Enhanced fluorescent cell simulation using texture mapping and statistical shape model

Manual evaluation of fluorescent microscopy cell experiments is time-consuming and tedious work. Therefore, often automated segmentation algorithms are applied to ensure an efficient evaluation with constantly high quality for the whole dataset. Image segmentation quality is usually validated and rated with manually annotated data, being the "gold-standard", but which is time-consuming, prone to errors and furthermore shows inter- and intra-labeler variances. As an alternative, cell simulation can be used for the generation of cell images and related real ground truth data. Within cell simulation frameworks, individual cells can be modeled with specific shape and texture. For the simulation of cell, we use statistical models computed from manually annotated real cell images. Different types of cell textures can be mapped from cells extracted from real micrographs into the simulated cell shapes. In this contribution we present resulting images for HeLa cells that show that our approach is suitable for the simulation of fluorescent micrographs.