Automatic segmentation and tracking of vasculature from confocal scanning laser fluorescence microscope using multi-orientation dissections

This paper describes automated techniques for segmenting, tracking and identifying the individual vessels within the vascular network of the tail of zebrafish embryo from serial images obtained from fluorescence Confocal Scanning Laser Microscope (CSLM). Our approach digitally dissects the voxel data and obtained multi-orientation serial sections along different cutting planes of the zebrafish to simplify the vessel segmentation and tracking processes. This approach is particularly effective for the zebrafish embryo tail as its vasculature exhibits distinctly horizontally and longitudinally oriented vascular components. The resulting algorithm successfully identified and labeled not only major vessels such as Caudal Vein and Caudal Artery but also minor ones such as Intersegmental Vessel. The labeled voxels are represented by an Attributed Vessel Representation Graph, which contains low level geometric information of the vessel as well as semantic information for high level analysis and inter-vasculature comparison.