3D Reconstruction And Biomechanical Analysis Of Spine Based On Improved MC Algorithm

With the rapid development of society,spinal diseases have become a disease with high incidence.In medicine,they are generally diagnosis and treatment by images which acquired from CT or MRI,but the diagnosis on the basis of two-dimensional images must have high professional knowledge and rich clinical experience.If the scan is displayed in a three-dimensional model,it is conducive to the doctor to make more accurate and intuitive diagnosis of the lesion location.At the same time,the3-dimensional model can relieve the pain of patients by simulating the preoperative and postoperative conditions,and can help doctors to communicate with others which can make progress together.This dissertation mainly focuses on the 3D reconstruction of spinal CT images,which is of great significance to the application of spinal preoperative diagnosis,intraoperative planning,postoperative simulation and other applications.The main research contents of this dissertation as follows:To improve the quality of reconstruction,a fusion segmentation algorithm is proposed to segment the target region.First,threshold segmentation is performed on the image with the threshold which obtained from the OTSU and histogram,which can increase the contrast between the target and non-target area.Then,through the reconstructing of the opening and closing operation to mark the target and fill the holes in target region.Finally,the watershed algorithm is used for effective segmentation.A feasibility study scheme is proposed for the problems of point cloud reconstruction and image reconstruction of marching cube.For point cloud reconstruction,a way to improve the normal vector ambiguity is proposed to improve the original marching cube algorithm.And replace the linear interpolation with the midpoint interpolation to solve the ambiguity problem of the reconstruction surface.In view of image reconstruction disadvantages,the problem of too many patches which due to the gradient difference is reduced by binaryzation the original images,and smooth along the normal vector direction to reduce the roughness of surface.The biomechanical model of the thoracic and lumbar spine is established by smoothing,grid division,fitting surface and assembly the model after image reconstruction.In order to verify the effectiveness of the biomechanical model,the geometric dimensions of its intervertebral discs are measured.And simulating the stress-strain situation under different working conditions,then analyzed the results.The experiments show that the proposed fusion segmentation algorithm not only effectively suppressed the over-segmentation of watershed segmentation,but also obtain a good segmentation effect.For the reconstruction of two data types,the improve algorithms proposed in this dissertation all reduce the complexity of the reconstruction and improve the reconstruction efficiency.The geometric dimensions of the intervertebral discs were measured to verify that they are similar to the actual dimensions.After assembling the overall model,through biomechanical analysis found that under the same working conditions,its stress-strain was proportional with load which verifying the effectiveness of its model.