Multi-view 3D skin feature recognition and localization for patient tracking in spinal surgery applications

Abstract Background Minimally invasive spine surgery is dependent on accurate navigation. Computer-assisted navigation increasingly used in minimally (MIS), but current solutions require the use of reference markers surgical field for both patient and instruments tracking. Purpose To improve reliability facilitate clinical workflow, this study proposes a new marker-free tracking framework based skin feature recognition. Methods Maximally Stable Extremal Regions (MSER) Speeded Up Robust Feature (SURF) algorithms are applied detection. The proposed multi-camera setup obtaining multi-view acquisitions area. Features can then be accurately detected using MSER SURF afterward localized by triangulation. triangulation error assessing localization quality 3D. Results was tested cadaver dataset eight cases. features entire datasets were found to have an overall 0.207 mm 0.204 SURF. accuracy compared system with conventional markers, serving as ground truth. An average 0.627 0.622 achieved SURF, respectively. Conclusions This demonstrates that setting feasible. technology shows promising results terms accuracy. In future, may further improved exploiting extended processing modern optical imaging techniques applications where crucial.