Real-time marking method and system on remote ultrasonic equipment picture

Abstract

The invention relates to the technical field of ultrasound, in particular to a method and a system for marking pictures of remote ultrasonic equipment in real time, and the method comprises the following steps: acquiring a real-time ultrasonic image acquired by the ultrasonic equipment; carrying out image preprocessing on the obtained ultrasonic image, wherein the image preprocessing comprises noise removal, image enhancement and region segmentation; transmitting the preprocessed image to an expert cooperation marking platform; multi-expert marking is carried out in the expert cooperation marking platform, and experts distribute different areas in marked images according to tasks; real-time conflict detection is carried out on the marking result of each expert, and if a marking conflict is detected, prompt information is generated and the experts are guided to carry out correction; based on the consistency detection algorithm, the consistency score of each expert mark is generated, the marking result is automatically optimized according to the consistency score, through a multi-expert remote cooperation marking mechanism, the limitation of a single expert mark in the prior art is overcome, and the accuracy and consistency of ultrasonic image marking are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of ultrasound, and more particularly to a method and system for real-time marking on a screen of a remote ultrasound device. Background Art

[0002] In the field of modern medical imaging diagnosis, ultrasound images are widely used in the examination and diagnosis of various diseases as a common non-invasive diagnostic tool. Real-time acquisition and analysis of ultrasound images can help doctors make quick judgments, especially in the fields of emergency, obstetrics, cardiovascular, etc. However, since the quality of ultrasound images is affected by many factors (such as patient body shape, image resolution, operation technology, etc.), the diagnostic accuracy of ultrasound images is often limited. Traditional ultrasound image marking methods usually rely on a single expert to mark the image. The marking content usually includes the location, size, shape, etc. of the lesion area. Since ultrasound images may contain multiple areas that need to be diagnosed and the image quality may be poor, the marking of a single expert is often highly subjective and limited. When faced with complex lesions or noise in the image, the marking of a single expert may be biased, resulting in inaccurate final diagnosis results, especially in some remote areas. Due to the lack of sufficient expert resources, patients may not be able to get professional diagnosis opinions in time. This has led to the inability to effectively guarantee the timeliness and accuracy of remote ultrasound diagnosis. Summary of the invention

[0003] To this end, the present invention provides a method and system for real-time marking on the screen of a remote ultrasound device, aiming to overcome the limitations of single expert marking in the prior art through a multi-expert remote collaborative marking mechanism, improve the accuracy and consistency of ultrasound image marking, and solve the problems of marking deviation, marking conflict and low efficiency of remote consultation in traditional methods.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A method for real-time marking on a remote ultrasound device screen comprises the following steps:

[0006] S1, obtaining a real-time ultrasound image collected by an ultrasound device;

[0007] S2, performing image preprocessing on the acquired ultrasound image, including noise removal, image enhancement, and region segmentation;

[0008] S3, transmitting the pre-processed image to the expert collaborative labeling platform;

[0009] S4, perform multi-expert labeling in the expert collaborative labeling platform, and experts label different areas in the image according to task assignment;

[0010] S5, performing real-time conflict detection on the marking results of each expert. If a marking conflict is detected, a prompt message is generated to guide the expert to make corrections;

[0011] S6. Generate a consistency score for each expert’s marking based on the consistency detection algorithm, and automatically optimize the marking results according to the consistency score;

[0012] S7. After the marking is completed, the final marking result is generated and submitted to the chief reviewer for review. After the review is passed, a diagnosis report is generated.

[0013] A system for real-time marking method on a remote ultrasound device screen, the system comprising:

[0014] Ultrasonic image acquisition module, used for real-time acquisition of ultrasonic images of ultrasonic equipment;

[0015] An image preprocessing module is used to remove noise, enhance the image and perform region segmentation on the acquired ultrasound image;

[0016] Expert collaborative labeling platform, used to support multi-expert collaborative labeling, where experts view and label real-time ultrasound images through a graphical interface;

[0017] The marking area allocation module is used to automatically allocate the marking area of ​​each expert according to the predetermined task to avoid marking overlap;

[0018] The conflict detection module is used to detect in real time whether the areas marked by the experts overlap. If a conflict is found, a prompt message is generated to guide the experts to make corrections;

[0019] The consistency detection module is used to generate a consistency score for each expert's marking and automatically optimize the marking results based on the consistency score;

[0020] The review module is used by the chief reviewer to review the marking results and generate a diagnosis report after the review is passed.

[0021] Further: the expert collaborative marking platform also includes:

[0022] The marking tool module has a variety of marking tools, and experts can select appropriate tools to mark images according to actual needs;

[0023] Real-time synchronous display module, the expert's marking operation is reflected in the image in real time, and other experts can view the latest marking progress synchronously, ensuring transparency and real-time collaboration in the marking process.

[0024] Further: the marking area allocation module includes:

[0025] Dynamic region allocation unit: Depending on the image content, the system can automatically identify key regions in the image and dynamically allocate labeling tasks based on the labeling requirements of these key regions;

[0026] Area allocation strategy unit: By setting area priority and marking order, it prevents multiple experts from marking the same area at the same time. The expert's marking area will be clearly marked when it is allocated, and the logo will be displayed on the image to prevent overlap;

[0027] Automatic marking area locking unit: After the task is assigned, the marking area of ​​each expert will be automatically locked. The locked area will be displayed as a "lock box" on the image. During the marking process, other experts cannot modify or mark the area, ensuring the independence of the marked area.

[0028] Further: the conflict detection module includes:

[0029] Marking area recognition and positioning unit: During the marking process, each area marked by the expert will be automatically recognized by the system, and the coordinates of the marked area will be recorded in real time;

[0030] Image registration and marker coordinate matching unit: The area marked by each expert is matched in the image through the image registration algorithm. The registration process ensures that the areas marked by different experts can be correctly aligned by extracting key feature points in the image to facilitate the detection of overlapping areas. The registered marker coordinates will be used to determine the overlap of different marked areas. If the intersection of multiple marked areas is greater than the set overlap threshold, it is considered a conflict and the system will generate a conflict warning.

[0031] Furthermore: the conflict detection module also includes a feedback unit for marking conflicts: when the system detects a marking conflict, it will send a warning message to the expert through the platform's instant message or pop-up prompt. In the prompt message, the conflict area is highlighted and the intersection with the marked area of ​​other experts is pointed out. The expert can correct the mark according to the feedback information of the platform, or communicate with other experts to resolve the conflict.

[0032] Further: the consistency detection module includes:

[0033] Consistency evaluation unit: automatically evaluates the consistency between expert marking results by calculating the overlap, shape similarity and pixel-level consistency of multiple expert-marked areas, and generates a consistency score;

[0034] Consistency optimization unit: Based on the consistency score of the expert marking results, the position, shape, and size of the marked area will be fine-tuned, and the marking results will be automatically optimized by weighted average. The optimized marked area will be displayed as an optimized box in the image;

[0035] Consistency score feedback unit: This unit tracks the consistency score of the expert in the marking process in real time, provides real-time feedback, and provides prompt information to the expert based on the consistency score of the marking, prompting the expert to make necessary adjustments during the marking process to improve the consistency of the marking results.

[0036] Consistency report generation unit: This unit generates a final consistency assessment report, which lists in detail the consistency score, deviation analysis, optimization suggestions and final optimization results of each expert's marking. The report will clearly show the areas with low marking consistency and provide improvement plans to help experts make further corrections.

[0037] Further: the consistency evaluation unit calculates the overlap ratio between the marked area of ​​each expert and the marked areas of other experts based on a geometric overlap algorithm, thereby evaluating the consistency of the marking;

[0038] The pixel comparison algorithm of structural similarity index is used to compare the regions marked by different experts at pixel level to evaluate their marking consistency.

[0039] Use contour detection and matching algorithms to compare the boundaries marked by experts and evaluate the consistency of the marked boundaries;

[0040] Further: the audit module includes:

[0041] Review task assignment unit: This unit is used to assign the marking results to the chief reviewer for review. It automatically assigns the appropriate chief reviewer according to the task type, expert field and workload, and ensures that each review task is assigned to an expert with relevant experience and qualifications.

[0042] Review process monitoring unit: This unit monitors the review process of the marking results by the chief reviewer in real time, records the review opinions and modification suggestions. If the marking results are found to have obvious errors or parts that need to be adjusted during the review process, the system will automatically remind the chief reviewer to make modifications;

[0043] Review feedback unit: When the chief reviewer reviews the marking results, the system will collect and feedback the reviewer's opinions, including the accuracy and consistency of the marking and necessary improvement suggestions. The review opinions will be fed back to the marking expert in the form of a report, and modification guidance will be provided;

[0044] Review confirmation and revision unit: The marking results that have been reviewed or modified by the chief expert will be submitted to the system for final confirmation. If the marking results need further revision, the system will guide the experts to modify and resubmit for review. The marking results that have passed the review will be locked and a final report will be generated.

[0045] Diagnostic report generation unit: Once the review is passed, the system will generate a formal diagnostic report based on the marking results. The report will list in detail each marked area, the marking expert's feedback and review opinions, summarize the diagnostic conclusions, and provide further diagnosis and treatment recommendations for key areas.

[0046] Further: the diagnostic report generation unit also includes a report storage and management module: the generated diagnostic reports will be stored in a database and classified and managed according to the patient or device number to ensure that the reports are easy to find, track and archive historical records;

[0047] Report update and correction module: If errors or new information are found during the subsequent marking or review process, the system can update and correct the report and generate a revised report. This module records the history of all version updates to ensure the transparency and traceability of the report modification process.

[0048] Compared with the shortcomings of the prior art, the beneficial effects of the present invention are:

[0049] Through the multi-expert collaborative marking mechanism, the professional knowledge and experience of multiple experts can be gathered to ensure more accurate marking of key areas in ultrasound images, thereby improving the accuracy of diagnostic results.

[0050] In particular, the use of automated marking area allocation and real-time conflict detection functions effectively avoids overlapping or conflicting areas when multiple experts are marking, reduces the occurrence of marking errors, and improves the efficiency and quality of remote marking collaboration.

[0051] In particular, a consistency score for each expert's marking is automatically generated through a consistency detection algorithm, and the marking results are optimized based on the score, ensuring that the marking results of different experts are highly consistent, thereby improving the reliability of the diagnosis report.

[0052] In particular, the efficiency of collaboration between experts has been greatly improved through automatic allocation of marked areas, real-time conflict detection and consistency optimization. Especially in remote ultrasound consultations, it can effectively overcome the limitations brought by the physical distance between experts, reduce the risk of misdiagnosis, and ensure the quality of diagnosis.

[0053] In particular, the combination of automated marking area allocation, real-time conflict detection and consistency optimization realizes the intelligent and automated marking of ultrasound images, enhances the operability and popularity of telemedicine, and is particularly suitable for areas or telemedicine environments that lack ultrasound diagnostic experts. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a flow chart of the method of the present invention;

[0055] Figure 2 Schematic diagram of the modules of the system of the present invention. DETAILED DESCRIPTION

[0056] Reference Figure 1 to Figure 2 The present invention is further described.

[0057] A method for real-time marking on a remote ultrasound device screen comprises the following steps:

[0058] The first step is to obtain real-time ultrasound images collected by the ultrasound device. The ultrasound device collects real-time ultrasound images of the patient's body through sensors. The images may include different areas and depths. These image data will be transmitted to the marking system in a digital way for subsequent processing and analysis.

[0059] The second step: image preprocessing of the acquired ultrasound image, including noise removal, image enhancement, and region segmentation: Through the image preprocessing module, the system first removes noise from the image and enhances the image contrast, making the important areas more visible. Region segmentation is used to distinguish different areas in the ultrasound image (such as tissues, organs, lesions, etc.), providing a basis for marking and analysis.

[0060] Step 3: Transfer the pre-processed images to the expert collaborative labeling platform;

[0061] Step 4: Multi-expert labeling is performed in the expert collaborative labeling platform, and experts label different areas in the image according to task assignments;

[0062] Step 5: Perform real-time conflict detection on the marking results of each expert. If a marking conflict is detected, a prompt message is generated to guide the expert to make corrections;

[0063] Step 6: Generate a consistency score for each expert’s marking based on the consistency detection algorithm, and automatically optimize the marking results based on the consistency score;

[0064] Step 7: After marking is completed, the final marking results are generated and submitted to the chief reviewer for review. After the review is passed, a diagnostic report is generated.

[0065] A system for implementing a real-time marking method on a remote ultrasound device screen, comprising:

[0066] Ultrasonic image acquisition module, used to collect ultrasonic images of ultrasonic equipment in real time: This module collects real-time image data from the patient's body through ultrasonic sensors, and converts these data into digital signals for processing and transmission. During the acquisition process, the ultrasonic equipment can adjust different frequencies, angles and acquisition depths to obtain images at different levels.

[0067] Image preprocessing module, used to remove noise, enhance image and segment regions of acquired ultrasound images: The preprocessing module uses various image processing algorithms (such as filtering algorithms, contrast enhancement algorithms and region segmentation algorithms) to process ultrasound images. These processes ensure that the labeling process is not affected by image quality and that each region in the image can be clearly displayed.

[0068] Expert collaborative marking platform is used to support multi-expert collaborative marking. Experts view and mark real-time ultrasound images through a graphical interface: Expert collaborative marking platform provides a collaborative environment for multiple experts. Each expert logs in to the platform and uses the marking tools provided by the platform (such as pen, frame selection tool, etc.) to mark the image. The platform supports multiple people to operate at the same time, and the marking content and progress of each expert will be synchronized to the platform in real time to ensure smooth team collaboration.

[0069] The expert collaborative marking platform also includes a marking tool module and a real-time synchronous display module.

[0070] The marking tool module has a variety of marking tools. Experts can select appropriate tools to mark images according to actual needs: This module provides various marking tools, including drawing tools (such as pen, straight line, rectangle, circle), text annotation tools, etc. Experts can flexibly select and mark on the image. In addition, it also provides functions such as zooming in, zooming out, and adjusting the viewing angle, so that experts can operate more accurately.

[0071] The real-time synchronous display module allows the expert's marking operations to be reflected in the image in real time, and other experts can view the latest marking progress synchronously, ensuring transparency and real-time collaboration in the marking process: This module uses image synchronization technology to ensure that each expert's marking operations are instantly reflected in the interface of other experts, avoiding misoperation caused by operation delays or information asymmetry.

[0072] The marking area allocation module is used to automatically allocate the marking area of ​​each expert according to the predetermined task to avoid overlapping marking: This module automatically allocates marking tasks to different experts by analyzing the image content and the expertise of the experts in the field. The system will ensure that the marking area of ​​each expert does not overlap and matches their task capabilities based on factors such as the complexity of the marking area and the expert's experience. The assigned marking area will be highlighted, and once an expert starts marking, the marking areas of other experts will be automatically locked to prevent overlap.

[0073] The conflict detection module is used to detect in real time whether the areas marked by experts overlap. If a conflict is found, a prompt message is generated to guide the expert to make corrections: This module monitors and analyzes the marking progress of each expert in real time, uses the image registration algorithm to compare the positions of the marked areas of each expert, and detects whether there is any overlap in marking. If a conflict is detected, the system will immediately generate a warning message, highlight the conflicting area, and guide the expert to make corrections.

[0074] The consistency detection module is used to generate a consistency score for each expert's marking and automatically optimize the marking results based on the consistency score: This module evaluates the consistency of each expert's marking results and calculates the consistency score by calculating indicators such as overlap and shape matching. The system automatically optimizes the marked area based on the consistency score to ensure that the final marking result reaches the optimal state.

[0075] The review module is used by the chief reviewer to review the marking results and generate a diagnosis report after the review is passed: After the marking work is completed, the chief reviewer will review the final marking results. This module supports reviewing whether the marked area is accurate and meets the clinical diagnosis standards, and makes modification suggestions for errors or inconsistencies in the expert's marking. After the review is completed, the system will generate a final diagnosis report and feedback the marking results and review opinions to the relevant experts.

[0076] The tag zone allocation module includes:

[0077] Dynamic region allocation unit: Depending on the image content, the system can automatically identify key regions in the image and dynamically allocate labeling tasks based on the labeling requirements of these key regions;

[0078] Area allocation strategy unit: By setting area priority and marking order, it prevents multiple experts from marking the same area at the same time. The expert's marking area will be clearly marked when it is allocated, and the logo will be displayed on the image to prevent overlap;

[0079] Automatic marking area locking unit: After the task is assigned, the marking area of ​​each expert will be automatically locked. The locked area will be displayed as a "lock box" on the image. During the marking process, other experts cannot modify or mark the area, ensuring the independence of the marked area.

[0080] The conflict detection module includes:

[0081] Marked area identification and positioning unit: The core task of this unit is to identify the areas marked by each expert in the image in real time. This is achieved by tracking and recording the coordinates of the marked areas. Each marked area will be assigned a unique identifier and the boundary coordinates of the area will be recorded. These coordinate information will be used for subsequent conflict detection. The identification and positioning of the marked area uses image segmentation technology. The outline and boundary of the marked area are detected by the algorithm, and then compared with the actual data in the image. The system uses these coordinates to determine the location, size and other information of the area, providing accurate data support for subsequent registration and overlap detection.

[0082] Image registration and marker coordinate matching unit: By applying the image registration algorithm, the system can find corresponding points of the same area in images marked by different experts and align these marked areas in space. The registration process includes extracting feature points or key points (such as edges, corners, etc.) in the image, matching these points and performing spatial transformation on them, so that the areas marked by different experts can be compared in the same coordinate system, ensuring that the areas marked by different experts can correspond correctly in physical space, thereby providing an accurate basis for subsequent overlap detection.

[0083] Feedback unit for labeling conflicts: Once the system has aligned the labeled areas of different experts through a deep learning-based image registration algorithm, it will determine whether the areas conflict based on a predetermined overlap threshold. Conflicts are determined by calculating the intersection of the labeled areas. If the intersection is greater than the set threshold, a labeling conflict is considered to have occurred. The system will generate an immediate warning message to alert the expert to the overlap of the labeled areas, and indicate the overlapping parts of the labels by highlighting the conflicting areas.

[0084] The image registration algorithm is specifically expressed as:

[0085] The source image is set as I1 and the target image is set as I2. They are ultrasound images labeled by different experts and may have problems such as geometric transformation and scale change.

[0086] The source image I1 and the target image I2 are used as input, and the feature map is obtained through the shared convolution encoder ,

[0087] The source image is I1 and the target image is respectively passed through a shared convolutional network Perform feature extraction:

[0088]

[0089] The decoder part of the network generates the transformation field through deconvolution (or transposed convolution) , which represents the non-rigid transformation from the source image to the target image. The transformation field D is the displacement vector field of each pixel. The output of this part is:

[0090]

[0091] Use Transform Field Perform spatial transformation on the source image I1 to obtain the registered image :

[0092]

[0093] Warp represents the spatial transformation operation of the image, which usually uses nearest neighbor interpolation or bilinear interpolation to resample pixel values.

[0094] The goal of network optimization is to minimize the difference between the source image I1 and the target image I2 while maintaining the smoothness of the transformation field. The loss function usually consists of the following parts: The difference between the target image I2 and the target image I2 is usually expressed as mean square error (MSE).

[0095]

[0096] This loss term is used to constrain the smoothness of the transformation field to avoid excessive deformation. Elastic regularization can be used:

[0097] in, Represents the computation of the second-order gradient of the transformed field.

[0098] The final loss function is the weighted sum of the registration loss and the regularization loss:

[0099]

[0100] in is the weight of the regularization term, and the network is trained using a gradient descent method (such as the Adam optimizer) to minimize the loss function To update the weights in the network. The registration task requires a large amount of labeled data for each pair of source image I1 and target image I2 to perform supervised learning through real annotation information (such as expert labels).

[0101] Conflict prompts can be sent to experts through the platform's instant messaging system or pop-up prompts. The prompt information will not only include the conflict area, but also display the intersection of other expert-marked areas to help experts correct the markings more accurately.

[0102] The consistency detection module includes:

[0103] Consistency evaluation unit: The consistency evaluation unit calculates the overlap ratio between each expert's marked area and other experts' marked areas based on the geometric overlap algorithm, thereby evaluating the consistency of the marking;

[0104] The pixel comparison algorithm of structural similarity index is used to compare the regions marked by different experts at pixel level to evaluate their marking consistency.

[0105] Use contour detection and matching algorithms to compare the boundaries marked by experts and evaluate the consistency of the marked boundaries;

[0106] The above is achieved through the following steps:

[0107] Suppose we have two expert-marked regions, R1 and R2, which are obtained through image segmentation or labeling algorithms, usually binary images or polygonal boundaries.

[0108] R1 represents the region marked by expert 1, and R2 represents the region marked by expert 2. We want to calculate the geometric overlap between these two regions.

[0109] represents the intersection area of ​​R1 and R2 (i.e. the area of ​​the overlapping area);

[0110] represents the union area of ​​R1 and R2 (i.e. the total area of ​​the two);

[0111] Geometric Overlap can be calculated using the following formula:

[0112]

[0113] The calculation of the overlapping area can be done in two ways:

[0114] Pixel-based calculation: For binary images, the overlapping area of ​​two images at the pixel level can be directly calculated. Set I1 and I2 as two binary images (1 represents the marked area, 0 represents the unmarked area), then the intersection can be calculated by pixel product.

[0115]

[0116] The union can be calculated by summing the pixels:

[0117]

[0118] Contour-based calculation: For polygonal marked areas, the intersection and union areas between two polygons can be calculated. Using polygon intersection algorithms, such as the Sutherland-Hodgman algorithm or the ray-casting algorithm, the intersection and union areas of two polygons can be calculated efficiently.

[0119] Geometric overlap can be further extended to the calculation of shape similarity. In addition to considering area overlap, the shape difference between the two regions must also be considered. Shape similarity measurement can be performed in the following ways:

[0120] Hausdorff distance: This distance measures the difference between the farthest points of two shapes and is an important indicator for evaluating shape consistency. Given the boundary point sets S1 and S2 of two regions, the Hausdorff distance is defined as:

[0121]

[0122] in, is the Euclidean distance between points S1 and S2, sup and inf represent the maximum and minimum distances between point sets, respectively.

[0123] Jaccard Similarity Index: It is used to evaluate the shape similarity of two regions and calculate the ratio of their intersection and union:

[0124]

[0125] Dice Similarity Coefficient (DSC): Similar to the Jaccard similarity coefficient, but it takes into account twice the intersection and the sum of the total area of ​​the two regions. The calculation formula is:

[0126]

[0127] In practical applications, deep learning-based feature extraction algorithms (such as convolutional neural networks) are used to extract high-level features of images, which can reduce the computational complexity of traditional methods (such as direct pixel-based methods). By matching these high-level features, noise can be reduced when calculating overlapping areas.

[0128] For the overlapping parts of the marked areas, they can be weighted according to the area of ​​the marked areas or the experience of the experts. For example, higher weights can be given to certain areas marked by experts (such as key information areas), which helps improve the evaluation of marking consistency.

[0129] Consistency Optimization Unit: This unit aims to automatically optimize the labeling results and improve the consistency of labeling. During this process, the system will fine-tune the position, size, and shape of the labeled area according to the consistency score of the expert labeling results. The optimization method usually uses a weighted average or a minimum deviation algorithm. According to the deviation of the areas marked by different experts, the system will dynamically adjust the labeling position to make the labeling results of multiple experts closer. Once the optimization is completed, the optimized labeled area will be displayed as an "optimization box" in the image and fed back to the expert for confirmation.

[0130] Consistency score feedback unit: The consistency score feedback unit will track the marking process of each expert in real time, collect the consistency score of the marked area, and provide real-time feedback to the expert through the platform. The feedback information includes the score of the current marked area, the difference between the current mark and other expert marks, and provides improvement suggestions for the differences. The goal of this module is to help experts understand the marking consistency in a timely manner during the marking process and make adjustments when necessary to improve the consistency of the marking results.

[0131] Consistency report generation unit: This unit generates a final consistency assessment report, which lists in detail the consistency score, deviation analysis, optimization suggestions and final optimization results of each expert's marking. The report will clearly show the areas with low marking consistency and provide improvement plans to help experts make further corrections.

[0132] The audit modules include:

[0133] Review task assignment unit: This unit is used to assign the marking results to the chief reviewer for review. It automatically assigns the appropriate chief reviewer according to the task type, expert field and workload, and ensures that each review task is assigned to an expert with relevant experience and qualifications.

[0134] Review process monitoring unit: This unit monitors the review process of the marking results by the chief reviewer in real time, records the review opinions and modification suggestions. If the marking results are found to have obvious errors or parts that need to be adjusted during the review process, the system will automatically remind the chief reviewer to make modifications;

[0135] Review feedback unit: The task of the review task allocation unit is to assign the marking results to the appropriate lead reviewer based on the task type, expert field, and workload. The system will select the most suitable expert for review based on the expert's field experience and historical performance, and ensure that the expert has sufficient qualifications and experience to handle the corresponding marking task. In this process, the system will also automatically consider the priority and complexity of the task and allocate the workload reasonably.

[0136] When the chief reviewer reviews the marking results, the system will collect and feedback the reviewer's opinions, including the accuracy and consistency of the marking and necessary improvement suggestions. The review opinions will be fed back to the marking expert in the form of a report, and modification guidance will be provided;

[0137] Review confirmation and revision unit: The marking results that have been reviewed or modified by the chief expert will be submitted to the system for final confirmation. If the marking results need further revision, the system will guide the experts to modify and resubmit for review. The marking results that have passed the review will be locked and a final report will be generated.

[0138] Diagnostic report generation unit: Once the review is passed, the system will generate a formal diagnostic report based on the marking results. The report will list in detail each marked area, the marking expert's feedback and review opinions, summarize the diagnostic conclusions, and provide further diagnosis and treatment recommendations for key areas.

[0139] The diagnostic report generation unit also includes a report storage and management module and a report update and correction module:

[0140] The report storage and management module is used to store the generated diagnostic reports in the database and classify them according to the patient or device number to ensure that the reports are easy to find, track and archive;

[0141] Based on the report update and correction module, if errors or new information are found during the subsequent marking or review process, the system can update and correct the report and generate a revised report. This module records the history of all version updates to ensure the transparency and traceability of the report modification process.

[0142] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Common changes and substitutions made by those skilled in the art within the scope of the technical solution of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for real-time marking on a remote ultrasound device screen, characterized in that: The steps include: S1, obtaining a real-time ultrasound image collected by an ultrasound device; S2, performing image preprocessing on the acquired ultrasound image, including noise removal, image enhancement, and region segmentation; S3, transmitting the pre-processed image to the expert collaborative labeling platform; S4, perform multi-expert labeling in the expert collaborative labeling platform, and experts label different areas in the image according to task assignment; S5, performing real-time conflict detection on the marking results of each expert. If a marking conflict is detected, a prompt message is generated to guide the expert to make corrections; S6. Generate a consistency score for each expert’s marking based on the consistency detection algorithm, and automatically optimize the marking results according to the consistency score; S7. After the marking is completed, the final marking result is generated and submitted to the chief reviewer for review. After the review is passed, a diagnosis report is generated.

2. A system for implementing the real-time marking method on a remote ultrasound device screen as described in claim 1 comprises: Ultrasonic image acquisition module, used for real-time acquisition of ultrasonic images of ultrasonic equipment; An image preprocessing module is used to remove noise, enhance the image and perform region segmentation on the acquired ultrasound image; Expert collaborative labeling platform, used to support multi-expert collaborative labeling, where experts view and label real-time ultrasound images through a graphical interface; The marking area allocation module is used to automatically allocate the marking area of ​​each expert according to the predetermined task to avoid marking overlap; The conflict detection module is used to detect in real time whether the areas marked by the experts overlap. If a conflict is found, a prompt message is generated to guide the experts to make corrections; The consistency detection module is used to generate a consistency score for each expert's marking and automatically optimize the marking results based on the consistency score; The review module is used by the chief reviewer to review the marking results and generate a diagnosis report after the review is passed.

3. The real-time marking system on the remote ultrasound device screen according to claim 2 is characterized in that: The expert collaborative marking platform also includes: The marking tool module has a variety of marking tools, and experts can select appropriate tools to mark images according to actual needs; Real-time synchronous display module, the expert's marking operation is reflected in the image in real time, and other experts can view the latest marking progress synchronously, ensuring transparency and real-time collaboration in the marking process.

4. The real-time marking system on the remote ultrasound device screen according to claim 3 is characterized in that: The marking area allocation module comprises: Dynamic region allocation unit: Depending on the image content, the system can automatically identify key regions in the image and dynamically allocate labeling tasks based on the labeling requirements of these key regions; Area allocation strategy unit: By setting area priority and marking order, it prevents multiple experts from marking the same area at the same time. The expert's marking area will be clearly marked when it is allocated, and the logo will be displayed on the image to prevent overlap; Automatic marking area locking unit: After the task is assigned, the marking area of ​​each expert will be automatically locked. The locked area will be displayed as a "lock box" on the image. During the marking process, other experts cannot modify or mark the area, ensuring the independence of the marked area.

5. The real-time marking system on the remote ultrasound device screen according to claim 2, characterized in that: The conflict detection module comprises: Marking area recognition and positioning unit: During the marking process, each area marked by the expert will be automatically recognized by the system, and the coordinates of the marked area will be recorded in real time; Image registration and marker coordinate matching unit: The area marked by each expert is matched in the image through the image registration algorithm. The registration process ensures that the areas marked by different experts can be correctly aligned by extracting key feature points in the image to facilitate the detection of overlapping areas. The registered marker coordinates will be used to determine the overlap of different marked areas. If the intersection of multiple marked areas is greater than the set overlap threshold, it is considered a conflict and the system will generate a conflict warning.

6. The real-time marking system on the remote ultrasound device screen according to claim 5, characterized in that: The conflict detection module also includes a feedback unit for marking conflicts: when the system detects a marking conflict, it will send a warning message to the expert through the platform's instant message or pop-up prompt. In the prompt message, the conflict area is highlighted and the intersection with the marked area of ​​other experts is pointed out. The expert can correct the mark according to the feedback information of the platform, or communicate with other experts to resolve the conflict.

7. The real-time marking system on the remote ultrasound device screen according to claim 2, characterized in that: The consistency detection module comprises: Consistency evaluation unit: automatically evaluates the consistency between expert marking results by calculating the overlap, shape similarity and pixel-level consistency of multiple expert-marked areas, and generates a consistency score; Consistency optimization unit: Based on the consistency score of the expert marking results, the position, shape, and size of the marked area will be fine-tuned, and the marking results will be automatically optimized by weighted average. The optimized marked area will be displayed as an optimized box in the image; Consistency score feedback unit: This unit tracks the consistency score of the expert in the marking process in real time, provides real-time feedback, and provides prompt information to the expert based on the consistency score of the marking, prompting the expert to make necessary adjustments during the marking process to improve the consistency of the marking results; Consistency report generation unit: This unit generates a final consistency assessment report, which lists in detail the consistency score, deviation analysis, optimization suggestions and final optimization results of each expert's marking. The report will clearly show the areas with low marking consistency and provide improvement plans to help experts make further corrections.

8. The real-time marking system on the remote ultrasound device screen according to claim 7, characterized in that: The consistency evaluation unit calculates the overlap ratio between the marked area of ​​each expert and the marked areas of other experts based on the geometric overlap algorithm, thereby evaluating the consistency of the marking; The pixel comparison algorithm of structural similarity index is used to compare the regions marked by different experts at pixel level to evaluate their marking consistency. Using contour detection and matching algorithms, the boundaries marked by experts are compared to evaluate the consistency of the marked boundaries.

9. The real-time marking system on the remote ultrasound device screen according to claim 5, characterized in that: The audit module includes: Review task assignment unit: This unit is used to assign the marking results to the chief reviewer for review. It automatically assigns the appropriate chief reviewer according to the task type, expert field and workload, and ensures that each review task is assigned to an expert with relevant experience and qualifications. Review process monitoring unit: This unit monitors the review process of the marking results by the chief reviewer in real time, records the review opinions and modification suggestions. If the marking results are found to have obvious errors or parts that need to be adjusted during the review process, the system will automatically remind the chief reviewer to make modifications; Review feedback unit: When the chief reviewer reviews the marking results, the system will collect and feedback the reviewer's opinions, including the accuracy and consistency of the marking and necessary improvement suggestions. The review opinions will be fed back to the marking expert in the form of a report, and modification guidance will be provided; Review confirmation and revision unit: The marking results that have been reviewed or modified by the chief reviewer will be submitted to the system for final confirmation. If the marking results need further revision, the system will guide the experts to make revisions and resubmit for review. The marking results that have passed the review will be locked and a final report will be generated; Diagnostic report generation unit: Once the review is passed, the system will generate a formal diagnostic report based on the marking results. The report will list in detail each marked area, the marking expert's feedback and review opinions, summarize the diagnostic conclusions, and provide further diagnosis and treatment recommendations for key areas.

10. The real-time marking system on the remote ultrasound device screen according to claim 9, characterized in that: The diagnostic report generation unit also includes a report storage and management module: the generated diagnostic reports will be stored in the database and classified and managed according to the patient or device number to ensure that the reports are easy to find, track and archive historical records; Report update and correction module: If errors or new information are found during the subsequent marking or review process, the system can update and correct the report and generate a revised report. This module records the history of all version updates to ensure the transparency and traceability of the report modification process.

Images