手术导航系统中的导管配准算法研究

CATHETER REGISTRATION ALGORITHMS IN SURGICAL NAVIGATION SYSTEM

目的研究手术导航系统中导管配准算法,并比较基于隐式马尔科夫模型的导管配准算法与基于形状相似度的导管配准算法的效果。方法选取由荷兰Rotterdam大学医学中心、法国Henri Mondor大学及意大利Circolo医院联合提供的开源数据集。在此数据集上分别执行基于隐式马尔科夫模型的导管配准算法和基于形状相似度的导管配准算法。对比两种算法配准后成对点之间的平均距离及配准花费时间。结果从成对点的平均距离上看,基于形状相似度的导管配准算法的成对点平均距离小于基于隐式马尔科夫模型的导管配准算法的成对点平均距离;从平均运行时间上看,基于形状相似度的导管配准算法的配准时间远长于基于隐式马尔科夫模型的导管配准算法的配准时间。结论基于隐式马尔科夫模型的导管配准算法虽然在精度上比基于形状相似度的导管配准算法要弱,但是精度上基本满足临床要求,而其配准时间远小于基于形状相似度的导管配准算法,并且可以满足在临床中导管实时配准的需求。基于隐式马尔科夫模型的导管配准算法具有更高的临床价值,值得临床推广应用。

Objective To investigate catheter registration algorithms in the surgical navigation system and the effectiveness of catheter registration algorithms based on hidden Markov model versus those based on shape similarity.Methods The open-source dataset jointly provided by Rotterdam University Medical Center in the Netherlands,Henri Mondor University in France,and Circolo Hospital in Italy was selected,and a catheter registration algorithm based on hidden Markov model or shape similarity was performed on this dataset.The two registration algorithms were compared in terms of the average distance between paired points after registration and time of registration.ResultsCompared with that based on hidden Markov model,the registration algorithm based on shape similarity had a significantly shorter average distance between paired points and a significantly longer time of registration.ConclusionAlthough the catheter registration algorithm based on hidden Markov model has a lower accuracy than that based on shape similarity,its accuracy basically meets clinical requirements.The catheter registration algorithm based on hidden Markov model has a significantly shorter time of registration than that based on shape similarity and can meet the needs of real-time catheter registration in clinical practice.The catheter registration algorithm based on hidden Markov model has a high clinical value and holds promise for clinical application.