Neurite mesh generation based on pre-processing

Abstract:
Generating realistic and watertight meshes for the 3D printing of neurons is often challenging. In neuroscience research, meshes are often generated based on one-dimensional morphological data. However, one-dimensional data of neurons have self-crossing and smoothing problems caused by upstream technical bottlenecks. The resulting overlapping and staggered neural mesh are not conducive to subsequent research. In this study, we propose a method to preprocess the one-dimensional structure of neurons. Firstly, self-intersections in the neuron skeleton were detected, and then different self-crossing conditions were adjusted, respectively. Finally, the mesh of neurons is generated from the adjusted skeleton by the mesh generation algorithm. Experiments verify that the proposed method solves the self-intersecting problem in one-dimensional data and generates smooth and watertight neuron meshes without self-intersections.