Brain Segmentation Of MRI Images And Its3D Visualization

Brain surgery has high risk and difficult features. Determine the entrance and path of thecraniotomy is first step in the surgical operation, which related to the success of surgerydirectly. The assisted surgical navigation devices are widely used by doctors to sketch thecraniotomy outline currently, planning surgical path, whereby guiding the surgery. Thesurgical navigation systems requires to visualize the brain structures from three-dimensionalperspective, including scalp, skull, brain parenchyma and tumor, so that doctors can visuallyobserve the relationship between surgery path and critical structures as while as functionalzones. Based on MRI image with high spatial resolution and good to soft tissue, it is widelyused in the diagnosis of the brain. This paper built a3D visualization platform for brainimages, designed the segmentation algorithms for scalp, skull, brain parenchyma and tumor,and added contour manual correction tool, the main work is as follows:The segmentation methods of scalp, skull and brain parenchyma, which based onthreshold and mathematical morphology are studied, as while as tumor segmentation byregion growing. The segmentation difficulties of MRI images are overcome, such as the skullimaging is not obvious; scalp tissue is coverage and thin thickness; brain parenchyma hascomplicated sulci; the diversity of tumor in morphology species and location and so on. Thesegmentation accuracy of scalp can reach sub-pixel level, while most of the brain parenchymasegmentation error is under0.2mm.The three-dimensional visualization platform is build, the entire data and brain structuresare reconstructed through synthetic volume rendering method and voxel level surfacerendering technique. Volume data visualization gives doctors a more complete and realdiagnosis environment. The brain structures are displayed in a window simultaneously bymeans of various different colors and transparency properties, so that the doctors can havebasises on quantitative thinking and comparing when determining the relationships betweensurgical path and critical structures as while as functional zones.The brain structure segmentation and3D visualization software is developed. Using theadvantages of QT in terms of interface design to design and landscape the user interface of thesoftware. With the help of ITK and VTK toolkit, the algorithms of brain structure segmentation and3D visualization are achieved, which are associate with the interfacecontrols making convenient for doctors to operation and use in clinical.The artificial correction tool of tissue contours is designed. By human intervention,through the means of added areas and minus areas change the over-segmentation andless-segmentation phenomena in the segmentation algorithms, improving the accuracy andapplicability of the segmentation results.