImageJ/Fiji functions combined with FluoRender interactive tools. Frequently accessed external functions will be converted to native FluoRender implementations to improve efficiency and accuracy. Specific Aim 4: Immersive volumetric data presentation. We will support the augmented reality (AR) headsets and holographic displays for immersive data analysis. These emerging display technologies will have more natural user interactions than the virtual reality (VR) devices and be advantageous for analyzing 3D data in scientific research.