Automatic Calculation Method and System for Standard Take-up Values ​​of DR Images

Abstract

This invention relates to the field of image analysis technology and discloses an automatic method and system for calculating the standard acquisition value of DR images. The method includes: acquiring DR images; performing grayscale fluctuation analysis on the DR images; calculating the degree of grayscale difference of each pixel; generating an attenuation fluctuation index representing the degree of local grayscale interference by artifacts; and analyzing the intensity of the impact on each pixel in the DR images based on the attenuation fluctuation index, generating an interference probability coefficient representing the probability of false attenuation of each pixel due to artifact contribution. This solution avoids the misjudgment of artifact regions and the omission of real lesion regions by traditional edge detection, improves the accuracy of region segmentation and standard acquisition value calculation, and enhances the analysis capability of DR images under complex artifact interference.

Description

Technical Field

[0001] This invention relates to the field of image analysis technology, specifically to a method and system for automatically calculating standard capture values ​​of DR images. Background Technology

[0002] When analyzing the degree of X-ray penetration attenuation in existing DR images, edge detection algorithms are typically used to extract the edges of the DR image, and after dividing the image into regions, the degree of X-ray penetration attenuation in each region is calculated to obtain a standard uptake value representing the degree of X-ray penetration attenuation in the DR image.

[0003] However, the above calculation method still has the following defects: the scattering artifacts in DR images can form false edges, and motion artifacts can cause the real tissue edges to become blurred, broken, or spatially misaligned. Under the interference of these two types of artifacts, traditional edge detection algorithms cannot effectively distinguish between real tissue edges and artifact edges, and are prone to misjudging artifact areas or missing real lesion areas, resulting in inaccurate regional localization and segmentation, causing deviations in the calculation of standard uptake values, and affecting the analysis of DR images. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an automatic calculation method and system for standard acquisition values ​​of DR images, thus solving the aforementioned problems.

[0005] The above-mentioned technical objective of the present invention is achieved through the following technical solution: Automatic calculation methods for standard acquisition values ​​of DR images include: Step S1: Acquire DR image, perform grayscale fluctuation analysis on DR image, calculate the degree of grayscale difference of each pixel, and generate attenuation fluctuation index representing the degree of local grayscale interference by artifacts. Step S2: Based on the attenuation fluctuation index, analyze the intensity of the impact on each pixel in the DR image and generate an interference probability coefficient representing the probability of false attenuation caused by artifact contribution for each pixel. Step S3: Based on the interference probability coefficient, suppress the features of the DR image and repair the broken edges to obtain an edge repair map that represents the actual tissue edge position and orientation. Step S4: Perform image region segmentation on the DR image based on the edge repair map to generate the real tissue region after eliminating artifact interference; Step S5: Calculate the grayscale value and interference probability coefficient of all pixels in the real tissue area to generate a standard uptake value that represents the degree of X-ray penetration attenuation in the real tissue area.

[0006] Furthermore, grayscale fluctuation analysis is performed on the DR images to calculate the degree of grayscale difference for each pixel, generating an attenuation fluctuation index that represents the degree of local grayscale interference from artifacts, including: For DR images, the degree of gray-level fluctuation within the neighborhood of each pixel is calculated, and a chaotic trend coefficient representing the degree of gray-level fluctuation is generated. The grayscale fluctuations in the neighborhood of each pixel are analyzed based on the chaotic trend coefficient to find the modulation ratio caused by artifacts and generate an attenuation fluctuation index that represents the degree of interference of local grayscale with artifacts.

[0007] Furthermore, based on the attenuation fluctuation index, the intensity of the impact on each pixel in the DR image is analyzed, and an interference probability coefficient representing the probability of false attenuation due to artifact contribution is generated for each pixel, including: Based on the decay fluctuation index, the cumulative intensity of each pixel affected by neighborhood artifacts is analyzed, and an artifact diffusion decay factor representing the degree of artifact energy propagation decay is generated. Based on the artifact diffusion attenuation factor, the spatial heterogeneity of background grayscale in the neighborhood of each pixel is analyzed, and a background depth coefficient representing the degree of background grayscale stretching caused by artifacts is generated.

[0008] Furthermore, based on the attenuation fluctuation index, the intensity of the impact on each pixel in the DR image is analyzed, and an interference probability coefficient representing the probability of false attenuation due to artifact contribution is generated for each pixel. This also includes: Based on the artifact diffusion attenuation factor and the background depth coefficient, the probability trend of the pixel's gray value being misjudged as tissue attenuation is calculated, and a false tendency index representing the pixel's susceptibility to being deceived by artifacts is generated. Based on the false tendency index, the impact on each pixel is analyzed, and an interference probability coefficient is generated to represent the probability of false attenuation caused by artifact contribution for each pixel.

[0009] Furthermore, based on the interference probability coefficient, the features of the DR image are suppressed, and the fracture edges are repaired to obtain an edge repair map representing the actual tissue edge location and orientation, including: Based on the interference probability coefficient, the gradient response of DR images severely contaminated by artifacts is suppressed, and a clean gradient intensity representing the gradient strength and direction after artifact suppression is generated.

[0010] Furthermore, features of the DR image are suppressed based on the interference probability coefficient, and fracture edges are repaired to obtain an edge repair map representing the location and orientation of the true tissue edge. This also includes: Based on the purification gradient intensity, trace and connect the broken edge segments, calculate the connection confidence between each pair of edge endpoints, and generate a bridging potential value that represents the overall continuity of the broken edge. Based on the bridging potential energy value, the fracture edge is bridged and reconstructed to generate an edge repair map that represents the actual tissue edge location and orientation.

[0011] Furthermore, based on the edge restoration map, image region segmentation is performed on the DR image to generate the true tissue region after eliminating artifact interference, including: Based on the edge restoration map and interference probability coefficient, the DR image is divided into regions to generate the real tissue region after eliminating artifact interference.

[0012] Furthermore, the grayscale values ​​and interference probability coefficients of all pixels within the real tissue region are calculated to generate a standard uptake value representing the degree of X-ray penetration attenuation in that real tissue region, including: Based on the grayscale values ​​and interference probability coefficients of all pixels within the real organization area, the effective weight of each pixel's contribution to the area attenuation is calculated, and a probability intensity representing the attenuation contribution of grayscale after being modulated by artifact probability is generated.

[0013] Furthermore, the grayscale values ​​and interference probability coefficients of all pixels within the real tissue region are calculated to generate a standard uptake value representing the degree of X-ray penetration attenuation in that real tissue region, which also includes: A co-analysis of probability intensity and spatial distribution of pixels within the real tissue region is performed to generate a standard uptake value representing the degree of X-ray penetration attenuation in the real tissue region.

[0014] Furthermore, the automatic calculation system for standard acquisition values ​​of DR images, applied to the above calculation method, includes: The grayscale analysis unit is used to acquire DR images, perform grayscale fluctuation analysis on DR images, calculate the degree of grayscale difference of each pixel, and generate an attenuation fluctuation index that represents the degree of local grayscale interference by artifacts. The interference analysis unit is used to analyze the intensity of the impact on each pixel in the DR image based on the attenuation fluctuation index, and generate an interference probability coefficient representing the probability of false attenuation of each pixel due to artifact contribution. The edge repair unit is used to suppress the features of DR images based on the interference probability coefficient and repair broken edges to obtain an edge repair map that represents the actual tissue edge location and orientation. The region segmentation unit is used to segment DR images into regions based on the edge restoration map, generating real tissue regions after eliminating artifact interference. The overall calculation unit is used to calculate the grayscale value and interference probability coefficient of all pixels in the real tissue area, and generate a standard intake value representing the degree of X-ray penetration attenuation in the real tissue area.

[0015] In summary, the present invention has the following main beneficial effects: By calculating the attenuation fluctuation index of each pixel, the degree of local grayscale interference by artifacts can be reflected. Furthermore, based on the interference probability coefficient, the gradient response severely contaminated by artifacts is adaptively suppressed to obtain the purified gradient intensity. The broken edges are then bridged and reconstructed by connecting the confidence level and the bridging potential value, generating an edge repair map. This effectively distinguishes between real tissue edges and false edges formed by scattering artifacts, while also repairing edge breaks and misalignments caused by motion artifacts. Then, dynamic thresholds are calculated and analyzed using the edge repair map to segment the region, obtaining the real tissue region after artifact interference is eliminated, and calculating the probability intensity. Simultaneously, a standard uptake value is generated by combining spatial distance weighting. This scheme avoids the misjudgment of artifact regions and the omission of real lesion regions in traditional edge detection, improves the accuracy of region segmentation and standard uptake value calculation, and enhances the analytical capabilities of DR images under complex artifact interference. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the automatic calculation method for standard acquisition values ​​of DR images according to the present invention. Figure 2 This is a schematic diagram of the automatic calculation system for standard acquisition values ​​of DR images according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] refer to Figure 1 and Figure 2 The automatic calculation method for standard acquisition values ​​of DR images includes: Step S1: Acquire DR image, perform grayscale fluctuation analysis on DR image, calculate the degree of grayscale difference of each pixel, and generate attenuation fluctuation index representing the degree of local grayscale interference by artifacts. Step S2: Based on the attenuation fluctuation index, analyze the intensity of the impact on each pixel in the DR image and generate an interference probability coefficient representing the probability of false attenuation caused by artifact contribution for each pixel. Step S3: Based on the interference probability coefficient, suppress the features of the DR image and repair the broken edges to obtain an edge repair map that represents the actual tissue edge position and orientation. Step S4: Perform image region segmentation on the DR image based on the edge repair map to generate the real tissue region after eliminating artifact interference; Step S5: Calculate the grayscale value and interference probability coefficient of all pixels in the real tissue area to generate a standard uptake value that represents the degree of X-ray penetration attenuation in the real tissue area.

[0019] In one embodiment, grayscale fluctuation analysis is performed on the DR image to calculate the degree of grayscale difference for each pixel and generate an attenuation fluctuation index representing the degree of local grayscale interference by artifacts, including: For DR images, the degree of gray-level fluctuation in the neighborhood of each pixel is calculated, and a chaotic trend coefficient representing the degree of gray-level fluctuation is generated. Specifically, for the current pixel in the DR image, a 3×3 neighborhood window is taken around it, and the gray-level values ​​of 9 pixels in the neighborhood window are extracted in the order from left to right and from top to bottom to form a gray-level sequence. The difference between any two adjacent gray values ​​in the gray-scale sequence is calculated, resulting in a total of 8 differences. When the difference is positive, the sign is assigned +1; when the difference is negative, the sign is assigned -1; and when the difference is 0, the sign is assigned 0, forming a sign sequence of length 8. The number of times the symbol values ​​of adjacent positions in the symbol sequence change is taken as the number of changes. The number of changes is divided by the maximum possible number of changes, 7, to obtain the chaotic trend coefficient, which represents the degree of gray-level fluctuation. Its value ranges from 0 to 1. The larger the chaotic trend coefficient, the more frequent the change in the direction of gray-level fluctuation in the neighborhood, the more disordered the gray-level sequence, and the more significant the local sawtooth fluctuations caused by artifact interference. When the symbol sequence is a sequence of all 0s or all the serial numbers are the same, the number of changes of adjacent symbols is 0, and the chaotic trend coefficient is 0.

[0020] The grayscale fluctuations within the neighborhood of each pixel are analyzed based on the chaotic trend coefficient to identify the modulation proportion caused by artifacts. An attenuation fluctuation index representing the degree of interference of local grayscale with artifacts is generated. Specifically, for the 3×3 neighborhood of the current pixel, for its grayscale value sequence and its 8 adjacent differences, the absolute values ​​of the 8 differences are added together to obtain the total fluctuation amplitude. Among the 8 differences, each pair of adjacent differences is checked in sequence. If their signs are different, the absolute value of the first difference is selected. The absolute values ​​of these selected differences are accumulated to obtain the abnormal modulation amplitude. If the total fluctuation amplitude is 0, then the attenuation fluctuation index is 0. Otherwise, the abnormal modulation amplitude is divided by the total fluctuation amplitude to obtain the attenuation fluctuation index, which represents the degree of interference of local grayscale by artifacts. Its value range is 0-1. Among the total fluctuation amplitude in the neighborhood, the proportion contributed by the alternation of signs, that is, the frequent changes in grayscale direction, is the proportion of high-frequency sawtooth modulation caused by artifacts.

[0021] By calculating the chaotic trend coefficient of the gray-level sequence in the neighborhood of each pixel, the frequency of changes in the direction of local gray-level fluctuations can be accurately reflected, effectively distinguishing between high-frequency sawtooth fluctuations caused by artifacts and the smooth transition of normal tissue edges. At the same time, by combining the decay fluctuation index, the abnormal modulation amplitude is compared with the total fluctuation amplitude to obtain the proportion of artifact interference in local gray-level changes, thereby directly extracting the modulation features caused by scattering artifacts and motion artifacts. Unlike traditional edge detection algorithms, this scheme does not rely on global edge assumptions, but identifies artifact interference from local gray-level fluctuation patterns, avoiding false edge misjudgment and segmentation deviation caused by blurring and breakage of real edges.

[0022] In one embodiment, based on the attenuation fluctuation index, the intensity of the impact on each pixel in the DR image is analyzed, and an interference probability coefficient representing the probability of false attenuation due to artifact contribution is generated for each pixel, including: Based on the attenuation fluctuation index, the cumulative intensity of each pixel affected by neighborhood artifacts is analyzed, and an artifact diffusion attenuation factor representing the degree of artifact energy propagation attenuation is generated. Specifically, this involves: traversing each pixel of the DR image, sorting the attenuation fluctuation index of all pixels from smallest to largest, and taking the value at the 90th percentile position as the dynamic threshold. Since artifact interference usually only affects local areas in the image rather than the whole, that is, the attenuation fluctuation index of most pixels is at a low level, and only a few pixels severely disturbed by artifacts have a high value. Therefore, choosing the value at the 90th percentile position as the dynamic threshold can effectively distinguish between normal background fluctuations and artifact-dominated fluctuations. Pixels in the DR image whose attenuation fluctuation index is greater than the dynamic threshold are marked as the source points of artifact energy release, and their initial intensity is the attenuation fluctuation index itself. For any pixel in the image, the Euclidean distance between it and the source point is calculated, and the reciprocal of the sum of the square of the Euclidean distance and 1 is used as the attenuation ratio. The initial intensity of each source point is multiplied by the attenuation ratio to obtain the contribution intensity of the source point to the target pixel. The contribution intensity of all source points is accumulated to obtain the original artifact diffusion value of the pixel. Find the maximum and minimum values ​​of the original artifact diffusion value in the DR image. Subtract the minimum value from the original artifact diffusion value of each pixel and divide by the difference between the maximum and minimum values ​​to obtain the artifact diffusion attenuation factor, which is in the range of 0-1 and represents the degree of artifact energy propagation attenuation.

[0023] Based on the artifact diffusion attenuation factor, the spatial heterogeneity of background grayscale in the neighborhood of each pixel is analyzed to generate a background depth coefficient representing the degree of background grayscale stretching caused by artifacts. Specifically, for each pixel in the DR image, a 5×5 neighborhood window is set with the pixel as the center. First, the arithmetic mean of the grayscale values ​​of all pixels in the neighborhood is calculated. Then, the absolute difference between the grayscale value of each pixel and the arithmetic mean is calculated. The absolute difference of each pixel is multiplied by its corresponding artifact diffusion attenuation factor to obtain the weighted absolute difference. Simultaneously, summing all weighted absolute differences within the neighborhood and dividing by the sum of all artifact diffusion attenuation factors within the neighborhood yields the weighted average absolute deviation. Dividing the weighted average absolute deviation by the arithmetic mean gives the background depth coefficient, which represents the degree of background grayscale stretching caused by artifacts. The larger the value of the background depth coefficient, the stronger the heterogeneity caused by background grayscale stretching by artifacts.

[0024] In one embodiment, based on the attenuation fluctuation index, the intensity of the impact on each pixel in the DR image is analyzed, and an interference probability coefficient representing the probability of false attenuation due to artifact contribution is generated for each pixel. The method further includes: Based on the artifact diffusion attenuation factor and the background depth coefficient, the probability trend of a pixel's grayscale value being misjudged as tissue attenuation is calculated, generating a false tendency index that indicates how easily a pixel is fooled by artifacts. Specifically, for each pixel, its artifact diffusion attenuation factor and background depth coefficient are multiplied to obtain the product result. At the same time, the artifact diffusion attenuation factor and background depth coefficient of the pixel are added together and the product result is subtracted to obtain the denominator value. The product result is divided by the denominator value, and the calculation result is normalized to the 0-1 range to obtain the false tendency index that indicates how easily a pixel is fooled by artifacts. The larger the false tendency index value, the higher the probability that the grayscale value of the pixel will produce false attenuation due to the contribution of artifacts, that is, the easier it is to be misjudged as real tissue attenuation.

[0025] Based on the false tendency index, the impact on each pixel is analyzed, and an interference probability coefficient representing the probability of false attenuation due to artifact contribution is generated. Specifically, for the current pixel, the false tendency index of all pixels in its surrounding 3×3 neighborhood is taken, and the arithmetic mean and harmonic mean of these 9 false tendency indices are calculated. Then, the harmonic mean is divided by the arithmetic mean to obtain the local uniformity coefficient of the current pixel. The local uniformity coefficient ranges from 0 to 1. The larger the local uniformity coefficient, the more uniform the distribution of the false tendency index in the neighborhood, and vice versa. The false tendency index of the current pixel is divided by the maximum value of all local uniformity coefficients in the neighborhood, and the calculation result is normalized to the 0-1 interval to obtain the relative intensity coefficient. Finally, the local uniformity coefficient and the relative intensity coefficient are multiplied to obtain the interference probability coefficient representing the probability of false attenuation due to artifact contribution of each pixel.

[0026] The calculated artifact diffusion attenuation factor reflects the degree of attenuation of artifact energy as it propagates with Euclidean distance. Combined with the background depth coefficient, it reflects the spatial heterogeneity of grayscale after artifact stretching within the neighborhood. Furthermore, the false tendency index assesses the probability that each pixel is easily misjudged as real tissue attenuation and generates an interference probability coefficient. This interference probability coefficient can accurately distinguish between false attenuation contributed by artifacts and real tissue attenuation from three levels: artifact diffusion, background distortion, and local consistency. It solves the problems of false edges formed by scattering artifacts and edge blurring and breakage caused by motion artifacts, avoiding misjudging artifact areas as lesions or missing real lesions, and improving the accuracy of region localization and segmentation.

[0027] In one embodiment, features of the DR image are suppressed based on the interference probability coefficient, and broken edges are repaired to obtain an edge repair map representing the location and orientation of the true tissue edge, including: Based on the interference probability coefficient, the gradient response of DR images severely contaminated by artifacts is suppressed, and a cleaned gradient intensity representing the gradient strength and direction after artifact suppression is generated. Specifically, for each pixel in the DR image, its original gradient strength and gradient direction are calculated using the Sobel operator; with the current pixel as the center, the median of the interference probability coefficients of all pixels in its surrounding 3×3 neighborhood is calculated. If the interference probability coefficient of the current pixel is greater than the median, an attenuation factor is calculated: if the interference probability coefficient of the current pixel is 0, the attenuation factor is set to 1; otherwise, the median is divided by the square of the interference probability coefficient of the current pixel to obtain the attenuation factor, and the original gradient intensity is multiplied by the attenuation factor to obtain the cleaned gradient intensity of the pixel. If the interference probability coefficient of the current pixel is not greater than the median, the cleaned gradient strength is directly the original gradient strength, and the gradient direction of all pixels remains unchanged. Thus, the cleaned gradient strength, which represents the gradient strength and direction after artifact suppression, can be obtained.

[0028] In one embodiment, the features of the DR image are suppressed based on the interference probability coefficient, and the broken edges are repaired to obtain an edge repair map representing the actual tissue edge location and orientation. The method further includes: Based on the purification gradient intensity, the broken edge segments are traced and connected, the connection confidence between each pair of edge endpoints is calculated, and a bridging potential value representing the overall continuity of the broken edge is generated. Specifically, for each pixel, it is determined whether its purification gradient intensity is greater than the purification gradient intensity of two adjacent pixels perpendicular to its gradient direction. If so, the pixel is marked as an edge seed point; otherwise, the pixel is not marked as an edge seed point and is skipped directly. Sort all edge seed points in descending order of purification gradient intensity, and then trace them sequentially: starting from the current seed point, search for a pixel on each side of the direction perpendicular to its gradient direction. If the pixel has been marked as a seed point, connect the pixel to the current edge line and continue to extend outward. Repeat this process until no seed point that meets the conditions can be found. Finally, several continuous edge lines are obtained. The two endpoints of each edge line are defined as follows: within the eight-neighborhood, the edge pixel has only one adjacent pixel belonging to the same edge line. For each pair of endpoints, connect the two endpoints to form a straight line segment, obtain all the pixels that the straight line segment passes through, calculate the cosine of the angle between the cleansing gradient direction and the direction of the straight line segment at each pixel, sum all the cosine values ​​and divide by the number of pixels to obtain the direction matching degree, then multiply the cleansing gradient intensities of each endpoint and divide by the square of the Euclidean distance between the endpoints to obtain the intensity distance factor, multiply the direction matching degree by the intensity distance factor, and normalize the calculation result to the 0-1 interval to obtain the connection confidence of the endpoint pair; Calculate the arithmetic mean of the connectivity confidence of all endpoint pairs, multiply the arithmetic mean by the ratio of the total number of endpoint pairs to the total number of pixels in the DR image to obtain the initial bridging potential value, divide the initial bridging potential value by the ratio of the total length of all edge lines in the DR image to the length of the diagonal of the DR image, and normalize the calculation result to the 0-1 interval to generate a bridging potential value representing the overall continuity of the broken edge.

[0029] Based on the bridging potential energy value, the fractured edge is bridged and reconstructed to generate an edge repair map representing the actual tissue edge position and direction. Specifically, this includes: creating a binary image with the same size as the original DR image and setting all pixel initial values ​​to 0, indicating that there are no edges in the initial state; setting the pixel values ​​corresponding to the existing edge lines to 1. Iterate through all endpoint pairs. For each endpoint pair, compare its connection confidence with the bridging potential value. If the connection confidence is greater than the bridging potential value, set the value of all pixels that pass through the straight line segment between the endpoints to 1; otherwise, do nothing. After performing the above operations on all endpoints, the pixel of the straight line segment is merged with the original edge line pixel. At this time, the position of the pixel value of 1 in the binary image represents the edge pixel, and the position of the pixel value of 0 represents the non-edge pixel. This binary image is the edge repair map that represents the position and direction of the real tissue edge. The position and direction of the continuous line with a value of 1 in the edge repair map represent the edge of the real tissue.

[0030] By calculating the attenuation factor, the gradient response in areas with severe artifact contamination is effectively suppressed, while the gradient corresponding to the real tissue edge is preserved. A cleansing gradient intensity is generated, and edge seed points are selected and continuous edge lines are tracked through the cleansing gradient intensity. The connection confidence between each pair of endpoints is further calculated, and finally an edge repair map is obtained. This scheme can significantly suppress the false edge response caused by scattering artifacts and actively repair the real edge breakage and spatial misalignment caused by motion artifacts, avoiding misjudging artifact areas as lesions or missing real lesions. Thus, an edge repair map that accurately represents the position and direction of the real tissue edge is obtained, improving the calculation accuracy of DR images under artifact interference.

[0031] In one embodiment, image region segmentation is performed on the DR image based on the edge restoration map to generate a real tissue region after eliminating artifact interference, including: Based on the edge restoration map and interference probability coefficient, the DR image is divided into regions to generate real tissue regions after excluding artifact interference. Specifically, the edge pixels in the edge restoration map are used as insurmountable boundaries, and the DR image is divided into several closed regions surrounded by edge lines. The pixels in each closed region are spatially continuous and are not crossed by the edge lines. For each closed region, the sum of the interference probability coefficients of all pixels in the closed region is divided by the total number of pixels in the closed region to obtain the average interference probability of the closed region; for any two adjacent closed regions, the absolute value of the difference between their average interference probabilities is calculated, and the absolute value is divided by the sum of their average interference probabilities, and the result is normalized to the 0-1 interval to obtain the difference coefficient. Calculate the arithmetic mean of the difference coefficients of all adjacent closed region pairs. Use this arithmetic mean as the merging threshold. Iterate through all adjacent closed region pairs. If the difference coefficient between the closed regions is less than or equal to the merging threshold, merge the two closed regions into a new region. Otherwise, do not merge. After completing one round of merging, recalculate the arithmetic mean of the difference coefficients of all newly formed adjacent region pairs and update the dynamic merging threshold. Repeat the merging process until there are no adjacent closed region pairs whose difference coefficients are less than or equal to the merging threshold. Each region obtained at this time is the real tissue region after eliminating artifact interference.

[0032] In one embodiment, the grayscale values ​​and interference probability coefficients of all pixels within the real tissue region are calculated to generate a standard uptake value representing the degree of X-ray penetration attenuation in the real tissue region, including: Based on the grayscale values ​​and interference probability coefficients of all pixels within the real organization area, the effective weight of each pixel's contribution to the area attenuation is calculated, generating a probability intensity representing the attenuation contribution of grayscale after being modulated by artifact probability. Specifically, this includes: using the natural constant e as the base and the negative of the interference probability coefficient of the pixel as the exponent, calculating the exponential function value to obtain the effective weight of the pixel. Through this calculation method, the effective weight of a pixel with a larger interference probability coefficient is closer to 0, and the effective weight of a pixel with a smaller interference probability coefficient is closer to 1, exhibiting non-linear exponential attenuation. Multiply the original gray value of each pixel by its effective weight to obtain the weighted gray value of that pixel. Then, sum the weighted gray values ​​of all pixels in the real tissue region and divide by the sum of the effective weights of all pixels in the real tissue region. This gives the probability intensity representing the contribution of gray value to attenuation after artifact probability modulation, which represents the contribution of X-ray penetration attenuation after artifact probability modulation of the real tissue region.

[0033] In one embodiment, the grayscale values ​​and interference probability coefficients of all pixels within the real tissue region are calculated to generate a standard uptake value representing the degree of X-ray penetration attenuation in the real tissue region, which further includes: A collaborative analysis of probability intensity and spatial distribution of pixels within the real tissue region is performed to generate a standard uptake value representing the degree of X-ray penetration attenuation in the real tissue region. Specifically, for the real tissue region, the centroid coordinates of the real tissue region are calculated. The centroid coordinates are the average of the horizontal coordinates and the average of the vertical coordinates of all pixels. For each pixel within the real organization region, multiply the probability intensity of that pixel by the reciprocal of the square of the Euclidean distance from that pixel to the centroid of the region to obtain the first value. Then, sum the first values ​​of all pixels to obtain the sum. The sum of the reciprocals of the squared distances of all pixels is calculated, and the sum is divided by the sum of the reciprocals to obtain the distance-weighted intensity. The distance-weighted intensity is then multiplied by the ratio of the total number of pixels in the real tissue region to the total number of pixels in the DR image to obtain the standard access value. The standard access value represents the degree of X-ray penetration attenuation in the real tissue region after artifact correction and spatial distribution modulation.

[0034] By using edge restoration maps and treating edge pixels as insurmountable boundaries, closed regions are delineated, resulting in the true tissue region after eliminating artifact interference. This effectively avoids the contamination of region segmentation caused by false edges formed by scattering artifacts and edge breaks and misalignments caused by motion artifacts. Simultaneously, the probability intensity is co-analyzed with the pixel distribution of DR images to generate a standard uptake value representing the degree of X-ray penetration attenuation in the true tissue region. This avoids misjudging false tissue or missing true lesions, ensuring that the standard uptake value accurately reflects the degree of X-ray penetration attenuation in the true tissue region, thus improving the reliability of DR image analysis under artifact interference.

[0035] In one embodiment, the automatic calculation system for standard acquisition values ​​of DR images, applied to the above calculation method, includes: The grayscale analysis unit is used to acquire DR images, perform grayscale fluctuation analysis on DR images, calculate the degree of grayscale difference of each pixel, and generate an attenuation fluctuation index that represents the degree of local grayscale interference by artifacts. The interference analysis unit is used to analyze the intensity of the impact on each pixel in the DR image based on the attenuation fluctuation index, and generate an interference probability coefficient representing the probability of false attenuation of each pixel due to artifact contribution. The edge repair unit is used to suppress the features of DR images based on the interference probability coefficient and repair broken edges to obtain an edge repair map that represents the actual tissue edge location and orientation. The region segmentation unit is used to segment DR images into regions based on the edge restoration map, generating real tissue regions after eliminating artifact interference. The overall calculation unit is used to calculate the grayscale value and interference probability coefficient of all pixels in the real tissue area, and generate a standard intake value representing the degree of X-ray penetration attenuation in the real tissue area.

[0036] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An automatic calculation method for standard acquisition values ​​of DR images, characterized in that, include: Step S1: Acquire DR image, perform grayscale fluctuation analysis on DR image, calculate the degree of grayscale difference of each pixel, and generate attenuation fluctuation index representing the degree of local grayscale interference by artifacts. Step S2: Based on the attenuation fluctuation index, analyze the intensity of the impact on each pixel in the DR image and generate an interference probability coefficient representing the probability of false attenuation caused by artifact contribution for each pixel. Step S3: Based on the interference probability coefficient, suppress the features of the DR image and repair the broken edges to obtain an edge repair map that represents the actual tissue edge position and orientation. Step S4: Perform image region segmentation on the DR image based on the edge repair map to generate the real tissue region after eliminating artifact interference; Step S5: Calculate the grayscale value and interference probability coefficient of all pixels in the real tissue area to generate a standard intake value representing the degree of X-ray penetration attenuation in the real tissue area.

2. The automatic calculation method for standard acquisition values ​​of DR images according to claim 1, characterized in that, Gray-level fluctuation analysis is performed on DR images to calculate the degree of gray-level difference for each pixel, generating a decay fluctuation index that represents the degree of local gray-level interference from artifacts, including: For DR images, the degree of gray-level fluctuation in the neighborhood of each pixel is calculated, and a chaotic trend coefficient representing the degree of gray-level fluctuation is generated. The grayscale fluctuations in the neighborhood of each pixel are analyzed based on the chaotic trend coefficient to find the modulation ratio caused by artifacts and generate an attenuation fluctuation index that represents the degree of interference of local grayscale with artifacts.

3. The automatic calculation method for standard acquisition values ​​of DR images according to claim 2, characterized in that, Based on the attenuation fluctuation index, the intensity of the impact on each pixel in the DR image is analyzed, and an interference probability coefficient representing the probability of false attenuation due to artifact contribution is generated for each pixel, including: Based on the decay fluctuation index, the cumulative intensity of each pixel affected by neighborhood artifacts is analyzed, and an artifact diffusion decay factor representing the degree of artifact energy propagation decay is generated. Based on the artifact diffusion attenuation factor, the spatial heterogeneity of background grayscale in the neighborhood of each pixel is analyzed, and a background depth coefficient representing the degree of background grayscale stretching caused by artifacts is generated.

4. The automatic calculation method for standard acquisition values ​​of DR images according to claim 3, characterized in that, Based on the attenuation fluctuation index, the intensity of the impact on each pixel in the DR image is analyzed, and an interference probability coefficient representing the probability of false attenuation due to artifact contribution is generated for each pixel. This also includes: Based on the artifact diffusion attenuation factor and the background depth coefficient, the probability trend of the pixel's gray value being misjudged as tissue attenuation is calculated, and a false tendency index representing the pixel's susceptibility to being deceived by artifacts is generated. Based on the false tendency index, the impact on each pixel is analyzed, and an interference probability coefficient is generated to represent the probability of false attenuation caused by artifact contribution for each pixel.

5. The automatic calculation method for standard acquisition values ​​of DR images according to claim 4, characterized in that, Based on the interference probability coefficient, the features of the DR image are suppressed, and the fracture edges are repaired to obtain an edge repair map representing the actual tissue edge location and orientation, including: Based on the interference probability coefficient, the gradient response of DR images severely contaminated by artifacts is suppressed, and a clean gradient intensity representing the gradient strength and direction after artifact suppression is generated.

6. The method for automatically calculating the standard acquisition value of DR images according to claim 5, characterized in that, Based on the interference probability coefficient, the features of the DR image are suppressed, and the fracture edges are repaired to obtain an edge repair map representing the location and orientation of the true tissue edges. This also includes: Based on the purification gradient intensity, trace and connect the broken edge segments, calculate the connection confidence between each pair of edge endpoints, and generate a bridging potential value that represents the overall continuity of the broken edge. Based on the bridging potential energy value, the fracture edge is bridged and reconstructed to generate an edge repair map that represents the actual tissue edge location and orientation.

7. The method for automatically calculating the standard acquisition value of DR images according to claim 6, characterized in that, Based on the edge restoration map, image region segmentation is performed on the DR image to generate the real tissue region after eliminating artifact interference, including: Based on the edge restoration map and interference probability coefficient, the DR image is divided into regions to generate the real tissue region after eliminating artifact interference.

8. The method for automatically calculating the standard acquisition value of DR images according to claim 7, characterized in that, The grayscale values ​​and interference probability coefficients of all pixels within the real tissue region are calculated to generate a standard uptake value representing the degree of X-ray penetration attenuation in that real tissue region, including: Based on the grayscale values ​​and interference probability coefficients of all pixels within the real organization area, the effective weight of each pixel's contribution to the area attenuation is calculated, and a probability intensity representing the attenuation contribution of grayscale after being modulated by artifact probability is generated.

9. The method for automatically calculating the standard acquisition value of DR images according to claim 8, characterized in that, The grayscale values ​​and interference probability coefficients of all pixels within the real tissue region are calculated to generate a standard uptake value representing the degree of X-ray penetration attenuation in that real tissue region. This also includes: A co-analysis of probability intensity and spatial distribution of pixels within the real tissue region is performed to generate a standard uptake value representing the degree of X-ray penetration attenuation in the real tissue region.

10. An automatic calculation system for standard acquisition values ​​of DR images, applied in the calculation method as described in any one of claims 1-9, characterized in that, include: The grayscale analysis unit is used to acquire DR images, perform grayscale fluctuation analysis on DR images, calculate the degree of grayscale difference of each pixel, and generate an attenuation fluctuation index that represents the degree of local grayscale interference by artifacts. The interference analysis unit is used to analyze the intensity of the impact on each pixel in the DR image based on the attenuation fluctuation index, and generate an interference probability coefficient representing the probability of false attenuation of each pixel due to artifact contribution. The edge repair unit is used to suppress the features of DR images based on the interference probability coefficient and repair broken edges to obtain an edge repair map that represents the actual tissue edge location and orientation. The region segmentation unit is used to segment DR images into regions based on the edge restoration map, generating real tissue regions after eliminating artifact interference. The overall calculation unit is used to calculate the grayscale value and interference probability coefficient of all pixels in the real tissue area, and generate a standard intake value representing the degree of X-ray penetration attenuation in the real tissue area.

Images