Desenvolvimento de algoritmos para construção automática de florestas de árvores arteriais

Abstract

Modern medical imaging devices do not have enough resolution to detect vascular segments with a diameter below a fraction of a millimeter. However, it is needed to have a geometric model of those vascular segments to study blood hemodynamics to aid in the diagnosis of disease and planning treatments. In this work, we propose new algorithms based on the Constrained Constructive Optimization (CCO) method. These algorithms generate a geometric model of the vascular system structure with many trees, considering physiological flow, pressure conditions, and minimum intravascular volume while constructing the model. We use those algorithms to generate vascular forests on convex and non-convex perfusion domains. The allometric and morphometric properties from the models are in concordance with those from real vascular trees reported on the specialized literature.