Micron-scale microscopic imaging method for tissue structure and function based on magnetic resonance imaging

Abstract

The invention relates to a micron-scale microscopic imaging method for an organization structure and functions based on magnetic resonance imaging. The micron-scale microscopic imaging method comprises the following steps: establishing a database 1 containing microscopic images and water molecule dispersion motion information on a microscopic scale; establishing a millimeter-scale magnetic resonance imaging image obtained based on a magnetic resonance dispersion principle, and a database 2 of dispersion motion rules and characteristics of all water molecules in voxels corresponding to each pixel in each image; and establishing a correlation mapping algorithm between the database 1 and the database 2, and obtaining microscopic image information corresponding to magnetic resonance macroscopic imaging based on the dispersion principle in a macroscopic scale through a bridge of water molecule dispersion motion characteristics in a microscopic level. According to the invention, the microscopic image for displaying cell and subcellular scale cell microstructures and functions is obtained noninvasively by adopting a magnetic resonance imaging technology with millimeter scale spatial resolution.

Description

Technical Field

[0001] The present invention relates to the field of magnetic resonance imaging technology, and in particular to a micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging. Background Art

[0002] Cells are the basic structural and functional units of human and mammalian tissues at the micrometer scale. If we can dynamically understand the structural and functional changes of cells, it will be extremely valuable for understanding the physiological functions of human and mammalian tissues, the mechanisms of disease occurrence and development, and the mechanism of action and effect evaluation of intervention measures such as drug treatment.

[0003] At present, optical microscopy is the main method to study and explore the microstructure and function of cells at the cellular and subcellular scales. However, the premise for optical cell imaging is that the imaging object needs to be optically transparent, and light needs to penetrate the imaging object before optical imaging can be achieved. However, most human tissues and mammalian tissues are optically opaque, and light cannot penetrate the human body, so it cannot be applied to optically opaque human or other biological tissues. At present, there is no non-invasive imaging method that can achieve micrometer-scale microscopic imaging of tissue microstructure and function under optically opaque conditions.

[0004] Diffusion magnetic resonance imaging is an imaging method that indirectly reflects information related to the diffusion movement of water molecules in tissues. It is a macroscopic non-invasive imaging method that can penetrate light-opaque human bodies or light-opaque biological tissues.

[0005] Diffusive motion is the diffusion displacement of water molecules at the micrometer scale. Diffusive motion can reflect the dynamic changes in the structure and function of cells in vivo. In the diffusion motion process of tens of milliseconds set in the diffusion magnetic resonance imaging sequence, the diffusion displacement of water molecules is distributed within the micrometer scale, which happens to fall within the cellular and subcellular scale. At the cellular and subcellular scale, the diffusion motion of water molecules is restricted by the cell membrane and by biological macromolecules such as proteins, deoxyribonucleic acid or ribonucleic acid. In addition, various cellular functions performed during processes such as activation, metabolism, proliferation, apoptosis or necrosis will also affect the diffusion motion of water molecules in the microenvironment.

[0006] Although diffusion motion itself reflects the information of tissue microstructure and function at the cellular and subcellular scales, it should be noted that the spatial resolution of current diffusion MRI is still limited to the millimeter scale, which is a macroscopic scale imaging. This means that there are millions of cells distributed in each pixel (representing each corresponding voxel) in the MRI image. Therefore, at present, diffusion MRI cannot obtain microscopic images at the cellular and subcellular scales.

[0007] Therefore, in view of the shortcomings of the existing technology, it is necessary to provide a micron-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging, which is based on the principle of magnetic resonance imaging of the diffusion motion of water molecules and can provide micron-scale information of tissue microstructure and function, so as to solve the shortcomings of the existing technology. Summary of the invention

[0008] The purpose of the present invention is to avoid the shortcomings of the prior art and provide a micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging, which can dynamically reflect the changes in the microstructure and function of cells.

[0009] The purpose of the present invention is achieved through the following technical measures.

[0010] A micrometer-scale microscopic imaging method for tissue structure and function based on magnetic resonance imaging is provided, comprising:

[0011] Establish a database containing microscopic images and information on the diffusion movement of water molecules at the microscopic scale1;

[0012] Establishing a millimeter-scale magnetic resonance imaging image based on the magnetic resonance diffusion principle, and a database of the diffusion motion laws and characteristics of all water molecules in the voxel corresponding to each pixel in each image 2; and

[0013] An association mapping algorithm is established between database 1 and database 2, and the microscopic image information corresponding to the macroscopic magnetic resonance macroscopic imaging based on the diffusion principle is obtained through the bridge of the diffusion motion characteristics of water molecules at the microscopic level.

[0014] Preferably, in the above-mentioned micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging, database 1 contains micrometer-scale microscopic images of various tissue types, various disease types, and various cell types, as well as characterization information reflecting the diffusion movement laws and characteristics of water in the microscopic images.

[0015] Preferably, the above-mentioned micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging, constructing database 1, specifically comprises the following steps:

[0016] S1: Using non-magnetic resonance imaging, a measurement technology that can directly measure the motion laws and related parameters of water molecule diffusion motion, combined with microscope imaging technology, to build a database based on microscopic images containing water molecule diffusion motion laws and related coefficients1;

[0017] S2: Use technology that can perform quantitative analysis on microscopic images to perform quantitative analysis on the various optical microscopic images obtained in S1, obtain the characteristics and parameters of the microscopic images, and integrate them into database 1.

[0018] Preferably, in the above-mentioned micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging, step S1 adopts coherent anti-Stokes Raman scattering microscopic imaging technology to construct a database of water molecule diffusion coefficients based on microscopic images.

[0019] Preferably, in the above-mentioned micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging, in S2, a technology capable of quantitatively analyzing microscopic images is specifically adopted to quantitatively analyze the various optical microscopic images obtained in S1, and a water molecule diffusion anisotropy parameter library based on microscopic images is obtained and integrated into database 1.

[0020] Preferably, in the above-mentioned micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging, a structural tensor analysis method is used in S2 to quantitatively analyze the microscopic image.

[0021] Preferably, in the above-mentioned micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging, database 2 contains millimeter-scale magnetic resonance imaging images of various tissue types and various disease types.

[0022] Preferably, in the above-mentioned micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging, the diffusion motion law and characteristics of water molecules in the voxel in the database 2 are specifically a set of probability density functions of the diffusion motion distribution of water molecules in the voxel,

[0023] The probability density function set of the diffusion motion distribution of water molecules within a voxel includes a probability density function set of the diffusion coefficient distribution of water molecules within a single voxel based on the diffusion magnetic resonance image and a probability density function set of the diffusion anisotropy distribution of water molecules within a single voxel based on the diffusion magnetic resonance image.

[0024] Preferably, the above-mentioned micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging establishes an association mapping algorithm between database 1 and database 2. Specifically, through a pattern recognition algorithm or a machine learning algorithm, the probability density function distribution set in database 2 is mapped to the corresponding specific set of diffusion motion characteristics of water corresponding to the microscopic image in database 1, thereby realizing micrometer-scale microscopic imaging technology of tissue microstructure and function based on magnetic resonance.

[0025] When a pattern recognition algorithm is used, the following steps are included:

[0026] S1: Discretize the probability density function of the distribution of diffusion characteristics within voxels at the millimeter scale for different tissue types and different diseases;

[0027] S2: Using the minimum mean square error criterion, through continuous iterative processing, from the micrometer-scale tissue microstructure and diffusion function database of different tissue types and different diseases, find the subset of probability density function values ​​of the distribution of diffusion characteristics of different tissue types and different diseases in the millimeter scale voxels after discretization that completely matches;

[0028] When using a machine learning algorithm, the following steps are involved:

[0029] S1: Discretize the probability density function of the distribution of diffusion characteristics within voxels at the millimeter scale for different tissue types and different diseases;

[0030] S2: using a part of the data in database 1 as a training set, using a feature recognition algorithm or a clustering algorithm to analyze the training set; using the diffusion characteristics as a link, combining the diffusion characteristic data in database 2, using a feature recognition algorithm to jointly establish a feature library, and establishing a mapping training set between database 2 and database 1;

[0031] S3: Using the mapping algorithm between database 2 and database 1 obtained by machine learning algorithm, based on the newly obtained images and diffusion data that meet the data characteristics of database 2, the corresponding microscopic imaging information in database 1 is obtained, that is, micrometer-scale microscopic imaging of tissue microstructure and function based on diffusion magnetic resonance imaging is realized.

[0032] Preferably, in the above-mentioned method for micrometer-scale microscopic imaging of tissue structure and function based on magnetic resonance imaging, database 1 is first established and then database 2 is established, or database 2 is first established and then database 1 is established, or database 1 and database 2 are established at the same time.

[0033] The present invention is a micrometer-scale microscopic imaging method for tissue structure and function based on magnetic resonance imaging, including: establishing a database 1 containing microscopic images and water molecule diffusion motion information at a microscopic scale; establishing a millimeter-scale magnetic resonance imaging image based on the magnetic resonance diffusion principle, and a database 2 of the diffusion motion laws and characteristics of all water molecules in the voxel corresponding to each pixel in each image; and establishing an association mapping algorithm between database 1 and database 2, and obtaining macroscopic image information corresponding to magnetic resonance macroscopic imaging based on the diffusion principle through the bridge of water molecule diffusion motion characteristics at the microscopic level. The problem of using magnetic resonance imaging technology with millimeter-scale spatial resolution to obtain microscopic images showing cell and subcellular scale cell microstructures and functions is solved.

[0034] The present invention constructs a database 1 containing microscopic images of tissue structures of different tissue types and different diseases at micrometer scale and corresponding information on the diffusion motion of water in tissues; constructs a database 2 of a set of probability density functions of magnetic resonance imaging of different tissue types and different diseases at millimeter scale and corresponding voxels reflecting the distribution of the diffusion motion characteristics of water molecules; a mapping relationship between database 1 and database 2 is established around the information on the diffusion motion state of water molecules; when magnetic resonance imaging data that conforms to database 2 is obtained, the corresponding microscopic imaging information in database 1 is obtained to realize diffusion magnetic resonance imaging of tissue microstructure and function at micrometer scale. The invention realizes the use of magnetic resonance imaging technology with millimeter-scale spatial resolution to obtain microscopic images showing cell and subcellular scale cell microstructure and function.

[0035] The method of the present invention can perform magnetic resonance imaging on objects that meet the database 2, including imaging tissues or organs in different parts of the human body, or organs and tissues in light-opaque organisms, etc., through magnetic resonance, and then obtain the corresponding microscopic imaging results of the imaging object that meet the data characteristics of database 1. Microscopic imaging of the tissue structure and function of light-opaque objects in a non-invasive imaging method is achieved at the micrometer scale. The method of the present invention can dynamically understand the structural and functional changes of cells, which is of great value for understanding the physiological functions of human tissues and mammalian tissues, the mechanism of disease occurrence and development, and the mechanism of action and effect evaluation of intervention measures such as drug treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The present invention is further described with reference to the accompanying drawings, but the contents in the accompanying drawings do not constitute any limitation to the present invention.

[0037] Figure 1 It is a flow chart of a micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging of the present invention. DETAILED DESCRIPTION

[0038] The present invention will be further described with reference to the following examples.

[0039] Example 1.

[0040] The present invention aims to solve the problem of how to use magnetic resonance imaging technology with millimeter-scale spatial resolution to obtain microscopic images showing cell and subcellular scale cell microstructure and function. In order to achieve the purpose of the present invention, the diffusion magnetic resonance imaging of tissue microstructure and function at the micrometer scale needs to overcome the following technical problems:

[0041] (1) How to construct a database 1 containing micron-scale tissue structure microscopic images of different tissue types and different diseases and the corresponding information on the diffusion motion of water in the tissue; (2) How to construct a database 2 containing a set of probability density functions of different tissue types and different diseases and the distribution of the diffusion motion characteristics of water molecules in the corresponding voxels; (3) How to establish a mapping relationship between database 1 and database 2 around the diffusion motion state information of water molecules; so that when obtaining personal magnetic resonance imaging data that conforms to database 2, the corresponding microscopic imaging information in database 1 can be obtained, thereby realizing diffusion magnetic resonance imaging of tissue microstructure and function at the micron scale.

[0042] In order to solve the above problems, this embodiment provides a micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging, comprising:

[0043] Establish a database containing microscopic images and information on the diffusion movement of water molecules at the microscopic scale1;

[0044] Establishing a millimeter-scale magnetic resonance imaging image based on the magnetic resonance diffusion principle, and a database of the diffusion motion laws and characteristics of all water molecules in the voxel corresponding to each pixel in each image 2; and

[0045] An association mapping algorithm is established between database 1 and database 2, and the microscopic image information corresponding to the macroscopic magnetic resonance macroscopic imaging based on the diffusion principle is obtained through the bridge of the diffusion motion characteristics of water molecules at the microscopic level.

[0046] It should be noted that the method for micrometer-scale microscopic imaging of tissue structure and function based on magnetic resonance imaging can first establish database 1 and then establish database 2, or first establish database 2 and then establish database 1, or establish database 1 and database 2 at the same time.

[0047] Database 1 contains microscopic images of various tissue types, various disease types, various cell types, and micrometer scale, as well as characterization information reflecting the diffusion movement law and characteristics of water in the microscopic images. The characterization information is information such as various curves, charts, function sets, etc.

[0048] Building database 1 specifically includes the following steps:

[0049] S1: Using non-magnetic resonance imaging, a measurement technology that can directly measure the motion laws and related parameters of water molecule diffusion motion, combined with microscope imaging technology, to build a database based on microscopic images that contains water molecule diffusion motion laws and related coefficients1. For example, using coherent anti-Stokes Raman scattering microscopic imaging technology to build a database of water molecule diffusion coefficients based on microscopic images.

[0050] S2: Use technology that can perform quantitative analysis on microscopic images, such as structural tensor analysis methods, to quantitatively analyze the various optical microscopic images obtained in S1, obtain the characteristics and parameters of the microscopic images, such as obtaining a water molecule diffusion anisotropy parameter library based on microscopic images, and integrate it into database 1.

[0051] The constructed "database 2" is a database obtained based on the principle of magnetic resonance diffusion, which contains millimeter-scale magnetic resonance imaging images of different tissue types and different disease types, as well as the diffusion motion laws and characteristics of all water molecules in the voxels corresponding to each pixel in each image, such as the probability density function set of the diffusion motion distribution of water molecules, etc. The probability density function set of the distribution of diffusion characteristics in millimeter-scale voxels of different tissue types and different diseases includes the probability density function set of the distribution of the diffusion coefficient of water molecules in a single voxel based on the diffusion magnetic resonance image, the probability density function set of the distribution of the anisotropy of the diffusion of water molecules in a single voxel based on the diffusion magnetic resonance image, etc.

[0052] The diffusion motion laws and characteristics of water molecules within voxels in database 2 can be a set of probability density functions of the diffusion motion distribution of water molecules within voxels. The set of probability density functions of the diffusion motion distribution of water molecules within voxels includes a set of probability density functions of the distribution of diffusion coefficients of water molecules within a single voxel based on diffusion magnetic resonance images, and a set of probability density functions of the distribution of diffusion anisotropy of water molecules within a single voxel based on diffusion magnetic resonance images.

[0053] The micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging of this embodiment establishes an association mapping algorithm between database 1 and database 2, and uses a pattern recognition algorithm or a machine learning algorithm to map the probability density function distribution set in database 2 to the corresponding specific set of water diffusion motion characteristics corresponding to the microscopic image in database 1, so as to realize the micrometer-scale microscopic imaging technology of tissue microstructure and function based on magnetic resonance. That is, through the bridge of the diffusion motion characteristics of water molecules at the microscopic level, the information such as the microscopic image corresponding to the magnetic resonance macroscopic imaging at the macroscopic scale based on the diffusion principle is obtained, that is, the micrometer-scale microscopic imaging technology of tissue microstructure and function based on magnetic resonance is realized.

[0054] When a pattern recognition algorithm is used, the following steps are included:

[0055] S1: Discretize the probability density function of the distribution of diffusion characteristics within voxels at the millimeter scale for different tissue types and different diseases;

[0056] S2: Using the minimum mean square error criterion, through continuous iterative processing, from the micrometer-scale tissue microstructure and diffusion function database of different tissue types and different diseases, find the subset of probability density function values ​​of the distribution of diffusion characteristics of different tissue types and different diseases in the millimeter scale voxels after discretization that completely matches;

[0057] When using a machine learning algorithm, the following steps are involved:

[0058] S1: Discretize the probability density function of the distribution of diffusion characteristics within voxels at the millimeter scale for different tissue types and different diseases;

[0059] S2: using a part of the data in database 1 as a training set, using a feature recognition algorithm or a clustering algorithm to analyze the training set; using the diffusion characteristics as a link, combining the diffusion characteristic data in database 2, using a feature recognition algorithm to jointly establish a feature library, and establishing a mapping training set between database 2 and database 1;

[0060] S3: Using the mapping algorithm between database 2 and database 1 obtained by machine learning algorithm, based on the newly obtained images and diffusion data that meet the data characteristics of database 2, the corresponding microscopic imaging information in database 1 is obtained, that is, micrometer-scale microscopic imaging of tissue microstructure and function based on diffusion magnetic resonance imaging is realized.

[0061] This embodiment constructs a database 1 containing microscopic images of tissue structures of different tissue types and different diseases at the micrometer scale and the corresponding information on the diffusion motion of water in the tissue; constructs a database 2 containing a set of probability density functions of magnetic resonance imaging of different tissue types and different diseases at the millimeter scale and the corresponding voxel that reflects the distribution of the diffusion motion characteristics of water molecules; a mapping relationship between database 1 and database 2 is established around the information on the diffusion motion state of water molecules; when magnetic resonance imaging data that conforms to database 2 is obtained, the corresponding microscopic imaging information in database 1 is obtained to realize diffusion magnetic resonance imaging of tissue microstructure and function at the micrometer scale. It is realized that the magnetic resonance imaging technology with millimeter-scale spatial resolution is used to obtain microscopic images showing the microstructure and function of cells at the cellular and subcellular scales.

[0062] The method of the present invention can perform magnetic resonance imaging on objects that meet the database 2, including imaging tissues or organs in different parts of the human body, or organs and tissues in light-opaque organisms, etc., through magnetic resonance, and then obtain the corresponding microscopic imaging results of the imaging object that meet the data characteristics of database 1. Microscopic imaging of the tissue structure and function of light-opaque objects in a non-invasive imaging method is achieved at the micrometer scale. The method of the present invention can dynamically understand the structural and functional changes of cells, which is of great value for understanding the physiological functions of human tissues and mammalian tissues, the mechanism of disease occurrence and development, and the mechanism of action and effect evaluation of intervention measures such as drug treatment.

[0063] Example 2.

[0064] This embodiment takes the micron-scale microscopic imaging of tissue microstructure and function based on diffusion magnetic resonance imaging of liver fibrosis tissue as an example to explain in detail the micron-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging of the present invention.

[0065] (I) Construct a database of micrometer-scale tissue microstructure and diffusion function for different tissue types and different diseases, namely "Database 1"

[0066] 1.1 Construction of a water molecule diffusion coefficient database based on coherent anti-Stokes Raman scattering microscopy images, namely “Database 1”

[0067] This step uses coherent anti-Stokes Raman scattering microscopy to measure the diffusion coefficient of water molecules in fresh liver tissue, and constructs a microscopic image library of liver tissue and a corresponding water molecule diffusion coefficient database. Coherent anti-Stokes Raman scattering microscopy uses the vibration states of different molecules to produce image contrast, and there is no need to label the tissue during imaging, so the sample is hardly affected during preparation and imaging.

[0068] First, a high-resolution coherent anti-Stokes Raman scattering microscopic image is obtained. In order to obtain a microscopic image of fresh liver tissue, the target object (slice sample of fresh liver tissue) is obtained. The slice is fixed on a slide and imaged using a coherent anti-Stokes Raman scattering microscope. The laser light source equipped with the microscope generates two laser pulses with a specific frequency difference, which are called pump laser beam and Stokes laser beam, respectively, and the frequency difference is adjusted to correspond to the characteristic vibration frequency of water molecules. When the pump laser beam and the Stokes laser beam are simultaneously irradiated to the target object (liver) tissue sample, a specific coherent anti-Stokes Raman scattering signal is generated. This signal responds to the vibration characteristics and distribution state of water molecules, so it can be used to evaluate the diffusion behavior of water molecules in tissues. In the experiment, the laser beam needs to be focused to a specific depth of the sample, and the optical system needs to be adjusted to obtain scattered signals at different depths. By changing the timing of the pump laser beam and the Stokes laser beam, the signal intensity at different time intervals can be obtained, and then the diffusion characteristics of water molecules are quantitatively analyzed using a diffusion model based on Fick's law. By analyzing the movement of water molecules at different time points, the diffusion coefficient of water molecules in the target object (liver tissue) can be obtained.

[0069] To further improve the measurement accuracy, the above measurements can be repeated under different temperature, pH value and osmotic pressure conditions to observe the effects of these factors on the diffusion of water molecules and establish a corresponding correction model.

[0070] By continuously acquiring the diffusion coefficients on coherent anti-Stokes Raman scattering images of a specific size, a complete water molecule diffusion coefficient database is constructed. In order to construct a sufficiently large water molecule diffusion coefficient database as much as possible, as many liver tissues as possible as the target object need to be processed.

[0071] Similarly, microscopic images of other different types of healthy target tissues and diseased tissues as well as the diffusion motion signals of water molecules therein can be obtained. Microscopic images of different types of cell suspensions as well as the diffusion motion signals of water molecules therein can also be obtained.

[0072] 1.2 Obtaining a water molecule diffusion anisotropy data set based on microscopic images

[0073] This step is explained using the three-dimensional structure tensor analysis of confocal microscopy images as an example. It is necessary to obtain a large number of target objects (fresh liver tissue slices), perform three-dimensional structure tensor analysis after confocal microscopy imaging, obtain water molecule diffusion anisotropy parameters, and then construct a water molecule diffusion anisotropy database of liver tissue. The advantage of three-dimensional structure tensor analysis is that it does not need to rely on the direction of the tissue slice plane.

[0074] First, obtain a high-resolution confocal microscopic image. In order to obtain a confocal microscopic image of the target object (fresh liver tissue, it is first necessary to prepare a target tissue sample stored in phosphate buffer to maintain the physiological state of the tissue. Next, use a vibrating microtome, a microknife, or a tissue cutter to slice the liver to an appropriate thickness, usually 5μm-5mm. Fix the slices on a slide. Finally, use a confocal microscope for imaging, adjust the laser intensity, scanning speed, and focal length of the microscope to ensure high-quality images, and set the magnification to 10x. The plane resolution of the final image acquisition is set to 0.25×0.25μm 2 , the layer thickness is set to 1μm-5μm. First, a 2D image mosaic of the entire tissue section is constructed, and then the 2D images acquired at different levels are stitched into a 3D image. After the image is converted to tif format, it is imported into MATLAB for subsequent structural tensor analysis. In order to analyze the diffusion anisotropy of water molecules, the image is preprocessed, including denoising, image enhancement, and contrast adjustment, to ensure the clarity of the tissue microstructural features.

[0075] Next, a three-dimensional structure tensor analysis method was applied to calculate the gradient information around each pixel of the confocal microscopy image. The structure tensor is a mathematical method that describes the local structural features of an image. The tensor matrix can be constructed by calculating pixel gradients to capture the anisotropic structural features in the tissue. For each pixel, the structure tensor reflects the morphological and orientation characteristics of the surrounding area. According to the calculated structure tensor, the anisotropic structure in the image is quantified. The eigenvalues ​​of the tensor are used to analyze the main axis direction and diffusion intensity of the anisotropic diffusion direction, thereby inferring the possible diffusion path of water molecules in the tissue. The results are quantified as anisotropy indices to describe the diffusion anisotropy characteristics of different regions. By continuously acquiring the diffusion anisotropy values ​​on confocal images of a specific size, a complete water molecule diffusion anisotropy data set is constructed and integrated into the "database 1". In order to construct a sufficiently large water molecule diffusion anisotropy data set as much as possible, it is necessary to process as many fresh liver tissues as possible.

[0076] Similarly, we can obtain microscopic images of other different types of healthy tissues, diseased tissues and the diffusion motion signals of water molecules therein. We can also obtain microscopic images of different types of cell suspensions and the diffusion motion signals of water molecules therein.

[0077] (ii) Constructing a database of probability density function data sets of different tissue types and diseases at the millimeter scale, including magnetic resonance imaging images and intra-voxel diffusion characteristics distribution, i.e., “Database 2”

[0078] 2.1 Construction of a probability density function library for the distribution of water molecule diffusion coefficients in a single voxel based on diffusion magnetic resonance images

[0079] For liver fibrosis tissue, in the area contained in each magnetic resonance voxel, a large number of water molecules are undergoing complex diffusion motion, and statistical methods are usually required to quantitatively describe the complex diffusion process. We can reasonably assume that in a specific micro-region within a voxel, a single diffusion coefficient D can be used to quantitatively describe the diffusion motion within this micro-region. In other micro-regions within voxels, due to differences in the microenvironment, the diffusion coefficient D must show significant differences. As the volume of the micro-region gradually decreases, the distribution of the diffusion coefficient within a single voxel approaches a continuous distribution. It can be expressed as:

[0080] S / S0=∫P(D)exp(-b·D)dD (1);

[0081] S and S0 represent diffusion weighting factors b≠0 and b=0 s / mm respectively 2 The diffusion magnetic resonance signal amplitude of a single voxel when . The diffusion weighting factor b is a scanning parameter that can be manually set before the magnetic resonance scan. It represents the degree of weighting of the diffusion coefficient for the diffusion magnetic resonance signal. Under the same diffusion coefficient D, the larger the diffusion weighting factor b, the smaller the diffusion magnetic resonance signal amplitude S, and the higher the attenuation S / S0 of the diffusion magnetic resonance signal. P(D) represents the probability density function of the diffusion coefficient distribution within the voxel. In order to better capture the dynamic changes of the diffusion coefficient distribution within the voxel in complex physiological and pathological processes, a multi-component Gaussian mixture model is used to describe the diffusion coefficient distribution within the voxel. P(D) can be expressed as:

[0082]

[0083] Where N represents the number of Gaussian distributions, w represents the weight of Gaussian distributions, and k σ The mean is The variance is σ 2 Gaussian distribution function:

[0084]

[0085] In order to fit a continuous P(D), the following three steps are required:

[0086] (1) Selecting a combination of b values ​​for multi-exponential fitting; (2) Sampling diffusion coefficients by continuous multi-exponential fitting; (3) Estimate the continuous diffusion coefficient distribution function by kernel density estimation.

[0087] First, select the b value combination for multi-exponential fitting. We set the b value range to three b value intervals: [100, 1500], [1600, 3000], [3100, 4500] s / mm 2. The number of exponential terms is set between 2 and 15, and the number of b values ​​corresponding to the multi-exponential fit is 3 times the number of exponential terms. When the number of exponential terms is M, 3M b values ​​need to be selected, and M b values ​​are selected from the three b value intervals of high, medium and low. When selecting, the highest M-1 b values ​​in each b value interval are used as the default initial b value, and then selected from other b values. Finally, for the multi-exponential fit of M exponential terms, (16-N) are generated. 3 Group b value combination.

[0088] Then, based on the b value combination selected in the previous step, the diffusion coefficient is sampled by performing a continuous multi-exponential fit. The discretized equation (1) is fitted using the non-negative regularized least squares method, and the discretized expression is:

[0089]

[0090] M represents the number of discretized diffusion coefficients and also the number of exponential terms. d (D m ) represents the probability density function after discretization. The non-negative regularized least squares fitting problem becomes:

[0091]

[0092] The first term and the second term in expression (5) represent the fitting error term and the regularization term, respectively, μ represents the regularization factor, and G represents the number of diffusion weighting factors b.

[0093] Finally, the probability density function P(D) of the continuous diffusion coefficient distribution is estimated by performing kernel density estimation on the sampled discrete diffusion coefficients.

[0094] The diffusion motion information obtained above is combined with magnetic resonance imaging Figure 1 One to one, and together included in "Database 2".

[0095] 2.2 Obtaining a set of probability density functions of the anisotropic distribution of water molecules in a single voxel based on diffusion magnetic resonance images

[0096] Obtaining the probability density function of the anisotropic distribution of water molecule diffusion in a single voxel is the same as constructing a set of probability density functions of the distribution of the diffusion coefficient of water molecules in a single voxel. The anisotropy of water molecule diffusion can be measured by relative anisotropy (RA), which is defined as:

[0097]

[0098] Among them, λ are the eigenvalues ​​of the diffusion tensor, and Var(λ) represents the variance of the eigenvalue λ of the diffusion tensor.

[0099] The probability density function of the anisotropic distribution of water molecules in a voxel can be expressed as P(RA). The steps for calculating P(RA) are as follows: First, it is necessary to select a suitable combination of b values. Secondly, it is necessary to continuously sample the discrete relative anisotropy. Finally, the probability density function of the anisotropic distribution of the diffusion of water molecules in a continuous voxel is constructed by the method of kernel density estimation. The calculation method is the same as that described above for constructing the probability density function of the diffusion coefficient distribution of water molecules in a continuous voxel.

[0100] (III) Constructing the mapping relationship between "Database 1" and "Database 2"

[0101] This embodiment constructs a pattern recognition algorithm for searching a subset of probability density functions that can be combined to form a distribution of water molecule diffusion coefficients in a single voxel from a water molecule diffusion coefficient database based on microscopic images, and searching a subset of probability density functions that can be combined to form a distribution of water molecule diffusion anisotropy in a single voxel from a water molecule diffusion anisotropy database based on microscopic images. Through this pattern recognition algorithm, a set of microscopic images corresponding to a single voxel is established.

[0102] First, the continuous probability density function is discretized. The number of intervals after discretization depends on the ratio of the size of a single magnetic resonance voxel to the size of a single microscopic image, and the interval interval is determined according to the number of intervals. The continuous function is approximated by the Euler method, and the discretized function value is solved by the recursive formula. In the actual implementation process, in order to improve the precision of discretization, the step size can be appropriately reduced so that the discretized probability density function value is closer to the true value of the continuous probability density function.

[0103] Then, the value closest to the discretized probability density function value is searched from the water molecule diffusion coefficient database and the water molecule diffusion anisotropy database, and the qualified value is determined by minimizing the mean square error between the two. By searching the values ​​one by one, the subset of the water molecule diffusion coefficient database and the subset of the water molecule diffusion anisotropy database that can form a complete discretized probability density function are determined.

[0104] This embodiment constructs a database 1 containing microscopic images of tissue structures of different tissue types and different diseases at the micrometer scale and the corresponding information on the diffusion motion of water in the tissue; constructs a database 2 containing a set of probability density functions of magnetic resonance imaging of different tissue types and different diseases at the millimeter scale and the corresponding voxel that reflects the distribution of the diffusion motion characteristics of water molecules; a mapping relationship between database 1 and database 2 is established around the information on the diffusion motion state of water molecules; when magnetic resonance imaging data that conforms to database 2 is obtained, the corresponding microscopic imaging information in database 1 is obtained to realize diffusion magnetic resonance imaging of tissue microstructure and function at the micrometer scale. It is realized that the magnetic resonance imaging technology with millimeter-scale spatial resolution is used to obtain microscopic images showing the microstructure and function of cells at the cellular and subcellular scales.

[0105] The method of the present invention can perform magnetic resonance imaging on objects that meet the database 2, including imaging tissues or organs in different parts of the human body, or organs and tissues in light-opaque organisms, etc., through magnetic resonance, and then obtain the corresponding microscopic imaging results of the imaging object that meet the data characteristics of database 1. Microscopic imaging of the tissue structure and function of light-opaque objects in a non-invasive imaging method is achieved at the micrometer scale. The method of the present invention can dynamically understand the structural and functional changes of cells, which is of great value for understanding the physiological functions of human tissues and mammalian tissues, the mechanism of disease occurrence and development, and the mechanism of action and effect evaluation of intervention measures such as drug treatment.

[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the essence and scope of the technical solution of the present invention.

Claims

1. A micrometer-scale microscopic imaging method of tissue structure and function based on magnetic resonance imaging, characterized in that: include: Establish a database containing microscopic images and information on the diffusion movement of water molecules at the microscopic scale1; Establishing a millimeter-scale magnetic resonance imaging image based on the principle of magnetic resonance diffusion, and a database of the diffusion motion laws and characteristics of all water molecules in the voxel corresponding to each pixel in each image2; as well as An association mapping algorithm is established between database 1 and database 2, and the microscopic image information corresponding to the macroscopic magnetic resonance macroscopic imaging based on the diffusion principle is obtained through the bridge of the diffusion motion characteristics of water molecules at the microscopic level.

2. The method for micrometer-scale microscopic imaging of tissue structure and function based on magnetic resonance imaging according to claim 1, characterized in that: Database 1 contains microscopic images of various tissue types, various disease types, and various cell types at a micrometer scale, as well as characterization information reflecting the diffusion movement laws and characteristics of water in the microscopic images.

3. The method for micrometer-scale microscopic imaging of tissue structure and function based on magnetic resonance imaging according to claim 2, characterized in that: Building database 1 specifically includes the following steps: S1: Using non-magnetic resonance imaging, a measurement technology that can directly measure the motion laws and related parameters of water molecule diffusion motion, combined with microscope imaging technology, to build a database based on microscopic images containing water molecule diffusion motion laws and related coefficients1; S2: Use technology that can perform quantitative analysis on microscopic images to perform quantitative analysis on the various optical microscopic images obtained in S1, obtain the characteristics and parameters of the microscopic images, and integrate them into database 1.

4. The method for micrometer-scale microscopic imaging of tissue structure and function based on magnetic resonance imaging according to claim 3, characterized in that: Step S1 uses coherent anti-Stokes Raman scattering microscopy imaging technology to build a database of water molecule diffusion coefficients based on microscopic images.

5. The method for micrometer-scale microscopic imaging of tissue structure and function based on magnetic resonance imaging according to claim 3, characterized in that: In S2, a technology capable of quantitatively analyzing microscopic images is specifically used to quantitatively analyze the various optical microscopic images obtained in S1, and a water molecule diffusion anisotropy parameter library based on the microscopic images is obtained and integrated into database 1.

6. The method for micrometer-scale microscopic imaging of tissue structure and function based on magnetic resonance imaging according to claim 5, characterized in that: In S2, the structural tensor analysis method is used to quantitatively analyze the microscopic images.

7. The method for micrometer-scale microscopic imaging of tissue structure and function based on magnetic resonance imaging according to claim 3, characterized in that: Database 2 contains millimeter-scale magnetic resonance imaging images of various tissue types and various disease types.

8. The method for micrometer-scale microscopic imaging of tissue structure and function based on magnetic resonance imaging according to claim 7, characterized in that: The diffusion motion law and characteristics of water molecules in the voxel in the database 2 are specifically a set of probability density functions of the diffusion motion distribution of water molecules in the voxel; The probability density function set of the diffusion motion distribution of water molecules within a voxel includes a probability density function set of the diffusion coefficient distribution of water molecules within a single voxel based on the diffusion magnetic resonance image and a probability density function set of the diffusion anisotropy distribution of water molecules within a single voxel based on the diffusion magnetic resonance image.

9. The method for micrometer-scale microscopic imaging of tissue structure and function based on magnetic resonance imaging according to any one of claims 1 to 8, characterized in that: Establishing an association mapping algorithm between database 1 and database 2, specifically, mapping the probability density function distribution set in database 2 to the corresponding specific set of water diffusion motion characteristics corresponding to the microscopic image in database 1 through a pattern recognition algorithm or a machine learning algorithm, so as to realize the micrometer-scale microscopic imaging technology of tissue microstructure and function based on magnetic resonance; When a pattern recognition algorithm is used, the following steps are included: S1: Discretize the probability density function of the distribution of diffusion characteristics within voxels at the millimeter scale for different tissue types and different diseases; S2: Using the minimum mean square error criterion, through continuous iterative processing, from the micrometer-scale tissue microstructure and diffusion function database of different tissue types and different diseases, find the subset of probability density function values ​​of the distribution of diffusion characteristics of different tissue types and different diseases in the millimeter scale voxels after discretization that completely matches; When using a machine learning algorithm, the following steps are involved: S1: Discretize the probability density function of the distribution of diffusion characteristics within voxels at the millimeter scale for different tissue types and different diseases; S2: using a part of the data in database 1 as a training set, using a feature recognition algorithm or a clustering algorithm to analyze the training set; using the diffusion characteristics as a link, combining the diffusion characteristic data in database 2, using a feature recognition algorithm to jointly establish a feature library, and establishing a mapping training set between database 2 and database 1; S3: Using the mapping algorithm between database 2 and database 1 obtained by machine learning algorithm, based on the newly obtained images and diffusion data that meet the data characteristics of database 2, the corresponding microscopic imaging information in database 1 is obtained, that is, micrometer-scale microscopic imaging of tissue microstructure and function based on diffusion magnetic resonance imaging is realized.

10. The method for micrometer-scale microscopic imaging of tissue structure and function based on magnetic resonance imaging according to claim 9, characterized in that: First create database 1 and then create database 2, or first create database 2 and then create database 1, or create database 1 and database 2 at the same time.

Images