基于分形维数特征的肺结节形状建模

摘要 针对肺结节形状建模的问题，提出了一种基于复杂网络的分形维数特征的肺结节形状建模的新方法.首先利用形状轮廓上的采样点之间的欧式距离进行网络化建模，然后利用局部二进制模式值对网络进行动态演化，并利用分形维数对复杂网络的复杂性进行分析.相较于传统形状建模方法，本文方法不仅考虑了形状的局部纹理特征，提高了形状发生非刚性形变的抗干扰能力，还不需要对样本形状进行对齐，提高了建模的效率.使用LIDC-IDRI数据库和沈阳盛京医院的CT资料，经仿真实验，结果表明本文方法能够建立表现良好的肺结节形状模型.
Abstract：Aiming at the modeling of pulmonary nodule shape， a new method to model the shape of pulmonary nodules based on the fractal dimension characteristic of complex networks was proposed. Firstly， the Euclidean distance between the sampling points on the shape contour is modeled by network， then the network is dynamically evolved with LBP(local binary patterns) values， and the complexity of the complex network is analyzed by using the fractal dimension. Compared with the traditional shape modeling method， the proposed method not only considers the local texture feature shape and improves the anti-interference ability to resist non-rigid shape deformation， but also improves the efficiency of modeling with no need of sample shape alignment. Based on the LIDC-IDRI database and the CT data of Shengjing hospital of Shenyang， simulation experiments show that this method performs well on shape model of pulmonary nodules.