Meshing of Spiny Neuronal Morphologies using Union Operators

Abstract

Neurons are characterized by thin and long interleaving arborizations in which creating accurate mesh models of their cellular membranes is challenging. While union operators are central for CAD/CAM modeling and computer graphics applications, their applicability to neuronal mesh generation has not been explored. In this work, we present the results of exploring the effectiveness of using union operators to generate high fidelity surface meshes of spiny neurons from their morphological traces. To improve the visual realism of the resulting models, a plausible shape of the cell body is also realized with implicit surfaces (metaballs). The algorithm is implemented in Blender based on its Python API and is integrated into NeuroMorphoVis, a neuroscience-specific framework for visualization and analysis of neuronal morphologies. Our method is applied to a dataset consisting of more than 600 neurons representing 60 morphological types reconstructed from the neocortex of a juvenile rat. The performance of our implementation is quantitatively analyzed, and the results are qualitatively compared to previous implementation. The resulting meshes are applicable in multiple contexts including visualization and analysis of full compartmental simulations and generation of high quality multimedia content for scientific visualization and visual computing (Figure 1).