ROI Boundary Detection Based on Geometric Active Contour Model in X-ray Skeletal Image

By utilizing the iterative solutions based on the level set method, the geometric active contour model (GACM) can be used to detect image boundaries and then solve the question of topology variation in curve evolution effectively, therefore, we adopt the GACM to investigate the boundary detection of digital X-ray skeletal images. According to the characteristics of medical images, we present an improved Chan-Vese method which combines regional information with image gradient information to increase the boundary detection capability. Theoretically speaking, the improved method can guarantee not only the speed and robustness of image division but also the accuracy of target extraction from the medical image background with abundant boundary layers. In practice, this method is proved to detect the boundaries of the epiphysis /metaphyseal regions of interest (EMROI) in X-ray skeletal images effectively and clearly.