The segmentation and analysis of biological objects in 3D is of increasing importance since laser scanning confocal microscopy, that produces highly resolutive 3D images, is commonly used by biologists. Beyond a proper 3D view of cellular processes, the segmentation and subsequent statistical analysis, including volume measurement and description of objects distribution, is thus highly relevant.Detecting spots in fluorescence microscopy can be a quite difficult task. The spots, that correspond to the localisation of interest molecules, often have a size under or near system resolution (around 200 nm in fluorescence microscopy). They may also gather together and usual tools for segmentation lead to the detection of one object where there may be two or more. An interesting application is the analysis of the nuclear organization in cells with an accurate description of the distribution of genes, Chromosomal territories and centromeres in the nucleus and with respect to nuclear subcompartments.
In this paper we present tools to process, segment and analyse 3D objects. We focus more on 3D spots but some tools or ideas can be easily adapted to any type of 3D objects. We present 3D filters particularly adapted to 3D spots detection like TopHat or Laplacian of Gaussian (LoG), we also present tools to separate close objects. We developed a segmentation procedure which respect spots topology and avoid merging of close spots. We also present tools to separate spots after segmentation, and other 3D methods like distance analysis, and finally we present statistical tools to analyse the spatial distribution of 3D spots. Statistical tools relate to the analysis of the distribution of distances between objects. More sophisticated methods are based on simulations of random points inside the objects. All presented tools can be automated trough the used of ImageJ macros. Some applications are presented with the analysis of FISH signals and synaptic receptors labelling.
Image Processing, Image Analysis, Nuclear organisation, 3D, 4D, 5D, FISH
Presenting author: Thomas Boudier
co-authors: P. Andrey, B. Amairi, J. Ollion, C. Escudé, N. Heck