An Improved Target Extraction Algorithm Based on Region Growing for Lung CT Image

An efficient target extraction algorithm for lung CT image is proposed in this paper. First, it selects the gray average of ROI as the gray of seed points, and hits the seed points by erosion algorithm. At the same time it adopts the average value and variance of the region as the similarity criteria to improve the accuracy of the algorithm. Finally, the targets are extracted gradually by multilevel gray processing method avoiding the disturbed of the other parts in lung. The experiment results demonstrate that our algorithm is efficient in lung nodule extraction of ROI, and can offer more information for pathological analysis and diagnosis.