Image Analysis: Segmentation and Registration
A local spatio-temporal coordination of cell growth and cell division plays a critical role in morphogenesis of both the plant and the animal tissues. The causal link between cell growth and cell division patterns and how they, in turn, affect organ formation is not well understood. Information such as rates and patterns of cell expansion play a critical role in explaining cell growth and deformation dynamics. However manual analysis is extremely tedious because of the high dimensionality and complexity of data. Therefore, the development of computational platforms that are capable of identification of cellular coordinates, automated tracking of cells and cell division events is important. Such computational platforms would facilitate the quantification of cellular parameters such as rates and patterns of cell expansion, cell division, and extraction of such information may lead to the development of growth models that can explain the causal relationships between cell deformation dynamics, cell growth and cell division patterns. This process is a computational challenge that has universal application to all developmental fields, both animals and plants. To achieve the goal of quantification of biological parameters and observing their evolution in time, advanced microscopy techniques are used to collect time lapse videos and quantify the behavior of hundreds of cells in a tissue over multiple days. One of these techniques is the Confocal Laser Scanning Microscopy (CLSM) based Live Cell Imaging. This technique allows us to take optical cross sections of the cells in the tissue over multiple observational time points to generate spatio-temporal 4D (X-Y-Z-T) image stacks. To analyze the details of the collected image data a fully automated image processing and analysis framework has been created, which comprises of three main parts – image registration, cell segmentation and cell tracking. Cell tracking module of this framework establishes cell correspondence across multiple slides and time windows and fuses these correspondences to obtain cell lineages, which contain cell life and division statistics. The quality of the cell lineage statistics depends heavily on cellular coordinate evaluations . Hence, without proper segmentation and registration the subsequent part in the image analysis system would fail.
The temporal and spatial growth of the living organisms are captured into 4D (X-Y-Z-T) image stacks with Confocal laser scanning microscopy. To keep the plant alive for a long period of time, it is necessary to limit its exposure to the laser. This results in poor image quality and presents significant challenges to image analysis since the segmentation and the tracking needs to be robust to the poor image quality. We use watershed segmentation and show how to optimally choose the parameters in the watershed algorithm for high quality segmentation results. Collected experimental results are compared to recent results in this area. Quantitative analysis shows that the proposed algorithms provide significantly longer cell lineages and more comprehensive identification of cell divisions.
The procedure capturing the spatio-temporal 4D (X-Y-Z-T) image stacks causes both temporal and spatial misalignments in this live imaging stacks. The spatial shifts between images from different temporal stacks are caused from the involvement of manual work in noncontinuous imaging procedure (physically moving the specimen from one place to another). Also, because of contin-uous growth of the living organism during the imaging procedure there is also a slice matching issue in consecutive stacks to be solved. The presence of these misalignments in image stacks causes collection of non accurate statistics. We present a fully automated 4D(X-Y-Z-T) registration method of live imaging stacks that takes care of both temporal and spatial misalignments. We present a novel landmark selection methodology where the shape features of individual cells are not of high quality and highly distinguishable. The proposed registration method finds the best image slice correspondence from consecutive image stacks to account for vertical growth in the tissue and the discrepancy in the choice of the starting focal point. Then it uses local graph-based approach to automatically find corresponding landmark pairs, and finally the registration parameters are used to register the entire image stack. The proposed registration algorithm combined with an existing tracking method is tested on multiple image stacks of tightly packed cells of Arabidopsis shoot apical meristem and the results show that it significantly improves the accuracy of cell lineages and division statistics.