A ridge scorewith application to piecewise linear neural reconstruction

摘要

We proposed a novel algorithm to extract connectivity information of neuronal arbors from 3D confocal images. The method is based on the use of ridge sampling and approximation using piecewise linear segments that conform to an inequality constraint that ensures shape accuracy. An automatic neuron structure reconstruction algorithm based on kernel interpolation of intensities usingmultiscale local Gaussian kernels has been utilized on a sample Olfactory neuron reconstruction problem successfully.