Medical image transmission method based on intelligent medical system

Abstract

The invention relates to a medical image transmission method based on an intelligent medical system. The method comprises the following steps: receiving original medical image data of the intelligent medical system in real time; acquiring patient information associated with the original medical image data, identifying and classifying the original medical image data through a preset image identification decision model, and deciding to generate an image transmission processing scheme; generating a control instruction according to the image transmission processing flow information, and processing the original medical image data; generating a control instruction according to the image transmission processing flow information, and transmitting the processed image data to a receiving end according to an image transmission protocol and an image transmission flow; and generating a control instruction according to the image transmission processing flow information, and controlling the receiving end to process and display the received and processed image data according to the image receiving flow. The method has the effect of effectively improving the medical image transmission processing efficiency and quality.

Description

Technical Field

[0001] This application relates to the field of image processing, and in particular to a medical image transmission method based on an intelligent medical system. Background Art

[0002] An intelligent medical system is a complex system that deeply integrates information technology, artificial intelligence technology, and medical services. It takes medical data as the core, uses Internet of Things devices to collect data, performs data storage, management, and analysis through cloud computing and big data analysis technologies, and combines artificial intelligence algorithms to assist medical decision-making, disease diagnosis, and treatment processes. Its purpose is to improve medical quality, enhance medical efficiency, improve the patient experience, and optimize the allocation of medical resources.

[0003] Existing intelligent medical systems include a medical device access layer, a data transmission layer, a data storage and processing layer, and a user access layer. The medical device access layer is used to connect various medical imaging devices, such as X-ray machines, CT scanners, MRI devices, etc. These devices need to have a digital image output interface and be able to output image data in a standard digital format (such as DICOM - Digital Imaging and Communications in Medicine standard). The data transmission layer is mainly responsible for establishing a secure and efficient network channel, including hospital internal local area network (LAN), wide area network (WAN) connections, and interfaces with external networks (such as the Internet). To ensure the stability of data transmission, high-performance network devices, such as routers, switches, etc., need to be configured, and appropriate network protocols, such as the TCP / IP protocol family, need to be adopted. The data storage and processing layer includes storage servers and data processing servers. The user access layer mainly provides access interfaces for medical staff, management personnel, and patients, including receiving ends such as computer terminals at doctor workstations, nurse stations, and application programs on mobile medical devices (such as tablets, smartphones).

[0004] However, medical images of different diseases have their unique imaging characteristics and diagnostic key points. The existing image transmission and processing process of intelligent medical systems is rigid, which is likely to mask lesion characteristics or produce false positive and false negative results, and there is a risk of missed diagnosis and misdiagnosis. Summary of the Invention

[0005] To solve the above problems, this application provides a medical image transmission method based on an intelligent medical system.

[0006] In a first aspect, this application provides a medical image transmission method based on an intelligent medical system, adopting the following technical solution:

[0007] A medical image transmission method based on an intelligent medical system includes the following steps:

[0008] Real-time receive the original medical image data of various medical imaging devices in the intelligent medical system;

[0009] Obtain patient information associated with the original medical image data, and identify and classify the original medical image data through a pre-set image recognition decision model, and generate an image transmission processing scheme by decision; the image transmission processing scheme includes image type information, image transmission type information, image transmission protocol, at least one image preprocessing model, at least one image segmentation model, at least one diagnostic assistance model, and image transmission processing flow information; the image transmission processing flow information includes image processing flow information, image transmission flow information, and image reception flow information;

[0010] Generate a control instruction according to the image transmission processing flow information, and control the image preprocessing model, the image segmentation model, and the diagnostic assistance model to process the original medical image data according to the image processing flow;

[0011] Generate a control instruction according to the image transmission processing flow information, and transmit the processed image data to the receiving end according to the image transmission protocol and in accordance with the image transmission flow;

[0012] Generate a control instruction according to the image transmission processing flow information, and control the receiving end to process the received and processed image data and display it according to the image reception flow.

[0013] Preferably, the specific steps for the real-time reception of the original medical image data of various medical imaging devices in the intelligent medical system are as follows:

[0014] Deploy an image data reception module at the bottom layer of the architecture of the intelligent medical system;

[0015] Based on the DICOM standard protocol, adopt a continuous listening mode to receive data, and capture in real time the image transmission requests at various medical imaging device ends in the intelligent medical system;

[0016] Based on the image transmission request, start a pre-set multi-threaded reception mechanism to receive image data from different medical imaging device ends in parallel

[0017] Perform integrity verification on the received image data. If it is found that the data is partially missing or damaged, send a retransmission request to the medical imaging device end. If the verification passes, the original medical image data is obtained.

[0018] Preferably, the specific steps for the above-mentioned obtaining of patient information associated with the original medical image data, identifying and classifying the original medical image data through a pre-set image recognition decision model, and generating an image transmission processing scheme by decision are as follows:

[0019] Deeply integrate with the HIS system, quickly retrieve and extract the associated patient information from the HIS system based on the unique identification information preset in the original medical image data. The patient information includes patient basic data, case information, and examination items applied for;

[0020] Determine the type range of the patient's original medical images based on the patient information;

[0021] According to the type range of the patient's original medical images, use the pre-set image recognition decision model to identify and classify the original medical image data, and clarify the image type information and image transmission type information of the original medical images;

[0022] The image recognition decision model selects an image transmission protocol according to the image transmission type information of the original medical images;

[0023] The image recognition decision model configures an image preprocessing model, an image segmentation model, and a diagnostic assistance model according to the image type information of the original medical images, and plans and generates image transmission processing flow information;

[0024] Integrate and package to generate an image transmission processing solution.

[0025] Preferably, the image recognition decision model configuring an image preprocessing model, an image segmentation model, and a diagnostic assistance model according to the image type information of the original medical images specifically includes the following steps:

[0026] The image recognition decision model obtains the enterprise's already constructed model information and generates a model library directory; the already constructed model information includes special models and general models;

[0027] The image recognition decision model precisely screens and matches the image type information of the original medical images in the model library directory, and decides to configure an image preprocessing model, an image segmentation model, and a diagnostic assistance model.

[0028] Preferably, the image recognition decision model obtaining the enterprise's already constructed model information further includes: obtaining model update information in real time, performing performance verification on the updated model, and updating the model library directory specifically includes the following steps:

[0029] Obtain model update information in real time and determine whether the updated model is a special model;

[0030] If it is a special model, randomly select several cases from the historical processing cases of the special model as test samples, perform performance tests on the updated special model and the previous version of the special model, and select the version of the special model with more excellent performance test results to update on the model library directory;

[0031] If it is not a dedicated model, then randomly select a number of test samples from each disease category currently adapted by the general model, perform performance tests on the updated general model and the previous version of the general model, and compare the performance test results of each disease category in turn to determine whether there is a version of the general model with better performance test results in all disease categories;

[0032] If there is, then select this version of the general model for update in the model library directory;

[0033] If not, then perform cracking classification on the general model according to the performance test results of each disease category, generate multiple versions of the general model and update them in the model library directory.

[0034] Preferably, the generating control instructions according to the image transmission processing flow information to control the image preprocessing model, the image segmentation model and the diagnostic assistance model to process the original medical image data according to the image processing flow specifically includes the following steps:

[0035] According to the image transmission processing flow information, compile standardized control instructions, and the control instructions clarify the start time of each model, the input data source, the running parameters and the output destination;

[0036] Based on the control instructions, control the image preprocessing model to read the original medical image data, preprocess the original medical image data according to the image processing flow, and transmit the processed image data to the input buffer of the image segmentation model in real time;

[0037] Based on the control instructions, control the image segmentation model to perform fine segmentation on the preprocessed image data according to the built-in segmentation algorithm according to the image processing flow, and output the segmentation result to the input buffer of the diagnostic assistance model;

[0038] Based on the control instructions, control the diagnostic assistance model to perform diagnostic analysis on the segmented image data according to the image processing flow to generate diagnostic suggestions.

[0039] Preferably, the image segmentation model performing fine segmentation on the preprocessed image data further includes: the image segmentation model rotates, scales and stitches the segmented image data according to the control instructions according to the image processing flow to form an image model of the organs in the image.

[0040] Preferably, the generating control instructions according to the image transmission processing flow information and transmitting the processed image data to the receiving end according to the image transmission protocol and the image transmission flow specifically includes the following steps:

[0041] Generate control instructions according to the image transmission processing flow information, and the control instructions include the transmission start point, the transmission receiving end, the image transmission protocol, the transmission priority, the bandwidth allocation strategy and the encryption method;

[0042] Encrypt the processed image data according to the control instruction;

[0043] According to the control instruction, transmit the encrypted image data from the transmission starting point to the receiving end according to the image transmission protocol and the image transmission processing flow.

[0044] Preferably, generating a control instruction according to the image transmission processing flow information to control the receiving end to process the received and processed image data according to the image receiving flow and display it specifically includes the following steps:

[0045] Generate a control instruction according to the image transmission processing flow information to control the receiving end to continuously monitor the arrival of image data at the specified port;

[0046] According to the control instruction, control the receiving end to process the received and processed image data according to the image receiving flow and verify the data integrity;

[0047] After the integrity verification passes, decode the received image data according to the control instruction according to the image transmission protocol to restore the original resolution and quality of the image;

[0048] Decrypt the received image data according to the control instruction;

[0049] Adjust the parameters of the decrypted image data according to the device parameters of the receiving end so that it is displayed on the receiving end.

[0050] Preferably, adjusting the parameters of the decrypted image data according to the device parameters of the receiving end so that it is displayed on the receiving end specifically includes the following steps:

[0051] Obtain the device parameters of the receiving end, and the device parameters include screen resolution and color mode;

[0052] Compare the resolution of the decrypted image data with the screen resolution of the receiving end to determine whether the resolution of the decrypted image data is higher than the screen resolution of the receiving end;

[0053] If it is higher, scale the decrypted image data based on the bilinear interpolation algorithm to obtain preliminary image data;

[0054] If it is not higher, process the decrypted image data through a pre-set super-resolution reconstruction model to obtain preliminary image data; the super-resolution reconstruction model is a machine learning model with a generative adversarial network (GAN) architecture and is obtained by deep learning training using high- and low-resolution medical images as sample data;

[0055] The preliminary image data is input into a preset parameter optimization model for contrast, brightness optimization adjustment, and color precision correction, and then displayed at the receiving end; the parameter optimization model is a convolutional neural network model obtained by iterative training with medical image data under different types and different lighting conditions.

[0056] In summary, the present application includes at least one of the following beneficial technical effects:

[0057] 1. When receiving the original medical image data of a patient generated by an intelligent medical system, obtaining the detailed patient information by integrating with the HIS system, being able to determine the image type range based on this information, and then using the image recognition decision model to identify the image to achieve precise classification, clarifying the image type and transmission type; then, according to the actual R & D situation of the medical image processing model of the medical institution, based on the image type and transmission type, the image recognition decision model decides to configure appropriate preprocessing, segmentation, and diagnostic assistance models, and plans detailed processing flow information and determines the image transmission protocol, enabling operations such as image preprocessing, segmentation, and diagnostic analysis to proceed orderly and efficiently; this personalized configuration avoids the problem that the fixed process in the existing system may cover up lesion characteristics or produce incorrect results, clearly highlighting the lesion characteristics on the processed medical image, and finally achieving stable transmission of the medical image based on the transmission protocol, helping doctors intuitively understand the patient's lesion information from the medical image, reducing the risk of missed diagnosis and misdiagnosis, and achieving the effect of effectively improving the efficiency and quality of medical image transmission and processing;

[0058] 2. Real-time obtaining of model update information, performing performance verification on the updated model and updating the model library directory; based on the actual situation of the model, for a dedicated model, samples are extracted from its historical processing cases for performance testing, comparing the new and old versions, and only retaining the version with better performance for updating to the model library directory, which ensures that the dedicated model always maintains the best performance state, can better process specific diseases or specific types of images, and provides more reliable support for clinical diagnosis; for a general model, sampling tests are respectively carried out in each adapted disease category to determine whether there is a new general model version that is comprehensively better than the old version. If there is, it is updated; if not, it is split and classified; this method not only ensures the overall performance optimization of the general model but also takes into account the special needs of different disease categories. Through split classification, more targeted general model variants can be formed for different diseases, improving its applicability and performance on various disease images;

[0059] 3. According to the device parameter characteristics of the receiving end, the bilinear interpolation algorithm and the super-resolution reconstruction model are used to intelligently scale the image data. Based on the color mode of the receiving device, the contrast, brightness optimization adjustment and color precise correction of the preliminary image data are carried out through the parameter optimization model (convolutional neural network model). By training the medical image data under different types and lighting conditions, the model can better adapt to various complex image display situations, make the image present the best display effect, help doctors more accurately identify lesion features, provide better visual support for clinical diagnosis, and achieve the effect of effectively improving the efficiency and quality of medical image transmission and processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is the flowchart of a medical image transmission method based on an intelligent medical system in an embodiment of the present application;

[0061] Figure 2 is the flowchart of the method for real-time receiving of original medical image data in an embodiment of the present application;

[0062] Figure 3 is the flowchart of the method for generating an image transmission and processing scheme by identifying and classifying and making decisions on the original medical image data in an embodiment of the present application;

[0063] Figure 4 is the flowchart of the method for decision-making and configuration of the image recognition decision model in an embodiment of the present application;

[0064] Figure 5 is the flowchart of the method for real-time obtaining of model update information to update the model library directory in an embodiment of the present application;

[0065] Figure 6 is the flowchart of the method for processing the original medical image data in an embodiment of the present application;

[0066] Figure 7 is the flowchart of the method for transmitting the processed image data to the receiving end in an embodiment of the present application;

[0067] Figure 8 is the flowchart of the method for displaying the processed image data at the receiving end in an embodiment of the present application;

[0068] Figure 9 is the flowchart of the method for parameter adjustment of the decrypted image data in an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0069] The following further describes the present application in detail Figures 1-9 with reference to the accompanying drawings.

[0070] An embodiment of the present application discloses a medical image transmission method based on an intelligent medical system. Refer to Figure 1 A medical image transmission method based on an intelligent medical system includes the following steps:

[0071] S1. Receive original medical image data: Receive the original medical image data of various medical imaging devices in the intelligent medical system in real time;

[0072] S2. Generate an image transmission processing scheme: Obtain the patient information associated with the original medical image data, and perform recognition and classification on the original medical image data through a pre-set image recognition decision model, and decide to generate an image transmission processing scheme; the image transmission processing scheme includes image type information, image transmission type information, an image transmission protocol, at least one image preprocessing model, at least one image segmentation model, at least one diagnostic assistance model, and image transmission processing flow information; the image transmission processing flow information includes image processing flow information, image transmission flow information, and image reception flow information;

[0073] S3. Image data processing: Generate a control instruction according to the image transmission processing flow information, and control the image preprocessing model, the image segmentation model, and the diagnostic assistance model to process the original medical image data according to the image processing flow;

[0074] S4. Image data transmission: Generate a control instruction according to the image transmission processing flow information, and transmit the processed image data to the receiving end according to the image transmission protocol and in accordance with the image transmission flow;

[0075] S5. Image display at the receiving end: Generate a control instruction according to the image transmission processing flow information, and control the receiving end to process the received and processed image data according to the image reception flow and display it. Through the above steps, when receiving the original medical image data of a patient generated by the intelligent medical system, by integrating with the HIS system to obtain the patient's detailed information, the range of image types can be determined based on this information, and then the image recognition decision model is used to identify and classify the image to achieve accurate classification and clarify the image type and transmission type; then, according to the actual research and development situation of the medical image processing models in medical institutions, the image recognition decision model is used to decide and configure appropriate preprocessing, segmentation, and diagnostic assistance models based on the image type and transmission type, and plan detailed processing flow information and determine the image transmission protocol, so that operations such as image preprocessing, segmentation, and diagnostic analysis can be carried out orderly and efficiently; this personalized configuration avoids the problem that the fixed process in the existing system may cover up lesion characteristics or produce incorrect results, clearly highlights the lesion characteristics on the processed medical image, and finally realizes the stable transmission of medical images based on the transmission protocol, which helps doctors intuitively understand the patient's lesion information from the medical image, reduces the risk of missed diagnosis and misdiagnosis, and achieves the effect of effectively improving the efficiency and quality of medical image transmission processing.

[0076] Specifically, some common image transmission protocols, image preprocessing models, image segmentation models, and diagnostic assistance models are exemplified. For example, image transmission protocols include the DICOM protocol, HTTP protocol, FTP protocol, etc.; image preprocessing models include mean filtering models, window width and window level adjustment models, Gaussian filtering models, histogram equalization models, etc. Based on this, dedicated models can also be developed through special optimization for some diseases according to their principles. For example, for lung CT images, a dedicated Gaussian filtering noise reduction model is set up to remove noise interference during the scanning process, and a window width and window level adjustment model is used to highlight key features such as lung nodules; for brain MRI images, an artifact removal model based on phase correction and a histogram equalization enhancement model are equipped to improve the image quality for subsequent diagnosis. Image segmentation models include threshold-based segmentation models, region-based segmentation models, U-Net models, etc. There are also dedicated models developed through special training for disease characteristics. Diagnostic assistance models include deep learning-based classification models, rule-based diagnostic models, case-based reasoning models, etc. It should be emphasized that although this application advocates training dedicated models to increase the accuracy of disease analysis and diagnosis for diagnostic assistance models, comprehensive diagnostic assistance models that can be applied to various disease analyses should also be within the scope of protection of this application.

[0077] Refer to Figure 2 , the specific steps for real-time receiving of the original medical image data of various medical imaging devices in the intelligent medical system are as follows:

[0078] A1. Deploy an image data receiving module: Deploy an image data receiving module at the bottom layer of the architecture of the intelligent medical system;

[0079] A2. Adopt a continuous listening mode: Based on the DICOM standard protocol, adopt a continuous listening mode to receive data and capture the image transmission requests of various medical imaging device terminals in the intelligent medical system in real time;

[0080] A3. Receive image data in parallel: Based on the image transmission request, start a pre-set multi-thread receiving mechanism to receive image data from different medical imaging device terminals in parallel;

[0081] A4. Integrity verification of image data: Perform integrity verification on the received image data. If partial loss or damage is found in the data, a retransmission request is sent to the medical imaging device side. If the verification passes, the original medical image data is obtained. By deploying a dedicated image data receiving module and continuously listening based on the DICOM standard protocol, the image transmission requests of the device can be captured in real time, and a multithreading mechanism is used to receive data in parallel, improving the efficiency and stability of data reception. This ensures that medical image data can enter the system quickly and accurately, reducing the risk of data loss or delay. Then, by performing integrity verification on the received image data and promptly requesting retransmission once data loss or damage is found, it is ensured that the data for subsequent processing is complete and accurate, providing a reliable data basis for subsequent image processing and diagnostic analysis, facilitating the stable and efficient transmission of the processed medical images to the doctor's receiving end, helping doctors intuitively understand the patient's lesion information based on the medical images, reducing the risk of missed diagnosis and misdiagnosis, and achieving the effect of effectively improving the efficiency and quality of medical image transmission and processing.

[0082] Refer to Figure 3 , obtaining the patient information associated with the original medical image data, and performing identification and classification on the original medical image data through a pre-set image recognition decision model. The decision to generate an image transmission and processing plan specifically includes the following steps:

[0083] B1. Obtaining the patient information associated with the original medical image data: Deeply integrate with the HIS system, and quickly retrieve and extract the associated patient information from the HIS system based on the unique identification information preset in the original medical image data. The patient information includes the patient's basic data, case information, and examination items applied for;

[0084] B2. Determining the type range of the patient's original medical images based on the patient information;

[0085] B3. Clarifying the image type information and image transmission type information of the original medical image: Through a pre-set image recognition decision model, according to the type range of the patient's original medical images, perform identification and classification on the original medical image data to clarify the image type information and image transmission type information of the original medical image;

[0086] B4. Selecting an image transmission protocol: The image recognition decision model selects an image transmission protocol according to the image transmission type information of the original medical image;

[0087] B5. Decision-configuring the model and planning to generate image transmission and processing flow information: The image recognition decision model decides to configure an image preprocessing model, an image segmentation model, and a diagnostic assistance model according to the image type information of the original medical image, and plans to generate image transmission and processing flow information;

[0088] B6. Integrate and package to generate an image transmission and processing solution. By deeply integrating with the HIS system, it not only ensures the accurate matching of images and patient information, provides rich and comprehensive clinical background data for generating the image transmission and processing solution for subsequent decision-making, makes the processing process no longer isolated, but is customized closely in combination with the actual situation of the patient, improves the pertinence and practicality of the entire processing process; avoids the problem that the fixed process in the existing system may cover up lesion characteristics or produce incorrect results, provides clear guidance for subsequent processing and transmission, and achieves the effect of effectively improving the efficiency and quality of medical image transmission and processing.

[0089] Refer to Figure 4 , the image recognition decision model configures the image preprocessing model, image segmentation model, and diagnostic assistance model according to the image type information of the original medical image. The specific steps are as follows:

[0090] C1. Obtain the information of the models already built by the enterprise: The image recognition decision model obtains the information of the models already built by the enterprise and generates a model library directory; the information of the models already built includes special models and general models.

[0091] C2. Configure the models by decision: The image recognition decision model accurately screens and matches the image type information of the original medical image in the model library directory, and configures the image preprocessing model, image segmentation model, and diagnostic assistance model. According to the actual R & D situation of the medical image processing models in medical institutions, through the image recognition decision model based on the image type and transmission type, it configures appropriate preprocessing, segmentation, and diagnostic assistance models, plans detailed processing process information, and determines the image transmission protocol, so that operations such as image preprocessing, segmentation, and diagnostic analysis can be carried out orderly and efficiently, helps doctors intuitively understand the lesion information of patients from medical images, reduces the risk of missed diagnosis and misdiagnosis, and achieves the effect of effectively improving the efficiency and quality of medical image transmission and processing.

[0092] Refer to Figure 5 , the above-mentioned image recognition decision model obtaining the information of the models already built by the enterprise also includes: obtaining model update information in real time, performing performance verification on the updated models, and updating the model library directory. The specific steps are as follows:

[0093] D1. Determine whether the updated model is a special model: Obtain model update information in real time and determine whether the updated model is a special model.

[0094] D2. Test and update the model: If it is a special model, randomly select several cases from the historical processing cases of the special model as test samples, perform performance tests on the updated special model and the previous version of the special model, compare the performance test results, and select the version of the special model with more excellent performance test results to update on the model library directory.

[0095] D3. Determine whether there is a general model version with better performance test results in all disease categories: If it is not a dedicated model, randomly select several test samples from each disease category currently adapted by the general model, conduct performance tests on the updated general model and the previous version of the general model, compare the performance test results of each disease category in turn, and determine whether there is a general model version with better performance test results in all disease categories;

[0096] D4. If there is, select this general model version for update in the model library list;

[0097] D5. If not, classify the general model according to the performance test results of each disease category, generate multiple general model versions and update them in the model library list. Through the above steps, obtain model update information in real time, perform performance verification on the updated model and update the model library list; based on the actual situation of the model, for the dedicated model, extract samples from its historical processing cases for performance testing, compare the new and old versions, and only retain the version with better performance for update to the model library list, which ensures that the dedicated model always maintains the best performance state, can better process specific diseases or specific types of images, and provides more reliable support for clinical diagnosis; for the general model, conduct sampling tests in each adapted disease category to determine whether there is a new general model version that is comprehensively better than the old version. If there is, update it; if not, classify it by splitting. This method not only ensures the performance optimization of the general model as a whole but also takes into account the special needs of different disease categories. Through classification by splitting, more targeted general model variants can be formed for different diseases, improving its applicability and performance on various disease images.

[0098] The mechanism of classifying the general model by splitting enables the system to subdivide the general model according to the characteristics of different disease categories. The medical image features of different diseases vary greatly. Through this method, model versions that better meet the needs of various diseases can be generated, effectively reducing the R & D pressure and saving R & D costs, while improving the system's adaptability to diverse clinical needs.

[0099] Refer to Figure 6 , the generating control instructions according to the image transmission and processing flow information to control the image preprocessing model, image segmentation model, and diagnostic assistance model to process the original medical image data according to the image processing flow specifically includes the following steps:

[0100] E1. Compile standardized control instructions: According to the image transmission and processing flow information, compile standardized control instructions, and the control instructions clarify the startup time, input data source, operation parameters, and output destination of each model;

[0101] E2. Preprocess the original medical image data: Based on the control instruction, control the image preprocessing model to read the original medical image data, preprocess the original medical image data according to the image processing process, and transmit the processed image data to the input buffer of the image segmentation model in real time;

[0102] E3. Perform fine segmentation on the preprocessed image data: Based on the control instruction, control the image segmentation model to perform fine segmentation on the preprocessed image data according to the built-in segmentation algorithm and the image processing process, and output the segmentation result to the input buffer of the diagnostic assistance model;

[0103] E4. Perform diagnostic analysis on the segmented image data to generate diagnostic suggestions: Based on the control instruction, control the diagnostic assistance model to perform diagnostic analysis on the segmented image data according to the image processing process to generate diagnostic suggestions. Compile standardized control instructions to clarify the startup time, input data source, operating parameters, and output destination of each model, so that operations such as image preprocessing, segmentation, and diagnostic analysis can be carried out orderly and efficiently. This precise control ensures that each processing link can be carried out according to the predetermined process and parameters, improving the accuracy and reliability of processing, and avoiding the loss of lesion characteristics or misdiagnosis caused by chaotic processing processes.

[0104] In addition, the above-mentioned image segmentation model's fine segmentation of the preprocessed image data also includes: The image segmentation model rotates, scales, and stitches the segmented image data according to the control instruction and the image processing process to form an image model of the internal organs of the image. For special medical images such as colonoscopy and gastroscopy images, the image segmentation model not only performs fine segmentation on the image, but also can rotate, scale, and stitch according to the instruction to form an image model of the internal organs of the image, providing more intuitive information for diagnosis. Cooperating with the diagnostic analysis model, it can provide strong decision-making support for doctors, helping to improve the accuracy of diagnosis and reducing the risk of missed diagnosis and misdiagnosis.

[0105] Refer to Figure 7 , The generation of control instructions according to the image transmission processing flow information and the transmission of the processed image data to the receiving end according to the image transmission protocol and the image transmission process specifically include the following steps:

[0106] F1. Generate control instructions: Generate control instructions according to the image transmission processing flow information. The control instructions include the transmission start point, transmission receiving end, image transmission protocol, transmission priority, bandwidth allocation strategy, and encryption method;

[0107] F2. Perform encryption processing: Perform encryption processing on the processed image data according to the control instruction;

[0108] F3. Transmit the image data to the receiving end: According to the control instruction, transmit the encrypted image data from the transmission starting point to the receiving end according to the image transmission protocol in accordance with the image transmission processing flow. Through the above steps, the security and efficiency of the data during transmission are ensured. The reasonable bandwidth allocation and transmission priority setting can ensure the priority transmission of important image data, meet the real-time requirements of medical diagnosis, and solve the problems of unstable, insecure data transmission and inability to meet real-time requirements in the existing system.

[0109] Refer to Figure 8 , the generation of the control instruction according to the image transmission processing flow information, controlling the receiving end to process the received and processed image data according to the image receiving process and display it specifically includes the following steps:

[0110] G1. Control the receiving end to continuously monitor the arrival of image data: Generate a control instruction according to the image transmission processing flow information, and control the receiving end to continuously monitor the arrival of image data at the specified port;

[0111] G2. Verify data integrity: According to the control instruction, control the receiving end to process the received and processed image data according to the image receiving process, and verify data integrity;

[0112] G3. Decode the received image data: After the integrity verification passes, decode the received image data according to the control instruction in accordance with the image transmission protocol to restore the original resolution and quality of the image;

[0113] G4. Perform decryption processing: Perform decryption processing on the received image data according to the control instruction;

[0114] G5. Display the image data on the receiving end: Adjust the parameters of the decrypted image data according to the device parameters of the receiving end to display it on the receiving end. Through the above steps, continuous monitoring is adopted to ensure data capture, double-check the stability of data transmission by verifying the integrity of the received data, then decode and decrypt the image data according to the transmission protocol, and adjust the image data according to the device parameters of the receiving end, so that the image can be displayed on different devices with the best effect, which helps doctors intuitively understand the lesion information of patients based on medical images, reduces the risk of missed diagnosis and misdiagnosis, and achieves the effect of effectively improving the efficiency and quality of medical image transmission and processing.

[0115] Refer to Figure 9 , the adjustment of the parameters of the decrypted image data according to the device parameters of the receiving end to display it on the receiving end specifically includes the following steps:

[0116] H1. Obtain the device parameters of the receiving end, and the device parameters include screen resolution and color mode;

[0117] H2. Determine whether the resolution of the decrypted image data is higher than the screen resolution of the receiving end: Compare the resolution of the decrypted image data with the screen resolution of the receiving end to determine whether the resolution of the decrypted image data is higher than the screen resolution of the receiving end;

[0118] H3. If it is higher, scale the decrypted image data based on the bilinear interpolation algorithm to obtain preliminary image data;

[0119] H4. If it is not higher, process the decrypted image data through a pre-set super-resolution reconstruction model to obtain preliminary image data; the super-resolution reconstruction model is a machine learning model with a generative adversarial network (GAN) architecture, which is obtained by performing deep learning training using high- and low-resolution medical images as sample data; it should be noted that the specific training steps of the above machine learning model are all prior arts and will not be elaborated here;

[0120] H5. Optimization of the display of preliminary image data: Input the preliminary image data into a pre-set parameter optimization model for contrast, brightness optimization adjustment, and color precision correction, and then display it on the receiving end; the parameter optimization model is a convolutional neural network model obtained by performing iterative training using medical image data under different types and different lighting conditions. According to the device parameter characteristics of the receiving end, the bilinear interpolation algorithm and the super-resolution reconstruction model are used to intelligently scale the image data. Based on the color mode of the receiving device, the parameter optimization model (convolutional neural network model) is used to perform contrast, brightness optimization adjustment, and color precision correction on the preliminary image data. By training with medical image data under different types and lighting conditions, this model can better adapt to various complex image display situations, making the image present the best display effect, helping doctors more accurately identify lesion characteristics, providing better visual support for clinical diagnosis, and achieving the effect of effectively improving the efficiency and quality of medical image transmission and processing.

[0121] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting the protection scope of the invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on these embodiments, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art can still, without conflict, make combinations, additions, deletions, or other adjustments to the features in the embodiments of the present invention according to the situation without creative efforts, so as to obtain different technical solutions that essentially do not deviate from the concept of the present invention, and these technical solutions also belong to the scope of protection of the present invention.

Claims

1. A medical image transmission method based on an intelligent medical system, characterized in that: The following steps are involved: Receive raw medical image data from various medical imaging devices in the intelligent medical system in real time; Acquire patient information associated with original medical image data, identify and classify the original medical image data through a preset image recognition decision model, and make a decision to generate an image transmission processing scheme; the image transmission processing scheme includes image type information, image transmission type information, image transmission protocol, at least one image preprocessing model, at least one image segmentation model, at least one diagnosis auxiliary model, and image transmission processing flow information; the image transmission processing flow information includes image processing flow information, image transmission flow information, and image receiving flow information; Generate control instructions based on the image transmission processing flow information, control the image preprocessing model, image segmentation model and diagnosis auxiliary model, and process the original medical image data according to the image processing flow; Generate control instructions according to the image transmission processing flow information, and transmit the processed image data to the receiving end according to the image transmission protocol and the image transmission flow; A control instruction is generated according to the image transmission processing flow information, and the receiving end is controlled to process the received and processed image data according to the image receiving flow and display it.

2. A medical image transmission method based on an intelligent medical system according to claim 1, characterized in that: The real-time receiving of raw medical image data of various medical imaging devices in the intelligent medical system specifically includes the following steps: Deploy the image data receiving module at the bottom layer of the intelligent medical system architecture; Based on the DICOM standard protocol, it adopts continuous monitoring mode to receive data and capture image transmission requests from various medical imaging devices in the intelligent medical system in real time; Based on the image transmission request, the pre-set multi-threaded receiving mechanism is started to receive image data from different medical imaging devices in parallel. The received image data is integrity checked. If it is found that part of the data is missing or damaged, a retransmission request is sent to the medical imaging device. If the check passes, the original medical image data is obtained.

3. The medical image transmission method based on an intelligent medical system according to claim 1, characterized in that: The step of obtaining patient information associated with the original medical image data, identifying and classifying the original medical image data using a preset image recognition decision model, and generating an image transmission processing solution by decision specifically includes the following steps: Deeply integrate with the HIS system, and quickly retrieve and extract the associated patient information from the HIS system based on the unique identification information preset in the original medical image data. The patient information includes basic patient information, case information, and application examination items; determining a range of types of original medical images of the patient based on the patient information; The original medical image data is identified and classified according to the type range of the original medical image of the patient through a preset image recognition decision model, and the image type information and image transmission type information of the original medical image are clarified; The image recognition decision model selects an image transmission protocol according to the image transmission type information of the original medical image; The image recognition decision model configures the image preprocessing model, the image segmentation model and the diagnosis auxiliary model according to the image type information of the original medical image, and plans and generates the image transmission processing flow information; Integrate and package to generate image transmission processing solutions.

4. The medical image transmission method based on the intelligent medical system according to claim 3, characterized in that: The image recognition decision model configures the image preprocessing model, the image segmentation model and the diagnosis auxiliary model according to the image type information of the original medical image, and specifically includes the following steps: The image recognition decision model obtains the information of the model that has been constructed by the enterprise and generates a model library directory; the constructed model information includes special models and general models; The image recognition decision model accurately screens and matches the image type information of the original medical image in the model library directory, and decides to configure the image preprocessing model, image segmentation model and diagnosis auxiliary model.

5. The medical image transmission method based on the intelligent medical system according to claim 4, characterized in that: The image recognition decision model acquires the information of the model that has been built by the enterprise and further includes: acquiring model update information in real time, performing performance verification on the updated model, and updating the model library directory, which specifically includes the following steps: Obtain model update information in real time and determine whether the updated model is a dedicated model; If it is a special model, a number of cases are randomly selected from the special model historical processing cases as test samples, and the updated special model is tested against the previous version of the special model. The special model version with better performance test results is selected and updated in the model library directory after comparing the performance test results. If it is not a dedicated model, a number of test samples are randomly selected from each disease category currently adapted by the general model, and the updated general model is tested against the previous version of the general model. The performance test results of each disease category are compared in turn to determine whether there is a general model version with better performance test results for all disease categories. If it exists, select the general model version and update it in the model library directory; If it does not exist, the general model will be cracked and classified according to the performance test results of each disease category, and multiple general model versions will be generated and updated in the model library directory.

6. The medical image transmission method based on an intelligent medical system according to claim 1, characterized in that: The generating of control instructions according to the image transmission processing flow information, controlling the image preprocessing model, the image segmentation model and the diagnosis auxiliary model, and processing the original medical image data according to the image processing flow specifically includes the following steps: Based on the image transmission processing flow information, standard control instructions are compiled, and the control instructions specify the start time, input data source, operation parameters and output destination of each model; Based on the control instructions, the image preprocessing model is controlled to read the original medical image data, the original medical image data is preprocessed according to the image processing flow, and the processed image data is transmitted to the image segmentation model input buffer in real time; Based on the control instructions, the image segmentation model is controlled to perform fine segmentation on the pre-processed image data according to the built-in segmentation algorithm and the image processing flow, and the segmentation results are output to the input buffer area of ​​the diagnosis auxiliary model; Based on the control instructions, the diagnosis auxiliary model is controlled to perform diagnostic analysis on the segmented image data according to the image processing process to generate diagnostic suggestions.

7. The medical image transmission method based on the intelligent medical system according to claim 6, characterized in that: The image segmentation model further comprises performing fine segmentation on the pre-processed image data: the image segmentation model rotates, scales and splices the segmented image data according to the control instruction and the image processing flow to form an image model of the organ in the image.

8. The medical image transmission method based on an intelligent medical system according to claim 1, characterized in that: The generating of control instructions according to the image transmission processing flow information and transmitting the processed image data to the receiving end according to the image transmission protocol and the image transmission flow specifically includes the following steps: Generate control instructions according to the image transmission processing flow information, the control instructions including the transmission starting point, the transmission receiving end, the image transmission protocol, the transmission priority, the bandwidth allocation strategy and the encryption method; Encrypting the processed image data according to the control instruction; According to the control instruction, the encrypted image data is transmitted from the transmission starting point to the receiving end according to the image transmission protocol and the image transmission processing flow.

9. The medical image transmission method based on an intelligent medical system according to claim 1, characterized in that: The generating of control instructions according to the image transmission processing flow information and controlling the receiving end to process the received and processed image data according to the image receiving flow and display the received and processed image data specifically comprises the following steps: Generate control instructions based on the image transmission processing flow information to control the receiving end to continuously monitor the arrival of image data at the designated port; According to the control instructions, the receiving end is controlled to process the received image data according to the image receiving process and verify the data integrity; After the integrity check is passed, the received image data is decoded according to the control instruction and the image transmission protocol to restore the original resolution and quality of the image; Decrypting the received image data according to the control instruction; The decrypted image data is parameterized according to the device parameters of the receiving end so that it can be displayed on the receiving end.

10. The medical image transmission method based on the intelligent medical system according to claim 9, characterized in that: The step of adjusting the parameters of the decrypted image data according to the device parameters of the receiving end so that the image data can be displayed on the receiving end specifically comprises the following steps: Obtaining device parameters of the receiving end, wherein the device parameters include screen resolution and color mode; Compare the resolution of the decrypted image data with the screen resolution of the receiving end to determine whether the resolution of the decrypted image data is higher than the screen resolution of the receiving end; If it is higher, the decrypted image data is scaled based on a bilinear interpolation algorithm to obtain preliminary image data; If it is not higher, the decrypted image data is processed by a preset super-resolution reconstruction model to obtain preliminary image data; the super-resolution reconstruction model is a machine learning model of a generative adversarial network architecture that uses high- and low-resolution medical images as sample data for deep learning training; The preliminary image data is input into a preset parameter optimization model for contrast and brightness optimization adjustment and color precision correction, and then displayed on the receiving end; the parameter optimization model is a convolutional neural network model obtained by iterative training of medical image data of different types and under different lighting conditions.

Images