Sternotomy and craniotomyd by an autonomous robot: Experimental evaluation on seven pigs

Background Autonomous robotic action may facilitate surgery. however, systematic evaluation data are lacking. The purpose of the study was to assess the feasibility, safety, and accuracy of sternotomy and craniotomy performed autonomously by a surgical robot, based on three dimensional (3D) planning. Methods In seven domestic pigs, the complete process of planning and robotic action were evaluated: 3D imaging (spiral CT scans, slice thickness (A) 2 mm vs. 4 mm, (B) 1.5 mm vs. 3 mm), 3D localization of markers, trajectory planning, matching of registration with 3D imaging, registration error, and effector forces (milling cutter). Robot milling trajectories for (A) median sternotomy and (B) craniotomy (n=28) were executed autonomously by a one-arm robot. 3D vector analysis of planned vs. executed trajectories was performed on CT scans (slice thickness 1.5 mm). Force limitation (35 N) was used. Results Robotic action was feasible in 93% of sternotomies and in 86% of craniotomies. No emergency stop occurred. Differences in marker detection were <2 mm, higher in craniotomies (p <0.05). Registration errors were higher in sternotomies (p = 0.03). Mean trajectory deviation was (A) 2.59 +/- 1.14 mm, (B) 2.56+1.21 mm (n.s.). For actions planned on image slices <= 3 mm, deviation was 2.51 +/- 1.16 vs 2.66 +/- 1.13 mm (slices >3 mm) (n.s.). Maximum forces (21-44 N) occurred at the end of trajectories. Conclusions Safety, feasibility and accuracy of the robotic tasks were acceptable. Our data may therefore serve as a base for future autonomic robotic procedures.