Minimally invasive surgery maneuver recognition based on surgeon model

This paper proposes a new user interface based on a maneuver recognition system, which models the surgeon behavior. This interface includes three different modules: data acquisition and coding, training system and on-line recognition system. The aim is defined as recognizing the surgeon movements while is performing a surgical maneuver, by using a 3D surgical tool tracker. The obtained measurements are converted in to movement symbols by means of a Wavelet transform and a fuzzy clustering. These symbols are used both for training HMM and for recognizing the current maneuver. The system has been tested in some in-vitro experiments performing a fictitious surgical protocol.