Skip to Main Content
Confocal fluorescence microscopy has become a standard tool to image thick 3D tissue samples, permitting the observation of cell behaviour, such as cell division within developing organs. However, robust and automatic extraction of nuclear shape and mitotic orientation may be hindered by a highly cluttered environment as for example in mammalian tissues. We propose a fast and automated framework for the segmentation of nuclei from cluttered 3D images, allowing robust quantification of various parameters such as number of cells, number of mitoses and mitotic orientation. We have applied this framework to scans of the developing mouse heart, and manual validation on three independent experiments indicates a detection rate of 93% in all cases. Moreover, the proposed tool permits fast, real-time 3D rendering of the data set during the analysis, and can be easily adapted to other applications related to dense tissue analysis.