A spatio-temporally integrated and nonlinear particle tracking system for live cell imaging

尽管光学显微镜的最新进展彻底改变了活细胞成像[1-2]，但现代显微镜的高灵敏度和高分辨率的全部潜力尚未实现。一个关键的障碍是目前细胞生物学家可用的图像分析技术的能力和范围有限。尽管商业程序承诺自动粒子识别和跟踪，并已成功地用于高信噪比环境下的辅助图像分析，但我们广泛的实验测试表明，它们不足以处理低信噪比、差和可变对比度的活细胞图像，并且通常包含多个靠近且运动不一致的粒子。在许多情况下，只能通过将光标放置在物体上一段时间来手动跟踪粒子。一个主要的改进将是利用先进的图像处理和分析算法来处理复杂的延时活细胞数据，就像新的去卷积算法在生物学bbb中显著改善了宽视场成像一样。在本项目中，我们建议研究和开发一种非线性偏微分方程（PDE）方法，作为跟踪活细胞中生物颗粒的新方法。与目前生物学家可用的所有商业软件相比，该方法的一个关键优势是充分利用时移数据中的时间和空间关系，以帮助克服严重的噪声影响和识别真实生物条件下的目标。具体而言，我们将基于PDE方法开发一个时空集成的非线性粒子跟踪系统，并将其集成到ImageJ图像分析套件[4]中，为生物学家提供一个友好的图形界面。

Despite recent advances in light microscopy revolutionizing live cell imaging [1-2], the full potential of the increasing high sensitivity and resolution of modern microscopes has yet to be realised. A key barrier is the limited power and scope of image analysis techniques currently available to cell biologists. Although commercial programs promise automated particle identification and tracking and have been successfully used to assist image analysis in high signal to noise ratio environments, our extensive experimental tests have shown that they are inadequate to deal with live cell images that are of low signal to noise ratio, poor and variable contrast, and often comprise multiple particles in close proximity and with inconsistent movement. In many cases, particles can only be tracked manually by placing the cursor on objects over time. A major improvement would be to exploit advanced image processing and analysis algorithms to deal with complex time-lapse live cell data, in much the same way that new de-convolution algorithms have significantly improved wide-field imaging in Biology [3]. In this project, we propose to research and develop a nonlinear partial differential equation (PDE) method as a new approach to tracking biological particles in live cells. A key advantage of this method over all commercial software currently available to biologists is to make full use of temporal and spatial relationships in time-lapse data to assist in overcoming severe noise effects and recognition of targets in real biological conditions. Specifically, we will develop a spatio-temporally integrated and nonlinear particle tracking system based on the PDE approach and integrate it to the ImageJ image analysis suite [4] to provide a user-friendly graphical interface to biologists.