Tumor positioning method and system based on radiotherapy image

Abstract

The invention relates to the field of brain radiotherapy image processing, in particular to a tumor positioning method and system based on radiotherapy images. The method comprises the following steps: obtaining an initial optimization factor according to gradient difference and distance distribution between each feature point and a corresponding feature point in an adjacent image; correcting the initial optimization factor in combination with the density feature, the direction feature and the organization feature of each feature point to obtain a final optimization factor of each feature point; obtaining the priority pairing degree of each brain radiotherapy image; pairing the brain radiotherapy images according to the priority pairing degree; and carrying out brain tumor positioning according to a brain radiotherapy image matching result. According to the invention, an accurate image registration result can be obtained, so that an accurate three-dimensional model is obtained, and tumor positioning can be more accurate.

Description

Technical Field

[0001] The present invention relates to the field of brain radiotherapy image processing, and in particular to a tumor positioning method and system based on radiotherapy images. Background Art

[0002] Radiotherapy imaging refers to the use of medical imaging equipment to scan and image the patient's tumor to obtain the three-dimensional structural information of the tumor, providing basic data for the formulation and implementation of radiotherapy plans. Localization of brain tumors usually requires reconstruction of a three-dimensional model of the patient's brain, and before that, it is necessary to align the patient's different brain radiotherapy images.

[0003] When registering a patient's brain radiotherapy images, the most commonly used method is the feature point matching algorithm, which achieves image registration by matching multiple slices of the radiotherapy image in sequence. Conventional feature point matching algorithms usually match adjacent slice images in pairs when matching feature points. However, the brain is an area with high activity and high tissue density, and the quality of each brain radiotherapy image is different. Low-quality brain radiotherapy images will directly affect the accuracy and stability of feature point matching, and errors may accumulate between adjacent brain radiotherapy images with the number of matches, resulting in poor final registration results, which directly affects the accuracy of the three-dimensional model and makes it difficult to obtain accurate tumor localization results. Summary of the invention

[0004] In order to solve the technical problem that low-quality brain radiotherapy images will directly affect the accuracy and stability of feature point matching, and the error may accumulate between adjacent brain radiotherapy images with the number of matches, resulting in poor final registration effect, directly affecting the accuracy of the three-dimensional model, and it is difficult to obtain accurate tumor localization results, the purpose of the present invention is to provide a tumor localization method and system based on radiotherapy images, and the technical solutions adopted are as follows:

[0005] A tumor localization method based on radiotherapy images, the method comprising:

[0006] Obtain all brain radiotherapy images;

[0007] Select any brain radiotherapy image as the target radiotherapy image, and select any feature point of the target radiotherapy image as the reference feature point; obtain the initial optimization factor of the reference feature point according to the gradient difference and distance distribution between the reference feature point and the feature point closest to the corresponding position of the reference feature point in all preset neighboring images; obtain the density feature index of the reference feature point according to the grayscale distribution in the preset first neighborhood of the reference feature point; obtain the direction feature index of the reference feature point according to the gradient distribution of the pixel points in the preset first neighborhood of the reference feature point; obtain the tissue feature index of the reference feature point according to the distribution of other feature points in the preset second neighborhood of the reference feature point; obtain the final optimization factor of the reference feature point according to the density feature index, the direction feature index and the tissue feature index of the reference feature point;

[0008] Obtaining a priority matching degree of the target radiotherapy image according to the final preferred factors of all feature points in the target radiotherapy image and the number of feature points of each brain radiotherapy image; pairing the brain radiotherapy images according to the priority matching degree of each brain radiotherapy image;

[0009] Brain tumor localization is performed based on the matching results of brain radiotherapy images.

[0010] According to the method for tumor localization based on radiotherapy images according to claim 1, the method for obtaining the initial optimization factor comprises:

[0011] The feature point in each preset neighboring image that is closest to the corresponding position of the reference feature point is used as the feature point to be compared in each preset neighboring image;

[0012] The initial optimal factor is obtained according to the initial optimal factor calculation formula, and the initial optimal factor calculation formula is as follows:

[0013]

[0014] Where μ represents the initial optimization factor of the reference feature point; I represents the number of preset neighboring images; T0 represents the gradient value of the reference feature point; T i represents the gradient value of the feature point to be compared in the i-th preset neighboring image; θ0 represents the gradient direction angle of the reference feature point, where the gradient direction angle is the angle between the gradient direction of the feature point and the horizontal direction; θ i represents the gradient direction angle of the feature point to be compared in the i-th preset neighboring image; R i G represents the distance between the corresponding position of the reference feature point and the feature point to be compared in the i-th preset neighboring image; i represents the image interval between the i-th preset adjacent image and the target radiotherapy image; || represents the absolute value function; exp[] represents the exponential function with a natural constant as the base.

[0015] According to the tumor localization method based on radiotherapy images according to claim 1, the method for obtaining the density characteristic index comprises:

[0016] The density characteristic index is obtained according to the density characteristic index calculation formula, and the density characteristic index calculation formula is as follows:

[0017]

[0018] Where ρ represents the density characteristic index of the reference feature point; J represents the number of pixels in the first neighborhood of the reference feature point; H0 represents the gray value of the reference feature point; H j Represents the gray value of the jth pixel in the preset first neighborhood of the reference feature point; Indicates the grayscale mean of the pixels in the first neighborhood preset by the reference feature point; Represents the grayscale mean of the target radiotherapy image; norm() represents the normalization function.

[0019] According to the method for tumor localization based on radiotherapy images according to claim 1, the method for obtaining the directional characteristic index comprises:

[0020] Obtaining a gradient histogram of all feature points in a second area preset by the reference feature point;

[0021] Sort all the gradient histogram columns in the gradient histogram according to their height from large to small;

[0022] The directional feature index of the reference feature point is obtained according to the gradient distribution of the feature points in the preset second neighborhood. The calculation formula is as follows:

[0023]

[0024] Where ω represents the directional characteristic index of the reference feature point; W max represents the peak value of the gradient histogram; K represents the number of gradient histogram columns in the gradient histogram; W k Represents the height of the kth histogram in the gradient histogram; W k+1 Represents the height of the k+1th histogram in the gradient histogram; norm[] represents the normalization function.

[0025] According to the method for tumor localization based on radiotherapy images according to claim 1, the method for obtaining the tissue characteristic index comprises:

[0026] Selecting a preset number of feature points closest to the reference feature points in a preset second area of ​​the reference feature points as target feature points;

[0027] The organizational characteristic index is obtained according to the organizational characteristic index calculation formula, and the organizational characteristic index calculation formula is as follows:

[0028]

[0029] In the formula, δ represents the tissue characteristic index of the reference feature point; V represents the preset number of target feature points; R v represents the distance between the vth target feature point and the reference feature point; σ(R v ) represents the distance variance between the target feature point and the reference feature point; F represents the number of feature points in the preset second neighborhood of the reference feature point; G represents the number of feature points in the target radiotherapy image; norm[] represents the normalization function; exp() represents the exponential function with a natural constant as the base.

[0030] According to the method for tumor localization based on radiotherapy images according to claim 1, the method for obtaining the final optimal factor comprises:

[0031] The final preferred factor is obtained according to the final preferred factor calculation formula, and the final preferred factor calculation formula is as follows:

[0032]

[0033] In the formula, μ′ represents the final optimization factor of the reference feature point; μ represents the initial optimization factor of the reference feature point; ρ represents the density characteristic index of the reference feature point; ω represents the direction characteristic index of the reference feature point; and μ represents the tissue characteristic index of the reference feature point.

[0034] According to the method for tumor localization based on radiotherapy images according to claim 1, the method for obtaining the priority matching degree comprises:

[0035] The priority matching degree is obtained according to the priority matching degree calculation formula, and the priority matching degree calculation formula is as follows:

[0036]

[0037] In the formula, γ represents the priority pairing degree of the target radiotherapy image; represents the final optimal factor mean of all feature points in the target radiotherapy image; P represents the number of feature points in the target radiotherapy image; P max Indicates the number of feature points in the brain radiotherapy image with the largest number of feature points.

[0038] A tumor localization system based on radiotherapy images, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned tumor localization method based on radiotherapy images when executing the computer program.

[0039] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned tumor localization method based on radiotherapy images are implemented.

[0040] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned tumor localization method based on radiotherapy images when executing the computer program.

[0041] The present invention has the following beneficial effects:

[0042] The present invention obtains all brain radiotherapy images for subsequent image registration and constructs a three-dimensional brain model; since if there are feature points with highly similar features in adjacent images, the reference feature point is more likely to be the real feature point in the target radiotherapy image, and should be selected preferentially when matching the feature points, the initial preferred factor of the reference feature point is obtained based on the gradient difference and distribution characteristics between the reference feature point and the feature point closest to the corresponding position of the reference feature point in all preset adjacent images; since there is a large density difference between the brain tumor area and other normal brain areas, there is a large difference in the gradient direction of the feature points in different areas, and the feature points in the brain tumor area are The distribution is also different from that of the normal brain area, so in order to distinguish the tumor area from the normal area, the density characteristics, direction characteristics and tissue characteristics of each feature point are analyzed to obtain the density characteristic index, direction characteristic index and tissue characteristic index of each feature point; since the higher the quality of the brain radiotherapy image, the more it needs to be retained when the feature points are matched later, so according to the final preferred factors of all feature points in the target radiotherapy image and the number of feature points of each brain radiotherapy image, the priority matching degree of the target radiotherapy image is obtained; using the priority matching degree of each brain radiotherapy image, all brain radiotherapy images are image matched; brain tumors are located according to the matching results of the brain radiotherapy images. The present invention can obtain more accurate image registration results, thereby obtaining a more accurate three-dimensional model, which can make tumor positioning more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1A flow chart of a tumor localization method based on radiotherapy images provided by one embodiment of the present invention;

[0045] Figure 2 A block diagram of a tumor localization system based on radiotherapy images provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation, structure, features and effects of a tumor localization method and system based on radiotherapy images proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0047] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0048] A specific solution of a tumor localization method based on radiotherapy images provided by the present invention is described in detail below with reference to the accompanying drawings.

[0049] See also Figure 1 , which shows a tumor localization method based on radiotherapy images provided by an embodiment of the present invention, the method comprising:

[0050] Step 1: Acquire all brain radiation therapy images.

[0051] The embodiment of the present invention is mainly applied to the brain tumor positioning scenario. In order to position the brain tumor, it is necessary to obtain a three-dimensional model of the brain. Before the three-dimensional model is established, it is necessary to first obtain brain radiotherapy images and perform feature point registration on all brain radiotherapy images.

[0052] In one embodiment of the present invention, an X-ray head CT machine is used to obtain a CT image of the patient's brain, and the brain CT image is used as the brain radiotherapy image that needs to be obtained, wherein the brain CT image is composed of multi-angle and multi-level slices, and the brain CT image has spatial continuity. The CT scan can generate continuous brain cross-sectional images, which can cover the entire brain area and provide a continuous view from the skull base to the skull top.

[0053] In one embodiment of the present invention, a semantic segmentation network is used to segment all brain radiotherapy images, remove irrelevant background areas, and retain only the brain area.

[0054] In one embodiment of the present invention, a SIFT algorithm is used to perform subsequent feature point matching operations.

[0055] It should be noted that the SIFT algorithm and the semantic segmentation network are technical means well known to those skilled in the art and will not be described in detail here.

[0056] Step S2: Select any brain radiotherapy image as the target radiotherapy image, and select any feature point of the target radiotherapy image as a reference feature point; obtain the initial optimization factor of the reference feature point based on the gradient difference and distance distribution between the reference feature point and the feature point closest to the corresponding position of the reference feature point in all preset neighboring images; obtain the density feature index of the reference feature point based on the grayscale distribution in the preset first neighborhood of the reference feature point; obtain the directional feature index of the reference feature point based on the gradient distribution of the pixel points in the preset first neighborhood of the reference feature point; obtain the tissue feature index of the reference feature point based on the distribution of other feature points in the preset second neighborhood of the reference feature point; obtain the final optimization factor of the reference feature point based on the density feature index, directional feature index and tissue feature index of the reference feature point.

[0057] In reality, in order to judge the quality of multiple brain radiotherapy images, it is necessary to first compare and analyze the feature points in each image. If the reference feature point in the target radiotherapy image can find a feature point with highly similar features in several adjacent brain radiotherapy images, it means that the authenticity and feature strength of the reference feature point are higher, and the greater the contribution to the image quality of the target radiotherapy image. Therefore, it is analyzed whether there is a feature point with highly similar features in the adjacent images to determine whether the reference feature point is a real feature point in the target radiotherapy image, that is, according to the gradient difference and distribution characteristics between the reference feature point and the feature point closest to the corresponding position of the reference feature point in all preset adjacent images, the initial preferred factor of the reference feature point is obtained.

[0058] Preferably, in one embodiment of the present invention, the method for obtaining the initial optimization factor includes:

[0059] The feature point in each preset neighboring image that is closest to the corresponding position of the reference feature point is used as the feature point to be compared in each preset neighboring image.

[0060] In one embodiment of the present invention, three brain radiotherapy images before and after the target radiotherapy image are used as preset adjacent images. It should be noted that in other embodiments of the present invention, the preset adjacent images can be set by the implementer, which is not limited here.

[0061] The initial preferred factor is obtained according to the initial preferred factor calculation formula. The initial preferred factor calculation formula is as follows:

[0062]

[0063] Where μ represents the initial optimization factor of the reference feature point; I represents the number of preset neighboring images; T0 represents the gradient value of the reference feature point; T i represents the gradient value of the feature point to be compared in the i-th preset neighboring image; θ0 represents the gradient direction angle of the reference feature point, where the gradient direction angle is the angle between the gradient direction of the feature point and the horizontal direction; θ i represents the gradient direction angle of the feature point to be compared in the i-th preset neighboring image; R i G represents the distance between the corresponding position of the reference feature point and the feature point to be compared in the i-th preset neighboring image; i represents the image interval between the i-th preset adjacent image and the target radiotherapy image, where the image interval is the number of images between the i-th preset adjacent image and the target radiotherapy image; | | represents the absolute value function; exp[] represents the exponential function with a natural constant as the base.

[0064] In the initial optimization factor calculation formula, the difference in gradient value and gradient direction angle (|T0-T i |×|θ0-θ i |) is smaller, indicating that the reference feature point has more similar features in the preset adjacent image. At this time, the reference feature point is closer to the real feature point, and the reference feature point can more affect the image quality of the target radiotherapy image. As the weight coefficient of the feature point to be compared in the i-th preset adjacent image, since the smaller the image interval between the i-th preset adjacent image and the target radiotherapy image, the smaller the acquisition time distance between the two images, the smaller the possible change between the reference feature point and the feature point to be compared between the two images, and the smaller the distance between the feature point to be compared in the i-th preset adjacent image and the corresponding position of the reference feature point in the image, the higher the feature similarity between the feature point to be compared in the i-th preset adjacent image and the reference feature point. At this time, (|T0-T i |×|θ0-θ i |) is more important, so As (|T0-T i |×|θ0-θ i |) weight coefficient, As the feature difference between the feature point to be compared in the i-th preset neighboring image and the reference feature point; calculate the average feature difference between the feature point to be compared in each preset neighboring image and the reference feature point, wherein the smaller the average feature difference, the greater the initial preferred factor of the reference feature point.

[0065] In brain radiotherapy images, the feature points corresponding to different brain regions contain different image information. For example, the feature points in the brain tumor region are critical for subsequent image matching, while the feature points in the normal brain region without lesions are relatively unimportant. Therefore, before matching the feature points, it is necessary to first analyze the location characteristics of the feature points, that is, to analyze the density characteristics, direction characteristics and tissue characteristics around the feature points.

[0066] Since the density of the brain tumor area is significantly different from that of other normal brain areas, the grayscale value of the brain tumor area is significantly different from that of the normal brain area in the image. Since tumor tissue has diversity and heterogeneity, the brain tumor area may show uneven density characteristics with larger grayscale values. Therefore, in an embodiment of the present invention, the grayscale distribution in the first neighborhood is preset according to the reference feature point to obtain the density feature index of the reference feature point.

[0067] The density characteristic index is obtained according to the density characteristic index calculation formula. The density characteristic index calculation formula is as follows:

[0068]

[0069] Where ρ represents the density characteristic index of the reference feature point; J represents the number of pixels in the first neighborhood of the reference feature point; H0 represents the gray value of the reference feature point; H j Represents the gray value of the jth pixel in the preset first neighborhood of the reference feature point; Indicates the grayscale mean of the pixels in the first neighborhood preset by the reference feature point; Represents the grayscale mean of the target radiotherapy image; norm() represents the normalization function.

[0070] In the density feature index calculation formula, the average grayscale difference between the grayscale value of the reference feature point and each pixel in its preset first neighborhood is The larger the value is, the more uneven the grayscale around the reference feature point is, and the more likely the reference feature point is a brain tumor area; and the difference between the grayscale mean of the preset first neighborhood and the overall grayscale mean of the target radiotherapy image is The larger the value is, the more abnormal the grayscale of the first neighborhood of the reference feature point is, and the more likely the first neighborhood of the reference feature point is to belong to the brain tumor area. Normalization is performed to quantify the density features of the reference feature points.

[0071] Since the tissue inside the brain tumor is unevenly distributed, it is reflected in the brain radiotherapy image that the gradient directions of the pixels surrounding the feature points in the brain tumor area are chaotic and disordered, while the gradient directions of the pixels surrounding the feature points in the normal brain area are mostly consistent. At this time, this direction is the main direction of the feature point. Therefore, in the embodiment of the present invention, the gradient distribution of the pixels in the first neighborhood is preset according to the reference feature point to obtain the directional feature index of the reference feature point.

[0072] Preferably, in one embodiment of the present invention, the method for obtaining the directional characteristic index includes:

[0073] Obtain the gradient histogram of all feature points in the second field preset by the reference feature point, in which the gradient direction corresponding to the peak value of the gradient histogram column is the main direction of the reference feature point. It should be noted that the gradient histogram is a technical means well known to those skilled in the art and will not be described in detail here.

[0074] In one embodiment of the present invention, a 25×25 rectangular area around the reference feature point is used as the preset second neighborhood. It should be noted that in other embodiments of the present invention, the preset second neighborhood can be set arbitrarily, and is not limited here.

[0075] In one embodiment of the present invention, all gradient histogram columns in the gradient histogram are sorted from large to small according to their heights, so as to facilitate the subsequent calculation of the height difference between two adjacent gradient histogram columns, wherein the smaller the height difference is, the less prominent the gradient directions corresponding to the two gradient histogram columns are.

[0076] The directional feature index of the reference feature point is obtained according to the gradient distribution of the feature points in the preset second neighborhood. The calculation formula is as follows:

[0077]

[0078] Where ω represents the directional characteristic index of the reference feature point; W max represents the peak value of the gradient histogram; K represents the number of gradient histogram columns in the gradient histogram; W k Represents the height of the kth histogram in the gradient histogram; W k+1 Represents the height of the k+1th histogram in the gradient histogram; norm[] represents the normalization function.

[0079] In the directional characteristic index calculation formula, the ratio of the gradient column to the total height of the gradient column is The smaller it is, the less prominent the main direction of the reference feature point is. At this time, the area to which the reference feature belongs is more likely to be a brain tumor area, and the larger the directional feature index should be. The average height difference between each two adjacent gradient histograms is The smaller it is, the more chaotic the gradient direction around the reference feature point is. At this time, the area to which the reference feature belongs is more likely to be a brain tumor area, and the larger the directional feature index should be.

[0080] Since brain tumors usually push or overburden surrounding brain tissues, the boundary features of the brain tumor area are clearer and more obvious than those of normal brain tissues, and the internal structural features are prominent, so the feature points in the brain tumor area are more concentrated inside the tumor and near the boundary. Therefore, in the embodiment of the present invention, the distribution of other feature points in the second neighborhood is preset according to the reference feature point to obtain the tissue feature index of the reference feature point.

[0081] Preferably, in one embodiment of the present invention, the method for obtaining tissue characteristic indicators includes:

[0082] A preset number of feature points closest to the reference feature point are selected as target feature points in the preset second area of ​​the reference feature point. In one embodiment of the present invention, the preset number is set to 5. It should be noted that in other embodiments of the present invention, the preset number can be set by itself, and the preset number is not greater than the total number of feature points in the preset second neighborhood, which is not limited here.

[0083] The organizational characteristic index is obtained according to the organizational characteristic index calculation formula. The organizational characteristic index calculation formula is as follows:

[0084]

[0085] In the formula, δ represents the tissue characteristic index of the reference feature point; V represents the preset number of target feature points; R v represents the distance between the vth target feature point and the reference feature point; σ(R v ) represents the distance variance between the target feature point and the reference feature point; F represents the number of feature points in the preset second neighborhood of the reference feature point; G represents the number of feature points in the target radiotherapy image; norm[] represents the normalization function; exp() represents the exponential function with a natural constant as the base.

[0086] In the organizational characteristic index calculation formula, the mean distance between the preset number of target characteristic points and the reference characteristic points is calculated. and distance variance σ(R v ), the smaller the distance mean and the smaller the distance variance, the higher the feature intensity around the reference feature point, and the larger the tissue feature index is at this time; calculate the ratio of the number of feature points in the second neighborhood preset for the reference feature point to the total number of feature points in the target radiotherapy image. The larger the ratio, the higher the feature intensity in the second neighborhood preset for the reference feature point, and the larger the tissue feature index is at this time.

[0087] Based on the density characteristic index, directional characteristic index and tissue characteristic index obtained in the above process, the initial optimization factor is corrected to obtain the final optimization factor. Therefore, in the embodiment of the present invention, the final optimization factor of the reference feature point is obtained according to the density characteristic index, directional characteristic index and tissue characteristic index of the reference feature point.

[0088] Preferably, in one embodiment of the present invention, the method for obtaining the final optimization factor includes:

[0089] The final preferred factor is obtained according to the final preferred factor calculation formula. The final preferred factor calculation formula is as follows:

[0090]

[0091] In the formula, μ′ represents the final optimization factor of the reference feature point; μ represents the initial optimization factor of the reference feature point; ρ represents the density characteristic index of the reference feature point; ω represents the direction characteristic index of the reference feature point; δ represents the tissue characteristic index of the reference feature point.

[0092] In the final optimization factor calculation formula, the mean values ​​of the density characteristic index, direction characteristic index and tissue characteristic index of the reference feature point are calculated. The larger the mean, the more the reference feature point conforms to the characteristics of the brain tumor area. At this time, the reference feature point is more likely to be the feature point of the brain tumor area. As the correction factor of the initial optimization factor, the final optimization factor is obtained.

[0093] Step S3: obtaining the priority matching degree of the target radiotherapy image according to the final preferred factors of all feature points in the target radiotherapy image and the number of feature points of each brain radiotherapy image; and pairing the brain radiotherapy images according to the priority matching degree of each brain radiotherapy image.

[0094] Since the more feature points there are in a brain radiotherapy image and the larger the final preferred factor of the feature point is, the higher the quality of the brain radiotherapy image is, and the more it needs to be retained when the feature points are matched later. In order to select brain radiotherapy images with higher image quality, in one embodiment of the present invention, the priority matching degree of the target radiotherapy image is obtained based on the final preferred factors of all feature points in the target radiotherapy image and the number of feature points in each brain radiotherapy image.

[0095] Preferably, in one embodiment of the present invention, the method for obtaining the priority pairing degree includes:

[0096] The priority matching degree is obtained according to the priority matching degree calculation formula. The priority matching degree calculation formula is as follows:

[0097]

[0098] In the formula, γ represents the priority pairing degree of the target radiotherapy image; represents the final optimal factor mean of all feature points in the target radiotherapy image; P represents the number of feature points in the target radiotherapy image; P max Indicates the number of feature points in the brain radiotherapy image with the largest number of feature points.

[0099] In the priority matching degree calculation formula, the larger the final preferred factor mean of all feature points in the target radiotherapy image, and the larger the ratio between the number of feature points of the target radiotherapy image and the number of feature points in the brain radiotherapy image with the largest number of feature points, the richer the feature information of the brain tumor area contained in the target radiotherapy image, and the more preferential matching should be performed on the target radiotherapy image, that is, the higher the priority matching degree of the target radiotherapy image.

[0100] Using the priority matching degree of each brain radiotherapy image, all brain radiotherapy images are matched. In one embodiment of the present invention, an image matching method is provided, which specifically includes:

[0101] Among the three brain radiotherapy images before and after the target radiotherapy image, the brain radiotherapy image with the highest degree of matching is selected to perform image matching with the target radiotherapy image;

[0102] All brain radiotherapy images are traversed to obtain the matching result of each brain radiotherapy image.

[0103] At this point, the image matching of the brain radiotherapy images is completed.

[0104] Step S4: localize the brain tumor according to the matching results of the brain radiotherapy images.

[0105] In one embodiment of the present invention, after obtaining the matching results of brain radiotherapy images, feature point matching is performed on all brain radiotherapy images to complete the registration of all brain radiotherapy images. After the registration is completed, the Level Set algorithm is used to perform three-dimensional reconstruction of the patient's brain to obtain a visual brain three-dimensional model with high accuracy and authenticity, thereby achieving accurate positioning of the patient's brain tumor.

[0106] It should be noted that image registration and Level Set algorithm are both technical means well known to those skilled in the art, and will not be described in detail here.

[0107] At this point, the patient's brain tumor has been localized.

[0108] In summary, all brain radiotherapy images are acquired; any one brain radiotherapy image is selected as the target radiotherapy image, and any feature point of the target radiotherapy image is selected as the reference feature point; according to the reference feature point, the gradient difference and distance distribution between the feature points closest to the corresponding positions of the reference feature point in all preset neighboring images are used to obtain the initial preferred factor of the reference feature point; according to the grayscale distribution in the preset first neighborhood of the reference feature point, the density feature index of the reference feature point is obtained; according to the gradient distribution of the pixel points in the preset first neighborhood of the reference feature point, the direction feature index of the reference feature point is obtained; according to the distribution of other feature points in the preset second neighborhood of the reference feature point, the tissue feature index of the reference feature point is obtained; according to the density feature index, direction feature index and tissue feature index of the reference feature point, the final preferred factor of the reference feature point is obtained; according to the final preferred factor of all feature points in the target radiotherapy image and the number of feature points of each brain radiotherapy image, the priority matching degree of the target radiotherapy image is obtained; according to the priority matching degree of each brain radiotherapy image, the brain radiotherapy images are paired; and brain tumors are located according to the matching results of brain radiotherapy images. The present invention can obtain a more accurate image registration result, thereby obtaining a more accurate three-dimensional model, and can make tumor positioning more accurate.

[0109] See also Figure 2 , which shows the second object of the present invention, a structural block diagram of a tumor localization system based on radiotherapy images, the system includes the following modules:

[0110] An image acquisition module 101 is used to acquire all brain radiotherapy images;

[0111] The final optimization factor analysis module 102 is used to select any brain radiotherapy image as the target radiotherapy image, and select any feature point of the target radiotherapy image as the reference feature point; obtain the initial optimization factor of the reference feature point according to the gradient difference and distance distribution between the reference feature point and the feature point closest to the corresponding position of the reference feature point in all preset neighboring images; obtain the density feature index of the reference feature point according to the grayscale distribution in the preset first neighborhood of the reference feature point; obtain the direction feature index of the reference feature point according to the gradient distribution of the pixel points in the preset first neighborhood of the reference feature point; obtain the tissue feature index of the reference feature point according to the distribution of other feature points in the preset second neighborhood of the reference feature point; obtain the final optimization factor of the reference feature point according to the density feature index, direction feature index and tissue feature index of the reference feature point;

[0112] The image pairing module 103 is used to obtain the priority pairing degree of the target radiotherapy image according to the final preferred factors of all feature points in the target radiotherapy image and the number of feature points of each brain radiotherapy image; and pair the brain radiotherapy images according to the priority pairing degree of each brain radiotherapy image;

[0113] The tumor localization module 104 is used to localize the brain tumor according to the brain radiotherapy image matching result.

[0114] The third objective of an embodiment of the present invention is to provide a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned tumor localization method based on radiotherapy images when executing the computer program.

[0115] The fourth objective of an embodiment of the present invention is to provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned tumor localization method based on radiotherapy images are implemented.

[0116] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0117] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A tumor localization method based on radiotherapy images, characterized in that: The method comprises: Obtain all brain radiotherapy images; Select any brain radiotherapy image as the target radiotherapy image, and select any feature point of the target radiotherapy image as the reference feature point; obtain the initial optimization factor of the reference feature point according to the gradient difference and distance distribution between the reference feature point and the feature point closest to the corresponding position of the reference feature point in all preset neighboring images; obtain the density feature index of the reference feature point according to the grayscale distribution in the preset first neighborhood of the reference feature point; obtain the direction feature index of the reference feature point according to the gradient distribution of the pixel points in the preset first neighborhood of the reference feature point; obtain the tissue feature index of the reference feature point according to the distribution of other feature points in the preset second neighborhood of the reference feature point; obtain the final optimization factor of the reference feature point according to the density feature index, the direction feature index and the tissue feature index of the reference feature point; Obtaining a priority matching degree of the target radiotherapy image according to the final preferred factors of all feature points in the target radiotherapy image and the number of feature points of each brain radiotherapy image; pairing the brain radiotherapy images according to the priority matching degree of each brain radiotherapy image; Brain tumor localization is performed based on the matching results of brain radiotherapy images.

2. A method for tumor localization based on radiotherapy images according to claim 1, characterized in that: The method for obtaining the initial optimization factor includes: The feature point in each preset neighboring image that is closest to the corresponding position of the reference feature point is used as the feature point to be compared in each preset neighboring image; The initial optimal factor is obtained according to the initial optimal factor calculation formula, and the initial optimal factor calculation formula is as follows: Where μ represents the initial optimization factor of the reference feature point; I represents the number of preset neighboring images; T0 represents the gradient value of the reference feature point; T i represents the gradient value of the feature point to be compared in the i-th preset neighboring image; θ0 represents the gradient direction angle of the reference feature point, where the gradient direction angle is the angle between the gradient direction of the feature point and the horizontal direction; θ i represents the gradient direction angle of the feature point to be compared in the i-th preset neighboring image; R i G represents the distance between the corresponding position of the reference feature point and the feature point to be compared in the i-th preset neighboring image; i represents the image interval between the i-th preset adjacent image and the target radiotherapy image; || represents the absolute value function; exp[] represents the exponential function with a natural constant as the base.

3. The method for tumor localization based on radiotherapy images according to claim 1, characterized in that: The method for obtaining the density characteristic index includes: The density characteristic index is obtained according to the density characteristic index calculation formula, and the density characteristic index calculation formula is as follows: Where ρ represents the density characteristic index of the reference feature point; J represents the number of pixels in the first neighborhood of the reference feature point; H0 represents the gray value of the reference feature point; H j Represents the gray value of the jth pixel in the preset first neighborhood of the reference feature point; Indicates the grayscale mean of the pixels in the first neighborhood preset by the reference feature point; represents the grayscale mean of the target radiotherapy image; norm( ) represents the normalization function.

4. The method for tumor localization based on radiotherapy images according to claim 1, characterized in that: The method for obtaining the directional characteristic index includes: Obtaining a gradient histogram of all feature points in a second area preset by the reference feature point; Sort all the gradient histogram columns in the gradient histogram according to their height from large to small; The directional feature index of the reference feature point is obtained according to the gradient distribution of the feature points in the preset second neighborhood. The calculation formula is as follows: Where ω represents the directional characteristic index of the reference feature point; W max represents the peak value of the gradient histogram; K represents the number of gradient histogram columns in the gradient histogram; W k Represents the height of the kth histogram in the gradient histogram; W k+1 Represents the height of the k+1th histogram in the gradient histogram; norm[] represents the normalization function.

5. The method for tumor localization based on radiotherapy images according to claim 1, characterized in that: The method for obtaining the tissue characteristic index includes: Selecting a preset number of feature points closest to the reference feature points in a preset second area of ​​the reference feature points as target feature points; The organizational characteristic index is obtained according to the organizational characteristic index calculation formula, and the organizational characteristic index calculation formula is as follows: In the formula, δ represents the tissue characteristic index of the reference feature point; V represents the preset number of target feature points; R v represents the distance between the vth target feature point and the reference feature point; σ(R v ) represents the distance variance between the target feature point and the reference feature point; F represents the number of feature points in the preset second neighborhood of the reference feature point; G represents the number of feature points in the target radiotherapy image; norm[ ] represents the normalization function; exp( ) represents the exponential function with a natural constant as the base.

6. The method for tumor localization based on radiotherapy images according to claim 1, characterized in that: The method for obtaining the final optimization factor includes: The final preferred factor is obtained according to the final preferred factor calculation formula, and the final preferred factor calculation formula is as follows: In the formula, μ′ represents the final optimization factor of the reference feature point; μ represents the initial optimization factor of the reference feature point; ρ represents the density characteristic index of the reference feature point; ω represents the direction characteristic index of the reference feature point; δ represents the tissue characteristic index of the reference feature point.

7. The method for tumor localization based on radiotherapy images according to claim 1, characterized in that: The method for obtaining the priority matching degree includes: The priority matching degree is obtained according to the priority matching degree calculation formula, and the priority matching degree calculation formula is as follows: In the formula, γ represents the priority pairing degree of the target radiotherapy image; represents the final optimal factor mean of all feature points in the target radiotherapy image; P represents the number of feature points in the target radiotherapy image; P max Indicates the number of feature points in the brain radiotherapy image with the largest number of feature points.

8. A tumor localization system based on radiotherapy images, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the tumor localization method based on radiotherapy images as described in any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a tumor localization method based on radiotherapy images as described in any one of claims 1 to 7 are implemented.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the tumor localization method based on radiotherapy images as described in any one of claims 1 to 7 are implemented.

Images