Label-guided snake for bone segmentation and its application on Medical Rapid Prototyping

Reconstruction of a 3D model from biomedical images is an intensively and extensively demanded technology for a broad variety of medical health care purposes. One of the major issues in these applications is the correctness of tissue segmentation. This paper presents a method for automatic bone segmentation and reconstruction from knee joint CT images. After the potential bone regions are detected by an initial threshold and binary images are labeled, the proposed Label-Guided Snake (L-Snake) based bone segmentation is performed to reconstruct a precise, smooth surface of bone. The experiments show that the proposed method produces robust results in cases where bones are cancellous or very close to each other. The final 3D model is rendered and validated in software and demonstrates its suitability for future Medical Rapid Prototyping and related applications.