【摘要】 针对磨玻璃肺结节(Ground Glass Opacity,GGO)边界对比度低、大小各异和灰度不均匀等造成分割准确率低的问题,提出一种基于边缘敏感的SLIC和二次密度聚类相结合的分割算法。将图像边缘检测结果与SLIC超像素算法相结合,并将其中含有边缘的超像素块用区域质心代替其原始聚类中心,改善SLIC边界黏连性较差的问题;针对密度聚类不能完整分割GGO的问题,提出二次密度聚类的方法,对密度聚类定位到的簇及其邻域簇进行二次密度聚类。实验结果表明,该算法分割GGO的平均准确率达90.17%,灵敏度达84%。更多还原

【Abstract】 Aiming at the low segmentation accuracy caused by low contrast, different sizes and uneven gray scale of ground glass pulmonary opacity(GGO), this paper proposes a segmentation algorithm based on edge sensitive SLIC and quadratic density clustering. The image edge detection results were combined with the SLIC super pixel algorithm, and the super pixel block with edge was replaced by the center of mass of the region for its original clustering center to improve the poor adhesion of the SLIC boundary. In addition, to solve the problem that GGO cannot be completely divided by density clustering, a method of quadratic density clustering is proposed to carry out the quadratic density clustering for the clusters located by density clustering and their neighborhood clusters. The experimental results show that the average accuracy and sensitivity of our algorithm for GGO segmentation are 90.17% and 84% respectively.更多还原