Joint 3D cell segmentation and classification in the Arabidopsis root using energy minimization and shape priors

This paper presents a discrete energy minimization approach to integrate different prior knowledge and image cues for simultaneous cell segmentation and classification. When there are multiple types of cells to segment, the segmentation of cells and the classification of the cell types are dependent on each other. The presented approach selects the optimal segmentations from hypotheses and infers the cell types in the same process. The approach is applied to the volumetric data of Arabidopsis roots.