Single-Cell Volume Determination for 3D Segmentation of Human Cancer and Nontumourigenic Cells to Identify Autophagic Phenotypes in High-Content Screening

High-content screening is the systematic analysis of hundreds or thousands or images to extract phenotypic data from cell-based assays. Most commonly, cells cultured on two-dimensional (2D) plastic substrata are studied by high-content imaging. Cells however behave differently when grown on tissue culture plastic, compared to a three-dimensional (3D) environment. For example, non-malignant and malignant cells show dramatic morphological and biochemical differences in 2D and 3D. Signal transduction pathways are integrated in different ways in the same cell line when cultured in monolayer or forming a 3D sphere on reconstituted basement membrane. Therefore 3D model systems of human cells present an obvious choice in high-content screening. The complexity of 3D image processing algorithms forms one of the critical bottlenecks in the screening when automatically quantifying phenotypic measures at cellular level. Here we present a local 3D segmentation approach, where each 3D confocal microscopic image stack was partitioned into several sub-volumes using nuclei as cell markers. As a result, image stacks with heterogenic fluorescence intensity cells can be segmented more precisely than using a global thresholding algorithm. We also present an automated 3D segmentation approach that can be used to segment cells that form 3D structures in confocal image stacks. This allows the measurement of single-cell 3D phenotypic features of interest, and classify cells in a more robust manner compared to 2D approach. Results of the image segmentation and object identification will be presented, merits and different aspects of the segmentation discussed. This research was supported by a Marie Curie International Reintegration Grant within the 7th European Community Framework Programme