Nervous system image analysis method and system

Abstract

The invention relates to the technical field of image analysis, and discloses a nervous system image analysis method and system.The nervous system image analysis system comprises a data acquisition unit, an image processing unit, a segmentation processing unit, a feature extraction unit, a sample database and an analysis and evaluation unit; then, a segmentation processing unit carries out segmentation according to a normalized gray value to obtain a plurality of pixel domains, a feature extraction unit extracts pixel domain features, a sample database stores pre-trained sample pixel domains and feature data, a diagnosis result label is added, during diagnosis analysis, the class degree of the to-be-analyzed image pixel domain and the sample pixel domains is calculated, and the class degree of the to-be-analyzed image pixel domain and the sample pixel domains is calculated. According to the method, the images can be accurately analyzed, the features can be comprehensively extracted, the sample database is used for assisting diagnosis, and the efficiency and the accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and in particular to a method and system for analyzing images of a nervous system. Background Art

[0002] Neurological diseases are a class of diseases that seriously affect human health. Early and accurate diagnosis is crucial for the treatment and prognosis of the disease. Currently, neurological imaging analysis plays a key role in the diagnosis of neurological diseases. However, existing neurological imaging analysis methods have some shortcomings.

[0003] Traditional neurological imaging analysis mainly relies on the doctor's experience for subjective judgment. This method is not only inefficient, but also easily affected by the doctor's subjective factors, making it difficult to ensure the accuracy and consistency of the diagnosis results. In addition, with the continuous development of medical imaging technology, the amount of neurological imaging data continues to increase, and traditional manual analysis methods can no longer meet the needs of fast and accurate diagnosis. In order to improve the accuracy and efficiency of neurological imaging analysis, a more objective and accurate analysis method is needed. In recent years, with the continuous development of computer technology and image processing technology, computer-based neurological imaging analysis methods have gradually become a research hotspot.

[0004] Therefore, in order to solve the problems existing in the existing nervous system image analysis methods, the present invention proposes a nervous system image analysis method and system. The method and system improve the accuracy and efficiency of nervous system image analysis by automatically segmenting, extracting features and conducting diagnostic analysis on nervous system images, and provide strong support for the early diagnosis and treatment of nervous system diseases. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for analyzing nervous system images, which solves the technical problems raised in the background technology.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A method and system for analyzing nervous system images, comprising:

[0008] A data acquisition unit, used for acquiring nervous system images to be analyzed;

[0009] An image processing unit, used for converting the nervous system image into a grayscale image, and then performing image normalization processing on the grayscale value of each pixel on the grayscale image, and obtaining the grayscale value of each pixel after normalization processing;

[0010] A segmentation processing unit, for performing segmentation processing on the nervous system image according to the grayscale values ​​of all pixels corresponding to the nervous system image after normalization processing, and obtaining a segmentation mark corresponding to each pixel, and then obtaining a plurality of pixel domains from the nervous system image according to the segmentation mark, wherein the pixel domain refers to an image block composed of pixels with the same segmentation mark;

[0011] Segmentation marks are specific labels given to pixels to distinguish different pixel domains during the neural system image segmentation process;

[0012] A feature extraction unit is used to perform feature extraction processing on all pixel domains in the nervous system image and obtain pixel domain feature data;

[0013] The pixel domain feature data includes: the area, perimeter and shape factor of the pixel domain;

[0014] A sample database, used to store pre-trained sample pixel domains and their corresponding sample feature data, wherein the sample pixel domains are annotated with diagnostic results;

[0015] The analysis and evaluation unit is used to perform diagnostic analysis on each pixel domain and its pixel domain feature data corresponding to the nervous system image to be analyzed and the sample pixel domain and sample feature data in the sample database, and determine the diagnostic result of the pixel domain corresponding to the nervous system image to be analyzed.

[0016] As a further solution of the present invention: the image normalization processing method is as follows:

[0017] StepA1: Get the grayscale value of each pixel in the grayscale image and mark the grayscale value of each pixel as G in the coordinate system. (i,j) , where (i, j) represents the pixel coordinates in the image, i represents the nth horizontal coordinate, j represents the nth vertical coordinate, i = 1, 2, ... n, j = 1, 2, ... m, n × m is the number of all pixels in the nervous system image;

[0018] Wherein, i=1, 2, ... n, n represents the number of all pixels;

[0019] StepA2: From the grayscale values ​​of all pixels, obtain the grayscale value with the largest value and the grayscale value with the smallest value, and mark them as G and min and G max ;

[0020] Step A3, through H (i,j) =(G (i,j) -G min ) / (G max -G min ), and obtain the gray value H of each pixel after normalization (i,j).

[0021] As a further solution of the present invention: the segmentation processing method is as follows:

[0022] Step B1: Select a pixel from the nervous system image as a reference pixel and obtain the gray value H corresponding to the reference pixel. (i,j) ;

[0023] At the same time, segmentation marks are added to the reference pixels;

[0024] Step B2, based on H (i,j) , obtain the grayscale values ​​H of multiple pixels adjacent to the reference pixel (i-1,j) , H (i+1,j) , H (i,j-1) , H (i,j+1) , H (i+1,j+1) , H (i+1,j-1) , H (i-1,j-1) , H (i-1,j+1) ;

[0025] Step B3, calculate the gray value H corresponding to the reference pixel points respectively (i,j) The grayscale values ​​H of the adjacent pixels are (i-1,j) , H (i+1,j) , H (i,j-1) , H (i,j+1) , H (i+1,j+1) , H (i+1,j-1) , H (i-1,j-1) , H (i-1,j+1) The absolute value of the difference between them is then compared with the preset grayscale difference threshold Hy:

[0026] H (i,j) and H (i+1,j) For example;

[0027] If |H (i,j) -H (i+1,j) |>Hy, then a segmentation mark different from the reference pixel point is added to the pixel point adjacent to the reference pixel point;

[0028] If |H (i,j) -H (i+1,j) |≤Hy, then the pixel point adjacent to the reference pixel point will be added with the same segmentation mark as the reference pixel point;

[0029] Step B4, using each adjacent pixel point as a reference pixel point in the manner of Step B3, and obtaining the segmentation mark of each pixel point adjacent to the reference pixel point;

[0030] And so on, until all pixels are marked with segmentation marks;

[0031] Step B5, based on the segmentation mark, all the pixels with the same segmentation mark are grouped into a pixel domain, where the pixel domain refers to an image block composed of pixels with the same segmentation mark.

[0032] As a further solution of the present invention: in Step B4, the pixel points to which the segmentation mark has been added are no longer required to have the segmentation mark added a second time. When each adjacent pixel point is used as a reference pixel point, the reference pixel point selected in the early stage is no longer required to have the segmentation mark added.

[0033] As a further solution of the present invention: the feature extraction processing method is as follows:

[0034] Select a pixel region;

[0035] Step C1, obtaining the number of all pixels in the pixel domain, and then recording the number as the area of ​​the pixel domain;

[0036] Step C2, obtaining the number of all pixel points on the boundary of the pixel domain, and then recording the number value as the perimeter of the pixel domain;

[0037] The determination method of whether a pixel point in the pixel domain is on the boundary is as follows:

[0038] Among multiple pixel points adjacent to a pixel point, if at least one of the adjacent pixel points is not in the pixel domain, the pixel point is determined to be on the boundary;

[0039] If all adjacent pixels are in the pixel domain, it is determined that the pixel is not on the boundary;

[0040] Step C3, calculate the shape factor of the pixel domain based on the area and perimeter of the pixel domain;

[0041] Calculation formula: X = 4π × S / C2

[0042] X represents the shape factor of the pixel domain, S represents the area of ​​the pixel domain, and C represents the perimeter of the pixel domain;

[0043] The shape factor is used to measure how close the shape of a pixel domain is to a circular shape. When the shape factor is closer to 1, the pixel domain is closer to a circle. When the difference between the shape factor and 1 is larger, the shape of the pixel domain is more irregular.

[0044] As a further solution of the present invention: wherein, the pre-training method of the sample pixel domain and the sample feature data is to collect a large amount of historical nervous system image data containing diagnosis results in advance, and obtain all pixel domains and their corresponding pixel domain feature data from the historical nervous system image data through an image processing unit, a segmentation processing unit, and a feature extraction unit, and the obtained pixel domain and its corresponding pixel domain feature data are respectively the sample pixel domain and the sample feature data;

[0045] At the same time, diagnostic result annotations are added to each pixel domain through the diagnostic results corresponding to the historical data of nervous system imaging; wherein, the diagnostic result annotations are added manually.

[0046] As a further solution of the present invention: wherein the historical data of nervous system images include normal images and images of different types of nervous system diseases.

[0047] As a further solution of the present invention: the diagnostic analysis method is as follows:

[0048] Step D1, select a pixel domain and its pixel domain feature data corresponding to a nervous system image to be analyzed, and a sample pixel domain and its corresponding sample feature data as an example;

[0049] Then, the shape factor, area, and perimeter in the pixel domain feature data corresponding to the nervous system image to be analyzed are obtained and marked as X0, S0, and C0 respectively;

[0050] Step D2, mark the shape factor, area, and perimeter of the sample feature data corresponding to a sample pixel domain as X y , S y and C y ;

[0051] Step D3, then pass

[0052] K=1 / (|X0-X y |×β1+|S0-S y |×β2+|C0-C y |×β3)

[0053] In the formula, β1, β2, and β3 are the corresponding preset weight coefficients respectively;

[0054] Calculate the similarity K between the pixel domain corresponding to the nervous system image to be analyzed and the sample pixel domain;

[0055] Step D4, according to the method of Step D1 to Step D3, calculate the similarity between the pixel domain corresponding to the nervous system image to be analyzed and the pixel domains of all samples;

[0056] Step D5. Compare the obtained similarities with the preset similarity threshold, and based on the comparison result, obtain the corresponding sample pixel domain when the similarity is greater than the similarity threshold, and then obtain a sample pixel domain with the largest similarity value, and then obtain the diagnostic result annotation of the sample pixel domain, and use the diagnostic result annotation of the sample pixel domain as the diagnostic result of the pixel domain corresponding to the nervous system image to be analyzed.

[0057] As a further solution of the present invention: before Step D4, X0 and X y The absolute value of the difference between y The absolute value of the difference between S0 and C0 y The absolute value of the difference between them is used to screen out the sample pixel domain involved in the similarity calculation:

[0058] Select a sample pixel domain;

[0059] If |X0-X y |>XL,|S0-S y |>SL,|C0-C y |>If any of the items in CL is true, the sample pixel domain will not be extracted to participate in the similarity calculation

[0060] If |X0-X y |>XL,|S0-S y |>SL,|C0-C y |>If none of CL is true, then the sample pixel domain is extracted to participate in the similarity calculation;

[0061] Among them, XL, SL, and CL are the corresponding preset difference screening thresholds;

[0062] As a further solution of the present invention: in Step D4, when all similarities are less than or equal to the similarity threshold, it means that the sample pixel domain does not contain a similar sample pixel domain in the sample database, and then the diagnosis result is determined by relevant personnel based on experience;

[0063] At the same time, the sample pixel domain and its corresponding pixel domain feature data are stored in the sample database as sample pixel domain and sample feature data.

[0064] A method for analyzing nervous system images, which is implemented by the nervous system image analysis system, comprises the following steps:

[0065] Step 1: Data Acquisition

[0066] Obtaining images of the nervous system to be analyzed;

[0067] Step 2: Image processing

[0068] The nervous system image is converted into a grayscale image, and then the grayscale value of each pixel on the grayscale image is normalized to obtain the normalized grayscale value of each pixel;

[0069] Step 3: Segmentation

[0070] According to the grayscale values ​​of all pixels corresponding to the nervous system image after normalization, the nervous system image is segmented, and the segmentation mark corresponding to each pixel is obtained, and then multiple pixel domains are obtained from the nervous system image according to the segmentation mark;

[0071] Step 4: Feature extraction

[0072] Perform feature extraction processing on all pixel domains in the nervous system image and obtain pixel domain feature data;

[0073] Step 5: Sample database preparation

[0074] A large amount of historical neurological imaging data containing diagnostic results is collected in advance, and all pixel domains and their corresponding pixel domain feature data are obtained from the historical neurological imaging data through image processing, segmentation processing, and feature extraction, and the obtained pixel domains and their corresponding pixel domain feature data are respectively sample pixel domains and sample feature data; at the same time, the diagnostic results obtained in advance from the historical neurological imaging data are used to add diagnostic result annotations to each pixel domain;

[0075] Step 6: Analysis and Evaluation

[0076] The pixel domains and their pixel domain feature data corresponding to the nervous system image to be analyzed are diagnosed and analyzed with the sample pixel domains and sample feature data in the sample database, and the diagnosis result of the pixel domain corresponding to the nervous system image to be analyzed is determined.

[0077] Beneficial effects of the present invention:

[0078] The present invention converts the nervous system images into grayscale images and performs normalization processing, thereby providing a standardized data basis for subsequent analysis, reducing interference caused by factors such as image brightness differences, and making the analysis results more accurate and reliable.

[0079] The present invention performs segmentation processing on the image, assigns segmentation labels to pixel points according to the normalized grayscale values, and then obtains multiple pixel domains. This segmentation method can divide complex nervous system images into different areas, making it easier to analyze the characteristics of each area in more detail.

[0080] In the present invention, the feature extraction unit extracts features from the pixel domain from three aspects: area, perimeter and shape factor. The area and perimeter can intuitively reflect the size and boundary conditions of the pixel domain, while the shape factor can measure the proximity of the pixel domain to a circle, thereby more comprehensively describing the morphological characteristics of the pixel domain.

[0081] The present invention uses a large amount of pre-collected historical data of nervous system images containing diagnostic results for pre-training to obtain sample pixel domains and their corresponding sample feature data, and stores them in a sample database. These sample data cover normal images and images of different types of nervous system diseases, providing rich references for diagnostic analysis.

[0082] In the present invention, the sample pixel domain in the sample database is annotated with the diagnostic result. By comparing and analyzing the pixel domain feature data of the image to be analyzed with the data in the sample database, the diagnostic result of the pixel domain to be analyzed can be quickly determined, thereby improving the diagnostic efficiency.

[0083] The present invention, before calculating the similarity, screens out the sample pixel domains participating in the similarity calculation by comparing the absolute value of the difference between the pixel domain to be analyzed and the sample pixel domain in terms of shape factor, area and perimeter with a preset difference screening threshold. This can avoid unnecessary calculations for sample pixel domains with large differences, reduce the amount of calculations, and improve calculation efficiency.

[0084] In the present invention, when all similarities are less than or equal to the similarity threshold, it indicates that there is no similar sample pixel domain in the sample database. At this time, the diagnosis result is determined by relevant personnel based on experience, and the sample pixel domain and its feature data are stored in the sample database. This not only provides a solution for the current diagnosis, but also continuously enriches the sample database, improves the adaptability and accuracy of the system, and provides more references for subsequent diagnostic analysis.

[0085] In summary, the nervous system image analysis method and system of the present invention can realize accurate analysis and diagnosis of nervous system images, improve diagnostic efficiency and accuracy, and have good scalability and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] The present invention will be further described below in conjunction with the accompanying drawings.

[0087] Figure 1 It is a system block diagram of a nervous system image analysis system of the present invention.

[0088] Figure 2 It is a schematic flow chart of a method for analyzing nervous system images of the present invention. DETAILED DESCRIPTION

[0089] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0090] Embodiment 1

[0091] See also Figure 1 As shown, the present invention is a method and system for analyzing nervous system images, comprising:

[0092] A data acquisition unit, used for acquiring nervous system images to be analyzed;

[0093] An image processing unit, used for converting the nervous system image into a grayscale image, and then performing image normalization processing on the grayscale value of each pixel on the grayscale image;

[0094] The method is as follows:

[0095] StepA1: Get the grayscale value of each pixel in the grayscale image and mark the grayscale value of each pixel as G in the coordinate system. (i,j) , where (i, j) represents the pixel coordinates in the image, i represents the nth horizontal coordinate, j represents the nth vertical coordinate, i = 1, 2, ... n, j = 1, 2, ... m, n × m is the number of all pixels in the nervous system image;

[0096] Wherein, i=1, 2, ... n, n represents the number of all pixels;

[0097] StepA2: From the grayscale values ​​of all pixels, obtain the grayscale value with the largest value and the grayscale value with the smallest value, and mark them as G and min and G max ;

[0098] Step A3, through H (i,j) =(G (i,j) -G min ) / (G max -G min ), and obtain the gray value H of each pixel after normalization (i,j) ;

[0099] A segmentation processing unit, used for segmenting the nervous system image according to the gray value of all pixels after normalization;

[0100] The specific method is as follows:

[0101] Step B1: Select a pixel from the nervous system image as a reference pixel and obtain the gray value H corresponding to the reference pixel. (i,j) ;

[0102] At the same time, segmentation marks are added to the reference pixels;

[0103] Step B2, based on H (i,j) , obtain the grayscale values ​​H of multiple pixels adjacent to the reference pixel (i-1,j) , H (i+1,j) , H (i,j-1) , H (i,j+1) , H (i+1,j+1) , H (i+1,j-1) , H (i-1,j-1) , H (i-1,j+1) ;

[0104] Step B3, calculate the gray value H corresponding to the reference pixel points respectively (i,j) The grayscale values ​​H of the adjacent pixels are (i-1,j) , H (i+1,j) , H (i,j-1) , H (i,j+1) , H (i+1,j+1) , H (i+1,j-1) , H (i-1,j-1) , H (i-1,j+1) The absolute value of the difference between them is then compared with the preset grayscale difference threshold Hy:

[0105] H (i,j) and H (i+1,j) For example;

[0106] If |H (i,j) -H (i+1,j) |>Hy, then a segmentation mark different from the reference pixel point is added to the pixel point adjacent to the reference pixel point;

[0107] If |H (i,j) -H (i+1,j) |≤Hy, then the pixel point adjacent to the reference pixel point will be added with the same segmentation mark as the reference pixel point;

[0108] Step B4, using each adjacent pixel point as a reference pixel point in the manner of Step B3, and obtaining the segmentation mark of each pixel point adjacent to the reference pixel point;

[0109] And so on, until all pixels are marked with segmentation marks;

[0110] Among them, the pixel points that have been added with the segmentation mark will not be added with the segmentation mark again. For example, when each adjacent pixel point is used as the reference pixel point, the reference pixel point selected in the early stage will not be added with the segmentation mark;

[0111] Step B5, according to the segmentation mark, all the pixels with the same segmentation mark are grouped into a pixel domain, where the pixel domain refers to an image block composed of the pixels with the same segmentation mark;

[0112] A feature extraction unit is used to perform feature extraction processing on all pixel domains in the nervous system image and obtain pixel domain feature data;

[0113] The pixel domain feature data includes: the area, perimeter and shape factor of the pixel domain;

[0114] The specific method is as follows:

[0115] Take a pixel domain as an example;

[0116] Step C1, obtaining the number of all pixels in the pixel domain, and then recording the number as the area of ​​the pixel domain;

[0117] Step C2, obtaining the number of all pixel points on the boundary of the pixel domain, and then recording the number value as the perimeter of the pixel domain;

[0118] The determination method of whether a pixel point in the pixel domain is on the boundary is as follows:

[0119] Among multiple pixel points adjacent to a pixel point, if at least one of the adjacent pixel points is not in the pixel domain, the pixel point is determined to be on the boundary;

[0120] If all adjacent pixels are in the pixel domain, it is determined that the pixel is not on the boundary;

[0121] Step C3, calculate the shape factor of the pixel domain based on the area and perimeter of the pixel domain;

[0122] Calculation formula: X = 4π × S / C2

[0123] X represents the shape factor of the pixel domain, S represents the area of ​​the pixel domain, and C represents the perimeter of the pixel domain;

[0124] The shape factor is used to measure the degree of closeness of the shape of a pixel domain to a circular shape, wherein the closer the shape factor is to 1, the closer the pixel domain is to a circle, and the greater the difference between the shape factor and 1, the more irregular the shape of the pixel domain;

[0125] A sample database, used to store pre-trained sample pixel domains and their corresponding sample feature data, wherein the sample pixel domains are annotated with diagnostic results;

[0126] The pre-training method of the sample pixel domain and the sample feature data is to collect a large amount of historical neurological imaging data containing diagnostic results in advance, and obtain all pixel domains and their corresponding pixel domain feature data from the historical neurological imaging data through an image processing unit, a segmentation processing unit, and a feature extraction unit, and the obtained pixel domain and its corresponding pixel domain feature data are the sample pixel domain and the sample feature data respectively;

[0127] At the same time, the diagnostic result annotation is added to each pixel domain through the diagnostic result corresponding to the historical data of the nervous system imaging; wherein, the diagnostic result corresponding to the historical data of the nervous system imaging is a known result obtained through experience; and the diagnostic result annotation is added manually;

[0128] Among them, the historical data of nervous system imaging include normal images and images of different types of nervous system diseases;

[0129] An analysis and evaluation unit, used for performing diagnostic analysis on each pixel domain and its pixel domain feature data corresponding to the nervous system image to be analyzed and the sample pixel domain and sample feature data in the sample database;

[0130] The specific method is as follows:

[0131] Step D1, select a pixel domain and its pixel domain feature data corresponding to a nervous system image to be analyzed, and a sample pixel domain and its corresponding sample feature data as an example;

[0132] Then, the shape factor, area, and perimeter in the pixel domain feature data corresponding to the nervous system image to be analyzed are obtained and marked as X0, S0, and C0 respectively;

[0133] Step D2, mark the shape factor, area, and perimeter of the sample feature data corresponding to a sample pixel domain as X y , S y and C y ;

[0134] Step D3, then pass

[0135] K=1 / (|X0-X y |×β1+|S0-S y |×β2+|C0-C y |×β3)

[0136] Calculate the similarity K between the pixel domain corresponding to the nervous system image to be analyzed and the sample pixel domain;

[0137] In the formula, β1, β2, and β3 are the corresponding preset weight coefficients respectively;

[0138] Step D4, according to the method of Step D1 to Step D3, calculate the similarity between the pixel domain corresponding to the nervous system image to be analyzed and the pixel domains of all samples;

[0139] Step D5, compare each obtained similarity with a preset similarity threshold, and according to the comparison result, obtain the corresponding sample pixel domain when the similarity is greater than the similarity threshold, and then obtain a sample pixel domain with the largest similarity value, and then obtain the diagnosis result annotation of the sample pixel domain, and use the diagnosis result annotation of the sample pixel domain as the diagnosis result of the pixel domain corresponding to the nervous system image to be analyzed;

[0140] This embodiment can realize automatic analysis and diagnosis of nervous system images by acquiring nervous system images, performing grayscale conversion, normalization, segmentation, feature extraction, and diagnostic analysis with a sample database. Accurately calculating feature data in the pixel domain and obtaining diagnostic results by comparing with the sample pixel domain can help improve the accuracy and efficiency of diagnosis and reduce human errors. At the same time, it provides a basic technical framework and analysis ideas for subsequent embodiments.

[0141] For example, in medical imaging diagnosis, it can quickly and accurately make preliminary judgments on neurological diseases, provide valuable references for doctors, and save diagnostic time.

[0142] Embodiment 2

[0143] As the second embodiment of the present invention, when the present application is implemented, compared with the first embodiment, the technical solution of the present embodiment is different from that of the first embodiment only in that:

[0144] In this embodiment, before Step D4 of the first embodiment, X0 and X y The absolute value of the difference between y The absolute value of the difference between S0 and C0 y The absolute value of the difference between them is used to screen out the sample pixel domain involved in the similarity calculation:

[0145] Select a sample pixel domain;

[0146] If |X0-X y |>XL,|S0-S y |>SL,|C0-C y |>If any of the items in CL is true, the sample pixel domain will not be extracted to participate in the similarity calculation

[0147] If |X0-X y |>XL,|S0-S y |>SL,|C0-C y|>If none of CL is true, then the sample pixel domain is extracted to participate in the similarity calculation;

[0148] Among them, XL, SL, and CL are the corresponding preset difference screening thresholds, which are used to determine whether the corresponding sample pixel domain participates in the similarity calculation;

[0149] Based on the first embodiment, this embodiment adds a step of screening out the sample pixel domain participating in the similarity calculation by the absolute value of the difference, which can reduce unnecessary calculation amount and improve the efficiency of diagnosis and analysis.

[0150] For example, when faced with a large number of sample pixel domains, it is able to quickly exclude samples with obvious differences and focus on calculating and analyzing samples that are more likely to be similar.

[0151] Embodiment 3

[0152] As the third embodiment of the present invention, when the present application is implemented, compared with the first and second embodiments, the technical solution of this embodiment is to combine the solutions of the first and second embodiments, and the difference between the technical solution of this embodiment and the first and second embodiments is only in this embodiment;

[0153] In this embodiment, in Step D4 of the first embodiment, when all similarities are less than or equal to the similarity threshold, it means that the sample pixel domain does not contain a similar sample pixel domain in the sample database, and then the diagnosis result is determined by relevant personnel based on experience;

[0154] At the same time, the sample pixel domain and the corresponding pixel domain feature data are stored in the sample database as the sample pixel domain and sample feature data;

[0155] This embodiment combines the advantages of the first and second embodiments, and when all similarities are less than or equal to the similarity threshold, the diagnosis result is determined by relevant personnel based on experience and stored in the sample database, which helps to enrich the sample database and improve the accuracy of subsequent diagnosis.

[0156] For example, for some special or rare neurological imaging conditions, it is possible to use experience to judge and supplement sample data, so that the system will have more references when facing similar situations in the future.

[0157] Embodiment 4

[0158] As the fourth embodiment of the present invention, when the present application is specifically implemented, compared with the first, second and third embodiments, the technical solution of this embodiment is to combine and implement the solutions of the above-mentioned first, second and third embodiments.

[0159] This embodiment combines the solutions of Embodiment 1, Embodiment 2 and Embodiment 3, and can give full play to the advantages of each step to achieve more comprehensive, accurate and efficient nervous system imaging analysis and diagnosis.

[0160] For example, in complex clinical diagnosis scenarios, various types of nervous system images can be processed more effectively, providing more reliable support for the diagnosis and treatment of diseases.

[0161] See also Figure 2 As shown, the present invention also provides a method for analyzing nervous system images, which is implemented by the nervous system image analysis system, and the method includes the following steps:

[0162] Step 1: Data Acquisition

[0163] Obtaining images of the nervous system to be analyzed;

[0164] Step 2: Image processing

[0165] The nervous system image is converted into a grayscale image, and then the grayscale value of each pixel on the grayscale image is normalized to obtain the normalized grayscale value of each pixel;

[0166] Step 3: Segmentation

[0167] According to the grayscale values ​​of all pixels corresponding to the nervous system image after normalization, the nervous system image is segmented, and the segmentation mark corresponding to each pixel is obtained, and then multiple pixel domains are obtained from the nervous system image according to the segmentation mark;

[0168] Step 4: Feature extraction

[0169] Perform feature extraction processing on all pixel domains in the nervous system image and obtain pixel domain feature data;

[0170] Step 5: Sample database preparation

[0171] A large amount of historical neurological imaging data containing diagnostic results is collected in advance, and all pixel domains and their corresponding pixel domain feature data are obtained from the historical neurological imaging data through image processing, segmentation processing, and feature extraction, and the obtained pixel domains and their corresponding pixel domain feature data are respectively sample pixel domains and sample feature data; at the same time, the diagnostic results obtained in advance from the historical neurological imaging data are used to add diagnostic result annotations to each pixel domain;

[0172] Step 6: Analysis and Evaluation

[0173] The pixel domains and their pixel domain feature data corresponding to the nervous system image to be analyzed are diagnosed and analyzed with the sample pixel domains and sample feature data in the sample database, and the diagnosis result of the pixel domain corresponding to the nervous system image to be analyzed is determined.

[0174] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0175] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A nervous system image analysis system, characterized in that: include: A segmentation processing unit, for performing segmentation processing on the nervous system image according to the grayscale values ​​of all pixels corresponding to the nervous system image, and obtaining a segmentation mark corresponding to each pixel, and then obtaining a plurality of pixel domains from the nervous system image according to the segmentation mark, wherein the pixel domain refers to an image block composed of pixels with the same segmentation mark; A feature extraction unit is used to perform feature extraction processing on all pixel domains in the nervous system image and obtain pixel domain feature data; A sample database, used to store pre-trained sample pixel domains and their corresponding sample feature data, wherein the sample pixel domains are annotated with diagnostic results; The analysis and evaluation unit is used to perform diagnostic analysis on each pixel domain and its pixel domain feature data corresponding to the nervous system image to be analyzed and the sample pixel domain and sample feature data in the sample database, and determine the diagnostic result of the pixel domain corresponding to the nervous system image to be analyzed.

2. A nervous system image analysis system according to claim 1, characterized in that: Also includes: A data acquisition unit, used for acquiring nervous system images to be analyzed; The image processing unit is used to convert the nervous system image into a grayscale image, and then perform image normalization on the grayscale value of each pixel on the grayscale image to obtain the grayscale value of each pixel after normalization.

3. A nervous system image analysis system according to claim 2, characterized in that: The image normalization process is as follows: StepA1: Get the grayscale value of each pixel in the grayscale image and mark the grayscale value of each pixel as G in the coordinate system. (i,j) , where (i, j) represents the pixel coordinates in the image, i represents the nth horizontal coordinate, j represents the nth vertical coordinate, i = 1, 2, ... n, j = 1, 2, ... m, n × m is the number of all pixels in the nervous system image; Wherein, i=1, 2, ... n, n represents the number of all pixels; StepA2: From the grayscale values ​​of all pixels, obtain the grayscale value with the largest value and the grayscale value with the smallest value, and mark them as G and min and G max ; Step A3, through H (i,j) =(G (i,j) -G min ) / (G max -G min ), and obtain the gray value H of each pixel after normalization (i,j) .

4. A nervous system image analysis system according to claim 3, characterized in that: The segmentation process is as follows: Step B1: Select a pixel from the nervous system image as a reference pixel, and obtain the normalized grayscale value H corresponding to the reference pixel. (i,j) ; At the same time, segmentation marks are added to the reference pixels; Among them, the segmentation mark is a specific mark assigned to pixels to distinguish different pixel domains during the neural system image segmentation process; Step B2, based on H (i,j) , obtain the grayscale values ​​H of multiple pixels adjacent to the reference pixel (i-1,j) , H (i+1,j) , H (i,j-1) , H (i,j+1) , H (i+1,j+1) , H (i+1,j-1) , H (i-1,j-1) , H (i-1,j+1) ; Step B3, calculate the gray value H corresponding to the reference pixel points respectively (i,j) The grayscale values ​​H of the adjacent pixels are (i-1,j) , H (i+1,j) , H (i,j-1) , H (i,j+1) , H (i+1,j+1) , H (i+1,j-1) , H (i-1,j-1) , H (i-1,j+1) The absolute value of the difference between them is then compared with the preset grayscale difference threshold Hy: Select H from (i,j) and H (i+1,j) ; If |H (i,j) -H (i+1,j) |>Hy, then a segmentation mark different from the reference pixel point is added to the pixel point adjacent to the reference pixel point; If |H (i,j) -H (i+1,j) |≤Hy, then the pixel point adjacent to the reference pixel point will be added with the same segmentation mark as the reference pixel point; Step B4, using each adjacent pixel point as a reference pixel point in the manner of Step B3, and obtaining the segmentation mark of each pixel point adjacent to the reference pixel point; And so on, until all pixels are marked with segmentation marks; Step B5, based on the segmentation mark, all the pixels with the same segmentation mark are grouped into a pixel domain, where the pixel domain refers to an image block composed of pixels with the same segmentation mark.

5. A nervous system image analysis system according to claim 4, characterized in that: In Step B4, the pixel points that have been added with the segmentation mark will not be added with the segmentation mark again. When the adjacent pixel points are used as the reference pixel points, the reference pixel points selected in the early stage will not be added with the segmentation mark.

6. A nervous system image analysis system according to claim 1, characterized in that: The pixel domain feature data includes: the area, perimeter and shape factor of the pixel domain; The feature extraction process is as follows: Select a pixel region; Step C1, obtaining the number of all pixels in the pixel domain, and then recording the number as the area of ​​the pixel domain; Step C2, obtaining the number of all pixel points on the boundary of the pixel domain, and then recording the number value as the perimeter of the pixel domain; The method for determining whether a pixel point in the pixel domain is on the boundary is as follows: Among multiple pixel points adjacent to a pixel point, if at least one of the adjacent pixel points is not in the pixel domain, the pixel point is determined to be on the boundary; If all adjacent pixels are in the pixel domain, it is determined that the pixel is not on the boundary; Step C3, calculate the shape factor of the pixel domain based on the area and perimeter of the pixel domain; Calculation formula: X = 4π × S / C2 X represents the shape factor of the pixel domain, S represents the area of ​​the pixel domain, and C represents the perimeter of the pixel domain.

7. A nervous system image analysis system according to claim 6, characterized in that: in, The pre-training method of the sample pixel domain and the sample feature data is to collect a large amount of historical neurological imaging data containing diagnostic results in advance, and obtain all pixel domains and their corresponding pixel domain feature data from the historical neurological imaging data through an image processing unit, a segmentation processing unit, and a feature extraction unit, and the obtained pixel domains and their corresponding pixel domain feature data are respectively the sample pixel domain and the sample feature data; At the same time, the diagnostic result obtained in advance through the historical data of nervous system imaging is used to add diagnostic result annotations to each pixel domain; wherein, the diagnostic result annotation is added manually; Among them, the historical data of nervous system imaging include normal images and images of different types of nervous system diseases.

8. A nervous system image analysis system according to claim 7, characterized in that: The diagnostic analysis is as follows: Step D1, select a pixel domain and its pixel domain feature data corresponding to a nervous system image to be analyzed, and a sample pixel domain and its corresponding sample feature data as an example; Then, the shape factor, area, and perimeter in the pixel domain feature data corresponding to the nervous system image to be analyzed are obtained and marked as X0, S0, and C0 respectively; Step D2, mark the shape factor, area, and perimeter of the sample feature data corresponding to a sample pixel domain as X y , S y and C y ; Step D3, then pass K=1 / (|X0-X y |×β1+|S0-S y |×β2+|C0-C y |×β3) Calculate the similarity K between the pixel domain corresponding to the nervous system image to be analyzed and the sample pixel domain; In the formula, β1, β2, and β3 are the corresponding preset weight coefficients respectively; Step D4, through X0 and X y The absolute value of the difference between y The absolute value of the difference between S0 and C0, y The absolute value of the difference between them is used to screen out the sample pixel domain involved in the similarity calculation: Select a sample pixel domain; If |X0-X y |>XL,|S0-S y |>SL,|C0-C y |>If any of the items in CL is true, the sample pixel domain will not be extracted to participate in the similarity calculation If |X0-X y |>XL,|S0-S y |>SL,|C0-C y |>If none of CL is true, then the sample pixel domain is extracted to participate in the similarity calculation; Among them, XL, SL, and CL are the corresponding preset difference screening thresholds; Then, according to Step D1 to Step D3, the similarity between the pixel domain corresponding to the nervous system image to be analyzed and the corresponding pixel domains of all samples is calculated; Step D5. Compare the obtained similarities with the preset similarity threshold, and based on the comparison result, obtain the corresponding sample pixel domain when the similarity is greater than the similarity threshold, and then obtain a sample pixel domain with the largest similarity value, and then obtain the diagnostic result annotation of the sample pixel domain, and use the diagnostic result annotation of the sample pixel domain as the diagnostic result of the pixel domain corresponding to the nervous system image to be analyzed.

9. A nervous system image analysis system according to claim 8, characterized in that: In Step D4, when all similarities are less than or equal to the similarity threshold, it means that the sample pixel domain does not contain a similar sample pixel domain in the sample database, and then the diagnosis result is determined by relevant personnel based on experience; At the same time, the sample pixel domain and its corresponding pixel domain feature data are stored in the sample database as sample pixel domain and sample feature data.

10. A method for analyzing nervous system images, characterized in that: The method is implemented by a nervous system image analysis system as described in any one of claims 1-9.

Images