A pregnancy risk prediction system for first trimester ultrasound examination

Abstract

The present application relates to the technical field of pregnancy risk prediction, and particularly relates to a pregnancy risk prediction system for early pregnancy ultrasonic examination, comprising a data acquisition module, a preprocessing module configured to preprocess detection data to obtain a user data group, a verification module configured to determine whether the operation of the data acquisition module is qualified based on an abnormal data rate of the user data group, and, when determining that the operation of the data acquisition module is abnormal, correct the acquisition of the user data group based on image representation parameters, a data correction module configured to correct preprocessing parameters for the detection data, and an acquisition correction module configured to correct acquisition parameters for each ultrasonic examination picture. The abnormal data rate of the user data group is analyzed to determine whether the operation of the data acquisition module is qualified, and when determining that the operation of the data acquisition module is abnormal, the preprocessing parameters for the detection data are corrected, so that the accuracy and reliability of the data are improved, and the precision of the pregnancy risk prediction is improved.

Description

Technical Field

[0001] This invention relates to the field of pregnancy risk prediction technology, and more particularly to a pregnancy risk prediction system for early pregnancy ultrasound examination. Background Technology

[0002] In early pregnancy, ultrasound examinations can obtain relevant images and clinical data, which can help predict potential adverse pregnancy outcomes. This is of great significance for clinicians to develop intervention measures in advance and protect the health of both the pregnant woman and the fetus. It is especially applicable to early pregnancy check-ups in medical institutions such as obstetrics and gynecology hospitals and maternal and child health hospitals.

[0003] Currently, traditional methods for predicting pregnancy risks in early pregnancy mainly rely on doctors' clinical experience and judgment based on single indicators, such as a rough risk assessment based solely on the pregnant woman's age and medical history. This method has limitations, lacking comprehensiveness and precision, and is difficult to accurately predict pregnancy outcomes.

[0004] With the development of information technology, some data-based prediction schemes have also emerged.

[0005] Chinese Patent Publication No. CN120089404B discloses a method and system for assisting in the prediction of preterm birth risk during pregnancy. This method involves acquiring historical pregnant women's physical data and delivery cycles, determining the degree of preterm birth based on the delivery cycle, and determining the preterm birth correlation of different types of physical data based on differences in the degree of preterm birth among different historical pregnant women and differences in the deviation of the same type of physical data during each physical examination. Based on the time differences in the deviation of the same type of physical data among different historical pregnant women, the method clusters the historical pregnant women, determines the overall correlation of each type of physical data based on the number of pregnant women in each cluster and the preterm birth correlation of the corresponding type of physical data, and then filters out the target type of physical data. Based on the target type of physical data from different physical examinations of historical pregnant women, a preterm birth prediction model is constructed to predict preterm birth. However, the above technical solution has the following problems: it does not consider analyzing the abnormal data rate based on user data groups to determine whether the data acquisition module is functioning correctly, and it does not consider correcting the preprocessing parameters for the detection data when the data acquisition module is found to be malfunctioning, affecting the accuracy and reliability of the data, and thus affecting the accuracy of pregnancy risk prediction. Summary of the Invention

[0006] Therefore, the present invention provides a pregnancy risk prediction system for early pregnancy ultrasound examination, which overcomes the problems in the prior art that do not consider analyzing the abnormal data rate based on user data groups to determine whether the data acquisition module is operating properly, and do not consider correcting the preprocessing parameters for the test data when the data acquisition module is found to be operating abnormally, which affects the accuracy and reliability of the data, and thus affects the accuracy of pregnancy risk prediction.

[0007] To achieve the above objectives, the present invention provides a pregnancy risk prediction system for early pregnancy ultrasound examination, comprising:

[0008] The data acquisition module acquires quantitative detection data from various ultrasound examination images and digital detection data from user medical records.

[0009] A preprocessing module, which is connected to the data acquisition module, is used to preprocess the detection data to obtain user data sets;

[0010] A verification module, connected to the preprocessing module, is used to determine whether the operation of the data acquisition module is qualified based on the abnormal data rate of the user data group, and, when it is determined that the operation of the data acquisition module is abnormal, to correct the acquisition parameters for the user data group based on the image representation parameters.

[0011] The data correction module is connected to the preprocessing module, the data acquisition module and the verification module respectively. It corrects the preprocessing parameters for the detection data, including the upper boundary threshold of the preprocessing module for outlier filtering based on image representation parameters, and the Canny operator double threshold of the edge detection algorithm for segmenting gestational sacs or embryos in each ultrasound image when the data acquisition module acquires quantitative detection data based on body mass index.

[0012] The acquisition and calibration module is connected to the data acquisition module and the calibration module respectively, and is used to correct the acquisition parameters for each ultrasound examination image, including correcting the image acquisition frame rate of the acquired ultrasound examination image, or redetermining the guidance scanning range.

[0013] Furthermore, the verification module is used to determine whether the operation of the data acquisition module is qualified based on the abnormal data rate of the user data group, including determining that the operation of the data acquisition module is abnormal when the abnormal data rate is greater than the preset abnormal data rate, and correcting the acquisition parameters of the user data group based on the image representation parameters.

[0014] Furthermore, the verification module is used to determine the abnormal data rate of the user data group, including obtaining the sum of the number of missing data and the number of significantly abnormal data in the user data group to obtain the number of potentially problematic data.

[0015] This is used to calculate the ratio of the number of potential data points to the preset necessary data points, thus obtaining the abnormal data rate.

[0016] Furthermore, the verification module is used to correct the acquisition parameters for the user data group based on the image representation parameters, including: determining to correct the acquisition parameters for each ultrasound examination image when the image representation parameters are less than or equal to the preset image representation parameters; and determining to correct the preprocessing parameters for the detection data when the image representation parameters are greater than the preset image representation parameters.

[0017] Furthermore, the verification module is used to determine image representation parameters, including:

[0018] This is used to obtain the ratio of the gradient mean of a single ultrasound image to the preset gradient mean, thus obtaining the clarity factor;

[0019] The ratio of the gradient variance of a single ultrasound image to a preset gradient variance is used to obtain the complete factor.

[0020] The image representation factor for a single ultrasound image is obtained by summing the corresponding weight coefficients assigned to the clarity factor and the integrity factor.

[0021] The average value of the image characterization factors for each ultrasound image is used to obtain the image characterization parameters.

[0022] Furthermore, the data correction module is used to correct the acquisition parameters for each ultrasound examination image, including:

[0023] Under the condition that the sharpness factor is less than or equal to the preset sharpness factor, the image acquisition frame rate of ultrasound examination images is corrected based on the average fetal heart rate.

[0024] Under the condition that the integrity factor is less than or equal to the preset integrity factor, the guiding scan range is redetermined based on the intensity gradient mutation value.

[0025] Furthermore, the data correction module is used to redetermine the guiding scan range based on the intensity gradient abrupt change value, including:

[0026] Sobel operator edge detection is performed on a single ultrasound image to obtain a band-shaped region extending 10% of the image width inward along the four boundaries of the ultrasound image, and the average gradient magnitude of each band-shaped region is calculated.

[0027] The average gradient amplitude of each ultrasound image is calculated to obtain the intensity gradient abrupt change value.

[0028] The increase in the guided scanning range is negatively correlated with the intensity gradient abrupt change value.

[0029] Furthermore, the data correction module is used to correct the image acquisition frame rate of the ultrasound examination images based on the average fetal heart rate, wherein:

[0030] The increase in the image acquisition frame rate of ultrasound examinations is positively correlated with the average fetal heart rate.

[0031] Furthermore, the data correction module corrects the upper boundary threshold for outlier filtering applied by the preprocessing module based on image representation parameters, wherein:

[0032] The increase in the upper boundary threshold for outlier filtering in the preprocessing module is positively correlated with the image representation parameters.

[0033] Furthermore, the data correction module is used to correct the Canny operator double threshold of the edge detection algorithm for segmenting gestational sacs or embryos in each ultrasound image based on body mass index, wherein:

[0034] The increase in the dual threshold of the Canny operator is positively correlated with the body mass index.

[0035] Compared with existing technologies, the beneficial effects of this invention are that the verification module determines whether the data acquisition module is operating correctly, and obtains the number of potential data points by summing the number of missing data points and significantly abnormal data points in the user data group. Missing data makes the data incomplete, while significantly abnormal data is caused by collection or input errors. Both missing data and significantly abnormal data affect data quality and the accuracy of subsequent predictions. The amount of problematic data in the data is comprehensively considered by combining missing data and significantly abnormal data. The preset necessary data quantity is the minimum amount of data required for accurate pregnancy risk prediction. The abnormal data rate reflects the degree of abnormality of the user data group relative to the preset necessary data quantity; the smaller the ratio, the more complete the data. When the abnormal data rate is greater than the preset abnormal data rate, it indicates that the data incompleteness is relatively serious, indicating that the data acquisition module is malfunctioning; when the abnormal data rate is less than or equal to the preset abnormal data rate, the data completeness meets the requirements, and the data acquisition module is operating correctly. By quantifying the abnormal data rate to judge the operating status of the data acquisition module, potential problems in the data acquisition process can be detected in a timely manner, ensuring the quality of data entering the subsequent prediction process, thereby improving the accuracy and reliability of pregnancy risk prediction.

[0036] Furthermore, the verification module corrects the acquisition of user data sets based on image representation parameters. The sharpness factor is the ratio of the gradient mean of a single ultrasound image to a preset gradient mean. The gradient mean reflects the image's sharpness; a low mean indicates overall image blurriness and low contrast. The sharpness factor reflects the sharpness of a single ultrasound image relative to a standard image. The gradient variance reflects the richness of image texture; low variance indicates a monotonous image texture, likely due to an incorrect scan area that only scanned uniform tissue. The completeness factor reflects the completeness of a single ultrasound image relative to a standard image. The image representation parameters collectively reflect the overall quality of all ultrasound images. A low gradient mean indicates overall image blurriness and low contrast; a low gradient variance indicates a monotonous image texture, likely due to an incorrect scan area that only scanned uniform tissue. If the image representation parameters are less than or equal to the preset image representation parameters, it indicates an image acquisition anomaly leading to incomplete data; the acquisition parameters for each ultrasound image are then corrected. If the image representation parameters are greater than the preset image representation parameters, it indicates the image is acceptable, but incomplete data is due to preprocessing issues; the preprocessing parameters for the detection data are then corrected. By quantitatively assessing the clarity and completeness of ultrasound images, the specific circumstances of incomplete data can be identified, allowing for targeted corrections, improving data quality and usability, and ultimately enhancing the accuracy of pregnancy risk prediction.

[0037] Furthermore, the data correction module adjusts the acquisition parameters for each ultrasound image. When the sharpness factor is less than or equal to the preset sharpness factor, it indicates poor image clarity, determined to be due to a conflict between the image acquisition frame rate and the Nyquist frequency of the fetal heart rate. The average fetal heart rate reflects the fetal heartbeat; the faster the fetal heart rate, the higher the image acquisition frame rate is needed to accurately capture fetal dynamics. The image acquisition frame rate is adjusted based on the average fetal heart rate to match the fetal heart rate, improving image clarity. The guiding scan range is redefined based on the intensity gradient mutation values ​​of each ultrasound image. When the integrity factor is less than or equal to the preset integrity factor, it indicates poor image integrity, determined to be due to inaccurate scanning range. Intensity gradient mutation values ​​reflect changes in tissue boundaries within the image. Analyzing these values ​​helps to redefine the guiding scan range, improving image integrity. Based on the issues of image clarity and integrity, the image acquisition parameters are corrected using both the average fetal heart rate and intensity gradient mutation values ​​to improve the clarity and integrity of ultrasound images, thereby obtaining more accurate quantitative detection data and providing better data support for pregnancy risk prediction.

[0038] Furthermore, the guided scanning range is redefined based on the intensity gradient abrupt change values ​​of each ultrasound image. Sobel operator edge detection and average gradient amplitude statistics for banded regions are used. The Sobel operator is a commonly used edge detection operator; by performing Sobel operator edge detection on a single ultrasound image, edge information in the image can be highlighted. Banded regions extending 10% of the image width inward along the four boundaries of the ultrasound image are obtained, and the average gradient amplitude of each banded region is calculated. Changes in gradient amplitude in the image edge regions reflect information about anatomical structures. If there are significant grayscale changes in the edge regions, it indicates that meaningful anatomical structure boundaries have been scanned. The average of the average gradient amplitudes of each ultrasound image is calculated to obtain the intensity gradient abrupt change value, which comprehensively reflects the grayscale changes in the edge regions of all ultrasound images. A low intensity gradient abrupt change value means that the grayscale changes in the image edge regions are not significant, indicating an insufficient scanning range and failure to cover the complete anatomical structure. Therefore, the increase in the guided scanning range is negatively correlated with the intensity gradient abrupt change value; the lower the intensity gradient abrupt change value, the greater the increase in the guided scanning range is needed to ensure that the complete target area is scanned. By analyzing the gradient amplitude of the edge regions of ultrasound images, the specific scanning range can be determined. The scanning range can be dynamically adjusted based on abrupt changes in intensity gradient values, improving the completeness of the ultrasound examination and obtaining more comprehensive quantitative data to provide a more accurate basis for pregnancy risk prediction.

[0039] Furthermore, the image acquisition frame rate for ultrasound images is adjusted based on the average fetal heart rate. The average fetal heart rate reflects the frequency of the fetal heartbeat. In ultrasound examinations, to accurately capture the fetal heart's motion signals, the image acquisition frame rate needs to match the fetal heart rate. According to the Nyquist sampling theorem, the sampling frequency must be at least twice the highest frequency of the signal to accurately reproduce the signal. The faster the fetal heart rate, the higher the frequency of its cardiac motion signals, requiring a higher image acquisition frame rate to avoid signal aliasing and ensure image clarity and accuracy. The increase in the image acquisition frame rate for acquiring ultrasound images is positively correlated with the average fetal heart rate. Dynamically adjusting the image acquisition frame rate based on the actual fetal heart rate improves the quality of ultrasound images, thereby obtaining more accurate fetal information and enabling more precise prediction of pregnancy risks. Attached Figure Description

[0040] Figure 1 This is a block diagram of a pregnancy risk prediction system for early pregnancy ultrasound examination according to an embodiment of the present invention;

[0041] Figure 2 This is a logic diagram of the verification module in an embodiment of the present invention, which determines whether the operation of the data acquisition module is qualified based on the abnormal data rate.

[0042] Figure 3This is a logical decision diagram for the verification module of the present invention to correct the acquired parameters of the user data group based on the image representation parameters.

[0043] Figure 4 This invention provides a logic diagram for correcting the acquisition parameters of each ultrasound examination image in the data correction module of this embodiment. Detailed Implementation

[0044] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0045] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0046] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0047] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0048] Please see Figure 1 The diagram shown is a block diagram of a pregnancy risk prediction system for early pregnancy ultrasound examination according to an embodiment of the present invention; the pregnancy risk prediction system for early pregnancy ultrasound examination according to an embodiment of the present invention includes:

[0049] The data acquisition module includes an image analysis unit for acquiring quantitative detection data from each ultrasound examination image based on the acquired user medical records, and a data acquisition unit for acquiring digital detection data from the user based on the user medical records.

[0050] A preprocessing module, which is connected to the data acquisition module, is used to preprocess the detection data to obtain user data sets;

[0051] A verification module, connected to the preprocessing module, is used to determine whether the operation of the data acquisition module is qualified based on the abnormal data rate of the user data group, and, when it is determined that the operation of the data acquisition module is abnormal, to correct the acquisition of the user data group based on the image representation parameters.

[0052] A data correction module, which is connected to the preprocessing module and the verification module respectively, is used to correct the preprocessing parameters for the detection data;

[0053] The acquisition and calibration module is connected to both the data acquisition module and the calibration module, and is used to correct the acquisition parameters for each ultrasound examination image.

[0054] Specifically, pregnancy risk prediction is based on qualified data sets.

[0055] Specifically, the method for predicting pregnancy risk based on a qualified dataset is not limited. It may include model building using a large amount of historical early pregnancy ultrasound data and corresponding pregnancy outcomes as a training set. The features are the standardized feature vectors of the user data set. Pregnancy outcomes include continued pregnancy into the mid-to-late stages, early miscarriage, and ectopic pregnancy. A neural network algorithm can be used for training, outputting the risk probabilities of various adverse pregnancy outcomes. A prediction output is then performed, calculating a risk score for the currently input user data set.

[0056] Specifically, digital testing data refers to data that can be directly read from a user's medical record, which may include basic patient information, obstetric history, current pregnancy information and embryo transfer information, clinical symptoms and signs, and serum biomarkers.

[0057] Patient basic information includes age, weight, and height. Obstetric history includes parity, parity, history of miscarriage, history of preterm birth, and history of live birth. Current pregnancy information includes date of last menstrual period, type of assisted reproductive technology and embryo information, serum biomarkers including serum human chorionic gonadotropin (hCG) and progesterone levels, and gestational age.

[0058] Specifically, the quantitative detection data includes detection data that can be read from ultrasound images, which may include the average diameter and location coordinates of the gestational sac; the diameter of the yolk sac; the crown-rump length of the embryo; and the size information of abnormal structures marked by image segmentation and recognition technology, including subchorionic hematoma.

[0059] Specifically, preprocessing includes outlier filtering of the detection data, missing value imputation based on a medical statistical model, Z-score standardization of numerical features, and the construction of key derived features to ultimately output a standardized user feature vector, i.e., the user data set. Key derived features include the gestational sac-embryo size ratio.

[0060] Outlier filtering employs a combination of hard filtering based on medical reference ranges and soft filtering based on statistical methods. The statistical method can be the 3σ principle, which is existing technology and will not be elaborated further.

[0061] Please see Figure 2 As shown, this is a logic diagram illustrating how the verification module of this invention determines the operational qualification of the data acquisition module based on the abnormal data rate. The verification module of this invention is used to determine the operational qualification of the data acquisition module based on the abnormal data rate of the user data group, including:

[0062] The number of missing data and the sum of the number of significantly abnormal data in the user data group are obtained to obtain the number of potentially problematic data. The ratio of the number of potentially problematic data to the preset number of necessary data is calculated to obtain the abnormal data rate.

[0063] When the abnormal data rate is less than or equal to the preset abnormal data rate, the data acquisition module is deemed to be operating successfully, and the user data group is marked as a successful data group.

[0064] When the abnormal data rate exceeds the preset abnormal data rate, the operation of the data acquisition module is determined to be abnormal, and the acquisition parameters of the user data group are corrected based on the image representation parameters.

[0065] Specifically, the preset abnormal data rate is selected within the range [0.04, 0.08]. Those skilled in the art can select this range themselves. Based on retrospective analysis of historical data, the minimum abnormal data rate threshold required to maintain more than 95% performance of the prediction model can be calculated. In this embodiment, 0.06 is preferably selected.

[0066] Specifically, each data point in the user data set has a pre-defined confidence range threshold. Data exceeding the corresponding pre-defined confidence range threshold is identified as significantly abnormal data. Each pre-defined confidence range threshold can be determined based on medical statistical results from a large-scale normal early pregnancy population, using the percentile method.

[0067] Specifically, the verification module determines whether the data acquisition module is functioning correctly by summing the number of missing data and significantly abnormal data in the user data set to obtain the number of potential data points. Missing data makes the data incomplete, while significantly abnormal data is caused by collection or input errors. Both missing and significantly abnormal data affect data quality and the accuracy of subsequent predictions. The amount of problematic data is assessed by comprehensively considering both missing and significantly abnormal data. The preset necessary data quantity is the minimum amount of data required for accurate pregnancy risk prediction. The abnormal data rate reflects the degree of abnormality of the user data set relative to the preset necessary data quantity; the smaller the ratio, the more complete the data. When the abnormal data rate is greater than the preset abnormal data rate, it indicates a serious data incompleteness, suggesting an abnormality in the data acquisition module. When the abnormal data rate is less than or equal to the preset abnormal data rate, the data completeness meets the requirements, and the data acquisition module is functioning correctly. By quantifying the abnormal data rate to judge the operational status of the data acquisition module, potential problems during data acquisition can be identified in a timely manner, ensuring the quality of data entering the subsequent prediction process, thereby improving the accuracy and reliability of pregnancy risk prediction.

[0068] Please see Figure 3 As shown, this is a logic decision diagram of the verification module of this invention for correcting the acquired parameters of a user data set based on image representation parameters. The verification module of this invention is used to correct the acquired parameters of a user data set based on image representation parameters, including:

[0069] For a single ultrasound image, the ratio of its gradient mean to the preset gradient mean is obtained to obtain the clarity factor.

[0070] For a single ultrasound image, the ratio of its gradient variance to a preset gradient variance is obtained to obtain the complete factor.

[0071] The image characterization factors for a single ultrasound image are obtained by summing the corresponding weight coefficients assigned to the clarity factor and the integrity factor.

[0072] The average value of the image characterization factors for each ultrasound image is calculated to obtain the image characterization parameters;

[0073] If the image representation parameter is less than or equal to the preset image representation parameter, then the acquisition parameters for each ultrasound image are determined to be corrected.

[0074] If the image representation parameters are greater than the preset image representation parameters, then the preprocessing parameters for the detection data are determined to be corrected.

[0075] Specifically, the preset gradient mean can be calculated by taking the average of each gradient mean from a large-scale, high-quality standard ultrasound image database of early pregnancy, and obtaining the preset benchmark value; the standard ultrasound image database of early pregnancy includes ultrasound examination images marked by experts as having excellent image quality;

[0076] The preset gradient variance can be obtained by calculating the average of each gradient variance from a large-scale, high-quality database of standard ultrasound images of early pregnancy.

[0077] Specifically, the weighting coefficients for both the clarity factor and the integrity factor are 0.5, in order to comprehensively consider the actual situation of the ultrasound images.

[0078] Specifically, the preset image representation parameters are selected within the range [0.75, 0.81]. Those skilled in the art can select and determine these parameters themselves. In quality control experiments, the image representation parameters can be compared with the binary evaluation of whether the image is suitable for diagnosis by experts using ROC analysis, and the critical value corresponding to the maximum Youden index can be selected as the preset image representation parameter. In this embodiment, preferably, the preset image representation parameter is 0.81.

[0079] Specifically, the verification module corrects the acquisition of user data groups based on image representation parameters. The sharpness factor is the ratio of the gradient mean of a single ultrasound image to a preset gradient mean. The gradient mean reflects the image's sharpness; a low mean indicates overall image blurriness and low contrast. The sharpness factor reflects the sharpness of a single ultrasound image relative to a standard image. The gradient variance reflects the richness of image texture; low variance indicates monotonous texture, likely due to an incorrect scan area that only scanned uniform tissue. The integrity factor reflects the completeness of a single ultrasound image relative to a standard image. The image representation parameters collectively reflect the overall quality of all ultrasound images. A low gradient mean indicates overall image blurriness and low contrast; a low gradient variance indicates monotonous texture, likely due to an incorrect scan area that only scanned uniform tissue. If the image representation parameters are less than or equal to the preset image representation parameters, it indicates an image acquisition anomaly leading to incomplete data; the acquisition parameters for each ultrasound image are then corrected. If the image representation parameters are greater than the preset image representation parameters, it indicates the image is acceptable, but incomplete data is due to preprocessing issues; the preprocessing parameters for the detection data are then corrected. By quantitatively assessing the clarity and completeness of ultrasound images, the specific circumstances of incomplete data can be identified, allowing for targeted corrections, improving data quality and usability, and ultimately enhancing the accuracy of pregnancy risk prediction.

[0080] Specifically, the data correction module is used to correct the preprocessing parameters for the detection data, including:

[0081] The upper boundary threshold for outlier filtering in the preprocessing module is corrected based on image representation parameters.

[0082] The Canny operator double threshold is used to correct the edge detection algorithm for segmenting gestational sacs or embryos in various ultrasound images based on body mass index.

[0083] Please see Figure 4 The diagram shown illustrates the logic decision of the data correction module in this embodiment of the invention for correcting the acquisition parameters of each ultrasound examination image. The data correction module of this invention is used to correct the acquisition parameters of each ultrasound examination image, including:

[0084] Under the condition that the clarity factor is less than or equal to the preset clarity factor, the frame rate of ultrasound examination image acquisition is corrected based on the average fetal heart rate.

[0085] Under the condition that the integrity factor is less than or equal to the preset integrity factor, the guiding scan range is redetermined based on the intensity gradient mutation value.

[0086] Specifically, the data correction module adjusts the acquisition parameters for each ultrasound image. When the sharpness factor is less than or equal to the preset sharpness factor, it indicates poor image clarity, determined to be due to a conflict between the image acquisition frame rate and the Nyquist frequency of the fetal heart rate. The average fetal heart rate reflects the fetal heartbeat; the faster the fetal heart rate, the higher the image acquisition frame rate is needed to accurately capture fetal dynamics. The image acquisition frame rate is adjusted based on the average fetal heart rate to match the fetal heart rate, improving image clarity. The guiding scan range is redefined based on the intensity gradient mutation values ​​of each ultrasound image. When the integrity factor is less than or equal to the preset integrity factor, it indicates poor image integrity, determined to be due to inaccurate scanning range. Intensity gradient mutation values ​​reflect changes in tissue boundaries within the image. Analyzing these values ​​helps to redefine the guiding scan range, improving image integrity. Based on the issues of image clarity and integrity, the image acquisition parameters are corrected using both the average fetal heart rate and intensity gradient mutation values ​​to improve the clarity and integrity of ultrasound images, thereby obtaining more accurate quantitative detection data and providing better data support for pregnancy risk prediction.

[0087] Specifically, the preset clarity factor is selected within the range [0.82, 0.85]. Those skilled in the art can select and determine it themselves. ROC analysis can be used to find the critical value that best distinguishes data that is unqualified due to clarity issues. In this embodiment, the preset clarity factor is preferably 0.85.

[0088] Specifically, the preset integrity factor is selected within the range [0.83, 0.87]. Those skilled in the art can select and determine it themselves. ROC analysis can be used to find the critical value that best distinguishes data non-compliance caused by abnormal data rate problems. In this embodiment, the preset integrity factor is preferably 0.85.

[0089] Specifically, the data correction module is used to redetermine the guiding scan range based on the intensity gradient abrupt changes in each ultrasound image, including:

[0090] Sobel operator edge detection is performed on a single ultrasound image to obtain a band-shaped region extending 10% of the image width inward along the four boundaries of the ultrasound image, and the average gradient magnitude of each band-shaped region is calculated.

[0091] The average gradient amplitude of each ultrasound image is calculated to obtain the intensity gradient abrupt change value.

[0092] The increase in the guided scanning range is negatively correlated with the intensity gradient abrupt change value.

[0093] Specifically, the guided scanning range is redefined based on the intensity gradient abrupt change values ​​of each ultrasound image. Sobel operator edge detection and average gradient amplitude statistics for banded regions are used. The Sobel operator is a commonly used edge detection operator; by applying Sobel operator edge detection to a single ultrasound image, edge information in the image can be highlighted. Banded regions extending 10% of the image width inward along the four boundaries of the ultrasound image are obtained, and the average gradient amplitude of each banded region is calculated. Changes in gradient amplitude in the image edge regions reflect information about anatomical structures. If there are significant grayscale changes in the edge regions, it indicates that meaningful anatomical structure boundaries have been scanned. The average of the average gradient amplitudes of each ultrasound image is calculated to obtain the intensity gradient abrupt change value, which comprehensively reflects the grayscale changes in the edge regions of all ultrasound images. A low intensity gradient abrupt change value means that the grayscale changes in the image edge regions are not significant, indicating an insufficient scanning range and failure to cover the complete anatomical structure. Therefore, the increase in the guided scanning range is negatively correlated with the intensity gradient abrupt change value; the lower the intensity gradient abrupt change value, the greater the increase in the guided scanning range is needed to ensure that the complete target area is scanned. By analyzing the gradient amplitude of the edge regions of ultrasound images, the specific scanning range can be determined. The scanning range can be dynamically adjusted based on abrupt changes in intensity gradient values, improving the completeness of the ultrasound examination and obtaining more comprehensive quantitative data to provide a more accurate basis for pregnancy risk prediction.

[0094] In this embodiment, optionally:

[0095] Compare the intensity gradient abrupt change value with the first preset abrupt change comparison value and the second preset abrupt change comparison value;

[0096] If the intensity gradient mutation value is less than or equal to the first preset mutation comparison value, the guidance scan range will be adjusted to 1.3 times the initial guidance scan range;

[0097] If the intensity gradient mutation value is less than or equal to the second preset mutation comparison value and greater than the first preset mutation comparison value, the guidance scan range will be adjusted to 1.2 times the initial guidance scan range.

[0098] If the intensity gradient mutation value is greater than the second preset mutation comparison value, the guidance scan range will be adjusted to 1.1 times the initial guidance scan range;

[0099] Specifically, the gradient magnitude output by the Sobel operator is normalized to a range between [0,1]. The first preset mutation comparison value is 0.15, and the second preset mutation comparison value is 0.30.

[0100] Specifically, methods for normalizing the gradient magnitude of the Sobel operator output can include:

[0101] To calculate the gradient magnitude, apply the Sobel operator in the horizontal (Gx) and vertical (Gy) directions to the ultrasound image for convolution, and calculate the gradient magnitude of each pixel: G = sqrt(Gx^2 + Gy^2).

[0102] The result is a matrix of the same size as the original image, with values ​​representing non-negative gradient strengths.

[0103] To determine the maximum value, iterate through the entire gradient magnitude matrix and find the maximum value G_max. This represents the most significant edge intensity in the current image.

[0104] Normalization is performed by dividing each value in the gradient magnitude matrix by the maximum value G_max, i.e., G_normalized = G / G_max. This ensures that all values ​​are linearly scaled to the interval [0, 1], where 0 represents no edges (uniform regions) and 1 represents the strongest edges in the current image. G_normalized is the normalized value obtained by dividing the gradient magnitude G calculated by the original Sobel operator by the maximum gradient magnitude G_max in the image.

[0105] If the intensity gradient abrupt change value is low, the scan is determined to be insufficient. A new suggested scan region is generated.

[0106] The intensity gradient abrupt change value reflects whether there is a significant gray-scale change in the edge region of a single image due to anatomical structure.

[0107] The guided scanning range is the area covered by which the operator is advised to move the probe.

[0108] Specifically, the data correction module is used to correct the image acquisition frame rate of ultrasound examination images based on the average fetal heart rate, wherein:

[0109] The increase in the image acquisition frame rate of ultrasound examinations is positively correlated with the average fetal heart rate.

[0110] Specifically, the average fetal heart rate can be obtained by analyzing the periodic motion signals of the embryonic heart region in the time-domain image sequence.

[0111] In this embodiment, optionally:

[0112] F_set = max(F_base, k * F_hr);

[0113] F_base is the preset base frame rate, which is optionally set to 60fps in a single embodiment. F_set is the adjusted image acquisition frame rate. k is a safety factor selected within the range [2, 3]. F_hr = 2 * (average fetal heart rate / 60). The average fetal heart rate is in beats per minute (bpm), divided by 60 to convert it to frequency (Hz).

[0114] Specifically, the image acquisition frame rate for ultrasound images is adjusted based on the average fetal heart rate. The average fetal heart rate reflects the frequency of the fetal heartbeat. In ultrasound examinations, to accurately capture the fetal heart's motion signals, the image acquisition frame rate needs to match the fetal heart rate. According to the Nyquist sampling theorem, the sampling frequency must be at least twice the highest frequency of the signal to accurately reproduce the signal. The faster the fetal heart rate, the higher the frequency of its cardiac motion signals, requiring a higher image acquisition frame rate to avoid signal aliasing and ensure image clarity and accuracy. The increase in the image acquisition frame rate for acquiring ultrasound images is positively correlated with the average fetal heart rate. Dynamically adjusting the image acquisition frame rate based on the actual fetal heart rate improves the quality of ultrasound images, thereby obtaining more accurate fetal information and enabling more precise prediction of pregnancy risks.

[0115] Specifically, the data correction module is used to correct the preprocessing parameters for the detection data, wherein:

[0116] The increase in the upper boundary threshold for outlier filtering in the preprocessing module is positively correlated with the image representation parameters.

[0117] Specifically, the preprocessing parameters for the test data are modified. Image characterization parameters comprehensively reflect quality information such as the clarity and completeness of ultrasound images. Higher image characterization parameters indicate higher image reliability and relatively higher data accuracy. In this case, a more lenient outlier filtering boundary can be used, as some seemingly abnormal data may appear to be actual physiological changes rather than acquisition errors. The increase in the upper boundary threshold for outlier filtering in the preprocessing module is positively correlated with the image characterization parameters. Dynamically adjusting the upper boundary threshold for outlier filtering based on image quality allows for more reasonable processing of test data, effectively excluding genuine outliers, improving data usability and the accuracy of pregnancy risk prediction.

[0118] In this embodiment, optionally:

[0119] Compare the image representation parameters with the first preset image comparison value and the second preset image comparison value;

[0120] When the image representation parameter is less than or equal to the first preset image comparison value, the upper boundary threshold for outlier filtering in the preprocessing module is adjusted to 1.09 times the corresponding upper boundary threshold.

[0121] When the image representation parameter is less than or equal to the second preset image comparison value and greater than the first preset image comparison value, the upper boundary threshold for outlier filtering in the preprocessing module is adjusted to 1.13 times the corresponding upper boundary threshold.

[0122] When the image representation parameter is greater than the second preset image comparison value, the upper boundary threshold for outlier filtering in the preprocessing module is adjusted to 1.19 times the corresponding upper boundary threshold.

[0123] The first preset image comparison value is 0.83, and the second preset image comparison value is 0.86.

[0124] Specifically, the data correction module is used to correct the Canny operator double threshold of the edge detection algorithm for segmenting gestational sacs or embryos in each ultrasound image based on body mass index, wherein:

[0125] The increase in the dual threshold of the Canny operator is positively correlated with the body mass index.

[0126] Specifically, this study modifies the Canny operator's dual thresholds for edge detection algorithms segmenting gestational sacs or embryos in various ultrasound images based on body mass index (BMI). BMI can serve as a proxy variable for fat thickness. In ultrasound examinations, higher BMI leads to more severe sound attenuation, resulting in greater interference from background noise on edge extraction. The Canny operator's dual thresholds (low and high thresholds) are used to determine edges in the image. By increasing the dual thresholds to suppress the interference of background noise caused by enhanced sound attenuation on edge extraction, edge detection becomes more accurate. The increase in the Canny operator's dual thresholds is positively correlated with BMI. Dynamically adjusting the Canny operator's dual thresholds based on the pregnant woman's BMI effectively suppresses the interference of background noise on edge extraction, improves the accuracy of gestational sac and embryo edge detection, and thus obtains more precise quantitative detection data of gestational sacs and embryos, providing a more reliable basis for pregnancy risk prediction.

[0127] In this embodiment, optionally:

[0128] Compare the body mass index with the first preset index comparison value and the second preset index comparison value;

[0129] When the body mass index is less than or equal to the first preset index comparison value, the two thresholds of the Canny operator are adjusted to 1.11 times the corresponding thresholds.

[0130] When the body mass index is less than or equal to the second preset index comparison value and greater than the first preset index comparison value, the two thresholds of the Canny operator are adjusted to 1.19 times the corresponding thresholds respectively;

[0131] When the body mass index is greater than the second preset index comparison value, the two thresholds of the Canny operator are adjusted to 1.23 times the corresponding thresholds.

[0132] The first preset index comparison value is 24 kg / m², and the second preset index comparison value is 28 kg / m².

[0133] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0134] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A pregnancy risk prediction system for early pregnancy ultrasound examination, characterized in that, include: The data acquisition module acquires quantitative detection data from various ultrasound examination images and digital detection data from user medical records. A preprocessing module, which is connected to the data acquisition module, is used to preprocess the detection data to obtain user data sets; A verification module, connected to the preprocessing module, is used to determine whether the data acquisition module is functioning correctly based on the abnormal data rate of the user data group, and, when the data acquisition module is determined to be malfunctioning, to correct the acquisition parameters for the user data group based on image representation parameters, including: For a single ultrasound image, the ratio of its gradient mean to the preset gradient mean is obtained to obtain the clarity factor. For a single ultrasound image, the ratio of its gradient variance to a preset gradient variance is obtained to obtain the complete factor. The image characterization factors for a single ultrasound image are obtained by summing the corresponding weight coefficients assigned to the clarity factor and the integrity factor. The average value of the image characterization factors for each ultrasound image is calculated to obtain the image characterization parameters; If the image representation parameter is less than or equal to the preset image representation parameter, then the acquisition parameters for each ultrasound image are determined to be corrected. If the image representation parameter is greater than the preset image representation parameter, then the preprocessing parameters for the detection data are determined to be corrected. A data correction module, which is connected to the preprocessing module, the data acquisition module, and the verification module, corrects the preprocessing parameters for the detection data. This includes correcting the upper boundary threshold of the preprocessing module for outlier filtering based on image representation parameters, and correcting the Canny operator double threshold of the edge detection algorithm for segmenting gestational sacs or embryos in each ultrasound image when the data acquisition module acquires quantified detection data based on body mass index. The increase in the upper boundary threshold of the preprocessing module for outlier filtering is positively correlated with the image representation parameters, and the increase in the Canny operator double threshold is positively correlated with the body mass index. An acquisition and correction module, connected to both the data acquisition module and the verification module, is used to correct the acquisition parameters for each ultrasound image, including: Under the condition that the clarity factor is less than or equal to the preset clarity factor, the frame rate of ultrasound examination image acquisition is corrected based on the average fetal heart rate. Under the condition that the integrity factor is less than or equal to the preset integrity factor, the guiding scan range is redetermined based on the intensity gradient mutation value.

2. The pregnancy risk prediction system for early pregnancy ultrasound examination according to claim 1, characterized in that, The verification module is used to determine whether the operation of the data acquisition module is qualified based on the abnormal data rate of the user data group. This includes determining that the operation of the data acquisition module is abnormal when the abnormal data rate is greater than the preset abnormal data rate, and correcting the acquisition parameters for the user data group based on the image representation parameters.

3. The pregnancy risk prediction system for early pregnancy ultrasound examination according to claim 2, characterized in that, The verification module is used to determine the abnormal data rate of the user data group, including obtaining the sum of the number of missing data and the number of significantly abnormal data in the user data group to obtain the number of potentially problematic data. This is used to calculate the ratio of the number of potential data points to the preset necessary data points, thus obtaining the abnormal data rate.

4. The pregnancy risk prediction system for early pregnancy ultrasound examination according to claim 3, characterized in that, The acquisition and correction module is used to redetermine the guiding scan range based on the intensity gradient abrupt change value, including: Sobel operator edge detection is performed on a single ultrasound image to obtain a band-shaped region extending 10% of the image width inward along the four boundaries of the ultrasound image, and the average gradient magnitude of each band-shaped region is calculated. The average gradient amplitude of each ultrasound image is calculated to obtain the intensity gradient abrupt change value. The increase in the guided scanning range is negatively correlated with the intensity gradient abrupt change value.

5. The pregnancy risk prediction system for early pregnancy ultrasound examination according to claim 4, characterized in that, The acquisition and correction module is used to correct the image acquisition frame rate of ultrasound examination images based on the average fetal heart rate, wherein: The increase in the image acquisition frame rate of ultrasound examinations is positively correlated with the average fetal heart rate.

Images