Abstract

The understanding of tumour angiogenesis and response to vascular-targeted drugs are of increasing interest in cancer research. We present 3D images of the in vivo tumour vasculature captured utilising multi-photon microscopy together with the results of manual and semi-automated delineation of the vascular network using novel in-house-developed software and algorithms. The software presented is aimed at aiding in these investigations and other problems where linear or dendritic structures are to be delineated from 3D data sets. A new algorithm, CHARM, based on a compact Hough transform and the formation of a radial map, has been used to automatically locate vessel centres and measure diameters. The robustness of this algorithm to image smoothing and noise has been investigated. Statistical information characterising the network in terms of vascular parameters as well as more complex analyses, such as fractal dimension, are now possible and examples are presented.

© 2003 SPIE