Synthesis and Modeling of Volumetric Cell Images for Enhanced Training of Machine Learning Classifiers – An Application for Selective Plane Illumination Microscopy (SPIM) - (Project Academy “Engineering Sciences” DFG-Call No. 56)

生物医学研究对显微技术的发展提出了越来越高的要求，以便能够对细胞和组织的生长进行定量分析。现有的3D荧光显微镜技术，如共聚焦显微镜，由于缺乏穿透深度或光毒效应，无法达到这些要求，无法长期监测活组织。选择性平面照明显微镜(SPIM)为这类问题提供了一种解决方案。自动样品定位和图像采集对于重复性、采集率和高通量是必不可少的。一个目标是实现机器学习算法和级联分类器，以实现快速和稳健的图像分析。主要目标是实现用于样品定位和图像采集的闭环反馈系统。

Biomedical research is demanding more and more on developments in microscopy technology to enable quantitative analysis of cell and tissue growth. Established 3D fluorescence microscopy technologies like confocal microscopy are not able to achieve these requirements due to a lack of penetration depth or photo toxic effects prohibiting the long term monitoring of living tissue. Selective Plane Illumination Microscopy (SPIM) offers a solution for this kind of problems. Automated sample positioning and image acquisition are essential for reproducibility, acquisition rate and high throughput. One goal is the implementation of machine learning algorithms and a cascaded classifier to achieve a fast and robust image analysis. The main goal is to realise of a closed loop feedback system both for sample positioning and image acquisition.