Research On3D Reconstruction Technologies Of Liver

As one of the non-invasive diagnostic methods, medical imaging carried on humanbody or parts of human body has gotten more and more attention from doctors andresearchers with the development of the technologies of3D reconstruction as well as3Dinteraction. Liver can be said to be one of the important organs for people’s health dueto the reason that it has the functions of detoxification, excretion, immunity, endocrinemetabolic and so forth. The diseases caused by liver lesions are often chronic diseases,or even fatal diseases. Digital liver model by3D reconstruction of CT, MRI etc., playsan important role in the diagnosis of hepatopathy and decision-making on treatments.The3D reconstruction of the liver can be implemented by either surface renderingor volume rendering. Although surface rendering is superior to volume rendering in thespeed of rendering and interaction, it has to generate intermediate geometric primitives(contour lines or facets), in other words, ROI (region of interest) must be segment fromthe background of the image. On the contrary, volume rendering is known as a directrendering method for no intermediate geometry primitives are generated during itsvisualization process. Besides, it can provide more details about anatomical informationof the liver. Reconstructing the liver segmented by the aforementioned methods merelydisplays the model, in order to achieve an interactive model it also needs to combinewith interactive technologies (i.e.,3D pickup,3D measurements, finite element analysis,etc.). The main tasks of the present paper are listed as follows:(1) Abdominal CT images are segmented by snake model, level set, graph theory,Markov random field approaches. Since the complexity of abdominal CT images whichis different organs not only have the same density but also adjacent to each other inspace, satisfying segmentation results cannot be obtained by a single segmentationmethod. Fortunately, by using prior knowledge as well as combining mathematicalmorphology with threshold segmentation and region labeling, a satisfying segmentationresult has been obtained in this paper.(2) The segmented result is reconstructed by Marching Cubes and Ray Casting.The result shows that surface rending is better than volume rending in the speed of3Dreconstruction and interaction. But volume rendering has a better effect on displayingand providing more space information on the digital liver model, that is, surfacerendering just shows the surface structure of the digital liver model, but volume rendering can also show the internal anatomical structure of the digital liver model.(3) By3D pickup,3D measurements and FEA (finite element analysis), themechanical model of the liver has been reconstructed. In the process of model creationfor finite element analysis, triangular facets need to be simplified due to the reason thatthe number of facets generated by Marching Cubes is too large. And for importing theliver model into ANSYS, an interface design will also be involved.