Parallel algorithms for segmentation of cellular structures in 2D+Time and 3D morphogenesis data

Abstract

In this paper we present a robust method to segment 3D and 2D+time biological data. Our method uses distance function and the Generalized Subjective Surfaces (GSUBSURF) level-set segmentation algorithm. We focus on memory-ecient implementation of distance function computation, a novel way of nite volume method discretization of GSUBSURF PDE and the optimized parallelization of our code via the MPI library, enabling us to take the advantage of high-performance computing servers. Several experiments, including optimized parallelization test, validation of parallel GSUBSURF implementation and visual check of method performance on real data, are presented as well.