CN110021003B - Image processing method, image processing apparatus, and nuclear magnetic resonance imaging device - Google Patents

Abstract

The invention relates to medical imaging technology, in particular to a method for realizing fusion of anisotropic images, diffusion envelope surface images and nerve distribution images. The method includes acquiring corresponding anisotropic images, diffusion envelope images and nerve distribution images according to nuclear magnetic resonance data, and then combining the respective characteristics of the three images to complete image fusion. The invention is suitable for analyzing and reconstructing the complex biological tissue structure in the brain, and has important application value in brain science, neuroscience, medical imaging and the like.

Description

Image processing method, image processing device and nuclear magnetic resonance imaging device

technical field

The invention relates to the cross field of neuroscience and medical image processing, in particular, to a method for processing anisotropic images in nuclear magnetic resonance imaging, a diffusion envelope image and a nerve distribution image, and a corresponding image processing device, as well as a method including the Magnetic resonance imaging equipment of an image processing apparatus. The method and device are suitable for the study of brain structure and function, the diagnosis and treatment of brain diseases, and the preoperative planning of clinical neurosurgery.

Background technique

Magnetic resonance imaging (MRI) is based on the principle of nuclear magnetic resonance. According to the different attenuation of the released energy in different structural environments inside the material, the electromagnetic wave emitted by the external gradient magnetic field can be detected, and the position of the atomic nucleus that constitutes the object can be known. and types, according to which the structure image inside the object can be drawn. Diffusion tensor imaging (DTI), a new method to describe the structure of the brain, is a special form of magnetic resonance imaging (MRI). That is, according to the diffusion movement of water molecules in biological tissues such as the brain, the tissue structure and the distribution of nerve fibers are judged. Diffusion tensor imaging maps can reveal how brain tumors affect nerve cell connections, leading to brain surgery. It can also reveal subtle abnormal changes associated with stroke, multiple sclerosis, schizophrenia, dyslexia, and more. In addition, there are high angular coordinate resolution imaging, enhanced diffusion tensor imaging and similar methods. Among them, enhanced diffusion tensor imaging is a medical imaging method invented by us that is suitable for reconstructing nerve fiber bundles in complex distribution areas of nerves. The diffusion motion of , and the direction of the number of fibers in the voxel are solved by combining the higher-order tensor decomposition theory.

Diffusion refers to the random and irregular motion of molecules, which is an important form of activity of water molecules in the human body, also known as Brownian motion. Diffusion is a three-dimensional process, and the distance of molecules diffusing in a certain direction in space is affected by the structure of biological tissues. The diffusion methods can be divided into two types: one means that in a completely homogeneous medium, the movement of molecules has no obstacles because there are no obstacles. The distance of movement in all directions is equal. This type of dispersion is called isotropic dispersion. For example, the dispersion of water molecules in pure water is isotropic dispersion. In human brain tissue, cerebrospinal fluid and cerebral gray matter The dispersion of water molecules is approximately isotropic. Another kind of dispersion is direction-dependent. In an organization arranged in a certain direction, the diffusion distance of molecules in all directions is unequal, which is called anisotropic dispersion. In order to describe the degree of anisotropy of the dispersion motion of water molecules, parameters such as fractional anisotropy (FA), relative anisotropy (RA), and volume ratio (VR) are often used to quantitatively analyze each Items of the opposite sex. Images based on anisotropic parameters throughout the brain have important applications in the diagnosis of brain diseases. Among them, the structure of white matter fibers in the brain observed by FA images is the clearest, and the demarcation between gray and white matter is good, and the FA value is positively correlated with the integrity of the myelin sheath, the compactness and parallelism of the fibers, so the FA value is the most widely used. In addition, the diffusion motion envelope surface is also an effective means to describe the diffusion motion of water molecules in the voxel, that is, the spatial envelope surface formed by the diffusion motion of water molecules released from the center of the voxel in unit time. Compared with the anisotropic parameters, the diffusion motion envelope surface reflects the influence of the diffusion motion of water molecules by the internal structure of biological tissue more completely and comprehensively, which has important physical and clinical significance. In traditional diffusion tensor imaging, the envelope of diffusion motion is an ellipsoid; in high-angle coordinate resolution imaging, the diffusion motion is a combination of a series of ellipsoids; in enhanced diffusion tensor imaging, the Diffuse motion envelope is one of the key steps to complete the reconstruction of nerve fiber bundles.

Fiber tractography (FT) is a new technology developed on the basis of diffusion tensor imaging, which can be used to non-destructively detect the direction and integrity of nerve fiber tracts in the brain. The basic principle is to start from any measurement voxel (that is, the seed point), travel along the direction of the nerve fiber bundle in the voxel for a specified length to the next voxel, and continue the above operation. Until the boundary of the measurement space is reached; or the anisotropy parameter in the voxel is lower than the threshold; or the angle between the nerve fiber bundles in the connected two voxels is greater than the threshold (usually 60 degrees). By connecting the directions of the nerve fiber bundles in this series of voxels in space, the overall direction of a nerve fiber bundle in space is obtained. Fiber tract tracing imaging technology can realize the three-dimensional reconstruction of the distribution of nerve fibers in the brain, which is of great significance for the study of brain structure and function, the diagnosis of clinical brain diseases, surgical navigation and preoperative planning.

The anisotropy image can give the overall distribution of gray and white matter in the slice plane; in comparison, the diffusion envelope image is more complete, and can comprehensively reflect the anisotropy of the diffusion motion in the slice plane, which can be roughly inferred. The tissue structure and nerve distribution images can help doctors and researchers to observe the spatial distribution and direction of nerve fibers in the brain intuitively, and describe the structure of nerve tissue in the brain and the connection of various functional areas in a detailed and complete manner. The three images respectively reflect the complex biological tissue structure inside the brain from three different aspects: gray and white matter distribution, local anisotropy and nerve fiber distribution. In order to make full use of these three kinds of information, it is necessary to perform certain fusion processing on the above three kinds of images. But the dimensions of the three images are different: the anisotropic image is a two-dimensional image, the neural distribution image is a three-dimensional image, and the diffusion envelope image is between two-dimensional and three-dimensional (the voxels to be measured are distributed on a two-dimensional plane, The diffusion envelope image within each voxel is three-dimensional), which causes certain difficulties for image fusion.

SUMMARY OF THE INVENTION

In order to solve the above problems, the present invention proposes a processing method for effectively completing the fusion of anisotropic images, diffusion envelope surface images and neural distribution images. Specifically, the present invention is implemented in this way.

A kind of image processing method, is used for realizing the fusion processing of anisotropic image, diffusion envelope surface image and nerve fiber distribution image, it is characterized in that comprising the steps:

Step 1. Obtain the NMR data of the object;

Step 2. Process the nuclear magnetic resonance data to obtain the diffusion motion envelope surface in each voxel of the object;

Step 3. Calculate the direction and anisotropy parameters of the nerve fiber bundles in each voxel;

Step 4. According to the calculated anisotropy parameters, generate an anisotropic image in the selected slice plane;

Step 5. In the selected slice plane, select the diffusion motion envelope surface in the voxel whose anisotropy parameter is greater than the threshold, and generate a diffusion envelope surface image;

Step 6. in the slice plane of the described selection, choose the voxel whose anisotropy parameter is greater than the threshold value to be the seed point, from the seed point, reconstruct the nerve fiber distribution image according to the direction of the nerve fiber bundle;

Step 7. Fusion of the anisotropic image, diffusion envelope image and neural distribution image in the selected slice plane.

According to one aspect of the present invention, an effective image processing device is provided for merging the anisotropic image, the diffusion envelope surface image and the nerve fiber distribution image, which is characterized by comprising the following units:

a sampling unit for acquiring nuclear magnetic resonance data of the object;

a calculation unit, used to obtain the diffuse motion envelope surface in each voxel of the object, and calculate the direction and anisotropy parameters of the nerve fiber bundle;

an image generation unit for generating anisotropic images, diffusion envelope images and nerve fiber distribution images respectively;

The fusion unit is used to realize the fusion of anisotropic images, diffusion envelope images and nerve fiber distribution images.

According to another aspect of the present invention, there is provided a magnetic resonance imaging apparatus including an image processing apparatus according to an embodiment of the present invention.

Description of drawings

1 is a flowchart of an image fusion method according to the present invention;

2 is a block diagram of an example configuration of an image fusion processing apparatus according to the present invention;

3 is a block diagram of a configuration example of a magnetic resonance imaging apparatus according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of the anisotropic image described in the implementation step S150 of the present invention;

5 is a partial schematic diagram of fusion of the anisotropic image and the diffusion envelope surface image described in the implementation step S180 of the present invention;

6 is a partial schematic diagram of the three types of image fusion described in the implementation step S180 of the present invention;

7 is a schematic diagram of fusion of anisotropic image, diffusion envelope surface image and neural distribution image finally realized by the present invention;

Detailed ways

In order to make the objects, technical solutions and advantages of the present invention clearer, the present invention will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.

As shown in Figure 1, it is a flow chart of the image fusion method of the present invention, which includes the following steps:

Step S110: Collect nuclear magnetic resonance data of the object. A finite number of sampling directions are selected on the unit sphere, and in each voxel, the attenuation intensity of the signal in the sampling direction is measured by nuclear magnetic resonance technology, that is, nuclear magnetic resonance data.

Step S120: Select a processing method for the nuclear magnetic resonance data, and obtain the diffusion motion envelope surface in each voxel of the object. Such as: diffusion tensor imaging (DTI), high angular coordinate resolution imaging (HARDI), enhanced diffusion tensor imaging (EDTI) and so on. Now take enhanced diffusion tensor imaging (EDTI) as an example:

(a) Calculate the dispersion coefficient D in each direction according to the signal attenuation intensity obtained in step S110,

b is the instrument parameter, S ₀ is the original signal strength, and S is the measured strength.

(b) Calculate the dispersion displacement x per unit time (1s) in each direction according to the dispersion coefficient D,

之间的 函数关系

即给定方向

可由

确定该方向上的径长r。而函数

可以展开成如下的形式：(c) Reconstructs the diffuse motion envelope within this voxel from the diffuse displacements in (b) in a finite number of directions. For a three-dimensional envelope surface, take its inner point to establish a coordinate system, then the graph can be expressed as a path length r and a direction unit vector

functional relationship between

i.e. a given direction

by

Determine the path length r in this direction. while the function

It can be expanded into the following form:

高阶导数替换为高阶张量)。根据高阶张量分解理论中的不可约分 解，任意高阶张量都可以分解为一系列不可约张量的组合，

最终可以表示为一系列不 可约张量和单位向量

的缩并之和。Among them, D ₁ , D ₂ ... D _r are vectors (first-order tensors), second-order tensors ... r-order tensors, respectively. The form of the above decomposition is similar to the Taylor expansion (x is replaced by

higher-order derivatives are replaced with higher-order tensors). According to the irreducible decomposition in the higher-order tensor decomposition theory, any higher-order tensor can be decomposed into a series of irreducible tensors,

can finally be represented as a series of irreducible tensors and unit vectors

The sum of the contractions.

为空间参数，则根据上述理论可做如下分解：Let the diffusion motion envelope be

is a spatial parameter, then according to the above theory, it can be decomposed as follows:

为三维空间中一组完备正交的展开基底。a_m为展开系数，由包络面

与给定基底

积分可得。具体的，展开基底可取为三维球谐函数，其数学表达式如下：in,

is a set of fully orthogonal unfolded bases in three-dimensional space. a _m is the expansion coefficient, determined by the envelope surface

with a given base

Points are available. Specifically, the expanded basis can be taken as a three-dimensional spherical harmonic function, and its mathematical expression is as follows:

in

为三维球谐函数(P_m,r为勒让德多项式)，a_m,r,b_m,r为展开系数。此外， 展开基底也可取为小波函数，脊波函数等完备正交的函数族。

is a three-dimensional spherical harmonic function (P _m,r is Legendre polynomial), a _m,r , b _m,r are expansion coefficients. In addition, the expansion base can also be taken as a complete orthogonal function family such as wavelet function and ridgelet function.

The specific steps for reconstructing the diffuse motion envelope surface through the diffuse displacement in a finite number of directions are as follows:

(以三维球谐展开为例)与节点方向上的弥散位移(在(b)中得到)已知，通过插值方法及离 散积分计算展开系数a_m,r,b_m,r：在每个网格内，由节点方向上的弥散位移通过线性插值得到网格中心处的弥散位移；用网格中心处的弥散位移，三角函数值和基底函数值代替整个网格 上的

sinrθ,cosrθ与

由函数值乘网格面积近似计算a_m,r,b_m,r表达式中的积分在该网格上的值；遍历所有网格，将其上积分值求和，即可得到展开系数a_m,r,b_m,r。(c1) The grid is divided on the unit sphere, and the nodes of the grid are the measurement directions. basis function

(taking the three-dimensional spherical harmonic expansion as an example) and the dispersion displacement in the nodal direction (obtained in (b)) are known, the expansion coefficients a _m,r , b _m,r are calculated by the interpolation method and discrete integration: in each net In the grid, the dispersion displacement at the center of the grid is obtained by linear interpolation from the dispersion displacement in the direction of the nodes; the dispersion displacement at the center of the grid, the trigonometric function value and the basis function value are used to replace the values of the entire grid.

sinrθ, cosrθ and

Calculate the value of the integral on the grid by _multiplying the function value by the grid area approximately; traverse all _grids and sum the integral values on them to get the expansion coefficient a _m,r ,b _m,r .

和(c1)中算出的a_m,r,b_m,r复原弥散运动包络面

(c2) by the substrate

and a _m,r , b _m,r calculated in (c1) to restore the diffuse motion envelope

For a general basis function, the above formula can be written as:

Among them, n is the expansion order, and it is necessary to comprehensively consider the basis function, accuracy requirements, calculation costs and other factors. .

的取值范围

等分为若干份，得 到单位球面上一系列均匀分布的点。遍历包络面，由

计算各点对应方向上的弥散位 移，并比较相邻点对应方向上弥散位移的大小；若某点对应方向上的弥散位移大于其所有相 邻点(对应方向上的弥散位移)，则该方向为包络面上弥散位移取极大值的方向；还可以通 过直接由

对

求导数，确定极大值方向。极大值方向即为体元内神经纤维束的方 向。Step S130: Calculate the direction of the nerve fiber bundle in the voxel according to the selected processing method of the nuclear magnetic resonance data, such as the enhanced diffusion tensor imaging mentioned above. set the parameter

range of values

Divide into several equal parts to obtain a series of uniformly distributed points on the unit sphere. Traverse the envelope surface, given by

Calculate the dispersion displacement in the corresponding direction of each point, and compare the size of the dispersion displacement in the corresponding direction of the adjacent points; if the dispersion displacement in the corresponding direction of a point is greater than all its adjacent points (the dispersion displacement in the corresponding direction), then the direction Take the direction of the maximum value for the diffusion displacement on the envelope surface; it can also be obtained directly by

right

Take the derivative to determine the direction of the maxima. The direction of the maximum value is the direction of the nerve fiber bundle in the voxel.

Step S140: According to the selected processing method of the nuclear magnetic resonance data, such as the enhanced diffusion tensor imaging mentioned in this embodiment, use the obtained diffusion motion envelope to calculate the anisotropy parameter. Note that the maximum value of the vector radius r on the diffuse motion envelope surface is maxr, the minimum value is minr, and the average value is meanr, then the anisotropy parameter f can be defined as:

f=(maxr-minr)/meanr. Note: The anisotropy parameter here can also be calculated in other ways.

Step S150: Select the slice currently under study, and make an anisotropic (parametric) image in the slice plane. As shown in Figure 4, the anisotropy parameter is calculated according to the restored diffusion motion envelope. The anisotropy parameter is as defined in S140. The value corresponds to 255), and the image with the anisotropy parameter as the gray value is output.

Step S160: In the selected slice plane, extract voxels whose anisotropy parameter is greater than the threshold value A. In this example, the threshold is taken as 0.2, and in general, the threshold can be taken as any real number between 0 and 1. The corresponding diffusion motion envelope surface obtained in step S120 is drawn in the selected voxel, that is, an image of the diffusion envelope surface in the plane is made.

Step S170: Select the voxel whose anisotropy parameter in the slice plane is greater than the threshold value A as the seed point, start from the seed point in sequence, and use the fiber tracing imaging method (note that other reconstruction methods can also be used) to complete the reconstruction of the nerve fiber . Travel along the direction of the nerve fiber bundle in the seed point for the specified length to the next voxel, and continue the above operation. Until the boundary of the measurement space is reached; or the anisotropy coefficient in the voxel is less than the threshold A; or the angle between the nerve fiber bundles in the two connected voxels is greater than the threshold B. In this example, the threshold B is set to be 60 degrees, and for a general case, the value range of the threshold B is 45-90 degrees. By connecting the directions of the nerve fiber bundles in this series of voxels in space, the overall direction of a nerve fiber bundle in space is obtained.

Step S180: As shown in Fig. 5, place the diffusion motion envelope image drawn in step S160 in the anisotropic image described in step S150 to complete the fusion of the anisotropic image and the diffusion envelope image; as shown in Fig. 5, and then Align the seed points of the reconstructed nerve fiber bundles in step S170 with the corresponding voxels in the anisotropic image in step S150 to complete the fusion of the anisotropic image, the diffusion envelope image and the nerve distribution image.

As shown in FIG. 2 , it is a schematic structural diagram of the image processing apparatus of the present invention. The image processing apparatus 200 in the embodiment of the present invention includes a sampling unit 210, a calculation unit 220, a drawing unit 230, and a fusion unit 240. In the above description of the image fusion method, the specific implementation process of some steps has been disclosed, and below, an overview of each unit of the image fusion processing apparatus is given without repeating some of the details that have been discussed. Specifically: the image fusion processing apparatus 200 includes:

Sampling unit 210: used to collect nuclear magnetic resonance data. A finite number of sampling directions are selected on the unit sphere, and in each voxel, the attenuation intensity of the signal in the sampling direction is measured by nuclear magnetic resonance technology.

Calculation unit 220: used to restore the diffusion motion envelope surface in the voxel, and calculate the direction and anisotropy parameters of the nerve fiber bundle. Choose an appropriate NMR data processing method, taking enhanced diffusion tensor imaging (EDTI) as an example:

(1) Calculate the dispersion coefficient D in each sampling direction in the voxel according to the signal attenuation strength;

(2) Calculate the dispersion displacement x per unit time in each sampling direction in the voxel according to the dispersion coefficient D;

以及展开阶数n，结合各采样方向上的弥散位移x复原所述体元内 的弥散运动包络面

设真实的弥散运动包络面为

做如下分 解；

为基底函数，根据选择的展开阶数n，由采样方向 上单位时间内的弥散位移x，通过线性插值，二次插值或多项式插值等方法和离散积分计算 展开系数a_m，结合基底函数

复原所述体元内的弥散运动包络面

(3) Select the basis function

and the expansion order n, combined with the diffusion displacement x in each sampling direction to restore the diffusion motion envelope surface in the voxel

Let the real diffuse motion envelope be

Do the following decomposition;

is the basis function, according to the selected expansion order n, from the dispersion displacement x per unit time in the sampling direction, through linear interpolation, quadratic interpolation or polynomial interpolation and other methods and discrete integration to calculate the expansion coefficient a _m , combined with the basis function

recover the diffuse motion envelope within the voxel

(4) Determine the direction of the nerve fiber bundle in the voxel according to the restored diffusion motion envelope surface: search for the maximum value on the restored diffusion motion envelope surface, and determine the direction of the maximum value; Find the derivative of the surface to determine the direction of the maximum value. The direction of the maximum value is the direction of the nerve fiber bundle in the voxel.

(5) The maximum value of the vector radius r on the envelope surface is maxr, the minimum value is minr, and the average value is meanr, then the anisotropy parameter f can be defined as: f=(maxr-minr)/meanr.

Image generation unit 230: used to draw anisotropic (parametric) images, diffusion envelope images and neural distribution images respectively:

Select the slice of the current study and make an image with the anisotropy parameter as the gray value;

In the selected slice plane, extract the voxels whose anisotropy parameter is greater than the threshold value, draw the corresponding diffusion motion envelope surface at the position of the voxel, and make the diffusion envelope surface image;

The voxels whose anisotropy parameters in the slice plane are greater than the threshold are selected as seed points, and the reconstruction of nerve fiber bundles is completed by the fiber tracing imaging method starting from the seed points. Follow the direction of the nerve fiber bundle in the seed point to travel the specified length to the next voxel, and continue the above operation. Until the boundary of the measurement space is reached; or the anisotropy coefficient in the voxel is less than the threshold A; or the angle between the nerve fiber bundles in the two connected voxels is greater than the threshold B. In this example, the threshold B is taken as 60 degrees, and in general, the threshold value ranges from 45 to 90 degrees. By connecting the directions of the nerve fiber bundles in this series of voxels in space, the overall direction of a nerve fiber bundle in space is obtained. Traverse all the seed points to get the neural distribution image.

Fusion unit 240: used to realize the fusion of the above three kinds of images. The drawn diffusion motion envelope image is placed in the anisotropic image, and then the seed points of the reconstructed nerve fiber bundles are aligned with the corresponding voxels in the anisotropic image to complete the anisotropic image. The fusion of network image and neural distribution image.

In addition, embodiments of the present disclosure also include magnetic resonance imaging apparatuses. As shown in FIG. 3 , the magnetic resonance imaging apparatus 300 includes an image fusion processing apparatus 200 . The image fusion processing apparatus 200 may be the configuration of the embodiment with reference to FIG. 2 .

As an example, each step of the above image fusion method and each component module and/or unit of the above image fusion processing apparatus may be implemented as software, firmware, hardware or a combination thereof. When implemented by software or firmware, a program constituting software for implementing the above-described method can be installed from a storage medium or a network to a computer having a dedicated hardware configuration, and the computer can execute various functions when various programs are installed. Wait.

The present invention also provides a program product storing machine-readable instruction codes. When the instruction code is read and executed by the machine, the above-mentioned neuroimaging method according to the embodiment of the present invention can be executed.

Correspondingly, a storage medium for carrying the above-mentioned program product storing the machine-readable instruction code is also included in the disclosure of the present invention. The storage medium includes, but is not limited to, a floppy disk, an optical disk, a magneto-optical disk, a memory card, a memory stick, and the like.

In the above description of specific embodiments of the invention, features described and/or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments as the features of the other embodiments. In combination with, or in place of, features in other embodiments.

It should be emphasized that the term "comprising/comprising" when used herein refers to the presence of a feature, element, step or component, but does not exclude the presence or addition of one or more other features, elements, steps or components.

In the above-described embodiments and examples, reference numerals composed of numbers are used to represent various steps and/or units. Those of ordinary skill in the art should understand that these reference numerals are only for convenience of description and drawing, and do not represent their order or any other limitation. While the present invention has been disclosed above by way of descriptions of specific embodiments thereof, it should be understood that all embodiments and examples described above are illustrative and not restrictive. Various modifications, improvements or equivalents of the present invention may be devised by those skilled in the art within the spirit and scope of the appended claims. Such modifications, improvements or equivalents should also be considered to be included within the scope of protection of the present invention.

Claims (6)

1. an image processing method, for realizing the fusion processing of anisotropic image, diffusion envelope surface image and nerve fiber distribution image, it is characterized in that comprising the steps: Step 1. Obtain the NMR data of the object; Step 2. Process the nuclear magnetic resonance data to obtain the diffusion motion envelope surface in each voxel of the object; Step 3. Calculate the direction and anisotropy parameters of the nerve fiber bundles in each voxel; Step 4. According to the calculated anisotropy parameters, generate an anisotropic image in the selected slice plane; Step 5. In the selected slice plane, select the diffusion motion envelope surface in the voxel whose anisotropy parameter is greater than the threshold value to generate a diffusion envelope surface image; Step 6. In the selected slice plane, select the voxel whose anisotropy parameter is greater than the threshold as the seed point, and from the seed point, reconstruct the nerve fiber distribution image according to the direction of the nerve fiber bundle; Step 7. Fusion of the anisotropic image, diffusion envelope image and neural distribution image in the selected slice plane. 2 . The image processing method according to claim 1 , wherein the anisotropy parameter is calculated according to the diffuse motion envelope. 3 . 3 . The image processing method according to claim 1 , wherein the threshold value is any real number from 0 to 1. 4 . 4. an image processing device, adopting the image fusion processing method of any one of claims 1-3 to fuse anisotropic images, diffusion envelope surface images and nerve fiber distribution images, it is characterized in that comprising the following units: a sampling unit for acquiring nuclear magnetic resonance data of the object; a calculation unit, used to obtain the diffuse motion envelope surface in each voxel of the object, and calculate the direction and anisotropy parameters of the nerve fiber bundle; an image generation unit for generating anisotropic images, diffusion envelope images and nerve fiber distribution images respectively; The fusion unit is used to realize the fusion of anisotropic images, diffusion envelope images and nerve fiber distribution images. 5 . The image processing apparatus according to claim 4 , wherein, in the calculation unit, the anisotropy parameter is calculated according to the diffuse motion envelope surface. 6 . 6. A nuclear magnetic resonance imaging apparatus comprising the image processing of any one of claims 4-5.

Images