Image processing equipment for breast screening
By designing image processing equipment for breast screening, using deep convolutional neural network and machine vision technology to analyze breast images, the problem of high error rate of breast screening in the prior art is solved, and higher accuracy and detection efficiency of breast disease screening are achieved.
Patent Information
- Application Number
- CN202510050188.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-13
- Publication Date
- 2025-06-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing breast screening technology has a high error rate in image recognition and detection, which is mainly due to the influence of image acquisition clarity and environmental factors, which lead to difficulties in identifying small tissues.
An image processing device for breast screening is designed, including an image acquisition unit, a data processing unit, a screening display unit and a report output unit. The device constructs a breast molybdenum target X-ray mass detection model through deep convolutional neural network, combines imaging theory and machine vision technology to segment and detect and analyze breast image data, generate early warning levels, and provide diagnostic support through multi-mode display and report output modules.
It improves the accuracy of breast disease screening, accurately detects tiny breast mass in breast mammography X-ray images, reduces the problem of inaccurate image detection caused by imaging differences of multiple models of equipment, and improves the detection and diagnosis efficiency of breast mass.
Smart Images
Figure CN120108655A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of health and medical technology, and in particular to an image processing device for breast screening. Background Art
[0002] Breast tumor is one of the most common malignant tumors in women. Early screening, detection and treatment are the key to improving the treatment effect of tumors. With the development of computer technology, existing breast screening can collect breast imaging data and use image analysis technology to assist in the diagnosis of breast tumor pathology. However, due to the influence of factors such as image acquisition clarity and acquisition environment, as well as the degree of recognition of small tissues, breast pathology image recognition and detection have a high error rate.
[0003] In summary, how to overcome the above-mentioned defects is a problem that those skilled in the art need to solve urgently. Summary of the invention
[0004] In response to the above-mentioned problems and needs, this solution proposes an image processing device for breast screening, which can solve the above-mentioned technical problems by adopting the following technical solutions.
[0005] To achieve the above object, the present invention provides the following technical solutions:
[0006] The image processing device for breast screening includes: an image acquisition unit, a data processing unit, a screening display unit and a report output unit;
[0007] The image acquisition unit is used to acquire breast image information, and send the image information to the data processing unit after preprocessing;
[0008] The data processing unit is used to perform segmentation detection and analysis on breast image data through a breast screening model, and to provide a warning level according to the analysis result;
[0009] The screening display unit is electrically connected to the data processing unit, and the screening display unit is used to display and remind the analysis results and warning information in multiple modes;
[0010] The report output unit is connected to the data processing unit, and is used to verify the identity of the management personnel, generate a report according to the analysis results and warning information, and print out the report.
[0011] Furthermore, the image acquisition unit includes an image receiving module and a data set construction module; the image receiving module is used to receive the breast molybdenum target X-ray image data to be identified and the breast molybdenum target X-ray image original data set, and convert the data format of the received breast molybdenum target X-ray image, and store the breast molybdenum target X-ray image after the format conversion; the data set construction module is connected to the image acquisition unit, and the data set construction module is used to expand the data to construct a data set, and perform image enhancement and data enhancement processing on the image data in the data set.
[0012] Furthermore, the data set construction module includes a labeling module, a contrast enhancement module and a data enhancement module;
[0013] The labeling module is used to label the position and category of the mammary gland mammary gland X-ray image masses in the original data set of mammary gland mammary gland X-ray images by using a labeling tool;
[0014] The contrast enhancement module is connected to the labeling module, and is used to perform local enhancement processing on the mass in the labeled data set by using the morphological top-hat and bottom-hat transformation method, and to crop the image;
[0015] The data enhancement module is connected to the contrast enhancement module, and is used to enhance the data set by adopting brightness and contrast change methods, translation, rotation, scale scaling and mirroring methods, Cutout and Cutmix image cropping methods and Mosaic image stitching methods, so as to expand the data volume and obtain an image data set.
[0016] Further, the data processing unit includes a model analysis module and a warning marking module;
[0017] The model analysis module is used to construct a breast mammography X-ray mass detection model using a deep convolutional neural network, and perform breast mass feature extraction, feature fusion and analysis output based on the breast mammography X-ray mass detection model;
[0018] The early warning marking module is connected to the model analysis module, and is used to perform early warning marking on the number and grade of detected lumps according to the analysis output results of the breast lumps.
[0019] Furthermore, the model analysis module includes a feature extraction module, a feature fusion module and a prediction classification module;
[0020] The feature extraction module uses the Darknet-53 backbone network to extract breast mass image features;
[0021] The feature fusion module is connected to the feature extraction module, and is used to obtain the feature maps of three different sizes obtained in the last three residual blocks of the Darknet-53 backbone network and the feature maps obtained after upsampling, and perform multi-scale feature fusion of the two feature maps through the MCFN feature fusion network. The MCFN feature fusion network cascades the FPN and PAN structures, and the sizes of the three predicted features are 52*52, 26*26, and 13*13 respectively;
[0022] The prediction and classification module is connected to the feature fusion module. The prediction and classification module is used to predict the results on multiple feature maps output by the FPN feature fusion network, obtain the information of prediction bounding boxes of different sizes, mass categories and confidence levels, the sizes of the three prediction features correspond to three groups of anchor boxes of different sizes, use the K-means anchor box clustering algorithm to perform cluster analysis on the bounding box labels in the data set, and use the non-maximum suppression algorithm to suppress redundant prediction boxes, and assign corresponding anchor boxes to the three prediction features shown.
[0023] Furthermore, the screening and display unit includes a mode selection module and a mode matching module;
[0024] The mode selection module is used for the screening personnel to select the early warning display mode of the early warning information of the mass, and the early warning display mode includes a voice mixed reminder mode, a projection display mode and an animation mark display module;
[0025] The pattern matching module is connected to the pattern selection module, and is used to receive warning mark information corresponding to the number and level of detected masses, and match and output the coding information corresponding to the warning mark information with the warning display mode selected by the user.
[0026] Furthermore, the report output unit includes a personnel identity verification module, a report generation module and a print output module;
[0027] The personnel identity authentication module includes an information registration module, an audit module and a login verification module. The information registration module is used for the screening personnel to enter basic identity information, and use the basic identity information and feedback information from the audit module to create unique account information, and send the entered basic identity information to the audit module. The audit personnel will review and confirm the identity information through the audit module and then feed it back to the information registration module. The login verification module is connected to the information registration module. The login verification module is used to compare the feature information corresponding to the obtained login information of the screening personnel with the basic identity information stored in the personnel database, and judge whether the identity of the personnel is legal according to the comparison result;
[0028] The report generation module is used to receive the breast mammography X-ray image mass detection results and early warning display information, and generate a detection report according to a preset report template;
[0029] The print output module is connected to the report generation module, and is used to transmit the report to be printed and the print control signal to the smart printer through the server for print output.
[0030] Furthermore, the report generation module includes an information acquisition module and a template matching module. The information acquisition module is used to obtain breast mammography X-ray image mass detection results and warning display information. The template matching module is connected to the information acquisition module. The template matching module is used to match the format of the test report according to the report template selected by the screening personnel, and fill the acquired information into the corresponding test report template.
[0031] It can be seen from the above technical scheme that the beneficial effects of the present invention are: the present invention is based on image processing technology for breast screening, and can accurately detect small breast masses in breast mammography X-ray images by applying imaging theory and machine vision technology to the recognition of breast tumor pathology images, thereby improving the accuracy of breast disease screening, avoiding the problem of inaccurate image detection caused by imaging differences among multiple models of equipment, and improving the detection and diagnosis efficiency of breast masses.
[0032] In addition to the objects, features and advantages described above, the best embodiments for implementing the present invention will be described in more detail below with reference to the accompanying drawings so that the features and advantages of the present invention can be easily understood. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] In order 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 of the present invention or the description of the prior art, wherein the drawings are only used to show some embodiments of the present invention, rather than limiting all embodiments of the present invention thereto.
[0034] Figure 1 The present invention is a schematic diagram of the composition structure of an image processing device for breast screening.
[0035] Figure 2 It is a schematic diagram of the composition structure of the image acquisition unit in the present invention.
[0036] Figure 3 It is a schematic diagram of the composition structure of the data processing unit in the present invention.
[0037] Figure 4 It is a schematic diagram of the composition structure of the report output unit in the present invention. DETAILED DESCRIPTION
[0038] In order to make the purpose, technical solution and advantages of the technical solution of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described in conjunction with the drawings of specific embodiments of the present invention. The same figure marks in the drawings represent the same parts. It should be noted that the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the described 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.
[0039] In recent years, with the rapid development of medical equipment in my country, medical imaging technology has made corresponding progress in theoretical methods and applications. The number of various types of medical images has increased day by day, and it has become one of the most important auxiliary technical means for clinical diagnosis.
[0040] like Figures 1 to 4 As shown, the present invention discloses an image processing device for breast screening, which is based on image processing technology for breast screening and can accurately detect small breast masses in breast mammary target X-ray images and improve the accuracy of breast disease screening. The image processing device for breast screening specifically includes: an image acquisition unit, a data processing unit, a screening display unit and a report output unit, wherein the image acquisition unit is used to acquire breast image information and send the image information to the data processing unit after pre-processing.
[0041] Specifically, the image acquisition unit includes an image receiving module and a data set construction module; the image receiving module is used to receive breast mammography X-ray image data to be identified and a breast mammography X-ray image original data set, and convert the data format of the received breast mammography X-ray image, and store the breast mammography X-ray image after format conversion; the data set construction module is connected to the image acquisition unit, and the data set construction module is used to expand the data to construct a data set, and perform image enhancement and data enhancement processing on the image data in the data set.
[0042] The data set construction module includes a labeling module, a contrast enhancement module and a data enhancement module; the labeling module is used to use a labeling tool to label the position and category of the breast mammography X-ray image masses in the original data set of the breast mammography X-ray image; the contrast enhancement module is connected to the labeling module, and the contrast enhancement module is used to use a morphological top-hat and bottom-hat transformation method to perform local enhancement processing on the masses in the labeled data set, and to crop the image; the data enhancement module is connected to the contrast enhancement module, and the data enhancement module is used to use a brightness and contrast change method, a translation, rotation, scale scaling and mirroring method, a Cutout and Cutmix image cropping method and a Mosaic image splicing method to enhance the data set, expand the data volume to obtain an image data set.
[0043] The screening method of existing breast cancer mainly contains mammography, ultrasonography and magnetic resonance imaging and digital breast tomosynthesis technology etc., wherein, mammography is the most widely used screening and diagnostic tool in current clinical and scientific fields.Therefore, in the present embodiment, mammography data is analyzed as detection object, but the contrast of the lump in general mammography and surrounding breast tissue is low, direct detection is not easy to obtain good detection performance, and unprocessed mammography data, easily cause the problem of overfitting in the model during training, therefore, need to pre-process the image to be identified and the data set image for training.Through morphological processing in the lump information in the enhanced structural element, and make the tissue information outside the structural element unchanged, and then reach the effect of local enhancement.Then the data set is expanded to improve the model detection performance by the method of multiple data enhancement.
[0044] The data processing unit is used to perform segmentation, detection and analysis on breast image data through a breast screening model, and to provide a warning level according to the analysis result.
[0045] Specifically, the data processing unit includes a model analysis module and a warning marking module; the model analysis module is used to construct a breast mammography X-ray mass detection model using a deep convolutional neural network, and perform breast mass feature extraction, feature fusion and analysis output based on the breast mammography X-ray mass detection model; the warning marking module is connected to the model analysis module, and the warning marking module is used to perform warning marking on the number and grade of detected masses according to the analysis output results of the breast masses.
[0046] More specifically, the model analysis module includes a feature extraction module, a feature fusion module and a prediction classification module; the feature extraction module uses the Darknet-53 backbone network to extract breast mass image features; the feature fusion module is connected to the feature extraction module, and the feature fusion module is used to obtain the feature maps of three different sizes obtained in the last three residual blocks of the Darknet-53 backbone network and the feature maps obtained after upsampling, and the two feature maps are multi-scale feature fused through the MCFN feature fusion network, and the MCFN feature fusion network is used to hierarchically combine the FPN and PAN structures. The sizes of the three prediction features are 52*52, 26*26, and 13*13 respectively; the prediction classification module is connected to the feature fusion module, and the prediction classification module is used to predict the results on multiple feature maps output by the FPN feature fusion network, and obtain the information of prediction bounding boxes of different sizes, mass categories and confidence levels. The sizes of the three prediction features correspond to three groups of anchor boxes of different sizes respectively. The K-means anchor box clustering algorithm is used to perform clustering analysis on the bounding box labels in the data set, and the non-maximum suppression algorithm is used to suppress redundant prediction boxes, and corresponding anchor boxes are assigned to the three prediction features shown.
[0047] In this embodiment, a detection model is constructed based on the YOLOv3 target detection model. The input image is analyzed and inferred by the detection model to obtain the categories and corresponding confidences of all targets in the image. The detection model extracts the corresponding image features through the backbone network Darknet-53, which contains 53 convolutional layers and uses a large number of residual units in the network. The detection model uses the feature maps extracted by the last three residual blocks for feature fusion, which are 8 times, 16 times and 32 times the input image, respectively. It can ensure network convergence and improve the classification and detection effects.
[0048] However, the single YOLOv3 convolutional network has low recognition accuracy for small masses and mutually occluded masses in breast mammography X-ray mass images. Therefore, a bottom-up feature fusion path is added after the traditional FPN structure to form a multi-scale cross-path feature fusion method MCFN, which adopts the cascade FPN and PAN structure to fully integrate the three feature maps of the backbone network. A huge number of candidate frames will be generated in the prediction stage, and a large number of redundant candidate frames need to be screened to improve the detection effect.
[0049] The screening display unit is electrically connected to the data processing unit, and is used to display and remind the analysis results and warning information in multiple modes. The screening display unit includes a mode selection module and a mode matching module; the mode selection module is used for the screening personnel to select the warning display mode of the mass warning information, and the warning display mode includes a voice mixed reminder mode, a projection display mode and an animation mark display module; the mode matching module is connected to the mode selection module, and the mode matching module is used to receive the warning mark information corresponding to the number and level of the detected masses, and match the coded information corresponding to the warning mark information with the warning display mode selected by the user for output.
[0050] The report output unit is connected to the data processing unit, and is used to verify the identity of the management personnel, generate a report according to the analysis results and warning information, and print out the report.
[0051] Specifically, the report output unit includes a personnel identity authentication module, a report generation module and a printing output module; the personnel identity authentication module includes an information registration module, an audit module and a login verification module, the information registration module is used for the screening personnel to enter basic identity information, and use the basic identity information and feedback information of the audit module to create a unique account information, and send the entered basic identity information to the audit module, through which the auditor reviews and confirms the identity information and then feeds it back to the information registration module, the login verification module is connected to the information registration module, the login verification module is used to compare the characteristic information corresponding to the obtained screening personnel login information with the basic identity information stored in the personnel database, and judge whether the personnel identity is legal according to the comparison result; the report generation module is used to receive the breast mammography X-ray image mass detection results and early warning display information, and generate a detection report according to a preset report template; the printing output module is connected to the report generation module, and the printing output module is used to transmit the report to be printed and the printing control signal to the smart printer through the server for printing output.
[0052] The report generation module includes an information acquisition module and a template matching module. The information acquisition module is used to obtain breast mammography X-ray image mass detection results and early warning display information. The template matching module is connected to the information acquisition module. The template matching module is used to match the format of the test report according to the report template selected by the screening personnel, and fill the acquired information into the corresponding test report template.
[0053] It should be noted that the embodiments described in the present invention are only preferred ways to implement the present invention, and any obvious modifications that belong to the overall concept of the present invention should fall within the protection scope of the present invention.
Claims
1. An image processing device for breast screening, characterized in that: include: Image acquisition unit, data processing unit, screening display unit and report output unit; The image acquisition unit is used to acquire breast image information, and send the image information to the data processing unit after preprocessing; The data processing unit is used to perform segmentation detection and analysis on breast image data through a breast screening model, and to provide a warning level according to the analysis result; The screening display unit is electrically connected to the data processing unit, and the screening display unit is used to display and remind the analysis results and warning information in multiple modes; The report output unit is connected to the data processing unit, and is used to verify the identity of the management personnel, generate a report according to the analysis results and warning information, and print out the report.
2. The image processing device for breast screening according to claim 1, characterized in that: The image acquisition unit includes an image receiving module and a data set building module; the image receiving module is used to receive the mammary gland mammography X-ray image data to be identified and the mammary gland mammography X-ray image original data set, and convert the data format of the received mammary gland mammography X-ray image, and store the mammary gland mammography X-ray image after the format conversion; The data set construction module is connected to the image acquisition unit, and is used to expand the data to construct a data set, and perform image enhancement and data enhancement processing on the image data in the data set.
3. The image processing device for breast screening according to claim 2, characterized in that: The data set construction module includes a labeling module, a contrast enhancement module and a data enhancement module; The labeling module is used to label the position and category of the mammary gland mammary gland X-ray image masses in the original data set of mammary gland mammary gland X-ray images by using a labeling tool; The contrast enhancement module is connected to the labeling module, and is used to perform local enhancement processing on the mass in the labeled data set by using the morphological top-hat and bottom-hat transformation method, and to crop the image; The data enhancement module is connected to the contrast enhancement module, and is used to enhance the data set by adopting brightness and contrast change methods, translation, rotation, scale scaling and mirroring methods, Cutout and Cutmix image cropping methods and Mosaic image stitching methods, so as to expand the data volume and obtain an image data set.
4. The image processing device for breast screening according to claim 1, characterized in that: The data processing unit includes a model analysis module and a warning marking module; The model analysis module is used to construct a breast mammography X-ray mass detection model using a deep convolutional neural network, and perform breast mass feature extraction, feature fusion and analysis output based on the breast mammography X-ray mass detection model; The early warning marking module is connected to the model analysis module, and is used to perform early warning marking on the number and grade of detected lumps according to the analysis output results of the breast lumps.
5. The image processing device for breast screening according to claim 4, characterized in that: The model analysis module includes a feature extraction module, a feature fusion module and a prediction classification module; The feature extraction module uses the Darknet-53 backbone network to extract breast mass image features; The feature fusion module is connected to the feature extraction module, and is used to obtain the feature maps of three different sizes obtained in the last three residual blocks of the Darknet-53 backbone network and the feature maps obtained after upsampling, and perform multi-scale feature fusion of the two feature maps through the MCFN feature fusion network. The MCFN feature fusion network cascades the FPN and PAN structures, and the sizes of the three predicted features are 52*52, 26*26, and 13*13 respectively; The prediction and classification module is connected to the feature fusion module. The prediction and classification module is used to predict the results on multiple feature maps output by the FPN feature fusion network, obtain the information of prediction bounding boxes of different sizes, mass categories and confidence levels, the sizes of the three prediction features correspond to three groups of anchor boxes of different sizes, use the K-means anchor box clustering algorithm to perform cluster analysis on the bounding box labels in the data set, and use the non-maximum suppression algorithm to suppress redundant prediction boxes, and assign corresponding anchor boxes to the three prediction features shown.
6. The image processing device for breast screening according to claim 5, characterized in that: The screening and display unit includes a mode selection module and a mode matching module; The mode selection module is used for the screening personnel to select the early warning display mode of the early warning information of the mass, and the early warning display mode includes a voice mixed reminder mode, a projection display mode and an animation mark display module; The pattern matching module is connected to the pattern selection module, and is used to receive warning mark information corresponding to the number and level of detected masses, and match and output the coding information corresponding to the warning mark information with the warning display mode selected by the user.
7. The image processing device for breast screening according to claim 6, characterized in that: The report output unit includes a personnel identity authentication module, a report generation module and a print output module; The personnel identity authentication module includes an information registration module, an audit module and a login verification module. The information registration module is used for the screening personnel to enter basic identity information, and use the basic identity information and feedback information from the audit module to create unique account information, and send the entered basic identity information to the audit module. The audit personnel will review and confirm the identity information through the audit module and then feed it back to the information registration module. The login verification module is connected to the information registration module. The login verification module is used to compare the feature information corresponding to the obtained login information of the screening personnel with the basic identity information stored in the personnel database, and judge whether the identity of the personnel is legal according to the comparison result; The report generation module is used to receive the breast mammography X-ray image mass detection results and early warning display information, and generate a detection report according to a preset report template; The print output module is connected to the report generation module, and is used to transmit the report to be printed and the print control signal to the smart printer through the server for print output.
8. The image processing device for breast screening according to claim 7, characterized in that: The report generation module includes an information acquisition module and a template matching module. The information acquisition module is used to obtain breast mammography X-ray image mass detection results and early warning display information. The template matching module is connected to the information acquisition module. The template matching module is used to match the format of the test report according to the report template selected by the screening personnel, and fill the acquired information into the corresponding test report template.