Based On Differential Evolution Algorithm GVF Snake Model For PET Medical Image Processing

PET is currently the world's highest level of nuclear medicine technology, which has been widely applied to various learning of medical research. As the PET's own characteristics and limitations of reconstruction technique, and its image quality is relatively low, far below the spatial resolution of CT and MRI, image contrast is low, poor positioning, background artifact is serious, there is a clear radial noise, which seriously affect the image information acquisition, is not conducive to a doctor for observation and diagnosis, but also seriously affect the image registration, fusion and other follow-up.Based on the traditional Snake model for PET medical image segmentation method is too sensitive to the presence of the initial outline, and it is difficult to converge to the target concave region and other issues. To this end, this paper adopts the improved algorithm GVF Snake model for PET image segmentation processing, in order to prevent their fall into the local optimum and the wrong convergence phenomenon, this article will use the differential evolution algorithm for global optimization of the GVF Snake model to optimize the segmentation results to improve segmentation accuracy. The experimental results show that the proposed method can accurately segment PET image lesion area, to avoid falling into local optimum and has a good real-time.