Local Variance-driven Level Set Model with Application to Segment Medical Images

The development of computer vision and medical treatment has driven the development of the medical image. At the same time, as one of the key technologies of medical image postprocessing, the accuracy and robustness of medical image segmentation play a key role in assisting doctors in diagnosis and treatment. In this study, we propose the local variance-driven level set model, which combines the local variance difference and the boundary information dynamically for segmenting medical images. The local variance difference is introduced into the area term to control the role of the area term in the overall energy functional and optimize the ability to resist noise. The edge detection function of is added to detect the edge of organs or lesions more accurately. The application to medical images shows that our local variance-driven level set model is more effective for medical image segmentation. Compared with other level set methods, the local variance-driven level set model is more accurate and more robust to noise.