Abstract
In the context of surgical navigation systems based on augmented reality (AR), the key challenge is to ensure the highest degree of realism in merging computer-generated elements with live views of the surgical scene. This paper presents an algorithm suited for wearable stereoscopic augmented reality video see-through systems for use in a clinical scenario. A video-based tracking solution is proposed that relies on stereo localization of three monochromatic markers rigidly constrained to the scene. A PnP-based optimization step is introduced to refine separately the pose of the two cameras. Video-based tracking methods using monochromatic markers are robust to non-controllable and/or inconsistent lighting conditions. The two-stage camera pose estimation algorithm provides sub-pixel registration accuracy. From a technological and an ergonomic standpoint, the proposed approach represents an effective solution to the implementation of wearable AR-based surgical navigation systems wherever rigid anatomies are involved.
Highlights
Augmented reality (AR) [1] is a ground-breaking technology in machine vision and computer graphics and may open the way for significant technological developments in the context of image-guided surgery (IGS)
The second row shows the results of Color Segmentation, right camera native frames are shown
The second row shows the results of Color Segmentation, and the third row shows the results of Circular Shape Recognition
Summary
Augmented reality (AR) [1] is a ground-breaking technology in machine vision and computer graphics and may open the way for significant technological developments in the context of image-guided surgery (IGS). AR in IGS allows merging of real views of the patient with computer-generated elements generally consisting of patient-specific three-dimensional (3D) models of anatomy extracted from medical datasets (Figure 1). In this way, AR establishes a functional and ergonomic integration between surgical navigation and virtual planning by providing physicians with a virtual navigation aid contextually blended within the real surgical scenario [2]. There has been a growing research interest in AR in medicine, which has driven a remarkable increase in the number of published papers. After 13 years, on 31 December 2008, the number of publications reached 255. During the last seven years, between 1 January 2009 and 30 April 2016, 647 papers were
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
