Problems of segmentation of pulmonary nodule candidates from CT images were studied in paper

the image segmentation method proposed by Chan Vese was analysed and improved. A new level set method based on circle dependent had been deduced

and then an algorithm for segmentation of pulmonary nodule candidates in CT images had been presented

it solved the problem of detecting multi circle areas with different sizes in an image. Experiments had been done for segmentation of synthetic and clinical pulmonary images by the proposed algorithm

the experiment results show that it can detect multi circles areas correctly and efficiently

and it is robust to resist noise disturbance. This algorithm has advantages for the realization of automatic detection for lung nodules in pulmonary CT images.