Methods for User-Controlled Synthesis of Blood Vessel Trees in Medical Applications: A Survey

Various applications in medicine require geometric models of the underlying blood vessel networks. This ranges from anatomical visualizations, via surgical training systems, to machine learning-based anatomical segmentation frameworks. Especially the latter require large amounts of data. However, both manual segmentation of vascular imaging data, as well as hand-crafting of vessel trees by 3D modellers is tedious and time-consuming. As a possible remedy, approaches for automated synthesis of vessel trees have been proposed in the past. These differ in controllabity by a user, the parameters affecting the visual results, and the computational performance. Therefore, in this paper we provide a survey of state-of-the-art methods for blood vessel synthesis, with a focus on user-control. We introduce several selected key approaches, survey options for steering the outcome, and discuss advantages and disadvantages for achieving certain characteristics of the synthesized trees. In the end, we aim to provide insights into optimizing and selecting vascular synthesis models for specific medical applications.