A generic model for ridges: A new framework to characterise biological planar structures

This work introduces a new generic framework to the analysis of ridges and its application to segmentation of planar structures in bioimaging. Based on a Gaussian model, local information associated to ridges is extracted, which actually characterise them. The theoretical basis of the framework is first presented for the one-dimensional case. Its extension to higher dimensions is then shown to be straightforward by applying the same basis over the maximum curvature direction. In the particular case of 3D, this framework allows characterisation of planar structures. The potential of the methodology to segment planar structures in 3D bioimaging is illustrated with several examples.