Characteristics of a good academic paper
Taking medical academic papers as an example, a good medical paper has the following characteristics: the birth of a good medical SCI paper requires not only a good topic selection and a good design, but also concrete implementation and serious summary. The author must grasp every link, be serious, rigorous and rigorous. Some people want to write a paper temporarily, but usually there is no title, no design, no material, let alone accumulation. How can they write papers temporarily? Therefore, medical SCI paper writing must pay attention to accumulation!
According to the source of medical papers:
It can be divided into original works (including papers, works and short reports) and compiled works (including textbooks, reference books, monographs, documents, reviews, lectures, special written talks, special discussions, etc. );
According to the purpose of writing a paper, it is divided into two categories: academic papers and degree papers;
According to the nature of medical disciplines and disciplines, it is divided into four categories: basic medicine, clinical medicine, preventive medicine and rehabilitation medicine;
According to the research content of the paper, it is divided into five categories: experimental research papers, investigation research papers, experimental research papers, data analysis papers and experience papers;
According to the genre of the paper, it is divided into: papers, documents, reviews, lectures, techniques and methods, case reports and medical popular science papers. Therefore, the author must choose the expression form of the corresponding genre according to his own research work and the content of research materials.
Every experiment or clinical observation should have strict plans and steps. Strictly operate and related procedures in the application, and do not arbitrarily change your own scientific research design and demonstration. Experts often see that many authors often use, possibly, roughly, estimate, or say that the data has obvious curative effect without statistics when writing articles. These words are not rigorous.
Medical academic papers
Application of Algorithm in 3D Reconstruction of Medical Images
abstract:
The three-dimensional reconstruction technology of medical images can be traced back to the early 1970s. Due to objective reasons such as the high price of medical imaging equipment integrated with 3D reconstruction platform, 3D visualization diagnosis of medical images started late in China, and it was not until 1990s that some universities began to conduct research at all levels [1]. With the development of computer technology, in just a few years, three-dimensional reconstruction technology has become an important means for people to explore the mysteries of life, disease diagnosis and surgical planning.
1 commonly used medical three-dimensional reconstruction materials
Computed tomography (CT) is a new diagnostic technique that combines computer and X-ray. Its main feature is high density resolution, which is 10 ~ 20 times higher than that of ordinary X-ray photos [2]. CT can accurately measure the tiny differences of radiation attenuation characteristics between different tissues on a certain plane, and display them in the form of digital images, which can distinguish the different densities of various soft tissues very finely, thus forming a contrast. For example, head X-ray can't distinguish brain tissue from cerebrospinal fluid, but CT can not only show ventricular system, but also distinguish gray matter and white matter of brain parenchyma. If contrast agent is introduced into CT to enhance the contrast, its resolution will be improved, thus broadening the diagnosis range of diseases and improving the accuracy of diagnosis.
Magnetic resonance imaging. Magnetic vibration imaging is a tomography technology, which uses magnetic vibration phenomenon to obtain electromagnetic signals of human body and reconstruct human body information. 1946, Flex Bloch of Stanford University and Edward purcell of Harvard University independently discovered the phenomenon of nuclear magnetic resonance. 1972, paul lauterbur developed a set of spatial coding methods for nuclear magnetic resonance signals, which can reconstruct human images. Magnetic resonance imaging technology has some similarities with other tomography technologies. For example, they can show the distribution of some physical quantities (such as density) in space. At the same time, magnetic resonance imaging also has its own characteristics, which can obtain cross-sectional images, three-dimensional volume images and even four-dimensional images of spatial spectrum distribution in any direction.
At present, the three-dimensional reconstruction methods of medical images mainly include surface rendering, volume rendering and reconstruction of three-dimensional geometric shapes from two-dimensional gray images on the surface of objects, or the method of restoring shapes through shading.
2 the basic principle of moving cube algorithm
Marching Cubes[3] algorithm is an isosurface construction method proposed by Lorensen et al. in 1987, which has been used up to now and is the representative of isosurface extraction technology in voxel unit [4]. The so-called isosurface refers to the set of all points in the grid space whose sampling value is equal to a given value. The essence of this algorithm is to treat a series of 2D slice data as a 3D data field, from which there will be
Extract substances with a certain threshold and connect them into triangular patches in some topological form.
Isoplane is the set of all voxel points with the same value in space, and the values of voxel points are obtained by trilinear interpolation of eight points V0~V7 in the voxel region. It can be expressed as: c is a constant. F(f) is the isosurface in volume data f, and the calculation formula can be expressed as:
⑴
Among them? 0,? 1,,? 7 is a constant determined by the values of eight fixed points from v0 to V7.
In the MC algorithm, it is assumed that the original data is a discrete three-dimensional spatial regular data field as shown in figure 1. Images produced by tomography (CT) and magnetic resonance imaging (MRI) for medical diagnosis belong to this type.
The basic idea of MC algorithm is to process voxels in data field one by one. As shown in fig. 2, the voxels intersecting the isosurface are classified, and the intersection points (V0~V7) between the isosurface and the edge of the voxel are calculated by interpolation. According to the relative position of each vertex in the voxel and the isosurface, the intersection of the isosurface and the edge of the cube is connected in a certain way to generate the isosurface as an approximate representation of the isosurface in the cube. After calculating the relevant parameters of the isosurface in the volume data field, the isosurface is drawn by using the commonly used graphics software package or the surface drawing function provided by hardware [5].
Generally, the method of binarization is used to draw the isosurface, that is, the point value (0 or 1) and the vertex density value are determined by comparing with the given threshold.
⑴ Read the three-dimensional discrete rule data field into the memory in layers.
⑵ Scan two layers of data, construct voxels one by one, and eight angles in each voxel are taken from two adjacent layers; Eight fixed points can be defined as (i, j, k), (i+ 1, j, k), (i+ 1, j+ 1, k), (i+ 1, j, k+/kloc.
⑶ Compare the function value of each corner of a voxel with the given equivalent face value c, and construct the state table of the voxel according to the comparison result.
(4) According to the state table, the boundary voxels intersecting with the isosurface are obtained.
5. Calculate the intersection of voxel edge and isosurface by linear interpolation method.
(6) Using the central difference method, the normal vector of each corner of the voxel is obtained, and then the normal vector of each vertex of the triangular patch is obtained by the linear interpolation method.
(7) Drawing an isosurface image according to the coordinates and normal vectors of each vertex on each triangular patch.
3. The judgment of the spatial equivalence point and the calculation of the intersection point between the equivalence surface and the voxel boundary.
Take any edge of a discrete grid and let two nodes on the edge be Mi(xi, yi, zi, qi) and Mj (xj, yj, zj, QJ) respectively; When (q-c)(q-c) is satisfied, the equivalent of the order of magnitude is c? 0 (expression of equivalence point judgment condition), then the equivalence point Mo is between Mi and Mj. Let the coordinates of the equivalent point Mo be (xo, yo, zo), and the equation (2) can be obtained by linear interpolation from the two points Mi and Mj:
⑵
Where k=(qi-c)(qj-c)? 0。 According to the judgment condition of isosurface (1) and the coordinate formula of equivalent points (2), we can search and judge the edges of the grid according to the discrete information of the structure, so as to find all the equivalent points of the structure in the specified area. After finding the equivalent points, you can connect them into triangles or polygons to form a part of the equivalent surface.
4 Calculation of Normal Vector of Isosurface
In order to display the isosurface image with graphics hardware, it is necessary to give the normal direction of the isosurface of the triangular patch, select the appropriate lighting model to render and generate realistic graphics. For each point on the isosurface, the gradient component along the tangent direction of the surface should be zero, so the gradient vector direction along this point also represents the normal direction of the isosurface at this point. Isoplane is often the interface of substances with different densities, so its gradient vector value is not zero, that is, formula (3):
⑶
It is time-consuming to directly calculate the normal of triangular patches. In order to eliminate the discontinuous change of brightness between triangular patches, we only need to give the normal direction of each vertex of triangular patches and draw each triangular patch with Gouraud model. Here we use the central interpolation method to calculate the gradient of each corner of each voxel. In the case of triangle, the normal vector of each triangle patch is calculated, then the normal vector of each vertex is obtained by using the normal vector of the triangular surface, and finally the normal vector of a point on the triangular surface is obtained by interpolating the three normal vectors of the three vertices of the triangle. There is a simple method to calculate the normal vector of the vertices of the isosurface. Considering that the gradient direction of contour line is perpendicular to the tangent of contour line, gradient vector can be used instead of the vertical line of contour line. In the three-dimensional case, the gradient direction of the isosurface is the normal direction of the isosurface. Thus, Formula (4) can be obtained:
⑷
Optimization of five-step cubic mesh model simplification algorithm
The mesh model simplification algorithm has achieved a series of results. At present, most simplification algorithms take the geometric position change of the model before and after edge folding as the folding cost to reduce the number of polygons and improve the operation efficiency. The purpose of mesh simplification algorithm is to improve efficiency on the premise of ensuring image accuracy as much as possible. Therefore, the principle of selecting coordinate points is to be as close as possible to the original grid. There are generally two kinds of subset selection methods: subset selection method and optimal selection method [6], that is, simply choose the less expensive one from the two endpoints of an edge. The optimal selection method is to select the point V with the smallest quadratic error as the folding point, and the quadratic error corresponding to this point is a quadratic equation. Finding its minimum value is to find the point where the derivative of the equation to x, y and z is zero. This process is equivalent to solving the matrix equation of Formula 5.
⑸
Measurement of folding cost
The calculation of folding cost is divided into two steps. Step 1: when calculating the quadratic error profile of each vertex, based on Garland's standard quadratic error measure, and considering the influence of the surrounding triangle area, calculate the average value of the quadratic error measure of each vertex; Step 2: When calculating the cost of edge folding, the length of edge and the degree of triangle shape change caused by edge folding are used as weighting factors.
The specific calculation method is as follows: in three-dimensional space, plane P can be expressed as ax+by+cz+d=0, or PTv=0, where P=[a, b, c]T is the unit normal vector of plane P and D is a constant. The square of the distance from any point v=[x, y, z, 1]T in the model space to the plane is the formula [6]:
⑹
Quadratic error of any point v=[x, y, z, 1]T in the grid model? (v) is defined as the sum of squares from the vertex to the plane related to the fixed point, which can be expressed as the formula once:
⑺
Among them, plane (V) represents the set of all triangular planes containing fixed point V, which is called the set of related planes of vertex V. In the initial state, the quadratic error of each point in the mesh model is 0, and the formula can be obtained after the above formula is deformed.
⑻
Where kp is the quadratic error metric of plane p.
⑼
A quadratic matrix called v=[x, y, z,1] t.
It is called the quadratic error of point V. When the edge is folded, an additional rule (Garland et al., 1987) can be used to obtain the quadratic error measure of point V, and the quadratic error value of the vertex is, that is, the folding cost of the edge.
Application of six-grid simplification algorithm in medical three-dimensional reconstruction
Grid algorithm is generally used to accelerate 3D reconstruction, but simple grid algorithm lacks practical value. Compared with its high-speed rendering, the loss of accuracy is unacceptable. So the mesh simplification algorithm is further optimized? Mesh simplification algorithm based on volume rendering.
Volume rendering is to display all substances (skin, bones, muscles, etc. ) A piece in a picture. But in the case of observing only bones, many triangular surface diagrams are meaningless. Ignoring those unnecessary triangles can effectively improve the reconstruction speed while ensuring the accuracy.
7 concluding remarks
MC algorithm determines the polygon of voxels by comparing thresholds, which often has the inevitable disadvantage of slow speed in the face of large-capacity data, but now various targeted improvements have made it have greater development potential, so MC algorithm is not only a simple algorithm, but also closer to? Voxel? This concept. Many popular 3D reconstruction algorithms are improved on the basis of MC to obtain the required specific 3D model. Medical image fusion algorithm based on wavelet transform, tomographic medical image interpolation algorithm, etc. It is mainly used to make CT and other data vulnerable to threshold segmentation in MC algorithm. Now, the use of OpenGL, VTK and other image function libraries makes 3D image modeling simple, and it is expected that the application of 3D reconstruction technology in medicine will have greater development.
References:
Puchao in Zhang Yumin. Three-dimensional medical image processing algorithm and its application [J]. Ordnance Automation, June 2004: 2 10 ~ 2 12.
Luo, Zhou and Shi Jiaoying. Simplified polyhedron model based on triangle removal criterion [J]. chinese journal of computers, China, February 2008:135 ~138.
[3] Nielsen General. Double marching cube. IEEE Visualization 2004.
Tian Jie, Bao Shanglian, Zhou, Medical image processing and analysis [M]. Electronic Industry Press, 2003.
Kim, Liu. Research on 3D Reconstruction of Medical Images [J]. China Medical Devices Journal, 2008.2:34.
Did you get a look at him? What are the characteristics of a good academic paper? People still see:
Characteristics of successful academic papers
2. Characteristics of Successful Academic Papers (2)
3. Understanding of the importance of academic papers
4. Experience of publishing academic papers
5. What is an academic paper? How to write academic papers?