Robust 3D Pose Estimation of a Laparoscopic Instrument with three Landmarks

Abstract

Knowing surgeon movements during laparoscopic training may provide useful data to speed up the learning process by means of instantaneous error correction and performance evaluation. The first step toward this knowledge is the identification of laparoscopic tool pose in the training environment. In this paper we propose a method to estimate in real time the 3D pose of laparoscopic instruments using a standard camera and three non invasive colored markers applied on the tool stem. The proposed method takes advantage of closed form solution for the problem which speeds up the computation and improves the precision and accuracy of the results. In addition the method handles occlusions even without any marker tracking algorithm thanks to the automatic identification of the insertion point. The method is evaluated in terms of precision, accuracy and real time execution. Results show that it can be effectively used in common training scenarios.