Using the visible human data set for segmentation and tumor removal surgery planning

摘要

The MR, CT and anatomical cross-section image data sets, scquired through the Visible Human Project, represent a common reference in various research fields such as the study of human anatomy, treatment planning, modeling in numericla dozimetry and for virtual reality applications in medicine. The usefulness of the sets is much improved by segmenting the raw iamge data and linking it to text-based data, thus creating a 3D anatomical atlas. The segmented sets can be used as a deformable atlas for automatic segmentation of patient specific CT or MR images. This is acheived by deforming the atlas model into the patient model, and linking the text-based data from atlas to achieved by deforming the atlas model into the patient mdoel, and linking the text-based data from atlas to aptient images. Such a deformation is based on mathematical modelling of linear elastic and viscous fluid materials and is very effiicent, but computationally expensive. To simpify it, the volume deformation can be exchanged for 3D surface model matching.An effective algorithm, resulting in 3D model of selected organs/tissues can significantly improve CT or MR scan based diagnostics. We demonstrate how to create a segmented, patient specific 3D model of the brain with a tumor, which can be interactively used by the surgeon for more efficient tumor removal surgery planning.