A PET-CT Lung Tumor Segmentation Method Based on Active Contour Model

To solve the problem that the doctors' clinical experience is not fully integrated into the algorithm design in PET-CT lung tumor segmentation, a hybrid active contour model named RSF_ML based on variational level set is proposed by combining with the PET Gaussian distribution prior, Region Scalable Fitting (RSF) model and Maximum Likelihood ratio Classification (MLC) criterion. Furthermore, referring to the important value of fusion image in the process of lung tumor manual delineation, a segmentation method for PET-CT lung tumor fusion image based on RSF_ML is proposed. Experiments show that the proposed method can achieve accurate segmentation of representative Non-Small Cell Lung Cancer (NSCLC), and the subjective and objective results are better than the comparison method, which can provide effective computer-aided segmentation results for clinic.