Automatic tracking of laparoscopic instruments for autonomous control of a cameraman robot

Abstract

BACKGROUND: An automated instrument tracking procedure was designed and developed for autonomous control of a cameraman robot during laparoscopic surgery. MATERIAL AND METHODS: The procedure was based on an innovative marker-free segmentation algorithm for detecting the tip of the surgical instruments in laparoscopic images. A compound measure of Saturation and Value components of HSV color space was incorporated that was enhanced further using the Hue component and some essential characteristics of the instrument segment, e.g., crossing the image boundaries. The procedure was then integrated into the controlling system of the RoboLens cameraman robot, within a triple-thread parallel processing scheme, such that the tip is always kept at the center of the image. RESULTS: Assessment of the performance of the system on prerecorded real surgery movies revealed an accuracy rate of 97% for high quality images and about 80% for those suffering from poor lighting and/or blood, water and smoke noises. A reasonably satisfying performance was also observed when employing the system for autonomous control of the robot in a laparoscopic surgery phantom, with a mean time delay of 200ms. CONCLUSION: It was concluded that with further developments, the proposed procedure can provide a practical solution for autonomous control of cameraman robots during laparoscopic surgery operations.