When
Where
Speaker: Woody March-Steinman
Title: Within Cells Interlinked
Abstract: We provide a tool for specialized annotation of cells during automated tracking and segmentation. Many experiments track the same cells throughout a movie, but drug screens and oxidative stress experiments, among others, may cause significant cell death that is necessary to annotate for accurate single cell trajectories and censoring for survival analysis. A hybrid tool consisting of a single-frame classifier and Hidden Markov Model sequence corrector is demonstrated on single cell time series data and can be applied on top of any existing cell tracking pipeline to provide accurate cell death annotation with included uncertainty estimates.
We also investigate spatiotemporal patterning of patient-derived colonic organoids. Data is taken from a kinase inhibitor screen on cells that have been grown in a 2-D monolayer and exhibit differential patterning of stem cell niches and epithelial cell regions. A neural network-based tool is developed to identify stem cell regions and to determine alterations to tissue homeostasis under drug perturbations and under uncertain ground-truth labeling. This tool enables more extensive use and analysis of this 2-D organoid system, which maintains some physiological relevance while providing a system that is easy to image over time.