Research Collection Doctoral Thesis Computational methods for quantitative single-cell time-lapse microscopy Author(s): Hilsenbeck, Oliver Publication Date: 2019 Permanent Link: https://doi.org/10.3929/ethz-b-000335898 Rights / License: In Copyright - Non-Commercial Use Permitted This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use. ETH Library DISS. ETH NO. 25681 Computational methods for quantitative single-cell time-lapse microscopy A thesis submitted to attain the degree of DOCTOR OF SCIENCES of ETH ZURICH (Dr. sc. ETH Zurich) presented by OLIVER HILSENBECK M.Sc., Technische Universität München and Ludwig-Maximilians-Universität München born on 07.08.1987 citizen of Germany accepted on the recommendation of Prof. Dr. Timm Schroeder, examiner Prof. Dr. Jörg Stelling, co-examiner Prof. Dr. Tanja Stadler, co-examiner 2019 Contents Summary ................................................................................................................................................. 4 Zusammenfassung ................................................................................................................................... 5 Chapter 1: Introduction .......................................................................................................................... 6 1.1 Motivation ..................................................................................................................................... 7 1.2 Scope and structure of this thesis .................................................................................................. 8 Chapter 2: fastER: a user-friendly tool for ultrafast and robust cell segmentation in large-scale microscopy .............................................................................................................................................. 9 2.1 Abstract ....................................................................................................................................... 10 2.2 Introduction ................................................................................................................................. 11 2.3 Materials and methods ................................................................................................................. 11 2.4 Results ......................................................................................................................................... 19 2.5 Discussion ................................................................................................................................... 25 2.6 Acknowledgements ..................................................................................................................... 26 Chapter 3: Software tools for single-cell tracking and quantification of cellular and molecular properties ............................................................................................................................................... 28 3.1 Software tools for single-cell tracking and quantification ........................................................... 29 3.2 Acknowledgements ..................................................................................................................... 35 3.3 Authors contributions .................................................................................................................. 35 Chapter 4: An open file format for cell lineage trees with continuous tracking and quantification data ............................................................................................................................................................... 37 4.1 An open file format for cell lineage trees .................................................................................... 38 Chapter 5: A probabilistic distance measure to locate errors in cell tracking data ............................... 42 5.1 Abstract ....................................................................................................................................... 43 5.2 Introduction ................................................................................................................................. 44 5.3 Methods ....................................................................................................................................... 45 5.4 Results ......................................................................................................................................... 54 5.5 Discussion and outlook ............................................................................................................... 56 5.6 Acknowledgments ....................................................................................................................... 56 Chapter 6: Discussion and conclusion .................................................................................................. 58 6.1 Discussion and conclusion .......................................................................................................... 59 Appendix A: Supplementary material to chapter 2 .............................................................................. 62 A.1 Additional algorithmic details .................................................................................................... 63 A.2 Additional evaluation details ...................................................................................................... 69 A.3 Population dynamics of in-vitro blood formation ...................................................................... 74 A.4 Supplementary figures ................................................................................................................ 75 A.5 Supplementary movies ............................................................................................................... 77 2 Appendix B: Supplementary material to chapter 3 .............................................................................. 79 B.1 Supplementary notes ................................................................................................................... 80 B.2 Supplementary tables .................................................................................................................. 84 B.3 Supplementary figures ................................................................................................................ 91 B.4 Supplementary videos ................................................................................................................. 98 Appendix C: Supplementary material to chapter 4 ............................................................................ 100 C.1 Tree file format ......................................................................................................................... 101 Acknowledgements ............................................................................................................................. 107 References ........................................................................................................................................... 108 3 Summary Quantitative long-term single-cell time-lapse microscopy is crucial to advance our understanding of cell behavior and its molecular control. Such experiments produce huge amounts of imaging data and thus require robust and efficient algorithms for automated analysis. Automated detection of cell outlines (cell segmentation) and cell tracking, however, are often challenging because high cell densities, cell-to-cell variability, complex cellular shapes, varying image illumination and low signal-to-noise ratios cannot always be avoided. To our knowledge, existing methods are mostly limited to specific analysis problems, lack usability and are either not robust enough or computationally expensive, thus limiting their application to large-scale microscopy. In addition, no method exists that produces perfect results – manual inspection and correction of automatically generated analysis results is therefore crucial to avoid wrong conclusions. Existing tools, however, provide mostly no or only very limited functionality for this purpose. In this cumulative thesis, we present fastER, a trainable tool for automated cell segmentation that is orders of magnitude faster than existing methods while producing state-of-the-art segmentation quality. It supports various cell types and image acquisition modalities but is easy- to-use even for non-experts. In addition, we present tTt and qTfy, which enable robust and efficient quantitative analysis of long-term single-cell imaging data. tTt allows efficient manual inspection and correction of existing tracking data (e.g. from auto-tracking tools) as well as fully manual analysis. The cell lineage trees exported by tTt can then be quantitatively analyzed with qTfy. Furthermore, we developed an open file format for cell lineage trees, including arbitrary quantification data on the level of track points, cells or trees. This is crucial to facilitate data exchange between different analysis tool and laboratories. Finally, we present a probabilistic distance measure to annotate existing tracking data with confidence levels so that errors can be located and corrected more efficiently with tTt. Quantitative evaluation results of their classification performance seem promising, but more tests are required to understand how useful this tool will be in