ANALYSIS SYSTEM AND METHOD BASED ON NUCLEAR MAGNETIC RESONANCE IMAGE

Abstract

An embodiment of the present invention provides an analysis system and method based on nuclear magnetic resonance images, wherein the analysis system includes: an acquisition module, used to acquire MRI images of pre-constructed experimental groups and control groups; an extraction module, used to extract multiple regions in the MRI images as analysis samples, and process the analysis samples to obtain feature data; an analysis module, used to screen and analyze the feature data using a preset screening strategy to obtain screened difference features; wherein the screening strategy includes a minimum classification error and a minimum average correlation coefficient method; the effect is: the obtained differences are re-screened to thereby improve the accuracy of the obtained difference features, in order to find objective biomarkers.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing, and in particular relates to an analysis system and method based on nuclear magnetic resonance images. Background Art

[0002] Magnetic resonance imaging (MRI), as a non-invasive neuroimaging technology, plays an important role in the medical field. In recent years, MRI has made significant progress in the study of depression.

[0003] Currently, there have been studies on the application of magnetic resonance imaging (MRI) in the assessment of depression; however, most of these methods use a comparative approach to obtain differences, which are then used as the basis for assessment. However, the differences themselves are not screened, thereby reducing the accuracy.

[0004] Therefore, it is of great significance to screen the obtained differences in order to find objective biomarkers. Summary of the invention

[0005] In view of the technical deficiencies mentioned in the background technology, an object of the embodiments of the present invention is to provide an analysis system and method based on nuclear magnetic resonance images.

[0006] In the first aspect, the present invention provides an analysis system based on nuclear magnetic resonance images, comprising:

[0007] An acquisition module, used to acquire pre-constructed MRI images of the experimental group and the control group;

[0008] An extraction module, used for extracting a plurality of regions in the MRI image as analysis samples, and processing the analysis samples to obtain feature data;

[0009] The analysis module is used to perform screening analysis on the feature data using a preset screening strategy to obtain the screened difference features; wherein the screening strategy includes the minimum classification error and the minimum average correlation coefficient method.

[0010] Preferably, the analysis sample is processed to obtain characteristic data, specifically including:

[0011] The extracted regions were statistically analyzed using the gray-level co-occurrence matrix;

[0012] The descriptive information obtained by the analysis is then combined with the statistical attributes, and the calculation results are used as the feature data.

[0013] Preferably, the MRI images include fMRI and sMRI, and the analysis module is further configured to:

[0014] Performing head motion correction on the fMRI to remove head motion interference;

[0015] Registration, aligning the structural image with the functional image;

[0016] Standardization and smoothing filtering.

[0017] Preferably, the filtering is followed by de-linearization and low-pass filtering.

[0018] Preferably, the analysis module further provides an external interface, through which an external reference image is acquired, and the analysis sample is re-determined in combination with the reference image; wherein the reference image is a brain image with electrical signals.

[0019] In the second aspect, the present invention further provides an analysis method based on nuclear magnetic resonance images, which is applied to the analysis system based on nuclear magnetic resonance images described in the first aspect, and the method comprises:

[0020] Obtain MRI images of pre-constructed experimental and control groups;

[0021] Extracting a plurality of regions in the MRI image as analysis samples, and processing the analysis samples to obtain feature data;

[0022] The characteristic data are screened and analyzed using a preset screening strategy to obtain the screened difference characteristics; wherein the screening strategy includes the minimum classification error and the minimum average correlation coefficient method.

[0023] Preferably, the analysis sample is processed to obtain characteristic data, specifically including:

[0024] The extracted regions were statistically analyzed using the gray-level co-occurrence matrix;

[0025] The descriptive information obtained by the analysis is then combined with the statistical attributes, and the calculation results are used as the feature data.

[0026] Preferably, the MRI images include fMRI and sMRI, and the method further comprises:

[0027] Performing head motion correction on the fMRI to remove head motion interference;

[0028] Registration, aligning the structural image with the functional image;

[0029] Standardization and smoothing filtering.

[0030] Preferably, the filtering is followed by de-linearization and low-pass filtering.

[0031] Preferably, the method further comprises:

[0032] An external reference image is acquired through the external interface, and the analysis sample is re-determined in combination with the reference image; wherein the reference image is a brain image with electrical signals.

[0033] It can be seen from the above technical solution that the present invention provides an analysis system and method based on nuclear magnetic resonance images, which first selects multiple areas of the MRI image as analysis samples, and processes the analysis samples to obtain characteristic data; then uses a preset screening strategy to screen and analyze the characteristic data to obtain the screened difference characteristics; thereby, on the basis of the difference acquisition, the obtained differences are re-screened, thereby improving the accuracy of the obtained difference characteristics, so that the obtained biomarkers are more objective. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0035] Figure 1 A structural block diagram of an analysis system based on nuclear magnetic resonance images provided by an embodiment of the present invention;

[0036] Figure 2 A flowchart of an analysis method based on nuclear magnetic resonance images provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The embodiments of the technical solution of the present invention will be described in detail below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore only used as examples, and the protection scope of the present invention cannot be limited by this. It should be understood that when used in this specification and the appended claims, the terms "include" and "comprise" indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their collections. The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described here are only used for illustration and are not intended to limit the present invention. In the following description, in order to provide a thorough understanding of the present invention, a large number of specific details are set forth. However, it is obvious to those of ordinary skill in the art that these specific details do not have to be adopted to implement the present invention.

[0038] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "in one embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. Furthermore, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any appropriate combinations and / or subcombinations.

[0039] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should have the common meanings understood by those skilled in the art to which the present invention belongs.

[0040] Embodiment 1:

[0041] Reference Figure 1 , an analysis system based on nuclear magnetic resonance images, comprising:

[0042] The acquisition module is used to obtain the pre-constructed MRI images of the experimental group and the control group.

[0043] Specifically, this embodiment selects a number of healthy brain MRI images with no history of mental illness as a control group, and a number of depression patients as an experimental group; this application is only used as an example for illustration.

[0044] The extraction module is used to extract multiple areas in the MRI image as analysis samples and process the analysis samples to obtain feature data.

[0045] Specifically, multiple ROI regions are extracted as analysis samples, and the ROI regions can be selected from the right insula and the bilateral middle occipital gyrus, lingual gyrus, calcarine gyrus, precentral / postcentral gyrus, supramarginal gyrus, superior temporal gyrus and middle cingulate gyrus, etc.; wherein the analysis samples are processed to obtain feature data, specifically including:

[0046] The extracted regions were statistically analyzed using the gray-level co-occurrence matrix;

[0047] The descriptive information obtained by the analysis is then combined with the statistical attributes, and the calculation results are used as the feature data.

[0048] Gray-Level Co-occurrence Matrix (GLCM) is a matrix that statistically describes the grayscale relationship between adjacent pixels in a local area or the entire area of ​​an image, or between two pixels within a certain distance. The element values ​​in this matrix represent the joint conditional probability density P(i, j|d, θ) between grayscale levels, which means that when the spatial distance d and direction θ are given, the grayscale starts at point i and the probability (i.e., frequency) of grayscale level j appears. These are descriptive information;

[0049] Normalize the co-occurrence matrix to obtain the co-occurrence probability matrix. The element in the co-occurrence probability matrix is ​​P, which represents the probability of the pixel gray level i and j appearing at the same time under the specified direction and distance;

[0050] According to the co-occurrence probability matrix, the statistical properties used to describe the texture features are calculated, such as contrast, correlation, energy, entropy and inverse variance; based on these values, the texture features of the image can be further analyzed;

[0051] Contrast: used to describe the contrast between different gray levels in image texture;

[0052] Correlation: used to describe the correlation of pixel gray levels in image texture;

[0053] Energy (also called contrast momentum): used to describe the degree of texture detail of image texture;

[0054] Entropy: used to describe the uncertainty and complexity of image texture;

[0055] Inverse variance: It is used to reflect the magnitude of local changes in image texture. If the image texture is relatively uniform in different areas and changes slowly, the inverse variance will be larger, otherwise it will be smaller.

[0056] The analysis module is used to perform screening analysis on the feature data using a preset screening strategy to obtain the screened difference features; wherein the screening strategy includes the minimum classification error and the minimum average correlation coefficient method.

[0057] Specifically, the effective screening of texture features through the screening strategy is based on the principle of minimizing the average correlation coefficient between texture features, thereby screening out multiple classification texture feature parameters.

[0058] When applied, the system can be combined with machine learning to improve the accuracy of processing differential brain network structures; machine learning uses SVM to solve the problem of small samples of fMRI data.

[0059] Furthermore, during implementation, in order to improve the accuracy of the data, the MRI image includes fMRI and sMRI, and the analysis module is also used for:

[0060] Performing head motion correction on the fMRI to remove head motion interference;

[0061] Registration, aligning the structural image with the functional image;

[0062] Standardization and smoothing filtering.

[0063] Among them, fMRI is functional magnetic resonance imaging, which scans a complete image data (a set of data) of the brain, and extracts the time series of voxels to represent the neural activity of the brain;

[0064] Common fMRI indicators include:

[0065] Amplitude of low-frequency oscillations (ALFF): ALFF is used to measure changes in blood oxygen in the low-frequency part of the brain caused by neural activity, and can be used to measure the strength of local neural activity in the brain;

[0066] Local consistency (ReHo), which can reflect the functional consistency of local blood oxygen levels during the resting period by calculating the Kendall concordance coefficient of a certain number of adjacent voxels at the voxel level;

[0067] Functional connectivity FC, Voxel-wise FC analyzes the synchronization between specific brain regions and voxels in the whole brain, considering the impact of specific regions on the whole brain;

[0068] Voxel-mirror homotopic connectivity (VMHC), which can reflect the connection between contralateral brain regions in the left and right hemispheres;

[0069] Functional connectivity density (FCD) is a measure of the degree of connection between each brain site (voxel) and every other site (voxel).

[0070] sMRI is structural magnetic resonance imaging;

[0071] Specifically, segmentation is performed after registration, and the structural image registered with the functional image is segmented into gray matter, white matter, etc., and standardized; standardization is to convert the individual bold image into a standard space;

[0072] Smoothing filtering is used to eliminate the subtle differences in brain structure of different subjects and improve the signal-to-noise ratio of the data; the filtering here is spatial filtering;

[0073] Then, temporal filtering is performed; that is, after the filtering, de-linearization and low-pass filtering are performed; wherein, de-linearization is high-pass filtering.

[0074] To improve the comprehensiveness of the processing, the analysis module is also used to:

[0075] The region corresponding to the characteristic data is marked, and a morphological difference analysis is performed on the marked region; wherein, when marking, the region is also expanded and reduced.

[0076] When implemented, the morphological difference analysis includes calculation of volume, surface area and curvature, and when any calculated value exceeds a corresponding preset value, it is determined to be abnormal;

[0077] Meanwhile, if the marked region determined to be abnormal is not in the extracted multiple regions, it is added to the analysis sample.

[0078] In another embodiment, the morphological difference analysis adopts a method of selecting reference points, and detects the significant morphological change areas of the depression group relative to the control group based on the calculated tensors and multiple distance values ​​of the two groups of brain structure networks.

[0079] The above scheme first selects multiple areas of the MRI image as analysis samples, and processes the analysis samples to obtain characteristic data; then the characteristic data is screened and analyzed using a preset screening strategy to obtain the screened difference characteristics; then, based on the difference acquisition, the obtained differences are re-screened, thereby improving the accuracy of the obtained difference characteristics, making the obtained biomarkers more objective.

[0080] Furthermore, in order to reduce omissions, the obtained information is complemented. In another embodiment, based on the above technical solution, the analysis module also provides an external interface, through which an external reference image is obtained, and the analysis sample is re-determined in combination with the reference image; wherein the reference image is a brain map with electrical signals.

[0081] The brain map may adopt magnetoencephalography, which detects brain activity with millisecond-level temporal and spatial resolutions. Its main advantage is that the skull and soft tissue between the brain and scalp will not distort or attenuate the magnetic signal, and there is no signal distortion. It has both high temporal and spatial resolutions, making it easy to find abnormal signals in neural networks and regions, thus complementing MRI images.

[0082] Furthermore, in order to improve the accuracy of MRI image processing, in another embodiment, when performing feature extraction, the analysis module also directly uses LBP (Local Binary Pattern) to extract texture features from multiple regions as analysis samples, specifically including:

[0083] Divide the detection window into small areas (cells) of 16×16;

[0084] For a pixel in each cell, the grayscale values ​​of the eight adjacent pixels are compared with it. If the surrounding pixel values ​​are greater than the central pixel value, the position of the pixel is marked as 1, otherwise it is 0; thus the LBP value of the central pixel of the window is obtained;

[0085] Then calculate the histogram of each cell, that is, the frequency of each number; then normalize the histogram;

[0086] Finally, the statistical histograms of each cell are connected to form a feature vector, which is the LBP texture feature vector of the entire image.

[0087] Then use SVM or other machine learning algorithms for classification; at the same time, use the genetic algorithm to optimize the penalty coefficient C and kernel function parameters in SVM to improve the classification performance, and then train based on sample data to improve the accuracy of analysis and processing.

[0088] Furthermore, the analysis module can also analyze the real-time MRI images using graph theory as a tool to construct a brain network with brain regions as nodes and the correlation of time series between brain regions as edges, so as to characterize the global and local characteristics of the brain network, including:

[0089] Functional network construction: based on the brain area division atlas of interest, the brain is divided into multiple regions, the average time signal within the region is extracted, and the correlation between two brain regions is calculated;

[0090] Based on the analysis of the link edge, the link edge can be directly analyzed to compare the inter-group differences in the link edge, examine the changes in the link edge over time, and the correlation with the scale score;

[0091] Analysis of graph properties, that is, analysis of global characteristics and local characteristics; among them, global characteristics include shortest path, clustering coefficient, global efficiency, synchronization, etc.; local characteristics include node efficiency, node degree centrality, node clustering coefficient, node shortest path, etc.

[0092] Embodiment 2:

[0093] Based on the same inventive concept, the embodiment of the present invention also provides an analysis method based on nuclear magnetic resonance images, referring to Figure 2 , applied to the above-mentioned analysis system based on nuclear magnetic resonance images, the method comprises:

[0094] S101, obtaining pre-constructed MRI images of the experimental group and the control group;

[0095] S102, extracting multiple regions in the MRI image as analysis samples, and processing the analysis samples to obtain feature data;

[0096] S103, using a preset screening strategy to perform screening analysis on the feature data to obtain screened difference features; wherein the screening strategy includes minimum classification error and minimum average correlation coefficient method.

[0097] The analysis sample is processed to obtain characteristic data, specifically including:

[0098] The extracted regions were statistically analyzed using the gray-level co-occurrence matrix;

[0099] The descriptive information obtained by the analysis is then combined with the statistical attributes, and the calculation results are used as the feature data.

[0100] Furthermore, the MRI image includes fMRI and sMRI, and the method further includes:

[0101] Performing head motion correction on the fMRI to remove head motion interference;

[0102] Registration, aligning the structural image with the functional image;

[0103] Standardization and smoothing filtering.

[0104] Furthermore, after the filtering, de-linear trending and low-pass filtering are performed.

[0105] Furthermore, the method further comprises:

[0106] An external reference image is acquired through the external interface, and the analysis sample is re-determined in combination with the reference image; wherein the reference image is a brain image with electrical signals.

[0107] It should be noted that for more specific workflows and effects of the method embodiments, please refer to the aforementioned system embodiment section, which will not be repeated here.

[0108] By first selecting multiple areas of the MRI image as analysis samples, and processing the analysis samples to obtain characteristic data; then using a preset screening strategy to screen and analyze the characteristic data to obtain the screened difference characteristics; then on the basis of the difference acquisition, the obtained differences are re-screened, thereby improving the accuracy of the obtained difference characteristics, in order to find objective biomarkers.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. An analysis system based on nuclear magnetic resonance images, characterized in that: include: An acquisition module, used to acquire pre-constructed MRI images of the experimental group and the control group; An extraction module, used for extracting a plurality of regions in the MRI image as analysis samples, and processing the analysis samples to obtain feature data; An analysis module, used to perform screening analysis on the feature data using a preset screening strategy to obtain the screened difference features; wherein the screening strategy includes a minimum classification error and a minimum average correlation coefficient method; The analysis module is also provided with an external interface, through which an external reference image is obtained, and the analysis sample is re-determined in combination with the reference image; wherein the reference image is a brain map with electrical signals; the brain map adopts magnetoencephalography, which explores brain activity with millisecond-level time resolution and spatial resolution; The analysis module is also used for: Marking the area corresponding to the characteristic data, and performing morphological difference analysis on the marked area; wherein, when marking, the area is also expanded and reduced; The morphological difference analysis includes the calculation of volume, surface area and curvature, and when any calculated value exceeds the corresponding preset value, it is determined to be abnormal; Meanwhile, if the marked region determined to be abnormal is not in the extracted multiple regions, it is added to the analysis sample; The morphological difference analysis adopts the method of selecting reference points, and detects the significant morphological change areas of the depression group relative to the control group based on the calculated tensors and multiple distance values ​​of the two groups of brain structure networks.

2. The analysis system based on nuclear magnetic resonance images according to claim 1, characterized in that: The analysis sample is processed to obtain characteristic data, specifically including: The extracted regions were statistically analyzed using the gray-level co-occurrence matrix; The descriptive information obtained by the analysis is then combined with the statistical attributes, and the calculation results are used as the feature data.

3. The analysis system based on nuclear magnetic resonance images according to claim 2, characterized in that: The MRI images include fMRI and sMRI, and the analysis module is further configured to: Performing head motion correction on the fMRI to remove head motion interference; Registration, aligning the structural image with the functional image; Standardization and smoothing filtering.

4. The analysis system based on nuclear magnetic resonance images according to claim 3, characterized in that: After the filtering, de-linear trending and low-pass filtering are also performed.

5. An analysis method based on nuclear magnetic resonance images, characterized in that: The method applied to the analysis system based on nuclear magnetic resonance images as described in claim 1 comprises: Obtain MRI images of pre-constructed experimental and control groups; Extracting a plurality of regions in the MRI image as analysis samples, and processing the analysis samples to obtain feature data; The characteristic data are screened and analyzed using a preset screening strategy to obtain screened difference characteristics; wherein the screening strategy includes a minimum classification error and a minimum average correlation coefficient method; The method further comprises: Acquire an external reference image through the external interface, and redetermine the analysis sample in combination with the reference image; wherein the reference image is a brain map with electrical signals; the brain map adopts magnetoencephalography, which explores brain activity with millisecond-level time resolution and spatial resolution; Marking the area corresponding to the characteristic data, and performing morphological difference analysis on the marked area; wherein, when marking, the area is also expanded and reduced; The morphological difference analysis includes the calculation of volume, surface area and curvature, and when any calculated value exceeds the corresponding preset value, it is determined to be abnormal; Meanwhile, if the marked region determined to be abnormal is not in the extracted multiple regions, it is added to the analysis sample; The morphological difference analysis adopts the method of selecting reference points, and detects the significant morphological change areas of the depression group relative to the control group based on the calculated tensors and multiple distance values ​​of the two groups of brain structure networks.

6. The analysis method according to claim 5, characterized in that The analysis sample is processed to obtain characteristic data, specifically including: The extracted regions were statistically analyzed using the gray-level co-occurrence matrix; The descriptive information obtained by the analysis is then combined with the statistical attributes, and the calculation results are used as the feature data.

7. The analysis method according to claim 6, characterized in that The MRI images include fMRI and sMRI, and the method further includes: Performing head motion correction on the fMRI to remove head motion interference; Registration, aligning the structural image with the functional image; Standardization and smoothing filtering.

8. The analysis method according to claim 7, characterized in that After the filtering, de-linear trending and low-pass filtering are also performed.

Images