Thyroid Tumor Ultrasound Image Segmentation based on Improved Graph Cut

摘要

Ultrasound image segmentation has strong pertinence, all kinds of algorithms are usually based on a particular area, specific imaging mode, and interested particular object. These problems make the ultrasonic image segmentation has no unified standard and common rules. Based on characteristics of C-V model and graph cut model, C-V model is discretized and combined with graph cut model to form new energy function. Finally experiments show that the improved method overcome the defect of re-initialization of the level set method in the segmentation of ultrasonic image, reduces the number of nodes, reduces the amount of calculation, and has high robustness and accurate segmentation result.