Piecewise linear cylinder models for 3-dimensional axon segmentation in Brainbow imagery

Generalized cylinder shapes are ubiquitous in biological systems and image processing techniques to identifies these tubular objects in 3D from biomedical imagery in various modalities is a general problem of interest. One such structure that exhibits branching tubular forms are neuronal networks; specifically, recent developments in microscopy imaging technology allow researchers to acquire massive amounts of 3D color images of neural structures that need to be tracked in 3D to extract structure for the purpose of studying function. In this paper, we propose a piecewise linear generalized cylinder tracing algorithm that exploits both edge and color information in order to automatically trace axons of neurons in Brainbow imagery. Results indicate that the proposed method can successfully trace multiple axons in dense neighborhoods.