An Evolutionary-Optimized Surgical Path Planner for a Programmable Bevel-Tip Needle

Abstract

Path planning algorithms for steerable needles in medical applications must guarantee the anatomical obstacle avoidance, reduce the insertion length and ensure the compliance with the needle kinematics. The majority of the solutions from the literature focus either on fast computation or on path optimality, the former at the expense of sub-optimal paths, the latter by making unbearable the computation in case of a high dimensional workspace. We implemented a 3D path planner for neurosurgical applications which keeps the computational cost consistent with standard pre-operative planning algorithms and fine-tunes the estimated pathways in accordance to multiple optimization objectives. From a user-defined entry point, our method confines a sample-based path search within a subsection of the original workspace considering the degree of curvature admitted by the needle. An evolutionary optimization procedure is used to maximize the obstacle avoidance and reduce the insertion length. The pool of optimized solutions is examined through a cost function to determine the best path. Simulations on one dataset showed the ability of the planner to save time and overcome the state of the art in terms of obstacle avoidance, insertion length and probability of failure, proving this algorithm as a valid planning method for complex environments.

Insights into Infusion-Based Targeted Drug Delivery in the Brain: Perspectives, Challenges and Opportunities

“…In the literature, there are many different approaches to preoperative planning of steerable needles. Besides more recent attempts exploiting evolutionary strategies [ 165 ] or machine learning techniques [ 166 ], the majority of the state-of-the-art planners are graph-based methods, such as A* [ 167 , 168 ], or sampling-based methods, such as Rapidly exploring Random Trees (RRT) [ 169 , 170 , 171 , 172 ] and Adaptive Fractal Trees (AFT) [ 173 , 174 ]. In [ 175 ], the authors presented a new searched-based planner based on [ 176 , 177 ] for steerable needles that guarantees completeness under some clinically reasonable assumptions.…”

A hybrid inductive learning-based and deductive reasoning-based 3-D path planning method in complex environments

“…Paths computed with the approaches mentioned above can be further refined using Bezier curves (Hoy et al, 2015), splines (Lau et al, 2009), polynomial basis functions (Qu et al, 2004), or with optimisation-based methods such as evolutionary algorithms, simulated annealing, and particle swarm (Besada-Portas et al, 2010) to obtain a smooth path. These approaches have the advantage of working properly in complex environments, as demonstrated in Favaro et al (2021); however, they require higher computational time than the sampling-based methods.…”

A Heuristically Accelerated Reinforcement Learning-Based Neurosurgical Path Planner

“…Optimization-based algorithms, especially the particle swarm optimization algorithm, and evolutionary algorithm have been adopted a lot. An evolutionary optimization-based 3D path planning algorithm [ 7 ] was adopted in a programmable bevel-tip needle. Segato et al.…”