A Reaction-diffusion Algorithm for Segmentation of Confocal Laser Scanning Microscope Images

抄録

The images obtained using a confocal laser scanning microscope (CLSM) alone cannot detect three-dimensional (3D) structures, for example of the endothelial cells that compose blood vessels, and require segmentation for 3D interpretation. Here, we investigated the segmentation of hepatic sinusiodal veins, which form periodic 3D networks. We propose a new approach for image segmentation based on the Turing reaction-diffusion (RD) model. We performed segmentation of CLSM images of sinusoidal endothelial cells using the proposed RD algorithm. Moreover, we discuss potential applications of this algorithm.