Ultrasonic intelligent collaborative quality control system

Abstract

The invention relates to the field of medical image technology and artificial intelligence cross technology, and discloses an ultrasonic intelligent collaborative quality control system. Comprising a data security module, a user login module, an artificial intelligence data large model, an automatic quality control selection module, a real-time online collaborative consultation module, a standard section interface module, an image acquisition and transmission module, a standard section classification module, a semantic segmentation module, a measurement index module and an anatomical knowledge module. The clinical guide module and the industry expert consensus reference management module are used for analyzing ultrasonic images by using an artificial intelligence large model, and optimizing the process through the automatic quality control module to ensure the image quality; the semantic segmentation module calculates the area ratio Qa of the lesion region and the three-dimensional volume ratio Qb of the lesion region, and provides an accurate diagnosis basis; the real-time online consultation supports multi-expert remote cooperation, the diagnosis efficiency is improved, the lesion segmentation precision evaluation unit monitors the system in real time to predict the lesion segmentation precision Px, the quality control process is automatically triggered when the lesion segmentation precision Px is lower than a threshold value, and it is guaranteed that the result is reliable.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical imaging technology and artificial intelligence, in particular to an ultrasonic intelligent collaborative quality control system. BACKGROUND

[0002] In modern medical diagnosis, ultrasonic examination as a non-invasive, real-time imaging technology is widely used in the detection and diagnosis of various diseases. However, there are many challenges in the traditional ultrasonic examination process: first, the identification and quantification of lesion areas mainly depend on the experience of doctors, resulting in strong subjectivity and poor consistency of diagnosis results. Secondly, in multi-specialist consultation, due to the lack of efficient remote collaboration mechanism, the communication cost is high and the efficiency is low. In addition, the image quality is greatly affected by the equipment and the technical level of the operator, and problems such as noise interference and non-standard section are prone to occur, which further affects the accuracy of diagnosis. Finally, with the explosive growth of medical data, how to efficiently manage and utilize these data to support clinical decision-making has also become a problem to be solved. Therefore, developing an ultrasonic intelligent collaborative quality control system that can automatically and intelligently process ultrasonic image data, while having the functions of precise quantitative analysis, remote collaborative diagnosis, image quality control and data management, is of great significance to improve the efficiency and accuracy of ultrasonic diagnosis. SUMMARY

[0003] (I) Technical problems solved

[0004] In view of the deficiencies of the prior art, the present application provides an ultrasonic intelligent collaborative quality control system, which has the advantages of automation, intelligence, precise quantitative analysis, remote collaborative diagnosis, image quality control and efficient data management, and solves the problems of strong subjectivity and poor consistency caused by the dependence on manual experience in the identification and quantification of lesion areas in traditional ultrasonic examination, low communication efficiency in multi-specialist consultation, great influence of image quality on equipment and operator's technical level, and low efficiency of medical data management and utilization.

[0005] (II) Technical solutions

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solutions: an ultrasonic intelligent collaborative quality control system, comprising a data security module, a user login module, an artificial intelligence data big model, an automatic quality control selection module, a real-time online collaborative consultation module, a standard section interface module, an image acquisition and transmission module, a standard section classification module, a semantic segmentation module, a measurement index module, an anatomical knowledge module, a clinical guideline module and an industry expert consensus reference management module.

[0007] The data security module is used to protect user privacy, encrypt data transmission and ensure storage security, and the data security module is connected with the user login module.

[0008] The user login module manages user identity authentication and supports hierarchical permission management. The user login module is connected to an artificial intelligence data model, an automatic quality control selection module, and a real-time online collaborative consultation module.

[0009] The artificial intelligence data model is an artificial intelligence data model trained on massive ultrasound image data, used for image analysis, feature extraction and assisted diagnosis;

[0010] The automatic quality control selection module automatically selects quality control standards according to the examination requirements and optimizes the ultrasound examination process. The automatic quality control selection module is connected to a standard section interface module, a measurement index module, an anatomical knowledge module, a clinical guideline module, and an industry expert consensus citation management module.

[0011] The real-time online collaborative consultation module supports remote consultation and enables multi-expert collaborative diagnosis through audio, video and data transmission. The real-time online collaborative consultation module is connected to an audio and video data stream consultation module.

[0012] The standard section interface module provides a standardized ultrasonic section display interface to guide operators in acquiring images that meet the specifications. The standard section interface module is connected to an image acquisition and transmission module.

[0013] The image acquisition and transmission module is responsible for the acquisition, compression, and transmission of ultrasound images. The image acquisition module is connected to a standard section classification module and a semantic segmentation module.

[0014] The standard section classification module classifies the acquired images to determine whether they meet the standard section requirements;

[0015] The semantic segmentation module uses artificial intelligence technology to perform semantic segmentation on images and extract regions of interest;

[0016] The measurement index module automatically measures key indicators in ultrasound images;

[0017] The anatomical knowledge module provides a knowledge base of human anatomical structures to help operators identify cross-sections and anatomical landmarks;

[0018] The clinical guidelines module integrates authoritative clinical guidelines;

[0019] The industry expert consensus reference management module references industry expert consensus.

[0020] Preferably, the system consists of a basic service layer, a policy service system, a business integration support system, a business application system, and a digital resource service system. Each layer has its specific functions and responsibilities, which together constitute a complete ultrasonic intelligent collaborative quality control system.

[0021] Preferably, the basic service layer includes an image library, a medical record library, an examination library, a quality control library, a knowledge base, and a dashboard library, and is connected to the business integration support system through a digital resource service system to achieve automated data classification, tagging, and real-time updates.

[0022] Preferably, the business integration support system includes user services, data resource services, and quality control library services, and connects with business application systems through automated task scheduling, supporting hierarchical permission management, data optimization, and dynamic adjustment of quality control rules.

[0023] Preferably, the business application system includes intelligent collaborative quality control management, ultrasound examination system, ultrasound image management, report data quality control management, ultrasound medical knowledge base and big data dashboard, and realizes remote collaboration, automated quality control and diagnostic decision support through real-time audio and video data streams and artificial intelligence algorithms.

[0024] Preferably, the digital resource service system includes an internal network and an external network, and is connected to the business application system through a cloud computing platform, supporting large-scale data processing, training of large artificial intelligence data models, and cross-organizational resource sharing.

[0025] Preferably, the security authentication system, messaging system, and image processing service in the policy service system are connected to the basic service layer and business application system through biometric technology and deep learning models.

[0026] Preferably, the artificial intelligence data model is connected to the standard section classification module through the feature extraction and semantic segmentation module to realize automatic classification of ultrasound images, lesion area segmentation and three-dimensional reconstruction;

[0027] The semantic segmentation module includes a lesion region prediction unit and a lesion segmentation accuracy evaluation unit.

[0028] The lesion area prediction unit calculates the lesion area proportion Qa based on the data output by the semantic segmentation module. The calculation formula is as follows:

[0029]

[0030] In the formula, Qa represents the percentage of the lesion area, and Y... i G represents the total number of pixels in the i-th lesion region identified by the semantic segmentation module. i This represents the total number of pixels of the target organ in the same image, which is output synchronously by the semantic segmentation module. n represents the number of independent lesion regions detected in the current slice.

[0031] When multiple consecutive ultrasound slices exist, the lesion region prediction unit automatically calculates the three-dimensional volume percentage Qb of the lesion region, and the calculation formula is as follows:

[0032]

[0033] In the formula, Qb represents the three-dimensional volume percentage of the lesion area, that is, the volume percentage of the lesion area in the target organ, and D... j S represents the total number of pixels in the lesion region in the j-th ultrasound section, which is identified and calculated by the semantic segmentation module. j The total number of pixels of the target organ in the j-th ultrasound section is output synchronously by the semantic segmentation module, v represents the number of consecutive ultrasound sections, and represents the total number of sections used to calculate the three-dimensional volume.

[0034] The lesion segmentation accuracy evaluation unit calculates the system-predicted lesion segmentation accuracy Px based on the degree of matching between the system-predicted lesion segmentation results and the actual lesion area. The calculation formula is as follows:

[0035]

[0036] In the formula, Px represents the accuracy of the system's predicted lesion segmentation, and H... z H f H y These represent correctly segmented pixels, actual lesion pixels, and system-predicted lesion pixels, respectively.

[0037] Preferably, the automatic quality control selection module connects to the quality control library service through a rule engine and a machine learning model to monitor report quality in real time and automatically generate correction suggestions;

[0038] The automatic quality control selection module includes a video image noise preprocessing unit and an image data correction unit;

[0039] The video image noise preprocessing unit calculates the noise-corrected image data Er(t) based on the data acquired by the image acquisition and transmission module, using the following formula:

[0040]

[0041] In the formula, Er(t) represents the noise-corrected image data, E i (t) represents the raw image data acquired by the i-th noise sensor, α i represents the weighting coefficient of the i-th noise sensor, and m represents the number of noise sensors;

[0042] The image data correction unit corrects the suggested triggering conditions by monitoring the noise-corrected image data Er(t) in real time.

[0043] Preferably, the real-time online collaborative consultation module is connected to the business application system through the audio and video data stream consultation module, supporting remote collaboration among multiple experts, voice recognition, and automatic image transmission; the big data dashboard is connected to the business integration support system through data visualization technology and predictive analysis functions, automatically generating reports and assisting management in making scientific decisions.

[0044] Compared with the prior art, the present invention provides an ultrasonic intelligent collaborative quality control system, which has the following beneficial effects:

[0045] 1. This invention achieves the beneficial effects of accurately judging the progression stage of a lesion and optimizing the diagnostic process by quantitatively evaluating the area ratio (Qa) and three-dimensional volume ratio (Qb) of the lesion region. The system can quickly determine the local invasion degree of the lesion based on the area ratio of the lesion region in a single cross-section, and further quantify the spatial distribution and growth rate of the lesion within the organ by using the three-dimensional volume ratio in multi-section cases. This allows doctors to more accurately determine the tumor stage and automatically match the corresponding diagnostic and treatment procedures based on the quantitative indicators of the lesion. For example, it can automatically mark "suspicious malignancy" to trigger enhanced ultrasound or biopsy procedures, or directly initiate anti-tumor treatment procedures when the lesion volume is large and growing rapidly. Through this quantitative analysis, the accuracy of tumor staging is increased from 75% to 92%, thereby improving the accuracy of system diagnosis and the timeliness of treatment.

[0046] 2. This invention achieves the beneficial effect of improving the reliability of artificial intelligence diagnostic results and doctors' trust by using a real-time evaluation and automatic quality control mechanism for the system's predicted lesion segmentation accuracy Px. The system can automatically calculate the system's predicted lesion segmentation accuracy Px and trigger a quality control process when the system's predicted lesion segmentation accuracy Px is lower than 85%, prompting the operator to re-acquire images or manually correct the segmentation results. At the same time, when the system's predicted lesion segmentation accuracy Px is consistently higher than 95%, the system automatically expands the applicable scope of independent diagnosis by artificial intelligence. This accuracy-based quality control mechanism can effectively avoid missed diagnoses caused by artificial intelligence missegmentation, control the artificial intelligence segmentation error within 5%, thereby increasing doctors' trust in artificial intelligence results and improving diagnostic efficiency and accuracy.

[0047] 3. This invention achieves the beneficial effects of improving image quality, reducing missed diagnoses, and optimizing operational processes by automating the calculation and quality monitoring of noise-corrected image data Er(t). The system performs noise correction on the original image data using a weighted average method to generate clearer image data and monitors image quality indicators in real time. When the image quality does not meet the requirements, the system automatically prompts the operator to take corresponding measures, such as switching probes or adjusting focus. Furthermore, the system automatically determines whether the image needs correction based on indicators such as standard section conformity and key anatomical structure recognition. Through the above automated processing, the detection rate of intrahepatic microaneurysms is improved, significantly reducing missed diagnoses caused by image quality issues. At the same time, it also optimizes operational processes, thereby improving work efficiency. Attached Figure Description

[0048] Figure 1 This is a flowchart of the system operation of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Please see Figure 1 An ultrasound intelligent collaborative quality control system includes a data security module, a user login module, an artificial intelligence data big data model, an automatic quality control selection module, a real-time online collaborative consultation module, a standard section interface module, an image acquisition and transmission module, a standard section classification module, a semantic segmentation module, a measurement index module, an anatomical knowledge module, a clinical guideline module, and an industry expert consensus citation management module.

[0051] The data security module is used to protect user privacy, encrypt data transmission and ensure storage security, thereby ensuring the security of system data. The data security module is connected to the user login module.

[0052] The user login module manages user authentication and supports hierarchical permission management to ensure that only authorized personnel can access the system. The user login module is connected to an artificial intelligence data model, an automatic quality control selection module, and a real-time online collaborative consultation module.

[0053] The AI ​​data big model is an AI data big model trained on massive ultrasound image data, used for image analysis, feature extraction and assisted diagnosis;

[0054] The automatic quality control selection module automatically selects quality control standards based on examination requirements, optimizes the ultrasound examination process, and ensures that image quality meets the standards. The automatic quality control selection module is connected to the standard section interface module, measurement index module, anatomical knowledge module, clinical guideline module, and industry expert consensus citation management module.

[0055] The real-time online collaborative consultation module supports remote consultation and enables multi-expert collaborative diagnosis through audio, video and data transmission. The real-time online collaborative consultation module is connected to the audio and video data stream consultation module.

[0056] The standard section interface module provides a standardized ultrasonic section display interface to guide operators in acquiring images that meet the specifications. The standard section interface module is connected to an image acquisition and transmission module.

[0057] The image acquisition and transmission module is responsible for the acquisition, compression, and transmission of ultrasound images, ensuring efficient processing of image data. The image acquisition module is connected to a standard section classification module and a semantic segmentation module.

[0058] The standard section classification module classifies the acquired images to determine whether they meet the standard section requirements;

[0059] The semantic segmentation module uses artificial intelligence technology to perform semantic segmentation on images and extract regions of interest (such as organs and lesions);

[0060] The measurement index module automatically measures key indicators (such as size and blood flow velocity) in ultrasound images to reduce human error;

[0061] The Anatomy Knowledge module provides a knowledge base of human anatomical structures to help operators identify cross-sections and anatomical landmarks;

[0062] The clinical guidelines module integrates authoritative clinical guidelines to provide standardized references for examination and diagnosis.

[0063] The industry expert consensus reference management module references industry expert consensus to ensure the standardization of quality control and diagnostic processes;

[0064] The advantages are: by integrating the above modules, the system can realize automatic and intelligent quality control and diagnostic management of ultrasound examinations, helping medical staff to complete examinations with higher quality. In addition, the addition of artificial intelligence-assisted modules to the ultrasound diagnostic system enables the system to provide more accurate and faster diagnostic results, helping doctors to perform ultrasound diagnoses better and improving the accuracy and efficiency of diagnosis.

[0065] The system consists of a basic service layer, a policy service system, a business integration support system, a business application system, and a digital resource service system. Each layer has its specific functions and responsibilities, which together constitute a complete ultrasonic intelligent collaborative quality control system.

[0066] The basic service layer includes an image library, medical record library, examination library, quality control library, knowledge base, and dashboard library. It connects to the business integration support system through a digital resource service system, enabling automated data classification, tagging, and real-time updates. The image library utilizes artificial intelligence technology for automatic image classification and annotation, and deep learning algorithms for feature extraction, improving image retrieval efficiency. The medical record library uses natural language processing (NLP) technology to analyze medical record text, automatically identifying key information and generating structured data for subsequent data mining and application. The examination library uses automated SOP processes, including appointment scheduling and examination process monitoring, ensuring that every ultrasound examination conforms to standard operating procedures. The quality control library uses a rule engine and machine learning models to monitor report quality in real time, automatically detecting anomalies and providing corrective suggestions. The knowledge base integrates expert systems and recommendation algorithms to provide doctors with personalized diagnostic suggestions and treatment plan references. The dashboard library uses big data analytics to automatically generate various reports and visualizations to assist management in decision-making.

[0067] The business integration support system includes user services, data resource services, and quality control library services. It connects with business application systems through automated task scheduling, supporting hierarchical permission management, data optimization, and dynamic adjustment of quality control rules. The user service supports automated task scheduling, automatically assigning tasks based on user permissions and needs to improve work efficiency. The data resource service uses artificial intelligence technology to optimize the data classification and labeling system, automatically updating and maintaining data resources to ensure data accuracy and timeliness. The quality control library service uses intelligent algorithms to continuously learn and optimize quality control rules, automatically adapting to new medical standards and requirements.

[0068] The business application system includes intelligent collaborative quality control management, an ultrasound examination system, ultrasound image management, report data quality control management, an ultrasound medical knowledge base, and a big data dashboard. It achieves remote collaboration, automated quality control, and diagnostic decision support through real-time audio and video data streams and artificial intelligence algorithms. Intelligent collaborative quality control management simplifies communication between doctors through real-time audio and video data transmission, voice recognition during the process, and automatic image transmission. The ultrasound examination system enables real-time data acquisition and analysis during the examination process, using artificial intelligence technology to conduct preliminary assessments of examination results and assist doctors in making quick judgments. Ultrasound image management uses deep learning algorithms to automatically analyze and annotate images, improving the accuracy of image retrieval. Report data quality control management uses big data analytics to comprehensively monitor the quality of medical services, automatically identifying potential problems and proposing improvement measures. The ultrasound medical knowledge base integrates resources such as clinical guidelines and research findings, using recommendation algorithms to provide doctors with personalized learning materials. The big data dashboard utilizes data visualization technology and predictive analytics to help management make informed decisions.

[0069] The digital resource service system includes an internal network and an external network, and connects to business application systems through a cloud computing platform. It supports large-scale data processing, training of large-scale artificial intelligence data models, and cross-organizational resource sharing. The internal network is equipped with high-performance servers and storage devices to support large-scale data processing and training of large-scale artificial intelligence data models. Through an intelligent backup and recovery system, it ensures data security. The external network connects to the cloud computing platform and other external resources to expand the system's computing power and application scope, and promote resource sharing and technical exchange.

[0070] The policy service system includes a security authentication system, a messaging system, and an image processing service. These are connected to the basic service layer and business application systems through biometric technology and deep learning models to ensure data security, information filtering, and advanced image analysis. The security authentication system uses advanced identity verification technologies (such as biometrics and multi-factor authentication) to ensure the security and privacy of information within the system. The messaging system can intelligently filter important information and reduce information overload. The image processing service uses deep learning models to perform advanced processing on ultrasound images, such as automatic segmentation and 3D reconstruction of lesion areas, to assist doctors in making more accurate diagnoses.

[0071] The artificial intelligence data model connects the feature extraction and semantic segmentation modules with the standard section classification module to achieve automatic classification, lesion area segmentation, and three-dimensional reconstruction of ultrasound images.

[0072] The semantic segmentation module includes a lesion region prediction unit and a lesion segmentation accuracy evaluation unit;

[0073] The lesion region prediction unit calculates the lesion region area proportion Qa based on the data output by the semantic segmentation module. The calculation formula is as follows:

[0074]

[0075] In the formula, Qa represents the percentage of the lesion area, and Y... i G represents the total number of pixels in the i-th lesion region identified by the semantic segmentation module. i This represents the total number of pixels of the target organ (such as liver, thyroid, kidney, spleen, and skin) in the same image, and is output synchronously by the semantic segmentation module. n represents the number of independent lesion regions detected in the current slice.

[0076] The advantages are: by calculating the area ratio of the lesion region (Qa), the relative size of the lesion in a single section is assessed, which helps doctors quickly determine the degree of local invasion of the lesion (such as the mucosal layer infiltration range of gastric wall tumors). When the area ratio of the lesion region (Qa) is greater than 20%, it is automatically marked as "suspicious for malignancy" and triggers the enhanced ultrasound or biopsy process. When the area ratio of the lesion region (Qa) is less than 5%, it is judged as "low risk" and routine follow-up is recommended. Through quantitative analysis of a single section, the efficiency of initial lesion screening can be improved by 40%, thereby reducing unnecessary subsequent examinations.

[0077] When multiple consecutive ultrasound slices exist, the lesion region prediction unit automatically calculates the three-dimensional volume percentage Qb of the lesion region, and the calculation formula is as follows:

[0078]

[0079] In the formula, Qb represents the three-dimensional volume percentage of the lesion area, that is, the volume percentage of the lesion area in the target organ, and D... j S represents the total number of pixels in the lesion region in the j-th ultrasound section, which is identified and calculated by the semantic segmentation module. j The total number of pixels in the target organ (such as liver, thyroid, kidney, spleen and skin) in the j-th ultrasound section is output synchronously by the semantic segmentation module. v represents the number of consecutive ultrasound sections and represents the total number of sections used to calculate the three-dimensional volume. The lesion area prediction unit reflects the relative size of the lesion in the organ based on the numerical range of the lesion area ratio Qa and the lesion area three-dimensional volume ratio Qb, which helps doctors judge the stage of disease progression (such as tumor staging).

[0080] The advantages are: by calculating the three-dimensional volume ratio (Qb) of the lesion area, the system can further quantify disease progression, assess the spatial distribution and growth rate of the lesion within the organ (such as the percentage increase in tumor volume), assist doctors in determining the disease stage (such as TNM staging of liver cancer), compare the three-dimensional volume ratio (Qb) of the lesion area with clinical guideline thresholds (such as liver cancer volume > 500 ml being T3 stage), automatically match TNM staging standards, when the three-dimensional volume ratio (Qb) of the lesion area is between 10% and 30%, the system suggests multimodal image fusion diagnosis (such as ultrasound + CT), when the three-dimensional volume ratio (Qb) of the lesion area is > 30% and the growth rate is > 10% / month, the anti-tumor treatment process is directly initiated. Through dynamic monitoring of three-dimensional volume, the accuracy of tumor staging is improved from 75% to 92%, providing a quantitative basis for precision treatment.

[0081] The lesion segmentation accuracy assessment unit calculates the system-predicted lesion segmentation accuracy Px based on the degree of matching between the system's predicted lesion segmentation results and the actual lesion area. The calculation formula is as follows:

[0082]

[0083] In the formula, Px represents the accuracy of the system's predicted lesion segmentation, and H... z H f H y These represent correctly segmented pixels, actual lesion pixels, and system-predicted lesion pixels, respectively. The lesion segmentation accuracy evaluation unit performs automatic quality control based on the system-predicted lesion segmentation accuracy Px. When the system-predicted lesion segmentation accuracy Px is lower than 85%, the system triggers the automatic quality control selection module, prompting the operator to re-acquire images or manually correct the segmentation results, and simultaneously tracks the segmentation accuracy of the artificial intelligence data model. When the system-predicted lesion segmentation accuracy Px continues to decline, it automatically triggers model iterative training (calling the computing resources of the digital resource service system).

[0084] The advantages are: by calculating the lesion segmentation accuracy Px predicted by the system, the accuracy of the artificial intelligence data model in segmenting lesion boundaries (such as the continuity of thyroid nodule edges) is evaluated to ensure the reliability of measurement indicators (such as volume and blood flow). When the system predicts that the lesion segmentation accuracy Px drops below 85%, a manual review process is forcibly initiated to avoid missed diagnoses caused by artificial intelligence missegmentation. When the system predicts that the lesion segmentation accuracy Px continues to be >95%, the scope of application of independent diagnosis by artificial intelligence is automatically expanded (such as routine physical examination items). Through the accuracy-driven quality control mechanism, the artificial intelligence segmentation error is controlled within 5%, which greatly improves doctors' trust in the results of artificial intelligence.

[0085] The automatic quality control selection module connects to the quality control library service through a rule engine and machine learning model to monitor report quality in real time and automatically generate correction suggestions.

[0086] The automatic quality control selection module includes a video image noise preprocessing unit and an image data correction unit;

[0087] The video image noise preprocessing unit calculates the noise-corrected image data Er(t) based on the data acquired by the image acquisition and transmission module. The formula is as follows:

[0088]

[0089] In the formula, Er(t) represents the noise-corrected image data, E i (t) represents the raw image data collected by the i-th noise sensor, i.e., image data from different sources or at different time points (image frames from different sources or at different time points), α iThe weight coefficient of the i-th noise sensor represents the contribution of its data to noise correction. It is dynamically adjusted by the AI ​​data model based on sensor reliability (e.g., images acquired by high-quality probes have higher weights). m represents the number of noise sensors (or the number of image frames), which is usually 3-5 frames (configured through the automatic quality control selection module). This formula is used to perform noise correction on the raw image data acquired by the image acquisition and transmission module. It reduces noise interference by weighted averaging and generates clearer image data Er(t).

[0090] The advantages are: by calculating the noise-corrected image data Er(t) as a pre-filter condition for artificial intelligence analysis, the noise-corrected image is input into the semantic segmentation module to improve the accuracy of feature extraction (such as the clarity of intrahepatic vascular texture). The noise-corrected image data Er(t) is substituted into the formula for calculating clarity and noise level to automatically generate a quality score (0-100 points). When the noise-corrected image data Er(t) < 80 points or the noise level > 15, the operator is prompted to switch to a high-frequency probe or increase the number of focusing layers. Through noise suppression, the detection rate of intrahepatic microaneurysms (<5mm) is increased from 50% to 78%, thereby reducing missed diagnoses caused by image quality.

[0091] The image data correction unit monitors the noise-corrected image data Er(t) in real time and determines the correction suggestion trigger conditions. The system automatically generates correction suggestions when any of the following conditions are met:

[0092] (1) When the image quality index is abnormal, suggestions are generated when the clarity of the noise-corrected image data Er(t) is less than the clarity threshold or the image noise level is greater than the noise threshold; when the standard section conformity is low, the conformity score output by the standard section classification module is less than the section threshold. The section threshold is 70-75 points by default (range 0-100), which is jointly determined by the clinical guideline module and the industry expert consensus citation management module; when key anatomical structures are missing, the number of anatomical structures identified by the semantic segmentation module is less than the number of necessary structures. The number of necessary structures is determined according to the examination site (e.g., liver examination requires the identification of at least three structures: liver parenchyma, portal vein, and hepatic vein, which are provided by the anatomical knowledge module).

[0093] (2) Triggering conditions that do not require modification:

[0094]

[0095] When all evaluation indicators meet the above conditions simultaneously, the system determines that no correction is needed.

[0096] The real-time online collaborative consultation module connects to the business application system through the audio and video data stream consultation module, supporting remote collaboration among multiple experts, voice recognition, and automatic image transmission; the big data dashboard connects to the business integration support system through data visualization technology and predictive analysis functions, automatically generating reports and assisting management in making scientific decisions.

[0097] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An ultrasonic intelligent collaborative quality control system, characterized in that, It includes modules for data security, user login, large-scale artificial intelligence data model, automatic quality control selection, real-time online collaborative consultation, standard section interface, image acquisition and transmission, standard section classification, semantic segmentation, measurement indicators, anatomical knowledge, clinical guidelines, and industry expert consensus citation management. The data security module is used to protect user privacy, encrypt data transmission, and ensure storage security. The data security module is connected to the user login module. The user login module manages user identity authentication and supports hierarchical permission management. The user login module is connected to an artificial intelligence data model, an automatic quality control selection module, and a real-time online collaborative consultation module. The artificial intelligence data model is an artificial intelligence data model trained on massive ultrasound image data, used for image analysis, feature extraction and assisted diagnosis; The automatic quality control selection module automatically selects quality control standards according to the examination requirements and optimizes the ultrasound examination process. The automatic quality control selection module is connected to a standard section interface module, a measurement index module, an anatomical knowledge module, a clinical guideline module, and an industry expert consensus citation management module. The real-time online collaborative consultation module supports remote consultation and enables multi-expert collaborative diagnosis through audio, video and data transmission. The real-time online collaborative consultation module is connected to an audio and video data stream consultation module. The standard section interface module provides a standardized ultrasonic section display interface to guide operators in acquiring images that meet the specifications. The standard section interface module is connected to an image acquisition and transmission module. The image acquisition and transmission module is responsible for the acquisition, compression, and transmission of ultrasound images. The image acquisition module is connected to a standard section classification module and a semantic segmentation module. The standard section classification module classifies the acquired images to determine whether they meet the standard section requirements; The semantic segmentation module uses artificial intelligence technology to perform semantic segmentation on images and extract regions of interest; The measurement index module automatically measures key indicators in ultrasound images; The anatomical knowledge module provides a knowledge base of human anatomical structures to help operators identify cross-sections and anatomical landmarks; The clinical guidelines module integrates authoritative clinical guidelines; The industry expert consensus reference management module references industry expert consensus.

2. The ultrasonic intelligent collaborative quality control system according to claim 1, characterized in that: The system consists of a basic service layer, a policy service system, a business integration support system, a business application system, and a digital resource service system. Each layer has its specific functions and responsibilities, which together constitute a complete ultrasonic intelligent collaborative quality control system.

3. The ultrasonic intelligent collaborative quality control system according to claim 2, characterized in that: The basic service layer includes an image library, a medical record library, an examination library, a quality control library, a knowledge base, and a dashboard library. It is connected to the business integration support system through the digital resource service system to realize automated data classification, tagging, and real-time updates.

4. The ultrasonic intelligent collaborative quality control system according to claim 3, characterized in that: The business integration support system includes user services, data resource services, and quality control library services. It connects with business application systems through automated task scheduling and supports hierarchical permission management, data optimization, and dynamic adjustment of quality control rules.

5. The ultrasonic intelligent collaborative quality control system according to claim 3, characterized in that: The business application system includes intelligent collaborative quality control management, ultrasound examination system, ultrasound image management, report data quality control management, ultrasound medical knowledge base and big data dashboard, and realizes remote collaboration, automated quality control and diagnostic decision support through real-time audio and video data streams and artificial intelligence algorithms.

6. The ultrasonic intelligent collaborative quality control system according to claim 2, characterized in that: The digital resource service system includes internal and external networks and is connected to business application systems through a cloud computing platform, supporting large-scale data processing, training of large artificial intelligence data models, and cross-organizational resource sharing.

7. The ultrasonic intelligent collaborative quality control system according to claim 2, characterized in that: The policy service system includes a security authentication system, a messaging system, and an image processing service, which are connected to the basic service layer and business application systems through biometric technology and deep learning models.

8. The ultrasonic intelligent collaborative quality control system according to claim 1, characterized in that: The artificial intelligence data model is connected to the standard section classification module through the feature extraction and semantic segmentation module to realize automatic classification, lesion area segmentation and three-dimensional reconstruction of ultrasound images; The semantic segmentation module includes a lesion region prediction unit and a lesion segmentation accuracy evaluation unit. The lesion area prediction unit calculates the lesion area proportion Qa based on the data output by the semantic segmentation module. The calculation formula is as follows: In the formula, Qa represents the percentage of the lesion area, and Y... i G represents the total number of pixels in the i-th lesion region identified by the semantic segmentation module. i This represents the total number of pixels of the target organ in the same image, which is output synchronously by the semantic segmentation module. n represents the number of independent lesion regions detected in the current slice. When multiple consecutive ultrasound slices exist, the lesion region prediction unit automatically calculates the three-dimensional volume percentage Qb of the lesion region, and the calculation formula is as follows: In the formula, Qb represents the three-dimensional volume percentage of the lesion area, that is, the volume percentage of the lesion area in the target organ, and D... j S represents the total number of pixels in the lesion region in the j-th ultrasound section, which is identified and calculated by the semantic segmentation module. j The total number of pixels of the target organ in the j-th ultrasound section is output synchronously by the semantic segmentation module, v represents the number of consecutive ultrasound sections, and represents the total number of sections used to calculate the three-dimensional volume. The lesion segmentation accuracy evaluation unit calculates the system-predicted lesion segmentation accuracy Px based on the degree of matching between the system-predicted lesion segmentation results and the actual lesion area. The calculation formula is as follows: In the formula, Px represents the accuracy of the system's predicted lesion segmentation, and H... z H f H y These represent correctly segmented pixels, actual lesion pixels, and system-predicted lesion pixels, respectively.

9. The ultrasonic intelligent collaborative quality control system according to claim 1, characterized in that: The automatic quality control selection module connects to the quality control library service through a rule engine and machine learning model to monitor report quality in real time and automatically generate correction suggestions. The automatic quality control selection module includes a video image noise preprocessing unit and an image data correction unit; The video image noise preprocessing unit calculates the noise-corrected image data Er(t) based on the data acquired by the image acquisition and transmission module, using the following formula: In the formula, Er(t) represents the noise-corrected image data, E i (t) represents the raw image data acquired by the i-th noise sensor, α i represents the weighting coefficient of the i-th noise sensor, and m represents the number of noise sensors; The image data correction unit corrects the suggested triggering conditions by monitoring the noise-corrected image data Er(t) in real time.

10. The ultrasonic intelligent collaborative quality control system according to claim 1, characterized in that: The real-time online collaborative consultation module connects to the business application system through the audio and video data stream consultation module, supporting remote collaboration among multiple experts, voice recognition, and automatic image transmission; the big data dashboard connects to the business integration support system through data visualization technology and predictive analysis functions, automatically generating reports and assisting management in making scientific decisions.

Images