A quality control detection method for CT images of arms raised combined with multi-view fusion

Abstract

The present invention discloses a double-arm-raised CT image quality control detection method combined with multi-view fusion, which performs technical quality control by performing CT scanning on the double-arm-raised posture from multiple views, fuses the multi-view images by using advanced image processing technology and algorithms, obtains more comprehensive quantitative evaluation of image information, and uses simple ellipse fitting to calculate the relationship between the normal vector of the ellipsoid and the humeral posture to determine the positional relationship according to the different body postures of patients, calculates the angle by the normal vector of the chest body ellipsoid plane and the direction vector of the humerus of both arms, and quantifies the angle into a quality control score to quantitatively evaluate the quality control effect, thereby further improving the accuracy and reliability of CT examinations.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and in particular to a double-arm-raised CT image quality control detection method combined with multi-view fusion. Background Art

[0002] CT technology is a non-invasive medical imaging method that uses X-rays to perform tomography on the human body and reconstruct a three-dimensional image of the human body through a computer. CT technology has the advantages of high precision, high resolution, non-destructiveness and rapid detection, and plays an important role in medical diagnosis. CT quality control testing is an important means to ensure the performance and image quality of CT equipment. By regularly performing quality control testing on CT equipment, potential problems with the equipment can be discovered and resolved in a timely manner, ensuring the accuracy and reliability of CT examinations. This is of great significance for improving medical quality and ensuring patient safety.

[0003] The arms-up posture is often used in chest examination quality control requirements during CT examinations. This posture helps reduce the obstruction of the examination site by the arms, allowing doctors to observe the structure and lesions of the target area more clearly. At the same time, the arms-up posture also requires the patient to maintain a certain degree of stability and cooperation to ensure the smooth progress of the examination process.

[0004] According to data, existing technologies obtain image data by performing CT scans on the target area. The quality control of the image is subjectively controlled by the radiologist. Image quality control technology is used to evaluate the performance and image quality of CT equipment to ensure the accuracy and reliability of the inspection results. Common quality control inspection indicators include spatial resolution, noise, CT value uniformity, etc. These indicators can reflect the overall performance and image quality of CT equipment. Although modern CT equipment has adopted a variety of technologies to reduce radiation dose, the multi-view CT scanning and three-dimensional reconstruction process are relatively complex and require professional technicians to operate and maintain. In addition, high-quality CT equipment and image processing software are usually expensive, which increases the operating costs of medical institutions.

[0005] Chinese patent document CN111402231A discloses an "automatic evaluation system and method for lung CT image quality". This automatic evaluation system can obtain artifacts in CT images with high accuracy and efficiency, and evaluate CT images objectively and impartially, which is convenient for hospitals and physical examination institutions to use and expand CT image quality control, and it can also creatively realize CT image posture discrimination, assisting image quality inspectors to judge whether protective detection is in place. The above technical solution lacks quantitative evaluation of quality control effects based on different patient postures. Summary of the invention

[0006] The present invention mainly solves the technical problem that the original technical solution lacks a quantitative evaluation of the quality control effect for different patient body postures, and provides a double-arm raised CT image quality control detection method combined with multi-view fusion, which performs technical quality control on the double-arm raised posture by CT scanning from multiple views, and uses advanced image processing technology and algorithms to fuse the multi-view images to obtain a more comprehensive quantitative evaluation of the image information, and uses a simple ellipse fitting to calculate the relationship between the normal vector of the ellipsoid and the humeral posture to determine the positional relationship for different patient body postures, calculates the angle through the normal vector of the chest body ellipsoid plane and the direction vector of the humerus of both arms, and quantifies the angle into a quality control score to quantitatively evaluate the quality control effect, thereby further improving the accuracy and reliability of CT examinations.

[0007] The above technical problem of the present invention is mainly solved by the following technical solution: The present invention comprises the following steps:

[0008] S1 inputs CT sequence images and selects chest quality control images for preprocessing;

[0009] S2 segments the preprocessed image to obtain a cross-sectional view;

[0010] S3 performs ellipse fitting on the rib pixels in the section view and straight line fitting on the humerus pixels in the section view to obtain the normal plane direction of the trunk and the direction of the arm humerus;

[0011] S4 calculates the humeral angle of both arms and obtains the quality control score of both arms posture, and decides the retention of CT sequence images based on the quality control score.

[0012] Technical quality control is performed by performing CT scanning on the raised arms posture from multiple views, and multi-view images are fused using advanced image processing technology and algorithms to obtain a more comprehensive quantitative evaluation of image information. In view of the different body postures of patients, a simple ellipse fitting is used to calculate the relationship between the normal vector of the ellipsoid and the humeral posture to determine the positional relationship. The angle is calculated through the normal vector of the chest body ellipsoid plane and the direction vector of the humerus of both arms, and the angle is quantified into a quality control score to quantitatively evaluate the quality control effect, thereby further improving the accuracy and reliability of CT examinations.

[0013] Preferably, the preprocessing in step S1 includes setting a window frame and a window level of the CT bone value, and converting the CT values ​​of the ribs and humerus of the trunk into voxel points of the chest rib image.

[0014] Preferably, the step S2 specifically includes inputting the number of divisions, obtaining the angle of the section view, and rotating the view section to map voxels to form view content.

[0015] Preferably, the voxel is Indicates, where X, Y, and Z represent three axes respectively.

[0016] Preferably, the step S2 further comprises: and Rotate the voxels of the Cube to form the view content, and finally get the cross-sectional view :

[0017]

[0018]

[0019]

[0020]

[0021] Where n and m are the pixels mapped from x and y from the viewing angle θ, is the minimum coordinate where n pixels are not empty, is the maximum coordinate of n pixels that are not empty, h and v are pixels in ZOθ view n, H and V are the new coordinate axes in ZOθ view, A symbol indicating whether the pixel is empty.

[0022] Preferably, the step S3 of performing ellipse fitting on the rib pixels in the cross-sectional view specifically comprises:

[0023]

[0024] Let F(A,H,V)=0,

[0025] Among them, A is the parameter (a, b, c, d, e, f); H, V are ; F(A,H,V) is the standard formula for ellipse fitting. The least squares method is used to fit F(A,H,V), and the extreme value derivative is used to obtain the coordinates of the two end points of the major axis of the ellipse;

[0026] Ellipse curvature formula

[0027]

[0028] The curvature at the pole is the largest, argmax(K,v,h), and we get , , get the vector And continue normalizing, we get Normal vector of the view .

[0029] Preferably, performing straight line fitting on the humerus pixels in the section view specifically comprises:

[0030] F(A,H,V)=ah+bv+c

[0031] Where A is the parameter (a, b, c), HV is (h, v, 1), and F(A, H, V) is the standard formula for straight line fitting;

[0032] Let F(A,H,V)=0, use the least squares method to fit F(A,H,V), and the direction vector , normalized to get .

[0033] Preferably, the step S3 further comprises: using the ellipse vector obtained in the multiple views The direction in the resultant space is taken as the normal vector of the minor axis plane of the ellipsoid:

[0034]

[0035] Among them, x, y, z are The coordinate of , h is the ordinate in the H plane;

[0036] Use the line direction vector obtained from multiple views As a direction in the composition space

[0037]

[0038] Among them, x, y, z are The coordinate of , h is the ordinate in the H plane;

[0039] As the direction vector of the humerus, the direction vectors of the humerus of the left and right arms are , .

[0040] Preferably, the step S4 of calculating the humeral angle of both arms specifically includes:

[0041] Normal plane direction to the torso , arm humerus direction vector , Calculate the space angle. The calculation formula for the space vector angle is , calculated , where the left arm , right arm humeral angle .

[0042] Preferably, the step S4 of calculating the quality control score specifically includes:

[0043]

[0044]

[0045]

[0046]

[0047] Among them, S_final is the final quality control score, S_left and S_right are the quality control scores of the left and right hand postures of both arms. When S=1, it is considered to be completely consistent with the posture of raising both arms during chest CT. The closer the quality control score is to 1, the better the quality control score is, and the closer the quality control score is to 0, the worse the quality control score is, and the less it meets the quality control standards.

[0048] The beneficial effects of the present invention are as follows: technical quality control is performed by performing CT scanning on the raised posture of both arms from multiple views, and multi-view images are fused using advanced image processing technology and algorithms to obtain a more comprehensive quantitative evaluation of image information. In view of the different body postures of patients, a simple ellipse fitting is used to calculate the relationship between the normal vector of the ellipsoid and the humeral posture to determine the positional relationship. The angle is calculated by the normal vector of the thoracic body ellipsoid plane and the direction vector of the humerus of both arms, and the angle is quantified into a quality control score to quantitatively evaluate the quality control effect, thereby further improving the accuracy and reliability of CT examinations. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flow chart of the present invention.

[0050] Figure 2 It is a schematic diagram of multiple views of CT images of the present invention.

[0051] Figure 3 This is a 0° view fitting effect diagram of an embodiment of the present invention.

[0052] Figure 4 This is a 45° view fitting effect diagram of an embodiment of the present invention.

[0053] Figure 5 This is a 90° view fitting effect diagram of an embodiment of the present invention.

[0054] Figure 6 This is a 135° view fitting effect diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the technical solution of the present invention is further described in detail below through embodiments and in combination with the accompanying drawings. It should be understood that the specific implementation method described here is only an optimal embodiment of the present invention, which is only used to explain the present invention and does not limit the scope of protection of the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0056] CT technology is a non-invasive medical imaging method that uses X-rays to perform tomography on the human body and reconstruct a three-dimensional image of the human body through a computer. CT technology has the advantages of high precision, high resolution, non-destructiveness and rapid detection, and plays an important role in medical diagnosis. CT quality control testing is an important means to ensure the performance and image quality of CT equipment. By regularly performing quality control testing on CT equipment, potential problems with the equipment can be discovered and resolved in a timely manner, ensuring the accuracy and reliability of CT examinations. This is of great significance for improving medical quality and ensuring patient safety.

[0057] The arms-up posture is often used in chest examination quality control requirements during CT examinations. This posture helps reduce the obstruction of the examination site by the arms, allowing doctors to observe the structure and lesions of the target area more clearly. At the same time, the arms-up posture also requires the patient to maintain a certain degree of stability and cooperation to ensure the smooth progress of the examination process.

[0058] Under normal circumstances, radiologists complete the image quality control of CT images through image observation and subjective evaluation. In order to further improve the accuracy and reliability of CT examinations, a multi-view CT quality control detection method for arms-up posture has emerged. The main idea of ​​this patent is to perform technical quality control on CT scanning of arms-up posture from multiple views, use advanced image processing technology and algorithms to fuse multi-view images, obtain more comprehensive quantitative evaluation of image information, and provide imaging technology quality control in the form of quality control scoring.

[0059] Existing technology obtains image data by performing CT scanning on the target area. The quality control of the image is subjectively controlled by the radiologist. Image quality control technology is used to evaluate the performance and image quality of CT equipment to ensure the accuracy and reliability of the examination results. Common quality control detection indicators include spatial resolution, noise, CT value uniformity, etc. These indicators can reflect the overall performance and image quality of CT equipment. Although modern CT equipment has adopted a variety of technologies to reduce radiation dose, the multi-view CT scanning and three-dimensional reconstruction process are relatively complex and require professional technicians to operate and maintain. In addition, high-quality CT equipment and image processing software are usually expensive, which increases the operating costs of medical institutions.

[0060] At present, the standardization and normalization of multi-view CT scanning and three-dimensional reconstruction are not high enough. When evaluating CT images, radiologists often rely on their own experience and subjective judgment. This subjectivity may lead to differences in the evaluation of the same image by different doctors, thus affecting the accuracy and consistency of quality control. It is a time-consuming task to observe and evaluate a large number of CT images one by one. This not only increases the workload of doctors, but also may affect the timeliness and efficiency of quality control. Radiologists need to rely on their own professional knowledge and experience to judge the image quality during the quality control process. However, there are differences in experience and level among different doctors, which may lead to inconsistent quality control results. Due to the large number of CT images, it may be difficult for doctors to fully cover all images during the quality control process, resulting in the omission of some potential problems. In the subjective evaluation process, doctors may pay more attention to the overall quality of the image and ignore some details and minor changes. These details and changes may be of great significance for the diagnosis and treatment of diseases.

[0061] To summarize, based on the shortcomings of existing manual quality control of CT images, this patent proposes the use of CT imaging technology to propose a double-arm-raised CT image quality control detection method combined with multi-view fusion. On the basis of combining the advantages of multi-view section image information processing and the quality control requirements of radiologists, it aims to further quantitatively detect the quality of CT images and the realization of quality control requirements through image processing and quality control posture computational model algorithms.

[0062] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the operations (or steps) as sequential processes, many of the operations (or steps) therein can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but can also have additional steps not included in the drawings; the process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0063] The technical solution of the present invention is further specifically described below through embodiments and in conjunction with the accompanying drawings.

[0064] Embodiment: This embodiment is a method for quality control detection of CT images of arms raised combined with multi-view fusion, such as Figure 1 As shown, the following steps are included:

[0065] S1 inputs CT sequence images and selects chest quality control images for preprocessing. The preprocessing includes setting the window frame and window level of CT bone values, and converting the CT values ​​of the ribs and humerus of the trunk into voxel points of the chest rib images.

[0066] S2 divides the preprocessed image to obtain a cross-sectional view, specifically including inputting the number of divisions, obtaining the angle of the cross-sectional view, and rotating the view cross-sectional view to map the voxels to form the view content. Indicates, where X, Y, and Z represent three axes respectively.

[0067] Viewing angle and Rotate the voxels of the Cube to form the view content, and finally get the cross-sectional view :

[0068]

[0069]

[0070]

[0071] }

[0072] Where n and m are the pixels mapped from x and y from the viewing angle θ, is the minimum coordinate where n pixels are not empty, is the maximum coordinate of n pixels that are not empty, h and v are pixels in ZOθ view n, H and V are the new coordinate axes in ZOθ view, A symbol indicating whether the pixel is empty.

[0073] S3 performs ellipse fitting on rib pixels in the cross-sectional view, specifically including:

[0074]

[0075] Let F(A,H,V)=0,

[0076] Among them, A is the parameter (a, b, c, d, e, f); H, V are ; F(A,H,V) is the standard formula for ellipse fitting. The least squares method is used to fit F(A,H,V), and the extreme value derivative is used to obtain the coordinates of the two end points of the major axis of the ellipse;

[0077] Ellipse curvature formula

[0078]

[0079] The curvature at the pole is the largest, argmax(K,v,h), and we get , , we get the vector And continue normalizing, we get Normal vector of the view .

[0080] Fit a straight line to the humerus pixels in the cut view, including:

[0081] F(A,H,V)=ah+bv+c

[0082] Where A is the parameter (a, b, c), HV is (h, v, 1), and F(A, H, V) is the standard formula for straight line fitting;

[0083] Let F(A,H,V)=0, use the least squares method to fit F(A,H,V), and the direction vector , normalized to get .

[0084] Obtain the normal plane direction of the torso and the humerus direction of the arm, including: using the ellipse vector obtained from multiple views The direction in the resultant space is taken as the normal vector of the minor axis plane of the ellipsoid:

[0085]

[0086] Among them, x, y, z are The coordinate of , h is the ordinate in the H plane;

[0087] Use the line direction vector obtained from multiple views As a direction in the composition space

[0088]

[0089] Among them, x, y, z are The coordinate of , h is the ordinate in the H plane;

[0090] As the direction vector of the humerus, the direction vectors of the humerus of the left and right arms are , .

[0091] S4 calculates the angle between the humerus of both arms, including:

[0092] Normal plane direction to the torso , arm humerus direction vector , Calculate the space angle. The calculation formula for the space vector angle is , calculated , where the left arm , right arm humeral angle .

[0093] Obtain the arm posture quality control score, and decide on the retention of CT sequence images based on the quality control score, including:

[0094]

[0095]

[0096]

[0097]

[0098] Among them, S_final is the final quality control score, S_left and S_right are the quality control scores of the left and right hand postures of both arms. When S=1, it is considered to be completely consistent with the posture of raising both arms during chest CT. The closer the quality control score is to 1, the better the quality control score is, and the closer the quality control score is to 0, the worse the quality control score is, and the less it meets the quality control standards.

[0099] like Figure 2 As shown in the multi-view diagram of CT images, the CT space matrix has three axes, X, Y, and Z, as shown in the left figure; the middle single-layer picture is a schematic diagram of the spatial perspective XOY plane, and θ (0°, 45°, 90°, 135°) is the view plane; the perspective is projected vertically onto this plane, and finally a ZOθ view is formed as shown in the right figure. The ellipse and straight line least squares fitting image is used to calculate the normal of the chest body and the direction vector of the arm humerus.

[0100] The quality control evaluation process of the multi-view double-arm raised posture mainly includes the following steps:

[0101] Input CT sequence images, select chest quality control images, and perform image preprocessing.

[0102] The CT values ​​of the ribs and humerus of the trunk are generally around 500-1500Hu.

[0103] window_center=1000

[0104] window_width=500

[0105] Among them, window_center is the window frame of CT bone value, window_width is the window position of CT bone value, and CT value is converted into voxel point of chest rib image. Here, voxel is represented by Cube(X,Y,Z), where X, Y, and Z represent three axes respectively.

[0106] Enter the number of segments N (N ≥ 2, the finer the N segmentation, the more accurate the result). For example, N = 4, so from 180 ≥ ≥0, the angle of obtaining the section view is divided into four sections, V (0°, 45°, 90°, 135°), where the viewing angle section is and The voxels of the Cube are rotated and mapped to form the view content.

[0107]

[0108]

[0109]

[0110]

[0111] Where n and m are the pixels mapped from x and y from the viewing angle θ, is the minimum coordinate where n pixels are not empty, is the maximum coordinate of n pixels that are not empty, h and v are pixels in ZOθ view n, H and V are the new coordinate axes in ZOθ view, A symbol indicating whether the pixel is empty.

[0112] Finally, we get the cross-sectional view .

[0113] right Ellipse fitting is performed on the rib pixels in the view.

[0114]

[0115] Let F(A,H,V)=0,

[0116] Among them, A is the parameter (a, b, c, d, e, f); H, V are ; F(A,H,V) is the standard formula for ellipse fitting. F(A,H,V) is obtained by least squares fitting, and the coordinates of the two end points of the major axis of the ellipse are obtained by extreme value derivation.

[0117] Ellipse curvature formula

[0118]

[0119] The curvature at the pole is the largest, argmax(K,v,h), and we get , , get the vector And continue normalizing, we get Normal vector of the view .

[0120] right A straight line is fitted to the humeral pixels in the view.

[0121] F(A,H,V)=ah+bv+c

[0122] Among them, A is the parameter (a, b, c), HV is (h, v, 1), and F(A, H, V) is the standard formula for straight line fitting.

[0123] Let F(A,H,V)=0, and use the least squares method to fit F(A,H,V). And the direction vector , normalized to get .

[0124] Use the ellipse vector T( ) is the direction in the resultant space, as the normal vector of the minor axis plane of the ellipsoid

[0125]

[0126] Among them, x, y, z are The coordinate of , h is the ordinate in the H plane.

[0127] Use the line direction vector U( ) as the direction in the synthetic space

[0128]

[0129] Among them, x, y, z are The coordinate of , h is the ordinate in the H plane.

[0130] As the direction vector of the humerus, the direction vectors of the humerus of the left and right arms are , .

[0131] Normal plane direction to the torso , arm humerus direction vector , Calculate the space angle. The calculation formula for the space vector angle is , calculated The left arm , right arm humeral angle .

[0132] Final calculation of quality control score

[0133]

[0134]

[0135]

[0136]

[0137] Among them, S_final is the final quality control score, S_left and S_right are the quality control scores of the left and right hand postures. When S=1, it is considered that the posture of raising arms during chest CT is completely in line with the condition. The closer the quality control score is to 1, the better the quality control score is. The closer the quality control score is to 0, the worse the quality control score is, and the less it meets the quality control standards.

[0138] The innovation of the present invention embodies novel ideas in the field of gesture recognition and behavior analysis, which are specifically embodied in the following aspects:

[0139] Multi-view fusion technology is a significant innovation in the process of fitting the torso posture. Traditional methods often rely on images or data from a single perspective, which easily leads to information loss or error accumulation. Multi-view fusion can comprehensively utilize view information from different angles and effectively integrate this information through algorithms, thereby improving the accuracy and robustness of posture recognition. This technology is particularly suitable for posture estimation in complex or occluded environments, providing strong support for the accurate fitting of torso posture.

[0140] This innovation reflects a deep understanding of the kinematic characteristics of the human body, based on the modeling of the angle between the direction of the humerus of both arms and the normal vector of the torso plane. The direction of the humerus of both arms not only reflects the movement state of the upper limbs, but is also closely related to the posture of the torso. By calculating the angle between the direction of the humerus of both arms and the normal vector of the torso plane, an intuitive and effective model can be constructed to describe the relative position relationship between the arms and the torso. This model is not only suitable for the recognition of specific behaviors such as raising arms, but can also be extended to other behavioral analyses involving coordinated movements of the upper limbs and torso.

[0141] Establishment of a quality control evaluation system. Based on the above model, the present invention further establishes a quality control evaluation system for the double-arm-raised behavior. This system can quantitatively evaluate the standardization and accuracy of the double-arm-raised behavior, and provides an important tool for sports training, rehabilitation therapy, human-computer interaction and other fields. Through real-time monitoring and feedback, users can adjust their movements to achieve better training effects or rehabilitation goals. At the same time, the quality control evaluation system also provides evaluation criteria for the automated behavior analysis system, which helps to improve the intelligence level of the system.

[0142] In summary, the innovation of the present invention is to improve the accuracy of posture recognition through multi-view fusion technology, use the angle between the direction of the humerus of both arms and the normal vector of the torso plane to build a model to describe the relative position relationship between the upper limbs and the trunk, and establish a quality control evaluation system for the arm-raising behavior based on this. These innovations not only enrich the inspection quality control methods of quality control posture recognition and behavior analysis, but also provide more powerful support for the application of quantitative CT image quality control. A quantitative quality control evaluation method for the arm-raising behavior was established, based on the angle modeling method between the direction of the humerus of both arms and the normal vector of the torso plane, and a multi-view fusion method was used to improve the accuracy of the modeling measurement calculation angle.

[0143] Beneficial effects:

[0144] The model is constructed by calculating the angle between the humeral direction of both arms and the normal vector of the torso plane and applied to CT image quality control, which can bring the following beneficial effects:

[0145] 1. Improve the accuracy of image analysis

[0146] Precise positioning: The model can accurately describe the relative position of the arms and the torso, which helps to more accurately locate the position and posture of the arms in CT images. Reduce errors: Traditional methods may produce errors due to human judgment or image blur, but this model reduces this subjectivity and uncertainty by calculating the angle, thereby improving the accuracy of the analysis.

[0147] 2. Optimize quality control process

[0148] Automated processing: The model can be integrated into CT image processing software to automatically identify and analyze the posture of both arms, thereby simplifying the quality control process and reducing manual intervention. Rapid feedback: Through model analysis, possible problems in the image can be quickly discovered, such as improper position of both arms, blurred images, etc., providing timely feedback to quality control personnel.

[0149] 3. Improve diagnostic efficiency

[0150] Save time: The automated quality control process can significantly save doctors’ time on image analysis, allowing them to focus more on disease diagnosis and treatment. Improve consistency: The application of this model helps achieve consistency in quality control standards and reduce differences in image analysis between different doctors or at different time points.

[0151] 4. Promote technological development

[0152] Promoting innovation, the application of this model demonstrates the potential of computer vision and image processing technology in the medical field, and helps promote the innovation and development of related image quality control technologies.

[0153] In summary, building a model by calculating the angle between the humeral direction of both arms and the normal vector of the torso plane and applying it to CT image quality control can bring many beneficial effects, including improving the accuracy of image analysis, optimizing the quality control process, improving diagnostic efficiency and promoting the development of quality control technology.

[0154] Example

[0155] like Figure 3 As shown, at 0° view:

[0156] T(0)=(0.99,0.13)

[0157] U_L(0)=(0.71,-0.70)

[0158] U_R(0)=(0.67,0.74)

[0159] The fitting diagram shows that the ellipse in the 0° view is the fit to the chest contour, and the straight line is the fit to the humerus of both arms.

[0160] Where T(0) is the ellipse pole vector. According to the normal vector formula Calculate and obtain, U_L(0), U_R(0) is the humeral direction vector, according to the formula Calculate acquisition.

[0161] like Figure 4 As shown, at 45° view:

[0162] T(45)=(0.99,0.07)

[0163] U_L(45)=(0.98,0.16)

[0164] U_R(45)=(0.97,-0.21)

[0165] T(45), U_L(45), U_R(45) are the vector results under the 45° view, and the formula is the same as the 0° view.

[0166] Where T(45) is the ellipse pole vector. According to the normal vector formula Calculate and obtain, U_L(45), U_R(45) is the humeral direction vector, according to the formula Calculate acquisition.

[0167] like Figure 5 As shown, at 90° view:

[0168] T(90)=(0.99,0.05)

[0169] U_L(90)=(0.97,-0.23)

[0170] U_R(90)=(0.98,-0.21)

[0171] T(90), U_L(90), U_R(90) are the vector results under the 90° view, and the formula is the same as the 0° view.

[0172] Where T(90) is the ellipse pole vector. According to the normal vector formula Calculate and obtain, U_L(90), U_R(90) is the humeral direction vector, according to the formula Calculate acquisition.

[0173] like Figure 6 As shown, at 135° view:

[0174] T(135)=(0.99,0.11)

[0175] U_L(135)=(0.97,-0.22)

[0176] U_R(135)=(0.97,-0.26)

[0177] T(135), U_L(135), U_R(135) are the vector results under the 135° view, and the formula is the same as the 0-degree view.

[0178] Where T(135) is the ellipse pole vector. According to the normal vector formula Calculate and obtain, U_L(135), U_R(135) is the humeral direction vector, according to the formula Calculate acquisition.

[0179] Use the ellipse vector T( ) Direction in synthetic space, action ellipsoid, normal vector of the minor axis plane

[0180]

[0181] Use the line direction vector obtained from multiple views Direction in composite space

[0182]

[0183] As the direction vector of the humerus, the direction vectors of the humerus of the left and right arms are , .

[0184] Calculate the ellipsoid normal:

[0185]

[0186] x=(0.13*sin(0)+0.07*sin(pi / 4)+0.05*sin(pi / 2)+0.11*sin(3*pi / 4)) / 4=0.04

[0187] y=(0.13*cos(0)+0.07*cos(pi / 4)+0.05*cos(pi / 2)+0.11*cos(3*pi / 4)) / 4=0.03

[0188] z=(0.99+0.99+0.99+0.99) / 4=0.99

[0189] y=(0.13*cos(0)+0.07*cos(pi / 4)+0.05*cos(pi / 2)+0.11*cos(3*pi / 4)) / 4=0.03

[0190] z=(0.99+0.99+0.99+0.99) / 4=0.99

[0191] Calculate the humeral direction vector:

[0192]

[0193] x=(-0.70*sin(0)+0.16*sin(pi / 4)+(-0.23)*sin(pi / 2)+(-0.22)*sin(3*pi / 4)) / 4=-0.07

[0194] y=(-0.70*cos(0)+0.16*cos(pi / 4)+(-0.23)*cos(pi / 2)+(-0.22)*cos(3*pi / 4)) / 4=-0.11

[0195] z=(0.71+0.98+0.97+0.97) / 4=0.91

[0196] Calculate the humeral direction vector:

[0197]

[0198] x=(0.74 * sin(0)+(-0.21)* sin(pi / 4)+(-0.21) * sin(pi / 2)+(-0.26) * sin(3*pi / 4) ) / 4=-0.13

[0199] y=(0.74 * cos(0)+(-0.21)* cos(pi / 4)+(-0.21) * cos(pi / 2)+(-0.26) * cos(3*pi / 4) ) / 4=0.19

[0200] z=(0.67+0.97+0.98+0.97) / 4=0.90

[0201] where pi is equivalent to , pi / 4 is 45°, pi / 2 is 90°, and 3*pi / 4 is 135°.

[0202] For example, if 0.99, 0.99, 0.99, 0.99 are the horizontal coordinates of T(0), T(45), T(90), and T(135), substitute the values ​​from the previous step.

[0203] For example, if 0.13, 0.79, 0.05, and 0.11 are the vertical coordinates of T(0), T(45), T(90), and T(135), substitute the values ​​from the previous step.

[0204] For example, 0.71, 0.98, 0.97, 0.97 are the horizontal axis coordinates of U_L(0), U_L(45), U_L(90), U_L(135). Substitute the values ​​from the previous step.

[0205] For example, -0.7, 0.16, 0.23, -0.22 are the vertical coordinates of U_L(0), U_L(45), U_L(90), U_L(135), substitute the values ​​from the previous step.

[0206] For example, if 0.67, 0.97, 0.98, 0.97 are the horizontal coordinates of U_R(0), U_R(45), U_R(90), U_R(135), substitute the values ​​from the previous step.

[0207] For example, 0.74, -0.21, -0.21, -0.26 are the vertical coordinates of U_R(0), U_R(45), U_R(90), and U_R(135). Substitute the values ​​from the previous step.

[0208] Compute the magnitude of a normal vector:

[0209] | |=(0.04**2+0.03**2+0.99**2)**0.5=0.99

[0210] Compute the magnitude of a direction vector:

[0211] | |=(0.07**2+0.11**2+0.91**2)**0.5=0.92

[0212] | |=(0.13**2+0.19**2+0.90**2)**0.5=0.93

[0213] | |,| |,| | are respectively the vectors in the previous step. , , Model.

[0214] For example, 0.04, 0.03, and 0.99 are Substitute the x, y, and z coordinate values ​​from the previous step into the numerical calculation.

[0215] For example, 0.07, 0.11, and 0.91 are Substitute the x, y, and z coordinate values ​​from the previous step into the numerical calculation.

[0216] For example, 0.13, 0.19, and 0.90 are Substitute the x, y, and z coordinate values ​​from the previous step into the numerical calculation.

[0217] Calculate the angle between the normal vector and the direction vector:

[0218]

[0219]

[0220] for , The angle between Obtained by calculation, and obtained by substituting numerical values ​​for calculation.

[0221] for , The angle between Obtained by calculation, and obtained by substituting numerical values ​​for calculation.

[0222] Convert the angle between the normal vector and the direction vector into a quality control score:

[0223] (0.06pi<0.5pi)

[0224] The angle between the normal vector and the direction vector is converted into a quality control score:

[0225] (0.09pi<0.5pi)

[0226] Comprehensive quality control scores of both arms:

[0227]

[0228] ， ， By formula Substitute 0.06pi and 0.09pi calculated in the previous step to calculate, where pi is equivalent to .

[0229] Evaluation results: The quality control result score is 1, which meets the quality control requirements of raising both arms. It is recommended to keep the image.

[0230] The specific embodiments described herein are merely examples of the spirit of the present invention. The above embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that technicians in the technical field to which the present invention belongs can make various modifications or supplements to the described specific embodiments or replace them in a similar manner, but will not deviate from the spirit of the present invention or exceed the scope defined in the attached claims. For ordinary technicians in this field, multiple deformations and improvements can also be made without departing from the concept of the present invention. Therefore, the scope of protection of the patent of the present invention shall be based on the attached claims.

Claims

1. A method for quality control of CT images of arms raised combined with multi-view fusion, characterized in that: The following steps are involved: S1 inputs CT sequence images and selects chest quality control images for preprocessing; S2 segments the preprocessed image to obtain a cross-sectional view; S3 performs ellipse fitting on the rib pixels in the section view and straight line fitting on the humerus pixels in the section view to obtain the normal plane direction of the trunk and the direction of the arm humerus; S4 calculates the humeral angle of both arms and obtains the quality control score of both arms posture, and determines the retention of CT sequence images according to the quality control score; The angle is calculated by the normal vector of the chest body ellipsoid plane and the direction vector of the humerus of both arms, and the angle is quantified into a quality control score to quantitatively evaluate the quality control effect.

2. According to the method for quality control of CT images of arms raised combined with multi-view fusion according to claim 1, it is characterized in that: The preprocessing in step S1 includes setting a window frame and a window level of the CT bone value, and converting the CT values ​​of the ribs and humerus of the trunk into voxel points of the chest rib image.

3. The method for quality control of CT images of arms raised combined with multi-view fusion according to claim 1 is characterized in that: The step S2 specifically includes inputting the number of cuts, obtaining the angle of the cut view, and rotating the view cut to map voxels to form view content.

4. The method for quality control of CT images of arms raised combined with multi-view fusion according to claim 2 or 3, characterized in that: The voxel is represented by Cube(X, Y, Z), wherein X, Y, and Z represent three axes respectively.

5. The method for quality control of CT images of arms raised combined with multi-view fusion according to claim 3 is characterized in that: The step S2 also includes: and The voxels of the Cube are rotated and mapped to form the view content, and finally the section view V(θ,H,V) is obtained: n=xsinθ-ycosθ m=xcosθ+ysinθ Where n and m are the pixels mapped from x and y from the viewing angle θ, is the minimum coordinate where n pixels are not empty, is the maximum coordinate of n pixels that are not empty, h and v are pixels in ZOθ view n, H and V are the new coordinate axes in ZOθ view, A symbol indicating whether the pixel is empty.

6. The method for quality control of CT images of arms raised combined with multi-view fusion according to claim 1, characterized in that: The step S3 of performing ellipse fitting on the rib pixels in the cross-sectional view specifically comprises: F(A,H,V)=ah 2 +bhv+cv 2 +d+ev+f Let F(A,H,V)=0, Among them, A is the parameter (a, b, c, d, e, f); H, V are (h 2 ,hv,v 2 ,h,v,1); F(A,H,V) is the standard formula for ellipse fitting. The least square method is used to fit F(A,H,V), and the extreme value derivative is used to obtain the coordinates of the two end points of the major axis of the ellipse; Ellipse curvature formula The curvature at the pole is the largest, argmax(K,v,h), and P1(v 1, h1),P2(v2h2),get vector And continue to normalize to get the normal vector of the V(θ,H,V) view 7. The method for quality control of CT images of arms raised combined with multi-view fusion according to claim 6, characterized in that: The straight line fitting of the humerus pixels in the section view specifically includes: F(A,H,V)=ah+bv+c Where A is the parameter (a, b, c), HV is (h, v, 1), and F(A, H, V) is the standard formula for straight line fitting; let F(A, H, V) = 0, use the least squares method to fit F(A, H, V), and the direction vector Normalize to get 8. The method for quality control of CT images of arms raised combined with multi-view fusion according to claim 7, characterized in that: The step S3 further includes using the direction of the ellipse vector T(θ) obtained in the multi-view synthesis space as the normal vector of the minor axis plane of the ellipsoid: Among them, x, y, z are the coordinates of γ, and h is the ordinate in the H plane; Use the straight line direction vector U(θ) obtained from multiple views as the direction in the synthesis space Among them, x, y, z are the coordinates of α, and h is the ordinate in the H plane; As the direction vector of the humerus, the direction vectors of the humerus of the left and right arms are α(x, y, z) and β(x, y, z) respectively.

9. The method for quality control of CT images of arms raised combined with multi-view fusion according to claim 8, characterized in that: The step S4 of calculating the humeral angle of both arms specifically includes: Calculate the spatial angle of the torso normal plane direction γ(x,y,z), the arm humerus direction vector α(x,y,z), β(x,y,z). The calculation formula of the spatial vector angle is Calculated The left arm is τ(α), and the right arm humeral angle is τ(β).

10. The method for quality control of CT images of arms raised combined with multi-view fusion according to claim 9, characterized in that: The step S4 of calculating the quality control score specifically includes: S_left=S(τ(α)) S_right=S(τ(β)) Among them, S_final is the final quality control score, S_left and S_right are the quality control scores of the left and right hand postures of both arms. When S=1, it is considered to be completely consistent with the posture of raising both arms during chest CT. The closer the quality control score is to 1, the better the quality control score is, and the closer the quality control score is to 0, the worse the quality control score is, and the less it meets the quality control standards.

Images