Supplementary MaterialsMathematical description of the MCSTracker algorithm rsif20160725supp1. perform tracking around the embryonic epidermis and compare cellCcell rearrangements to previous studies in other tissues. Our implementation is usually open source and generally relevant to epithelial tissues. embryo, expressing DE-Cadherin::GFP. Observe Experimental methods for details. (studies where phototoxicity provides a barrier Roy-Bz to high-temporal resolution imaging [28C30]. To address this limitation, we propose a novel algorithm for cell tracking that uses only the connectivity of cell apical surfaces (physique?1). By representing the cell sheet as a physical network in which each pair of adjacent cells shares an edge, we show that cells can be tracked between successive frames by finding the (MCS) of the two networks: the largest network of connected cells that is contained in these two consecutive frames. It is then possible to track any remaining cells based on their adjacency to cells tracked using the MCS. Our algorithm does not require the tuning of parameters to a specific application, and scales in sub-quadratic time with the number of cells in the sheet, making it amenable to the analysis of large tissues. We demonstrate here that our algorithm resolves tissue movements, cell neighbour exchanges, cell division and cell removal (for example, by delamination, extrusion or death) in a large number of datasets, and successfully songs cells across sample segmented frames from Roy-Bz microscopy data of a stage-11 embryo. We further show how our algorithm may be used to gain insight into tissue homeostasis by measuring, for example, the rate of cell rearrangement in the tissue. In particular, we find a large amount of cell rearrangement within the observed dataset despite the absence of gross morphogenetic movement. The remainder of the paper is usually structured as follows. In 2, we describe the algorithm for cell tracking. In 3, we analyse the overall performance of Rabbit polyclonal to IFNB1 the algorithm on and datasets. Finally, in 4, we discuss future extensions and potential applications. 2.?Material and methods In this section, we provide a conceptual overview of the core principles underlying our cell tracking algorithm. We focus on providing an accessible, non-technical description rather than including all details required to implement the algorithm from scrape. A comprehensive mathematical description of the algorithm is usually provided in the electronic supplementary material. The input to the algorithm is usually a set of segmented images obtained from a live-imaging microscopy dataset of the apical surface of an epithelial cell sheet. For each image, the segmentation is usually assumed to have correctly recognized which cells are adjacent and the locations of junctions where three or more cells meet. Numerous publicly available segmentation tools can be used for this segmentation step, for example, SeedWaterSegmenter [10] or ilastik [18]. The segmentation is used to generate a polygonal approximation to the cell tessellation (physique?1embryo, taken 5 min apart. Observe Experimental methods for details. There are several cell neighbour exchanges between these images. Black: overlay of the network of cells that this algorithm uses for cell tracking. Cells in the tessellation correspond to network vertices that are connected by an edge if the cells are adjacent. (are tracked correctly by the MCS. Roy-Bz Three cells in each frame are marked by a yellow Roy-Bz (bright) dot. Within the two cell networks, these cells are users of the MCS. However, these cells are not tracked correctly by the MCS. This mismatch occurs as the MCS is found based on the connectivity of cells within the network alone. The fewer connections a cell has.
