3D Tube-like Extraction Of CT Images

These days,3D structures’ extraction and visualization is a hot topic in the digitalsurgical field. Knowing the spatial location of blood vessels can make a difference whendoctors do the surgery plan before the surgery. So the vessel extraction is important inComputer-aided diagnosis and it also meaningful when treat the organic diseases.In this paper, we survey3D vessel extraction methods in medical images andanalyze digital anatomical structure of liver. A new vessel extraction method ispresented. It can extract the multi-scale vessels and show the spatial location objectivelyand directly.Firstly, we introduce the multi-slice spiral CT angiography (MSCTA), DICOMformat, noise reduction and preprocessing of vessel segmentation. We present relatedresearch technologies which are widely used in medical images.Secondly, Hessian matrix is proposed to enhance the multi-scale vessels of the liver.Traditional threshold is applied to obtain the initial vessel data before the enhancement.Vessel is enhanced by calculating the eigenvalues and eigenvectors of Hessian matrix.Vessel enhanced should plus with the initial vessel in order to avoid the fragmentation.And the experiment shows the enhancement is effectively.Thirdly, GMM-EM algorithm is proposed to further segment after the enhancement.The theory of GMM and EM is introduced and its effective has proved in ourexperiment. At last, a vas tree tracing method based on spatial connection is presentedto remove noise. The system now is used for the liver and heart system. At last,3Dvisualization technology, including the volume rendering and surface rendering arepresented. The system is implemented to visualize the digital liver model to display theinitial CT slices, inner anatomy structure and also show the spatial location relationshipbetween the extracted vessels and tumor.Experiments are carried out to evaluate the performance of the methods above.Therefore, the algorithm and system can help surgeons optimize operation plans. Itsupports the feasibility of these methods.