Design And Implementation Of REST Web Service For MRI Ventricle Image Segmentation And 3D Reconstruction

In order to meet the Hospital XX’s application requirements of clinical diagnosis based on MRI brain images, we finished MRI(Magnetic Resonance Imaging) brain image segmentation and 3D reconstruction. We then designed and implemented the REST(Representational State Transfer) Web service for MRI ventricle image segmentation and 3D reconstruction. We also designed and implemented a client application system of ventricle segmentation and 3D reconstruction. The main work and contributions of this thesis are summerized as the following.(1) Design and Implementation of REST Web Service for MRI Ventricle Segmentation.To meet the demands of accurate segmentation of MRI ventricle images, several typical image segmentation algorithms via region-based active contour models are firstly analyzed. Then the LGDF(Local Gaussian Distribution Fitting) method based on active contour model is implemeneted, and the ventricle area is also estimated based on the segmentation results. After that, a dynamic linking library is encapsulated by combining C++ and Matlab programming ways, and the REST Web service for MRI brain image segmentation is designed and implemented based on JNI and Jersey framework.(2) Design and Implementation of REST Web Service for 3D Reconstruction.To meet the application demand for effective rendering and real-time interaction during the ventricle 3D reconstruction, several typical redenring methods including surface-based and volume-based are analyzed, in which surface-based methods such as Marching Cubes, Dividing Cubes are investigated and implemented especially based on VTK lib. Finally the REST Web Service for 3D reconstruction and rendering is designed and implemented.(3) Client Application System Design and Implementation for Ventricle Segmentation and 3D Reconstruction.To meet the Hospital XX’s applicable need about MRI ventricle image segmentation and 3D reconstruction, a B/S structure system based on Struts, Spring and Hibernate framework has been finally implemented and put into use. The main functions of the system include MRI image segmentation, 3D reconstruction of ventricle sections, visualization of the reconstructed results, and human-machine interaction.The application in Hospital XX demonstrates the effectiveness of the system.