Numerical algorithm for tracking cell dynamics in 4D biomedical images (Q2515718): Difference between revisions

The article is motivated by the recent research in biology and medicine where the reconstruction of cell population dynamics is crucial for obtaining the cell lineage trees and study of formation and evolution of morphogenetic structures. Such research is related to embryonic development of organisms as well as to anticancer drug design. The authors present the new method for tracking cells in 4D biomedical images (3D+time image sequences), based on the extraction of the cell trajectories as centered paths inside 4D spatio-temporal tree structures obtained by segmentation of 4D images. At first the approach to the trajectories extraction and numerical approaches used in the tracking method are introduced. Then numerical experiments devoted to testing the tracking method are discussed together with processing of the real 3D+time image sequence of the early zebrafish brain embryogenesis. The cell tracking algorithm for the cell trajectories extraction is composed of the following steps: {\parindent=6mm \begin{itemize} \item[(1)] construction of a 4D segmentation yielding the 4D spatio-temporal tree structure inside which two (constrained) distance functions will be computed, \item [(2)] computation of the first constrained distance function D giving distance of any point to the most far cell identifier to which it is continuously connected inside the 4D segmentation, \item [(3)] computation of the second constrained distance function \(D_B\) giving distance of any point to the boundary of the 4D segmentation, \item [(4)] building a potential field \(V\) for tracking by using a suitable combination of two computed distance functions, \item [(5)] extraction of the steepest descent paths of the potential field, \item [(6)] centering the extracted paths inside the 4D spatio-temporal trees in order to get the cell trajectories. \end{itemize}} The naturally parallelizable discretization of the eikonal equation which is used for computing distance functions is given and results of the tracking method for real 4D image data are presented and discussed.