Medical image diagnosis method and system based on cloud platform

Abstract

The invention relates to the technical field of medical image processing, in particular to a medical image diagnosis method and system based on a cloud platform, and the method comprises the following steps: based on an input medical image, analyzing the texture density of a plurality of positions in the image, calculating the gray gradient change direction and rate of pixel points, and extracting boundary points of a focus region in the image; segmenting the lesion image to obtain lesion boundary data; according to the method, the accurate segmentation of the lesion area is realized by calculating the texture density of a plurality of positions in the image, and the spatial distribution of the lesion is predicted by using the tissue density value of the lesion area and the density change rate between adjacent pixels, so that the dynamic change of the lesion area is quantified, and the dynamic image sequence is analyzed; the lesion activity level grading is realized, the lesion development trend is more intuitive, the resolution of the image loading area is dynamically adjusted in combination with real-time user operation and the network state, the remote image access experience is optimized, and the image calling is more efficient and smoother.

Description

Technical Field

[0001] The present invention relates to the field of medical image processing technology, and in particular to a medical image diagnosis method and system based on a cloud platform. Background Art

[0002] The field of medical image processing technology includes the acquisition, storage, transmission, analysis, and visualization of medical images. The core content of this technical field includes processing medical image data through computer technology to assist doctors in clinical diagnosis and treatment decisions. Medical image processing involves multiple modes of medical imaging technology, such as X-ray, computed tomography CT, magnetic resonance imaging MRI, and ultrasound imaging. This field covers multiple aspects of medical image denoising, enhancement, segmentation, feature extraction, image registration, and three-dimensional reconstruction. It is used in disease screening, lesion detection, surgical planning, and treatment monitoring. Combined with the development of cloud computing, big data, and artificial intelligence technology, medical image processing is gradually developing towards intelligence and remoteness, making the analysis of medical image data more accurate and efficient.

[0003] Among them, the medical imaging diagnosis method based on cloud platform refers to the use of cloud computing architecture and distributed computing technology to remotely process, analyze and diagnose medical imaging data. The method covers technical means such as cloud storage, distributed computing, deep learning model reasoning and remote access of medical images. Relying on the computing resources of the cloud platform, through image preprocessing, feature extraction, classification and recognition, intelligent analysis of medical images is realized. Medical imaging data is acquired through image acquisition equipment and uploaded to the cloud storage module through the data interface. The cloud server is used to convert the image format, denoise and enhance it. The classification model based on convolutional neural network is used for target recognition and lesion segmentation. The diagnosis results and image analysis data are provided to the end users through the remote access module.

[0004] Traditional medical imaging diagnostic technology has shortcomings in the processing, analysis and remote access of medical images. The lesion identification process lacks refined texture density analysis, which leads to blurred lesion boundaries and affects the accuracy of segmentation. The spatial prediction of lesions mainly relies on global feature analysis, fails to fully consider local gradient changes, and reduces the accuracy of extension trend judgment. It mainly uses single-frame data and fails to effectively combine the change characteristics of multiple time points, making the assessment of lesion activity lack temporal coherence. The image storage method is relatively fixed and fails to classify according to multi-dimensional features such as lesion attributes and patient information, which reduces the accuracy and efficiency of image retrieval. Remote image access uses a fixed resolution loading method and fails to make dynamic adjustments based on access identity, user interaction and network status, so that image calls are limited by network bandwidth fluctuations, resulting in loading delays and a decreased access experience. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a medical imaging diagnosis method and system based on a cloud platform.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a medical imaging diagnosis method based on a cloud platform, comprising the following steps: S1: Based on the input medical image, analyze the texture density of multiple positions in the image, calculate the gray gradient change direction and rate of the pixel points, extract the boundary points of the lesion area in the image, segment the lesion image, and obtain the lesion boundary data; S2: Based on the lesion boundary data, the tissue density value of the lesion area is obtained, the density change rate between adjacent pixels is calculated, and the tissue density distribution data is obtained. By calculating the density gradient change trend in the lesion area, the expansion direction of the lesion area is analyzed, and the spatial distribution prediction result of the lesion area is constructed to establish the distribution position prediction result; S3: calling the distribution position prediction result, analyzing the patient's medical images at multiple time points, extracting the grayscale change information of the lesion area in the dynamic image sequence, calculating the signal change rate of multiple lesion areas in the image frame, identifying the activity levels of multiple lesions, and outputting the activity rate classification data; S4: Based on the activity rate classification data, a hierarchical index rule library is established according to the type, location, activity level, and patient information of the lesion. By matching the image content and index rules, the medical images are classified and stored in the cloud platform to generate an image data index.

[0007] As a further scheme of the present invention, the lesion boundary data includes the coordinates of the lesion contour points, the boundary grayscale gradient direction, and the boundary pixel change rate; the distribution position prediction result is specifically the lesion expansion direction, the lesion spatial distribution area, and the lesion boundary change trend; the activity rate grading data includes the lesion grayscale change rate, the lesion signal change level, and the lesion activity classification result; the image data index specifically refers to the image classification label, the image storage path, and the image type identifier.

[0008] As a further solution of the present invention, the step of acquiring the lesion boundary data is specifically as follows: S111: based on the input medical image, obtaining the pixel grayscale value of the medical image, calculating the grayscale change rate of each pixel point within the neighborhood range, obtaining the grayscale gradient value, and calculating the grayscale change amplitude of the pixel point in each direction to generate texture density data; S112: Based on the texture density data, calculate the density gradient of the pixel points in multiple directions, obtain the gradient change rate and calculate the gradient ratio between pixels, calibrate the boundary candidate points, and establish the boundary point extraction result; S113: Based on the boundary point extraction result, the gradient change consistency of adjacent boundary points is calculated using the formula: ; Calculate the adjusted position of the boundary point, optimize the boundary shape, segment the lesion image, and generate lesion boundary data; in, Represents the optimized The coordinates of the boundary points, For the original The coordinates of the boundary points, For the The gray value of the boundary point, For the The grayscale values ​​of adjacent boundary points are For the The gradient change rate of the boundary point is For the The gradient change rate of adjacent boundary points is is the weight coefficient of the boundary point, is the smoothing adjustment coefficient, Indicates the current boundary point number, Indicates the next adjacent boundary point of the current boundary point.

[0009] As a further solution of the present invention, the step of obtaining the distribution position prediction result is specifically: S211: extracting gray values ​​of pixels in the lesion area based on the lesion boundary data, constructing a gray value matrix, evaluating changes in tissue density in the lesion area, and generating tissue density distribution data; S212: Based on the tissue density distribution data, calculate the density gradient change rate between adjacent pixel points, calculate the gray value difference between adjacent pixel points, obtain the density change gradient matrix, analyze the expansion direction of the lesion area, and generate expansion direction trend data; S213: Based on the expansion direction trend data, density change rate calculation is performed on the boundary pixel points of the lesion area, using the formula: ; Calculate the spatial expansion trend of the lesion area, combine the gradient change rate of the pixel points, establish the spatial expansion prediction map of the lesion area, and obtain the distribution position prediction result; in, is the spatial expansion trend of the lesion area, For the The density gradient value of each pixel, For the The density gradient value of adjacent pixels, For the The gray value of a pixel, For the The grayscale values ​​of adjacent pixels, is the total number of pixels at the boundary of the lesion area, where is the number of the current boundary pixel, For and number The next adjacent boundary pixel.

[0010] As a further solution of the present invention, the step of acquiring the activity rate classification data is specifically as follows: S311: calling the distribution position prediction result, analyzing the patient's medical images at multiple time points, obtaining the lesion area in each frame of the image, extracting the grayscale value of the lesion area, calculating the average grayscale of the lesion area in multiple image frames, and generating grayscale change data of the lesion area; S312: Based on the grayscale change data of the lesion area, the grayscale change amplitude between adjacent frames is calculated to obtain the grayscale change rate of the lesion area in the image sequence, using the formula: ; Obtaining the signal change rate of the lesion area by calculation, and establishing lesion signal change rate data; in, Representative The signal change rate of the lesion area is For the The lesion area is The average gray value of the frame, For the The average gray value of the frame, is the standard deviation of grayscale changes in the lesion area between all frames, is the time interval between adjacent frames, is the total number of frames in the image sequence, Indicates the serial number of the lesion area. Indicates the time index of the current frame, Indicates the next adjacent frame time index; S313: Identify the activity levels of multiple lesions in the patient's medical image based on the lesion signal change rate data, and generate activity rate grading data.

[0011] As a further solution of the present invention, the steps of obtaining the image data index are specifically as follows: S411: extracting the type, location, activity level, and patient information of the lesion based on the activity rate classification data, forming key items of the lesion index, and constructing a hierarchical index rule base; S412: Based on the hierarchical index rule base, the characteristic parameters of the medical image data are analyzed, and the image data and the index rules are matched, using the formula: ; Calculate the image classification weight value, identify the categories of multiple medical images in the index library, and generate image classification matching results; in, is the image classification weight value, For the The matching degree between the image data and the index rules, For the The classification weight of image data, For the The reference value of the image data, For the The target adjustment value of the image data, is the total number of image data, is the index of the image data; S413: Based on the image classification and matching results, extract the index parameters of the classified images, establish index association relationships, and store them in the cloud platform to generate image data indexes.

[0012] As a further embodiment of the present invention, the method further comprises: S5: calling the image data index, analyzing the access request information of the cloud platform, extracting user permissions according to the access identity, matching the accessible image resolution, dynamically adjusting the resolution of the image loading area in combination with the user's real-time operation and network status, and generating image access management parameters; The image access management parameters include user authority level, image resolution matching rules, and dynamic loading area range.

[0013] As a further solution of the present invention, the steps of obtaining the image access management parameters are specifically as follows: S511: calling the image data index, parsing the identity information of the user access request, obtaining the user authority level, and generating user authority matching data; S512: According to the user authority matching data and the authority level of the user's current access request, matching the image resolution loading parameters for the user to generate an image resolution permission value; S513: Based on the image resolution permission value, real-time user operation information and current network status parameters are collected in real time, using the formula: ; Obtain the resolution adjustment amount of the image loading area by calculation, dynamically adjust the resolution of the image loading area, and generate image access management parameters; in, The amount to adjust the resolution of the image loading area. is the image resolution permission value corresponding to the user's permission. For user The adjustment ratio of real-time operation, The effect of operation type on image resolution. is the current network bandwidth value, is the network bandwidth stability adjustment coefficient, The total amount of real-time operation data for users. It is the index number of the user's real-time operation data.

[0014] A cloud-based medical imaging diagnosis system, which is used to perform the cloud-based medical imaging diagnosis method, comprises: The image boundary detection module is based on the input medical image, analyzes the texture density distribution of the image, performs grayscale gradient calculation, obtains the gradient change direction and rate, extracts the boundary information of the lesion area, and obtains the lesion boundary data; The spatial distribution prediction module extracts the tissue density value of the lesion area based on the lesion boundary data, calculates the density change rate of adjacent pixels, obtains tissue density distribution data, analyzes the trend of the lesion expansion direction, calculates the expansion probability of the lesion boundary point, constructs the spatial distribution prediction result of the lesion area, and obtains the distribution position prediction result; The lesion dynamic analysis module calls the distribution position prediction result, extracts the grayscale change data of the lesion area in the dynamic image sequence for the patient's medical images at multiple time points, calculates the intensity change rate of the lesion signal between adjacent frames, identifies the activity level of the lesion, and generates activity rate classification data; The classification storage management module establishes a classification index rule base based on the activity rate classification data, combined with the type, location, activity level, and basic information of the lesion, and classifies and stores the images by matching the image content and the index rules to generate an image data index; The remote access management module calls the image data index, parses the access request information of the cloud platform, matches the image loading resolution parameters according to the access user's identity authority, dynamically adjusts the resolution of the image loading area in combination with the user's interactive behavior and network status, and generates image access management parameters.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by calculating the texture density of multiple positions in the image, accurate segmentation of the lesion area is achieved. The spatial distribution of the lesion is predicted by using the tissue density value of the lesion area and the density change rate between adjacent pixels, so that the dynamic changes of the lesion area can be quantified. The analysis of dynamic image sequences realizes the classification of lesion activity levels, making the development trend of the lesion more intuitive. Combined with real-time user operations and network status, the resolution of the image loading area is dynamically adjusted, the experience of remote image access is optimized, and image calls are made more efficient and smooth. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the workflow of the present invention; Figure 2 A flow chart for acquiring lesion boundary data of the present invention; Figure 3 A flow chart for obtaining the distribution location prediction result of the present invention; Figure 4 The activity rate classification data acquisition flow chart of the present invention; Figure 5 A flowchart for obtaining an image data index of the present invention; Figure 6 This is a flow chart of obtaining image access management parameters of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0019] See also Figure 1 The present invention provides a technical solution: a medical imaging diagnosis method based on a cloud platform, comprising the following steps: S1: Based on the input medical image, analyze the texture density of multiple positions in the image, calculate the gray gradient change direction and rate of the pixel points, extract the boundary points of the lesion area in the image, segment the lesion image, and obtain the lesion boundary data; S2: Based on the lesion boundary data, the tissue density value of the lesion area is obtained, the density change rate between adjacent pixels is calculated, and the tissue density distribution data is obtained. By calculating the density gradient change trend in the lesion area, the expansion direction of the lesion area is analyzed, and the spatial distribution prediction result of the lesion area is constructed to establish the distribution position prediction result; S3: Call the distribution position prediction results, analyze the patient's medical images at multiple time points, extract the grayscale change information of the lesion area in the dynamic image sequence, calculate the signal change rate of multiple lesion areas in the image frame, identify the activity levels of multiple lesions, and output the activity rate classification data; S4: Based on the activity rate classification data, a hierarchical index rule library is established according to the type, location, activity level, and patient information of the lesion. By matching the image content and index rules, the medical images are classified and stored in the cloud platform to generate an image data index; S5: Call the image data index, analyze the access request information of the cloud platform, extract user permissions based on the access identity, match the accessible image resolution, dynamically adjust the resolution of the image loading area based on the user's real-time operation and network status, and generate image access management parameters.

[0020] The lesion boundary data include the coordinates of the lesion contour points, the direction of the boundary grayscale gradient, and the change rate of the boundary pixels. The distribution position prediction results are specifically the lesion expansion direction, the lesion spatial distribution area, and the lesion boundary change trend. The activity rate classification data include the lesion grayscale change rate, the lesion signal change level, and the lesion activity classification results. The image data index specifically refers to the image classification label, image storage path, and image type identification. The image access management parameters include user authority level, image resolution matching rules, and dynamic loading area range.

[0021] See also Figure 2 , the specific steps for obtaining the lesion boundary data are: S111: based on the input medical image, obtaining the pixel grayscale value of the medical image, calculating the grayscale change rate of each pixel point within the neighborhood range, obtaining the grayscale gradient value, and calculating the grayscale change amplitude of the pixel point in each direction to generate texture density data; Based on the input medical image, the pixel grayscale value is obtained. The grayscale value of each pixel is between Represents black, Represents white. After obtaining the gray value of a single pixel, a neighborhood calculation window centered on the pixel is established, which is usually set to Pixel unit to ensure that sufficient local texture information is included in the calculation. In this window, for each pixel point, traverse its neighboring pixels and calculate its grayscale change rate. During the calculation, perform differential operations on the grayscale values ​​of adjacent pixels, record the differential results and calculate their average to characterize the grayscale changes around the pixel point. At the same time, calculate the mean of the grayscale values ​​of all pixels in the window as a reference benchmark, calculate the deviation of the grayscale values ​​of all pixels from the benchmark mean, and calculate the sum of the squares of the deviations and take the average to obtain the grayscale variance of the window. The variance can characterize the degree of grayscale change in the pixel area. The variance calculation formula is as follows: ; in, is the variance, For the The gray value of a pixel, is the average gray value of the pixels in the window, is the total number of pixels in the window.

[0022] Set a pixel The neighborhood window contains the following pixel grayscale values: 120, 125, 130,128, 124,126,127, 130, 123,122,121, 129, 130,128, 126, 127,125, 124,123,121,122, 120,126,129,130; Calculate the average gray value: ; Calculate the variance: ; The variance value is used to characterize the grayscale change of the pixel area, and finally the texture density data is obtained.

[0023] S112: Based on the texture density data, calculate the density gradient of the pixel in multiple directions, obtain the gradient change rate and calculate the gradient ratio between pixels, calibrate the boundary candidate points, and establish the boundary point extraction result; Based on the texture density data, the density gradient of each pixel in different directions is calculated. Four directions are set: horizontal, vertical, left diagonal, and right diagonal. The gradient change rate is calculated for each direction. When calculating, for the current pixel, the grayscale value of the adjacent pixel is obtained, the difference between the two is calculated, and the difference is divided by the pixel distance between the two pixels to obtain the gradient change rate. The calculation formula is as follows: ; middle, For direction The gradient change rate on is the distance between pixels, usually set to 1 pixel unit, is the gray value of the adjacent pixel, is the gray value of the current pixel.

[0024] Calculate the gradient ratio between adjacent pixels: ; in, is the gradient ratio of adjacent pixels, is the gradient change rate of adjacent pixels, is the gradient change rate of the current pixel.

[0025] Set the current pixel , adjacent pixels , the distance between pixels : ; Assume that the next pixel , calculate the gradient ratio between adjacent pixels: ; If the gradient ratio Exceeding the set threshold , then the pixel is marked as a boundary candidate point. Finally, the boundary point extraction result is established.

[0026] S113: Based on the boundary point extraction result, the gradient change consistency of adjacent boundary points is calculated using the formula: ; Calculate the adjusted position of the boundary point, optimize the boundary shape, segment the lesion image, and generate lesion boundary data; in, Represents the optimized The coordinates of the boundary points, For the original The coordinates of the boundary points, For the The gray value of the boundary point, For the The grayscale values ​​of adjacent boundary points are For the The gradient change rate of the boundary point is For the The gradient change rate of adjacent boundary points is is the weight coefficient of the boundary point, is the smoothing adjustment coefficient, Indicates the current boundary point number, Indicates the next adjacent boundary point of the current boundary point; Based on the boundary point extraction results, the gradient change consistency of adjacent boundary points is calculated, and gradient smoothing is used to calculate the boundary adjustment parameters to correct the boundary point position and make the boundary smoother. The calculation formula is as follows: ; in, After optimization The coordinates of the boundary points, For the original The coordinates of the boundary points, For the The gray value of the boundary point, For the The grayscale values ​​of adjacent boundary points are For the The gradient change rate of the boundary point is For the The gradient change rate of adjacent boundary points is is the weight coefficient of the boundary point, is the smoothing adjustment factor.

[0027] Set boundary point coordinates , , , , , ,set up : ; Finally, the optimized boundary point coordinate set is obtained , used for further segmentation of the lesion area.

[0028] See also Figure 3 , the specific steps for obtaining the distribution location prediction results are: S211: based on the lesion boundary data, extract the grayscale values ​​of the pixels in the lesion area, construct a grayscale value matrix, evaluate the change of tissue density in the lesion area, and generate tissue density distribution data; Based on the lesion boundary data, the grayscale values ​​of all pixels in the lesion area are extracted, and the grayscale matrix is ​​established according to the pixel coordinates. The grayscale mean of each pixel in the neighborhood is calculated, and the difference between the grayscale value of each pixel and the neighborhood mean is calculated to obtain the local grayscale deviation. The local grayscale deviation mean of all pixels in the lesion area is calculated to obtain the overall tissue density distribution of the area, and the local grayscale deviation of all pixels is standardized to remove the influence of extreme values ​​and ensure uniform data distribution.

[0029] The calculation formula is: ; Calculate the tissue density distribution within the lesion area, where Represents the average gray value of all pixels in the area.

[0030] Assuming that the lesion area contains 4 pixels, whose grayscale values ​​are 100, 120, 110 and 130 respectively, the average grayscale value is calculated as follows: ; Then calculate the grayscale deviation of each pixel, respectively. , , and . Sum all the deviations and take the average: ; The calculation results show that the tissue density value of the lesion area is 10, and the subsequent steps will perform gradient calculation based on this value.

[0031] S212: Based on the tissue density distribution data, the density gradient change rate between adjacent pixels is calculated, the gray value difference between adjacent pixels is calculated, the density change gradient matrix is ​​obtained, the expansion direction of the lesion area is analyzed, and the expansion direction trend data is generated; Based on the tissue density distribution data, the density gradient change rate between adjacent pixels is calculated, the tissue density value of each pixel is extracted, the density change difference between it and the adjacent pixel is calculated, the density gradient change matrix is ​​formed, and the gradient change rate is calculated. The gradient change data is normalized using a standardization coefficient to keep the density data of different scales consistent.

[0032] The calculation formula is: ; Calculate the density gradient change rate and obtain the gradient change trend.

[0033] Assume that there are three consecutive pixels in a lesion area, and their density values ​​are 20, 25, and 30 respectively. Calculate the density gradient change rate between pixel 1 and pixel 2: ; Calculate the rate of change of density gradient between pixel 2 and pixel 3: ; The calculation results show that the density gradient change trend of the lesion area is slightly reduced between the front and rear pixels. The subsequent steps predict the lesion expansion trend based on this data.

[0034] S213: Based on the expansion direction trend data, density change rate calculation is performed on the boundary pixel points of the lesion area using the formula: ; Calculate the spatial expansion trend of the lesion area, combine the gradient change rate of the pixel points, establish the spatial expansion prediction map of the lesion area, and obtain the distribution position prediction result; in, is the spatial expansion trend of the lesion area, For the The density gradient value of each pixel, For the The density gradient value of adjacent pixels, For the The gray value of a pixel, For the The grayscale values ​​of adjacent pixels, is the total number of pixels at the boundary of the lesion area, where is the number of the current boundary pixel, For and number The next adjacent boundary pixel; Based on the expansion direction trend data, the density change rate of the boundary pixels of the lesion area is calculated, and combined with the gradient change trend of the pixels, a spatial expansion prediction map of the lesion area is established to predict the spatial distribution of the lesion area. The formula is: ; Calculate the spatial expansion trend of the lesion area , quantifying the diffusion rate of the lesion area.

[0035] The lesion area is assumed to contain 5 key pixel points, and the extracted pixel density gradient values ​​are as follows: Table 1 Pixel density gradient values ; Referring to Table 1, the spatial expansion trend of each pixel pair is calculated: ; ; ; ; ; ; ; The calculation results show that the spatial expansion trend of the lesion area The value is 0.0602, which is used for the calculation of subsequent lesion extension prediction.

[0036] See also Figure 4 ,The specific steps for obtaining activity rate classification data are: S311: calling the distribution position prediction result, analyzing the patient's medical images at multiple time points, obtaining the lesion area in each frame of the image, extracting the grayscale value of the lesion area, calculating the average grayscale of the lesion area in multiple image frames, and generating grayscale change data of the lesion area; After acquiring the dynamic image sequence data, first decompose the image sequence into frames, extract the image frames corresponding to all time points, and determine the pixel range of the lesion area. In each frame, locate the center point of the lesion area, select all pixels in the lesion area, record their grayscale values, and construct a grayscale value matrix of the lesion area, which stores the grayscale distribution of the lesion area at different time points. Next, traverse the image frames and perform pixel grayscale mean calculation on the lesion area of ​​each frame to reduce the impact of individual pixel abnormalities on the data and form a time series grayscale mean sequence. During the calculation process, the grayscale value of each pixel in the lesion area is summed up and divided by the total number of pixels in the lesion area, using the formula: ; The average grayscale value of the lesion area in the current frame is calculated and stored in the time series grayscale database to form a complete grayscale change information data set to obtain the grayscale change information of the lesion area.

[0037] in, For the lesion area In time The average gray value at For the lesion area In time Place The gray value of a pixel, is the total number of pixels in the lesion area.

[0038] set up , the pixel gray value of the lesion area is , , , , .calculate: ; The calculation results show that the average grayscale value of the lesion area in the current frame is 111.6.

[0039] S312: Based on the grayscale change data of the lesion area, the grayscale change amplitude between adjacent frames is calculated to obtain the grayscale change rate of the lesion area in the image sequence, using the formula: ; Obtaining the signal change rate of the lesion area by calculation, and establishing lesion signal change rate data; in, Representative The signal change rate of the lesion area is For the The lesion area is The average gray value of the frame, For the The average gray value of the frame, is the standard deviation of grayscale changes in the lesion area between all frames, is the time interval between adjacent frames, is the total number of frames in the image sequence, Indicates the serial number of the lesion area. Indicates the time index of the current frame, Indicates the next adjacent frame time index; Based on the grayscale change information of the lesion area, the grayscale value change amplitude of the lesion area at different time points is calculated, the grayscale difference between adjacent frames is calculated by frame-by-frame comparison method, and the grayscale change trend matrix of the lesion area is established. For each time frame of the lesion area, the grayscale value change rate of the frame relative to the previous frame is calculated, and normalized to reduce the signal amplitude difference between different lesion areas. Finally, the signal change rate of the lesion area in the entire image sequence is calculated using the formula: ; After the calculation is completed, the signal change rate data of the lesion area is stored to form complete time series signal change data, and the lesion signal change rate data is obtained.

[0040] in, For the lesion area The signal change rate, For the lesion area In time The average gray value at For time The average gray value at is the standard deviation of grayscale changes in the lesion area between all frames, is the time interval between adjacent frames, is the total number of image frames.

[0041] set up , , standard deviation The grayscale mean of the lesion area is , , , , , , , , , .calculate: ; ; ; The calculation results show that the signal change rate of the lesion area is 7.45.

[0042] S313: identifying the activity levels of multiple lesions in the patient's medical image according to the lesion signal change rate data, and generating activity rate classification data; First, the signal change rate values ​​of all lesion areas are extracted, and the mean of the signal change rate of these lesion areas is calculated to obtain the central tendency of the entire data set. When calculating the mean, the signal change rate values ​​of all lesions are accumulated and divided by the total number of lesion areas to obtain the overall level of lesion signal change. Secondly, a classification threshold for dividing the lesion activity level is set, which is used to determine whether the lesion area belongs to a low-activity lesion, a medium-activity lesion, or a high-activity lesion. During the classification process, the upper and lower limits of the activity level threshold are set so that the lesion area with a lesion signal change rate lower than the mean minus the threshold is classified as a low-activity lesion, the lesion area with a lesion signal change rate higher than the mean plus the threshold is classified as a high-activity lesion, and the lesion area between the two is classified as a medium-activity lesion. In the specific classification process, the signal change rate data of all lesion areas are traversed, and the signal change rate of each lesion area is compared with the calculated mean and classification threshold to determine its activity level. For lesion areas with a signal change rate less than the lower limit of classification, they are classified as low-activity lesions. These lesion areas show a relatively stable grayscale change trend in the image frames at multiple time points, which usually means that the lesions develop slowly or are in a stable state. For lesion areas with a signal change rate greater than the upper limit of classification, they are classified as high-activity lesions. These lesion areas show a significant grayscale change trend in the image sequence, which usually means that the lesion tissue has undergone a large change in a short period of time, and there is a situation of malignant growth or inflammation diffusion. Lesion areas between the two are classified as medium-activity lesions. The signal change rate of this type of lesion is in a relatively moderate range, and is in a stage of slow diffusion or slight change. After completing the classification of lesion activity levels, the classification results of all lesion areas are sorted and stored in the lesion activity level database, so that the activity status of different lesions can be quantitatively described. Finally, the lesion activity level classification is completed based on the signal change rate of the lesion area, and the lesion activity level classification data is output.

[0043] See also Figure 5 , the steps for obtaining the image data index are as follows: S411: extracting the type, location, activity level, and patient information of the lesion based on the activity rate classification data, forming key items of the lesion index, and constructing a hierarchical index rule base; Based on the activity rate classification data, the type, location, activity level and patient information of the lesion are first extracted to determine the key index items in the medical imaging data. For the lesion type, the tissue structure characteristics of the lesion are analyzed, and the boundary morphology, density distribution, signal change pattern and other parameters of the lesion area are extracted. According to the physical characteristics of the lesion and the medical classification standards, the lesion category number is assigned to form a lesion category index. For the lesion location, the two-dimensional plane coordinates or three-dimensional space coordinates of the lesion are extracted based on the spatial coordinate system of the medical image, and the position information is converted into standardized index parameters and stored in the index rule library. For the lesion activity level, the lesion signal change pattern in the image frame sequence is analyzed to extract the lesion signal at different time points. Lesion signal value, calculate the dynamic activity trend of the lesion, and set the lesion activity level index according to the signal fluctuation of the lesion. For patient information, call the patient's medical record data, extract the patient's age, gender, medical history, treatment record and other information, normalize the patient's characteristic parameters, and map them to the index database, so that the patient information and lesion data are interconnected to form a complete lesion index system, and finally integrate all index key items to establish a hierarchical index rule library to ensure that the lesion data has complete grading rules in subsequent image matching and classification indexing, so that the image data can be accurately classified according to different lesion types, locations and activity states, and provide a standardized basis for subsequent image data storage and retrieval.

[0044] S412: Based on the hierarchical index rule base, the characteristic parameters of the medical image data are analyzed, and the image data and the index rules are matched, using the formula: ; Calculate the image classification weight value, identify the categories of multiple medical images in the index library, and generate image classification matching results; in, is the image classification weight value, For the The matching degree between the image data and the index rules, For the The classification weight of image data, For the The reference value of the image data, For the The target adjustment value of the image data, is the total number of image data, is the index of the image data; Based on the hierarchical index rule base, the characteristic parameters of medical image data are analyzed, and the gray value, tissue density, signal change trend and other parameters of the lesion area in the image data are called. According to the hierarchical index rules, the image data is matched with the lesion index items in the rule base, and the key features of each image data are extracted. The matching degree between the image and the index rules is calculated using the formula: ; Calculate the image classification weight value, screen the image data based on the classification weight value, determine the index category of the image data, classify it into the corresponding category of the index rule library, and finally identify the categories of multiple medical images in the index library to generate image classification matching results.

[0045] in, is the image classification weight value, For the The matching degree between the image data and the index rules, For the The classification weight of image data, For the The reference value of the image data, For the The target adjustment value of the image data, is the total number of image data, The index number of the image data.

[0046] set up , , , , . Substitute into the calculation: ; ; ; ; ; The calculation results show that the image classification weight value is This value is used to measure the matching degree between the image data and the index library classification. If the classification weight value is greater than the set classification threshold, the image data belongs to the corresponding classification and is stored in the index rule library, finally forming the image classification matching result.

[0047] S413: Based on the image classification and matching results, extract the index parameters of the classified images, establish index association relationships, and store them in the cloud platform to generate image data indexes; Based on the image classification and matching results, first extract the index parameters of the classified images, parse the storage attributes of the image data, extract the classification matching scores of the images, set the image storage priority according to the scores, screen highly relevant image data, establish image index associations, match the index parameters with the image data, adjust the index hierarchy according to the image category, and store the image data with high matching in the cloud platform index database first. At the same time, generate a unique index code for the image data to ensure that the image data can be quickly retrieved. For low matching image data, adjust the image storage path according to the index optimization rules, set the image storage label, and associate the image index history. Finally, complete the cloud platform storage of the image data based on the index optimization value, ensure that the classification index of the image data can meet the hierarchical storage requirements, generate the image data index, make the subsequent image data query and call more efficient, and at the same time improve the organizational management capabilities of the medical image database and form a structured image data index system.

[0048] See also Figure 6 , the specific steps for obtaining image access management parameters are: S511: calling the image data index, parsing the identity information of the user access request, obtaining the user authority level, and generating user authority matching data; To call the image data index, first read the image data index library stored in the cloud platform, which contains the classification information, resolution, access permission settings and data storage path of all medical images. Then, the system receives the user's access request and parses the identity information field in the request data packet, including user ID, access source, device type, etc. The system compares the user database and extracts the permission level corresponding to the user. The permission level can be divided into three categories: basic access permission, diagnostic permission and advanced analysis permission. Each permission level is associated with the accessible image type, resolution and access range. After the parsing is completed, the user permission information is matched with the image data index, and the image list that meets the user's access level is extracted and stored in the temporary data cache area for subsequent access processing. In this process, the image data index matching degree is calculated using the formula: ; in, is the matching degree of image data index, For the The matching coefficient between image data and user permissions, For the The access priority of image data. The total number of image data within the current user's authority.

[0049] set up , , , , . Substitute into the calculation: ; ; ; ; The final calculated image data index matching degree is 3.45, which is used for further matching of user access rights to ensure the accuracy of image access control.

[0050] S512: Matching the image resolution loading parameters for the user according to the user authority matching data and the authority level of the user's current access request, and generating an image resolution permission value; Based on the user permission matching data, the system determines the image resolution range that the user can load under the current access environment. For different permission levels, the medical images stored in the image data index library usually contain multiple resolution versions, such as 512×512, 1024×1024 and 2048×2048. The system matches the corresponding image resolution according to the user's permission level. For example, users with basic permissions can access images with a resolution of 512×512, while users with advanced permissions can access images with a resolution of 2048×2048. During the matching process, the system further combines the type of terminal device accessed by the user. If the user is using a mobile device, only medium-resolution images are allowed to be loaded to reduce bandwidth consumption. On the contrary, if the user is using a high-performance computing terminal, the highest-resolution image can be loaded. The system uses the calculation formula: ; in, is the image resolution allowed value, is the basic image resolution corresponding to the user's permissions, is the user priority of the current access request, The total number of users accessing the image queue.

[0051] set up , , Substitute into the calculation: ; ; The calculated image resolution permission value is 1331, which is used to determine the maximum loadable image resolution of the user in the current access session.

[0052] S513: Based on the image resolution permission value, real-time collection of user's real-time operation information and current network status parameters, using the formula: ; Obtain the resolution adjustment amount of the image loading area by calculation, dynamically adjust the resolution of the image loading area, and generate image access management parameters; in, The amount to adjust the resolution of the image loading area. is the image resolution permission value corresponding to the user's permission. For user The adjustment ratio of real-time operation, The effect of operation type on image resolution. is the current network bandwidth value, is the network bandwidth stability adjustment coefficient, The total amount of real-time operation data for users. It is the index number of the user's real-time operation data; When users browse images, the system collects user interactions in real time, including zooming, moving, rotating, and other behaviors, and monitors current network status parameters, such as bandwidth utilization, packet loss rate, and latency. The resolution of the image loading area is adjusted based on these data using the calculation formula: ; set up , , , , , , . Substitute into the calculation: ; ; ; The calculated adjustment resolution of the image loading area is 1380. The system dynamically adjusts the resolution of the image loading area based on the result to ensure the clarity and smoothness of the image display during user interaction, and finally generates the image access management parameters.

[0053] A cloud-based medical imaging diagnosis system, which is used to perform the cloud-based medical imaging diagnosis method, includes: The image boundary detection module is based on the input medical image, analyzes the texture density distribution of the image, performs grayscale gradient calculation, obtains the gradient change direction and rate, extracts the boundary information of the lesion area, and obtains the lesion boundary data; The spatial distribution prediction module extracts the tissue density value of the lesion area based on the lesion boundary data, calculates the density change rate of adjacent pixels, obtains tissue density distribution data, analyzes the trend of the lesion expansion direction, calculates the expansion probability of the lesion boundary point, constructs the spatial distribution prediction result of the lesion area, and obtains the distribution position prediction result; The lesion dynamic analysis module calls the distribution position prediction results, extracts the grayscale change data of the lesion area in the dynamic image sequence for the patient's medical images at multiple time points, calculates the intensity change rate of the lesion signal between adjacent frames, identifies the activity level of the lesion, and generates activity rate classification data; The classification storage management module establishes a classification index rule base based on the activity rate classification data, combined with the type, location, activity level and basic information of the lesion, and classifies and stores the images by matching the image content and index rules to generate an image data index; The remote access management module calls the image data index, parses the access request information of the cloud platform, matches the image loading resolution parameters according to the access user's identity authority, dynamically adjusts the resolution of the image loading area based on the user's interaction behavior and network status, and generates image access management parameters.

[0054] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A medical imaging diagnosis method based on a cloud platform, characterized in that: The following steps are involved: S1: Based on the input medical image, analyze the texture density of multiple positions in the image, calculate the gray gradient change direction and rate of the pixel points, extract the boundary points of the lesion area in the image, segment the lesion image, and obtain the lesion boundary data; S2: Based on the lesion boundary data, the tissue density value of the lesion area is obtained, the density change rate between adjacent pixels is calculated, and the tissue density distribution data is obtained. By calculating the density gradient change trend in the lesion area, the expansion direction of the lesion area is analyzed, and the spatial distribution prediction result of the lesion area is constructed to establish the distribution position prediction result; S3: calling the distribution position prediction result, analyzing the patient's medical images at multiple time points, extracting the grayscale change information of the lesion area in the dynamic image sequence, calculating the signal change rate of multiple lesion areas in the image frame, identifying the activity levels of multiple lesions, and outputting the activity rate classification data; S4: Based on the activity rate classification data, a hierarchical index rule library is established according to the type, location, activity level, and patient information of the lesion. By matching the image content and index rules, the medical images are classified and stored in the cloud platform to generate an image data index.

2. The cloud platform-based medical imaging diagnosis method according to claim 1, characterized in that: The lesion boundary data includes the coordinates of the lesion contour points, the boundary grayscale gradient direction, and the boundary pixel change rate. The distribution position prediction result is specifically the lesion expansion direction, the lesion spatial distribution area, and the lesion boundary change trend. The activity rate grading data includes the lesion grayscale change rate, the lesion signal change level, and the lesion activity classification result. The image data index specifically refers to the image classification label, image storage path, and image type identification.

3. The cloud platform-based medical imaging diagnosis method according to claim 1, characterized in that: The steps for obtaining the lesion boundary data are specifically as follows: S111: based on the input medical image, obtaining the pixel grayscale value of the medical image, calculating the grayscale change rate of each pixel point within the neighborhood range, obtaining the grayscale gradient value, and calculating the grayscale change amplitude of the pixel point in each direction to generate texture density data; S112: Based on the texture density data, calculate the density gradient of the pixel points in multiple directions, obtain the gradient change rate and calculate the gradient ratio between pixels, calibrate the boundary candidate points, and establish the boundary point extraction result; S113: Based on the boundary point extraction result, the gradient change consistency of adjacent boundary points is calculated using the formula: ； Calculate the adjusted position of the boundary point, optimize the boundary shape, segment the lesion image, and generate lesion boundary data; in, Represents the optimized The coordinates of the boundary points, For the original The coordinates of the boundary points, For the The gray value of the boundary point, For the The grayscale values ​​of adjacent boundary points are For the The gradient change rate of the boundary point is For the The gradient change rate of adjacent boundary points is is the weight coefficient of the boundary point, is the smoothing adjustment coefficient, Indicates the current boundary point number, Indicates the next adjacent boundary point of the current boundary point.

4. The cloud platform-based medical imaging diagnosis method according to claim 1, characterized in that: The steps for obtaining the distribution position prediction result are specifically as follows: S211: extracting gray values ​​of pixels in the lesion area based on the lesion boundary data, constructing a gray value matrix, evaluating changes in tissue density in the lesion area, and generating tissue density distribution data; S212: Based on the tissue density distribution data, calculate the density gradient change rate between adjacent pixel points, calculate the gray value difference between adjacent pixel points, obtain the density change gradient matrix, analyze the expansion direction of the lesion area, and generate expansion direction trend data; S213: Based on the expansion direction trend data, density change rate calculation is performed on the boundary pixel points of the lesion area, using the formula: ； Calculate the spatial expansion trend of the lesion area, combine the gradient change rate of the pixel points, establish the spatial expansion prediction map of the lesion area, and obtain the distribution position prediction result; in, is the spatial expansion trend of the lesion area, For the The density gradient value of each pixel point, For the The density gradient value of adjacent pixels, For the The gray value of a pixel, For the The grayscale values ​​of adjacent pixels, is the total number of pixels at the boundary of the lesion area, where is the number of the current boundary pixel, For and number The next adjacent boundary pixel.

5. The cloud platform-based medical imaging diagnosis method according to claim 1, characterized in that: The steps for obtaining the activity rate classification data are specifically as follows: S311: calling the distribution position prediction result, analyzing the patient's medical images at multiple time points, obtaining the lesion area in each frame of the image, extracting the grayscale value of the lesion area, calculating the average grayscale of the lesion area in multiple image frames, and generating grayscale change data of the lesion area; S312: Based on the grayscale change data of the lesion area, the grayscale change amplitude between adjacent frames is calculated to obtain the grayscale change rate of the lesion area in the image sequence, using the formula: ； Obtaining the signal change rate of the lesion area by calculation, and establishing lesion signal change rate data; in, Representative The signal change rate of the lesion area is For the The lesion area is The average gray value of the frame, For the The average gray value of the frame, is the standard deviation of grayscale changes in the lesion area between all frames, is the time interval between adjacent frames, is the total number of frames in the image sequence, Indicates the serial number of the lesion area. Indicates the time index of the current frame, Indicates the next adjacent frame time index; S313: Identify the activity levels of multiple lesions in the patient's medical image based on the lesion signal change rate data, and generate activity rate grading data.

6. The cloud platform-based medical imaging diagnosis method according to claim 1, characterized in that: The steps of obtaining the image data index are specifically as follows: S411: extracting the type, location, activity level, and patient information of the lesion based on the activity rate classification data, forming key items of the lesion index, and constructing a hierarchical index rule base; S412: Based on the hierarchical index rule base, the characteristic parameters of the medical image data are analyzed, and the image data and the index rules are matched, using the formula: ； Calculate the image classification weight value, identify the categories of multiple medical images in the index library, and generate image classification matching results; in, is the image classification weight value, For the The matching degree between the image data and the index rules, For the The classification weight of image data, For the The reference value of the image data, For the The target adjustment value of the image data, is the total number of image data, is the index of the image data; S413: Based on the image classification and matching results, extract the index parameters of the classified images, establish index association relationships, and store them in the cloud platform to generate image data indexes.

7. The cloud platform-based medical imaging diagnosis method according to claim 1, characterized in that: The method further comprises: S5: calling the image data index, analyzing the access request information of the cloud platform, extracting user permissions according to the access identity, matching the accessible image resolution, dynamically adjusting the resolution of the image loading area in combination with the user's real-time operation and network status, and generating image access management parameters; The image access management parameters include user authority level, image resolution matching rules, and dynamic loading area range.

8. The cloud platform-based medical imaging diagnosis method according to claim 7, characterized in that: The steps for obtaining the image access management parameters are specifically as follows: S511: calling the image data index, parsing the identity information of the user access request, obtaining the user authority level, and generating user authority matching data; S512: According to the user authority matching data and the authority level of the user's current access request, matching the image resolution loading parameters for the user to generate an image resolution permission value; S513: Based on the image resolution permission value, real-time user operation information and current network status parameters are collected in real time, using the formula: ； Obtain the resolution adjustment amount of the image loading area by calculation, dynamically adjust the resolution of the image loading area, and generate image access management parameters; in, The amount to adjust the resolution of the image loading area. is the image resolution permission value corresponding to the user's permission. For user The adjustment ratio of real-time operation, The effect of operation type on image resolution. is the current network bandwidth value, is the network bandwidth stability adjustment coefficient, The total amount of real-time operation data for users. It is the index number of the user's real-time operation data.

9. A cloud-based medical imaging diagnostic system, characterized in that: According to any one of claims 1 to 8, the cloud platform-based medical imaging diagnosis method, the system comprising: The image boundary detection module is based on the input medical image, analyzes the texture density distribution of the image, performs grayscale gradient calculation, obtains the gradient change direction and rate, extracts the boundary information of the lesion area, and obtains the lesion boundary data; The spatial distribution prediction module extracts the tissue density value of the lesion area based on the lesion boundary data, calculates the density change rate of adjacent pixels, obtains tissue density distribution data, analyzes the trend of the lesion expansion direction, calculates the expansion probability of the lesion boundary point, constructs the spatial distribution prediction result of the lesion area, and obtains the distribution position prediction result; The lesion dynamic analysis module calls the distribution position prediction result, extracts the grayscale change data of the lesion area in the dynamic image sequence for the patient's medical images at multiple time points, calculates the intensity change rate of the lesion signal between adjacent frames, identifies the activity level of the lesion, and generates activity rate classification data; The classification storage management module establishes a classification index rule base based on the activity rate classification data, combined with the type, location, activity level, and basic information of the lesion, and classifies and stores the images by matching the image content and the index rules to generate an image data index; The remote access management module calls the image data index, parses the access request information of the cloud platform, matches the image loading resolution parameters according to the access user's identity authority, dynamically adjusts the resolution of the image loading area in combination with the user's interactive behavior and network status, and generates image access management parameters.

Images