Approach for automated segmentation and detection of dendrites and spines from fluorescence confocal image based on skeleton tree

This paper proposed an approach for automated segmentation and detection of dendrites and spines from fluorescence confocal image.It presented a new adaptive region growing algorithm to the segmentation of the dendrite spines for the preprocessing at the beginning.And then,proposed path planning algorithm based on seed points to compute the shortest path between the origin point and the target point to get the initial skeleton.Used minimum spanning tree based on weight as the description module of the skeleton of dendrite spine.In this foundation,the research of detection of spines focused on the change of the vector angle in seed points and the vertices distance of a spine.Experiment shows that the proposed approach can effectively extract the skeleton of the dendrite and achieve a better result on spines detection.