3D Geometric And Feature Mapping Model For Bladder

Bladder cancer is the fourth prone cancer for the man all over the world with the very high death rate and very frequent recurrence rate. The early diagnosis and treatment of bladder abnormalities is crucial to prevention and cure. As a non-invasive and painless method, Virtual Endoscopy is a promising tumor-detection approach which is adopted widely in medical examination. Computer aided tumor detection for the bladder based on virtual endoscopy consists of the3D organ reconstruction, the description and modelling of tumor shape and texture., which gives the doctors more intuitive reference of examination and diagnosis.This thesis focuses on the methodology for the modeling of the bladder shape and texture based on surface parameterization. The main contents of the thesis include:(1) A semi-automatic segmentation method which called inter-slice spread Snake algorithm is presented for effectively bladder wall segmentation from the volumetric MRI data.(2) A new surface reconstruction algorithm for points cloud is proposed. Iterative free-form deformation technique is employed to fit the initialized ellipsoidal model for point cloud configuration. And then the deformable parametric model is generated for the bladder. Due to the global features of the parametric model, it is with the ease to achieve quantitative comparison beyond the visualization.(3) Feature Projection technique is proposed to establish the map between the geometric and texture features and the surface of parametric model. The resultant2D feature model facilitates the computer aided tumor detection.(4) The unified model for bladder is presented which normalize any bladder into the corresponding sphere model for uniform representation and quantitative comparison.Both the theoretical derivation and experimental results have shown that the proposed model has the ability to present global shape information as well as the local details of bladder surface which provide doctors a measurable and advanced visualization method and lay a better foundation for the computer aided detection later.