Automatic detection of endoscope in intraoperative CT image: Application to AR guidance in laparoscopic surgery

Abstract

Augmented reality in minimally invasive surgery has rapidly grown over the recent years. Commonly, the surgical scene is augmented through the endoscopic view with a 3D model extracted from a preoperative acquisition. Nevertheless, due to the probable pneumoperitoneum and the patient displacement, the organs of interest often drastically change in shape and place. Methods exist that attempt to recover or simulate the distortion between the pre- and intraoperative states, but require yet another intraoperative 3D acquisition in order to reach the required accuracy. In this article, we propose a new approach to automatically registering the reconstruction from an intraoperative CT acquisition with the static endoscopic view, by locating the endoscope tip consciously intruding in the model. This method does not require any external tracking apparatus nor analysis of the endoscopic image. In the last section, we also present some early quantitative and qualitative results proving the feasibility and clinical potential of our approach.