Image segmentation method and system based on medical risk

By denoising and enhancing medical images, combined with Hessian matrix extreme point detection and undirected graph segmentation algorithm, the problem of isolated point segmentation in the image is solved, clearer image segmentation is achieved, and the reading of medical image data is improved. Efficiency and accuracy.

CN119006512AInactive Publication Date: 2024-11-22XIEHE HOSPITAL ATTACHED TO TONGJI MEDICAL COLLEGE HUAZHONG SCI & TECH UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411073978.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing medical image segmentation method is based on graph theory, which is easy to segment the set of isolated points in the image, affecting the quality of image segmentation.

Method used

Using the image segmentation method based on medical risk, after noise reduction and enhancement processing, Hessian matrix extreme point detection and undirected graph segmentation algorithm are used, combined with histogram equalization method, to overcome the problem of isolated point segmentation and achieve clearer image segmentation .

Benefits of technology

It significantly improves the accuracy and clarity of image segmentation, reduces the impact of irrelevant information, improves the reading efficiency and accuracy of medical image data, and reduces medical risks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119006512A_ABST
    Figure CN119006512A_ABST
Patent Text Reader

Abstract

The invention discloses a medical risk-based image segmentation method and system, and particularly relates to the technical field of image processing, and the method comprises the following steps: obtaining to-be-processed medical image data based on a medical imaging device; performing noise reduction and enhancement processing on the to-be-processed medical image; and segmenting the medical image after noise reduction and enhancement processing by using a segmentation algorithm. According to the method, the image is segmented by adopting the segmentation algorithm, so that the foreground and the background in the image can be separated more clearly, the segmentation effect is clear and obvious, the image part required by the treatment process is obtained, the time required for identifying other irrelevant information is shortened, wrong guidance caused by other information is reduced, and the treatment efficiency is improved. Therefore, the reading of the medical image data is more accurate and faster, and the medical risk caused by the misreading of the image information is reduced to a certain extent.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and more specifically, to a medical risk-based image segmentation method and system. Background Art

[0002] Medical image segmentation is an important step in medical image processing and analysis. Its ultimate goal is to use a series of features (such as image grayscale, texture, color, and local statistical features) to extract the "region of interest" in the image to assist doctors in diagnosis, treatment planning, and clinical research. However, factors such as noise and field offset effects during the imaging process make the obtained images blurred and uneven. In addition, the medical image itself has low contrast and blurred boundaries between tissues and lesions, which greatly affects the results of the segmentation algorithm.

[0003] Several advantages of medical image segmentation:

[0004] (1) Segment the organs, tissues, and lesions in the human body from medical images and accurately measure their size and distance, which will help doctors to accurately diagnose and analyze the symptoms, and provide technical support for patients to make efficient and accurate treatment plans.

[0005] (2) Providing materials for three-dimensional reconstruction of the human body, and then providing assistance to doctors during surgery, can greatly reduce surgical risks and reduce the burden on doctors during surgery.

[0006] (3) Medical image segmentation can also remove a large amount of useless information in the image and segment out only the required parts. This can reduce the amount of image data without compressing the image, thereby improving efficiency and medical level.

[0007] However, the existing medical image segmentation is based on graph theory. Although it has a more obvious global energy optimization, the normalized cut and minimum cut are studied based on it as the basic idea. Among them, the minimum cut has the problem of easily segmenting a set of isolated points in the image. Therefore, a medical risk-based image segmentation method and system are needed to solve the above problems. Summary of the invention

[0008] In order to overcome the above-mentioned defects, the present invention provides an image segmentation method and system based on medical risks. The technical problem to be solved by the present invention is: using the minimum cut method for medical-related images, it is easy to segment a set of isolated points in the image, which affects the quality of image segmentation.

[0009] To achieve the above object, the present invention provides the following technical solution: an image segmentation method based on medical risk, comprising the following steps:

[0010] Obtaining medical image data to be processed based on a medical imaging device;

[0011] Perform noise reduction and enhancement processing on the medical images to be processed;

[0012] The medical images after denoising and enhancement are segmented using a segmentation algorithm.

[0013] As a further solution of the present invention: the image enhancement method is a histogram equalization method;

[0014] The image noise reduction process adopts the following method:

[0015] First, a digital image acquisition model is constructed, and the sensor is imaged using the three-angle uniform traversal method. Assuming that a point P(x, y) in the digital image imaging area F is a pixel point, the Hessian matrix of the digital image on the scale σ can be defined as:

[0016] Perform M-1 iterations and take the critical point P(x,y) of the Hessian matrix of the noisy digital image. We have:

[0017] Extract the multivariate feature data set from the noisy digital image at two moments and extract the initial network pattern information. At this time, the Hessian matrix is ​​transformed into:

[0018]

[0019] After simplification,

[0020] In the feature subspace containing noisy information, the Hessian matrix extreme point detection method is used to extract the corner feature of the digital image, and the 9×9 median filter is used to uniformly traverse the noise points in the center of the digital image pixels. The digital images collected in a strong interference environment usually contain a lot of noise, which makes it difficult to perform effective detail analysis. In order to reduce the amount and complexity of calculations, digital image denoising filtering is performed, and the Hessian matrix is ​​constructed to enhance the details of the digital image. The Hessian matrix of the digital image enhancement processing is:

[0021] H=D xx D yy -(ωD yy ) 2

[0022] The original pixel feature extraction and information preprocessing of the digital image are performed, and the grayscale pixel feature c of the digital image is obtained as follows:

[0023]

[0024] Weight coefficient m for detail recognition of digital images j(k) is set to 0.9, and H′ can be calculated to obtain local performance points through adaptive eigendecomposition; D xx <0, then the grayscale pixel feature point P(x,y) of the digital image is a local maximum point.

[0025] As a further solution of the present invention: the image segmentation algorithm is specifically:

[0026] First, the image is expressed as an undirected graph G = (V, E), where V is the set of nodes, E is the set of edges connecting nodes, and the basis of V is N = |V|; the edge connecting each two nodes is given a weight w = (u, n), which measures the similarity between nodes u and v; the image G = (V, E) can be divided into two non-intersecting sets A and B by simply removing the edges connecting the two parts, and satisfying A∪B = V, A∩B = Φ. The dissimilarity of the two parts is calculated by the sum of the weights of the removed edges, defined as,

[0027] Determine the normalized segmentation criterion: Normalized segmentation adds the ratio of the nodes in each region to the sum of the weights of all nodes on the basis of the minimum cut to balance the segmentation. The formula for normalized segmentation is expressed as:

[0028]

[0029] where assoc(A,V)=∑ u∈A,t∈V w(u,t) is the sum of the weights of the edges from all nodes in A to all nodes in the image, and assoc(B,V) is the sum of the weights of the edges from all nodes in B to all nodes in the image. In the case of segmenting isolated points, the value in the normalized segmentation is relatively large:

[0030]

[0031] Among them, assoc(A,A) and assoc(B,B) represent the sum of the weights of the edges connecting the nodes in A and B respectively. The above formula reflects the balance of the definition and represents the association degree of the nodes in each group;

[0032] The Ncut criterion can overcome the problem of dividing isolated points. The division corresponding to the minimum Ncut value is the optimal division of graph G. Minimizing Ncut can be transformed into the following standard feature system:

[0033]

[0034] Where D is a diagonal matrix of NXN, and its diagonal elements are (i, j), W is a symmetric matrix whose elements are w(i, j), λ and z are the corresponding eigenvalues ​​and eigenvectors;

[0035] The eigenvector corresponding to the second minimum eigenvalue of the eigensystem is used to complete the optimal division of the entire image and obtain a segmentation result of the corresponding image.

[0036] A medical risk-based image segmentation system, comprising:

[0037] A data acquisition unit, used for acquiring medical image data to be processed based on a medical imaging device;

[0038] An image preprocessing unit, used for performing noise reduction and enhancement processing on the medical image to be processed;

[0039] The image segmentation unit is used to segment the medical image after noise reduction and enhancement processing using a segmentation algorithm to obtain a final target image after segmentation.

[0040] As a further solution of the present invention: the image enhancement method is a histogram equalization method.

[0041] An electronic device comprises a processor and a memory storing a computer program, wherein when the processor executes the computer program, the medical risk-based image segmentation method according to any one of claims 1 to 7 is implemented.

[0042] A non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the medical risk-based image segmentation method as claimed in any one of claims 1 to 7.

[0043] The purpose of image segmentation is to extract the target from the background. The segmentation process is mainly based on the inherent characteristics of the image, such as grayscale, texture, contrast brightness, color features, etc., to divide the image into homogeneous regions with their own characteristics. Medical image segmentation is a necessary prerequisite for subsequent operations on medical images. For medical images, the image itself contains a lot of complex information, and clinical diagnosis is often only interested in a certain part of it, and the other parts are secondary information. How to extract these interesting and meaningful areas from medical images is the research content of medical image segmentation.

[0044] The beneficial effects of the present invention are:

[0045] 1. The present invention uses a segmentation algorithm to segment the image, which can more clearly separate the foreground and background in the image. The segmentation effect is clear and obvious, and the image part required for the treatment process is obtained, which reduces the time required to identify other irrelevant information and reduces the misleading caused by other information, making the reading of medical image data more accurate and quick, and reducing the medical risks caused by misreading image information to a certain extent;

[0046] 2. The present invention adopts denoising processing on the image. The denoised image retains the detailed information of the original image and has a clearer visual effect. The histogram equalization method is subsequently used to enhance the image. The equalized image makes the boundaries between different parts of the image clearer and more obvious in terms of overall visual effect, which facilitates a clearer and more accurate understanding of the information displayed in the medical image. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 Schematic diagram of the connection of the system of the present invention. DETAILED DESCRIPTION

[0048] 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.

[0049] Embodiment 1:

[0050] As shown in the figure, the present invention provides an image segmentation method based on medical risk, comprising the following steps:

[0051] Obtaining medical image data to be processed based on a medical imaging device;

[0052] Perform noise reduction and enhancement processing on the medical images to be processed;

[0053] The medical images after denoising and enhancement are segmented using a segmentation algorithm.

[0054] The image enhancement method is the histogram equalization method;

[0055] The specific image segmentation algorithm is:

[0056] First, the image is expressed as an undirected graph G = (V, E), where V is the set of nodes, E is the set of edges connecting nodes, and the basis of V is N = |V|; the edge connecting each two nodes is given a weight w = (u, n), which measures the similarity between nodes u and v; the image G = (V, E) can be divided into two non-intersecting sets A and B by simply removing the edges connecting the two parts, and satisfying A∪B = V, A∩B = Φ. The dissimilarity of the two parts is calculated by the sum of the weights of the removed edges, defined as,

[0057] Determine the normalized segmentation criterion: Normalized segmentation adds the ratio of the nodes in each region to the sum of the weights of all nodes on the basis of the minimum cut to balance the segmentation. The formula for normalized segmentation is expressed as:

[0058]

[0059] where assoc(A,V)=∑ u∈A,t∈V w(u,t) is the sum of the weights of the edges from all nodes in A to all nodes in the image, and assoc(B,V) is the sum of the weights of the edges from all nodes in B to all nodes in the image. In the case of segmenting isolated points, the value in the normalized segmentation is relatively large:

[0060]

[0061] Among them, assoc(A,A) and assoc(B,B) represent the sum of the weights of the edges connecting the nodes in A and B respectively. The above formula reflects the balance of the definition and represents the association degree of the nodes in each group;

[0062] The Ncut criterion can overcome the problem of dividing isolated points. The division corresponding to the minimum Ncut value is the optimal division of graph G. Minimizing Ncut can be transformed into the following standard feature system:

[0063]

[0064] Where D is a diagonal matrix of NXN, and its diagonal elements are (i, j), W is a symmetric matrix whose elements are w(i, j), λ and z are the corresponding eigenvalues ​​and eigenvectors;

[0065] The eigenvector corresponding to the second minimum eigenvalue of the eigensystem is used to complete the optimal division of the entire image and obtain a segmentation result of the corresponding image.

[0066] A medical risk-based image segmentation system, comprising:

[0067] A data acquisition unit, used for acquiring medical image data to be processed based on a medical imaging device;

[0068] An image preprocessing unit, used for performing noise reduction and enhancement processing on the medical image to be processed;

[0069] The image segmentation unit is used to segment the medical image after noise reduction and enhancement processing using a segmentation algorithm to obtain a final target image after segmentation.

[0070] The image enhancement method is histogram equalization method.

[0071] An electronic device comprises a processor and a memory storing a computer program, wherein when the processor executes the computer program, the medical risk-based image segmentation method according to any one of claims 1 to 7 is implemented.

[0072] A non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the medical risk-based image segmentation method as claimed in any one of claims 1 to 7.

[0073] Embodiment 2:

[0074] As shown in the figure, the present invention provides an image segmentation method based on medical risk, comprising the following steps:

[0075] Obtaining medical image data to be processed based on a medical imaging device;

[0076] Perform noise reduction and enhancement processing on the medical images to be processed;

[0077] The medical images after denoising and enhancement are segmented using a segmentation algorithm.

[0078] The image enhancement method is the histogram equalization method;

[0079] Image noise reduction is done using the following methods:

[0080] First, a digital image acquisition model is constructed, and the sensor is imaged using the three-angle uniform traversal method. Assuming that a point P(x, y) in the digital image imaging area F is a pixel point, the Hessian matrix of the digital image on the scale σ can be defined as:

[0081] Perform M-1 iterations and take the critical point P(x,y) of the Hessian matrix of the noisy digital image. We have:

[0082] Extract the multivariate feature data set from the noisy digital image at two moments and extract the initial network pattern information. At this time, the Hessian matrix is ​​transformed into:

[0083]

[0084] After simplification,

[0085] In the feature subspace containing noisy information, the Hessian matrix extreme point detection method is used to extract the corner feature of the digital image, and the 9×9 median filter is used to uniformly traverse the noise points in the center of the digital image pixels. The digital images collected in a strong interference environment usually contain a lot of noise, which makes it difficult to perform effective detail analysis. In order to reduce the amount and complexity of calculations, digital image denoising filtering is performed, and the Hessian matrix is ​​constructed to enhance the details of the digital image. The Hessian matrix of the digital image enhancement processing is:

[0086] H=D xx Dyy -(ωD yy ) 2

[0087] The original pixel feature extraction and information preprocessing of the digital image are performed, and the grayscale pixel feature c of the digital image is obtained as follows:

[0088]

[0089] Weight coefficient m for detail recognition of digital images j (k) is set to 0.9, and H′ can be calculated to obtain local performance points through adaptive eigendecomposition; D xx <0, then the grayscale pixel feature point P(x,y) of the digital image is a local maximum point.

[0090] The specific image segmentation algorithm is:

[0091] First, the image is expressed as an undirected graph G = (V, E), where V is the set of nodes, E is the set of edges connecting nodes, and the basis of V is N = |V|; the edge connecting each two nodes is given a weight w = (u, n), which measures the similarity between nodes u and v; the image G = (V, E) can be divided into two non-intersecting sets A and B by simply removing the edges connecting the two parts, and satisfying A∪B = V, A∩B = Φ. The dissimilarity of the two parts is calculated by the sum of the weights of the removed edges, defined as,

[0092] Determine the normalized segmentation criterion: Normalized segmentation adds the ratio of the nodes in each region to the sum of the weights of all nodes on the basis of the minimum cut to balance the segmentation. The formula for normalized segmentation is expressed as:

[0093]

[0094] where assoc(A,V)=∑ u∈A ,t∈Vw(u,t) is the sum of the weights of the edges from all nodes in A to all nodes in the image, and assoc(B,V) is the sum of the weights of the edges from all nodes in B to all nodes in the image; in the case of segmenting isolated points, the value in the normalized segmentation is relatively large:

[0095]

[0096] Among them, assoc(A,A) and assoc(B,B) represent the sum of the weights of the edges connecting the nodes in A and B respectively. The above formula reflects the balance of the definition and represents the association degree of the nodes in each group;

[0097] The Ncut criterion can overcome the problem of dividing isolated points. The division corresponding to the minimum Ncut value is the optimal division of graph G. Minimizing Ncut can be transformed into the following standard feature system:

[0098]

[0099] Where D is a diagonal matrix of NXN, and its diagonal elements are (i, j), W is a symmetric matrix whose elements are w(i, j), λ and z are the corresponding eigenvalues ​​and eigenvectors;

[0100] The eigenvector corresponding to the second minimum eigenvalue of the eigensystem is used to complete the optimal division of the entire image and obtain a segmentation result of the corresponding image.

[0101] A medical risk-based image segmentation system, comprising:

[0102] A data acquisition unit, used for acquiring medical image data to be processed based on a medical imaging device;

[0103] An image preprocessing unit, used for performing noise reduction and enhancement processing on the medical image to be processed;

[0104] The image segmentation unit is used to segment the medical image after noise reduction and enhancement processing using a segmentation algorithm to obtain a final target image after segmentation.

[0105] The image enhancement method is histogram equalization method.

[0106] An electronic device comprises a processor and a memory storing a computer program, wherein when the processor executes the computer program, the medical risk-based image segmentation method according to any one of claims 1 to 7 is implemented.

[0107] A non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the medical risk-based image segmentation method as claimed in any one of claims 1 to 7.

[0108] Secondly: In the drawings of the embodiments disclosed in the present invention, only the structures related to the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other;

[0109] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A medical risk-based image segmentation method, characterized in that: The following steps are involved: Obtaining medical image data to be processed based on a medical imaging device; Perform noise reduction and enhancement processing on the medical images to be processed; The medical images after denoising and enhancement are segmented using a segmentation algorithm.

2. The medical risk-based image segmentation method according to claim 1, characterized in that: The image enhancement method is a histogram equalization method; The image noise reduction process adopts the following method: First, a digital image acquisition model is constructed, and the sensor is imaged using the three-angle uniform traversal method. Assuming that a point P(x, y) in the digital image imaging area F is a pixel point, the Hessian matrix of the digital image on the scale σ can be defined as: Perform M-1 iterations and take the critical point P(x,y) of the Hessian matrix of the noisy digital image. We have: Extract the multivariate feature data set from the noisy digital image at two moments and extract the initial network pattern information. At this time, the Hessian matrix is ​​transformed into: After simplification, In the feature subspace containing noisy information, the Hessian matrix extreme point detection method is used to extract the corner feature of the digital image, and the 9×9 median filter is used to uniformly traverse the noise points in the center of the digital image pixels. The digital images collected in a strong interference environment usually contain a lot of noise, which makes it difficult to perform effective detail analysis. In order to reduce the amount and complexity of calculations, digital image denoising filtering is performed, and the Hessian matrix is ​​constructed to enhance the details of the digital image. The Hessian matrix of the digital image enhancement processing is: H=D xx D yy -(ωD yy ) 2 The original pixel feature extraction and information preprocessing of the digital image are performed, and the grayscale pixel feature c of the digital image is obtained as follows: Weight coefficient m for detail recognition of digital images j (k) is set to 0.9, and H′ can be calculated to obtain local performance points through adaptive eigendecomposition; D xx <0, then the grayscale pixel feature point P(x,y) of the digital image is a local maximum point.

3. The medical risk-based image segmentation method according to claim 1, characterized in that: The image segmentation algorithm is specifically: First, the image is expressed as an undirected graph G = (V, E), where V is the set of nodes, E is the set of edges connecting nodes, and the basis of V is N = |V|; the edge connecting each two nodes is given a weight w = (u, n), which measures the similarity between nodes u and v; the image G = (V, E) can be divided into two non-intersecting sets A and B by simply removing the edges connecting the two parts, and satisfying A∪B = V, A∩B = Φ. The dissimilarity of the two parts is calculated by the sum of the weights of the removed edges, defined as, Determine the normalized segmentation criterion: Normalized segmentation adds the ratio of the nodes in each region to the sum of the weights of all nodes on the basis of the minimum cut to balance the segmentation. The formula for normalized segmentation is expressed as: where assoc(A,V)=∑ u∈A,t∈V w(u,t) is the sum of the weights of the edges from all nodes in A to all nodes in the image, and assoc(B,V) is the sum of the weights of the edges from all nodes in B to all nodes in the image. In the case of segmenting isolated points, the value in the normalized segmentation is relatively large: Among them, assoc(A,A) and assoc(B,B) represent the sum of the weights of the edges connecting the nodes in A and B respectively. The above formula reflects the balance of the definition and represents the association degree of the nodes in each group; The Ncut criterion can overcome the problem of dividing isolated points. The division corresponding to the minimum Ncut value is the optimal division of graph G. Minimizing Ncut can be transformed into the following standard feature system: Where D is a diagonal matrix of NXN, and its diagonal elements are (i, j), W is a symmetric matrix whose elements are w(i, j), λ and z are the corresponding eigenvalues ​​and eigenvectors; The eigenvector corresponding to the second minimum eigenvalue of the eigensystem is used to complete the optimal division of the entire image and obtain a segmentation result of the corresponding image.

4. A medical risk-based image segmentation system, characterized in that: include: A data acquisition unit, used for acquiring medical image data to be processed based on a medical imaging device; An image preprocessing unit, used for performing noise reduction and enhancement processing on the medical image to be processed; The image segmentation unit is used to segment the medical image after noise reduction and enhancement processing using a segmentation algorithm to obtain a final target image after segmentation.

5. The medical risk-based image segmentation method and system according to claim 4, characterized in that: The image enhancement method is a histogram equalization method.

6. An electronic device comprising a processor and a memory storing a computer program, characterized in that: When the processor executes the computer program, the medical risk-based image segmentation method according to any one of claims 1 to 7 is implemented.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the medical risk-based image segmentation method according to any one of claims 1 to 7 is implemented.