Automatic entry point planning for robotic post-mortem CT-based needle placement.

Introduction: Post-mortem computed tomography guided placement of co-axial introducer needles allows for the extraction of tissue and liquid samples for histological and toxicological analyses. Automation of this process can increase the accuracy and speed of the needle placement, thereby making it more feasible for routine examinations. To speed up the planning process and increase safety, we developed an algorithm that calculates an optimal entry point and end-effector orientation for a given target point, while taking constraints such as accessibility or bone collisions into account.

Technique: The algorithm identifies the best entry point for needle trajectories in three steps. First, the source CT data is prepared and bone as well as surface data are extracted and optimized. All vertices of the generated surface polygon are considered to be potential entry points. Second, all surface points are tested for validity within the defined hard constraints (reachability, bone collision as well as collision with other needles) and removed if invalid. All remaining vertices are reachable entry points and are rated with respect to needle insertion angle. Third, the vertex with the highest rating is selected as the final entry point, and the best end-effector rotation is calculated to avoid collisions with the body and already set needles.

Discussion: In most cases, the algorithm is sufficiently fast with approximately 5-6 s per entry point. This is the case if there is no collision between the end-effector and the body. If the end-effector has to be rotated to avoid collision, calculation times can increase up to 24 s due to the inefficient collision detection used here. In conclusion, the algorithm allows for fast and facilitated trajectory planning in forensic imaging.