The Research And Implementation For Interactive Medical Image 3D Reconstruction System Based On VTK

Medical image (CT, MRI) visualization and interactive techniques have become the research hotspots in medical image processing, and the reconstructed 3D images can display the shapes of the organs, tissues or lesions directly and stereoscopically, therefore through human-computer interaction technique, we can achieve ancillary therapy, surgical planning, prosthetic design and so on. However, there are still some difficult issues to be studied further, including several key techniques: interaction designing techniques, new accelerating reconstruction techniques, new improvement techniques, all of which are within the main topics of this paper.In order to solve the blurriness problems rooted in medical images, a new algorithm focusing on medical image multi-resolution enhancement is presented in this paper. The main idea of the algorithm is to denoise and enhance the medical images by adding windows to the selective masking filtering algorithm, and then adopting wavelet transform algorithm, spreading the spectrum of regions of interest (ROI) and enhancing the layering.With regard to the image segmentation issues, the traditional snake algorithm is sensitive to the initial outline position. In this paper, the improved Snake active contour algorithm combine the region growing algorithm with Snake active contour algorithm is presented by adopting the strategy that takes the contour of region growing segmentation result as the Snake initial contour. The algorithm makes full use of the region growing algorithm as well as the snake active contour algorithm. The experiment result demonstrates that the accuracy of medical image segmentation is greatly improved with no obvious computational speed reduction.The MC (Marching Cubes) algorithm is traditionally utilized for reconstruction of medical images, which will inherently cause the ambiguous results. This paper proposes an improved MC algorithm to both eliminate the ambiguity problems and handle the computational bottleneck within it. Firstly, by linear interpolation, the intersections of the ambiguity surface and the iso-surface are positioned. Then, the equivalent points on the faced edge of ambiguity surface are connected, to form two intersecting lines, which will uniquely determine the connection contours based on the state value of joint point for the two lines. Referring to performance bottleneck of Ray-Casting algorithm, we propose the algorithm which will early terminate the calculation on the basis of segmented series medical images, at the step of image synthesis, setting an appropriate threshold of opacity that calculation is valid to reduce the computation, therefore the reconstructing speed is greatly enhanced.As for the three dimensional interaction and system designing techniques, advanced interaction methods including medical orthogonal three sections displaying and virtual endoscopy and its roaming are discussed in detail and implemented by VTK, which is the development tool of the system,with its data models and interaction mechanisms.Finally, we developed a medical image 3D reconstruction system with interactive function based on the above research. The results show that this system has a good interaction performance, real-time effect and good 3D reconstruction results.