Image report push method, device and computing equipment based on RPA and AI

Abstract

The present invention discloses a method, device and computing equipment for pushing image reports based on RPA and AI. The method includes: using a preset lung segmentation model to segment the lung medical image to be detected sent by the RPA robot to obtain a target medical image containing only the lung area, using a preset lung nodule detection model to detect the target medical image to obtain attribute information of each suspected lung nodule, using a preset lung nodule segmentation model to segment the area where each suspected lung nodule is located to obtain three-dimensional contour information of each suspected lung nodule; generating a suspected lung nodule image report based on the attribute information and three-dimensional contour information of each suspected lung nodule and sending it to the hospital platform through the RPA robot. In this way, the lung nodules are detected by AI image analysis technology to obtain a suspected lung nodule image report, and the report is sent to the hospital platform through the RPA robot, which reduces the time doctors spend identifying lung nodules and improves efficiency.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and more specifically, to a method, apparatus, and computing device for pushing image reports based on RPA and AI. Background Art

[0002] RPA (Robotic Process Automation) software robots can simulate human operations on the mouse and keyboard on the computer to perform automated office work like real people. They can work 24 hours a day, automatically executing processes or a series of tasks according to rules, freeing users from repetitive and tedious work.

[0003] AI (Artificial Intelligence) is a technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence.

[0004] Currently, when doctors perform lung cancer tests, they need to examine lung medical images containing lung nodules to identify them. Doctors typically rely on experience when identifying lung nodules, which is time-consuming and inefficient. Summary of the Invention

[0005] This invention provides a method, device, and computing device for pushing imaging reports based on RPA and AI to address the low efficiency of manual lung nodule detection. The specific technical solution is as follows.

[0006] In a first aspect, the present invention provides a method for pushing image reports based on RPA and AI, which is applied to an image analysis platform integrated with a Robotic Process Automation (RPA) robot. The method includes:

[0007] S1. Receive a lung medical image to be detected sent by the RPA robot;

[0008] S2. Segment the lung medical image to be detected using a preset lung segmentation model to obtain a target medical image containing only the lung area, detect the target medical image using a preset lung nodule detection model to obtain attribute information of each suspected lung nodule in the target medical image, segment the area where each suspected lung nodule is located using the preset lung nodule segmentation model to obtain three-dimensional contour information of each suspected lung nodule, wherein the preset lung nodule detection model is trained by lung nodule sample images and lung nodule attribute information in a preset lung nodule image sample set to establish an association relationship between each lung nodule sample image and the lung nodule attribute information;

[0009] S3. Generate a suspected lung nodule imaging report based on the attribute information and three-dimensional contour information of each suspected lung nodule, and send the suspected lung nodule imaging report to the hospital platform through the RPA robot.

[0010] Optionally, the step S2 of segmenting the lung medical image to be detected using a preset lung segmentation model to obtain a target medical image containing only the lung area specifically includes:

[0011] S21. Filtering the radiomics features of the lung region from the lung medical image to be detected using a filtered back projection reconstruction algorithm FBP and a sinogram-determined iterative reconstruction algorithm SAFIRE, reconstructing the radiomics features to obtain a target medical image containing only the lung region.

[0012] Optionally, the preset lung nodule detection model in step S2 is obtained by:

[0013] S22, obtaining lung nodule sample images from a preset lung nodule image sample set and attribute information of the lung nodules contained in the lung nodule sample images;

[0014] S23, inputting the lung nodule sample image and the attribute information of the lung nodules contained in the lung nodule sample image into an initial convolutional neural network for feature extraction, to obtain reference attribute information of the lung nodules contained in the lung nodule sample image;

[0015] S24, calculating a difference value between the reference attribute information and the attribute information of the lung nodules included in the lung nodule sample image;

[0016] S25, optimizing the parameters of the initial convolutional neural network based on the difference value, and returning to step S22;

[0017] S26. When the number of iterations reaches a preset number, the training is completed, and a preset lung nodule detection model is obtained that associates the lung nodule sample image with the attribute information of the lung nodule.

[0018] Optionally, the preset lung nodule segmentation model in step S2 is a three-dimensional semantic segmentation convolutional neural network 3DU-Net model.

[0019] Optionally, the attribute information of each suspected pulmonary nodule includes pulmonary nodule location, pulmonary nodule size, and pulmonary nodule density.

[0020] Optionally, step S3 specifically includes:

[0021] S31, marking each suspected pulmonary nodule in the to-be-detected lung medical image according to the pulmonary nodule location in the attribute information of each suspected pulmonary nodule;

[0022] S32. Generate a density histogram based on the lung nodule density in the attribute information of each suspected lung nodule;

[0023] S33. Generate a suspected lung nodule image report including the labeled lung medical image to be detected, the density histogram, and the size of each suspected lung nodule, and send the suspected lung nodule image report to the hospital platform through the RPA robot.

[0024] Optionally, the lung medical image to be detected is recognized by using an optical character recognition (OCR) method to obtain image content.

[0025] In a second aspect, the present invention provides an image report push device based on RPA and AI, which is applied to an image analysis platform integrated with a Robotic Process Automation (RPA) robot. The device includes:

[0026] A receiving module, configured to receive the lung medical image to be detected sent by the RPA robot;

[0027] a segmentation module for segmenting the lung medical image to be detected using a preset lung segmentation model to obtain a target medical image containing only the lung area, detecting the target medical image using a preset lung nodule detection model to obtain attribute information of each suspected lung nodule in the target medical image, and segmenting the area where each suspected lung nodule is located using the preset lung nodule segmentation model to obtain three-dimensional contour information of each suspected lung nodule, wherein the preset lung nodule detection model is trained using lung nodule sample images and lung nodule attribute information in a preset lung nodule image sample set to establish an association relationship between each lung nodule sample image and the lung nodule attribute information;

[0028] The imaging report push module is used to generate an imaging report of suspected lung nodules based on the attribute information and three-dimensional contour information of each suspected lung nodule, and send the imaging report of suspected lung nodules to the hospital platform through the RPA robot.

[0029] Optionally, the segmentation module is specifically configured to:

[0030] The radiomics features of the lung area are screened out from the lung medical image to be detected using the filtered back projection reconstruction algorithm FBP and the sinusoidal graph determined iterative reconstruction algorithm SAFIRE, and the radiomics features are reconstructed to obtain a target medical image containing only the lung area.

[0031] Optionally, the RPA and AI-based imaging report push device further includes a model training module, which is used to train a preset lung nodule detection model. The model training module includes:

[0032] An acquisition submodule, configured to acquire a lung nodule sample image from a preset lung nodule image sample set and attribute information of the lung nodules contained in the lung nodule sample image;

[0033] A sample input submodule is used to input the lung nodule sample image and the attribute information of the lung nodules contained in the lung nodule sample image into the initial convolutional neural network for feature extraction, so as to obtain the reference attribute information of the lung nodules contained in the lung nodule sample image;

[0034] a calculation submodule, configured to calculate a difference value between the reference attribute information and the attribute information of the lung nodules contained in the lung nodule sample image;

[0035] A parameter optimization submodule, configured to optimize the parameters of the initial convolutional neural network based on the difference value, and trigger the acquisition submodule;

[0036] The training completion submodule is used to complete the training when the number of iterations reaches a preset number, and obtain a preset lung nodule detection model that associates the lung nodule sample image with the attribute information of the lung nodule.

[0037] Optionally, the preset lung nodule segmentation model in the segmentation module is a three-dimensional semantic segmentation convolutional neural network 3DU-Net model.

[0038] Optionally, the attribute information of each suspected pulmonary nodule includes pulmonary nodule location, pulmonary nodule size, and pulmonary nodule density.

[0039] Optionally, the imaging report push module specifically includes:

[0040] a labeling submodule, configured to label each suspected pulmonary nodule in the lung medical image to be detected according to the location of the pulmonary nodule in the attribute information of the suspected pulmonary nodule;

[0041] A density histogram generation submodule is used to generate a density histogram based on the lung nodule density in the attribute information of each suspected lung nodule;

[0042] The push submodule is used to generate a suspected lung nodule image report including the labeled lung medical image to be detected, the density histogram and the size of each suspected lung nodule, and send the suspected lung nodule image report to the hospital platform through the RPA robot.

[0043] Optionally, the lung medical image to be detected is recognized by using an optical character recognition (OCR) method to obtain image content.

[0044] In a third aspect, an embodiment of the present invention further provides a computing device, comprising a storage device and a processor, wherein the storage device is used to store a computer program, and the processor runs the computer program to enable the computing device to execute any of the above-mentioned RPA and AI-based image report push methods.

[0045] In a fourth aspect, an embodiment of the present invention further provides a storage medium on which a computer program is stored, which, when executed by a processor, implements any of the above-mentioned RPA and AI-based image report push methods.

[0046] From the above content, it can be seen that the RPA and AI-based image report push method, device and computing device provided in the embodiments of the present invention can receive the lung medical image to be detected sent by the RPA robot; use the preset lung segmentation model to segment the lung medical image to be detected to obtain a target medical image containing only the lung area, use the preset lung nodule detection model to detect the target medical image to obtain the attribute information of each suspected lung nodule in the target medical image, use the preset lung nodule segmentation model to segment the area where each suspected lung nodule is located to obtain the three-dimensional contour information of each suspected lung nodule, wherein the preset lung nodule detection model is trained by the lung nodule sample images and the attribute information of the lung nodules in the preset lung nodule image sample set to establish the association relationship between each lung nodule sample image and the attribute information of the lung nodule; generate a suspected lung nodule image report according to the attribute information and three-dimensional contour information of each suspected lung nodule, and send the suspected lung nodule image report to the hospital platform through the RPA robot. The technical solution provided in this embodiment automatically detects lung nodules through the image analysis technology of the artificial intelligence AI possessed by the image analysis platform to obtain the attribute information and three-dimensional contour information of each suspected lung nodule, and then generates a suspected lung nodule imaging report based on the attribute information and three-dimensional contour information of each suspected lung nodule, and automatically sends the suspected lung nodule imaging report to the hospital platform through the RPA robot so that the doctor can confirm whether each suspected lung nodule is a true lung nodule. Since the suspected lung nodule imaging report contains the attribute information and three-dimensional contour information of each suspected lung nodule, the doctor can quickly confirm whether each suspected lung nodule is a true lung nodule based on the suspected lung nodule imaging report, reducing the time spent by the doctor to identify lung nodules and improving efficiency. Of course, implementing any product or method of the present invention does not necessarily require achieving all the advantages described above at the same time.

[0047] The innovative features of the embodiments of the present invention include:

[0048] 1. The image analysis platform uses artificial intelligence (AI) to automatically detect lung nodules and obtain the attribute information and three-dimensional contour information of each suspected lung nodule. Then, a suspected lung nodule imaging report is generated based on the attribute information and three-dimensional contour information of each suspected lung nodule. The suspected lung nodule imaging report is automatically sent to the hospital platform through the RPA robot so that doctors can confirm whether each suspected lung nodule is a true lung nodule. Since the suspected lung nodule imaging report contains the attribute information and three-dimensional contour information of each suspected lung nodule, doctors can quickly confirm whether each suspected lung nodule is a true lung nodule based solely on the suspected lung nodule imaging report, reducing the time doctors spend identifying lung nodules and improving efficiency.

[0049] 2. The RPA robot automatically obtains the lung medical images to be tested, and automatically analyzes the lung medical images to be tested using the artificial intelligence (AI) image analysis technology of the image analysis platform to generate an image report of suspected lung nodules. The RPA robot then automatically sends the image report of suspected lung nodules to the hospital platform, eliminating the need for human intervention throughout the process and improving work efficiency.

[0050] 3. By screening out only the imaging genomics features of the lung area from the lung medical image to be detected, and then reconstructing the imaging genomics features, a target medical image containing only the lung area is obtained, avoiding the interference of areas outside the lung on subsequent lung nodule detection.

[0051] 4. By training the initial convolutional neural network, a preset lung nodule detection model can be obtained that associates the lung nodule sample image with the attribute information of the lung nodule. Suspected lung nodules can be detected through the preset lung nodule detection model.

[0052] 5. The image analysis platform facilitates doctors' review and improves efficiency by marking each suspected lung nodule in the lung medical images to be tested.

[0053] 6. Since the generated suspected lung nodule imaging report contains the labeled lung medical imaging image to be detected, density histogram and the size of each suspected lung nodule, doctors can quickly obtain relevant information about each suspected lung nodule based on the suspected lung nodule imaging report, greatly improving the doctor's work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely some embodiments of the present invention. Those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0055] Figure 1 A schematic diagram of a process for pushing imaging reports provided by the present invention;

[0056] Figure 2 This is a flowchart of a method for pushing image reports based on RPA and AI, provided in Example 1 of the present invention;

[0057] Figure 3 This is a module diagram of an RPA- and AI-based imaging report push device provided in Example 2 of the present invention;

[0058] Figure 4 This is a structural diagram of a computing device provided in Example 3 of the present invention. DETAILED DESCRIPTION

[0059] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] It should be noted that the terms "including" and "having" and any variations thereof in the embodiments of the present invention and the accompanying drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to the process, method, product, or apparatus.

[0061] In the description of the embodiments of the present invention, the term "medical image of the lung to be detected" refers to an image of lung tissue obtained from the human lung in a non-invasive manner and requiring further detection of whether there are lung nodules. The image not only includes the lung area, but also includes images of surrounding lung tissue collected during the non-invasive acquisition. For example, the medical image of the lung to be detected may be a CT (Computed Tomography) image of the lung to be detected.

[0062] In the description of the embodiment of the present invention, the term "target medical image" refers to a medical image that only contains the lung area and is segmented from the lung medical image to be detected.

[0063] In the description of the embodiment of the present invention, the term "suspected lung nodule" refers to a lung nodule detected by the preset lung nodule detection model provided by the embodiment of the present invention. Since whether it is a real lung nodule still needs to be judged by the doctor based on medical experience, in the embodiment of the present invention, the lung nodule detected by the preset lung nodule detection model is named as suspected lung nodule.

[0064] In the description of the embodiments of the present invention, the term "image analysis platform" refers to a platform with artificial intelligence (AI) image analysis technology that is integrated with a robotic process automation (RPA) robot. The platform includes a preset lung segmentation model, a preset lung nodule detection model, and a preset lung nodule segmentation model.

[0065] In the description of the embodiments of the present invention, the term "preset lung segmentation model" is a model used to segment the lung medical image to be detected to obtain a target medical image containing only the lung area to avoid interference from areas outside the lungs in subsequent lung nodule detection.

[0066] In the description of the embodiments of the present invention, the term "preset lung nodule detection model" refers to a model for detecting lung nodules, and the output of the model is attribute information of each suspected lung nodule.

[0067] In the description of the embodiment of the present invention, the term "preset lung nodule segmentation model" is a model used to segment each suspected lung nodule separately from the target medical image, thereby obtaining three-dimensional contour information of each suspected lung nodule.

[0068] In the description of the embodiments of the present invention, the term "three-dimensional contour information" refers to information describing the three-dimensional appearance of an object, and may include three-dimensional shape information and three-dimensional edge information of the object.

[0069] In the description of the embodiment of the present invention, the term "hospital platform" refers to an information platform used by the hospital to integrate hospital resources and realize the hospital's human, financial, and material management and electronic diagnosis and treatment.

[0070] In the description of the embodiments of the present invention, the term "long diameter and short diameter" is a general term for the longest diameter and the shortest diameter of a pulmonary nodule.

[0071] In the description of the embodiments of the present invention, the term "density histogram" refers to a graph that represents the distribution of density data by a series of vertical stripes or line segments of varying heights. Generally, the horizontal axis represents the data type, and the vertical axis represents the density distribution. For example, in the embodiments of the present invention, the horizontal axis represents each suspected pulmonary nodule, and the vertical axis represents the pulmonary nodule density of each suspected pulmonary nodule.

[0072] In the description of the embodiments of the present invention, the term "radiological features" refers to the features used to describe lesions in medical imaging, which may include lesion diameter, shape, edge, density, cavity and calcification.

[0073] In the description of the embodiment of the present invention, the term "attribute information of lung nodules" is a feature used to describe the characteristics of lung nodules, which may include the location of lung nodules, the size of lung nodules, and the density of lung nodules.

[0074] In order to explain the contents of each embodiment of the present invention more clearly and clearly, the basic working principles of the embodiments of the present invention are briefly introduced below.

[0075] Robotic Process Automation (RPA) uses specific "robot software" to simulate human operations on computers and automatically execute process tasks according to rules.

[0076] AI (Artificial Intelligence) is the abbreviation of artificial intelligence. It is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence.

[0077] In related businesses, when doctors perform lung cancer tests, they need to observe medical images of the lungs containing lung nodules to determine the size of the lung nodules. When doctors identify the size of lung nodules, they usually rely on experience, which results in a long time and low efficiency. In order to solve the above problems, the inventors of this application proposed an image report push method based on RPA and AI. An image analysis platform integrated with a robotic process automation (RPA) robot automatically detects lung nodules and generates an image report of suspected lung nodules. The RPA robot automatically sends the image report of suspected lung nodules to the hospital platform so that doctors can make the final test results, thereby improving efficiency.

[0078] The following combination Figure 1 , for the application scenario of image report push, a brief introduction to the image analysis platform and RPA operations is given.

[0079] See also Figure 1 , the RPA robot extracts the lung medical imaging images to be tested from the hospital platform Figure 1 The image a in the image is extracted and sent to the image analysis platform. The image analysis platform receives the lung medical image to be tested sent by the RPA robot. The image analysis platform then uses the preset lung segmentation model to segment the lung medical image to be tested, and obtains the target medical image containing only the lung area, i.e. Figure 1 The b-image in the image is used to avoid interference of areas outside the lung on subsequent lung nodule detection.

[0080] See also Figure 1The image c in the image, the image analysis platform then uses the preset lung nodule detection model to detect the target medical image, and obtains the attribute information of each suspected lung nodule in the target medical image, wherein the attribute information of each suspected lung nodule includes multiple types. In one implementation, the attribute information of each suspected lung nodule includes the location of the lung nodule and the size of the lung nodule, that is, Figure 1 The position and diameter of the suspected nodule in the embodiment of the present invention are displayed by the diameter.

[0081] If the location and diameter of the suspected pulmonary nodule can be output after the test, the nature of the test result can be determined as a suspected pulmonary nodule. Figure 1 The output diameter of the suspected pulmonary nodule can include the long and short diameters, i.e. Figure 1 Calculate the major and minor diameters in D.

[0082] See also Figure 1 The image analysis platform uses the preset lung nodule segmentation model to segment the area where each suspected lung nodule is located, and obtains the three-dimensional contour information of each suspected lung nodule, that is, extracts the contour of each suspected lung nodule and displays it in a three-dimensional form. Figure 1 Nodule contour extraction in D in , where the rectangular box in the d image is a suspected lung nodule. The three-dimensional shape of the suspected lung nodule can be seen by rotating the rectangular box with the mouse.

[0083] Since the attribute information of each suspected pulmonary nodule may also include the pulmonary nodule density, a density histogram may also be generated based on the pulmonary nodule density in the attribute information of each suspected pulmonary nodule. Figure 1 The density histogram in D in is used to mark each suspected lung nodule in the lung medical image to be detected according to the lung nodule position in the attribute information of each suspected lung nodule, and a suspected lung nodule image report and a JSON (JavaScript Object Notation) result are generated, which include the marked lung medical image to be detected, the density histogram and the lung nodule size of each suspected lung nodule. The suspected lung nodule image report and the JSON result contain the same content, except that the JSON result is a non-visual result generated in the background of the image analysis platform, while the suspected lung nodule image report is a visual result generated in the foreground of the image analysis platform.

[0084] The RPA robot obtains the suspected lung nodule imaging report and JSON results and sends them to the hospital platform. Figure 1 Through the regional health platform in the hospital, doctors can view the imaging report of the suspected lung nodule and then make the final test results.

[0085] In the above process, the image analysis platform integrated with the Robotic Process Automation (RPA) robot automatically detects lung nodules and generates a suspected lung nodule imaging report. The RPA robot automatically sends the suspected lung nodule imaging report to the hospital platform so that doctors can confirm whether each suspected lung nodule is a true lung nodule. Since the suspected lung nodule imaging report contains the attribute information and three-dimensional contour information of each suspected lung nodule, doctors can quickly confirm whether each suspected lung nodule is a true lung nodule based solely on the suspected lung nodule imaging report, which saves time and improves efficiency.

[0086] The following is a detailed introduction to the process of pushing image reports based on RPA and AI from the perspective of the image analysis platform.

[0087] Example 1

[0088] Figure 2 This is a flowchart of a method for pushing image reports based on RPA and AI provided by the first embodiment of the present invention. The execution subject of this method is an image analysis platform integrated with a robotic process automation (RPA) robot. Figure 2 As shown, the method may include:

[0089] S1. Receive the lung medical image to be tested sent by the RPA robot.

[0090] When a doctor performs a lung cancer test, the RPA robot extracts the medical images of the lungs to be tested from the hospital platform, and then sends the extracted medical images of the lungs to be tested to the image analysis platform. The image analysis platform receives the medical images of the lungs to be tested sent by the RPA robot.

[0091] S2. Use a preset lung segmentation model to segment the lung medical image to be detected to obtain a target medical image that only contains the lung area, use a preset lung nodule detection model to detect the target medical image to obtain attribute information of each suspected lung nodule in the target medical image, use the preset lung nodule segmentation model to segment the area where each suspected lung nodule is located to obtain three-dimensional contour information of each suspected lung nodule, wherein the preset lung nodule detection model is trained by lung nodule sample images and lung nodule attribute information in a preset lung nodule image sample set to establish a correlation between each lung nodule sample image and the attribute information of the lung nodule.

[0092] Since the lung medical image to be tested does not only contain the lung area but may also contain other human body areas, in order to facilitate the detection of lung nodules, the lung area needs to be detected separately. Since the image analysis platform contains a preset lung segmentation model, a preset lung nodule detection model and a preset lung nodule segmentation model, after the image analysis platform receives the lung medical image to be tested sent by the RPA robot, it can use the preset lung segmentation model to segment the lung medical image to obtain the target medical image containing only the lung area.

[0093] In a specific embodiment, step S2 of segmenting the lung medical image to be detected using a preset lung segmentation model to obtain a target medical image containing only the lung region may specifically include:

[0094] S21. Filtered back projection reconstruction algorithm FBP and sinogram-determined iterative reconstruction algorithm SAFIRE are used to screen out radiomic features of the lung region from the lung medical image to be detected, and the radiomic features are reconstructed to obtain a target medical image containing only the lung region.

[0095] In an embodiment of the present invention, the preset lung segmentation model is a model that includes a filtered back projection reconstruction algorithm (FBP) and a sinusoidal graph determined iterative reconstruction algorithm (SAFIRE). Therefore, when segmenting a lung medical image to be detected, the filtered back projection reconstruction algorithm (FBP) and the sinusoidal graph determined iterative reconstruction algorithm (SAFIRE) can be used to filter out radiomic features of the lung region from the lung medical image to be detected, and then the radiomic features are reconstructed to obtain a target medical image containing only the lung region. Exemplarily, the number of radiomic features of the lung region that are filtered out is 11.

[0096] Therefore, by screening out only the imaging genomics features of the lung area from the lung medical image to be detected and then reconstructing the imaging genomics features, a target medical image containing only the lung area is obtained, thereby avoiding interference of areas outside the lung on subsequent lung nodule detection.

[0097] Among them, using the filtered back projection reconstruction algorithm FBP and the sinusoidal graph determination iterative reconstruction algorithm SAFIRE to screen out the imaging omics features of the lung area from the lung medical image to be detected can include: using the optical character recognition OCR method to identify the lung medical image to be detected to obtain the image content, and using the filtered back projection reconstruction algorithm FBP and the sinusoidal graph determination iterative reconstruction algorithm SAFIRE to screen out the imaging omics features of the lung area from the image content.

[0098] After obtaining a target medical image containing only the lung area, the preset lung nodule detection model can be used to detect the target medical image to obtain the attribute information of each suspected lung nodule in the target medical image, wherein the preset lung nodule detection model is trained by the lung nodule sample images and the attribute information of the lung nodules in the preset lung nodule image sample set to establish the association relationship between each lung nodule sample image and the attribute information of the lung nodule.

[0099] For example, the attribute information of each suspected pulmonary nodule may include pulmonary nodule location, pulmonary nodule size, and pulmonary nodule density. The pulmonary nodule size is usually represented by the diameter of the pulmonary nodule, which may include the major and minor axes.

[0100] In a specific embodiment, the preset lung nodule detection model in step S2 can be obtained by:

[0101] S22, obtaining lung nodule sample images in a preset lung nodule image sample set and attribute information of the lung nodules contained in the lung nodule sample images;

[0102] S23, inputting the lung nodule sample image and the attribute information of the lung nodules contained in the lung nodule sample image into the initial convolutional neural network for feature extraction, thereby obtaining reference attribute information of the lung nodules contained in the lung nodule sample image;

[0103] S24, calculating a difference value between the reference attribute information and the attribute information of the lung nodule contained in the lung nodule sample image;

[0104] S25. Optimize the parameters of the initial convolutional neural network based on the difference value, and return to step S22;

[0105] S26. When the number of iterations reaches a preset number, the training is completed, and a preset lung nodule detection model is obtained that associates the lung nodule sample image with the attribute information of the lung nodule.

[0106] In order to train a model that can detect lung nodules, various images containing lung nodules can be collected, and the collected images can be stored as lung nodule sample images in a preset lung nodule image sample set, wherein the preset lung nodule image sample set contains various lung nodule sample images with different attribute information.

[0107] When training the model, it is necessary to obtain the lung nodule sample images in the preset lung nodule image sample set and the attribute information of the lung nodules contained in the lung nodule sample images.

[0108] After obtaining the lung nodule sample images and the attribute information of the lung nodules contained in the lung nodule sample images in the preset lung nodule image sample set, the lung nodule sample images and the attribute information of the lung nodules contained in the lung nodule sample images are input into the initial convolutional neural network for feature extraction to obtain the reference attribute information of the lung nodules contained in the lung nodule sample images.

[0109] When the number of iterations does not reach the preset number, it means that the network at this time has not yet adapted to most of the lung nodule sample images. At this time, it is necessary to calculate the difference value between the reference attribute information and the attribute information of the lung nodules contained in the lung nodule sample image, and then optimize the parameters of the initial convolutional neural network based on the difference value, and return to execute the step of obtaining the lung nodule sample images in the preset lung nodule image sample set and the attribute information of the lung nodules contained in the lung nodule sample images.

[0110] During training, all sample lung nodule images can be looped through, and the parameters of the initial convolutional neural network can be continuously adjusted. When the number of iterations reaches the preset number, it indicates that the network can adapt to the majority of lung nodule sample images and obtain accurate results. At this point, training is considered complete, and a preset lung nodule detection model is obtained that associates the lung nodule sample images with lung nodule attribute information.

[0111] It can be seen that by training the initial convolutional neural network through the above training method, a preset lung nodule detection model can be obtained that associates the lung nodule sample image with the attribute information of the lung nodule.

[0112] For example, the lung nodule sample images used in training contain attribute information of lung nodules in which the size of the lung nodules is less than 5 mm, so that the preset lung nodule detection model can detect suspected lung nodules with a size of less than 5 mm.

[0113] Since the target medical image is detected using a preset lung nodule detection model, suspected lung nodules are obtained. Whether each suspected lung nodule is a real lung nodule still requires a doctor to make a professional judgment to obtain the detection result. In order to facilitate subsequent doctors to view the three-dimensional contours of each suspected lung nodule, after obtaining the attribute information of each suspected lung nodule in the target medical image, it is also necessary to use the preset lung nodule segmentation model to segment the area where each suspected lung nodule is located to obtain the three-dimensional contour information of each suspected lung nodule.

[0114] Exemplarily, the preset lung nodule segmentation model can be a three-dimensional semantic segmentation convolutional neural network 3DU-Net model.

[0115] S3. Generate a suspected lung nodule imaging report based on the attribute information and three-dimensional contour information of each suspected lung nodule, and send the suspected lung nodule imaging report to the hospital platform through the RPA robot.

[0116] To help doctors determine whether each suspected lung nodule is a true lung nodule, after obtaining the attribute information and three-dimensional contour information of each suspected lung nodule, a suspected lung nodule imaging report is generated based on the attribute information and three-dimensional contour information of each suspected lung nodule, and the suspected lung nodule imaging report is sent to the hospital platform through the RPA robot.

[0117] In one implementation, when the attribute information of each suspected pulmonary nodule includes pulmonary nodule location, pulmonary nodule size, and pulmonary nodule density, step S3 may specifically include:

[0118] S31, marking each suspected pulmonary nodule in the lung medical image to be detected according to the pulmonary nodule location in the attribute information of each suspected pulmonary nodule;

[0119] S32. Generate a density histogram based on the lung nodule density in the attribute information of each suspected lung nodule;

[0120] S33. Generate a suspected lung nodule image report including the labeled lung medical image to be detected, a density histogram, and the size of each suspected lung nodule, and send the suspected lung nodule image report to the hospital platform through the RPA robot.

[0121] In order to facilitate doctors to check the location of each suspected lung nodule, the image analysis platform marks each suspected lung nodule in the lung medical image to be detected according to the lung nodule location in the attribute information of each suspected lung nodule. The method of marking each suspected lung nodule in the lung medical image to be detected can be: using a rectangular frame to frame each suspected lung nodule, of course, it can also be any other prominent marking method, and the embodiment of the present invention does not impose any limitation on this.

[0122] Therefore, the image analysis platform facilitates doctors' review and improves efficiency by marking each suspected lung nodule in the lung medical image to be tested.

[0123] To facilitate doctors in checking the density of each suspected lung nodule, the image analysis platform generates a density histogram based on the lung nodule density in the attribute information of each suspected lung nodule, so that doctors can check the density of all suspected lung nodules through only one density histogram.

[0124] After labeling each suspected lung nodule and generating a density histogram, the labeled lung medical image to be detected, the density histogram, and the size of each suspected lung nodule are integrated to generate a suspected lung nodule image report containing the labeled lung medical image to be detected, the density histogram, and the size of each suspected lung nodule. The suspected lung nodule image report is then sent to the hospital platform via the RPA robot.

[0125] Since the generated suspected lung nodule imaging report contains the labeled lung medical imaging image to be detected, the density histogram and the size of each suspected lung nodule, the doctor can quickly obtain relevant information about each suspected lung nodule based on the suspected lung nodule imaging report, which greatly improves the doctor's work efficiency.

[0126] As can be seen from the above content, this embodiment can receive the lung medical image to be detected sent by the RPA robot; use the preset lung segmentation model to segment the lung medical image to be detected to obtain a target medical image containing only the lung area, use the preset lung nodule detection model to detect the target medical image to obtain the attribute information of each suspected lung nodule in the target medical image, use the preset lung nodule segmentation model to segment the area where each suspected lung nodule is located to obtain the three-dimensional contour information of each suspected lung nodule, wherein the preset lung nodule detection model is trained by the lung nodule sample images and the attribute information of the lung nodules in the preset lung nodule image sample set to establish the association relationship between each lung nodule sample image and the attribute information of the lung nodule; generate a suspected lung nodule image report based on the attribute information and three-dimensional contour information of each suspected lung nodule, and send the suspected lung nodule image report to the hospital platform through the RPA robot. The technical solution provided in this embodiment automatically detects lung nodules through the image analysis technology of the artificial intelligence AI possessed by the image analysis platform to obtain the attribute information and three-dimensional contour information of each suspected lung nodule, and then generates a suspected lung nodule imaging report based on the attribute information and three-dimensional contour information of each suspected lung nodule, and automatically sends the suspected lung nodule imaging report to the hospital platform through the RPA robot so that doctors can confirm whether each suspected lung nodule is a true lung nodule. Since the suspected lung nodule imaging report contains the attribute information and three-dimensional contour information of each suspected lung nodule, doctors can quickly confirm whether each suspected lung nodule is a true lung nodule based only on the suspected lung nodule imaging report, reducing the time spent by doctors to identify lung nodules and improving efficiency.

[0127] In addition, the RPA robot automatically obtains the medical images of the lungs to be tested, and automatically analyzes the medical images of the lungs to be tested through the artificial intelligence AI image analysis technology of the image analysis platform to generate an image report of suspected lung nodules. The image report of suspected lung nodules is then automatically sent to the hospital platform through the RPA robot. No human intervention is required throughout the process, which improves work efficiency.

[0128] Example 2

[0129] Figure 3 The figure shows a module diagram of an image report push device based on RPA and AI. Figure 3As shown, an embodiment of this specification provides an image report push device based on RPA and AI, which is applied to an image analysis platform integrated with a robotic process automation (RPA) robot. The device may include:

[0130] A receiving module 310 is configured to receive the lung medical image to be detected sent by the RPA robot;

[0131] a segmentation module 320 for segmenting the lung medical image to be detected using a preset lung segmentation model to obtain a target medical image containing only the lung area, detecting the target medical image using a preset lung nodule detection model to obtain attribute information of each suspected lung nodule in the target medical image, and segmenting the area where each suspected lung nodule is located using the preset lung nodule segmentation model to obtain three-dimensional contour information of each suspected lung nodule, wherein the preset lung nodule detection model is trained using lung nodule sample images and lung nodule attribute information in a preset lung nodule image sample set to establish an association relationship between each lung nodule sample image and the lung nodule attribute information;

[0132] The imaging report push module 330 is used to generate an imaging report of suspected lung nodules based on the attribute information and three-dimensional contour information of each suspected lung nodule, and send the imaging report of suspected lung nodules to the hospital platform through the RPA robot.

[0133] The RPA and AI-based image report push device provided in an embodiment of the present invention can receive a lung medical image to be detected sent by an RPA robot; use a preset lung segmentation model to segment the lung medical image to be detected to obtain a target medical image containing only the lung area, use a preset lung nodule detection model to detect the target medical image to obtain attribute information of each suspected lung nodule in the target medical image, use a preset lung nodule segmentation model to segment the area where each suspected lung nodule is located to obtain three-dimensional contour information of each suspected lung nodule, wherein the preset lung nodule detection model is trained by using lung nodule sample images and lung nodule attribute information in a preset lung nodule image sample set to establish an association between each lung nodule sample image and the attribute information of the lung nodule; generate a suspected lung nodule image report based on the attribute information and three-dimensional contour information of each suspected lung nodule, and send the suspected lung nodule image report to the hospital platform through the RPA robot. The technical solution provided in this embodiment automatically detects lung nodules through the image analysis technology of the artificial intelligence AI possessed by the image analysis platform to obtain the attribute information and three-dimensional contour information of each suspected lung nodule, and then generates a suspected lung nodule imaging report based on the attribute information and three-dimensional contour information of each suspected lung nodule, and automatically sends the suspected lung nodule imaging report to the hospital platform through the RPA robot so that doctors can confirm whether each suspected lung nodule is a true lung nodule. Since the suspected lung nodule imaging report contains the attribute information and three-dimensional contour information of each suspected lung nodule, doctors can quickly confirm whether each suspected lung nodule is a true lung nodule based only on the suspected lung nodule imaging report, reducing the time spent by doctors to identify lung nodules and improving efficiency.

[0134] In one implementation, the segmentation module 320 may be specifically configured to:

[0135] The radiomics features of the lung area are screened out from the lung medical image to be detected using the filtered back projection reconstruction algorithm FBP and the sinusoidal graph determined iterative reconstruction algorithm SAFIRE, and the radiomics features are reconstructed to obtain a target medical image containing only the lung area.

[0136] In one implementation, the RPA and AI-based imaging report push device further includes a model training module, which is used to train a preset lung nodule detection model. The model training module includes:

[0137] An acquisition submodule, configured to acquire a lung nodule sample image from a preset lung nodule image sample set and attribute information of the lung nodules contained in the lung nodule sample image;

[0138] A sample input submodule is used to input the lung nodule sample image and the attribute information of the lung nodules contained in the lung nodule sample image into the initial convolutional neural network for feature extraction, so as to obtain the reference attribute information of the lung nodules contained in the lung nodule sample image;

[0139] a calculation submodule, configured to calculate a difference value between the reference attribute information and the attribute information of the lung nodules contained in the lung nodule sample image;

[0140] A parameter optimization submodule, configured to optimize the parameters of the initial convolutional neural network based on the difference value, and trigger the acquisition submodule;

[0141] The training completion submodule is used to complete the training when the number of iterations reaches a preset number, and obtain a preset lung nodule detection model that associates the lung nodule sample image with the attribute information of the lung nodule.

[0142] In one implementation, the preset lung nodule segmentation model in the segmentation module 320 is a three-dimensional semantic segmentation convolutional neural network 3DU-Net model.

[0143] In one implementation, the attribute information of each suspected pulmonary nodule includes pulmonary nodule location, pulmonary nodule size, and pulmonary nodule density.

[0144] In one implementation, the image report pushing module 330 may specifically include:

[0145] a labeling submodule, configured to label each suspected pulmonary nodule in the lung medical image to be detected according to the location of the pulmonary nodule in the attribute information of the suspected pulmonary nodule;

[0146] A density histogram generation submodule is used to generate a density histogram based on the lung nodule density in the attribute information of each suspected lung nodule;

[0147] The push submodule is used to generate a suspected lung nodule image report including the labeled lung medical image to be detected, the density histogram and the size of each suspected lung nodule, and send the suspected lung nodule image report to the hospital platform through the RPA robot.

[0148] In one implementation, the optical character recognition (OCR) method is used to recognize the lung medical image to be detected to obtain the image content.

[0149] Example 3

[0150] See also Figure 4 , Figure 4 This is a schematic diagram of the structure of a computing device provided by the third embodiment of the present invention. Figure 4 As shown, the computing device may include:

[0151] A memory 401 storing executable program code;

[0152] a processor 402 coupled to the memory 401;

[0153] The processor 402 calls the executable program code stored in the memory 401 to execute the RPA and AI-based image report push method provided in any embodiment of the present invention.

[0154] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the RPA and AI-based image report push method provided in any embodiment of the present invention.

[0155] In various embodiments of the present invention, it should be understood that the size of the serial numbers of the above-mentioned processes does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0156] In the embodiments provided herein, it should be understood that "B corresponding to A" means that B is associated with A and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B based solely on A; B can also be determined based on A and / or other information.

[0157] In addition, the functional units in the embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0158] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible memory. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several requests for causing a computer device (which can be a personal computer, server, or network device, specifically a processor in the computer device) to execute some or all of the steps of the above-mentioned methods of various embodiments of the present invention.

[0159] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program. The program can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0160] Those skilled in the art will appreciate that the accompanying drawings are merely schematic diagrams of an embodiment, and the modules or processes in the accompanying drawings are not necessarily required to implement the present invention.

[0161] Those skilled in the art will appreciate that the modules in the apparatuses of the embodiments may be distributed in the apparatuses of the embodiments as described in the embodiments, or may be located in one or more apparatuses different from the embodiments with corresponding changes. The modules in the above embodiments may be combined into one module or further divided into multiple sub-modules.

[0162] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for pushing image reports based on RPA and AI, characterized in that: The method is applied to an image analysis platform integrated with a Robotic Process Automation (RPA) robot, and includes: S1. Receive a lung medical image to be detected sent by the RPA robot; S2. Use a preset lung segmentation model to segment the lung medical image to be detected to obtain a target medical image containing only the lung area, use a preset lung nodule detection model to detect the target medical image to obtain attribute information of each suspected lung nodule in the target medical image, and use the preset lung nodule segmentation model to segment the area where each suspected lung nodule is located to obtain three-dimensional contour information of each suspected lung nodule, wherein the preset lung nodule detection model is trained by lung nodule sample images and lung nodule attribute information in a preset lung nodule image sample set to establish a correlation between each lung nodule sample image and the lung nodule attribute information, and the preset lung nodule detection model can detect suspected lung nodules smaller than 5 mm in size; S3. Generate a suspected lung nodule imaging report based on the attribute information and three-dimensional contour information of each suspected lung nodule, and send the suspected lung nodule imaging report to the hospital platform through the RPA robot; The preset lung nodule detection model in step S2 is obtained by: S22, obtaining lung nodule sample images from a preset lung nodule image sample set and attribute information of the lung nodules contained in the lung nodule sample images, wherein the size of the lung nodules in the attribute information of the lung nodules contained in the lung nodule sample images used during training is less than 5 mm; S23, inputting the lung nodule sample image and the attribute information of the lung nodules contained in the lung nodule sample image into an initial convolutional neural network for feature extraction, to obtain reference attribute information of the lung nodules contained in the lung nodule sample image; S24, calculating a difference value between the reference attribute information and the attribute information of the lung nodules included in the lung nodule sample image; S25, optimizing the parameters of the initial convolutional neural network based on the difference value, and returning to step S22; S26. When the number of iterations reaches a preset number, the training is completed, and a preset lung nodule detection model is obtained that associates the lung nodule sample image with the attribute information of the lung nodule.

2. The method according to claim 1, wherein The step S2 of segmenting the lung medical image to be detected using a preset lung segmentation model to obtain a target medical image containing only the lung area specifically includes: S21. Filtering the radiomics features of the lung region from the lung medical image to be detected using a filtered back projection reconstruction algorithm FBP and a sinogram-determined iterative reconstruction algorithm SAFIRE, reconstructing the radiomics features to obtain a target medical image containing only the lung region.

3. The method according to claim 1, wherein The preset lung nodule segmentation model in step S2 is a three-dimensional semantic segmentation convolutional neural network 3DU-Net model.

4. The method according to claim 1, wherein The attribute information of each suspected pulmonary nodule includes pulmonary nodule location, pulmonary nodule size and pulmonary nodule density.

5. The method according to claim 4, wherein The step S3 specifically includes: S31, marking each suspected pulmonary nodule in the to-be-detected lung medical image according to the pulmonary nodule location in the attribute information of each suspected pulmonary nodule; S32. Generate a density histogram based on the lung nodule density in the attribute information of each suspected lung nodule; S33. Generate a suspected lung nodule image report including the labeled lung medical image to be detected, the density histogram, and the size of each suspected lung nodule, and send the suspected lung nodule image report to the hospital platform through the RPA robot.

6. The method according to any one of claims 1 to 5, characterized in that: The optical character recognition (OCR) method is used to recognize the lung medical image to be detected to obtain the image content.

7. An image report push device based on RPA and AI, characterized in that: Applied to an image analysis platform integrated with a Robotic Process Automation (RPA) robot, the device includes: A receiving module, configured to receive the lung medical image to be detected sent by the RPA robot; a segmentation module for segmenting the lung medical image to be detected using a preset lung segmentation model to obtain a target medical image containing only the lung area, detecting the target medical image using a preset lung nodule detection model to obtain attribute information of each suspected lung nodule in the target medical image, and segmenting the area where each suspected lung nodule is located using a preset lung nodule segmentation model to obtain three-dimensional contour information of each suspected lung nodule, wherein the preset lung nodule detection model is trained using lung nodule sample images and lung nodule attribute information in a preset lung nodule image sample set to establish an association between each lung nodule sample image and the lung nodule attribute information, and the preset lung nodule detection model can detect suspected lung nodules smaller than 5 mm in size; An imaging report push module is used to generate an imaging report of suspected lung nodules based on the attribute information and three-dimensional contour information of each suspected lung nodule, and send the imaging report of suspected lung nodules to the hospital platform through the RPA robot; The above-mentioned imaging report push device based on RPA and AI also includes a model training module, which is used to train a preset lung nodule detection model. The model training module includes: An acquisition submodule is used to obtain lung nodule sample images from a preset lung nodule image sample set and attribute information of the lung nodules contained in the lung nodule sample images, wherein the size of the lung nodules in the attribute information of the lung nodules contained in the lung nodule sample images used during training is less than 5 mm; A sample input submodule is used to input the lung nodule sample image and the attribute information of the lung nodules contained in the lung nodule sample image into the initial convolutional neural network for feature extraction, so as to obtain the reference attribute information of the lung nodules contained in the lung nodule sample image; a calculation submodule, configured to calculate a difference value between the reference attribute information and the attribute information of the lung nodules contained in the lung nodule sample image; A parameter optimization submodule, configured to optimize the parameters of the initial convolutional neural network based on the difference value, and trigger the acquisition submodule; The training completion submodule is used to complete the training when the number of iterations reaches a preset number, and obtain a preset lung nodule detection model that associates the lung nodule sample image with the attribute information of the lung nodule.

8. The device according to claim 7, wherein The segmentation module is specifically used to: The radiomics features of the lung area are screened out from the lung medical image to be detected using the filtered back projection reconstruction algorithm FBP and the sinusoidal graph determined iterative reconstruction algorithm SAFIRE, and the radiomics features are reconstructed to obtain a target medical image containing only the lung area.

9. The device according to claim 7, wherein The invention also includes a model training module, which is used to train a preset lung nodule detection model. The model training module includes: An acquisition submodule, configured to acquire a lung nodule sample image from a preset lung nodule image sample set and attribute information of the lung nodules contained in the lung nodule sample image; A sample input submodule is used to input the lung nodule sample image and the attribute information of the lung nodules contained in the lung nodule sample image into the initial convolutional neural network for feature extraction, so as to obtain the reference attribute information of the lung nodules contained in the lung nodule sample image; a calculation submodule, configured to calculate a difference value between the reference attribute information and the attribute information of the lung nodules contained in the lung nodule sample image; a parameter optimization submodule, configured to optimize the parameters of the initial convolutional neural network based on the difference value, and trigger the acquisition submodule; The training completion submodule is used to complete the training when the number of iterations reaches a preset number, and obtain a preset lung nodule detection model that associates the lung nodule sample image with the attribute information of the lung nodule.

10. A computing device, characterized in that The method comprises a storage device and a processor, wherein the storage device is used to store a computer program, and the processor runs the computer program to enable the computing device to perform the steps of the method according to any one of claims 1 to 6.

11. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images