Automated Cellular Phenotyping for Large-Scale Drug and siRNA Knockdown Screening by Live Cell Microscopy
Roland Eils
University of Heidelberg, BIOQUANT, IPMB, and DKFZ Heidelberg, Dept. Bioinformatics and Functional Genomics
The increasing degree of laboratory automation allows to perform high-content screens within relatively short time frames. However, the analysis of the acquired image data still constitutes a bottleneck and requires automated image analysis techniques. Here, we present a fully automated approach to quantify delays in mitotic phases based on 3D fluorescence microscopy image sequences from high-content experiments. Our image analysis workflow comprises segmentation, tracking, image feature extraction, and classification of cells into seven mitotic phases and five morphological phenotype classes. Additionally, we present an automatic approach to check and recover the consistency of the resulting phase sequences and to determine mitotic phase lengths. For the first time our overall approach allows to measure mitotic phase lengths and to detect treatment-induced phase length changes fully automatically which is essential for screening of mitotic delay phenotypes. We validated our scheme on image sequences from experiments with nocodazole and RNAi knockdown experiments. In particular, statistical analysis of the resulting phase lengths provided new insights into effects of nocodazole on mitotic progression.