Inductive and Deductive Reasoning for Robotic Steerable Needle in Neurosurgery

Keyhole approaches in brain surgery require the surgeon to reach targets located in deep brain region. Endpoints can be difficult to reach by traditional rigid needles, without damaging the adjacent tissues. Steerable needles can represent a breakthrough in neurosurgery, as they grant access to sensitive destinations. Computing the optimal trajectory in the preoperative phase consists in a path planning problem, which can be tackled exploiting classical methods, such as graph-based, search-based and learning-based approaches. However these techniques present some limitations: the first two require a trade-off between completeness and efficiency, while the latter needs large datasets to successfully train models. To overcome these drawbacks, we propose to model the path planning problem for steerable needle in neurosurgery combining deductive and inductive reasoning. In particular our system, depicted in figure 1, exploits Answer Set Programming (ASP) semantics to model the brain environment and thus implement an artificial intelligent agent able to move within it, satisfying requirements, which can be customized depending on the specific application and based on the preferences expressed by domain experts, as surgeons and clinicians.