A method for improving the accuracy of tumor radiotherapy target area delineation

Abstract

The present invention relates to the technical field of tumor image processing, and particularly relates to a method for improving the accuracy of tumor radiotherapy target area delineation. The present invention obtains the edge connection lines connecting the central pixel points and different edge pixel points within the possible tumor regions; according to the gray-scale distribution of the pixel points on each edge connection line, obtains the pixel gray-scale uniformity of each edge connection line, and obtains the connection line clustering clusters; and then analyzes the position distribution characteristics of the edge connection lines within different connection line clustering clusters and the number of connection line clustering clusters, to obtain the overall gray-scale uniformity of each possible tumor region; combines the gray-scale distribution of the boundary pixel points within each possible tumor region and the boundary morphological characteristics of each possible tumor region, to obtain the tumor credibility of each possible tumor region; obtains the optimal tumor region, and delineates the tumor target area. The present invention improves the accuracy of target area delineation by obtaining the accurate gray-scale range of the tumor region.

Description

Technical Field

[0001] The present invention relates to the technical field of tumor image processing, and particularly relates to a method for improving the accuracy of tumor radiotherapy target area delineation. Background Art

[0002] The tumor radiotherapy target area refers to the area that needs to be irradiated during tumor radiotherapy. The delineation of the tumor radiotherapy target area is a key step in the formulation of the radiotherapy plan. Accurate target area delineation can ensure that all tumor tissues can receive sufficient dose of irradiation, kill tumor cells to the greatest extent, and at the same time minimize the irradiation of surrounding normal tissues and organs to protect normal tissues.

[0003] In the prior art, the tumor target area is divided based on thresholds, and semi-automatic target area delineation technology is used for delineation. However, due to the low contrast between some tumor tissues and normal tissues, such as in the magnetic resonance images of some brain tumors, the signal difference between the tumor and the surrounding edema tissue is not obvious, and it is difficult to clearly identify the tumor boundary, resulting in poor accuracy of target area delineation. Summary of the Invention

[0004] In order to solve the technical problems of low contrast between some tumor tissues and normal tissues and poor accuracy of target area delineation, the purpose of the present invention is to provide a method for improving the accuracy of tumor radiotherapy target area delineation. The specific technical solution adopted is as follows:

[0005] The present invention proposes a method for improving the accuracy of tumor radiotherapy target area delineation. The method includes:

[0006] Obtain the grayscale image of the patient's brain MRI;

[0007] According to the grayscale distribution characteristics of the pixel points in the grayscale image, obtain multiple possible tumor regions; for any possible tumor region, obtain the edge connections of the central pixel point and different edge pixel points; according to the grayscale distribution of the pixel points on each edge connection, obtain the pixel grayscale uniformity of each edge connection; according to the pixel grayscale uniformity of each edge connection, cluster all edge connections to obtain connection clustering clusters; according to the position distribution characteristics of the edge connections within different connection clustering clusters, the pixel grayscale uniformity of all edge connections, and the number of connection clustering clusters, obtain the overall grayscale uniformity of each possible tumor region.

[0008] According to the grayscale distribution of the boundary pixel points within each possible tumor region, obtain the boundary clarity of each possible tumor region; according to the boundary morphological characteristics, overall grayscale uniformity, and boundary clarity of each possible tumor region, obtain the tumor credibility of each possible tumor region; according to the grayscale values and tumor credibility of the pixel points in different possible tumor regions, obtain the optimal tumor region.

[0009] The tumor target area is delineated based on the optimal tumor area.

[0010] Furthermore, the method for obtaining the possible tumor area includes:

[0011] The Otsu algorithm is used to obtain the optimal segmentation threshold in the grayscale image. If the grayscale value of a pixel point is greater than the optimal segmentation threshold, the corresponding pixel point is regarded as a possible tumor pixel point. Connected component detection is performed on all possible tumor pixel points to obtain multiple pixel point connected components as the possible tumor areas.

[0012] Furthermore, the method for obtaining the pixel grayscale uniformity includes:

[0013] The grayscale value sequence of pixel points on each edge connection line is obtained in the position order. If there is a grayscale value in the grayscale value sequence that is greater than or less than the adjacent grayscale values before and after, the corresponding grayscale value is regarded as the grayscale extreme value.

[0014] The average difference between all adjacent grayscale values in the grayscale value sequence is obtained, the product of the average difference and the number of grayscale extreme values is calculated, and negative correlation mapping is performed as the pixel grayscale uniformity of each edge connection line.

[0015] Furthermore, the method for obtaining the overall grayscale uniformity includes:

[0016] According to the position distribution characteristics of the edge connection lines within each connection cluster, the connection aggregation degree of each connection cluster is obtained.

[0017] According to the connection aggregation degree of different connection clusters within each possible tumor area, the pixel grayscale uniformity of all edge connection lines, and the number of connection clusters, the overall grayscale uniformity of each possible tumor area is obtained. The connection aggregation degree and the pixel grayscale uniformity are both positively correlated with the overall grayscale uniformity, and the number of connection clusters is negatively correlated with the overall grayscale uniformity.

[0018] Furthermore, the method for obtaining the connection aggregation degree includes:

[0019] For each connection cluster, any edge connection line is selected as the starting point, and all edge connection lines are numbered in the clockwise direction. If there are other edge connection lines within the neighborhood range of an edge connection line in the clockwise direction, the corresponding edge connection line is regarded as the adjacent edge connection line. All adjacent edge connection lines of each edge connection line are obtained to form an adjacent edge connection line group. The cumulative sum of the differences in the numbers between adjacent edge connection lines within all adjacent edge connection line groups is obtained as the connection clustering degree.

[0020] Furthermore, the method for obtaining the boundary clarity includes:

[0021] Obtain the variance of the gray values of all boundary pixel points within each possible tumor region as the gray fluctuation feature;

[0022] Obtain the average gray gradient of all boundary pixel points within each possible tumor region as the gray change rate;

[0023] Obtain the ratio between the gray change rate and the gray fluctuation feature of each possible tumor region as the boundary clarity of each possible tumor region.

[0024] Furthermore, the method for obtaining the tumor credibility includes:

[0025] Obtain the difference between the 8-chain code values of each boundary pixel point and its adjacent boundary pixel points, and perform normalization. If the normalized result is greater than the preset difference threshold, regard the corresponding boundary pixel point as an irregular point; form the corresponding connection lines for all adjacent irregular points as irregular lines;

[0026] Obtain the product of the average value of the number of irregular points on all irregular lines and the number of irregular points as the first product; obtain the ratio of the first product to the average value of the number of boundary pixel points between all adjacent irregular lines as the irregular density;

[0027] According to the overall gray uniformity, boundary clarity, and irregular density of each possible tumor region, obtain the tumor credibility of each possible tumor region. The irregular density is positively correlated with the tumor credibility, and both the overall gray uniformity and the boundary clarity are negatively correlated with the tumor credibility.

[0028] Furthermore, the method for obtaining the best tumor region includes:

[0029] Screen out the target tumor regions according to the tumor credibility of each possible tumor region;

[0030] According to the tumor credibility of the target tumor regions, adjust the initial minimum gray value and the initial maximum gray value of the pixel points in all target tumor regions to obtain the minimum gray transformation value and the maximum gray transformation value;

[0031] In the gray image, obtain the corresponding region within the range formed by the minimum gray transformation value and the maximum gray transformation value as the best tumor region.

[0032] Furthermore, the method for obtaining the target tumor regions includes:

[0033] If the tumor credibility of a possible tumor region is greater than the preset credibility threshold, regard the corresponding possible tumor region as a target tumor region.

[0034] Furthermore, the method for obtaining the minimum gray transformation value and the maximum gray transformation value includes:

[0035] Obtain the mean tumor credibility of all target tumor regions as the average credibility level;

[0036] Obtain the difference between the positive integer 1 and the average credibility level as the first difference; calculate the product of the initial minimum gray value and the first difference and round down to obtain the minimum gray value for gray value transformation;

[0037] Obtain the difference between the standard maximum gray value and the initial maximum gray value as the second difference; calculate the product of the second difference and the average credibility level as the first product; calculate the sum of the initial maximum gray value and the first product and round down to obtain the maximum gray value for gray value transformation.

[0038] The present invention has the following beneficial effects:

[0039] Based on the gray value distribution characteristics of pixel points in a gray image, the present invention obtains multiple possible tumor regions to initially locate the regions in the image that may be related to tumors; in order to quantify the shape, size, and distribution of tumor regions, for any possible tumor region, obtain the edge connection lines connecting the central pixel point and different edge pixel points; according to the gray value distribution of pixel points on each edge connection line, obtain the pixel gray value uniformity of each edge connection line to reflect the gray value change of the edge region; according to the pixel gray value uniformity of each edge connection line, cluster all edge connection lines to obtain connection line clustering clusters. Through clustering, edge connection lines with similar gray value characteristics can be grouped for discussion; according to the position distribution characteristics of edge connection lines within different connection line clustering clusters, the pixel gray value uniformity of all edge connection lines, and the number of connection line clustering clusters, obtain the overall gray value uniformity of each possible tumor region to more comprehensively quantify the gray value distribution of the entire possible tumor region; according to the gray value distribution of boundary pixel points within each possible tumor region, obtain the boundary clarity of each possible tumor region to analyze the boundary characteristics of the possible tumor region, which helps to evaluate the tumor boundary characteristics of the possible tumor region; according to the boundary morphological characteristics, overall gray value uniformity, and boundary clarity of each possible tumor region, obtain the tumor credibility of each possible tumor region to evaluate the possibility of a certain region being a tumor; according to the gray value and tumor credibility of pixel points in different possible tumor regions, obtain the optimal tumor region to make the tumor region more prominent in the image; outline the tumor target area. The present invention improves the accuracy of target area outlining by obtaining the accurate gray value range of the tumor region. Description of the Drawings

[0040] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0041] Figure 1 It is a flowchart of a method for improving the accuracy of tumor radiotherapy target area delineation provided by an embodiment of the present invention.

[0042] Figure 2 It is a flowchart of a method for obtaining tumor credibility provided by an embodiment of the present invention.

[0043] Figure 3 It is a flowchart of a method for obtaining grayscale transformation values provided by an embodiment of the present invention. Detailed implementation manners

[0044] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following, in combination with the accompanying drawings and preferred embodiments, details the specific implementation manners, structures, features, and effects of a method for improving the accuracy of tumor radiotherapy target area delineation according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0045] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0046] The following specifically describes the specific solution of a method for improving the accuracy of tumor radiotherapy target area delineation provided by the present invention in combination with the accompanying drawings.

[0047] Please refer to Figure 1 , which shows a flowchart of a method for improving the accuracy of tumor radiotherapy target area delineation provided by an embodiment of the present invention. The specific method includes:

[0048] Step S1: Obtain the grayscale image of the patient's brain MRI.

[0049] In the embodiment of the present invention, in order to accurately delineate the tumor target area, it is necessary to identify the accurate tumor area; first, perform an MRI examination on the patient to obtain the patient's brain MRI image and ensure the quality of the obtained image;

[0050] It should be noted that, for the convenience of subsequent image processing, in an embodiment of the present invention, the brain MRI image is grayscale processed to obtain a grayscale image of the brain MRI, and then the processed grayscale image is analyzed. It should be noted that grayscale processing, as a well-known technical means to those skilled in the art, can be specifically set according to the specific implementation scenario. In an embodiment of the present invention, the mean grayscale algorithm is used to obtain the grayscale image, which simplifies the image information, improves the operation speed, and helps to improve the accuracy of image recognition.

[0051] Step S2: According to the grayscale distribution characteristics of the pixel points in the grayscale image, obtain multiple possible tumor regions; for any possible tumor region, obtain the edge connections connecting the central pixel point and different edge pixel points; according to the grayscale distribution of the pixel points on each edge connection, obtain the pixel grayscale uniformity of each edge connection; according to the pixel grayscale uniformity of each edge connection, cluster all the edge connections to obtain connection clustering clusters; according to the position distribution characteristics of the edge connections within different connection clustering clusters, the pixel grayscale uniformity of all the edge connections, and the number of connection clustering clusters, obtain the overall grayscale uniformity of each possible tumor region.

[0052] Since tumor cells are metabolically active and have a high water content, in the grayscale image, they usually show a lighter grayscale, while the non-tumor regions around have a slightly darker grayscale. Therefore, according to the grayscale distribution characteristics of the pixel points in the grayscale image, multiple possible tumor regions are obtained.

[0053] Preferably, in an embodiment of the present invention, the method for obtaining possible tumor regions includes:

[0054] Use the Otsu algorithm to obtain the optimal segmentation threshold in the grayscale image. If the grayscale value of a pixel point is greater than the optimal segmentation threshold, the corresponding pixel point is regarded as a possible tumor pixel point; perform connected component detection on all possible tumor pixel points to obtain multiple pixel point connected components as possible tumor regions.

[0055] It should be noted that the specific Otsu algorithm and connected component algorithm are well-known technical means to those skilled in the art and will not be elaborated here.

[0056] In order to quantify the shape, size, and distribution of the tumor region, for any possible tumor region, obtain the edge connections connecting the central pixel point and different edge pixel points.

[0057] Due to the differences in cell components, blood vessel distribution, and the presence or absence of necrosis and cystic changes within the tumor, the grayscale values in different regions are inconsistent, while the grayscale distribution of the brain tissue in the non-tumor region is relatively uniform. The grayscale change characteristics of the pixel points are reflected by the grayscale distribution of the pixel points on each edge connection. According to the grayscale distribution of the pixel points on each edge connection, obtain the pixel grayscale uniformity of each edge connection.

[0058] Preferably, in one embodiment of the present invention, the method for obtaining pixel gray-scale uniformity includes:

[0059] Obtain the gray-scale value sequence of pixel points on each edge connection line in position order. If there is a gray-scale value in the gray-scale value sequence that is greater than or less than the adjacent gray-scale values before and after, take the corresponding gray-scale value as the gray-scale extreme value;

[0060] Obtain the average difference between all adjacent gray-scale values in the gray-scale value sequence, calculate the product of the average difference and the number of gray-scale extreme values, and perform a negative correlation mapping, which is used as the pixel gray-scale uniformity of each edge connection line.

[0061] In one embodiment of the present invention, the formula for pixel gray-scale uniformity is expressed as:

[0062] ;

[0063] Wherein, represents the pixel gray-scale uniformity of the th edge connection line; represents the number of gray-scale extreme values of the th edge connection line; represents the difference between the th gray-scale value and the next adjacent gray-scale value; represents the number of gray-scale values on the gray-scale value sequence corresponding to the th edge connection line; represents the exponential function with the natural constant as the base.

[0064] In the formula for pixel gray-scale uniformity, is negatively correlated through the exponential function with the natural constant as the base; represents the average difference between all adjacent gray-scale values on the gray-scale value sequence corresponding to the th edge connection line. The larger the average difference, the larger the gray-scale value between adjacent pixel points, and the more uneven the gray-scale distribution of pixel points; the larger the number of gray-scale extreme values of the th edge connection line, the greater the gray-scale fluctuation, and the smaller the pixel gray-scale uniformity.

[0065] Through clustering, edge connection lines with similar gray-scale distribution characteristics can be grouped into one category, and a large number of edge connection lines can be divided into a few clusters, simplifying the analysis process; according to the pixel gray-scale uniformity of each edge connection line, all edge connection lines are clustered to obtain a connection line clustering cluster.

[0066] It should be noted that, in an embodiment of the present invention, the method for obtaining the connection clustering clusters includes: performing K-means clustering on all edge connections according to the pixel gray-scale uniformity of each edge connection to obtain multiple connection clustering clusters; in other embodiments of the present invention, clustering can also be performed by existing clustering methods such as mean shift clustering, and the specific means are well-known technical means to those skilled in the art and will not be elaborated here.

[0067] The position distribution characteristics of the edge connections within different connection clustering clusters reflect the shape and structural complexity of the tumor region, and the pixel gray-scale uniformity reflects the gray-scale change situation of the possible tumor regions. The greater the pixel gray-scale uniformity, the more uniform the pixel gray-scale distribution, and the more likely it is a normal tissue region; the larger the number of connection clustering clusters, the greater the difference in pixel gray-scale uniformity of different edge connections, and the more uneven the gray-scale of the possible tumor regions; based on the position distribution characteristics of the edge connections within different connection clustering clusters, the pixel gray-scale uniformity of all edge connections, and the number of connection clustering clusters, the overall gray-scale uniformity of each possible tumor region is obtained.

[0068] Preferably, in an embodiment of the present invention, the method for obtaining the overall gray-scale uniformity includes:

[0069] Based on the position distribution characteristics of the edge connections within each connection clustering cluster, the connection aggregation degree of each connection clustering cluster is obtained;

[0070] Preferably, in an embodiment of the present invention, the method for obtaining the connection aggregation degree includes:

[0071] For each connection clustering cluster, select any edge connection as the starting point and number all edge connections in the clockwise direction. If there are other edge connections within the neighborhood range of the edge connection in the clockwise direction, the corresponding edge connection is used as the neighboring edge connection, and all neighboring edge connections of each edge connection are obtained to form a neighboring edge connection group; obtain the cumulative sum of the differences in numbers between adjacent edge connections within all neighboring edge connection groups as the connection clustering degree.

[0072] It should be noted that, in an embodiment of the present invention, the neighborhood range is based on the edge pixel points of the edge connection on the boundary, and the range where the Euclidean distance between the corresponding edge pixel points of other edge connections is less than a preset distance is obtained, where the preset distance is 0.3; in other embodiments of the present invention, the neighborhood range can be specifically set according to specific situations and will not be limited and elaborated here.

[0073] According to the connection aggregation degree of different connection clustering clusters within each possible tumor region, the pixel gray level uniformity of all edge connections, and the number of connection clustering clusters, the overall gray level uniformity of each possible tumor region is obtained. The connection aggregation degree and the pixel gray level uniformity are both positively correlated with the overall gray level uniformity, and the number of connection clustering clusters is negatively correlated with the overall gray level uniformity.

[0074] In an embodiment of the present invention, the formula for the overall gray level uniformity is expressed as:

[0075] ;

[0076] Wherein, represents the overall gray level uniformity of each possible tumor region; represents the pixel gray level uniformity of the th edge connection; represents the number of edge connections in the possible tumor region; represents the th connection aggregation degree of the connection clustering cluster; represents the number of connection clustering clusters.

[0077] In the formula for the overall gray level uniformity, represents calculating the average value of the pixel gray level uniformity of all edge connections in the possible tumor region; represents the average value of the connection aggregation degrees of all connection clustering clusters in the possible tumor region. The larger the pixel gray level uniformity and the connection aggregation degree, the more uniform the gray level distribution on the edge connection, indicating that the overall gray level uniformity is larger, showing a positive correlation. The more the number of connection clustering clusters, the greater the difference in the gray level uniformity of different edge connections, and the worse the overall gray level uniformity, showing a negative correlation.

[0078] Step S3: According to the gray level distribution of the boundary pixel points within each possible tumor region, obtain the boundary clarity of each possible tumor region; according to the boundary morphological characteristics, overall gray level uniformity, and boundary clarity of each possible tumor region, obtain the tumor credibility of each possible tumor region; according to the gray level values and tumor credibility of the pixel points in different possible tumor regions, obtain the optimal tumor region.

[0079] The boundaries of benign tumors may be relatively clear, with a certain demarcation from surrounding tissues, but they may also appear as a gradual transition of grayscale at the boundaries on grayscale images. The boundaries of malignant tumors are usually blurred and grow in an invasive manner. The grayscale boundary with the surrounding normal tissues is not obvious, and it is difficult to accurately define the scope of the tumor. The edge of the tumor may experience sudden changes in grayscale values ​​or irregular grayscale transition zones. The boundaries of normal brain tissues are relatively clear and regular in shape. There is generally no blurred or abnormal boundary shape, and the grayscale value changes are relatively stable. The boundary clarity of each possible tumor area is obtained based on the grayscale distribution of boundary pixels in each possible tumor area.

[0080] Preferably, in one embodiment of the present invention, the method for acquiring boundary definition includes:

[0081] Obtain the grayscale value variance of all boundary pixels in each possible tumor area as the grayscale fluctuation feature;

[0082] The grayscale gradient mean of all boundary pixels in each possible tumor area is obtained as the grayscale change rate;

[0083] The ratio between the grayscale change rate and the grayscale fluctuation characteristics of each possible tumor region is obtained as the boundary clarity of each possible tumor region.

[0084] The larger the grayscale value variance, the more uneven the grayscale distribution, the larger the grayscale fluctuation characteristics, and the more regular the boundary shape. Conversely, the more uniform the grayscale distribution, the smaller the grayscale fluctuation characteristics, and the more irregular the boundary. The larger the grayscale gradient mean, the more drastic the grayscale change at the boundary, the greater the contrast at the boundary, and the clearer the boundary. Therefore, the smaller the grayscale fluctuation characteristics, the greater the grayscale change rate, and the greater the boundary clarity.

[0085] It should be noted that when calculating the ratio between the grayscale change rate and the grayscale fluctuation characteristics, in order to avoid the denominator of the formula being 0 and the formula being meaningless, a threshold value, such as 0.01, can be added artificially. The specific means are technical means well known to technical personnel in this field and will not be elaborated here.

[0086] The boundary morphology of the brain tumor area is irregular, and the boundary is fuzzy relative to the normal brain tissue area; the overall grayscale uniformity reflects the uniformity of the pixel grayscale values ​​in the entire possible tumor area. The greater the overall grayscale uniformity, the smaller the possibility of a tumor; therefore, the tumor credibility of each possible tumor area is obtained based on the boundary morphological characteristics, overall grayscale uniformity and boundary clarity of each possible tumor area.

[0087] Preferably, in one embodiment of the present invention, the method for obtaining the tumor credibility is as follows: Figure 2 , which shows a flow chart of a method for obtaining tumor credibility, including:

[0088] Step S201: Obtain the differences between the 8-chain code values of each boundary pixel point and its adjacent boundary pixel points, and perform normalization. If the normalized result is greater than the preset difference threshold, regard the corresponding boundary pixel point as an irregular point; form the corresponding connection lines for all adjacent irregular points as irregular lines.

[0089] The 8-chain code of a boundary pixel point can represent the boundary trend of the possible tumor region. The smaller the difference between the 8-chain code values of adjacent boundary pixel points, the smaller the direction change between adjacent boundary pixel points, and the relatively more regular the boundary. On the contrary, the larger the difference, the larger the direction change between adjacent boundary pixel points, and the relatively more irregular the boundary, indicating that there may be a tumor in the region, which is marked as an irregular point.

[0090] It should be noted that in an embodiment of the present invention, the size of the preset difference threshold is 0.1; in other embodiments of the present invention, the size of the preset difference threshold can be specifically set according to specific situations, and no limitation and elaboration are made here.

[0091] Step S202: Obtain the product of the mean value of the number of irregular points on all irregular lines and the number of irregular points as the first product; obtain the ratio of the first product to the mean value of the number of boundary pixel points between all adjacent irregular lines as the irregular density.

[0092] The more the number of irregular points, the more the area occupied by the irregular lines on the boundary, the fewer the boundary pixel points between adjacent irregular lines, the more discontinuous the boundary, and the greater the tumor credibility.

[0093] Step S203: According to the overall gray level uniformity, boundary clarity, and irregular density of each possible tumor region, obtain the tumor credibility of each possible tumor region. The irregular density is positively correlated with the tumor credibility, and both the overall gray level uniformity and the boundary clarity are negatively correlated with the tumor credibility.

[0094] In an embodiment of the present invention, the formula for the tumor credibility is expressed as:

[0095] ;

[0096] Wherein, represents the tumor credibility of the th possible tumor region; represents the overall gray level uniformity of the th possible tumor region; represents the boundary clarity of the th possible tumor region; represents the mean value of the number of irregular points on all irregular lines in the th possible tumor region; represents the The number of irregular points within a possible tumor region; Denotes the average number of boundary pixel points between all adjacent irregular lines; Denotes a normalization function; Denotes the exponential function with the natural constant as the base.

[0097] In the formula for tumor credibility, Denotes obtaining the product of the mean and the number of irregular points on all irregular lines within the th possible tumor region as the first product; obtaining the ratio of the first product to the average number of boundary pixel points between all adjacent irregular lines, i.e., the irregular density. The greater the irregular density, the more irregular points there are, the more area the irregular lines occupy on the boundary, the fewer boundary pixel points between adjacent irregular lines, the less continuous the boundary, and the greater the tumor credibility; performing a negative correlation mapping on

[0098] through the exponential function with the natural constant as the base. The greater the boundary clarity and the greater the overall gray uniformity, the smaller the possibility of being a tumor and the smaller the tumor credibility.

[0099] Preferably, in an embodiment of the present invention, for the method of obtaining the optimal tumor region, please refer to Figure 3 , which shows a flowchart of a method for obtaining the optimal tumor region, including:

[0100] Step S301: Screening out the target tumor region according to the tumor credibility of each possible tumor region.

[0101] Preferably, in an embodiment of the present invention, the method for obtaining the target tumor region includes:

[0102] If the tumor credibility of a possible tumor region is greater than the preset credibility threshold, taking the corresponding possible tumor region as the target tumor region.

[0103] It should be noted that, in an embodiment of the present invention, the size of the preset credibility threshold is 0.8; in other embodiments of the present invention, the size of the preset credibility threshold can be specifically set according to specific circumstances, which will not be limited and elaborated herein.

[0104] Step S302: According to the tumor credibility of the target tumor area, the grayscale initial minimum value and the grayscale initial maximum value of the pixel points in all the target tumor areas are adjusted to obtain the grayscale transformation minimum value and the grayscale transformation maximum value.

[0105] Preferably, in one embodiment of the present invention, the method for obtaining the grayscale transformation minimum value and the grayscale transformation maximum value includes:

[0106] The mean tumor credibility of all target tumor regions is obtained as the average credibility level;

[0107] Obtain the difference between the positive integer 1 and the average confidence level as the first difference; calculate the product of the grayscale initial minimum value and the first difference, and round it down to obtain the grayscale transformation minimum value;

[0108] The difference between the grayscale standard maximum value and the grayscale initial maximum value is obtained as the second difference; the product of the second difference and the average credible level is calculated as the first product; the sum of the grayscale initial maximum value and the first product is calculated and rounded down as the grayscale transformation maximum value.

[0109] In one embodiment of the present invention, the formulas for the grayscale transformation minimum value and the grayscale transformation maximum value are expressed as:

[0110] ;

[0111] ;

[0112] in, Represents the minimum value of grayscale transformation; Indicates the maximum value of grayscale transformation; Indicates the initial minimum value of grayscale; Grayscale initial maximum value; It represents the mean tumor credibility of all target tumor areas, that is, the average credibility level; Indicates the maximum value of grayscale standard; Indicates the floor symbol.

[0113] In the formulas of the maximum grayscale transformation and the minimum grayscale transformation, the greater the credibility, the larger the maximum grayscale transformation is adjusted, the smaller the minimum grayscale transformation is adjusted, and the larger the grayscale range is, so as to avoid missing tumor pixels with smaller or larger grayscales.

[0114] It should be noted that the grayscale standard maximum value is the maximum value within the grayscale value standard range of the grayscale image, that is, 255.

[0115] Step S303: In the grayscale image, a corresponding area within the range formed by the grayscale transformation minimum value and the grayscale transformation maximum value is obtained as the optimal tumor area.

[0116] Step S4: outlining the tumor target area based on the optimal tumor area.

[0117] The optimal tumor area reflects a more complete tumor region and contributes to more accurate target delineation.

[0118] Using the commonly used semi-automatic target delineation technology to delineate the tumor target area of ​​the patient's optimal tumor area helps to improve the delineation accuracy of the tumor target area. The specific means are well known to those skilled in the art and will not be described in detail here.

[0119] In summary, for any possible tumor area, the present invention obtains edge lines connecting the central pixel and different edge pixels; according to the grayscale distribution of the pixels on each edge line, the pixel grayscale uniformity of each edge line is obtained, and the line clustering cluster is obtained; according to the position distribution characteristics of the edge lines in different line clustering clusters, the pixel grayscale uniformity of all edge lines and the number of line clustering clusters, the overall grayscale uniformity of each possible tumor area is obtained; according to the grayscale distribution of the boundary pixels in each possible tumor area, the boundary clarity of each possible tumor area is obtained; combined with the analysis of the boundary morphological characteristics of each possible tumor area, the optimal tumor area is obtained, and the tumor target area is delineated. The present invention improves the accuracy of target area delineation by obtaining an accurate grayscale range of the tumor area.

[0120] 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.

[0121] 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 method for improving the accuracy of tumor radiotherapy target area delineation, characterized in that: The method comprises: Obtain grayscale images of the patient's brain magnetic resonance imaging; According to the grayscale distribution characteristics of the pixels in the grayscale image, multiple possible tumor areas are obtained; for any possible tumor area, edge lines connecting the central pixel and different edge pixels are obtained; according to the grayscale distribution of the pixels on each edge line, the pixel grayscale uniformity of each edge line is obtained; according to the pixel grayscale uniformity of each edge line, all edge lines are clustered to obtain line clustering clusters; according to the position distribution characteristics of the edge lines in different line clustering clusters, the pixel grayscale uniformity of all edge lines and the number of line clustering clusters, the overall grayscale uniformity of each possible tumor area is obtained; According to the grayscale distribution of the boundary pixels in each possible tumor area, the boundary clarity of each possible tumor area is obtained; according to the boundary morphological characteristics, overall grayscale uniformity and boundary clarity of each possible tumor area, the tumor credibility of each possible tumor area is obtained; according to the grayscale values ​​of the pixels and the tumor credibility in different possible tumor areas, the optimal tumor area is obtained; Tumor target volumes were delineated based on the optimal tumor region.

2. A method for improving the accuracy of tumor radiotherapy target area delineation according to claim 1, characterized in that: The method for obtaining the possible tumor area includes: The Otsu algorithm is used to obtain the optimal segmentation threshold in the grayscale image. If the grayscale value of a pixel is greater than the optimal segmentation threshold, the corresponding pixel is regarded as a possible tumor pixel. Connected domain detection is performed on all possible tumor pixels to obtain multiple pixel connected domains as possible tumor areas.

3. A method for improving the accuracy of tumor radiotherapy target area delineation according to claim 1, characterized in that: The method for obtaining pixel grayscale uniformity includes: Obtain the grayscale value sequence of the pixel points on each edge line in order of position. If there is a grayscale value in the grayscale value sequence that is greater than or less than the adjacent grayscale values, the corresponding grayscale value is taken as the grayscale extreme value. The difference mean between all adjacent gray values ​​in the gray value sequence is obtained, the product of the difference mean and the number of gray extreme values ​​is calculated, and a negative correlation mapping is performed as the pixel gray uniformity of each edge line.

4. The method for improving the accuracy of tumor radiotherapy target area delineation according to claim 1, characterized in that: The method for obtaining the overall grayscale uniformity includes: According to the position distribution characteristics of the edge links in each link cluster, the link aggregation degree of each link cluster is obtained; According to the connection aggregation of different connection clusters in each possible tumor area, the pixel grayscale uniformity of all edge connections and the number of connection clusters, the overall grayscale uniformity of each possible tumor area was obtained. The connection aggregation and pixel grayscale uniformity were positively correlated with the overall grayscale uniformity, and the number of connection clusters was negatively correlated with the overall grayscale uniformity.

5. A method for improving the accuracy of tumor radiotherapy target area delineation according to claim 4, characterized in that: The method for obtaining the connection concentration degree includes: For each link cluster, select any edge link as the starting point and number all edge links in a clockwise direction. If there are other edge links in the neighborhood of the edge link in the clockwise direction, the corresponding edge link will be used as the adjacent edge link, and all adjacent edge links of each edge link will be obtained to form an adjacent edge link group; the cumulative sum of the differences in the numbers of adjacent edge links in all adjacent edge link groups is obtained as the link clustering degree.

6. The method for improving the accuracy of tumor radiotherapy target area delineation according to claim 1, characterized in that: The method for obtaining the boundary clarity includes: Obtain the grayscale value variance of all boundary pixels in each possible tumor area as the grayscale fluctuation feature; The grayscale gradient mean of all boundary pixels in each possible tumor area is obtained as the grayscale change rate; The ratio between the grayscale change rate and the grayscale fluctuation characteristics of each possible tumor region is obtained as the boundary clarity of each possible tumor region.

7. The method for improving the accuracy of tumor radiotherapy target area delineation according to claim 1, characterized in that: The method for obtaining the tumor credibility includes: Obtain the difference between the 8-chain code value of each boundary pixel and the adjacent boundary pixel, and normalize it. If the normalized result is greater than the preset difference threshold, the corresponding boundary pixel is regarded as an irregular point; all adjacent irregular points are connected to form corresponding lines as irregular lines; The product of the mean value of the number of irregular points on all irregular lines and the number of irregular points is obtained as the first product; the ratio of the first product and the mean value of the number of boundary pixel points between all adjacent irregular lines is obtained as the irregular density; The tumor credibility of each possible tumor area is obtained according to the overall grayscale uniformity, boundary clarity and irregular density of each possible tumor area. The irregular density is positively correlated with the tumor credibility, while the overall grayscale uniformity and boundary clarity are negatively correlated with the tumor credibility.

8. The method for improving the accuracy of tumor radiotherapy target area delineation according to claim 1, characterized in that: The method for obtaining the optimal tumor region comprises: According to the tumor credibility of each possible tumor area, the target tumor area is screened out; According to the tumor credibility of the target tumor area, the initial minimum grayscale value and the initial maximum grayscale value of the pixel points in all the target tumor areas are adjusted to obtain the minimum grayscale transformation value and the maximum grayscale transformation value; In the grayscale image, the corresponding area within the range formed by the grayscale transformation minimum value and the grayscale transformation maximum value is obtained as the optimal tumor area.

9. A method for improving the accuracy of tumor radiotherapy target area delineation according to claim 8, characterized in that: The method for acquiring the target tumor area comprises: If the tumor credibility of the possible tumor region is greater than the preset credibility threshold, the corresponding possible tumor region is used as the target tumor region.

10. The method for improving the accuracy of tumor radiotherapy target area delineation according to claim 8, characterized in that: The method for obtaining the grayscale transformation minimum value and the grayscale transformation maximum value comprises: The mean tumor credibility of all target tumor regions is obtained as the average credibility level; Obtain the difference between the positive integer 1 and the average confidence level as the first difference; calculate the product of the grayscale initial minimum value and the first difference, and round it down to obtain the grayscale transformation minimum value; The difference between the grayscale standard maximum value and the grayscale initial maximum value is obtained as the second difference; the product of the second difference and the average credible level is calculated as the first product; the sum of the grayscale initial maximum value and the first product is calculated and rounded down as the grayscale transformation maximum value.

Images