This paper focuses on the structure and the concept of the framework
used in the vesselness filters benchmark that was recently
introduced. Vesselness filters are used to detect the presence of
vessels in an image. There exists a wide variety of such filters and
comparing their respective strengths and weaknesses is a non-trivial
task, especially given the different contexts in which they are
published. This benchmark was designed to ease such comparison
process whereas remaining easy to customize. More specifically, this
paper presents the benchmark structure and architecture. It also
shows how to integrate new vesselness filters and/or new metrics in
the benchmark with the requirements for future comparisons and online
demonstrations.