The Reconstruction Of Three-dimensional Model And Finite Element Analysis Of Human Spine

In recent years,with the increasing pressure on people's work and life,the spine disease of our country has a growing trend.According to incomplete statistics,cases of spinal diseases in the mainland are still increasing at a rate of 3 million people per year.In the past decade,with the development of science and technology,spine surgery and diagnostic techniques have achieved considerable development.However,due to the complexity of the skeletal structure of the spine,the difficulty of operation and the risk factor are still high,so it is urgent to find a solution to improve the precision and safety of the surgery.The purpose of this paper is to fully analyze the biological properties of the spine and help doctors better understand the pathological conditions of the spine,so that doctors can make more accurate,specific and efficient feedback on spinal diseases,and then solve the problems caused by spinal diseases to people.At the same time,the increase or decrease of the 3D printing model material can be obtained through the finite element analysis of the model.At present,there are certain achievements in the modeling of the spine and finite element analysis.But from spine segmentation to final modeling and finite element analysis,in general,each step is performed using software.And when researchers are working on CT images,It is also necessary to rely on accumulated experience for some manual work,which easily leads to lack of objectivity and accuracy in the conclusions obtained.In view of the above,this paper intends to work in the following parts: medical image segmentation,three-dimensional model reconstruction,finite element model construction,and finite element analysis,ensure the integrity of the entire process.Using deep learning and three-dimensional reconstruction algorithms can quickly and effectively segment the spine and reconstruct high-quality results.Through training,it can automatically segment and reduce impurity interference,and the manual operation is greatly reduced during the reconstruction process.Finally,the displacement and stress state of different loads under the same force direction are used as one of the criteria for evaluating the validity of the model.Medical image segmentation: According to the characteristics of the CT image of spine medicine,in this paper,firstly,the deep learning method is used to segment the object of study from the spine CT image sequence,that is,the region of interest is extracted first.Next,further skeletal segmentation is performed on the extracted area.The advantage of doing so is to reduce the interference of other impurities,to get a complete and high-quality segmentation results,and to be able to fully understand the characteristics of the human spine image,rapid learning,and strong versatility.Three-dimensional model reconstruction: The modeling results obtained from the previous stage are modeled using two methods: volume rendering and surface rendering.Compare the quality of the models under the two methods and model modeling time and subsequent modeling needs to consider internal details.It was determined that the method of using the improved MC surface rendering method was used to model it.The traditional MC algorithm needs to scan every voxel in the volume data.This wastes scan time on the empty cube.To solve this problem,this paper first selects the seed cube and then scans based on the regional growth mechanism,saves scan time.Next,the model is smoothed to get a complete spin modeling result.Finite element model construction: After the 3D model is built,the physical model needs to be established before the corresponding finite element analysis can be calculated.Solid model modeling follows the NURBS method to create a surface model,then it is transformed into a solid model by surface fitting.This is due to the complexity of the spine structure,and NURBS can well control the direction and curvature of the curve,and then get a more realistic model.Finite element analysis: The construction of a total finite element model of the spine has a large workload and is limited by the computational capabilities of the experimental platform.Therefore,the L3-L4 section model is selected for functional verification and performance verification.The displacements and stress states of different loads under the same stress direction were studied,and it was concluded that the displacement stress and load magnitude were linear within a range that the spine could bear.In order to verify the accuracy and effectiveness of the experimental results in this paper,the results were compared with those of the same FEA model set in previous literature.This method can be further extended to the total vertebral joint if the computational capabilities permit.