Simultaneous segmentation of bone regions using multiple-level threshold

Conventional approaches to segmenting skeletal structures include extracting 2D contour profiles on the 2D image domain and performing 3D region growing to acquire the 3D bone region directly. However, as far as multiple pieces of bones segmentation are concerned, these methods are not accurate or efficient enough as each bone must be processed one by one. The purpose of this study is to present a method for multiple bones segmentation. A seed regions generation algorithm is proposed to automatically generate an initial region on each bone of interest. The seed regions are then expanded iteratively. In this process, simultaneous region growing is employed to control the moving speed of the fronts, which can reduce the occurrence of overflow. An investigation of the initial threshold and target threshold is also conducted and appropriate methods to optimize the values are provided. Moreover, five cases of lower limbs are employed to demonstrate the feasibility of the proposed method.