$3D+t$ Morphological Processing: Applications to Embryogenesis Image Analysis

摘要

We propose to directly process $3D+t$ image sequences with mathematical morphology operators, using a new classification of the $3D+t$ structuring elements. Several methods (filtering, tracking, segmentation) dedicated to the analysis of $3D+t$ datasets of zebrafish embryogenesis are introduced and validated through a synthetic dataset. Then, we illustrate the application of these methods to the analysis of datasets of zebrafish early development acquired with various microscopy techniques. This processing paradigm produces spatio-temporal coherent results as it benefits from the intrinsic redundancy of the temporal dimension, and minimizes the needs for human intervention in semi-automatic algorithms.