Abstract: The segmentation of lung parenchyma is the foundation of chest CT image processing, such as lung nodule detection, quantitative analysis of lung function, three-dimensional reconstruction, and visualization analysis. This paper uses an edge detection method based on genetic algorithm to segment the lung parenchyma of original chest CT image. With global searching capacity and the largest variance between clusters as the fitness function, this method can search the optimal threshold of edge detection automatically, and extract the edge of lung parenchyma by combining morphologic processing to realize the segmentation of lung parenchyma. Experiment shows that the method can not only simplify the segmentation of lung parenchyma, but also achieve a good segmentation effect. It has a good foreground in application.

Key words: genetic algorithm, edge detection, largest variance between clusters, segmentation of lung parenchyma

摘要： 肺组织分割是肺结节检测、肺功能定量分析、三维重建与可视化计算等胸部CT图像分析处理的基础。该文采用了一种基于遗传算法的边缘检测方法直接分割原始胸部CT图像的肺组织，利用遗传算法的全局寻优能力，以最大类间方差为适应度函数自动搜索最佳边缘检测阈值，并结合形态学处理提取肺组织边缘以实现肺组织分割。实验结果表明，该方法能简化分割处理，且分割效果较好，有不错的应用前景。

关键词: 遗传算法, 边缘检测, 最大类间方差, 肺组织分割

CLC Number: 

• TN911.73