Auxiliary ultrasonic diagnosis quality control system and method

Through the automated processing and quality control of auxiliary ultrasound diagnosis quality control system, the problems of high misdiagnosis rate and low work efficiency in ultrasound diagnosis are solved, and high accuracy and reliability of ultrasound imaging analysis is achieved, which is suitable for large-scale clinical applications.

WO2024239937A1PCT designated stage expired Publication Date: 2024-11-28SHANGHAI SOUNDWISE TECHNOLOGY CO LTD

Patent Information

Application Number
PCT/CN2024/090776
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-19
Filing Date
2024-04-30
Publication Date
2024-11-28

AI Technical Summary

Technical Problem

The existing ultrasound diagnosis quality control methods have problems such as high misdiagnosis rate, low working efficiency and high cost. The application scope and applicability of different methods are different, making it difficult to effectively improve the accuracy and reliability of ultrasound diagnosis.

Method used

A auxiliary ultrasound diagnosis quality control system is adopted, and the ultrasound image data is preprocessed and segmented through the segmentation module. The mass score calculation module performs quality analysis. The film retention module conducts comprehensive analysis based on the nodule position and benign and malignant classification scores to calculate the film retention score. To determine the retained image frames, combine artificial intelligence technology to achieve automated processing and quality control.

Benefits of technology

It realizes automatic identification and analysis of ultrasound images, reduces the cost and risk of manual intervention, improves the accuracy and reliability of ultrasound diagnosis, and is suitable for large-scale clinical applications, with high practicality and economic benefits.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2024090776_28112024_PF_FP_ABST
    Figure CN2024090776_28112024_PF_FP_ABST
Patent Text Reader

Abstract

The present application provides an auxiliary ultrasonic diagnosis quality control system and method. The method comprises: preprocessing and segmenting an image frame of ultrasonic image data, so as to obtain a target part mask and a nodule mask; analyzing the quality of the target part mask, and determining a quality score corresponding to the target part mask; and when the quality score is not less than a preset quality control threshold value, performing comprehensive analysis on the basis of a location area of a nodule obtained by means of processing based on the nodule mask, a nodule benign and malignant classification score as well as a target part position classification score, so as to obtain an image reservation score, and determining reserved image frames in the ultrasonic image data according to the image reservation score.
Need to check novelty before this filing date? Find Prior Art

Description

Assisted ultrasound diagnosis quality control system and method

[0001] Related applications

[0002] This application claims priority to Chinese patent application number 202310574704.9, filed on May 19, 2023, entitled “A system and method for auxiliary ultrasound diagnosis quality control based on artificial intelligence,” the entire text of which is hereby incorporated by reference. Technical Field

[0003] The present application relates to the technical field of ultrasonic diagnosis quality control, and in particular to an auxiliary ultrasonic diagnosis quality control system and method. Background Art

[0004] Ultrasonic diagnosis is a diagnostic method that applies ultrasound technology to the human body. By measuring physiological or tissue structure data and morphology, it can detect diseases and provide diagnostic information. Ultrasound diagnosis is a non-invasive, painless, convenient, and intuitive diagnostic method. B-ultrasound, in particular, is widely used and has a significant impact. It is considered one of the four major medical imaging technologies, along with X-rays, CT scans, and magnetic resonance imaging.

[0005] During the ultrasound diagnosis process, misdiagnosis or missed diagnosis may occur due to factors such as operator experience and equipment performance. Therefore, ultrasound quality control is needed to evaluate and control the quality of ultrasound images, thereby improving the accuracy and reliability of ultrasound diagnosis.

[0006] Currently, common ultrasound image quality control methods include manual inspection and quantitative assessment. Manual inspection involves the operator directly examining ultrasound images to check quality indicators such as clarity, noise, and artifacts to determine whether the images meet standards. While simple and easy to use, this method requires significant manual intervention, resulting in low efficiency and a high misdiagnosis rate due to the operator's experience and subjective factors. Quantitative assessment, on the other hand, involves calculating the numerical values ​​of image quality indicators to quantitatively assess ultrasound image quality. This method can objectively and rapidly assess image quality, reducing manual intervention, but it can only assess a few quality indicators and requires extensive upfront data processing and model training, resulting in high costs.

[0007] The above methods each have their own advantages and disadvantages. In addition, since ultrasound image quality assessment involves multiple quality indicators, the application scope and applicability of different methods are also different.

[0008] Summary of the Invention

[0009] According to various embodiments of the present application, the present application provides an auxiliary ultrasound diagnosis quality control system and method.

[0010] In a first aspect, the present application provides an auxiliary ultrasound diagnosis quality control system, comprising:

[0011] A segmentation module is used to pre-process and segment the image frames of the ultrasound image data obtained by scanning the target part to obtain a target part mask plate and a nodule mask plate;

[0012] a quality score calculation module, connected to the segmentation module, for performing a quality analysis on the target portion mask to determine a quality score corresponding to the target portion mask;

[0013] The retention module is used to perform a comprehensive analysis based on the nodule area, benign and malignant classification score of the nodule, and target site location classification obtained by processing the nodule mask when the quality score is not less than a preset quality control threshold, to obtain a retention score, and determine the image frames in the ultrasound image data to be retained based on the retention score.

[0014] In one embodiment, the retaining module is further configured to retain the current image frame when the quality score is less than the preset quality control threshold.

[0015] In one embodiment, the auxiliary ultrasound diagnosis quality control system further includes:

[0016] a first data scaling module, connected to the segmentation module, configured to scale the current image frame to a first image processing frame of a first preset size;

[0017] The segmentation module is used to perform segmentation processing on the first image processing frame to obtain the target part mask and the nodule mask.

[0018] In one embodiment, the quality score calculation module includes:

[0019] a target portion ratio calculation unit, configured to calculate the target portion ratio according to a ratio of the number of pixels with a grayscale value of 255 in the target portion mask to the total number of pixels in the target portion mask;

[0020] A mass score calculation unit is connected to the target site proportion calculation unit and is used to calculate the mass score according to the target site proportion, where the mass score is a predetermined multiple of the target site proportion.

[0021] In one embodiment, the auxiliary ultrasound diagnosis quality control system further includes: a nodule location determination module, which is respectively connected to the segmentation module, the quality score calculation module, and the film retention module, and is used to obtain the area where the nodule is located according to the nodule mask when the quality score is not less than the preset quality control threshold; the nodule location determination module includes:

[0022] a contour determining unit, configured to determine a nodule contour based on pixels with a grayscale value of 255 in the nodule mask;

[0023] The nodule region determination unit is connected to the contour determination unit and is used to calculate the minimum circumscribed rectangle of the nodule contour and determine the region where the nodule is located by a fitting method.

[0024] In one embodiment, the auxiliary ultrasound diagnosis quality control system further includes:

[0025] The nodule benign and malignant judgment module is respectively connected to the segmentation module, the quality score calculation module and the film retention module, and is used to obtain the nodule benign and malignant classification score based on the image frame preprocessed by the ultrasound image data and the area where the nodule is located based on a pre-trained nodule benign and malignant classification model when the quality score is not less than the preset quality control threshold.

[0026] In one embodiment, the auxiliary ultrasound diagnosis quality control system further includes: a data reorganization module connected to the nodule benign or malignant judgment module, the data reorganization module including:

[0027] a first cutting unit, configured to cut all nodules in the image frame of the ultrasound image data according to the region where the nodules are located, to obtain a first nodule image group;

[0028] a second cutting unit, configured to cut all nodules in the nodule mask according to the area where the nodules are located, to obtain a second nodule image group;

[0029] A recombining unit is connected to the first cutting unit and the second cutting unit respectively, and is used to replace the G channel in the first nodule image group with the second nodule image group to obtain a third image group, and input it into the benign and malignant nodule classification model as the preprocessed image frame.

[0030] In one embodiment, the auxiliary ultrasound diagnosis quality control system further includes:

[0031] A second data scaling module, configured to scale the current image frame to a second image processing frame of a second preset size;

[0032] The target part position classification module is respectively connected to the second data scaling module, the quality score calculation module and the film retention module, and is used to process the second image processing frame based on the pre-trained target part position classification model when the quality score is not less than the preset quality control threshold to obtain the target part position classification.

[0033] In one embodiment, the retained fragment fraction is calculated using the following formula:

[0034] SL=a*Pt*L+b*Pn / Pt+c*SBM

[0035] Among them, a, b, and c represent weights respectively; Pt represents the proportion of the target part; L represents the predetermined multiple of the proportion of the target part; Pn represents the proportion of nodules; SBM represents the benign and malignant classification score of nodules; SL represents the film retention score.

[0036] In a second aspect, the present application further provides an auxiliary ultrasound diagnosis quality control method, which is applied to the above-mentioned auxiliary ultrasound diagnosis quality control system, comprising:

[0037] Preprocessing and segmenting the image frames of ultrasound image data obtained by scanning the target part to obtain a target part mask and a nodule mask;

[0038] Performing a quality analysis on the target portion mask to determine a quality score corresponding to the target portion mask;

[0039] When the quality score is not less than the preset quality control threshold, a comprehensive analysis is performed based on the nodule area, the benign and malignant classification score of the nodule, and the target site location classification obtained by processing the nodule mask to obtain a retention score, and the image frames in the ultrasound image data to be retained are determined based on the retention score.

[0040] The details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are merely embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the disclosed drawings without any creative work.

[0042] FIG1 is a schematic diagram of the architecture of an auxiliary ultrasound diagnosis quality control system in some embodiments of the present application.

[0043] FIG2 is a flow chart of a method for assisting ultrasound diagnosis quality control in some embodiments of the present application. DETAILED DESCRIPTION

[0044] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of this application.

[0045] It should be noted that, unless there is any conflict, the embodiments and features in the embodiments of this application can be combined with each other.

[0046] The present application will be further described below with reference to the accompanying drawings and specific embodiments, but they are not intended to limit the present application.

[0047] Referring to FIG1 , in some embodiments of the present application, a system for assisting ultrasound diagnosis quality control is provided, comprising:

[0048] Segmentation module 1, used to pre-process and segment the image frames of ultrasound image data obtained by scanning the target part to obtain a target part mask and a nodule mask;

[0049] The quality score calculation module 2 is connected to the segmentation module 1 and is used to perform quality analysis on the target part mask and determine the quality score corresponding to the target part mask;

[0050] The retention module 3 is used to perform a comprehensive analysis based on the nodule area, benign and malignant classification score of the nodule, and target site location classification obtained by processing the nodule mask when the quality score is not less than the preset quality control threshold, to obtain the retention score, and determine the image frames in the retained ultrasound image data based on the retention score.

[0051] Specifically, in this embodiment, the above-mentioned ultrasound image data is obtained by scanning the target part through a local B-ultrasound device. The target part can be the thyroid gland, or the breast or other parts of the human body with a nodule area. In the embodiment of the present application, the ultrasound image data of the thyroid gland is obtained by scanning. The ultrasound image data consists of several continuous image frames, and the image frames are color images. The ultrasound image data is transmitted to the auxiliary ultrasound diagnosis quality control system at the local edge end. For each image frame, the image frame is segmented by the segmentation module 1 to obtain a thyroid mask plate and a nodule mask plate. The thyroid mask plate and the nodule mask plate are binary images; then the thyroid mask plate is quality analyzed to obtain a corresponding quality score; the quality score is compared with the preset quality control threshold by the quality score quality threshold judgment module 5. The preset quality control threshold can be set according to the actual application. As an example and not a limitation, the preset quality control threshold is 0.2. When the quality score is not less than the preset quality control threshold, the film retention module 3 performs a comprehensive analysis based on the nodule area, the benign and malignant classification score of the nodule, and the target site location classification, and calculates the film retention score, thereby determining the image frames in the retained ultrasound image data based on the film retention score.

[0052] The embodiments of the present application use artificial intelligence technology to automatically identify and analyze ultrasound images, achieve automated processing and quality control, reduce the cost and risk of manual intervention, and have high accuracy and reliability.

[0053] As an optional implementation, the retaining module 3 is further configured to directly retain the current image frame when the quality score is less than a preset quality control threshold.

[0054] Specifically, in this embodiment, when the quality score is less than the preset quality control threshold, it means that there is no thyroid gland or the area is small in the thyroid mask obtained by segmenting the current image frame, that is, the image quality is poor. However, since the ultrasound image data is a continuous video and has contextual association, the current image frame with a quality score less than the preset quality control threshold is directly retained in the film retention module and will not be directly discarded.

[0055] As an optional implementation manner, the method further includes:

[0056] A first data scaling module 4, connected to the segmentation module 1, for scaling the current image frame to a first image processing frame of a first preset size;

[0057] The segmentation module 1 is used to perform segmentation processing on the first image processing frame to obtain a target part mask and a nodule mask.

[0058] Specifically, in this embodiment, before segmentation module 1 segments the image frame, it first scales the current image frame to a predetermined size. This first predetermined size can be set based on actual application, such as the segmentation accuracy of the segmentation module. By way of example and not limitation, the first predetermined size is 256*256*3. The scaled first processed image frame is then input to segmentation module 1 for segmentation. The segmentation module can be implemented using a Unet segmentation model.

[0059] As an optional implementation, the quality score calculation module 2 includes:

[0060] A target portion ratio calculation unit (not shown in the figure) is used to calculate the target portion ratio based on the ratio of the number of pixels with a grayscale value of 255 in the target portion mask to the total number of pixels in the target portion mask;

[0061] The mass score calculation unit (not shown in the figure) is connected to the target part proportion calculation unit and is used to calculate the mass score according to the target part proportion. The mass score is a predetermined multiple of the target part proportion.

[0062] Specifically, in this embodiment, the mass fraction can be calculated using the following formula: SQ = Pt*L; Pt = N Pi=255 / N MASK ;

[0063] Wherein, Pt represents the thyroid ratio; L represents the predetermined multiple of the thyroid ratio. The specific value of L can be set according to actual conditions. As an example but not a limitation, L can be selected as 5; N Pi=255 Indicates the number of pixels with a grayscale value of 255 in the thyroid mask; N MASK represents the total number of pixels in the thyroid mask; SQ represents the quality score, where the quality score is ≤ 1. It should be noted that the above formula is applicable to other parts of the human body with nodules.

[0064] Furthermore, the auxiliary ultrasound diagnosis quality control system also includes: a data recovery module 7, which is respectively connected to the segmentation module 1 and the nodule position determination module 6, and is used to perform data recovery on the segmentation of the nodule area output by the segmentation module 1 to obtain a binary nodule mask plate to improve the image accuracy of the nodule mask plate.

[0065] As an optional embodiment, the auxiliary ultrasound diagnosis quality control system further includes: a nodule location determination module 6, which is respectively connected to the segmentation module 1, the quality score calculation module 2 and the film retention module 3, and is used to obtain the area where the nodule is located according to the nodule mask when the quality score is not less than a preset quality control threshold; the nodule location determination module includes:

[0066] A contour determination unit (not shown in the figure) is used to determine the nodule contour based on pixels with a grayscale value of 255 in the nodule mask;

[0067] The nodule region determination unit (not shown in the figure) is connected to the contour determination unit and is used to calculate the minimum circumscribed rectangle of the nodule contour and determine the nodule region by fitting method.

[0068] Specifically, in this embodiment, the contour of the thyroid nodule is determined by calculating the pixels with a grayscale value of 255 in the nodule mask. One or more thyroid nodules may exist in a nodule mask. For each nodule contour, the minimum circumscribed rectangle of each contour is calculated, and then the area where each thyroid nodule is located is determined by fitting, and the fitting method can adopt the ellipse fitting method.

[0069] As an optional implementation, the auxiliary ultrasound diagnosis quality control system further includes:

[0070] The nodule benign or malignant judgment module 9 is respectively connected to the segmentation module 1, the quality score calculation module 2 and the film retention module 3, and is used to obtain the benign or malignant classification score of the nodule based on the pre-trained benign or malignant nodule classification model according to the image frames preprocessed by the ultrasound image data and the area where the nodule is located when the quality score is not less than the preset quality control threshold.

[0071] Specifically, in this embodiment, the nodule benign and malignant judgment module 9 is implemented using a two-classification network model. The nodule benign and malignant classification model identifies whether the nodule is benign or malignant, and then outputs the benign and malignant classification score of the nodule based on the benign and malignant classification. Specifically, it outputs 0 when the nodule is benign and outputs 1 when the nodule is malignant.

[0072] Furthermore, the aforementioned benign and malignant nodule classification model can be implemented using a related benign and malignant nodule classification model, such as at least one of the VGG network (Visual Geometry Group), the Resnet network (Residual Neural Network), the Dense Convolutional Network (Dense Convolutional Network), and the EfficientNet network. The classification model training and model application processes are not further described here.

[0073] As an optional embodiment, the auxiliary ultrasound diagnosis quality control system further includes: a data reorganization module 8 connected to the nodule benign or malignant judgment module 6, and the data reorganization module 8 includes:

[0074] A first cutting unit (not shown in the figure) is used to cut all nodules in the image frame of the ultrasound image data according to the area where the nodules are located, to obtain a first nodule image group;

[0075] A second cutting unit (not shown in the figure) is used to cut all nodules in the nodule mask according to the area where the nodules are located to obtain a second nodule image group;

[0076] The reorganization unit (not shown in the figure) is connected to the first cutting unit and the second cutting unit, respectively, and is used to replace the G channel in the first nodule image group with the second nodule image group to obtain a third image group, which is input as a preprocessed image frame into the nodule benign and malignant classification model.

[0077] Specifically, in this embodiment, before the nodule is judged to be benign or malignant, the current image frame is clipped and reorganized by the data reorganization module 8 to improve the image accuracy. Furthermore, the color image frame is cut according to the nodule area (i.e., nodule position) output by the nodule position determination module 6 to obtain the square where each nodule image is located. When the image is larger than the side length 224, it is scaled to 224; when the image side length is less than 224, the data outside the nodule is intercepted and supplemented to 224 to obtain the first nodule image group; the binary image of the nodule mask is cut in the same way to obtain the second nodule image group; for each nodule area, the position number corresponds one to one, and the G channel of the RGB channel component in the first nodule image group is replaced with the second nodule image group. After the replacement, a third image group is formed and used as the input for the nodule benign or malignant judgment.

[0078] As an optional implementation, the auxiliary ultrasound diagnosis quality control system further includes:

[0079] A second data scaling module 10 is configured to scale the current image frame to a second image processing frame of a second preset size;

[0080] The target part position classification module 11 is respectively connected to the second data scaling module 10, the quality score calculation module 2 and the film retention module 3, and is used to process the second image processing frame based on the pre-trained target part position classification model when the quality score is not less than the preset quality control threshold to obtain the target part position classification.

[0081] Furthermore, when the target part location classification result is not the target classification result, a preset number of consecutive image frames before and after the current image frame are searched. If the target classification result appears in any of the preset number of consecutive image frames before and after the current image frame, the current image frame is retained; otherwise, the current image frame is discarded. Furthermore, the percentage of image frames in which the target classification result appears among the preset number of consecutive image frames is obtained. If the percentage exceeds a preset value, the current image frame is retained; otherwise, the current image frame is discarded.

[0082] Specifically, in this embodiment, taking the thyroid gland as an example, the thyroid gland is divided into six position categories: isthmus, left transverse section, left longitudinal section, right transverse section, right longitudinal section, and other. That is, the target part position classification model is a six-category network model. When the classification result is isthmus, left transverse section, left longitudinal section, right transverse section, or right longitudinal section, the image is entered into the film retention module 3 for film retention processing. When the classification result is other, the image frames of the preset number of consecutive frames before and after the current image frame are searched. If the image frames with the classification results of isthmus, left transverse section, left longitudinal section, right transverse section, or right longitudinal section appear in the image frames of the preset number of consecutive frames before and after, the current frame is processed for film retention; otherwise, it is discarded. Furthermore, the proportion of image frames with the classification results of isthmus, left transverse section, left longitudinal section, right transverse section, or right longitudinal section in the preset number of consecutive image frames is obtained. If the proportion exceeds a preset value, the image is processed for film retention; otherwise, it is discarded.

[0083] Furthermore, the target location classification model can be implemented using a related thyroid location classification model, such as at least one of the following: a VGG network (Visual Geometry Group), a ResNet network (Residual Neural Network), a Dense Convolutional Network (Dense Convolutional Network), or an EfficientNet network. The classification model training and application processes are not detailed here.

[0084] Furthermore, the target part position classification module 11 is also connected to the output of the segmentation module 1 for classification based on the binary image of the thyroid mask; when the thyroid mask image does not exist, a new thyroid mask with mask set to 0 is created.

[0085] As an optional embodiment, the retained fraction is calculated using the following formula: SL = a*Pt*L+b*Pn / Pt+c*SBM

[0086] Among them, a, b, and c represent weights respectively, and the weight values ​​can be set according to actual applications. They are optional but not limited. a=0.3, b=0.3, and c=0.4; Pt represents the target site ratio; L represents the predetermined multiple of the target site ratio, and L can be selected as 5; Pn represents the nodule ratio; SBM represents the benign and malignant classification score of the nodule; SL represents the film retention score.

[0087] Furthermore, when the ultrasound image data is an ultrasound examination of a normal healthy person and there is no nodule, the SBM is replaced by 0 to retain the thyroid image of the normal healthy person.

[0088] Furthermore, the auxiliary ultrasound diagnosis quality control system also includes: a data sorting module 12, connected to the film retention module 3, for sorting the retained image frames and forming a diagnosis report, which is sent to the doctor's local device to assist in ultrasound diagnosis.

[0089] In the above embodiment, after the local B-ultrasound equipment scans and obtains ultrasound image data, it is transmitted to the auxiliary ultrasound diagnosis quality control system at the local edge. The quality score calculation module 2 performs preliminary quality control based on the quality control AI to ensure that the data quality meets the scanning standards. The quality control AI network is a classification network including but not limited to ResNet, Desnet, and Transformer networks, and outputs a quality score. A quality score greater than 0.8 is selected as qualified. The data is then transmitted to the cloud via the network, and a preliminary diagnosis is performed by the auxiliary diagnosis AI to determine the location of the nodule, the judgment of benign and malignant nodules, and the classification of the thyroid location. The auxiliary diagnosis AI network generates a structured report based on the pre-analysis parameters.

[0090] Furthermore, the diagnostic report can be sampled and reviewed to ensure its accuracy and reliability. The final report is sent back to the doctor's local device via the Internet for reference and use.

[0091] It should be understood by those skilled in the art that the modules or steps of the present application described above can be implemented using a general-purpose computing device, they can be concentrated on a single computing device, or distributed across a network composed of multiple computing devices, optionally, they can be implemented using program code executable by the computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be performed in a different order than herein, or they can be made into separate integrated circuit modules, or multiple modules or steps can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0092] 2 , the present application further provides an auxiliary ultrasound diagnosis quality control method, which is applied to the above-mentioned auxiliary ultrasound diagnosis quality control system, comprising:

[0093] Step 1: Preprocess and segment the image frames of ultrasound image data obtained by scanning the target part to obtain a target part mask and a nodule mask;

[0094] Step 2: Perform a quality analysis on the target area mask to determine the quality score corresponding to the target area mask;

[0095] Step 3: When the quality score is not less than the preset quality control threshold, a comprehensive analysis is performed based on the nodule area, benign and malignant classification score of the nodule, and target site location classification obtained by nodule mask processing to obtain the retention score, and the image frames in the retained ultrasound image data are determined based on the retention score.

[0096] The advantages or beneficial effects of adopting the above-mentioned technical solution are: this application uses deep learning technology to realize automated processing and quality control of ultrasound images, which not only greatly reduces the cost and risk of manual intervention, but also can intelligently analyze and identify ultrasound images, thereby improving the accuracy and reliability of ultrasound diagnosis; at the same time, this application adopts digital processing, which can quickly and accurately process a large number of ultrasound images, is suitable for large-scale clinical applications, and has high practicality and economic benefits.

[0097] In some embodiments, the method for assisting ultrasound diagnosis quality control further includes:

[0098] When the quality score is less than the preset quality control threshold, the current image frame is retained.

[0099] In some embodiments, preprocessing and segmenting the image frames of ultrasound image data obtained by scanning the target part to obtain the target part mask and the nodule mask further includes:

[0100] Scaling the current image frame to a first image processing frame of a first preset size;

[0101] The first image processing frame is segmented to obtain the target part mask and the nodule mask.

[0102] In some embodiments, performing a quality analysis on the target portion mask to determine a quality score corresponding to the target portion mask further comprises:

[0103] The target portion ratio is calculated based on the ratio of the number of pixels with a grayscale value of 255 in the target portion mask to the total number of pixels in the target portion mask;

[0104] The mass fraction is calculated according to the target site proportion, and the mass fraction is a predetermined multiple of the target site proportion.

[0105] In some embodiments, the method for assisting ultrasound diagnosis quality control further includes:

[0106] Determining the nodule contour according to pixels with a grayscale value of 255 in the nodule mask;

[0107] The minimum circumscribed rectangle of the nodule outline is calculated, and the region where the nodule is located is determined by a fitting method.

[0108] In some embodiments, the method for assisting ultrasound diagnosis quality control further includes:

[0109] When the quality score is not less than the preset quality control threshold, the benign and malignant nodule classification score is obtained based on a pre-trained benign and malignant nodule classification model and the image frames preprocessed by the ultrasound image data and the area where the nodule is located.

[0110] In some embodiments, the method for assisting ultrasound diagnosis quality control further includes:

[0111] Cutting all nodules in the image frame of the ultrasound image data according to the area where the nodules are located to obtain a first nodule image group;

[0112] cutting all nodules in the nodule mask according to the area where the nodules are located to obtain a second nodule image group;

[0113] The G channel in the first nodule image group is replaced with the second nodule image group to obtain a third image group, which is input into the benign and malignant nodule classification model as the preprocessed image frame.

[0114] In some embodiments, the method for assisting ultrasound diagnosis quality control further includes:

[0115] Scaling the current image frame to a second image processing frame of a second preset size;

[0116] When the quality score is not less than the preset quality control threshold, the image frames of the ultrasound image data are processed based on a pre-trained target part position classification model to obtain the target part position classification.

[0117] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0118] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0119] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. An auxiliary ultrasound diagnosis quality control system, It is characterized in that include: A segmentation module is used to pre-process and segment the image frames of the ultrasound image data obtained by scanning the target part to obtain a target part mask plate and a nodule mask plate; A quality score calculation module, connected to the segmentation module, for performing quality analysis on the target part mask plate to determine the quality score corresponding to the target part mask plate; The film retention module is used to perform a comprehensive analysis based on the nodule area, benign and malignant classification scores of nodules, and target site location classification obtained by processing the nodule mask plate when the quality score is not less than a preset quality control threshold, to obtain a film retention score, and determine the image frames in the ultrasound image data to be retained based on the film retention score.

2. The auxiliary ultrasound diagnosis quality control system according to claim 1, in, The retaining module is further configured to retain the current image frame when the quality score is less than the preset quality control threshold.

3. The auxiliary ultrasound diagnosis quality control system according to claim 1, in, Also includes: A first data scaling module, connected to the segmentation module, configured to scale the current image frame to a first image processing frame of a first preset size; The segmentation module is used to perform segmentation processing on the first image processing frame to obtain the target part mask and the nodule mask.

4. The auxiliary ultrasound diagnosis quality control system according to claim 1, in, The quality score calculation module comprises: A target part proportion calculation unit, used to calculate the target part proportion according to the ratio of the number of pixels with a gray value of 255 in the target part mask to the total number of pixels in the target part mask; A mass score calculation unit is connected to the target part proportion calculation unit and is used to calculate the mass score according to the target part proportion, wherein the mass score is a predetermined multiple of the target part proportion.

5. The auxiliary ultrasound diagnosis quality control system according to claim 1, in, Also includes: a nodule location determination module, connected to the segmentation module, the quality score calculation module and the film retention module, respectively, and configured to obtain the area where the nodule is located according to the nodule mask when the quality score is not less than the preset quality control threshold; The nodule location determination module comprises: A contour determination unit, used for determining a nodule contour according to pixels with a gray value of 255 in the nodule mask; The nodule region determination unit is connected to the contour determination unit and is used to calculate the minimum circumscribed rectangle of the nodule contour and determine the region where the nodule is located by a fitting method.

6. The auxiliary ultrasound diagnosis quality control system according to claim 1, in, Also includes: The nodule benign or malignant judgment module is respectively connected to the segmentation module, the quality score calculation module and the retention module, and is used to obtain the nodule benign or malignant classification score based on the image frames preprocessed by the ultrasound image data and the area where the nodule is located based on a pre-trained nodule benign or malignant classification model when the quality score is not less than the preset quality control threshold.

7. The auxiliary ultrasound diagnosis quality control system according to claim 6, in, Also includes: A data reorganization module is connected to the nodule benign or malignant judgment module, and the data reorganization module includes: A first cutting unit, configured to cut all nodules in the image frame of the ultrasound image data according to the area where the nodules are located, to obtain a first nodule image group; A second cutting unit is used to cut all nodules in the nodule mask according to the area where the nodules are located to obtain a second nodule image group; The reorganization unit is connected to the first cutting unit and the second cutting unit respectively, and is used to replace the G channel in the first nodule image group with the second nodule image group to obtain a third image group, and input it into the nodule benign and malignant classification model as the preprocessed image frame.

8. The auxiliary ultrasound diagnosis quality control system according to claim 1, in, Also includes: A second data scaling module, used for scaling the current image frame to a second image processing frame of a second preset size; The target part position classification module is respectively connected to the second data scaling module, the quality score calculation module and the film retention module, and is used to process the second image processing frame based on a pre-trained target part position classification model when the quality score is not less than the preset quality control threshold to obtain the target part position classification.

9. The auxiliary ultrasound diagnosis quality control system according to claim 5, in, The retained film fraction is calculated using the following formula: SL = a*Pt*L+b*Pn / Pt+c*SBM Among them, a, b, and c represent weights respectively; Pt represents the proportion of the target part; L represents the predetermined multiple of the proportion of the target part; Pn represents the proportion of nodules; SBM represents the benign and malignant classification score of the nodules; SL represents the retention score.

10. A method for quality control of auxiliary ultrasound diagnosis, It is characterized in that The auxiliary ultrasonic diagnosis quality control system as claimed in any one of claims 1 to 9 comprises: Preprocessing and segmenting the image frames of the ultrasound image data obtained by scanning the target part to obtain a target part mask plate and a nodule mask plate; Performing a quality analysis on the target portion mask plate to determine a quality score corresponding to the target portion mask plate; When the quality score is not less than the preset quality control threshold, the area where the nodule is located obtained by processing the nodule mask The ultrasound image data is retained by performing a comprehensive analysis of the domain, the benign and malignant classification score of the nodule, and the target site location classification to obtain a retention score, and determining the image frames in the ultrasound image data to be retained according to the retention score.

Citation Information

Patent Citations

  • Auxiliary ultrasonic diagnosis quality control system and method based on artificial intelligence

    CN116646073A

  • Machine learning based non-invasive diagnosis of thyroid disease

    US10993653B1

  • B-mode ultrasound intelligent auxiliary acquisition method and system

    WO2020118535A1

  • Ultrasound image processing method and system, and computer readable storage medium

    WO2021232320A1

  • Periodic parameter analysis method and ultrasonic imaging system

    WO2022141083A1

Cited By

  • Thyroid hyperspectral image risk prediction method based on quality control and uncertainty

    CN122435417A