基于Freeman链码的病变肺实质分割

Pathological lung segmentation for images based on Freeman chain code

针对传统的肺实质分割方法对临床上的大面积病变肺的分割效果不理想,提出一种结合改进模糊C均值聚类与Freeman链码算法的肺实质分割方法。用改进模糊C均值聚类算法对CT图像粗分割,结合Freeman链码算法生成的三链码差对缺失的肺实质边缘进行修复,获得完整的肺实质区域。从LIDC数据库中选取20个CT序列图像进行实验,平均分割精度为96%。实验结果表明,无论肺部有无大面积病变,该算法对肺部CT图像均具有理想的分割效果,无需人工干预,算法鲁棒性强。

The traditional algorithms of pulmonary segmentation are not ideal for the segmentation of large lung lesions in clinical practice.To solve the problem,a lung segmentation algorithm based on improved fuzzy C-means clustering and Freeman chain code algorithm was proposed.The improved fuzzy C-means clustering algorithm was used to obtain rough CT images.The Freeman chain code method was used to repair the internal contours.Twenty CT images were selected from the LIDC database for experimentation.The average segmentation accuracy is 96%.Experimental result show that the proposed method has ideal accuracy and strong robustness.It is fully automatic on lung CT images regardless of whether there is large pathological lung.