Risk-based path planning for a steerable flexible probe for neurosurgical intervention

Minimally invasive surgery has become increasingly common in neurosurgical intervention. Recently, various types of flexible needles have been developed to reach deep targets while avoiding important anatomical regions. In a planning stage, it is required to compute a curvilinear trajectory from an entry point on the skull of the patient to the target lesion. We present a path planner that provides a feasible path for a steerable flexible probe which takes into account its mechanical properties and soft tissue morphology surrounding the target to produce a “path of least risk” for the patient. To meet such requirements, a well known Rapidly-exploring Random Trees (RRT)-based method is adopted since it is capable of producing a curvature-constrained path in complex brain risk maps. To alleviate the sub-optimality of RRTs, a multiple growth of trees that ensures several solutions is introduced so that the clinician can evaluate and select a path that minimizes the cost of different weighted competing objectives, such as the length of the path, the clearance from vessels or nervous bundles and, most importantly, the accumulated risk in crossing different brain regions. Simulations using the risk map of a brain's coronal slice and experimental tests using the path planner integrated with the flexible probe, showed that the approach is promising and that the main objectives of the planning method have been achieved.