Angiography image analysis method and device for digital diagnosis and treatment equipment

Abstract

The invention provides a contrast image analysis method and device for digital diagnosis and treatment equipment. The method comprises the following steps: predicting a disease progress condition based on a focus volume characteristic parameter, a density statistical parameter and a quantitative index of a focus surrounding tissue density difference; judging whether the prediction result of the disease progress condition reaches a preset risk threshold range or not, if the prediction result is within the preset risk threshold range, judging that the disease progress condition is in a risk state, and generating high-risk early warning information; and judging the severity of the illness state according to the judgment result of the illness state progress condition prediction and in combination with the fitting data of the focus volume and density changing along with time, performing stage classification on the focus progress process, and generating a quantitative illness state diagnosis report.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method and device for analyzing angiographic images used in digital diagnosis and treatment equipment. Background Art

[0002] When analyzing the angiographic image sequence, it is affected by different incremental granularities, such as pixel level, voxel level and regional level. At the pixel level, how to accurately define the lesion area is a key technical step; first, how to explore image registration technology to achieve spatial alignment of lesion areas at different time points and eliminate the position offset introduced by patient position changes and respiratory movements; secondly, how to obtain the change data of the lesion area volume over time based on the fusion of voxel level and pixel number is a data processing problem to be solved; at the same time, the pixel gray value is related to the density of the lesion area, and how to obtain the change data of the lesion area density over time is also a major difficulty; in addition, after obtaining quantitative indicators such as the number of pixels, average density and gray value distribution of the lesion area at each time point, how to establish the association between time series data, explore the law of lesion progression, and further analyze the morphological characteristics of the lesion area is an important step to characterize the morphological complexity and invasion range of the lesion from multiple angles. Finally, how to predict the development of the disease and judge the risk of the disease to generate a disease diagnosis report is an important goal that this scheme needs to deal with. Summary of the invention

[0003] The present invention provides a method for analyzing contrast images for digital diagnosis and treatment equipment, which mainly includes: Obtain the contrast image sequences collected from the target patient at different time points, and segment the lesion area based on the contrast images at different time points based on pixel-level features, voxel-level features, and region-level features, respectively, to obtain the number of pixels, average density, and grayscale value distribution of the lesion area at different time points; Extract and describe the features of the lesion area at different time points, generate feature descriptors, perform ratio tests on the feature descriptors, screen out feature matching points, calculate the transformation matrix between the lesion areas through the feature matching points, and obtain the changes between the lesion areas at different time points; After alignment and registration, the change data of lesion volume over time are obtained according to the number of pixels in the lesion area at each time point. At the same time, the average density of the lesion area is calculated according to the gray value of pixels in the lesion area to obtain the change data of lesion density over time; By fitting the data of the change of lesion volume over time and the change of lesion density over time, the law of the change of lesion volume and density over time is analyzed, and the volume dynamic characteristic parameters and density statistical parameters are obtained to describe the dynamic change law of the lesion progression process; According to the dynamic change law of the lesion during its progression, the irregularity of the lesion shape and the smoothness of the boundary are calculated to evaluate the morphological complexity and boundary clarity of the lesion. At the same time, the density difference between the lesion area and the surrounding normal tissue is compared to obtain a quantitative index of the density difference of the tissue around the lesion. Predict the progression of the disease based on lesion volume characteristic parameters, density statistical parameters, and quantitative indicators of the difference in density of tissues around the lesion; Determine whether the predicted result of the disease progression reaches the preset risk threshold range. If the predicted result is within the preset risk threshold range, the disease progression is judged to be a risk state and a high-risk warning information is generated; Based on the predicted results of the disease progression and the fitted data of the changes in lesion volume and density over time, the severity of the disease is judged, the lesion progression process is staged and graded, and a quantitative disease diagnosis report is generated.

[0004] The present invention provides a contrast image analysis device for digital diagnosis and treatment equipment, which mainly includes: Image processing and feature extraction module, used to acquire and analyze the angiographic image sequences at different time points; Lesion area change analysis module, used to calculate the changes between lesion areas at different time points; Lesion dynamic feature analysis module, used to analyze the law of changes in lesion volume and density over time; Lesion morphology assessment module, used to assess the morphological complexity and boundary clarity of lesions; Disease prediction and risk assessment module, used to predict disease progression and generate high-risk warning information; The disease diagnosis and grading module is used to determine the severity of the disease and generate a quantitative diagnosis report.

[0005] The technical solution provided by the embodiment of the present invention may have the following beneficial effects: The present invention discloses a method and device for analyzing contrast images for digital diagnosis and treatment equipment. The method obtains contrast image sequences of patients at different time points, segments lesions at pixel, voxel and regional levels, extracts features and performs registration. By calculating the change data of lesion volume and density over time, a dynamic change law model is established to obtain volume dynamic feature parameters and density statistical parameters. Combined with indicators such as lesion morphological complexity, boundary clarity and surrounding tissue density difference, a pre-established disease progression prediction model is input to achieve prediction of disease progression. The method can also generate high-risk warning information based on the prediction results, stage and grade the lesion progression process, and generate a quantitative disease diagnosis report. The present invention improves the accuracy and reliability of lesion progression prediction through multi-dimensional feature analysis and dynamic modeling, and provides an important reference for clinical diagnosis and treatment decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Figure 1 The present invention is a flowchart of a method for analyzing angiographic images used in digital diagnosis and treatment equipment.

[0007] Figure 2 It is a schematic diagram of an angiographic image analysis method for digital diagnosis and treatment equipment of the present invention.

[0008] Figure 3 It is another schematic diagram of an angiographic image analysis method for digital diagnosis and treatment equipment of the present invention.

[0009] Figure 4 The present invention is a schematic structural diagram of an angiographic image analysis device for digital diagnosis and treatment equipment. DETAILED DESCRIPTION

[0010] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0011] like Figure 1-4 In this embodiment, a contrast image analysis method for a digital diagnosis and treatment device may specifically include: S101. Obtain a sequence of angiography images collected from a target patient at different time points. For the angiography images at different time points, segment the lesion area based on pixel-level features, voxel-level features, and region-level features to obtain the number of pixels, average density, and grayscale value distribution of the lesion area at different time points.

[0012] The image is subjected to Gaussian smoothing according to the angiography image acquisition sequence, and an edge tracking tool based on pixel grayscale gradient is used to obtain a lesion boundary contour map; the point with the maximum grayscale value is selected as the seed point for the lesion boundary contour map, and the region is filled according to the rule that the difference in pixel density of eight neighborhoods is less than a preset threshold value, so as to obtain a lesion region density distribution map; the lesion region density distribution map is used to establish three-dimensional point cloud data of a preset voxel size, and the noise points of the point cloud data are eliminated to obtain three-dimensional feature data; the centroid coordinates of the lesion region are extracted from the three-dimensional feature data to establish a registration reference point, and the lesion boundary points are paired with the nearest neighbors according to a preset distance threshold to obtain the lesion region volume data and a grayscale distribution histogram.

[0013] Specifically, the original image sequence of the lesion area was obtained according to the contrast image acquisition interval, and each image in the image sequence was enhanced according to the Gaussian smoothing operator. The enhanced image was used to extract the lesion boundary using an edge tracking tool based on the pixel grayscale gradient. The grayscale histogram of the lesion area was extracted by setting the binarization threshold to 128, and the initial contour map of the lesion area was calculated from the grayscale histogram. The maximum grayscale value point in the center of the lesion was selected as the seed point from the initial contour map, and the growth radius was set to 8 pixels. The area was filled based on the expansion rule that the difference in pixel density of 8 neighborhoods was less than 10%, and the density distribution map of the lesion area was obtained. The lesion boundary was defined according to the range of plus or minus 20% of the mean density, and the corrected two-dimensional segmentation result map was obtained. For the corrected two-dimensional segmentation result map, the voxel size was set to 2×2×2 cubic millimeters, and the three-dimensional spatial coordinates were obtained from the corresponding positions of the adjacent layer images. The three-dimensional point cloud data was constructed based on the grayscale value of the voxel. The point cloud data was screened using the minimum voxel bounding box, and the number of voxels and the average density value of the voxels were extracted to form the three-dimensional feature data. For the three-dimensional feature data obtained at different time points, the centroid coordinates of the lesion area were extracted to establish the registration reference points. The lesion boundary points were paired based on the nearest neighbor principle, and the maximum distance threshold between the point pairs was set to 3 mm. The lesion area volume was calculated for the registered data, and the number of pixels, average density value and grayscale distribution histogram of the lesion area were counted from the registration results. The contrast image sequence corresponds to the lesion area with density change characteristics at different time points. The image was enhanced by the Gaussian smoothing operator, and the standard deviation was set to 1.5 and the kernel window size was 3×3, which can effectively reduce the influence of image noise, maintain the edge characteristics of the lesion, and improve the accuracy of edge tracking. In the liver tumor contrast images, there is a density difference between the lesion area boundary and the surrounding tissue. The edge tracking tool identifies the boundary position based on the pixel grayscale gradient change, and the grayscale difference is greater than 30. The grayscale histogram of the lesion area reflects the tissue density distribution characteristics. The grayscale value range is 0-255, and the bimodal distribution morphology corresponds to the lesion parenchyma area and the edge transition area. The selection of seed points affects the regional growth results. The point with the maximum gray value in the center of the lesion is selected as the starting position to ensure that the expansion starts from the lesion parenchyma tissue. In the pulmonary nodule angiography image, the 8-neighborhood expansion rule is based on the density similarity criterion. When the density difference of adjacent pixels is less than 10%, the target area is included to achieve adaptive segmentation of the lesion area. The lesion boundary correction adopts a threshold range of plus or minus 20% of the density mean to exclude the mis-segmented background tissue and improve the accuracy of the segmentation boundary. In the three-dimensional spatial reconstruction, the voxel size is set to 2×2×2 cubic millimeters to adapt to the slice thickness resolution of the CT image and achieve accurate expression of the morphological characteristics of the lesion. In the brain tumor angiography sequence, the adjacent layer images establish spatial associations through the corresponding position coordinates to construct a three-dimensional point cloud data set. The minimum voxel bounding box screening eliminates free noise points, retains the continuous and complete lesion tissue area, and calculates the number of voxels to reflect the change in lesion volume.The registration of lesion regions at different time points is based on the principle of feature point matching, and the centroid coordinates are used as reference points to establish a spatial mapping relationship. For example, in the dynamic enhanced scanning of breast masses, the boundary points of the lesions are paired using the nearest neighbor criterion, and the point pair distance threshold is set to 3 mm to control the matching error, and the lesion morphology is achieved through registration. After the registration is completed, quantitative measurements are performed to calculate the volume change rate of the lesion area, the statistical average density value reflects the change in tissue blood supply, and the grayscale distribution histogram describes the internal heterogeneity characteristics of the lesion. In the follow-up evaluation of multiple metastatic tumors, the contrast image sequence at multiple consecutive time points reflects the dynamic change process of the lesion. By quantitatively analyzing the characteristic parameters of the lesion area at different phases, comparing the lesion morphology and density changes at adjacent time points, and combining clinical diagnosis and treatment information to evaluate the treatment effect. Different types of tumor lesions have specific contrast characteristics, the density change curve morphology reflects the state of tissue angiogenesis, and the grayscale distribution coefficient of variation characterizes the uniformity of the internal structure of the lesion.

[0014] By extracting regional level features of angiography images at different time points, calculating the feature quantization value of the lesion area, comparing the preset feature quantization threshold with the feature quantization value of the lesion area, and preliminarily segmenting the target area, the seed point selection conditions are preset, and a number of seed points are selected from the target area according to the selection conditions, and the segmented area is expanded from the seed points until the preset expansion conditions are met to achieve the final segmentation.

[0015] A Gaussian smoothing kernel function is used to perform noise reduction processing on the angiography image sequence, and a candidate area is obtained by projection calculation based on the grayscale value distribution matrix of the denoised image; the seed point position is obtained based on the local grayscale value mean and variance of the candidate area through a preset uniform area judgment condition, and the grayscale difference threshold is used to expand the region from the seed point to obtain an extended segmentation area; edge contour lines are extracted from the extended segmentation area, and feature points are obtained by calculating the edge gradient data in the detection window, and regional segmentation is performed according to the curvature of the feature points until the area converges to obtain the final segmentation area; shape feature parameters are extracted from the final segmentation area, and the difference in regional morphological features is judged by the feature point pairing distance threshold. If the difference in shape feature parameters is within a preset range and the feature point pairing rate exceeds a preset threshold, it is determined to be a corresponding area.

[0016] Specifically, the original image data was obtained from the angiographic image sequence at different time points. The original image data was denoised using a Gaussian smoothing kernel function with a window size of 3×3 and a standard deviation of 1.5. The regional grayscale value distribution matrix was extracted from the denoised image. The quantified value of the lesion region characteristics was calculated by horizontal and vertical integral projection. If the quantified value was within the range of plus or minus 15% of the preset threshold, the region was taken as a candidate region. The regional validity was judged against the preset quantification threshold to obtain a preliminary segmentation region. The local grayscale value mean and variance were calculated within the preliminary segmentation region. The grayscale standard deviation within the 5×5 pixel block was set to be less than 10 as the uniform region judgment condition. The position with the largest grayscale value was selected from the uniform region as the starting seed point. The grayscale difference of adjacent pixels within the 8-neighborhood of the seed point was set to be less than 20% as the extension growth threshold. The grayscale value-based regional expansion tool was used to expand outward from the seed point position to generate an extended segmentation region. For the extended segmented area, extract the edge contour of the area, set the detection window size to 7 pixels, calculate the first-order difference of the grayscale value in the window in the vertical and horizontal directions to obtain the edge gradient data, select the edge points with gradient values ​​greater than 30 as feature points, build an edge list based on the 8-neighborhood connectivity of the feature points, obtain the edge point with the largest local curvature from the list as a new seed point, and continue segmenting according to the extended growth threshold until the area change rate between two adjacent segmentations is less than 5%. Extract the three shape feature parameters of area, perimeter, and roundness from the final segmented area, compare the morphological features of the target areas at different time points, pair the edge feature points based on the nearest neighbor criterion, set the maximum distance threshold between the feature point pairs to 3 pixels, and if the shape feature parameter difference between the two areas is within 10% and the feature point pairing rate is greater than 80%, they are determined to be corresponding areas, and the automatic registration of the target areas at different time points is completed. During the acquisition of medical contrast images, the lesion area presents a specific grayscale change pattern at different time points. The Gaussian smoothing kernel function is used to reduce random noise interference. In the dynamic enhanced scanning of the liver, a 3×3 window with a 1.5 standard deviation parameter is used to achieve image smoothing while maintaining edge features. During the quantification of lesion area features, the horizontal and vertical integral projections reflect the regional grayscale distribution characteristics. The grayscale distribution of normal liver parenchyma tissue is uniform, and the lesion area shows local grayscale abnormalities. When the quantization value deviation exceeds 15%, it indicates the presence of lesions. The local grayscale statistical characteristics correspond to the tissue density distribution. In pancreatic space-occupying lesions, the grayscale standard deviation within the 5×5 pixel block is less than 10, indicating that the tissue structure is uniform. The maximum grayscale value position is selected as the seed point to ensure expansion from the center of the lesion. Vascular-rich tumors show obvious enhancement in the arterial phase. The grayscale difference of neighboring pixels is less than 20%, which corresponds to tissue density continuity. The segmented area is gradually expanded based on the grayscale similarity principle. A 7-pixel detection window was used for lesion edge feature extraction. In dynamic enhancement of pulmonary nodules, an edge gradient value exceeding 30 indicated a tissue interface, and a complete edge contour was constructed through 8-neighborhood connectivity analysis.Edge curvature reflects the change in contour morphology. The point with the maximum local curvature often corresponds to the morphological characteristics of the lesion. Continuing segmentation as a new seed point can improve the accuracy of boundary positioning. The area change rate of two consecutive segmentations is less than 5%, indicating that the segmented area tends to be stable. Comparison of morphological characteristics of lesion regions at different time points. For example, in the breast mass angiography sequence, the area represents the size of the lesion, the perimeter reflects the complexity of the boundary, and the roundness describes the regularity of the morphology. In the nearest neighbor feature point pairing, the distance threshold of 3 pixels is adapted to the slight deformation of the lesion, the tolerance range of 10% of the shape feature parameter considers the dynamic change of density during the angiography process, and the feature point pairing rate of 80% ensures the accuracy of the lesion region correspondence. In the tracking and evaluation of multiple metastases, the angiography images at multiple consecutive time points record the dynamic change process of the lesion. By quantitatively analyzing the characteristic parameters of the lesion region at different phases, the morphology and density change rules of the lesion are displayed. Different types of tumor lesions have specific angiography manifestations. The morphology of the density change curve corresponds to the characteristics of tissue blood supply. The changes in regional morphological parameters reflect the infiltration and growth characteristics of the lesion. The shape feature registration results support the evaluation of the lesion evolution process. In the segmentation of heterogeneous tumor regions, the edge gradient and curvature features are combined to improve the accuracy of segmentation boundaries, and the multiple seed point expansion strategy adapts to the complex lesion morphology.

[0017] S102, extracting and describing the features of the lesion regions at different time points, generating feature descriptors, performing ratio tests on the feature descriptors, screening out feature matching points, calculating the transformation matrix between the lesion regions through the feature matching points, and obtaining the changes between the lesion regions at different time points.

[0018] The original image of the lesion area is smoothed using a Gaussian blur tool, edge contours are extracted from the smoothed image using an edge detection operator, and a feature descriptor is obtained by calculating a gradient histogram based on the edge contours; the Euclidean distance value between corresponding positions at different time points is calculated based on the feature descriptor, and an initial matching point set is obtained by comparing the Euclidean distance value with a preset distance threshold; matching point pairs are randomly selected from the initial matching point set to calculate an affine transformation matrix, and the number of internal points that satisfy the affine transformation matrix is ​​counted to obtain the optimal transformation parameters; the lesion area image is spatially mapped based on the optimal transformation parameters, and the transformed image is resampled using a bilinear interpolation tool, and the registration result is determined by calculating the mutual information value of the images before and after resampling.

[0019] Specifically, according to the original images of the lesion area at different time points, the Gaussian blur tool with a window size of 3×3 and a standard deviation of 1.5 was used to smooth the images. The edge contour of the lesion area was extracted by the Sobel edge detection operator. The grayscale gradient amplitude and direction in the 8-neighborhood of the edge point were calculated. The gradient histogram in the 16×16 pixel window was extracted from each edge point. The gradient distribution was statistically calculated according to the 8-directional intervals. The gradient histogram value was normalized by dividing the histogram sum to obtain a 128-dimensional feature descriptor. For the normalized feature descriptors, the Euclidean distance between the feature descriptors at the corresponding positions at different time points was calculated, and the distance threshold was set to 0.6. For each feature point, the two candidate matching points with the smallest distance were selected. If the ratio of the minimum distance to the second smallest distance was less than 0.8, the matching point pair was retained. All feature points were traversed and matched by the point pair matching tool based on the distance threshold to generate an initial matching point set. Three pairs of matching points were randomly selected from the initial matching point set, and the corresponding affine transformation matrix was calculated. The number of inliers that satisfied the transformation matrix was counted. The maximum number of iterations was set to 100, and the inlier threshold was 3 pixels. The random sampling process was repeated until the number of inliers was greater than 60% or the maximum number of iterations was reached. The transformation matrix with the largest number of inliers was selected as the optimal solution, and the rotation angle, translation vector, and scaling factor were calculated from the transformation matrix. The images of the lesion area at different time points were spatially mapped according to the optimal transformation matrix, and the bilinear interpolation tool with a 16×16 pixel window was used to resample the transformed images. The mutual information values ​​of the images before and after resampling were calculated. If the mutual information value was greater than the preset threshold of 0.8, the registration was determined to be valid. The overlap coefficient of the lesion area was calculated from the registered image, and the area ratio and boundary distance of the registered area were obtained. In medical image processing, the lesion area shows density and morphological changes at different time points, and accurate feature extraction is crucial for registration. Gaussian blurring uses a 3×3 window with a 1.5 standard deviation parameter. In the dynamic enhanced scanning of the liver, pixel-level smoothing is used to reduce random noise interference and maintain the edge structural characteristics of the lesion. The sobel operator is directional sensitive to edge detection. In the CT image of lung nodules, the horizontal and vertical gradient operators are used to jointly extract the edge contour, and the 8-neighborhood gradient statistics reflect the local structural characteristics. The 16×16 pixel window contains sufficient local feature information, the 8-directional gradient histogram describes the edge distribution law, and the normalization process eliminates the influence of image brightness and contrast changes. In the feature matching process, the Euclidean distance measures the similarity between feature descriptors. In the follow-up of pancreatic space-occupying lesions, the distance threshold of 0.6 distinguishes similar and non-similar feature points. The minimum distance ratio test excludes fuzzy matching, and the ratio threshold of 0.8 strikes a balance between ensuring matching accuracy and matching quantity. As shown in the dynamic enhanced sequence of breast masses, the distance-based point pair matching gives priority to corresponding points with high local feature similarity to form a stable initial matching set.The affine transformation matrix solution uses three pairs of matching points to construct the equation group. In the multiple metastasis tracking assessment, 100 iterations of sampling improve the probability of searching for the optimal solution, and the 3-pixel inlier threshold allows slight deformation. The 60% inlier ratio requirement ensures that the transformation matrix has statistical reliability. The rotation angle and translation vector describe the change in the spatial posture of the lesion, and the scaling factor reflects the volume change. In the evaluation of the registration results, the bilinear interpolation of the 16×16 pixel window maintains the grayscale continuity of the image. In the enhanced scanning sequence of brain glioma, the mutual information value of 0.8 or above indicates that the registration result is reliable. The overlap coefficient of the lesion area quantifies the registration accuracy, the area ratio reflects the volume change, and the boundary distance describes the degree of contour matching. Different types of tumor lesions show specific morphological changes during treatment, and the registration results support the analysis of the lesion evolution process. In the feature extraction of heterogeneous tumor regions, multi-scale gradient features improve the expression ability of the descriptor, and the combination of local feature matching and global transformation constraints ensures the accuracy of the registration results. In dynamic enhanced imaging, the grayscale change of the lesion area reflects the characteristics of tissue blood supply, and the change of morphological parameters corresponds to the invasive growth characteristics. The image sequence of continuous time points records the dynamic process of the lesion, and the changes in the characteristic parameters of the lesion area are quantitatively analyzed, combined with clinical follow-up information to evaluate the treatment effect. The feature descriptor construction and matching strategy are adapted to the morphological characteristics of different types of lesions, and the registration result verification ensures the reliability of the lesion evolution analysis.

[0020] S103, after alignment and registration, obtain the data of the change of the lesion volume over time according to the number of pixels in the lesion area at each time point, and calculate the average density of the lesion area according to the gray value of the pixels in the lesion area to obtain the data of the change of the lesion density over time.

[0021] The lesion boundary contour line is obtained by a seed point-based region growing tool, and the lesion volume value is obtained by converting the number of pixels in the connected area within the contour line and the image resolution; for the image area within the lesion boundary contour line, the histogram equalization tool is used for enhancement processing, the gray value matrix is ​​extracted from the enhanced image, and the regional average density value is obtained by projection integral operation; time series data is constructed according to the lesion volume value and the regional average density value, and linear interpolation correction is performed on abnormal points in the time series data that exceed preset values ​​to obtain a normalized volume density change sequence; for the normalized volume density change sequence, the change rate is obtained by calculating the difference between adjacent time points, the mean and standard deviation of the change rate are used to construct a lesion progression feature parameter set, and the feature parameter set is normalized to obtain a standardized lesion progression feature.

[0022] Specifically, according to the images of the lesion area at different time points after registration, the lesion boundary contour line was extracted by the seed point-based regional growth tool, the seed point position was set at the maximum gray value of the region, and the growth threshold was that the gray value difference of the neighboring pixels was less than 15%. If the 8 neighborhoods between two pixels were connected and the gray value difference was less than the threshold, they were judged to belong to the same region, and the number of pixels in the connected area within the contour line was calculated. The actual volume of the lesion was converted according to the image resolution of 0.5 mm per pixel, and the volume change of the lesion over time was recorded in the format of timestamp plus volume value. For the pixels in the lesion area, the contrast-limited histogram equalization tool was used to enhance the image, the contrast limit threshold was set to 0.02, and the block size was 8×8 pixels. The gray value matrix within the lesion area was extracted from the enhanced image, and the average density value of the region was calculated by horizontal and vertical projection integrals, and the time series data table was constructed in combination with the time series information. The volume and density values ​​in the time series data were processed by 5-point median filtering, and a sliding window of 11 data points was set to calculate the standard deviation of the data in the window. If the single point value exceeded 2 times the standard deviation, it was marked as an abnormal point. The abnormal point data was corrected by linear interpolation of the previous and next time points. The data was normalized according to the normalization coefficient of 0.05 to obtain a standardized volume density change sequence. The difference between adjacent time points was extracted from the standardized sequence as the change rate, the mean and standard deviation of the change rate were calculated, and a feature data set containing volume change rate, density change rate, volume standard deviation, and density standard deviation was constructed. The data was normalized according to the numerical distribution range of the feature data to obtain the normalized lesion progression feature parameters. In medical image analysis, the volume and density changes of the lesion area reflect the disease progression state, and accurate extraction of regional features is crucial to evaluate the treatment effect. The seed point was selected at the maximum grayscale value of the region to ensure that it started to grow from the lesion parenchyma tissue. In the dynamic enhanced scanning of the liver, the grayscale difference threshold of 15% was adapted to the gradient characteristics of tissue density. The 8-neighborhood connectivity judgment was used to consider the spatial position relationship of pixels to achieve regional integrity segmentation. The resolution setting of 0.5 mm per pixel is suitable for routine CT scans and supports quantitative calculation of lesion volume. Contrast-limited histogram equalization shows good results in dynamic enhancement of breast masses. The contrast limit threshold of 0.02 avoids over-enhancement artifacts, and the block size of 8×8 pixels optimizes grayscale distribution in local areas. The horizontal and vertical projection integrals reflect the regional density distribution characteristics. For example, in the evaluation of pancreatic space-occupying lesions, the integral curve morphology corresponds to the uniformity of tissue density. The time series data table records the dynamic changes of lesions during follow-up and supports disease progression analysis. Discrete noise points are eliminated by 5-point median filtering. In the tracking and evaluation of multiple metastases, the 11-point sliding window statistics capture the trend of data changes. The standard deviation 2 times threshold identifies abnormal fluctuations, and linear interpolation supplements data to maintain sequence continuity. The normalization coefficient of 0.05 is used to map parameters of different dimensions to a unified numerical interval, which is convenient for comprehensive analysis of characteristic parameters.The difference calculation between adjacent time points reflects the rate of lesion progression. For example, in the follow-up evaluation of brain glioma, the volume change rate corresponds to the tumor growth state, and the density change rate reflects the change in tissue characteristics. The mean and standard deviation of the change rate describe the stability of lesion progression, and the characteristic data are normalized to eliminate the parameter scale difference, and a quantitative indicator system for lesion progression is constructed. In the time series feature analysis of lesions, volume and density parameters complement each other and jointly characterize the law of disease progression. For example, in the monitoring of lung cancer radiotherapy, volume changes reflect tumor load, and density changes indicate necrosis and liquefaction. Follow-up data at multiple consecutive time points record the treatment response process, and the efficacy is evaluated by quantitatively analyzing the changes in lesion characteristic parameters. Different types of tumor lesions show specific progression patterns, and the volume density change curve describes tissue pathological changes. Standardized characteristic parameters support lesion progression prediction and provide data support for clinical treatment decisions. In the analysis of heterogeneous tumors, local area feature extraction and global parameter statistics are combined to accurately characterize the characteristics of lesion progression. The continuity processing and outlier correction of follow-up data ensure the reliability of characteristic parameters, and normalization processing realizes the unified expression of multidimensional features. In the construction of characteristic parameters of lesion progression, static morphological characteristics and dynamic change indicators complement each other to form a complete quantitative evaluation system.

[0023] According to the number of pixels in the lesion area at each time point, combined with the voxel-level features of the contrast image, the voxel-level features include voxel grayscale value, voxel spatial position and voxel shape size, the number of pixels and voxel volume at each time point are determined, the total volume of the lesion is calculated, the total volume of the lesion at each time point is recorded, and the data on the change of lesion volume over time are obtained.

[0024] A Gaussian smoothing kernel function is used to perform noise reduction processing on the angiography image data, and a fixed threshold is used to perform binary segmentation on the smoothed image, and a set of voxels in the initial lesion area is obtained by a region growing algorithm; a three-dimensional grid is used to divide the space for the initial lesion area, and the voxels are classified according to the Euclidean distance from the voxel to the center point of the grid. The connectivity is determined by the distance and grayscale value difference between adjacent grid units to obtain a three-dimensional grid connectivity map of the lesion area; a set of boundary points is extracted from the three-dimensional grid connectivity map, and a bounding box is constructed according to the maximum and minimum coordinate values ​​of the boundary points in three directions, and the maximum diameter and volume of the lesion are obtained by calculating the diagonal length of the bounding box and counting the total number of voxels in the connected domain; a time series data table is constructed for the lesion volume, and the change data of the lesion volume over time are obtained by linear interpolation.

[0025] Specifically, based on the contrast image data, the Gaussian smoothing kernel function was used to reduce the noise of the image. The kernel window size was set to 3×3 pixels and the standard deviation was set to 1.5. The grayscale value range of the voxels was extracted from the smoothed image and a grayscale histogram was established. The voxels were binarized based on a fixed threshold of 128, and the seed point position was set at the maximum grayscale value of the region. Regional growth was performed according to the criterion that the grayscale difference of the neighboring voxels was less than 10%, and the voxel set of the initial lesion area was obtained. For the initial lesion area, a 2×2×2 cubic millimeter three-dimensional grid division space was set, and the absolute coordinate value of each grid center point was recorded. Voxels were classified based on the Euclidean distance from the voxel center point to the grid center point. If the distance between the center points of adjacent grid units was less than 3 voxels and the grayscale value difference was within 15%, the two grid units were judged to belong to the same connected domain, and a three-dimensional grid connectivity map of the lesion area was constructed. The set of boundary points of the lesion area was extracted from the three-dimensional grid connectivity map. The three-dimensional bounding box was constructed by searching the maximum and minimum coordinate values ​​of the boundary points in the x, y, and z directions. The diagonal length of the bounding box was calculated as the maximum diameter of the lesion. The actual spatial size was converted by combining the voxel resolution of 0.5 mm per pixel. The total number of voxels in the connected domain was counted and multiplied by the volume of a single voxel of 0.125 cubic millimeters to obtain the total volume of the lesion. For the lesion volume value obtained at each time point, a time series data table was constructed in the format of timestamp plus volume value. The volume data sequence was smoothed by a 5-point median filter. The volume change rate of adjacent time points was calculated, and the change rate threshold was set to 20%. The abnormal points exceeding the threshold were marked, and the data was corrected by linear interpolation of the four valid data points before and after the abnormal point to obtain the change curve of the lesion volume over time. In the process of medical image analysis, voxel feature extraction is the basis for three-dimensional reconstruction of lesions. For example, Gaussian smoothing is used in dynamic enhanced scanning of the liver. A 3×3 window with a 1.5 standard deviation parameter is used to retain the edge structural characteristics of the lesion and effectively suppress random noise. Gray value distribution reflects tissue density characteristics. For example, in pulmonary nodule angiography images, a binary threshold of 128 distinguishes the lesion area from the background tissue. The seed point is selected at the maximum gray value to ensure that the growth starts from the lesion parenchyma tissue, and the gray value difference threshold of 10% adapts to the gradual change of tissue density. In three-dimensional grid division, the grid size of 2×2×2 cubic millimeters matches the commonly used clinical CT scanning resolution. For example, in the analysis of pancreatic space-occupying lesions, the spatial coordinates record the voxel position distribution, and the Euclidean distance calculation reflects the spatial relationship between voxels. The connectivity judgment threshold of 3 voxels takes into account the influence of the scanning layer thickness, the gray value difference tolerance of 15% adapts to the inhomogeneity of tissue density, and the grid connectivity map describes the spatial morphological characteristics of the lesion. The lesion boundary feature extraction is expressed by a three-dimensional bounding box. For example, in the dynamic enhancement sequence of breast masses, the maximum coordinates in the x, y, and z directions determine the spatial range. The resolution setting of 0.5 mm per pixel corresponds to high-resolution thin-layer scanning, and the voxel volume of 0.125 cubic millimeters supports accurate measurement.Voxel counting combined with actual size conversion accurately reflects the change in lesion size. In time series data processing, 5-point median filtering eliminates discrete noise. In the tracking and evaluation of multiple metastases, a 20% volume change rate threshold identifies significant changes. Linear interpolation of the four data points before and after maintains the smoothness of the curve, and the volume change curve depicts the dynamic evolution of the lesion. Three-dimensional reconstruction of lesions involves multiple processing links, and different types of tumor lesions have specific imaging manifestations. For example, in the follow-up evaluation of brain gliomas, voxel feature analysis supports accurate lesion positioning, and connected domain construction ensures regional integrity. In the processing of areas with uneven tissue density, grid expression reduces the impact of local noise, and bounding box calculation adapts to irregular shape features. Follow-up data at continuous time points records lesion changes during treatment, and the efficacy is evaluated by quantitative analysis of volume parameter changes. For example, in monitoring chemotherapy for lung cancer, the volume curve morphology reflects the treatment response, and abnormal point correction eliminates the impact of acquisition errors. In the analysis of different imaging sequences, voxel feature extraction and three-dimensional reconstruction parameters are adjusted according to scanning conditions to ensure the accuracy of measurement results. In heterogeneous tumor volume measurement, multi-level feature extraction and spatial relationship analysis are combined to improve the accuracy of regional division.

[0026] S104. By fitting the data of changes in lesion volume over time and the data of changes in lesion density over time, the law of changes in lesion volume and density over time is analyzed to obtain volume dynamic characteristic parameters and density statistical parameters to describe the dynamic change law in the process of lesion progression.

[0027] The cubic spline interpolation tool is used to smooth the time series data of lesion volume and density, and the derivative value is calculated according to the smoothed data series, and the characteristic point position is marked by the sign change of the derivative value; the lesion volume data is fitted with a second-order polynomial function according to the characteristic point position, and the quadratic term coefficient and the linear term coefficient are extracted from the fitting function to obtain the volume dynamic characteristic parameter; for the lesion density data, the density data is divided into a high-value segment, an intermediate segment and a low-value segment using a preset threshold, and the density change characteristic parameter is obtained through the statistics of the data segment; a feature vector is constructed according to the volume dynamic characteristic parameter and the density change characteristic parameter, and the feature vector is normalized using the minimum and maximum normalization method to obtain a lesion progression feature set.

[0028] Specifically, based on the time series data of lesion volume and density, the cubic spline interpolation tool was used to smooth the data sequence, the spline node interval was set to 3 time points, the boundary derivative value was set to 0, and the first-order derivative value was calculated point by point for the interpolation curve. If the derivative values ​​of two adjacent points have opposite signs and the change amplitude exceeds 20%, it is marked as an inflection point, if the derivative value is greater than zero and the next derivative value is less than zero, it is marked as a maximum point, if the derivative value is less than zero and the next derivative value is greater than zero, it is marked as a minimum point, and the slope of the curve between two adjacent feature points is calculated to obtain the characteristic parameter set. For the volume change data, a second-order polynomial function was used to fit the time series data points, the coefficients of the polynomial function were calculated, and the quadratic term coefficients were extracted as acceleration parameters and the linear term coefficients as growth rate parameters. The threshold was set to 10% based on the fitting error between the function and the data point, and the 95% confidence interval of the parameter was calculated by 1000 random resamplings. The fluctuation range of the volume dynamic characteristics was obtained from the upper and lower limits of the confidence interval. Two fixed thresholds were set for the density change data sequence, and 20% of the data mean was taken as the segmentation standard. The density data was divided into three intervals: high value segment, middle segment and low value segment. The proportion, mean and standard deviation of the number of data points in each interval were counted, the density change rate of adjacent time points was calculated, and the maximum, minimum and median of the density change rate were extracted as statistical feature parameters. Combining volume feature parameters and density statistical parameters, the feature vector of lesion progression was constructed, including parameters such as volume acceleration, growth rate, density segment ratio and density change rate. The feature values ​​were mapped to the 0-1 interval by the minimum and maximum standardization method, and the standardized feature parameter set was obtained to record the progression process characteristics of the lesion in chronological order. In the time series analysis of medical images, the changes in lesion volume and density reflect the law of disease progression. Cubic spline interpolation uses the node interval of 3 time points to balance smoothness and detail preservation. For example, in the dynamic enhanced scanning of liver tumors, the boundary derivative value is set to 0 to avoid oscillation at both ends of the curve. The sign change of the derivative value exceeds 20%, marking the inflection point. For example, in the follow-up evaluation of lung nodules, it corresponds to the key moment of treatment response. For example, a sudden drop in density indicates necrosis, and a sudden decrease in volume reflects effective treatment. Second-order polynomial fitting reflects the dynamic characteristics of lesion growth. For example, in the follow-up analysis of breast masses, the coefficient of the quadratic term reflects the growth acceleration. A positive value indicates accelerated progression, and a negative value indicates suppressed growth. The 10% fitting error threshold balances the model complexity and fitting accuracy. 1000 random resamplings are used to construct a 95% confidence interval to evaluate the reliability of parameter estimation. The volume change reflects the overall load of the lesion, and the shape of the growth curve corresponds to different stages of progression. The segmented density analysis uses 20% above and below the data mean as a threshold. For example, in the follow-up of pancreatic space-occupying lesions, the high-value segment corresponds to solid tissue, and the low-value segment indicates cystic necrosis. The interval data ratio reflects the tissue composition, and the density change rate describes the speed of tissue transformation. The maximum, minimum, and median constitute the dynamic characteristics of density, which characterize the evolution of the internal structure of the lesion.The feature vector integrates multi-dimensional parameter information. For example, in the monitoring of brain glioma progression, volume acceleration characterizes the growth dynamics, and the density segment ratio reflects the degree of heterogeneity. The minimum and maximum standardization eliminates the dimension effect, and the feature is mapped to the 0-1 interval for comprehensive analysis. The time series feature record describes the trajectory of lesion progression and supports prognostic evaluation. In dynamic enhanced sequence analysis, volume and density features confirm each other and jointly reflect the nature of the lesion. For example, in the follow-up of chemotherapy for lung cancer, the reduction of volume accompanied by the decrease of density indicates that the treatment is effective, and the increase of density heterogeneity indicates the risk of drug resistance. The process of lesion change is tracked and recorded at continuous time points, and the treatment effect is evaluated by quantitative feature analysis. Different types of tumor lesions show specific progression patterns, and the combination of feature parameters reflects the characteristics of biological behavior. In the feature analysis of lesion progression, static morphological parameters and dynamic change indicators complement each other to build a complete quantitative evaluation system. During data processing, smoothing and fitting parameters are adjusted according to the time resolution to ensure the accuracy of feature extraction. In the analysis of heterogeneous lesions, multi-level feature extraction and parameter fusion are combined to comprehensively characterize the progression law. Standardization processing realizes feature comparison between different cases and supports clinical group research.

[0029] S105. According to the dynamic change law during the progression of the lesion, the irregularity of the lesion shape and the smoothness of the boundary are calculated to evaluate the morphological complexity and boundary clarity of the lesion. At the same time, the density difference between the lesion area and the surrounding normal tissue is compared to obtain a quantitative index of the density difference of the tissue around the lesion.

[0030] The sobel operator is used to calculate the gradient of the grayscale image of the lesion area, and a boundary point set is obtained according to a preset threshold, and a closed contour line is generated through the boundary point set; a reference point sequence is obtained for the closed contour line, a discrete curvature value sequence is calculated according to the reference point sequence, and the proportion of points exceeding the curvature threshold in the discrete curvature value sequence is counted to obtain the boundary curvature; an annular area is constructed according to the closed contour line, and the annular area is obtained by expanding the lesion boundary outward by a number of pixels, radial sampling points are obtained from the annular area along the boundary sampling, and radial sampling lines are constructed through the radial sampling points; a density change curve is fitted according to the density sequence on the radial sampling line, a boundary clarity index and a density difference parameter are calculated through the density change curve, and a lesion invasion degree index is obtained according to the boundary clarity index and the density difference parameter.

[0031] Specifically, based on the grayscale image of the lesion area, the Sobel operator was used to calculate the image gradient, and the boundary point set was extracted by using 50% of the maximum gradient value as the threshold. The boundary points were connected in a clockwise direction to form a closed contour line, and the Graham scanning tool was used to calculate the minimum circumscribed polygon. The number of pixels in the contour line was counted to obtain the actual area, and the ratio of the actual area to the area of ​​the minimum circumscribed polygon was calculated to obtain the shape irregularity index. For the boundary contour line, a reference point was taken every 5 pixels along the contour line, and a circular window with a radius of 7 pixels was taken with the reference point as the center. The discrete curvature value of the boundary point in the window was calculated, and the curvature threshold was set to 0.3. The proportion of the number of points exceeding the threshold to the total number of reference points was counted to obtain the boundary curvature. The 5-point mean filter was used to smooth the boundary point coordinate sequence, and the ratio of the boundary length before and after smoothing was calculated to obtain the boundary smoothness parameter. The boundary of the lesion area was extended outward by 10 pixels to construct an annular area. The Sobel operator was used to calculate the gradient map in the annular area. A sampling point was taken every 3 pixels along the boundary. 5 pixels were extended from the sampling point to the inside and outside directions to form a radial sampling line. The density value sequence on the radial sampling line was extracted, and the mean and standard deviation of the sequence were calculated to obtain the density distribution characteristics. Based on the density sequence on the radial sampling line, the density change curve was fitted by the least squares method. The slope of the curve at the midpoint of the sampling line was calculated as the density gradient value. The gradient threshold was set to 20% of the density mean. The proportion of sampling lines greater than the threshold was statistically obtained to obtain the boundary clarity index. The mean value of the inner and outer density differences was calculated to obtain the tissue density difference parameter. The boundary clarity index and the density difference parameter were multiplied to obtain a quantitative indicator of the degree of lesion invasion. In medical image analysis, the morphological characteristics and boundary characteristics of lesions reflect their biological behavior. The Sobel operator has directional sensitivity in edge detection. For example, in dynamic enhanced scanning of the liver, the 50% maximum gradient threshold retains the main boundary and suppresses noise interference. The boundary points are connected clockwise to form a closed contour. The minimum circumscribed polygon is constructed by the Graham scanning tool to reflect the appearance characteristics of the shape, and the area ratio quantifies the degree of shape irregularity. In the boundary morphology analysis, the sampling interval of every 5 pixels balances the calculation amount and accuracy requirements. In the CT image of the lung nodule, the curvature calculation window with a radius of 7 pixels adapts to the local changes of the boundary. The curvature threshold of 0.3 distinguishes between smooth boundaries and sharp turns, and the curvature statistics reflect the complexity of the boundary. The 5-point mean filter eliminates boundary noise. For example, in the ultrasound image of breast masses, the ratio of the length before and after smoothing reflects the degree of boundary irregularity. The annular area analysis extends the range of 10 pixels. In the evaluation of pancreatic space-occupying lesions, the transition zone of the lesion boundary is included. The radial sampling of 3 pixels intervals ensures complete boundary coverage, and the inner and outer extensions are 5 pixels each to obtain the change of tissue density. The sampling line density sequence statistics reflect both local characteristics and overall distribution, and reveal the characteristics of boundary invasion in lymph node examination. The density change curve fitting describes the characteristics of tissue transition. For example, in the enhanced scanning of brain glioma, the midpoint slope reflects the sharpness of the tissue interface.The gradient threshold of 20% density mean distinguishes clear and fuzzy boundaries, indicating the depth of invasion in colorectal cancer staging. Boundary clarity and density difference comprehensively reflect the degree of invasion and support staging diagnosis. In the analysis of lesion morphology and boundaries, irregularity and density difference support each other. For example, in multi-parameter imaging of prostate cancer, irregular boundaries are often accompanied by changes in density gradient. In the analysis of continuous lesions, morphological parameters reflect overall characteristics, and density characteristics reflect local changes. In time series follow-up, changes in boundary characteristics reflect lesion progression and support treatment monitoring. In the analysis of heterogeneous lesions, multi-level feature extraction confirms each other. In the evaluation of soft tissue tumors, morphological complexity and boundary fuzziness jointly indicate the degree of malignancy. In the quantification of boundary features, geometric parameters and density parameters complement each other to build a complete quantitative evaluation system. In the analysis of different imaging sequences, parameter settings are adjusted according to image characteristics to ensure the accuracy of feature extraction. In the evaluation of lesion invasiveness, morphological features and density features are used together to provide objective basis.

[0032] S106. Predicting the progression of the disease based on the lesion volume characteristic parameters, density statistical parameters, and quantitative indicators of the density difference of tissues surrounding the lesion.

[0033] Volume feature parameters and density statistical parameters are extracted from the lesion time series images, and normalized feature data are obtained by minimum and maximum standardization processing. A correlation coefficient matrix is ​​calculated based on the normalized feature data, and a feature subset is obtained from the correlation coefficient matrix; feature means and standard deviations in different disease progression categories are calculated based on the feature subsets, and features with a discrimination degree higher than a threshold are obtained by the ratio of inter-class discreteness to intra-class discreteness, and feature vectors are obtained by sorting in descending order of information gain rate; the training data are cross-validated using the feature vectors, and a validation index is calculated using a support vector machine classifier, and an optimal parameter combination is selected according to the validation index to train a classification model; new feature data are received, the new feature data are standardized, and a distance from a feature point to a classification boundary is calculated based on the classification model, and the disease progression category and confidence are determined by the classification boundary distance.

[0034] Specifically, based on the volume feature parameters and density statistical parameters extracted from the lesion time series images, the minimum and maximum standardization tools were used to normalize the feature values, and the normalization interval was set to 0 to 1. The correlation coefficient matrix between the two features after standardization was calculated, and the correlation coefficient threshold was set to 0.8. One of the feature pairs with a correlation coefficient higher than the threshold was retained, and the other was eliminated to obtain the feature subset after eliminating redundancy. For each feature in the feature subset, the mean and standard deviation of the feature in different disease progression categories were calculated, and the discrimination threshold was set to 0.5. The ratio of the inter-class dispersion to the intra-class dispersion of the feature value distribution was statistically calculated. The features with a discrimination exceeding the threshold were selected from the calculation results, and the features were sorted in descending order according to the information gain rate. The features ranked in the top 75% were retained to construct the feature vector. The data set was divided by 5-fold cross validation, and a support vector machine classifier was constructed for each fold of data. The Gaussian kernel function parameter was set to 0.1 and the penalty parameter was set to 10. The classification accuracy, sensitivity, and specificity were calculated on the validation set. The parameter combination with the best validation index was selected as the final model parameter, and the classification model was retrained through the entire training set. The input new feature data was standardized, and the distance from the feature point to the classification boundary was calculated using the trained classification model. The boundary threshold was set to positive or negative 0.5. If the distance was greater than 0.5, it was judged as accelerated progression, if the distance was less than -0.5, it was judged as slowed progression, and if the distance was between -0.5 and 0.5, it was judged as stable progression. The disease progression category and confidence were obtained from the judgment results. The confidence values ​​of each disease progression category were obtained and converted into corresponding predicted probability values ​​by methods such as the softmax function. In medical image feature analysis, data standardization is crucial to eliminate dimensional differences. For example, in the follow-up evaluation of liver tumors, the volume change rate and density distribution parameters are comparable after being standardized from 0 to 1. The correlation coefficient matrix reflects the strength of the association between features. For example, in the dynamic enhancement of nodules, the volume growth rate and the density change rate often show a strong correlation. The correlation threshold of 0.8 balances the independence of features and the completeness of information. The feature selection process focuses on the ability to distinguish between categories. For example, in the classification of pancreatic space-occupying lesions, benign and malignant lesions show significant differences in density distribution characteristics. The discrimination threshold of 0.5 selects features with identification value, and the discrimination calculation considers intra-class aggregation and inter-class separability. The 75% feature retention ratio strikes a balance between information preservation and dimensionality reduction simplification. In the prediction of lesion progression, morphological features and density features complement each other. Cross-validation uses 5-fold division to ensure model stability. For example, in the prognostic evaluation of breast masses, different folds of data reflect the consistency of feature expression. The Gaussian kernel function parameter of 0.1 adapts to nonlinear feature mapping, and the penalty parameter of 10 weighs the generalization ability of the model. The validation index comprehensively evaluates the classification performance. The accuracy reflects the overall judgment accuracy, the sensitivity reflects the ability to identify progression, and the specificity represents the accuracy of stability judgment.The distance threshold setting of the classification boundary reflects the prediction confidence. In the monitoring of brain glioma progression, the positive and negative thresholds of 0.5 distinguish between clear progression and waiting to be observed. Distance calculation quantifies the reliability of classification. The larger the absolute value, the more certain the prediction. The determination of progression category combines the changes in features at multiple time points to dynamically reflect the evolution trend of the disease. In the extraction of lesion features, time series data reflects the law of dynamic changes. In the follow-up of colorectal cancer surgery, the shape of the volume growth curve corresponds to different prognostic types. In the feature selection process, statistical characteristics and clinical relevance support each other to build a stable and reliable prediction index system. In classifier training, parameter optimization and performance evaluation promote each other to improve prediction accuracy. In multimodal feature fusion analysis, features of different dimensions reflect different attributes of lesions. For example, in the prediction of lymph node metastasis, morphological changes and density changes jointly indicate prognostic risks. Data preprocessing and feature selection complement each other to ensure the quality of input features. Model training and verification complement each other to establish a reliable prediction tool. In the prediction of disease progression, quantitative features and classification judgments support each other and provide a basis for clinical decision-making. In the related feature processing flow, the parameter settings of each link are based on clinical practice experience to build a complete prediction system. In the prediction of different types of lesions, the feature extraction and model training parameters are adjusted according to the specific application scenarios to ensure the reliability of the prediction results.

[0035] S107: Determine whether the prediction result of the disease progression reaches a preset risk threshold range. If the prediction result is within the preset risk threshold range, determine that the disease progression is in a risk state and generate high-risk warning information.

[0036] A standardized prediction value is obtained based on the predicted probability value using the minimum and maximum standardization tool, and the standardized prediction value is within the interval of zero to one; an upper limit value and a lower limit value of the risk threshold are set for the standardized prediction value, and a risk status mark is obtained by calculating the positional relationship between the standardized prediction value and the risk threshold interval; a confidence interval is calculated for the standardized prediction value using an interval estimation tool based on the standard deviation, and an overlapping ratio value is obtained by counting the overlapping range of the confidence interval and the risk threshold interval; the risk level is determined based on the overlapping ratio value, and if the overlapping ratio value exceeds the preset overlapping threshold, the risk level is increased, otherwise the risk level is reduced, and a risk score is constructed through the risk level and the overlapping ratio value; a warning level mark is set based on the risk score, and the warning level mark includes a high-risk red mark, a medium-risk yellow mark, and a low-risk green mark, and risk warning information is generated through the warning level mark and the timestamp.

[0037] Specifically, according to the predicted probability value obtained by predicting the progression of the disease, the minimum and maximum standardization tools are used to map the predicted value to the interval of 0 to 1, and the upper limit of the standardized risk threshold is set to 0.8 and the lower limit is set to 0.6. The position relationship between the current predicted value and the risk threshold interval is calculated. If the predicted value falls within the threshold interval, the risk status mark is recorded, and the relative position percentage of the predicted value in the interval is calculated. For the confidence interval of the predicted value, the interval estimation tool based on the standard deviation is used to calculate the 95% confidence interval, and the overlap ratio of the confidence interval and the risk threshold interval is counted. The overlap ratio threshold is set to 0.3. If the overlap ratio exceeds the threshold, the risk level is increased, otherwise the risk level is reduced. The risk score is constructed from the risk level and the overlap ratio. The risk score is divided into three levels, with 90 points or more as the high risk level, 70 to 90 points as the medium risk level, and 70 points or less as the low risk level. The corresponding risk level is selected according to the risk score calculation result, and the red, yellow and green three-color warning signs are set according to the risk level to generate a risk mark containing the warning level and timestamp. The warning information is constructed in the key-value pair format, and six key fields are set: warning title, predicted value, risk score, risk level, warning time, and confidence interval. The warning prompt in the standard format is generated by string concatenation. Different prompt level tags are set according to the warning level to complete the generation of risk warning information. In the medical warning system, the standardization of risk assessment is crucial. For example, in the prediction of liver cancer recurrence risk, the predicted probability is standardized from 0 to 1 to facilitate threshold setting. The double threshold design of 0.8 and 0.6 constructs a buffer zone to avoid frequent alarms due to prediction fluctuations. The relative position of the predicted value in the threshold interval reflects the degree of risk. For example, the predicted value of 0.75 is located at 75% of the interval, indicating a high degree of risk. Confidence interval analysis reflects the reliability of prediction. In the follow-up evaluation of lung nodules, the probability that the 95% confidence interval contains the real risk is high. The overlap ratio threshold of 0.3 balances sensitivity and specificity. For example, in the prediction of breast mass progression, a large overlap ratio indicates that the prediction result is uncertain and vigilance is required. The risk score comprehensively considers the predicted value and uncertainty to form a more robust basis for warning. The three-level risk classification corresponds to the needs of clinical practice. In the monitoring of pancreatic cystic tumors, a high-risk score of more than 90 points indicates the need for timely intervention. The red, yellow and green warning signs intuitively reflect the risk level, and the moderate risk status of 70 to 90 points requires enhanced follow-up. The risk level is closely related to clinical treatment, and different levels trigger different warning responses. Standardization of warning information improves the efficiency of message transmission. For example, in postoperative monitoring, the key-value pair format contains complete warning elements. The predicted value, risk score, and confidence interval constitute a quantitative description, and the timestamp records the timing of the warning. Multi-dimensional information supports clinical decision-making, and different levels of marking guide intervention measures. In the warning of disease progression, prediction accuracy and warning timeliness complement each other. For example, in the follow-up evaluation of brain glioma, the dynamic comparison of the predicted probability and the risk threshold reflects the trend of disease changes.Confidence interval analysis reveals the reliability of predictions, and risk scoring quantifies the urgency of early warning. Standardization of early warning information facilitates clinical application, and the standard format facilitates information transmission and processing. In the risk assessment system, multi-level early warning mechanisms support each other. For example, in lymphoma efficacy monitoring, a predicted value exceeding 0.8 directly triggers a high-risk warning, while a value between 0.6 and 0.8 is based on confidence judgment. The overlap of the 95% confidence interval and the risk threshold indicates a boundary situation, which requires a comprehensive assessment in combination with clinical manifestations. The three-level risk division corresponds to different follow-up strategies, and the standardization of early warning information promotes multidisciplinary collaboration. In the monitoring of heterogeneous lesions, risk assessment standards need to be adjusted dynamically. There are differences in the risk characteristics of different types of tumors, and the setting of early warning parameters should be optimized according to the specific disease. Standardized early warning mechanisms ensure the accuracy of information transmission and support clinical risk management. In the early warning process, quantitative indicators and risk levels complement each other to build a comprehensive early warning system.

[0038] S108. Based on the judgment result of the disease progression prediction and the fitted data of the changes in lesion volume and density over time, the severity of the disease is judged, the lesion progression process is staged and graded, and a quantitative disease diagnosis report is generated.

[0039] A quadratic polynomial fitting tool is used to perform piecewise fitting on the curves of lesion volume and density changes over time, and the volume change rate and density change rate in each interval are calculated based on the curve; the lesion progression grade is judged based on the volume change rate and density change rate, and a severity score is obtained for the progression grade, and the severity score is composed of a volume progression score, a density change score, and a predicted probability score; the disease stage status is determined based on the changing trend of the progression grade at consecutive time points, if the progression grade increases for several consecutive time points, it is determined to be a progressive stage, and if the progression grade is stable for several consecutive time points, it is determined to be a stable stage; a diagnosis report content is generated based on the disease stage status and severity score, and the diagnosis report content includes the volume density change rate, the severity score, and the disease stage result.

[0040] Specifically, according to the prediction results of disease progression and the data of changes in lesion volume density, the quadratic polynomial fitting tool was used to fit the curves of volume and density changes over time in segments, and the length of the fitting interval was set to 3 months. The change rate in each interval was calculated. For the volume change rate, 20% per month was set as the rapid progression threshold, 10% as the moderate progression threshold, and 15% per month as the significant change threshold, and 5% as the mild change threshold for the density change rate. The progression level of volume and density was obtained from the change rate values. According to the volume density progression level and predicted probability value, a severity scoring system was constructed, with a volume progression score interval of 0-40 points, a density change score interval of 0-30 points, and a predicted probability score interval of 0-30 points. The total score was calculated according to the score corresponding to the progression level, and a total score of 90-100 points was divided into extremely severe, 75-89 points into severe, 60-74 points into moderate, and below 60 points into mild. The severity of the disease was determined from the scoring results. According to the change of progression level at continuous time points, the progression level of three consecutive time points is set as the progression period, the progression level of three consecutive time points is set as the stable period, and the progression level of three consecutive time points is set as the remission period. The disease is staged in combination with the severity score, and a standardized diagnosis result including the progression speed, severity, and staging stage is generated. The structured format is used to construct the content of the diagnosis report, and the patient information, examination time, lesion location, size and shape are set as basic information, the volume change rate, density change amplitude, severity score, and progression stage are set as diagnostic points, and the risk prediction probability and warning level are set as prompt information. The corresponding descriptive statements are generated through numerical indicator mapping, and the disease progression process is recorded in chronological order to complete the generation of the diagnosis report. In the quantitative analysis of medical images, the dynamic change characteristics of the lesion reflect the state of disease progression. Quadratic polynomial fitting can capture the acceleration or deceleration trend of the lesion. For example, in the follow-up evaluation of liver cancer, the fitting interval of 3 months balances sensitivity and stability. The rapid progression threshold of 20% volume growth per month corresponds to a shortened tumor doubling time, which indicates changes in biological behavior in nodule monitoring. The significant change threshold of 15% decrease in density reflects tissue necrosis in pancreatic cystic lesions. The severity scoring system integrates multi-dimensional characteristics. For example, in the dynamic enhancement of breast masses, volume progression accounts for 40% of the weight to reflect changes in tumor burden, and density changes and predicted probability each account for 30% of the weight to balance morphology and risk prediction. Extremely severe scores above 90 points correspond to a high-risk state of rapid progression, which indicates a poor prognosis in the evaluation of gliomas. Severe grades of 75-89 points require close follow-up, and moderate grades of 60-74 points indicate that the disease is controllable. Staging judgment is based on observations at consecutive time points. For example, in postoperative monitoring of colorectal cancer, accelerated progression at three consecutive points indicates an increased risk of recurrence. Characteristics of the progressive stage include rapid increase in volume and significant changes in density, which correspond to treatment resistance in the evaluation of lymphoma chemotherapy. The stable stage shows that the disease is under control, and the remission stage reflects the effectiveness of treatment.Severity scores and staging judgments are mutually verified to build a complete disease assessment system. Standardization of diagnostic reports improves the efficiency of information transmission. For example, in the follow-up of soft tissue tumors, basic information records the spatiotemporal information of lesions, and diagnostic points quantify the progression characteristics. A volume change rate of 20% / month is converted into a description of rapid progression, and a density change of 15% / month corresponds to a significant change prompt. A severity score of 85 points generates a description of severe progression, and a risk prediction probability of 0.8 triggers a high-risk warning. In the process of disease assessment, quantitative indicators and qualitative descriptions complement each other. For example, in prostate cancer imaging omics, volume density changes reflect histological changes, and the rate of progression corresponds to the degree of malignancy. Continuous time point observations reveal the evolution of the disease, and multi-dimensional scoring ensures the reliability of judgment. Standardized reports promote multidisciplinary collaboration and support clinical decision-making. In the diagnosis of different diseases, the evaluation criteria need to be optimized and adjusted. Tumors with high malignancy use more sensitive progression thresholds, and slow-growing lesions choose longer observation intervals. The severity score weight can be adjusted according to the characteristics of the disease to ensure the clinical relevance of the evaluation results. In the report generation process, the selection of descriptive statements strictly corresponds to quantitative indicators to ensure the accuracy of expression. The setting of early warning information is based on clinical intervention needs, and a standardized follow-up management system is established.

[0041] The above embodiment is only one of the preferred implementation modes of the present invention and should not be used to limit the protection scope of the present invention. Any changes or modifications that are made to the main design concept and spirit of the present invention and have no substantive significance, and the technical problems they solve are still consistent with the present invention, should be included in the protection scope of the present invention.

Claims

1. A method for analyzing contrast images for digital diagnosis and treatment equipment, characterized in that: The method comprises: Obtain the contrast image sequences collected from the target patient at different time points, and segment the lesion area based on the contrast images at different time points based on pixel-level features, voxel-level features, and region-level features, respectively, to obtain the number of pixels, average density, and grayscale value distribution of the lesion area at different time points; Extract and describe the features of the lesion area at different time points, generate feature descriptors, perform ratio tests on the feature descriptors, screen out feature matching points, calculate the transformation matrix between the lesion areas through the feature matching points, and obtain the changes between the lesion areas at different time points; After alignment and registration, the change data of lesion volume over time are obtained according to the number of pixels in the lesion area at each time point. At the same time, the average density of the lesion area is calculated according to the gray value of pixels in the lesion area to obtain the change data of lesion density over time; By fitting the data of the change of lesion volume over time and the change of lesion density over time, the law of the change of lesion volume and density over time is analyzed, and the volume dynamic characteristic parameters and density statistical parameters are obtained to describe the dynamic change law of the lesion progression process; According to the dynamic change law of the lesion during its progression, the irregularity of the lesion shape and the smoothness of the boundary are calculated to evaluate the morphological complexity and boundary clarity of the lesion. At the same time, the density difference between the lesion area and the surrounding normal tissue is compared to obtain a quantitative index of the density difference of the tissue around the lesion. Predict the progression of the disease based on lesion volume characteristic parameters, density statistical parameters, and quantitative indicators of the difference in density of tissues around the lesion; Determine whether the predicted result of the disease progression reaches the preset risk threshold range. If the predicted result is within the preset risk threshold range, the disease progression is judged to be a risk state and a high-risk warning information is generated; Based on the predicted results of the disease progression and the fitted data of the changes in lesion volume and density over time, the severity of the disease is judged, the lesion progression process is staged and graded, and a quantitative disease diagnosis report is generated.

2. The method according to claim 1, characterized in that The method comprises: obtaining a sequence of contrast images collected from a target patient at different time points, and segmenting the lesion area based on pixel-level features, voxel-level features, and region-level features for the contrast images at different time points, to obtain the number of pixels, average density, and grayscale value distribution of the lesion area at different time points, including: Performing Gaussian smoothing on the image according to the contrast image acquisition sequence, and using an edge tracking tool based on pixel grayscale gradient to obtain a lesion boundary contour map; For the lesion boundary contour map, the point with the maximum gray value is selected as the seed point, and the area is filled according to the rule that the difference of pixel density in eight neighborhoods is less than a preset threshold value, so as to obtain a lesion area density distribution map; Using the lesion area density distribution map to establish three-dimensional point cloud data of a preset voxel size, removing noise points from the point cloud data to obtain three-dimensional feature data; Extracting the centroid coordinates of the lesion area from the three-dimensional feature data to establish a registration reference point, performing nearest neighbor pairing on the lesion boundary points according to a preset distance threshold, and obtaining the volume data and grayscale distribution histogram of the lesion area; Also includes: By extracting regional level features of contrast images at different time points, calculating the feature quantization value of the lesion area, comparing the preset feature quantization threshold with the feature quantization value of the lesion area, and preliminarily segmenting to obtain the target area; presetting the selection conditions of seed points, selecting a number of seed points from the target area according to the selection conditions, and expanding the segmented area from the seed points until the preset expansion conditions are reached to achieve final segmentation, specifically including: using Gaussian smoothing kernel function to perform noise reduction processing on the contrast image sequence, and obtaining the candidate area through projection calculation according to the gray value distribution matrix of the denoised image; according to the local gray value mean and variance of the candidate area, obtaining the seed point position through the preset uniform area judgment condition, and using the gray difference threshold to perform regional expansion from the seed point to obtain the extended segmented area; extracting edge contour lines from the extended segmented area, obtaining feature points by calculating the edge gradient data in the detection window, and performing regional segmentation according to the curvature of the feature points until the area converges to obtain the final segmented area; extracting shape feature parameters from the final segmented area, judging the difference in regional morphological features through the feature point pairing distance threshold, and determining the corresponding area if the shape feature parameter difference is within the preset range and the feature point pairing rate exceeds the preset threshold.

3. The method according to claim 1, characterized in that The extracting and describing the features of the lesion regions at different time points, generating feature descriptors, performing ratio tests on the feature descriptors, screening out feature matching points, calculating the transformation matrix between the lesion regions through the feature matching points, and obtaining the changes between the lesion regions at different time points, comprises: The original image of the lesion area is smoothed by using a Gaussian blur tool, edge contour lines are extracted from the smoothed image by using an edge detection operator, and a gradient histogram is calculated according to the edge contour lines to obtain a feature descriptor; Calculating the Euclidean distance value between corresponding positions at different time points according to the feature descriptor, and obtaining an initial matching point set by comparing the Euclidean distance value with a preset distance threshold; Randomly select matching point pairs from the initial matching point set to calculate an affine transformation matrix, and count the number of interior points that satisfy the affine transformation matrix to obtain optimal transformation parameters; The lesion area image is spatially mapped according to the optimal transformation parameters, the transformed image is resampled using a bilinear interpolation tool, and the registration result is determined by calculating the mutual information value of the image before and after the resampling.

4. The method according to claim 1, characterized in that: After the alignment and registration, the change data of the lesion volume over time are obtained according to the number of pixels in the lesion area at each time point, and the average density of the lesion area is calculated according to the gray value of the pixels in the lesion area to obtain the change data of the lesion density over time, including: The lesion boundary contour line is obtained by using a region growing tool based on a seed point, and the lesion volume value is obtained by converting the number of pixels in the connected area within the contour line and the image resolution; For the image area within the lesion boundary contour line, a histogram equalization tool is used to perform enhancement processing, a gray value matrix is ​​extracted from the enhanced image, and a regional average density value is obtained by projection integral operation; Constructing time series data according to the lesion volume value and the average density value of the region, performing linear interpolation correction on abnormal points in the time series data that exceed preset values, and obtaining a normalized volume density change sequence; For the normalized volume density change sequence, the change rate is obtained by calculating the difference between adjacent time points, the mean and standard deviation of the change rate are used to construct a lesion progression feature parameter set, and the feature parameter set is normalized to obtain a standardized lesion progression feature; Also includes: According to the number of pixels in the lesion area at each time point, combined with the voxel-level features of the contrast image, the voxel-level features include voxel grayscale value, voxel spatial position and voxel shape size, the number of pixels and voxel volume at each time point are determined, the total volume of the lesion is calculated, the total volume of the lesion at each time point is recorded, and the change data of the lesion volume over time are obtained, which specifically includes: using Gaussian smoothing kernel function to perform noise reduction processing on the contrast image data, using a fixed threshold to perform binary segmentation on the smoothed image, and obtaining the voxel set of the initial lesion area through the region growing algorithm; The initial lesion area is divided into three-dimensional grids, and voxels are classified according to the Euclidean distance from the voxel to the grid center point. The connectivity is determined by the distance and gray value difference between adjacent grid units to obtain a three-dimensional grid connectivity map of the lesion area. A set of boundary points is extracted from the three-dimensional grid connectivity map, and a bounding box is constructed according to the maximum and minimum coordinate values ​​of the boundary points in three directions. The maximum diameter and volume of the lesion are obtained by calculating the diagonal length of the bounding box and counting the total number of voxels in the connected domain. A time series data table is constructed for the lesion volume, and the change data of the lesion volume over time are obtained by linear interpolation.

5. The method according to claim 1, characterized in that The method includes performing data fitting on the change data of lesion volume over time and the change data of lesion density over time, analyzing the change law of lesion volume and density over time, obtaining volume dynamic characteristic parameters and density statistical parameters, and describing the dynamic change law in the process of lesion progression, including: Using a cubic spline interpolation tool to smooth the time series data of the lesion volume and density, calculating derivative values ​​according to the smoothed data sequence, and marking the position of the feature point by the change of the sign of the derivative value; Fitting a second-order polynomial function to the lesion volume data according to the characteristic point positions, extracting quadratic term coefficients and linear term coefficients from the fitting function to obtain volume dynamic characteristic parameters; For the lesion density data, a preset threshold is used to divide the density data into a high value segment, an intermediate segment and a low value segment, and density change characteristic parameters are obtained through statistics of the data segments; A feature vector is constructed according to the volume dynamic feature parameter and the density change feature parameter, and the feature vector is normalized by using a minimum-maximum normalization method to obtain a lesion progression feature set.

6. The method according to claim 1, characterized in that The method comprises calculating the irregularity of the lesion shape and the smoothness of the boundary according to the dynamic change law during the progression of the lesion, evaluating the morphological complexity and boundary clarity of the lesion, and comparing the density difference between the lesion area and the surrounding normal tissue to obtain a quantitative index of the density difference of the tissue around the lesion, including: The sobel operator is used to calculate the gradient of the grayscale image of the lesion area, a boundary point set is obtained according to a preset threshold, and a closed contour line is generated through the boundary point set; A reference point sequence is obtained for the closed contour line, a discrete curvature value sequence is calculated according to the reference point sequence, and a proportion of points exceeding a curvature threshold in the discrete curvature value sequence is counted to obtain a boundary curvature; Constructing an annular region according to the closed contour line, wherein the annular region is obtained by expanding the lesion boundary outward by a number of pixels, sampling along the boundary of the annular region to obtain radial sampling points, and constructing radial sampling lines through the radial sampling points; A density variation curve is fitted according to the density sequence on the radial sampling line, a boundary clarity index and a density difference parameter are calculated through the density variation curve, and a lesion invasion degree index is obtained according to the boundary clarity index and the density difference parameter.

7. The method according to claim 1, characterized in that The method predicts the disease progression based on the lesion volume characteristic parameters, density statistical parameters and the quantitative index of the density difference of the tissue around the lesion, including: Extracting volume feature parameters and density statistical parameters from the lesion time series images, obtaining normalized feature data by minimum and maximum standardization processing, calculating a correlation coefficient matrix based on the normalized feature data, and obtaining a feature subset from the correlation coefficient matrix; Calculating the feature means and standard deviations in different disease progression categories based on the feature subsets, obtaining features with a discrimination degree higher than a threshold value by the ratio of the inter-class dispersion to the intra-class dispersion, and obtaining feature vectors by sorting in descending order of information gain rate; The feature vector is used to perform cross-validation on the training data, a validation index is calculated by a support vector machine classifier, and an optimal parameter combination is selected according to the validation index to train the classification model; New feature data is received, the new feature data is standardized, a distance from a feature point to a classification boundary is calculated according to the classification model, and a disease progression category and a confidence level are determined by the classification boundary distance.

8. The method according to claim 1, characterized in that The step of determining whether the predicted result of the disease progression reaches a preset risk threshold range, and if the predicted result is within the preset risk threshold range, determining that the disease progression is a risk state, and generating high-risk warning information, includes: A standardized prediction value is obtained by using a minimum-maximum standardization tool according to the predicted probability value, wherein the standardized prediction value is within the range of zero to one; Setting a risk threshold upper limit and a risk threshold lower limit for the standardized predicted value, and obtaining a risk status mark by calculating a positional relationship between the standardized predicted value and the risk threshold interval; The confidence interval is calculated for the standardized predicted value using an interval estimation tool based on the standard deviation, and the overlapping ratio value is obtained by counting the overlapping range of the confidence interval and the risk threshold interval; The risk level is determined according to the overlap ratio value. If the overlap ratio value exceeds a preset overlap threshold, the risk level is increased, otherwise the risk level is decreased, and a risk score is constructed by the risk level and the overlap ratio value; A warning level identifier is set according to the risk score, the warning level identifier includes a high-risk red identifier, a medium-risk yellow identifier and a low-risk green identifier, and risk warning information is generated through the warning level identifier and a timestamp.

9. The method according to claim 1, characterized in that: The judgment result predicted according to the disease progression is combined with the fitting data of the lesion volume and density changing over time to judge the severity of the disease, stage and grade the lesion progression process, and generate a quantitative disease diagnosis report, including: A quadratic polynomial fitting tool is used to perform segmented fitting on the curve of lesion volume and density change over time, and the volume change rate and density change rate in each interval are calculated based on the curve; Determine the progression grade of the lesion according to the volume change rate and the density change rate, and obtain a severity score for the progression grade, wherein the severity score is composed of a volume progression score, a density change score, and a predicted probability score; The disease stage is determined according to the changing trend of the progression level at consecutive time points. If the progression level increases at consecutive time points, it is determined to be a progressive stage; if the progression level is stable at consecutive time points, it is determined to be a stable stage; A diagnosis report is generated based on the disease stage status and severity score, and the diagnosis report includes the volume density change rate, severity score and disease stage result.

10. An angiographic image analysis device for digital diagnosis and treatment equipment, characterized in that: The device comprises: Image processing and feature extraction module, used to acquire and analyze the angiographic image sequences at different time points; Lesion area change analysis module, used to calculate the changes between lesion areas at different time points; Lesion dynamic feature analysis module, used to analyze the law of changes in lesion volume and density over time; Lesion morphology assessment module, used to assess the morphological complexity and boundary clarity of lesions; Disease prediction and risk assessment module, used to predict disease progression and generate high-risk warning information; The disease diagnosis and grading module is used to determine the severity of the disease and generate a quantitative diagnosis report.

Images