Breast cancer prevention and treatment method and system based on ultrasonic artificial intelligence and 5G internet of medicine

Abstract

The invention discloses a breast cancer prevention and treatment method and system based on ultrasonic artificial intelligence and 5G internet of medicine, and is applied to the technical field of breast cancer prevention and treatment. The method comprises the following steps: formulating a breast cancer risk factor questionnaire, performing breast cancer related risk factor investigation on a research object, and establishing a risk prediction model; acquiring mammary gland color Doppler ultrasound image data of the research object; a breast nodule screening system based on the ultrasonic artificial intelligence technology is established, the breast nodule screening system conducts artificial intelligence technology analysis on the breast color ultrasound image data, and a preliminary diagnosis result is given; and obtaining a breast nodule screening rate, a breast cancer early cancer screening rate and an early diagnosis rate according to an analysis result of the breast nodule screening system. The artificial intelligence ultrasonic inspection is more convenient, rapid in input, high in inspection speed and capable of adapting to various complex environments.

Description

Technical Field

[0001] The present invention relates to the technical field of breast cancer prevention and treatment, and more specifically to a breast cancer prevention and treatment method and system based on ultrasonic artificial intelligence and 5G medical networking. Background Art

[0002] Ultrasound is the main means of breast cancer screening, but the accuracy of two-dimensional ultrasound and the diagnosis of the nature of breast nodules are still insufficient. In addition, the lack of experience and skills of ultrasound physicians in grassroots hospitals can easily lead to misdiagnosis and missed diagnosis. For this reason, ultrasound artificial intelligence algorithm diagnosis based on two-dimensional ultrasound diagnosis has been increasingly used in clinical practice. AI breast ultrasound uses artificial intelligence algorithms represented by deep learning neural networks to simulate the human brain, and has super adaptability and learning ability. It can automatically identify and classify breast lesions and simulate clinical physicians to diagnose and evaluate the prognosis of breast cancer. Innovative artificial intelligence algorithms can better identify lesions, avoid missed diagnosis and misdiagnosis, improve the efficiency of examinations and the degree of automation of implementation, and improve the accuracy of examinations, speed up the work process, and save time.

[0003] The breast nodule patients in the Huyanghe City Medical Community of the Seventh Division of the Xinjiang Production and Construction Corps cover many areas such as the Tianbei New District and the regiment. The area is vast and sparsely populated, and the diagnosis and treatment level of grassroots companies is relatively low. In order to solve the problem of medical treatment, we have adopted the method of sinking superior medical resources to the grassroots level to provide breast nodule screening for employees in their place of residence.

[0004] Therefore, it is an urgent problem for technical personnel in this field to propose a breast cancer prevention and treatment method and system based on ultrasonic artificial intelligence and 5G medical networking to solve the difficulties existing in the existing technology. Summary of the invention

[0005] In view of this, the present invention provides a breast cancer prevention and treatment method and system based on ultrasonic artificial intelligence and 5G medical networking, and innovates artificial intelligence algorithms to improve the sensitivity and specificity of ultrasonic diagnosis of breast nodules.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A method for preventing and treating breast cancer based on ultrasound artificial intelligence and 5G medical networking, comprising the following steps:

[0008] S1. Model establishment steps: formulate a breast cancer risk factor questionnaire, investigate breast cancer-related risk factors on the research subjects, and establish a risk prediction model;

[0009] S2, data collection step: collecting breast color Doppler ultrasound image data of the research subjects;

[0010] S3, system establishment and data analysis steps: establish a breast nodule screening system based on ultrasound artificial intelligence technology, the breast nodule screening system performs artificial intelligence technology analysis on breast color ultrasound image data and gives a preliminary diagnosis result;

[0011] S4. Data aggregation step: Based on the analysis results of the breast nodule screening system, the breast nodule screening rate, breast cancer early cancer screening rate and early diagnosis rate are obtained.

[0012] Optionally, S1 also includes the establishment of a breast nodule ultrasound database, which provides ultrasound examination reports based on BI-RADS-US standard terminology, as well as ultrasound examination and pathological diagnosis data approved by SAS; uses the XGBoost algorithm for feature selection, uses the CNN deep learning model for breast nodule prediction, and uses cross-validation to evaluate the model to verify the accuracy, sensitivity, specificity, and precision of the prediction model.

[0013] Optionally, the algorithm in the breast nodule screening system in S3 includes but is not limited to the following clinical parameters: mass morphology, mass margin, internal echo, calcification, mass aspect ratio, capsule and blood flow.

[0014] Optionally, the breast nodule screening system in S3 uses artificial intelligence technology to analyze breast color ultrasound image data, and the results given include not only the classification of breast nodules, but also graded diagnosis and treatment based on the type of breast nodules.

[0015] A breast cancer prevention and treatment system based on ultrasonic artificial intelligence and 5G medical networking, a breast cancer prevention and treatment method based on ultrasonic artificial intelligence and 5G medical networking using any of the above, comprising a model building module, a data acquisition module, a system building and data analysis module, and a data aggregation module connected in sequence; wherein,

[0016] The model building module is used to develop a breast cancer risk factor questionnaire, conduct a breast cancer-related risk factor survey on the research subjects, and establish a risk prediction model;

[0017] A data acquisition module, used to collect breast color ultrasound image data of the research object;

[0018] System establishment and data analysis module, used to establish a breast nodule screening system based on ultrasound artificial intelligence technology. The breast nodule screening system uses artificial intelligence technology to analyze breast color ultrasound image data and gives preliminary diagnosis results;

[0019] The data summary module is used to obtain the breast nodule screening rate, breast cancer early cancer screening rate and early diagnosis rate based on the analysis results of the breast nodule screening system.

[0020] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a method and system for preventing and treating breast cancer based on ultrasonic artificial intelligence and 5G medical networking, which has the following beneficial effects:

[0021] The artificial intelligence ultrasound examination of the present invention is more convenient, has fast input and examination speed, and can adapt to various complex environments. Compared with traditional examinations, it has the following advantages:

[0022] (1) Efficiency: Artificial intelligence has promoted the rapid development of automated ultrasound examinations, greatly improving examination efficiency, even to 2-3 times faster than manual inspections.

[0023] (2) Accuracy: AI has obvious advantages in terms of accuracy, with an accuracy of one thousandth of an inch. It is not subject to subjective control, and as long as the parameter settings are the same, multiple machines with the same configuration can maintain the same accuracy.

[0024] (3) Objectivity: Manual inspections are subject to fatigue and have a fatal flaw, which is the subjectivity caused by emotions. The results of inspections carried out by artificial intelligence are naturally more objective and reliable.

[0025] (4) Digitalization: All image data generated during the AI ​​ultrasound examination can be copied independently or via a network connection to facilitate the summary, statistics and analysis of screening results.

[0026] In summary, artificial intelligence used in breast nodule screening has the advantages of digitization, traceability, and high precision, which greatly improves the sensitivity and specificity of breast cancer screening. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0028] Figure 1 A flow chart of a method for breast cancer prevention and treatment based on ultrasonic artificial intelligence and 5G medical networking provided by the present invention;

[0029] Figure 2 The three types of networking modes of the 5G private network provided by the present invention;

[0030] Among them, 2a is the premium mode, 2b is the exclusive mode, and 2c is the premium mode;

[0031] Figure 3 The logical architecture diagram of the 5G medical private network provided by the present invention;

[0032] Figure 4 This is a networking diagram of the 5G high-definition remote diagnosis and treatment platform provided by the present invention;

[0033] Figure 5 This is a diagram of the implementation process of the algorithm for detecting lesions in the MammoWorks ultrasonic AI detection system of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. 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.

[0035] Reference Figure 1 As shown, the present invention discloses a method for preventing and treating breast cancer based on ultrasonic artificial intelligence and 5G medical networking, comprising the following steps:

[0036] S1. Model establishment steps: formulate a breast cancer risk factor questionnaire, investigate breast cancer-related risk factors on the research subjects, and establish a risk prediction model;

[0037] S2, data collection step: collecting breast color Doppler ultrasound image data of the research subjects;

[0038] S3, system establishment and data analysis steps: establish a breast nodule screening system based on ultrasound artificial intelligence technology, the breast nodule screening system performs artificial intelligence technology analysis on breast color ultrasound image data and gives a preliminary diagnosis result;

[0039] S4. Data aggregation step: derive the breast nodule screening rate, breast cancer early cancer screening rate and early diagnosis rate based on the analysis results of the breast nodule screening system.

[0040] Specifically, the subjects were investigated, and the main contents of the questionnaire included: gender, age, marital status, menarche, age at first birth, body mass index (BMI), menopause time, breastfeeding history, number of pregnancies, miscarriage history, number of full-term births, passive smoking history, education level, occupation, lifestyle and behavior (including smoking and drinking), physical exercise, estrogen use, radiation exposure history, tumor disease history, breastfeeding history, health status survey (including family history and history of benign breast disease), emotional and psychological factors, and the definition of diet (high-fat diet: mainly fried foods, fatty meat, animal offal and cream products, crude fiber diet: beans, fruits, vegetables and fungi-based foods); definition of physical exercise (never: never do physical exercise, occasionally: exercise <3 times / week, each <30min, often: exercise ≥3 times / week, each ≥_30min); definition of drinking: drinking liquor at least once a week for more than six months (occasionally: average alcohol consumption ≤100ml / day; often: average alcohol consumption >100ml / day); definition of smoking: at least one cigarette a day for more than six months (occasionally: smoking ≤7 cigarettes / week, often: smoking >7 cigarettes / week). Analyze the breast cancer risk factors obtained from the survey, establish a risk prediction model, and evaluate the model.

[0041] The establishment of this model can assist breast nodule stakeholders to take early intervention measures or policy decisions to reduce the risk of breast cancer among women in the Seventh Division of the Xinjiang Production and Construction Corps, and can also assist health management departments in allocating medical resources more rationally.

[0042] Furthermore, S1 also includes the establishment of a breast nodule ultrasound database, which provides ultrasound examination reports based on the BI-RADS-US standard terminology, as well as ultrasound examination and pathological diagnosis data recognized by SAS; the XGBoost algorithm is used for feature selection, the CNN deep learning model is used for breast nodule prediction, and the cross-validation evaluation model is used to verify the accuracy, sensitivity, specificity, and precision of the prediction model.

[0043] Specifically, a breast nodule ultrasound database is established to provide research data for ultrasound analysis with descriptions and reports of breast nodules. According to the standard terminology of breast ultrasound reports and data systems, a database containing information such as the ultrasound manifestations and classifications of breast nodules and their pathological results is created using MySQL. This information is replaced with data for recognition by SAS software. Result patient information can be obtained from the hospital information system or image storage and transmission system, or directly entered from the PATIENT form. After the ultrasound examination is completed, the ultrasound examination information can be entered from the US-EXAM or PATIENT form, and then the standard ultrasound report can be output or printed. The pathological results of breast nodule lesions can also be entered into the database. As needed, ultrasound and pathological information can be transferred from the database to SAS for research and analysis.

[0044] Furthermore, the algorithm in the breast nodule screening system in S3 includes the following clinical parameters: including but not limited to the following clinical parameters: (1) Tumor morphology. Irregular tumor morphology is a common manifestation of breast cancer and is also the most sensitive ultrasound sign for diagnosing breast cancer. For example, crab-like and other irregular shapes are typical morphologies of breast cancer. A quasi-round or oval tumor is more likely to be a benign breast nodule. (2) Tumor margin. The border of breast cancer is often unclear. Typical features are peripheral spiculation and strong echo halo, which are the infiltration and growth of breast cancer into surrounding tissues. Benign breast nodules often have smooth margins. (3) Internal echo. Compared with breast glands and adipose tissue, the internal echo of breast cancer is often significantly low echo. Breast cancer lesions may be accompanied by echo attenuation behind the tumor. Benign breast nodules such as breast hyperplasia and mastitis may show echo-free areas on ultrasound. Mastitis may also show uneven internal echoes. (4) Calcification. Breast cancer lumps are often accompanied by tiny calcifications or sand-like calcifications, while benign breast nodules are usually not accompanied by calcifications. (5) Aspect ratio of the mass. It is generally believed that the aspect ratio of breast cancer masses is greater than 1, that is, the growth of the mass is not parallel or perpendicular to the axis of the breast gland, and the height is greater than the width. The aspect ratio of benign breast nodules is less than 1 or less than 0.8. (6) Capsule. Benign breast nodules often have a complete capsule, while malignant breast nodules often lack a complete capsule. (7) Blood flow. Benign breast nodules have less blood flow, while malignant breast nodules have more abundant blood flow. At the same time, the blood flow resistance of malignant breast nodules is often higher than that of benign nodules. The breast ultrasound artificial intelligence device is used to diagnose the breast ultrasound images of the enrolled patients. If the result is a recommendation for follow-up, it indicates a possible benign condition. If the result is a recommendation for examination, it indicates a possible malignancy. The specific test results must be combined with clinical practice.

[0045] The core of the ultrasound examination algorithm model is an advanced object detection framework based on convolutional neural network (CNN) for detecting, locating, and classifying objects in images, such as the VGG16 network. In a branch of the convolutional neural network, the ROI (Region of Interest) objects on the image are detected and located through convolutional layers. According to the category of the object (nodule, calcification, ductal dilation, lymph node, etc.), default detection boxes of different sizes and aspect ratios are used to find objects of different sizes and shapes. The default box with the highest score is the final rectangular box marked on the image. Another branch network of this neural network evaluates the information of each proposed region from the last convolutional layer and resamples it to a fixed size. The presence of lesions is detected through these two branches, and the boundary box regression task is used to refine the boundaries of the objects existing in the region. From the detected overlapping objects, the Non-Maximum Suppression (NMS) algorithm is selected to eliminate redundant (cross-repeated) windows and find the best detection position.

[0046] As Figure 5 shown, the specific method is to perform operations on the input breast ultrasound image through multiple convolutional layers and pooling layers of the neural network, so as to achieve automatic extraction of target features. After obtaining the feature target image (Feature map), it is input into a region proposal network (regionproposal network, RPN) network to automatically extract the target lesion region (ROI), and map the ROI region to the feature map (ROI pooling). Finally, the mapped feature map is used to predict the lesion category and location. The model outputs the bounding box of each detected lesion and a score, which reflects the confidence of the algorithm in evaluating the lesion category.

[0047] To facilitate network training, the overall loss of the RPN network includes classification loss (cls) and bounding box regression loss (reg), which are specifically defined as follows:

[0048]

[0049] In the above formula, i represents the i-th anchor, p i represents the probability that the i-th anchor is predicted as the true label, represents the corresponding GT predict probability (that is, when the IoU between the i-th anchor and GT > 0.7, it is considered that this anchor is positive, conversely, when IoU < 0.3, it is considered that this anchor is negative, as for those anchors with 0.3 < IoU < 0.7, they do not participate in training); t iRepresents the regression parameters for predicting the i-th anchor bounding box, Represents the GT box corresponding to the i-th anchor.

[0050] Furthermore, the breast nodule screening system in S3 uses artificial intelligence technology to analyze breast color ultrasound image data, and the results given include not only the classification of breast nodules, but also graded diagnosis and treatment based on the type of breast nodules.

[0051] Furthermore, hierarchical diagnosis and treatment specifically includes:

[0052] For BI-RADS 0-3, regular review is recommended, with follow-up once a year; for BI-RADS 4-5, further review is recommended at a higher-level hospital. After the review, if the nodule is considered benign, review and follow-up every 3 months is recommended; if the nodule is considered malignant, surgery or puncture biopsy is recommended to confirm the pathological diagnosis, and follow-up every 3 months; if the diagnosis cannot be made or there is doubt, expert remote consultation can be conducted through Internet telemedicine, and diagnosis and treatment recommendations can be given, with follow-up every 3 months.

[0053] Specifically, the present invention uses ultrasonic AI technology from the ultrasonic AI system of Tsinghua University's industry-university-research transformation product. The frequency of the breast examination probe is 5.0-12.0MHz. Physicians in primary hospitals perform ultrasonic examinations on research subjects and collect images. AI breast ultrasound can automatically identify and classify breast lesions. The breast BI-RADS classification standard refers to the BI-RADS of the American College of Radiology in 2013. Nodules can be divided into 0-6 categories. For BI-RADS 0-3, regular review is recommended, and follow-up is once a year; for breast nodules of BI-RADS 4-5, it is recommended that a higher-level hospital conduct further review. After the review, if benign nodules are considered, it is recommended to conduct a review and follow-up every 3 months; if malignant nodules are considered, surgery or puncture biopsy is recommended to confirm the pathological diagnosis, and follow-up every 3 months; if a diagnosis cannot be made or there is a doubt, expert remote consultation can be conducted through Internet telemedicine, and diagnosis and treatment recommendations can be given, with follow-up every 3 months.

[0054] The breast ultrasound examination results, biopsy puncture or surgery results, and treatment status of the research subjects were evaluated and analyzed, and the breast nodule screening rate, breast cancer early cancer screening rate, and early diagnosis rate in Huyanghe City, the Seventh Division of the Corps were obtained through statistical analysis;

[0055] Specifically, the present invention builds a remote consultation medical networking system based on 5G Internet key technologies to construct a hierarchical diagnosis and treatment information-based cancer prevention system for breast cancer-specific medical alliances:

[0056] For data transmission in remote consultation medical networking systems, training, and audio and video communications, 5G public networks are used to meet the high bandwidth and low latency requirements for concurrent data and video streams. For patient information, including clinical data, ultrasound imaging data, and ultrasound real-time dynamic video, 5G private networks are used to achieve security and privacy in data transmission.

[0057] 5G private network includes three types of networking modes: using three private network modes of "premium, exclusive, and exclusive". Figure 2 As shown in the figure, 2a is the preferred mode: (1) reuse of large network: no need to build new resources, (2) configuration data to achieve differentiated services: by configuring different 5QI and CNN to achieve rate and priority differences; 2b is the exclusive mode: (1) enhanced coverage on the wireless side: add points and optimize on demand, (2) UPF and MEP sink on demand: exclusive use of core network media plane resources; 2c is the premium mode: (1) dedicated base station / frequency: exclusive use of wireless resources to achieve high isolation and high reliability; (2) dedicated SA core network: control plane resources (AMF / SMF / UDM) are provided on demand, etc., providing different dedicated solutions for wireless network and core network resources, gradually superimposing network capabilities, gradually improving network specialization, and gradually enhancing network value.

[0058] 5G base stations cover a wide range Figure 3 As shown in the figure, multi-point expert collaborative consultation can be carried out in the region. The base station AAU and BBU meet the bandwidth conditions and reach the gbps level. The audio and video system, data transmission system and same-screen collaboration system are connected to the 5G network through China Mobile CPE equipment. Figure 4 The mobile bearer network needs to build a connection from the base station to the core network gateway, and the connection from the base station side to the core network gateway includes an access ring and an aggregation ring.

[0059] A breast cancer prevention and treatment system based on ultrasound artificial intelligence and 5G medical networking, applying the above-mentioned breast cancer prevention and treatment method based on ultrasound artificial intelligence and 5G medical networking, including a model building module, a data acquisition module, a system building and data analysis module, and a data aggregation module connected in sequence; wherein,

[0060] The model building module is used to develop a breast cancer risk factor questionnaire, conduct a breast cancer-related risk factor survey on the research subjects, and establish a risk prediction model;

[0061] A data acquisition module, used to collect breast color ultrasound image data of the research object;

[0062] System establishment and data analysis module, used to establish a breast nodule screening system based on ultrasound artificial intelligence technology. The breast nodule screening system uses artificial intelligence technology to analyze breast color ultrasound image data and gives preliminary diagnosis results;

[0063] The data summary module is used to obtain the breast nodule screening rate, breast cancer early cancer screening rate and early diagnosis rate based on the analysis results of the breast nodule screening system.

[0064] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0065] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for preventing and treating breast cancer based on ultrasonic artificial intelligence and 5G medical networking, characterized in that: The following steps are involved: S1. Model establishment steps: formulate a breast cancer risk factor questionnaire, investigate breast cancer-related risk factors on the research subjects, and establish a risk prediction model; S2, data collection step: collecting breast color Doppler ultrasound image data of the research subjects; S3, system establishment and data analysis steps: establish a breast nodule screening system based on ultrasound artificial intelligence technology, the breast nodule screening system performs artificial intelligence technology analysis on breast color ultrasound image data and gives a preliminary diagnosis result; S4. Data aggregation step: Based on the analysis results of the breast nodule screening system, the breast nodule screening rate, breast cancer early cancer screening rate and early diagnosis rate are obtained.

2. A method for preventing and treating breast cancer based on ultrasonic artificial intelligence and 5G medical networking according to claim 1, characterized in that: S1 also includes the establishment of a breast nodule ultrasound database, which provides ultrasound examination reports based on the BI-RADS-US standard terminology, as well as ultrasound examination and pathological diagnosis data approved by SAS; the use of the XGBoost algorithm for feature selection, the use of the CNN deep learning model for breast nodule prediction, and the use of cross-validation evaluation models to verify the accuracy, sensitivity, specificity, and precision of the prediction model.

3. The method for preventing and treating breast cancer based on ultrasonic artificial intelligence and 5G medical networking according to claim 1, characterized in that: The algorithms in the breast nodule screening system in S3 include but are not limited to the following clinical parameters: mass morphology, mass margin, internal echo, calcification, mass aspect ratio, capsule and blood flow.

4. The method for preventing and treating breast cancer based on ultrasonic artificial intelligence and 5G medical networking according to claim 1, characterized in that: The breast nodule screening system in S3 uses artificial intelligence technology to analyze breast color ultrasound image data, and the results given include not only the classification of breast nodules, but also graded diagnosis and treatment based on the type of breast nodules.

5. A breast cancer prevention and treatment system based on ultrasonic artificial intelligence and 5G medical networking, characterized in that: A method for preventing and treating breast cancer based on ultrasonic artificial intelligence and 5G medical networking as described in any one of claims 1 to 5 is applied, comprising a model building module, a data acquisition module, a system building and data analysis module, and a data aggregation module connected in sequence; wherein, The model building module is used to develop a breast cancer risk factor questionnaire, conduct a breast cancer-related risk factor survey on the research subjects, and establish a risk prediction model; A data acquisition module, used to collect breast color ultrasound image data of the research object; System establishment and data analysis module, used to establish a breast nodule screening system based on ultrasound artificial intelligence technology. The breast nodule screening system uses artificial intelligence technology to analyze breast color ultrasound image data and gives preliminary diagnosis results; The data summary module is used to obtain the breast nodule screening rate, breast cancer early cancer screening rate and early diagnosis rate based on the analysis results of the breast nodule screening system.

Images