Heart attack early warning system based on image recognition

Abstract

The invention provides a heart attack early warning system based on image recognition. The heart attack early warning system comprises an image acquisition and processing system, data transmission equipment, a central server and a user terminal, the image acquisition and processing system is used for acquiring a medical image, performing image processing and initial anomaly detection on the medical image at an edge end, acquiring an initial anomaly detection result, and performing initial anomaly detection early warning when the initial anomaly detection result meets an initial early warning requirement; the data transmission equipment is used for establishing data connection and performing data transmission between the image acquisition and processing system and the central server and between the central server and the user terminal; the central server is used for performing data processing and risk assessment on the initial anomaly detection result which does not meet the initial early warning requirement to obtain a risk assessment result; and the user terminal is used for receiving the risk assessment result sent by the central server in real time and visually displaying the risk assessment result.

Description

Technical Field

[0001] The invention proposes a heart attack warning system based on image recognition, belonging to the technical field of risk warning. Background Art

[0002] With the aging of the population and the change of lifestyle, heart disease has become one of the major diseases threatening human health worldwide. Early warning and timely intervention of heart disease are crucial to reducing mortality and improving the quality of life of patients. Traditional methods for diagnosing heart disease mainly rely on electrocardiogram (ECG), blood tests and clinical symptom analysis, but the above methods often have lags and are difficult to provide timely warnings before acute heart attacks. In recent years, with the rapid development of computer vision and artificial intelligence technology, image recognition technology has been increasingly used in the medical field, providing new ideas for early warning of heart disease. Medical images, such as echocardiogram, CT (computed tomography) and MRI (magnetic resonance imaging), can intuitively display the structure and functional state of the heart, providing doctors with rich diagnostic information. However, traditional manual interpretation of medical images is not only time-consuming and labor-intensive, but also easily affected by factors such as doctor experience and fatigue, and there is a risk of misdiagnosis and missed diagnosis. Therefore, it is particularly important to develop a heart attack warning system based on image recognition technology. The system can automatically, quickly and accurately analyze medical images, identify potential heart abnormalities, and issue warning signals when necessary, so that patients can seek medical treatment in time or take corresponding measures.

[0003] At present, there are some medical auxiliary diagnosis systems based on image recognition on the market, but most of them focus on auxiliary diagnosis of diseases rather than early warning. The above systems often need to run on high-performance computers or cloud servers, and lack support for application scenarios with high requirements for real-time and portability. In addition, the existing systems rely on complex algorithm models in data processing and risk assessment, which are computationally intensive and time-consuming, and cannot meet the urgent needs of heart attack early warning.

[0004] In response to the above problems, the present invention proposes a heart attack warning system based on image recognition. The system integrates image acquisition and processing systems, data transmission equipment, central servers and user terminals to realize the full chain of automated processes from medical image acquisition, preliminary processing and warning at the edge, data transmission, in-depth analysis and risk assessment at the central server to real-time reception and visual display at the user terminal. The system not only improves the accuracy and timeliness of heart disease warnings, but also reduces the burden on the central server through edge computing technology, and enhances the real-time and portability of the system. Summary of the invention

[0005] The present invention provides a heart attack warning system based on image recognition to solve the technical problems in the above-mentioned prior art. The technical problems adopted are as follows: A heart attack warning system based on image recognition, comprising an image acquisition and processing system, a data transmission device, a central server and a user terminal; wherein the image acquisition and processing system is data-connected to the central server via the data transmission device; and the central server is data-connected to the user terminal via the data transmission device; The image acquisition and processing system is used to acquire medical images, perform image processing and initial anomaly detection on the medical images at the edge, obtain initial anomaly detection results, and perform initial anomaly detection warning when the initial anomaly detection results meet the initial warning requirements; The data transmission device is used to establish a data connection and perform data transmission between the image acquisition and processing system and the central server, and between the central server and the user terminal; The central server is used to perform data processing and risk assessment on the initial anomaly detection results that do not meet the initial warning requirements, and obtain risk assessment results; The user terminal is used to receive the risk assessment results sent by the central server in real time and perform visual display corresponding to the risk assessment results.

[0006] Furthermore, the image acquisition and processing system includes multiple medical image acquisition devices, an image standard processing module, a first edge computing device and a second edge computing device; wherein the data signal output end of the medical image acquisition device is connected to the signal input end of the image standard processing module; the signal output end of the image standard processing module is connected to the data signal input end of the first edge computing device; the signal output end of the first edge computing device is connected to the signal input end of the second edge computing device; the signal output end of the second edge computing device is the signal output end of the image acquisition and processing system.

[0007] Further, the medical image acquisition device is used to acquire cardiac medical images, obtain medical images for heart attack warning, and input the medical images for heart attack warning into the image standard processing module; The image standard processing module is used to perform standardization processing on medical images acquired by multiple medical image acquisition devices, acquire the standardized medical images, and send the standardized medical images to an edge computing device; The first edge computing device is configured to monitor the operating status of the second edge computing device in real time, and determine whether to send the standardized medical image to the second edge computing device according to the operating status of the second edge computing device, and, when the load of the second edge computing device exceeds a preset load requirement, activate its own initial anomaly detection function to perform initial anomaly detection on the standardized medical image for heart attack warning simultaneously with the second edge computing device; The second edge computing device is used to perform initial anomaly detection on the standardized medical images used for heart attack warning, and to perform initial anomaly detection warning when the initial anomaly detection results meet the initial warning requirements.

[0008] Furthermore, the image standardization processing process of the image standard processing module includes: Receiving original medical image data acquired by the plurality of medical image acquisition devices in real time, and performing noise reduction processing on the original medical image data to obtain original medical image data after noise reduction processing; Performing data image cropping and correction processing on the original medical image after the noise reduction processing according to preset image requirement information to obtain a medical image after cropping and correction processing; Performing initial standardized adjustment on the cropped and corrected medical image according to a preset resolution and size to obtain an initially standardized medical image; A secondary normalization adjustment is performed on the initially normalized medical image according to preset image contrast and brightness to obtain a normalized medical image.

[0009] Furthermore, the operation process of the first edge computing device includes: Monitoring the operating status parameters of the second edge computing device in real time, wherein the operating status parameters include CPU utilization, memory utilization, and processing queue length; Obtaining a load index of the current second edge computing device according to the CPU utilization, memory utilization, and processing queue length; When the load index of the second edge computing device does not exceed a preset load index threshold, sending the standardized medical image to the second edge computing device; When the load index of the second edge computing device exceeds a preset load index threshold, real-time monitoring is performed on a first duration during which the load index of the second edge computing device exceeds the preset load index threshold; When the load index of the second edge computing device exceeds the preset load index threshold for a first duration exceeding a preset time length, the first edge computing device automatically starts the initial abnormality detection function, and obtains the number of image data processed by the first edge computing device according to the number of standardized medical images to be detected contained in the processing queue length of the second edge computing device and the current load index of the second edge computing device; wherein the number of image data processed is obtained by the following formula: Where, N represents the number of image data processed, and N is rounded up; K represents the load index; K y represents a preset load index threshold; m represents the number of standardized processed medical images to be detected contained in the processing queue length; The first edge computing device retrieves, according to the number of processed image data, the same number of standardized medical images to be detected as the number of processed image data from the processing queue length of the second edge computing device as the target medical images, and performs initial anomaly detection on the target medical images to obtain initial anomaly detection results; When the load index of the second edge computing device drops below a preset load index threshold, and the duration of being below the preset load index threshold exceeds a preset second duration, the first edge computing device automatically turns off the initial anomaly detection function.

[0010] Furthermore, the load index is obtained as follows: Extracting an operating status parameter corresponding to each unit time of the second edge computing device, wherein the operating status parameter includes CPU utilization, memory utilization, and processing queue length; The parameter matrix A corresponding to each unit time is generated using the CPU utilization, memory utilization and processing queue length included in the operating state parameters corresponding to each unit time; wherein the structure of the parameter matrix A is as follows: Among them, A represents the parameter matrix corresponding to each unit time; a 11 、a 12 ……a n3 represents all elements contained in the parameter matrix A; n represents the number of unit times experienced by the corresponding second edge computing device under each unit time; Extract the parameter matrix A corresponding to each unit time that the second edge computing device has experienced in history and the parameter matrix A corresponding to the current unit time; Obtain the load index of the current second edge computing device by using the parameter matrix A corresponding to each unit time that the second edge computing device has experienced in history and the parameter matrix A corresponding to the current unit time; The load index is obtained by the following formula: Where K represents the load index; σ max Indicates the maximum singular number corresponding to the parameter matrix A corresponding to the current unit time; Indicates the norm of the parameter matrix A corresponding to the current unit time; A represents the parameter matrix A corresponding to the current unit time T The determinant of the matrix A; z and p Represents the median and mean of the determinant of the parameter matrix corresponding to all unit times that the second edge computing device has experienced in history; Var represents the variance of the determinant of the parameter matrix corresponding to all unit times that the second edge computing device has experienced in history; ε represents the minimum constant used to prevent the independent variable of the logarithmic function from being 0; x represents the base of the preset logarithmic function, and x≥2.

[0011] Furthermore, the process of the first edge computing device and the second edge computing device performing initial anomaly detection includes: Extracting regions from different parts of the heart in the standardized medical image for early warning of heart attack, and obtaining a plurality of heart region images corresponding to the different parts of the heart; Acquire characteristic data corresponding to different parts of each heart from the plurality of heart region images, wherein the characteristic data include the size of the ventricles and atria, the shape of the ventricles and atria, the thickness of the heart wall and the shape of the heart valve; Retrieving standard data corresponding to the size of the ventricle and atrium, the shape of the ventricle and atrium, the thickness of the heart wall and the shape of the heart valve; Comparing the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape contained in the feature data with the standard data corresponding to the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape, and obtaining a similarity value between the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape contained in the feature data and the standard data corresponding to the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape; The abnormal index parameter is obtained by using the similarity value between the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape contained in the characteristic data and the standard data corresponding to the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape, wherein the abnormal index parameter is obtained by the following formula: Where E represents the anomaly index parameter; r represents the number of feature data types contained in the feature data; S i represents the similarity value between the i-th feature data and its corresponding standard data; S min and S max They represent the minimum and maximum similarity values ​​among r types of feature data respectively; w min Indicates the weight value of the feature data corresponding to the minimum similarity value; w mx represents the minimum weight among r types of feature data; w max Indicates the weight value of the feature data corresponding to the maximum similarity value; w my Represents the maximum weight among r types of feature data; The abnormal index parameter is compared with the preset abnormal index threshold. When the abnormal index parameter is lower than the preset abnormal index threshold, it is determined that there is no abnormal situation; when the abnormal index parameter is not lower than the preset abnormal index threshold, it is determined that there is an abnormal situation and an initial abnormal detection warning is performed.

[0012] Further, the central server includes a data receiving module, a data processing module, a risk assessment module and an early warning module; wherein the signal input end of the data receiving module is the signal input end of the central server; the signal output end of the data receiving module is connected to the signal input end of the data processing module; the signal output end of the data processing module is connected to the signal input end of the risk assessment module; the signal output end of the risk assessment module is connected to the signal input end of the early warning module; The data receiving module is used to receive in real time the medical image for heart attack warning after standardized processing that does not meet the initial warning requirements when the medical image for heart attack warning after standardized processing does not meet the initial warning requirements; The data processing module is used to perform data preprocessing on the medical image for heart attack warning after standardization that does not meet the initial warning requirements to obtain a preprocessed medical image; wherein the data preprocessing includes noise reduction processing and data cleaning processing; The risk assessment module is used to perform a risk assessment of a heart attack based on the preprocessed medical image to obtain a risk assessment result; The early warning module is used to issue a heart attack early warning when the risk assessment result indicates that there is a risk of heart attack.

[0013] Furthermore, the process of the risk assessment module performing heart attack risk assessment based on the preprocessed medical image includes: In the case where the initial warning requirements are not met, the pre-processed medical images corresponding to the requirements not meeting the initial warning requirements are retrieved; Retrieving a standardized medical image for heart attack warning from the historical record of the patient corresponding to the pre-processed medical image that does not meet the initial warning requirements; sorting the medical images according to the acquisition time of the standardized medical images for heart attack warning in the patient's historical records to generate a medical image set; The risk of heart attack is assessed using the abnormal index parameters of every two adjacent medical images in the medical image set.

[0014] Further, using the abnormal index parameters of every two adjacent medical images in the medical image set to perform a risk assessment for a heart attack includes: Extracting abnormality index parameters of every two adjacent medical images in the medical image set; Compare the abnormal index parameters of each two adjacent medical images to obtain an abnormal index parameter difference, and extract from the abnormal index parameter difference an abnormal index parameter difference corresponding to a case where the abnormal index parameter of the medical image preceding the medical image is lower than the abnormal index parameter of the medical image following the medical image; Compare the abnormal index parameter difference corresponding to each abnormal index parameter of the previous medical image being lower than the abnormal index parameter of the subsequent medical image with a preset difference threshold; When the abnormal index parameter difference corresponding to the abnormal index parameter of the previous medical image is lower than the abnormal index parameter of the later medical image exceeds a preset difference threshold, the previous medical image and the later medical image corresponding to the abnormal index parameter difference exceeding the preset difference threshold are marked as a target abnormal image group; Retrieving similarity values ​​of the sizes of the ventricles and atria, the shapes of the ventricles and atria, the thickness of the heart wall and the shape of the heart valves of the medical images before and after the medical images included in each target abnormal image group; The risk determination value is obtained by using the similarity values ​​of the size of the ventricles and atria, the shape of the ventricles and atria, the thickness of the heart wall and the shape of the heart valves of the medical images before and after the medical images included in all target abnormal image groups; wherein the risk determination value is obtained by the following formula: Wherein, Y represents the risk determination value; x represents the base of the preset logarithmic function, and x≥2; β represents the positive offset, and the value range of the positive offset is 0.28-3.63; k represents the number of target abnormal image groups; t j represents the time difference between the later medical image in the jth target abnormal image group and the current time T; Q represents the similarity parameter corresponding to all target abnormal image groups, and the similarity parameter is obtained by the following formula: Where Q represents the similarity parameter corresponding to all target abnormal image groups; k represents the number of target abnormal image groups; r j represents the number of feature data types contained in the feature data of the jth target abnormal image group; S hi and S xi represents the similarity value between the i-th feature data and the standard data corresponding to the previous medical image and the later medical image of the j-th target abnormal image group; E hj and E xj Indicates the abnormal index parameters corresponding to the previous medical image and the subsequent medical image of the j-th target abnormal image group; When the risk determination value exceeds a preset risk determination threshold, it is determined that there is a risk of a heart attack, and a heart attack warning is issued.

[0015] Beneficial effects of the present invention: The present invention proposes a heart attack warning system based on image recognition. By integrating image recognition technology and risk assessment models, the system can more accurately identify the potential risk of heart attack and reduce the possibility of misdiagnosis and missed diagnosis. The initial processing and early warning at the edge and the efficient data transmission mechanism ensure that the system can respond to the patient's medical image data in real time and issue a warning signal at the first time. The real-time reception and visual display function of the user terminal enables patients and medical staff to understand the patient's risk status anytime and anywhere, enhancing the ease of use and convenience of the system. Through the automated early warning process, the system can help medical institutions allocate medical resources more effectively and provide patients with timely intervention and treatment services. The system's remote data transmission and visual display functions make telemedicine possible and provide patients with a wider range of medical service options. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The system block diagram of the system of the present invention. DETAILED DESCRIPTION

[0017] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0018] The embodiment of the present invention proposes a heart attack warning system based on image recognition, such as Figure 1 As shown, the heart attack warning system based on image recognition includes an image acquisition and processing system, a data transmission device, a central server and a user terminal; wherein the image acquisition and processing system is data-connected to the central server via the data transmission device; and the central server is data-connected to the user terminal via the data transmission device; The image acquisition and processing system is used to acquire medical images, perform image processing and initial anomaly detection on the medical images at the edge, obtain initial anomaly detection results, and perform initial anomaly detection warning when the initial anomaly detection results meet the initial warning requirements; The data transmission device is used to establish a data connection and perform data transmission between the image acquisition and processing system and the central server, and between the central server and the user terminal; The central server is used to perform data processing and risk assessment on the initial anomaly detection results that do not meet the initial warning requirements, and obtain risk assessment results; The user terminal is used to receive the risk assessment results sent by the central server in real time and perform visual display corresponding to the risk assessment results.

[0019] The working principle of the above technical solution is: the above-mentioned heart attack warning system based on image recognition in this embodiment, through a highly integrated process, realizes a fully automated warning process from real-time acquisition of patient medical images, preliminary processing and warning at the edge, rapid transmission of data, in-depth analysis and risk assessment of the central server, to real-time reception and visual display at the user terminal.

[0020] First, the image acquisition and processing system is responsible for capturing the patient's medical images in real time. The images are processed and feature extracted at the edge, and initial anomaly detection is performed. If the detected anomaly meets the preset initial warning requirements, the system will immediately trigger the initial warning signal to alert the patient or medical staff.

[0021] Subsequently, the above images and their preliminary processing results will be quickly transmitted to the central server through efficient data transmission equipment. After receiving the data, the central server uses advanced image recognition technology and risk assessment models to conduct in-depth processing and risk assessment of abnormal results that do not meet the initial warning requirements but require further analysis. This process comprehensively considers the patient's various medical information and historical data to obtain a more comprehensive and accurate risk assessment result.

[0022] Finally, the central server sends the risk assessment results to the user terminal in real time. The user terminal displays the above results through an intuitive graphical interface, allowing patients and medical staff to quickly understand the patient's heart attack risk and take necessary intervention and treatment measures.

[0023] The effect of the above technical solution is: by integrating image recognition technology and risk assessment models, the system can more accurately identify the potential risk of heart attack and reduce the possibility of misdiagnosis and missed diagnosis. The initial processing and early warning at the edge and the efficient data transmission mechanism ensure that the system can respond to the patient's medical image data in real time and issue early warning signals at the first time. The real-time reception and visual display functions of the user terminal enable patients and medical staff to understand the patient's risk status anytime and anywhere, enhancing the ease of use and convenience of the system. Through the automated early warning process, the system can help medical institutions allocate medical resources more effectively and provide patients with timely intervention and treatment services. The system's remote data transmission and visual display functions make telemedicine possible and provide patients with a wider range of medical service options.

[0024] In summary, this heart attack warning system based on image recognition provides strong support for early warning and timely intervention of heart disease through its unique working principle and technical advantages.

[0025] In one embodiment of the present invention, the image acquisition and processing system includes multiple medical image acquisition devices, an image standard processing module, a first edge computing device and a second edge computing device; wherein the data signal output end of the medical image acquisition device is connected to the signal input end of the image standard processing module; the signal output end of the image standard processing module is connected to the data signal input end of the first edge computing device; the signal output end of the first edge computing device is connected to the signal input end of the second edge computing device; the signal output end of the second edge computing device is the signal output end of the image acquisition and processing system.

[0026] Wherein, the medical image acquisition device is used to acquire cardiac medical images, obtain medical images for heart attack warning, and input the medical images for heart attack warning into the image standard processing module; The image standard processing module is used to perform standardization processing on medical images acquired by multiple medical image acquisition devices, acquire the standardized medical images, and send the standardized medical images to an edge computing device; The first edge computing device is configured to monitor the operating status of the second edge computing device in real time, and determine whether to send the standardized medical image to the second edge computing device according to the operating status of the second edge computing device, and, when the load of the second edge computing device exceeds a preset load requirement, activate its own initial anomaly detection function to perform initial anomaly detection on the standardized medical image for heart attack warning simultaneously with the second edge computing device; The second edge computing device is used to perform initial anomaly detection on the standardized medical images used for heart attack warning, and to perform initial anomaly detection warning when the initial anomaly detection results meet the initial warning requirements.

[0027] The working principle of the above technical solution is: use multiple medical image acquisition devices (such as ultrasound cardiographs, MRI scanners, etc.) to acquire cardiac medical images. The original medical images acquired by the above devices are input into the image standard processing module.

[0028] The image standard processing module performs unified standardization processing on medical images acquired by multiple medical image acquisition devices to improve the consistency and comparability of the images. The standardized medical images are sent to the first edge computing device for further processing.

[0029] The first edge computing device is responsible for real-time monitoring of the operating status of the second edge computing device, including its load. If the load of the second edge computing device exceeds the preset load requirement, the first edge computing device will start its own initial anomaly detection function and process the standardized medical images in parallel with the second edge computing device. The above mechanism ensures that the system can continue to effectively perform initial anomaly detection under high load conditions.

[0030] The second edge computing device (or the first edge computing device started at the same time) performs initial anomaly detection on the standardized medical image. If the detection results meet the preset initial warning requirements (such as detecting abnormal image features that may be related to a heart attack), the system will issue an initial abnormality detection warning to alert medical staff or patients.

[0031] The effects of the above technical solutions are: unified processing of medical images through standardized processing modules reduces the complexity and error rate of subsequent processing. The real-time monitoring and load management mechanism of edge computing devices ensures that the system can maintain stable operation under high load conditions. The initial anomaly detection function can quickly analyze medical images in real time and identify abnormal features that may be related to heart attacks. Parallel processing of multiple edge computing devices improves the accuracy and reliability of detection.

[0032] The real-time monitoring function of the first edge computing device can detect abnormal conditions of the second edge computing device in a timely manner and take corresponding measures to ensure the continuous operation of the system. The modular design of the system allows each part to work independently, improving the fault tolerance of the entire system. Through automated initial anomaly detection and early warning, the system can detect potential heart attack risks in a timely manner, providing strong support for the rational allocation of medical resources. The application of edge computing reduces data transmission delays and bandwidth costs, while improving the real-time and response speed of the system. The timeliness of initial anomaly detection and early warning enables patients to seek medical treatment in a timely manner and avoid higher medical costs caused by worsening conditions.

[0033] In one embodiment of the present invention, the image standardization processing process of the image standard processing module includes: A1. receiving original medical image data acquired by the plurality of medical image acquisition devices in real time, and performing noise reduction processing on the original medical image data to obtain original medical image data after noise reduction processing; A2. Performing data image cropping and correction processing on the original medical image after the noise reduction processing according to preset image requirement information to obtain a cropped and corrected medical image; A3, performing initial standardization adjustment on the cropped and corrected medical image according to a preset resolution and size to obtain an initially standardized medical image; A4. Perform secondary standardization adjustment on the initially standardized medical image according to preset image contrast and brightness to obtain a standardized medical image.

[0034] The working principle of the above technical solution is: receiving raw medical image data from multiple medical image acquisition devices in real time. Performing noise reduction processing on the raw medical image data to remove noise and interference signals that may exist in the image. Noise reduction processing can improve the clarity and signal-to-noise ratio of the image, providing better image quality for subsequent processing.

[0035] According to the preset image requirement information (such as the region of interest, image direction, etc.), the original medical image after noise reduction is cropped and corrected. Cropping can remove irrelevant parts of the image and retain key information; correction is used to adjust the geometric shape and position of the image to ensure the correctness and consistency of the image.

[0036] The cropped and corrected medical images are initially standardized according to the preset resolution and size. This step aims to unify the size and resolution of the images and provide standardized input data for subsequent processing. The initial standardization adjustment can ensure that medical images from different sources and types are comparable in subsequent processing.

[0037] The medical image after initial standardization is subjected to secondary standardization adjustment according to the preset image contrast and brightness. This step further adjusts the image contrast and brightness to improve the visual effect and readability of the image. Secondary standardization adjustment can ensure that the image can maintain good display effect under different display devices and environmental conditions.

[0038] The effect of the above technical solution is: through the steps of noise reduction processing, cropping correction and standardization adjustment, the noise and interference signals in the image are effectively removed, and the clarity and signal-to-noise ratio of the image are improved. At the same time, by adjusting the contrast and brightness of the image, the visual effect and readability of the image are improved, providing a better image quality basis for subsequent processing and analysis. Through a unified cropping, correction and standardization adjustment process, it is ensured that medical images of different sources and types are comparable in subsequent processing. It can reduce processing errors and deviations caused by image differences and improve the accuracy and reliability of the overall processing results. The standardized processing process realizes the automated processing and analysis of medical images, reducing the need for manual intervention and manual adjustment. It can improve processing efficiency and speed, and reduce processing costs and time costs. The medical images after standardized processing have better consistency and comparability, providing a more reliable data basis for subsequent processing and analysis. It can achieve more accurate clinical application needs such as disease diagnosis, condition assessment and treatment effect evaluation.

[0039] In one embodiment of the present invention, the operation process of the first edge computing device includes: B1. Monitor the operating status parameters of the second edge computing device in real time, wherein the operating status parameters include CPU utilization, memory utilization, and processing queue length; B2. Obtaining a load index of the current second edge computing device according to the CPU utilization, memory utilization, and processing queue length; B3. When the load index of the second edge computing device does not exceed a preset load index threshold, sending the standardized medical image to the second edge computing device; B4. When the load index of the second edge computing device exceeds a preset load index threshold, real-time monitoring is performed on a first duration during which the load index of the second edge computing device exceeds the preset load index threshold; B5. When the load index of the second edge computing device exceeds the preset load index threshold for a first duration exceeding the preset time length, the first edge computing device automatically starts the initial abnormality detection function, and obtains the number of image data processed by the first edge computing device according to the number of standardized medical images to be detected contained in the processing queue length of the second edge computing device and the current load index of the second edge computing device; wherein the number of image data processed is obtained by the following formula: Where, N represents the number of image data processed, and N is rounded up; K represents the load index; K y represents a preset load index threshold; m represents the number of standardized processed medical images to be detected contained in the processing queue length; B6. The first edge computing device retrieves, according to the number of image data to be processed, the same number of standardized medical images to be detected as the number of image data to be processed from the processing queue length of the second edge computing device as the target medical images, and performs initial anomaly detection on the target medical images to obtain initial anomaly detection results; B7. When the load index of the second edge computing device drops below a preset load index threshold, and the duration of being below the preset load index threshold exceeds a preset second duration, the first edge computing device automatically turns off the initial anomaly detection function.

[0040] The working principle of the above technical solution is as follows: the first edge computing device monitors the operating status parameters of the second edge computing device in real time, including CPU utilization, memory utilization and processing queue length. The current load index of the second edge computing device is calculated according to the above parameters to evaluate whether its processing capacity is saturated. If the load index of the second edge computing device does not exceed the preset load index threshold, the medical image after normalization is directly sent to the second edge computing device for further processing. If the load index exceeds the threshold and lasts for more than the preset time length, the first edge computing device starts the initial anomaly detection function and processes part of the medical images by itself to reduce the burden of the second edge computing device. The first edge computing device calculates and determines the number of medical images that it needs to process (i.e., the number of image data processed) according to parameters such as the load index and the processing queue length. The corresponding number of medical images are retrieved from the processing queue of the second edge computing device for initial anomaly detection to quickly identify possible abnormal situations. When the load of the second edge computing device drops below the preset threshold and lasts for more than the second preset time length, the first edge computing device turns off the initial anomaly detection function and resumes normal working state.

[0041] The effect of the above technical solution is: through real-time monitoring and load evaluation, the system can still run stably under high load conditions, avoiding system crashes or data loss caused by overload of a single device. The initial anomaly detection function can timely discover and handle potential problems and reduce the possibility of system failure. When the load of the second edge computing device is too high, the first edge computing device can automatically take on part of the processing work to achieve dynamic allocation and optimal utilization of resources. By calculating the number of image data processed, it is ensured that the first edge computing device will not have too much impact on its own performance when processing additional tasks. The initial anomaly detection function can quickly identify and process key medical images, improving the overall processing efficiency and response speed of the system. After the load of the second edge computing device is reduced, the first edge computing device can quickly turn off the anomaly detection function, resume normal working procedures, and reduce unnecessary resource consumption. The above technical solution can be flexibly configured and expanded according to actual needs to adapt to application scenarios of different scales and complexities. By adjusting the preset load index threshold, time length and other parameters, fine control and optimization of system performance can be achieved.

[0042] By real-time monitoring of the operating status parameters (CPU utilization, memory utilization, and processing queue length) of the second edge computing device, the first edge computing device can dynamically evaluate its load. When the load index does not exceed the preset threshold, the task (i.e., the medical image after standardization) is normally assigned to the second edge computing device for processing; when the load index exceeds the threshold, further measures are taken, such as starting the initial anomaly detection function, to reduce the burden on the second edge computing device. When the load of the second edge computing device is too high and the duration exceeds the preset value, the first edge computing device can automatically start the initial anomaly detection function and calculate the number of images to be processed based on the processing queue length of the second edge computing device and the current load index. The above method not only avoids the waste of resources, but also ensures the timely processing of key tasks (such as anomaly detection). According to the calculated number of image data to be processed, the first edge computing device can retrieve the corresponding number of images to be detected from the processing queue of the second edge computing device for initial anomaly detection. The above flexible task allocation method can optimize the overall processing flow and improve processing efficiency.

[0043] When the load index of the second edge computing device drops below the preset threshold and lasts for more than the preset second duration, the first edge computing device can automatically turn off the initial anomaly detection function. The above-mentioned automatic recovery mechanism can quickly restore the system to a normal state and avoid unnecessary resource consumption. Through the above-mentioned dynamic adjustment mechanism, the above-mentioned technical solution can ensure that the system can still operate stably under high load conditions, avoiding system crashes or performance degradation caused by excessive load. At the same time, through the initial anomaly detection function, potential abnormal problems can also be discovered and handled in a timely manner to improve the reliability of the system. For users using the system, dynamic load balancing and efficient resource utilization mean faster processing speeds and more stable system performance. It can further enhance user satisfaction and trust in the system.

[0044] In summary, the above technical solution effectively improves the stability, reliability, processing efficiency and response speed of the system through real-time monitoring, load assessment, diversion processing and anomaly detection mechanisms, while enhancing the scalability and flexibility of the system. At the same time, the above technical solution realizes efficient resource utilization and flexible task allocation by dynamically monitoring and adjusting the load of edge computing devices. This not only improves the stability and reliability of the system, but also optimizes the user experience. For scenarios that require processing of complex tasks such as a large number of medical images, the above technical solution has important practical value.

[0045] In one embodiment of the present invention, the process of obtaining the load index is as follows: Extracting an operating status parameter corresponding to each unit time of the second edge computing device, wherein the operating status parameter includes CPU utilization, memory utilization, and processing queue length; The parameter matrix A corresponding to each unit time is generated using the CPU utilization, memory utilization and processing queue length included in the operating state parameters corresponding to each unit time; wherein the structure of the parameter matrix A is as follows: Among them, A represents the parameter matrix corresponding to each unit time; a 11 、a 12 ……a n3 represents all elements contained in the parameter matrix A; n represents the number of unit times experienced by the corresponding second edge computing device under each unit time; Extract the parameter matrix A corresponding to each unit time that the second edge computing device has experienced in history and the parameter matrix A corresponding to the current unit time; Obtain the load index of the current second edge computing device by using the parameter matrix A corresponding to each unit time that the second edge computing device has experienced in history and the parameter matrix A corresponding to the current unit time; The load index is obtained by the following formula: Where K represents the load index; σ max Indicates the maximum singular number corresponding to the parameter matrix A corresponding to the current unit time; Indicates the norm of the parameter matrix A corresponding to the current unit time; A represents the parameter matrix A corresponding to the current unit time T The determinant of the matrix A; z and p Represents the median and mean of the determinant of the parameter matrix corresponding to all unit times that the second edge computing device has experienced in history; Var represents the variance of the determinant of the parameter matrix corresponding to all unit times that the second edge computing device has experienced in history; ε represents the minimum constant used to prevent the independent variable of the logarithmic function from being 0; x represents the base of the preset logarithmic function, and x≥2.

[0046] The working principle of the above technical solution is: extract the operating status parameters of the second edge computing device in each unit time in real time, including CPU utilization, memory utilization and processing queue length. The above parameters reflect the performance status of the device at different time points. Use the operating status parameters in each unit time to construct the corresponding parameter matrix A. Each row of the matrix A represents all parameter values ​​in one unit time. The structure of the matrix A allows subsequent mathematical operations to extract key information.

[0047] Extract and integrate the parameter matrix A corresponding to each unit time that the second edge computing device has experienced in history, as well as the parameter matrix A corresponding to the current unit time. This provides a comprehensive data basis for calculating the load index. Using the parameter matrix A corresponding to the current unit time and historical data, the load index K is calculated using the above formula. The formula involves the maximum singular number, norm, determinant of the ATA matrix of the parameter matrix A, and the median, mean, and variance of the determinant in the historical data. The minimum constant ε and the logarithmic function base x are introduced to ensure the stability and accuracy of the calculation.

[0048] The effect of the above technical solution is: by comprehensively considering the operating status parameters of multiple dimensions such as CPU utilization, memory utilization and processing queue length, as well as historical data, the current load status of the second edge computing device can be more accurately evaluated. The calculation method of the load index takes into account the diversity and dynamics of the data, and improves the accuracy and reliability of the evaluation. The above technical solution can adjust the calculation of the load index according to the operating status parameters of the device in real time, thereby realizing dynamic monitoring and evaluation of the device load. The system can respond to load changes in a timely manner and take corresponding measures to optimize resource allocation and avoid overload. By accurately evaluating the device load, the system can allocate tasks and resources more intelligently to avoid system crashes or performance degradation caused by overload of a single device. It can improve the stability and reliability of the entire system and ensure that the processing of key data such as medical images is not affected. At the same time, when the load index exceeds the preset threshold, the system can automatically start the abnormal detection or task diversion mechanism to transfer some tasks to other devices for processing. It can optimize resource utilization and improve the overall processing capacity and efficiency of the system. The above technical solution can be flexibly configured and expanded according to actual needs to adapt to application scenarios of different scales and complexities. By adjusting the parameters and thresholds in the formula, fine control and optimization of system performance can be achieved.

[0049] At the same time, the above technical solution can fully reflect the operating status of the device by extracting the operating status parameters (CPU utilization, memory utilization and processing queue length) corresponding to each unit time of the second edge computing device. The above parameters are key indicators for evaluating the load condition of the equipment and can achieve accurate load management. The parameter matrix A is generated using the operating status parameters corresponding to each unit time, which provides a data basis for the subsequent load index calculation. The structure of the parameter matrix A is clear and easy to process and calculate. At the same time, by extracting the parameter matrix A corresponding to each unit time that has been experienced in history and the parameter matrix A corresponding to the current unit time, continuous monitoring and dynamic evaluation of the operating status of the equipment are achieved.

[0050] At the same time, the load index K is calculated by the above formula, which comprehensively considers the maximum singular number, norm, determinant of the ATA matrix of the parameter matrix A corresponding to the current unit time, and the median, mean and variance of the determinant of the parameter matrix corresponding to all unit times that have been experienced in history. The above calculation method takes into account both the current operating status and the historical operating data, and realizes the accurate calculation of the load index. At the same time, the minimum constant ε is introduced into the formula to prevent the independent variable of the logarithmic function from being 0. This ensures the stability and reliability of the load index calculation and avoids erroneous results caused by abnormal calculations. The base x of the logarithmic function is preset, and x≥2, which provides flexible parameter settings for the calculation of the load index. The above setting method can be adjusted according to actual needs to adapt to different application scenarios and load management requirements.

[0051] By accurately calculating the load index, the solution can provide strong data support for resource allocation and load balancing. When the load index exceeds the preset threshold, corresponding measures (such as starting backup equipment, optimizing task allocation, etc.) can be taken to reduce the equipment load and improve the overall performance and stability of the system. The above technical solution is not only applicable to the current system environment, but also can adapt to future expansion needs by adjusting parameters and formulas. It can maintain the advancement and competitiveness of the system and provide strong support for future technological development.

[0052] In summary, the above technical solution achieves accurate evaluation and management of the load of the second edge computing device through comprehensive operation status monitoring, construction and utilization of parameter matrix, accurate calculation of load index and flexible parameter setting. This is of great significance for improving system performance, optimizing resource allocation and enhancing system reliability.

[0053] In one embodiment of the present invention, the process of performing initial anomaly detection by the first edge computing device and the second edge computing device includes: C1. Extracting regions from different parts of the heart in the standardized medical image for early warning of heart attack, and obtaining a plurality of heart region images corresponding to the different parts of the heart; C2. Acquire characteristic data corresponding to different parts of each heart from the multiple heart region images, wherein the characteristic data includes the size of the ventricles and atria, the shape of the ventricles and atria, the thickness of the heart wall, and the shape of the heart valve; C3, retrieve standard data corresponding to the size of the ventricle and atrium, the shape of the ventricle and atrium, the thickness of the heart wall and the shape of the heart valve; C4. Compare the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape contained in the feature data with the standard data corresponding to the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape, and obtain a similarity value between the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape contained in the feature data and the standard data corresponding to the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape; C5. Obtain an abnormality index parameter using the similarity value between the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape contained in the characteristic data and the standard data corresponding to the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape, wherein the abnormality index parameter is obtained by the following formula: Where E represents the anomaly index parameter; r represents the number of feature data types contained in the feature data; S i represents the similarity value between the i-th feature data and its corresponding standard data; S min and S max They represent the minimum and maximum similarity values ​​among r types of feature data respectively; w min Indicates the weight value of the feature data corresponding to the minimum similarity value; w mx represents the minimum weight among r types of feature data; w max Indicates the weight value of the feature data corresponding to the maximum similarity value; w my Represents the maximum weight among r types of feature data; C6. Compare the abnormal index parameter with the preset abnormal index threshold. When the abnormal index parameter is lower than the preset abnormal index threshold, it is determined that there is no abnormal situation; when the abnormal index parameter is not lower than the preset abnormal index threshold, it is determined that there is an abnormal situation, and an initial abnormal detection warning is performed.

[0054] The working principle of the above technical solution is: from the standardized heart attack warning medical images, accurate regional extraction is performed for different parts of the heart to generate multiple heart region images. This step is the basis for subsequent analysis and ensures the accurate capture of the characteristics of various parts of the heart. Key feature data is extracted from each heart region image, including the size and shape of the ventricles and atria, the thickness of the heart wall, and the shape of the heart valves. The above feature data is an important indicator for evaluating heart health.

[0055] Retrieve standard data corresponding to the above feature data. The above standard data is usually obtained based on the statistics of cardiac images of a large number of healthy people and represents the range of various indicators of a normal heart. Compare the extracted feature data with the standard data one by one and calculate the similarity value between them. This step aims to quantify the degree of deviation of various parts of the heart from the normal state.

[0056] The abnormality index parameter is calculated by a specific formula using the similarity value between the characteristic data and the standard data, combined with the weight of each characteristic data. This formula comprehensively considers the similarity and weight of multiple characteristic data to comprehensively evaluate the degree of abnormality of the heart.

[0057] The calculated abnormal index parameter is compared with the preset abnormal index threshold. If the abnormal index is lower than the threshold, it is determined that the current state of the heart is normal and there is no abnormality; if it is not lower than the threshold, it is determined that there is an abnormality and the initial abnormality detection warning mechanism is triggered.

[0058] The effect of the above technical solution is: through precise area extraction and feature data acquisition, combined with comparison and analysis with standard data, it is possible to more accurately assess the health of the heart and improve the accuracy of heart attack warning. The above technical solution takes into account the comprehensive evaluation of multiple parts of the heart and multiple feature data, which can more comprehensively reflect the overall state of the heart and reduce the possibility of missed diagnosis and misdiagnosis. Once an abnormality is detected, the system can immediately trigger the early warning mechanism to provide doctors with timely diagnostic basis and intervention suggestions, so as to detect and treat heart disease early. Through the automated initial abnormality detection process, the workload of doctors can be reduced, the utilization efficiency of medical resources can be improved, and more patients can obtain effective medical services in a timely manner. The application of the above technical solutions has promoted the development of medical image processing and edge computing technologies, and provided strong support for the intelligence and precision of the medical field.

[0059] On the other hand, by accurately extracting regional images of different parts of the heart from standardized medical images and obtaining the corresponding feature data of each part (such as the size and shape of the ventricles and atria, the thickness of the heart wall, the shape of the heart valves, etc.), a detailed data basis is provided for subsequent abnormality detection. The above-mentioned precise regional extraction and feature data acquisition methods can improve the accuracy and reliability of abnormality detection. Comparing the extracted feature data with the standard data and calculating the similarity value can intuitively reflect the degree of difference between various parts of the heart and the normal state. The above-mentioned comparison method can not only detect potential abnormalities, but also provide a basis for quantitative evaluation of the degree of abnormality.

[0060] The abnormal index parameter is calculated by comprehensively considering the similarity values ​​and weight values ​​of multiple feature data (size and shape of ventricles and atria, thickness of heart walls, shape of heart valves, etc.). The above comprehensive calculation method can more comprehensively and objectively reflect the overall abnormality of the heart, avoiding the one-sidedness and misleading nature of single feature data. Flexible abnormality judgment is performed based on the comparison results between the abnormal index parameter and the preset abnormal index threshold. When the abnormal index parameter is lower than the threshold, it is judged that there is no abnormality; when it is not lower than the threshold, it is judged that there is an abnormality, and an initial abnormality detection warning is performed. The above judgment mechanism can not only ensure the accuracy of abnormality detection, but also issue warnings in a timely manner, providing timely and effective support for subsequent abnormality processing.

[0061] The above technical solution can achieve accurate early warning of heart attacks. Once abnormalities are detected in various parts of the heart, early warning signals can be issued in time to remind medical staff and patients to take necessary intervention measures. The accuracy and timeliness of early warning of heart attacks can be improved, and the risks and hazards of heart attacks can be reduced. Through accurate abnormality detection and early warning, unnecessary waste of medical resources and excessive examinations can be avoided. At the same time, for patients with abnormal conditions, targeted treatment and management can be carried out in a timely manner, thereby reducing medical costs and improving medical efficiency.

[0062] In summary, the above technical solution achieves accurate early warning of heart attack through precise region extraction, feature data acquisition, standard data comparison, abnormal index parameter calculation and abnormality determination. It can effectively improve the accuracy and timeliness of early warning of heart attack, optimize the allocation of medical resources and reduce medical costs, and has important practical application value and social significance.

[0063] In one embodiment of the present invention, the central server includes a data receiving module, a data processing module, a risk assessment module and an early warning module; wherein the signal input end of the data receiving module is the signal input end of the central server; the signal output end of the data receiving module is connected to the signal input end of the data processing module; the signal output end of the data processing module is connected to the signal input end of the risk assessment module; the signal output end of the risk assessment module is connected to the signal input end of the early warning module; The data receiving module is used to receive in real time the medical image for heart attack warning after standardized processing that does not meet the initial warning requirements when the medical image for heart attack warning after standardized processing does not meet the initial warning requirements; The data processing module is used to perform data preprocessing on the medical image for heart attack warning after standardization that does not meet the initial warning requirements to obtain a preprocessed medical image; wherein the data preprocessing includes noise reduction processing and data cleaning processing; The risk assessment module is used to perform a risk assessment of a heart attack based on the preprocessed medical image to obtain a risk assessment result; The early warning module is used to issue a heart attack early warning when the risk assessment result indicates that there is a risk of heart attack.

[0064] The working principle of the above technical solution is as follows: when the medical image used for heart attack warning after standardized processing does not meet the initial warning requirements, the image will be sent to the data receiving module of the central server in real time. The data receiving module, as the signal input end of the central server, is responsible for receiving the image data that does not meet the initial conditions. The received image data is then transmitted to the data processing module for preprocessing. The preprocessing process includes noise reduction processing and data cleaning processing, which aims to improve the image quality, remove noise and irrelevant information, so that the subsequent risk assessment module can perform analysis more accurately. The preprocessed medical image is sent to the risk assessment module. The module uses advanced algorithms and models to conduct in-depth analysis of the image to assess the patient's risk of heart attack. This process may involve image recognition, feature extraction, machine learning and other technologies to identify abnormal image features related to heart attack. Based on the output results of the risk assessment module, the warning module will determine whether the patient is at risk of heart attack. If the assessment results indicate that there is a risk, the warning module will immediately trigger the warning mechanism and send an alarm to the doctor or relevant medical personnel so that they can take timely intervention measures.

[0065] The effect of the above technical solution is: by preprocessing and in-depth analysis of medical images that do not meet the initial warning requirements, the above technical solution can more accurately assess the patient's risk of heart attack, reduce false alarms and missed alarms, and improve the accuracy of the warning. The system can receive and process medical image data in real time to ensure that abnormal conditions are discovered in time before the patient has a heart attack, and buy valuable treatment time for doctors. The entire warning process is highly automated, reducing the need for manual intervention and improving work efficiency and accuracy. Doctors can focus more on the treatment and care of patients without spending a lot of time manually analyzing image data. The warning results can provide doctors with important clinical decision-making basis, helping them better assess the patient's condition and develop personalized treatment plans. The implementation of the above technical solution has promoted the application and development of medical image processing and artificial intelligence technology in the field of heart disease warning, and provided strong support for the intelligence and precision of the medical field.

[0066] In one embodiment of the present invention, the process of the risk assessment module performing heart attack risk assessment based on the preprocessed medical image includes: S1. In the case where the initial warning requirements are not met, retrieve the pre-processed medical image corresponding to the non-compliance with the initial warning requirements; S2, retrieving a standardized medical image for heart attack warning from the historical records of the patient corresponding to the pre-processed medical image that does not meet the initial warning requirements; S3, sorting the medical images according to the acquisition time of the standardized medical images for heart attack warning in the historical records of the patient to generate a medical image set; S4. Using the abnormal index parameters of every two adjacent medical images in the medical image set to perform heart attack risk assessment.

[0067] Wherein, using the abnormal index parameters of each two adjacent medical images in the medical image set to perform heart attack risk assessment includes: S401, extracting abnormality index parameters of every two adjacent medical images in the medical image set; S402, comparing the abnormality index parameters of each two adjacent medical images to obtain an abnormality index parameter difference, and extracting from the abnormality index parameter difference an abnormality index parameter difference corresponding to a case where the abnormality index parameter of the previous medical image is lower than the abnormality index parameter of the subsequent medical image; S403, comparing the abnormal index parameter difference corresponding to each abnormal index parameter of the previous medical image being lower than the abnormal index parameter of the subsequent medical image with a preset difference threshold; S404: when the abnormal index parameter difference corresponding to the abnormal index parameter of the earlier medical image is lower than the abnormal index parameter of the later medical image and exceeds a preset difference threshold, the earlier medical image and the later medical image corresponding to the abnormal index parameter difference exceeding the preset difference threshold are marked as a target abnormal image group; S405, retrieving similarity values ​​of the sizes of the ventricles and atria, the shapes of the ventricles and atria, the thickness of the heart wall, and the shape of the heart valves of the medical images before and after the medical images included in each target abnormal image group; S406, obtaining a risk determination value using the similarity values ​​of the sizes of the ventricles and atria, the shapes of the ventricles and atria, the thickness of the heart wall, and the shapes of the heart valves of the medical images before and after the medical images in all target abnormal image groups; wherein the risk determination value is obtained by the following formula: Wherein, Y represents the risk determination value; x represents the base of the preset logarithmic function, and x≥2; β represents the positive offset, and the value range of the positive offset is 0.28-3.63; k represents the number of target abnormal image groups; t j represents the time difference between the later medical image in the jth target abnormal image group and the current time T; Q represents the similarity parameter corresponding to all target abnormal image groups, and the similarity parameter is obtained by the following formula: Where Q represents the similarity parameter corresponding to all target abnormal image groups; k represents the number of target abnormal image groups; r j represents the number of feature data types contained in the feature data of the jth target abnormal image group; S hi and S xi represents the similarity value between the i-th feature data and the standard data corresponding to the previous medical image and the later medical image of the j-th target abnormal image group; E hj and E xj Indicates the abnormal index parameters corresponding to the previous medical image and the subsequent medical image of the j-th target abnormal image group; S407: When the risk determination value exceeds a preset risk determination threshold, it is determined that there is a risk of a heart attack, and a heart attack warning is issued.

[0068] The working principle of the above technical solution is: when the initial warning requirements are not met, the corresponding pre-processed medical images are first retrieved. Then, all the standardized medical images for heart attack warning in the patient's historical records are retrieved and sorted according to the acquisition time to generate an ordered medical image set.

[0069] Extract the abnormal index parameters of every two adjacent medical images from the medical image set. Compare the abnormal index parameters of the adjacent images to find out the situation where the abnormal index parameter of the medical image in the previous time is lower than that of the medical image in the next time, and calculate the difference between them. Compare the difference with the preset difference threshold to identify the target abnormal image group with significantly increased abnormal index.

[0070] For each target abnormal image group, retrieve the similarity values ​​of the size, shape, heart wall thickness and heart valve shape of the ventricles and atria of the medical images before and after the medical images contained therein. Using the above similarity values, combined with factors such as time difference and number of feature data types, a risk determination value is calculated by a specific formula. The formula takes into account the combined effects of logarithmic function, positive offset, time difference and similarity parameters. Compare the calculated risk determination value with the preset risk determination threshold to determine whether there is a risk of heart attack. If the risk determination value exceeds the preset threshold, the system determines that there is a risk of heart attack and triggers a heart attack warning mechanism.

[0071] The effect of the above technical solution is: by comprehensively considering the changes in abnormal index parameters of the patient's historical medical images and the changes in the similarity of the characteristics of various parts of the heart, the above technical solution can more accurately assess the risk of heart attack. By real-time monitoring and analysis of medical image data, the above technical solution can detect abnormal conditions in time before a heart attack, provide doctors with early warnings, and thus gain more treatment time. The above technical solution performs risk assessment based on the patient's historical medical image data, which can reflect the patient's individual characteristics and changes in the condition and achieve personalized assessment. The entire risk assessment process is highly automated, reducing the need for manual intervention and improving work efficiency. At the same time, by processing a large amount of medical image data through computer algorithms, the assessment results can be obtained more quickly. The risk assessment results can provide doctors with important clinical decision-making basis, helping them better assess the patient's condition and formulate personalized treatment plans.

[0072] On the other hand, by extracting the abnormal index parameters of each two adjacent medical images in the medical image set and comparing their differences, the changing trend of abnormal heart conditions can be finely monitored. This method helps to capture subtle changes in heart conditions and provides the possibility for timely intervention measures. Focusing on the situation where the abnormal index parameters of the medical image in the previous time are lower than the abnormal index parameters of the medical image in the later time, that is, the trend of worsening abnormal conditions, the target abnormal image group is identified. This method can screen out medical images that really need attention and reduce unnecessary analysis work. At the same time, the size, shape, heart wall thickness, and similarity values ​​of the heart valve shape of the medical images contained in the target abnormal image group, as well as the time factor, are used to calculate the risk determination value through a complex mathematical formula. This method takes into account multiple factors, including the changing trend of abnormal heart conditions, the degree of abnormality, and the urgency of time, and can more accurately assess the risk of heart attack. The parameters such as the base x of the logarithmic function and the positive offset β in the above formula can be adjusted according to actual conditions to adapt to different application scenarios and risk assessment requirements. This flexibility enables the above technical solution to be more widely used in various heart attack risk assessment scenarios.

[0073] At the same time, when the risk determination value exceeds the preset risk determination threshold, the system can automatically determine that there is a risk of heart attack and issue an early warning. This timely early warning mechanism helps medical staff and patients take necessary intervention measures in a timely manner to reduce the risk and harm of heart attack. Through fine abnormal change monitoring and targeted abnormal image group identification, combined with a comprehensive risk assessment method, the above technical solution can significantly improve the accuracy and efficiency of heart attack diagnosis. This helps to reduce misdiagnosis and missed diagnosis, and improve medical quality and patient satisfaction. Through accurate risk assessment, the above technical solution can screen out patients and medical images that really need attention, thereby optimizing the allocation of medical resources. This helps to avoid unnecessary waste of medical resources and improve the efficiency and quality of medical services.

[0074] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if the above-mentioned modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include the above-mentioned modifications and variations.

Claims

1. A heart attack warning system based on image recognition, characterized in that: The heart attack warning system based on image recognition includes an image acquisition and processing system, a data transmission device, a central server and a user terminal; wherein the image acquisition and processing system is connected to the central server through the data transmission device; and the central server is connected to the user terminal through the data transmission device. The image acquisition and processing system is used to acquire medical images, perform image processing and initial anomaly detection on the medical images at the edge, obtain initial anomaly detection results, and perform initial anomaly detection warning when the initial anomaly detection results meet the initial warning requirements; The data transmission device is used to establish a data connection and perform data transmission between the image acquisition and processing system and the central server, and between the central server and the user terminal; The central server is used to perform data processing and risk assessment on the initial anomaly detection results that do not meet the initial warning requirements, and obtain risk assessment results; The user terminal is used to receive the risk assessment results sent by the central server in real time and perform visual display corresponding to the risk assessment results.

2. The heart attack warning system based on image recognition according to claim 1, characterized in that: The image acquisition and processing system includes multiple medical image acquisition devices, an image standard processing module, a first edge computing device and a second edge computing device; wherein the data signal output end of the medical image acquisition device is connected to the signal input end of the image standard processing module; the signal output end of the image standard processing module is connected to the data signal input end of the first edge computing device; the signal output end of the first edge computing device is connected to the signal input end of the second edge computing device; the signal output end of the second edge computing device is the signal output end of the image acquisition and processing system.

3. The heart attack warning system based on image recognition according to claim 2, characterized in that: The medical image acquisition device is used to acquire cardiac medical images, obtain medical images for heart attack warning, and input the medical images for heart attack warning into the image standard processing module; The image standard processing module is used to perform standardization processing on medical images acquired by multiple medical image acquisition devices, acquire the standardized medical images, and send the standardized medical images to an edge computing device; The first edge computing device is configured to monitor the operating status of the second edge computing device in real time, and determine whether to send the standardized medical image to the second edge computing device according to the operating status of the second edge computing device, and, when the load of the second edge computing device exceeds a preset load requirement, activate its own initial anomaly detection function to perform initial anomaly detection on the standardized medical image for heart attack warning simultaneously with the second edge computing device; The second edge computing device is used to perform initial anomaly detection on the standardized medical images used for heart attack warning, and to perform initial anomaly detection warning when the initial anomaly detection results meet the initial warning requirements.

4. The heart attack warning system based on image recognition according to claim 2 or 3, characterized in that: The image standardization processing process of the image standard processing module includes: Receiving original medical image data acquired by the plurality of medical image acquisition devices in real time, and performing noise reduction processing on the original medical image data to obtain original medical image data after noise reduction processing; Performing data image cropping and correction processing on the original medical image after the noise reduction processing according to preset image requirement information to obtain a medical image after cropping and correction processing; Performing initial standardized adjustment on the cropped and corrected medical image according to a preset resolution and size to obtain an initially standardized medical image; A secondary normalization adjustment is performed on the initially normalized medical image according to preset image contrast and brightness to obtain a normalized medical image.

5. The heart attack warning system based on image recognition according to claim 2 or 3, characterized in that: The operation process of the first edge computing device includes: Monitoring the operating status parameters of the second edge computing device in real time, wherein the operating status parameters include CPU utilization, memory utilization, and processing queue length; Obtaining a load index of the current second edge computing device according to the CPU utilization, memory utilization, and processing queue length; When the load index of the second edge computing device does not exceed a preset load index threshold, sending the standardized medical image to the second edge computing device; When the load index of the second edge computing device exceeds a preset load index threshold, real-time monitoring is performed on a first duration during which the load index of the second edge computing device exceeds the preset load index threshold; When the load index of the second edge computing device exceeds the preset load index threshold for a first duration exceeding a preset time length, the first edge computing device automatically starts the initial abnormality detection function, and obtains the number of image data processed by the first edge computing device according to the number of standardized medical images to be detected contained in the processing queue length of the second edge computing device and the current load index of the second edge computing device; wherein the number of image data processed is obtained by the following formula: Where, N represents the number of image data processed, and N is rounded up; K represents the load index; K y represents a preset load index threshold; m represents the number of standardized processed medical images to be detected contained in the processing queue length; The first edge computing device retrieves, according to the number of processed image data, the same number of standardized medical images to be detected as the number of processed image data from the processing queue length of the second edge computing device as the target medical images, and performs initial anomaly detection on the target medical images to obtain initial anomaly detection results; When the load index of the second edge computing device drops below a preset load index threshold, and the duration of being below the preset load index threshold exceeds a preset second duration, the first edge computing device automatically turns off the initial anomaly detection function.

6. The heart attack warning system based on image recognition according to claim 5, characterized in that: The process of obtaining the load index is as follows: Extracting an operating status parameter corresponding to each unit time of the second edge computing device, wherein the operating status parameter includes CPU utilization, memory utilization, and processing queue length; The parameter matrix A corresponding to each unit time is generated using the CPU utilization, memory utilization and processing queue length included in the operating state parameters corresponding to each unit time; wherein the structure of the parameter matrix A is as follows: Among them, A represents the parameter matrix corresponding to each unit time; a 11 、a 12 ……a n3 represents all elements contained in the parameter matrix A; n represents the number of unit times experienced by the corresponding second edge computing device under each unit time; Extract the parameter matrix A corresponding to each unit time that the second edge computing device has experienced in history and the parameter matrix A corresponding to the current unit time; Obtain the load index of the current second edge computing device by using the parameter matrix A corresponding to each unit time that the second edge computing device has experienced in history and the parameter matrix A corresponding to the current unit time; The load index is obtained by the following formula: Where K represents the load index; σ max Indicates the maximum singular number corresponding to the parameter matrix A corresponding to the current unit time; Indicates the norm of the parameter matrix A corresponding to the current unit time; A represents the parameter matrix A corresponding to the current unit time T The determinant of the matrix A; z and p Represents the median and mean of the determinant of the parameter matrix corresponding to all unit times that the second edge computing device has experienced in history; Var represents the variance of the determinant of the parameter matrix corresponding to all unit times that the second edge computing device has experienced in history; ε represents the minimum constant used to prevent the independent variable of the logarithmic function from being 0; x represents the base of the preset logarithmic function, and x≥2.

7. The heart attack warning system based on image recognition according to claim 2, characterized in that: The process of performing initial anomaly detection by the first edge computing device and the second edge computing device includes: Extracting regions from different parts of the heart in the standardized medical image for early warning of heart attacks, and obtaining a plurality of heart region images corresponding to the different parts of the heart; Acquire characteristic data corresponding to different parts of each heart from the plurality of heart region images, wherein the characteristic data include the size of the ventricles and atria, the shape of the ventricles and atria, the thickness of the heart wall and the shape of the heart valve; Retrieving standard data corresponding to the size of the ventricle and atrium, the shape of the ventricle and atrium, the thickness of the heart wall and the shape of the heart valve; Compare the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape contained in the feature data with the standard data corresponding to the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape, and obtain a similarity value between the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape contained in the feature data and the standard data corresponding to the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape; The abnormal index parameter is obtained by using the similarity value between the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape contained in the characteristic data and the standard data corresponding to the size of the ventricles and atria, the shape of the ventricles and atria, the heart wall thickness and the heart valve shape, wherein the abnormal index parameter is obtained by the following formula: Where E represents the anomaly index parameter; r represents the number of feature data types contained in the feature data; S i represents the similarity value between the i-th feature data and its corresponding standard data; S min and S max They represent the minimum and maximum similarity values ​​among r types of feature data respectively; w min Indicates the weight value of the feature data corresponding to the minimum similarity value; w mx represents the minimum weight among r types of feature data; w max Indicates the weight value of the feature data corresponding to the maximum similarity value; w my Represents the maximum weight among r types of feature data; The abnormal index parameter is compared with the preset abnormal index threshold. When the abnormal index parameter is lower than the preset abnormal index threshold, it is determined that there is no abnormal situation; when the abnormal index parameter is not lower than the preset abnormal index threshold, it is determined that there is an abnormal situation and an initial abnormal detection warning is performed.

8. The heart attack warning system based on image recognition according to claim 1, characterized in that: The central server includes a data receiving module, a data processing module, a risk assessment module and an early warning module; wherein the signal input end of the data receiving module is the signal input end of the central server; the signal output end of the data receiving module is connected to the signal input end of the data processing module; the signal output end of the data processing module is connected to the signal input end of the risk assessment module; the signal output end of the risk assessment module is connected to the signal input end of the early warning module; The data receiving module is used to receive in real time the medical image for heart attack warning after standardized processing that does not meet the initial warning requirements when the medical image for heart attack warning after standardized processing does not meet the initial warning requirements; The data processing module is used to perform data preprocessing on the medical image for heart attack warning after standardization that does not meet the initial warning requirements to obtain a preprocessed medical image; wherein the data preprocessing includes noise reduction processing and data cleaning processing; The risk assessment module is used to perform a risk assessment of a heart attack based on the preprocessed medical image to obtain a risk assessment result; The early warning module is used to issue a heart attack early warning when the risk assessment result indicates that there is a risk of heart attack.

9. The heart attack warning system based on image recognition according to claim 8, characterized in that: The process of the risk assessment module performing heart attack risk assessment based on the preprocessed medical image includes: In the case where the initial warning requirements are not met, the pre-processed medical images corresponding to the requirements not meeting the initial warning requirements are retrieved; Retrieving a standardized medical image for heart attack warning from the historical record of the patient corresponding to the pre-processed medical image that does not meet the initial warning requirements; sorting the medical images according to the acquisition time of the standardized medical images for heart attack warning in the patient's historical records to generate a medical image set; The risk of heart attack is assessed using the abnormal index parameters of every two adjacent medical images in the medical image set.

10. The heart attack warning system based on image recognition according to claim 9, characterized in that: Using the abnormal index parameters of each two adjacent medical images in the medical image set to assess the risk of heart attack includes: Extracting abnormality index parameters of every two adjacent medical images in the medical image set; Compare the abnormal index parameters of each two adjacent medical images to obtain an abnormal index parameter difference, and extract from the abnormal index parameter difference an abnormal index parameter difference corresponding to a case where the abnormal index parameter of the medical image preceding the medical image is lower than the abnormal index parameter of the medical image following the medical image; Compare the abnormal index parameter difference corresponding to each abnormal index parameter of the previous medical image being lower than the abnormal index parameter of the subsequent medical image with a preset difference threshold; When the abnormal index parameter difference corresponding to the abnormal index parameter of the previous medical image is lower than the abnormal index parameter of the later medical image exceeds a preset difference threshold, the previous medical image and the later medical image corresponding to the abnormal index parameter difference exceeding the preset difference threshold are marked as a target abnormal image group; Retrieving similarity values ​​of the sizes of the ventricles and atria, the shapes of the ventricles and atria, the thickness of the heart wall and the shape of the heart valves of the medical images before and after the medical images included in each target abnormal image group; The risk determination value is obtained by using the similarity values ​​of the size of the ventricles and atria, the shape of the ventricles and atria, the thickness of the heart wall and the shape of the heart valves of the medical images before and after the medical images included in all target abnormal image groups; wherein the risk determination value is obtained by the following formula: Wherein, Y represents the risk determination value; x represents the base of the preset logarithmic function, and x≥2; β represents the positive offset, and the value range of the positive offset is 0.28-3.63; k represents the number of target abnormal image groups; t j represents the time difference between the later medical image in the jth target abnormal image group and the current time T; Q represents the similarity parameter corresponding to all target abnormal image groups, and the similarity parameter is obtained by the following formula: Where Q represents the similarity parameter corresponding to all target abnormal image groups; k represents the number of target abnormal image groups; r j represents the number of feature data types contained in the feature data of the jth target abnormal image group; S hi and S xi represents the similarity value between the i-th feature data and the standard data corresponding to the previous medical image and the later medical image of the j-th target abnormal image group; E hj and E xj Indicates the abnormal index parameters corresponding to the previous medical image and the subsequent medical image of the j-th target abnormal image group; When the risk determination value exceeds a preset risk determination threshold, it is determined that there is a risk of a heart attack, and a heart attack warning is issued.

Images