New snake algorithm to track neuronal structure in microscopy image

We present a new method using snake to segment elongated objects in optical microscopy image. Such objects include neuronal dendrites, which play an important role in regulating neurological function and have manifest changes in many neurodegenerative diseases, including Alzheimer's disease and Parkinson's disease. Based on our experience, standard gradient vector flow (GVF) snake has problem to track dendrites. The snake tends to stop too early due to the inhomogeneity of the dendrite signal. In the new method, we introduced an additional external force that aims to restart the region growing process and increase the capture range of the GVF snake. We tested the new method on real neuronal images and obtained better segmentation results. We demonstrate the performance of our method by examples.