3D Reconstruction Of Head MRI Based On One-Class Support Vector Machine With Immune Algorithm

3D reconstruction of medical images is a hot research area. It is an important application of computer graphics and image processing in biomedicine engineering. 3D reconstruction of medical images is a multidisciplinary subject. It relates to the subjects of digital image processing, graphics and some related knowledge of medical. It is widely used in diagnostic, surgery simulating, plastic and artificial limb surgery, radiotherapy planning, and teaching in anatomy.There are two major traditional methods of 3D reconstruction. One category is set by splicing surface using modules to describe the 3D structure of the object, which is called surface rendering, also known as indirect rendering method; the other is directly the volume rendering method, which known as direct mapping method.The highly non-linear 2D boundary of encephalic tissue must results in highly irregular 3D curve surface which mainly represents in the remarkable and frequent change of curvature and the incontinuity of curve surface. It is so hard to reconstruct the surfaces with the traditional methods.Support Vector Machine (SVM) has a great advantage in dealing with the highly non-linear problems. According to the careful analysis about the problem and the further study of the SVM, One-Class Support Vector Machine was firstly explored into the application of 3D reconstruction to deal with some special configuration problem. Theoretically, One-Class Support Vector Machine, which tries to find the smallest hyper-sphere enclosing target data in high dimensional space by kernel function, is an excellent way to solve this problem. In order to reduce the blindness of parameter selection and perfect the One-Class Support Vector Machine theory, Immune Algorithm (IA) and K-fold cross validation were introduced to intelligently search optimal parameters to reconstruct each tissue. Because there is not any definite evaluation criterion in the reconstruction effect, in this paper, modeling accuracy is firstly explored into the application of 3D reconstruction effects, i.e., the number of the support vectors which locate on the boundary of the real tissue divide the number of the total support vectors of each tissue. Based on these research methods, we finished the seven tissues successfully and got satisfactory reconstruction accuracy. Finally, professional 3D plotting software was used for the process of visualization of the seven tissues with good vision effect.