一种结合CV模型与贝叶斯的肺实质分割方法

Combination of CV Model and the Bayesian Method for Lung Parenchyma Segmentation

胸膜结节的灰度与肺实质外围灰度十分接近,很难运用传统算法分割这种病变部位.针对胸膜结节难以精确分割的问题,在本文中提出了一种结合CV模型与贝叶斯模型的优化算法,本算法采用CV模型进行初分割,并在分割结果基础上采用了贝叶斯方法:通过CT图像上一帧来预测并更新胸膜结节信息,最后将筛选出的病变区域添加到初始分割轮廓上,完成肺实质的自动分割.运用本文提出的方法,对来自LIDC公开数据集中的32位病人共计234张CT样本图像进行仿真实验,综合得到本文算法对此类结节分割准确率、召回率和F值分别为99.6%、93.6%、96.5%,较文中所对比算法有明显提升.实验结果表明,该算法具有不错的适应性和鲁棒性,并且提高了此类结节的分割精度.

The gray scale of the pleural nodules is very close to the gray level of the lung parenchyma.It is difficult to segment the lesions by traditional algorithms.In order to solve the problem that pleural nodules are difficult to be accurately segmented,an optimization algorithm combining CV model and Bayesian model is proposed in this paper.This algorithm uses CV model for initial segmentation,and adopts Bayesian method based on segmentation results:The pleural nodule information is predicted and updated by one frame on the CT image,and finally the selected lesion area is added to the initial segmentation contour to complete the automatic segmentation of the lung parenchyma.Using the method proposed in this paper,a total of 234 CT sample images from 32 patients in the LIDC public dataset were simulated.The accuracy,recall and F value of the algorithm were 99.6%and 93.6 respectively.%,96.5%,compared with the algorithm compared in the text has a significant improvement.The experimental results show that the algorithm has good adaptability and robustness,and improves the segmentation accuracy of such nodules.