Laparoscopic image analysis for automatic tracking of surgical tools

Resumen

Laparoscopy is a surgical technique nowadays embedded in the clinical routine. Recent researches have been focused on analysing video information captured by the endoscope for extracting cues useful for surgeons, such as depth information. In particular, the 3D pose estimation of the surgical tools presents three important added values: (1) to extract objective parameters for the surgical training stage, (2) to develop an image-guided surgery based on the knowledge of the surgery tools localization, (3) to design new roboticsystems for an automatic laparoscope positioning, according to the visual feedback. Tool’s shape and orientation in the image is the key to get its 3D position. This work presents an image analysis for automatic laparoscopic tool’s detection along the recorded video without extra tool markers, using an edges detection strategy. Also, this analysis includes a previous stage of barrel distortion correction for videoendoscopic image.