Fetal gestational week evaluation method and device based on multi-section ultrasonic image measurement

Abstract

The invention relates to the technical field of medical image analysis and diagnosis, in particular to a fetal gestational week evaluation method and device based on multi-section ultrasonic image measurement, and the method comprises the steps: determining a semantic segmentation training data set based on a skull cross section, an abdominal section and a femur section in a target fetal ultrasonic image, constructing a target measurement model meeting a first preset condition; determining a gestational week prediction data set based on a plurality of measurement values in the target fetal ultrasound image and gestational week data corresponding to the plurality of measurement values so as to construct a gestational week prediction model meeting a second preset condition; and splicing the target measurement model and the gestational week prediction model to obtain a fetal gestational week evaluation model for evaluating the gestational week of the fetus. Therefore, the problems that in the prior art, manual examination and analysis of doctors mainly depend on, a large amount of labor cost needs to be consumed, the examination results of different doctors are different, and the accuracy of fetal gestational week evaluation is reduced are solved.

Description

Technical Field

[0001] This application relates to the technical field of medical image analysis and diagnosis, and particularly relates to a method and device for evaluating fetal gestational age based on multi-plane ultrasound image measurements. Background Art

[0002] In related technologies, the fetal gestational age assessment method can be based on the last menstrual cycle. This method determines the conception date as 14 days before the last menstrual date. However, factors such as irregular menstruation of pregnant women, unclear last menstrual date, and use of oral contraceptives will all affect the accuracy of this method. Therefore, the fetal gestational age can also be evaluated through ultrasound examination. This method first performs an ultrasound examination on the fetus, and then analyzes and processes the ultrasound images to evaluate the fetal gestational age. For example, the commonly used Hadlock formula in clinical practice. Doctors input the biometric measurement values obtained from the ultrasound images into this regression formula to calculate the fetal gestational age value.

[0003] However, in related technologies, the method of evaluating fetal gestational age through ultrasound examination mainly relies on manual examination and analysis by doctors, which requires a large amount of labor costs, and the examination standards of different doctors are not unified, resulting in differences in examination results, reducing the accuracy and reliability of fetal gestational age assessment, and urgently need to be solved. Summary of the Invention

[0004] This application is based on the inventor's recognition of the following problems:

[0005] The gestational age of a fetus refers to the time from conception to the present. Accurately and quickly determining the fetal gestational age is crucial for prenatal care and the diagnosis of diseases such as fetal developmental abnormalities and premature birth. Taking premature birth as an example, premature birth will bring the risk of lifelong disability to the child and increase the subsequent health care costs. If accurate gestational age assessment can be achieved, it will help to identify the premature birth risk of the fetus at an early stage, so as to take effective measures for intervention.

[0006] The traditional methods for assessing the gestational age of a fetus are mainly the following two: (1) The assessment method based on the last menstrual cycle. This method determines the conception date as 14 days before the date of the last menstrual period. However, factors such as the irregular menstruation of the pregnant woman, the uncertainty about her own last menstrual date, and the use of oral contraceptives will all affect the accuracy of this method. (2) The assessment method based on ultrasound examination. This method first performs an ultrasound examination on the fetus, and then analyzes and processes the ultrasound images to assess the gestational age of the fetus. For example, the Hadlock formula commonly used in clinical practice. Doctors input the biometric measurement values obtained from the analysis of the ultrasound images into this regression formula to calculate the gestational age value of the fetus. Although the accuracy of this method is better than that of the assessment method based on the last menstrual cycle, it relies on the manual examination and analysis of doctors, requires a large amount of labor costs, and is easily affected by the differences in examination habits among doctors. Therefore, an automated method that can perform ultrasound image analysis and gestational age assessment is very necessary.

[0007] With the rapid development of deep learning, deep neural networks have been widely used in the field of medical imaging for the analysis and diagnosis of medical images. However, there are few methods for using deep neural network technology to assess gestational age. On the one hand, existing deep neural network technologies require a large amount of manually labeled data for training. The acquisition of fetal ultrasound images requires a large amount of labor costs, and the data labeling also needs to be manually annotated by ultrasound doctors, which is quite difficult, resulting in a lack of sufficient data for training and evaluation. On the other hand, existing methods for using deep learning to assess gestational age use traditional medical segmentation models to segment the biometric features of ultrasound images. However, traditional medical segmentation models have disadvantages such as limited receptive fields, insufficient fusion of local and global information, and being greatly affected by ultrasound image noise, resulting in poor accuracy of medical segmentation. In addition, the labeling habits of ultrasound doctors are not uniform, resulting in certain differences in data labeling. Therefore, how to make the deep learning model balance the labeling differences among different doctors and achieve standardized and unified biometric feature extraction is also a very challenging task.

[0008] This application provides a method and device for assessing the gestational age of a fetus based on multi-plane ultrasound image measurements, so as to solve the problems in the related technologies. The method of assessing the gestational age of a fetus by ultrasound examination mainly relies on the manual examination and analysis of doctors, requires a large amount of labor costs, and the examination standards of different doctors are not unified, resulting in differences in examination results and reducing the accuracy and reliability of fetal gestational age assessment.

[0009] The first aspect embodiment of the present application provides a method for evaluating the gestational age of a fetus based on multi-plane ultrasound image measurements, which is applied to the model construction stage and includes the following steps: determining a semantic segmentation training data set based on the transverse cranial plane, abdominal plane, and femoral plane in the target fetal ultrasound image, and constructing a target measurement model that meets the first preset condition by using the semantic segmentation training data set; determining a gestational age prediction data set based on multiple measurement values in the target fetal ultrasound image and the gestational age data corresponding to the multiple measurement values, and constructing a gestational age prediction model that meets the second preset condition by using the gestational age prediction data set; splicing the target measurement model and the gestational age prediction model to obtain a fetal gestational age evaluation model for evaluating the gestational age of a fetus.

[0010] Optionally, in an embodiment of the present application, the determining a semantic segmentation training data set based on the transverse cranial plane, abdominal plane, and femoral plane in the target fetal ultrasound image includes: using the transverse cranial plane, abdominal plane, and femoral plane in the target fetal ultrasound image as the target key planes for predicting the gestational age; determining the target key parts corresponding to the target key planes, where the key part of the transverse cranial plane is the skull, the key part of the abdominal plane is the abdomen, and the key part of the femoral plane is the femur; drawing a target area in the corresponding first target plane sample of the target key part to generate a mask image including the target key part, and corresponding the first target plane sample with the mask image to determine the semantic segmentation training data set.

[0011] Optionally, in an embodiment of the present application, the constructing a target measurement model that meets the first preset condition by using the semantic segmentation training data set includes: constructing a measurement training data set based on the semantic segmentation training data set, and performing data preprocessing on the measurement training data set to determine the processed measurement training data set; combining a target point regression deep neural network model with a pre-constructed target semantic segmentation network model to construct a measurement neural network; combining the measurement neural network with a semantic segmentation objective function to determine a measurement hybrid objective function; training the measurement neural network by using the processed measurement training data set and the measurement hybrid objective function to obtain a trained measurement neural network; determining a target measurement model that meets the first preset condition according to the trained measurement neural network.

[0012] Optionally, in an embodiment of the present application, the data preprocessing of the measurement training dataset to determine the processed measurement training dataset includes: taking the head circumference and biparietal diameter of the cranial cross-section, the abdominal circumference of the abdominal section, and the femur length of the femoral section in the target fetal ultrasound image as target key measurement values for predicting the gestational week; using the target key measurement values to determine the target measurement key points of the target fetal ultrasound image, where the measurement key points of the head circumference and biparietal diameter are the perimeter of the head and the maximum distance between the two parietal bones on both sides of the head, the measurement key point of the abdominal circumference is the perimeter of the abdomen, and the measurement key point of the femur length is the longest axis length of the femur; drawing target line segments and target points of the target measurement key points in the corresponding second target section samples to obtain a plotted image, recording the data in the second target section samples and the data in the plotted image, and combining with the semantic segmentation training dataset to determine the processed measurement training dataset.

[0013] Optionally, in an embodiment of the present application, the constructing a gestational week prediction model that meets the second preset condition by using the gestational week prediction dataset includes: performing data preprocessing on the gestational week prediction dataset to obtain a processed gestational week prediction dataset; constructing a target multi-layer perceptron neural network model and determining the objective function of the target multi-layer perceptron neural network model; training the target multi-layer perceptron neural network model by using the processed gestational week prediction dataset and the objective function to obtain a trained multi-layer perceptron neural network model; and determining a gestational week prediction model that meets the second preset condition according to the trained multi-layer perceptron neural network model.

[0014] An embodiment of the second aspect of the present application provides a method for evaluating the gestational week of a fetus based on multi-section ultrasound image measurements, which is applied to the online evaluation stage and includes the following steps: obtaining the current fetal ultrasound image of a target patient; inputting the current fetal ultrasound image of the target patient into the fetal gestational week evaluation model to output the gestational week evaluation result of the target patient.

[0015] Optionally, in an embodiment of the present application, the inputting the current fetal ultrasound image of the target patient into the fetal gestational week evaluation model to output the gestational week evaluation result of the target patient includes: inputting the current fetal ultrasound image into the target measurement model in the fetal gestational week evaluation model to output the target segmentation measurement value of the current fetal ultrasound image; and inputting the target segmentation measurement value into the gestational week prediction model in the fetal gestational week evaluation model to output the gestational week evaluation result of the target patient.

[0016] In the third aspect of the present application, an embodiment provides a fetal gestational age assessment device based on multi-plane ultrasound image measurements, which is applied to the model construction stage and includes: a first determination module, configured to determine a semantic segmentation training data set based on the cranial transverse plane, abdominal plane, and femoral plane in the target fetal ultrasound image, and construct a target measurement model that meets the first preset condition by using the semantic segmentation training data set; a second determination module, configured to determine a gestational age prediction data set based on multiple measurement values in the target fetal ultrasound image and the gestational age data corresponding to the multiple measurement values, and construct a gestational age prediction model that meets the second preset condition by using the gestational age prediction data set; a model construction module, configured to splice the target measurement model and the gestational age prediction model to obtain a fetal gestational age assessment model for assessing the fetal gestational age.

[0017] Optionally, in an embodiment of the present application, the first determination module includes: a processing unit, configured to use the cranial transverse plane, the abdominal plane, and the femoral plane in the target fetal ultrasound image as the target key planes for predicting the gestational age; a first determination unit, configured to determine the target key parts corresponding to the target key planes, where the key part of the cranial transverse plane is the skull, the key part of the abdominal plane is the abdomen, and the key part of the femoral plane is the femur; a second determination unit, configured to draw a target area in the corresponding first target plane sample of the target key part to generate a mask image including the target key part, and correspond the first target plane sample with the mask image to determine the semantic segmentation training data set.

[0018] Optionally, in an embodiment of the present application, the first determination module includes: a first construction unit, configured to construct a measurement training data set based on the semantic segmentation training data set, and perform data preprocessing on the measurement training data set to determine the processed measurement training data set; a second construction unit, configured to combine a target point regression deep neural network model with a pre-constructed target semantic segmentation network model to construct a measurement neural network; a third determination unit, configured to combine the measurement neural network with a semantic segmentation objective function to determine a measurement hybrid objective function; a first training unit, configured to train the measurement neural network by using the processed measurement training data set and the measurement hybrid objective function to obtain a trained measurement neural network; a fourth determination unit, configured to determine a target measurement model that meets the first preset condition according to the trained measurement neural network.

[0019] Optionally, in an embodiment of the present application, the first construction unit includes: a processing subunit, configured to use the head circumference and biparietal diameter of the cranial cross-section, the abdominal circumference of the abdominal cross-section, and the femoral length of the femoral cross-section in the target fetal ultrasound image as target key measurement values for predicting the gestational age; a first determination subunit, configured to use the target key measurement values to determine target measurement key points of the target fetal ultrasound image, where the measurement key points of the head circumference and biparietal diameter are the perimeter of the head and the maximum distance between the two parietal bones on both sides of the head, the measurement key point of the abdominal circumference is the perimeter of the abdomen, and the measurement key point of the femoral length is the longest axis length of the femur; a second determination subunit, configured to draw a target line segment and target points in a corresponding second target cross-section sample for the target measurement key points to obtain a plotted image, record the data in the second target cross-section sample and the data in the plotted image, and combine the semantic segmentation training dataset to determine the processed measurement training dataset.

[0020] Optionally, in an embodiment of the present application, the second determination module includes: an acquisition unit, configured to perform data preprocessing on the gestational age prediction dataset to obtain a processed gestational age prediction dataset; a third construction unit, configured to construct a target multi-layer perceptron neural network model and determine the target function of the target multi-layer perceptron neural network model; a second training unit, configured to use the processed gestational age prediction dataset and the target function to train the target multi-layer perceptron neural network model to obtain a trained multi-layer perceptron neural network model; a fifth determination unit, configured to determine a gestational age prediction model that meets the second preset condition according to the trained multi-layer perceptron neural network model.

[0021] An embodiment of the fourth aspect of the present application provides a fetal gestational age evaluation device based on multi-cross-section ultrasound image measurements, which is applied to the online evaluation stage and includes: an acquisition module, configured to acquire the current fetal ultrasound image of a target patient; an evaluation module, configured to input the current fetal ultrasound image of the target patient into the fetal gestational age evaluation model to output the gestational age evaluation result of the target patient.

[0022] Optionally, in an embodiment of the present application, the evaluation module includes: a first evaluation unit, configured to input the current fetal ultrasound image into the target measurement model in the fetal gestational age evaluation model to output the target segmentation measurement value of the current fetal ultrasound image; a second evaluation unit, configured to input the target segmentation measurement value into the gestational age prediction model in the fetal gestational age evaluation model to output the gestational age evaluation result of the target patient.

[0023] An embodiment of the fifth aspect of the present application provides an electronic device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor. The processor executes the program to implement the fetal gestational age evaluation method based on multi-plane ultrasound image measurements as described in the above embodiments.

[0024] An embodiment of the sixth aspect of the present application provides a computer-readable storage medium storing a computer program, which when executed by a processor, implements the fetal gestational age evaluation method based on multi-plane ultrasound image measurements as described above.

[0025] An embodiment of the seventh aspect of the present application provides a computer program product, including a computer program, which when executed, is used to implement the fetal gestational age evaluation method based on multi-plane ultrasound image measurements as described above.

[0026] Embodiments of the present application can determine a semantic segmentation training data set based on the transverse cranial section, abdominal section, and femoral section in the fetal ultrasound image, thereby constructing a target measurement model. Then, based on multiple measurement values in the fetal ultrasound image and the gestational age data corresponding to the multiple measurement values, a gestational age prediction data set is determined, thereby constructing a gestational age prediction model. Secondly, the target measurement model and the gestational age prediction model are spliced to obtain a fetal gestational age evaluation model for evaluating the fetal gestational age, so that the biological characteristics related to gestational age evaluation in the ultrasound image can be accurately extracted through a unified inspection standard, and then the precise gestational age evaluation value can be inferred using the biological characteristics. Thus, the problem in the related art that mainly relies on manual inspection and analysis by doctors, which requires a large amount of labor costs and the inspection results of different doctors are different, reducing the accuracy of fetal gestational age evaluation, is solved.

[0027] Additional aspects and advantages of the present application will be given in part in the following description, become apparent in part from the following description, or be understood through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of the embodiments in conjunction with the drawings, where:

[0029] Figure 1 is a flowchart of a fetal gestational age evaluation method based on multi-plane ultrasound image measurements according to an embodiment of the present application;

[0030] Figure 2 is a flowchart of another fetal gestational age evaluation method based on multi-plane ultrasound image measurements according to an embodiment of the present application;

[0031] Figure 3 is a schematic diagram of the principle of a measurement model in a specific embodiment of the present application;

[0032] Figure 4 Schematic diagram of the principle of the gestational age prediction model for a specific embodiment of the present application;

[0033] Figure 5 Schematic diagram of the structure of a fetal gestational age assessment device based on multi-plane ultrasound image measurements according to an embodiment of the present application;

[0034] Figure 6 Schematic diagram of the structure of another fetal gestational age assessment device based on multi-plane ultrasound image measurements according to an embodiment of the present application;

[0035] Figure 7 Schematic diagram of the structure of an electronic device according to an embodiment of the present application. Detailed implementation manners

[0036] The embodiments of the present application will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present application and should not be construed as a limitation to the present application.

[0037] The fetal gestational age assessment method and device based on multi-plane ultrasound image measurements according to the embodiments of the present application will be described below with reference to the accompanying drawings. In view of the problem in the related art mentioned in the above background art that mainly relies on the manual examination and analysis by doctors, which requires a large amount of labor cost, and the examination results of different doctors are different, reducing the accuracy of fetal gestational age assessment, the present application provides a fetal gestational age assessment method based on multi-plane ultrasound image measurements. In this method, the target measurement model constructed based on the target fetal ultrasound image and the constructed gestational age prediction model can be spliced to obtain a fetal gestational age assessment model for assessing the fetal gestational age, so that the biological characteristics related to gestational age assessment in the ultrasound image can be accurately extracted through a unified examination standard, and then the accurate gestational age assessment value can be inferred using the biological characteristics. Thus, the problem in the related art that mainly relies on the manual examination and analysis by doctors, which requires a large amount of labor cost, and the examination results of different doctors are different, reducing the accuracy of fetal gestational age assessment, is solved.

[0038] Specifically, Figure 1 Schematic diagram of the process of a fetal gestational age assessment method based on multi-plane ultrasound image measurements provided by an embodiment of the present application.

[0039] As Figure 1 shown, this fetal gestational age assessment method based on multi-plane ultrasound image measurements, applied to the model construction stage, includes the following steps:

[0040] In step S101, based on the transverse cranial section, abdominal section, and femoral section in the target fetal ultrasound image, a semantic segmentation training dataset is determined, and the target measurement model that meets the first preset condition is constructed using the semantic segmentation training dataset.

[0041] In the embodiment of the present application, the transverse cranial section, abdominal section, and femoral section in the target fetal ultrasound image are a large number of pre-collected ultrasound section pictures related to the fetal gestational age, and the data is labeled under the guidance of professional doctors, providing a sufficient and effective dataset for the training of the deep learning model.

[0042] It can be understood that in the embodiment of the present application, based on the transverse cranial section, abdominal section, and femoral section in the target fetal ultrasound image. For example, first, according to professional books and expert opinions, the transverse cranial section, abdominal section, and femoral section can be selected as the key sections for predicting the gestational age, and the key parts of the three sections are learned to determine the semantic segmentation training dataset in the following steps, and the target measurement model that meets the first condition in the following steps is constructed using the semantic segmentation training dataset. That is, the trained measurement neural network is used as an end-to-end measurement model, effectively improving the accuracy of biometric feature extraction in fetal ultrasound images.

[0043] Among them, in one embodiment of the present application, determining the semantic segmentation training dataset based on the transverse cranial section, abdominal section, and femoral section in the target fetal ultrasound image includes: using the transverse cranial section, abdominal section, and femoral section in the target fetal ultrasound image as the target key sections for predicting the gestational age; determining the target key parts corresponding to the target key sections. Among them, the key part of the transverse cranial section is the skull, the key part of the abdominal section is the abdomen, and the key part of the femoral section is the femur; drawing the target area in the corresponding first target section sample of the target key part to generate a mask image containing the target key part, and corresponding the first target section sample with the mask image to determine the semantic segmentation training dataset.

[0044] In the actual execution process, in the embodiment of the present application, according to professional books and expert opinions, the transverse cranial section, abdominal section, and femoral section in the fetal ultrasound image can be selected as the key sections for predicting the gestational age, and the key parts of the three sections are learned. It is obtained that the required parts of the sections are the skull, abdomen, and femur respectively. Then, the target area is drawn in the corresponding section sample according to the target part to generate the segmentation mask image of the corresponding part, and the original image of the section sample is corresponded with the target mask image one by one to construct the semantic segmentation training dataset, effectively improving the accuracy and reliability of the semantic segmentation training dataset, and helping to improve the segmentation performance of the model.

[0045] Next, the semantic segmentation training dataset can be preprocessed. First, the mask image is placed on the corresponding channels (each channel corresponds to a semantic category, and the values of the channels other than this category are all zero) to generate a multi-channel mask image. Then, the sizes of the ultrasound image and the mask image are uniformly scaled to 224×224 to obtain the preprocessed semantic segmentation training set.

[0046] Secondly, a semantic segmentation network can be constructed based on the preprocessed semantic segmentation training dataset. Specifically, the semantic segmentation network structure continues the classic U-Net framework, including an encoder (downsampling), a decoder (upsampling), and skip connections. The difference is that at the beginning, the position embedding layer uses multi-layer convolution for sequence encoding to expand the receptive field of the model, and each layer uses a neural network that mixes convolution and Transformer to better fuse local and global information, and a threshold function is used for denoising at the bottleneck, that is, if the positive value is lower than the set threshold, it is set to zero; the objective function adopts a combination of Dice loss and BCE (Binary Cross Entropy) loss.

[0047] Thirdly, a part of the data in the semantic segmentation training dataset can be selected to form a pre-training set. Based on the semantic segmentation network structure and the objective function in the above steps, the semantic segmentation network is pre-trained. The training process aims to minimize the value of the objective function, and a trained semantic segmentation network model is obtained according to the final weight parameters of the encoder and the decoder, so as to provide prior segmentation knowledge for the subsequent combined training with the point regression model.

[0048] In the embodiment of the present application, the adopted semantic segmentation model has the characteristics of high precision, high efficiency, and light weight, effectively solving the difficulties of the traditional medical segmentation model such as limited receptive field, insufficient fusion of local and global information, and being greatly affected by ultrasound image noise, and providing reliable results for subsequent measurements.

[0049] Optionally, in an embodiment of the present application, a target measurement model that meets the first preset condition is constructed using the semantic segmentation training dataset, including: constructing a measurement training dataset based on the semantic segmentation training dataset, and performing data preprocessing on the measurement training dataset to determine the processed measurement training dataset; combining the target point regression deep neural network model with the pre-constructed target semantic segmentation network model to construct a measurement neural network; combining the measurement neural network with the semantic segmentation objective function to determine the mixed objective function of the measurement; training the measurement neural network using the processed measurement training dataset and the mixed objective function of the measurement to obtain the trained measurement neural network; and determining the target measurement model that meets the first preset condition according to the trained measurement neural network.

[0050] As a possible implementation manner, the embodiments of the present application can construct a measurement training dataset based on a semantic segmentation training dataset, perform data preprocessing on the measurement training dataset, select a deep neural network model for point regression, combine it with a semantic segmentation network to construct a measurement neural network, and combine a segmentation objective function to construct a hybrid objective function for measurement. Thus, based on the processed measurement training dataset, measurement neural network, and objective function, the training process of the measurement neural network can be determined to complete the training, and the trained measurement neural network is used as an end-to-end measurement model.

[0051] Specifically, first, point regression prediction is performed on the line segment endpoints of the femur length and biparietal diameter. The structure of the point regression model is based on ResNet50. The core structure of ResNet50 is based on residual learning, and deep feature extraction is achieved through the stacking of residual blocks. Each residual block contains a skip connection, allowing gradients to flow directly through the network, effectively alleviating the problem of gradient disappearance in deep networks. A bottleneck structure is used to reduce the computational amount while retaining sufficient expressive power. This network extracts multi-scale features through gradual downsampling, and finally reduces the feature map to a fixed size through a global average pooling layer and performs final target point prediction through a fully connected layer. The objective function is the MSE (Mean Squared Error) loss; for the head circumference and abdominal circumference, a post-processing method is used. According to the segmentation masks corresponding to the head circumference and abdominal circumference, the corresponding ellipses are fitted, and the ellipse perimeter is used as the predicted perimeter length. Finally, the post-processing and point regression model are spliced behind the pre-trained semantic segmentation model to construct a measurement model. Among them, the objective function is a hybrid loss function composed of Dice loss, BCE loss, and MSE loss, aiming to optimize the segmentation and measurement tasks.

[0052] Next, according to the processed measurement training dataset, measurement neural network, and objective function, the measurement neural network can be trained. The training process aims to minimize the value of the objective function, and a trained network model is obtained according to the final weight parameters of each layer of the two networks. The trained network model is used as a measurement model.

[0053] In the embodiments of the present application, the adopted point regression model balances the influence of differences in doctors' examination habits, improves the measurement accuracy, and is easy to expand to the measurement tasks of other medical ultrasound images.

[0054] Optionally, in an embodiment of the present application, data preprocessing is performed on the measurement training data set to determine the processed measurement training data set, including: using the head circumference and biparietal diameter of the transverse cranial section, the abdominal circumference of the abdominal section, and the femur length of the femoral section in the target fetal ultrasound image as the target key measurement values for predicting the gestational age; using the target key measurement values to determine the target measurement key points of the target fetal ultrasound image, where the measurement key points of the head circumference and biparietal diameter are the perimeter of the head and the maximum distance between the two parietal bones on both sides of the head, the measurement key point of the abdominal circumference is the perimeter of the abdomen, and the measurement key point of the femur length is the longest axis length of the femur; drawing target line segments and target points in the corresponding second target section samples for the target measurement key points to obtain a plotted image, recording the data in the second target section samples and the data in the plotted image, and combining with the semantic segmentation training data set to determine the processed measurement training data set.

[0055] In some embodiments, the embodiments of the present application may select the head circumference and biparietal diameter of the transverse cranial section, the abdominal circumference of the abdominal section, and the femur length of the femoral section in the fetal ultrasound image as the key measurement values for predicting the gestational age according to professional books and expert opinions, and learn the measurement methods of these four measurement values to obtain the measurement key points as follows: the perimeter of the head and the maximum distance between the two parietal bones on both sides of the head, the perimeter of the abdomen, and the longest axis length of the femur; then draw target line segments and target points in the section samples according to the measurement key points, and correspond the original image of the section sample with the data records containing the target line segments and target points one by one. Finally, combine with the semantic segmentation training data set to determine the processed measurement training data set, thereby improving the data quality and ensuring the accuracy and reliability of subsequent analysis.

[0056] In step S102, based on multiple measurement values in the target fetal ultrasound image and the gestational age data corresponding to the multiple measurement values, a gestational age prediction data set is determined, and a gestational age prediction model that meets the second preset condition is constructed using the gestational age prediction data set.

[0057] It can be understood that the embodiments of the present application can be based on multiple measurement values in the fetal ultrasound image and the gestational age data corresponding to the multiple measurement values. For example, based on the processed measurement training data set, the gestational age data of the patients included therein can be collected, and the data records of the head circumference and biparietal diameter of the transverse cranial section, the abdominal circumference of the abdominal section, and the femur length of the femoral section of the same patient are corresponded with the gestational age data records one by one to construct a training set for gestational age prediction, that is, a gestational age prediction data set, and the gestational age prediction data set is preprocessed to construct a multi-layer perceptron neural network structure and an objective function for gestational age prediction based on the processed gestational age prediction data set, and the multi-layer perceptron neural network is trained, so that a gestational age prediction model can be constructed, effectively improving the accuracy and reliability of gestational age prediction.

[0058] Optionally, in an embodiment of the present application, constructing a gestational age prediction model that meets the second preset condition using a gestational age prediction data set includes: performing data preprocessing on the gestational age prediction data set to obtain a processed gestational age prediction data set; constructing a target multi-layer perceptron neural network model and determining the objective function of the target multi-layer perceptron neural network model; training the target multi-layer perceptron neural network model using the processed gestational age prediction data set and the objective function to obtain a trained multi-layer perceptron neural network model; and determining a gestational age prediction model that meets the second preset condition according to the trained multi-layer perceptron neural network model.

[0059] In some embodiments, the embodiments of the present application can collect the gestational age data of the patients included therein based on the processed measurement training data set, and correspond the data records of the head circumference and biparietal diameter of the transverse cranial section, the abdominal circumference of the abdominal section, and the femur length of the femoral section of the same patient with their gestational age data records one by one to construct a training set for gestational age prediction, that is, a gestational age prediction data set, and perform data preprocessing on the gestational age prediction data set to obtain a processed gestational age prediction data set.

[0060] Next, a multi-layer perceptron network model for gestational age prediction can be constructed. The overall network structure is a three-layer multi-layer perceptron, and the number of nodes in the hidden layers are 256, 512, and 256 respectively, which is used to learn the regression relationship between the measured values and the gestational age, and a dropout layer is adopted in each layer, that is, each time during training, effective nodes will be selected for calculation according to the set dropout rate of each layer to avoid overfitting during training; the objective function is the MSE loss.

[0061] Finally, based on the processed gestational age prediction data set, the multi-layer perceptron neural network structure, and the objective function, train the multi-layer perceptron neural network. The training process aims to minimize the value of the objective function, and obtain a trained multi-layer perceptron neural network model according to the final weight parameters of the nodes in each hidden layer, and use the trained multi-layer perceptron neural network model as the gestational age prediction model.

[0062] In step S103, splicing the target measurement model and the gestational age prediction model to obtain a fetal gestational age evaluation model for evaluating the gestational age of a fetus.

[0063] It can be understood that the embodiments of the present application can splice the target measurement model and the gestational age prediction model, that is, splice the gestational age prediction model behind the target measurement model, so as to obtain a fetal gestational age evaluation model for evaluating the gestational age of a fetus based on the ultrasonic image measurement values, and further can realize automatic and rapid segmentation and measurement of ultrasonic images and evaluation of gestational age, effectively solving the problem of accurate and rapid determination of the gestational age of a fetus in clinical diagnosis.

[0064] The fetal gestational age assessment method based on multi-plane ultrasound image measurements proposed in the embodiments of the present application can splice the target measurement model constructed based on the ultrasound image of the target fetus and the constructed gestational age prediction model to obtain a fetal gestational age assessment model for evaluating the fetal gestational age. Thus, through a unified inspection standard, biological characteristics related to gestational age assessment in ultrasound images can be accurately extracted, and then the precise gestational age assessment value can be inferred using the biological characteristics. Thereby, it solves the problems in the related art that mainly rely on manual inspection and analysis by doctors, which requires a large amount of labor costs, and the inspection results of different doctors are different, reducing the accuracy of fetal gestational age assessment.

[0065] And, as Figure 2 shown, Figure 2 is a schematic flowchart of another fetal gestational age assessment method based on multi-plane ultrasound image measurements provided by the embodiments of the present application.

[0066] As Figure 2 shown, this fetal gestational age assessment method based on multi-plane ultrasound image measurements is applied to the online assessment stage and includes the following steps:

[0067] In step S201, the current fetal ultrasound image of the target patient is obtained.

[0068] In the embodiments of the present application, the target patient is the patient currently undergoing fetal gestational age assessment.

[0069] It can be understood that the embodiments of the present application can obtain the current fetal ultrasound image of the patient. For example, clear fetal ultrasound cross-sectional images are collected from a medical imaging device, and data related to the images (such as acquisition time, device parameters, and basic patient information, etc.) are recorded to provide comprehensive and accurate data support for subsequent analysis.

[0070] In step S202, the current fetal ultrasound image of the target patient is input into the fetal gestational age assessment model to output the gestational age assessment result of the target patient.

[0071] It can be understood that the embodiments of the present application can input the current fetal ultrasound image of the patient collected from a medical imaging device into the fetal gestational age assessment model. First, the gestational age assessment model can automatically analyze and output the gestational age assessment result of the target patient based on the key anatomical features and measurement data in the current fetal ultrasound image, thereby realizing automated gestational age inference based on measurements and meeting the accurate and rapid requirements of clinical fetal gestational age assessment.

[0072] Optionally, in an embodiment of the present application, the current fetal ultrasound image of the target patient is input into the fetal gestational age assessment model to output the gestational age assessment result of the target patient, including: inputting the current fetal ultrasound image into the target measurement model in the fetal gestational age assessment model to output the target segmentation measurement value of the current fetal ultrasound image; inputting the target segmentation measurement value into the gestational age prediction model in the fetal gestational age assessment model to output the gestational age assessment result of the target patient.

[0073] As a possible implementation manner, as Figure 3 shown, first, the embodiment of the present application inputs the current fetal ultrasound image into the target measurement model in the fetal gestational age assessment model to determine the biparietal diameter, head circumference, abdominal circumference, and femur length of the current fetus according to the transverse cranial section, abdominal section, and femoral section of the current fetus. Then, as Figure 4 shown, the biparietal diameter, head circumference, abdominal circumference, and femur length of the current fetus are input into the gestational age prediction model in the fetal gestational age assessment model to determine the actual gestational age of the current fetus and output the gestational age assessment result of the patient, which can quickly and accurately evaluate the gestational age, reduce manual measurement errors, improve the diagnostic efficiency, and at the same time provide objective decision support for clinicians, helping to achieve more precise pregnancy management.

[0074] According to the fetal gestational age assessment method based on multi-section ultrasound image measurements proposed in the embodiment of the present application, the current fetal ultrasound image of the target patient collected can be input into the fetal gestational age assessment model to output the gestational age assessment result of the target patient, which can quickly and accurately evaluate the gestational age, reduce manual measurement errors, and improve the diagnostic efficiency. Thus, it solves the problem in the related art that mainly relies on the manual examination and analysis of doctors, which requires a large amount of labor costs, and the examination results of different doctors are different, reducing the accuracy of fetal gestational age assessment.

[0075] Secondly, as Figure 5 shown, it is a block diagram of a fetal gestational age assessment device based on multi-section ultrasound image measurements according to an embodiment of the present application.

[0076] As Figure 5 shown, the fetal gestational age assessment device 10 based on multi-section ultrasound image measurements applied to the model construction stage includes: a first determination module 100, a second determination module 200, and a model construction module 300.

[0077] Specifically, the first determination module 100 is used to determine the semantic segmentation training data set based on the transverse cranial section, abdominal section, and femoral section in the target fetal ultrasound image, and construct a target measurement model that meets the first preset condition by using the semantic segmentation training data set.

[0078] The second determination module 200 is configured to determine a gestational age prediction data set based on multiple measurement values and gestational age data corresponding to the multiple measurement values in the target fetal ultrasound image, and construct a gestational age prediction model that meets the second preset condition by using the gestational age prediction data set.

[0079] The model construction module 300 is configured to splice the target measurement model and the gestational age prediction model to obtain a fetal gestational age evaluation model for evaluating the fetal gestational age.

[0080] Optionally, in an embodiment of the present application, the first determination module 100 includes: a processing unit, a first determination unit, and a second determination unit.

[0081] Among them, the processing unit is configured to use the transverse cranial section, abdominal section, and femoral section in the target fetal ultrasound image as the target key sections for predicting the gestational age.

[0082] The first determination unit is configured to determine the target key parts corresponding to the target key sections. Among them, the key part of the transverse cranial section is the skull, the key part of the abdominal section is the abdomen, and the key part of the femoral section is the femur.

[0083] The second determination unit is configured to draw a target area in the corresponding first target section sample of the target key part to generate a mask image including the target key part, and correspond the first target section sample with the mask image to determine a semantic segmentation training data set.

[0084] Optionally, in an embodiment of the present application, the first determination module 100 includes: a first construction unit, a second construction unit, a third determination unit, a first training unit, and a fourth determination unit.

[0085] Among them, the first construction unit is configured to construct a measurement training data set based on the semantic segmentation training data set, and perform data preprocessing on the measurement training data set to determine the processed measurement training data set.

[0086] The second construction unit is configured to combine the target point regression deep neural network model with the pre-constructed target semantic segmentation network model to construct a measurement neural network.

[0087] The third determination unit is configured to combine the measurement neural network with the semantic segmentation objective function to determine the measurement hybrid objective function.

[0088] The first training unit is configured to train the measurement neural network by using the processed measurement training data set and the measurement hybrid objective function to obtain the trained measurement neural network.

[0089] The fourth determination unit is configured to determine a target measurement model that meets the first preset condition according to the trained measurement neural network.

[0090] Optionally, in an embodiment of the present application, the first construction unit includes: a processing subunit, a first determination subunit, and a second determination subunit.

[0091] Among them, the processing subunit is configured to use the head circumference and biparietal diameter of the cranial cross-section, the abdominal circumference of the abdominal section, and the femur length of the femoral section in the target fetal ultrasound image as the target key measurement values for predicting the gestational week.

[0092] The first determination subunit is configured to use the target key measurement values to determine the target measurement key points of the target fetal ultrasound image. Among them, the measurement key points of the head circumference and biparietal diameter are the perimeter of the head and the maximum distance between the two parietal bones on both sides of the head, the measurement key point of the abdominal circumference is the perimeter of the abdomen, and the measurement key point of the femur length is the longest axis length of the femur.

[0093] The second determination subunit is configured to draw a target line segment and target points in the corresponding second target section sample for the target measurement key points to obtain a plotted image, record the data in the second target section sample and the data in the plotted image, and combine the semantic segmentation training data set to determine the processed measurement training data set.

[0094] Optionally, in an embodiment of the present application, the second determination module 200 includes: an acquisition unit, a third construction unit, a second training unit, and a fifth determination unit.

[0095] Among them, the acquisition unit is configured to perform data preprocessing on the gestational week prediction data set to obtain a processed gestational week prediction data set.

[0096] The third construction unit is configured to construct a target multi-layer perceptron neural network model and determine the target function of the target multi-layer perceptron neural network model.

[0097] The second training unit is configured to use the processed gestational week prediction data set and the target function to train the target multi-layer perceptron neural network model to obtain a trained multi-layer perceptron neural network model.

[0098] The fifth determination unit is configured to determine a gestational week prediction model that meets the second preset condition according to the trained multi-layer perceptron neural network model.

[0099] It should be noted that the foregoing explanation of the embodiment of the fetal gestational week evaluation method based on multi-section ultrasound image measurements also applies to the fetal gestational week evaluation device based on multi-section ultrasound image measurements in this embodiment, and will not be elaborated here.

[0100] According to the fetal gestational age assessment device based on multi-plane ultrasound image measurements proposed in the embodiments of the present application, the target measurement model constructed based on the target fetal ultrasound image and the constructed gestational age prediction model can be spliced to obtain a fetal gestational age assessment model for evaluating the fetal gestational age. Thus, through a unified inspection standard, the biological characteristics related to gestational age assessment in the ultrasound image can be accurately extracted, and then the precise gestational age assessment value can be inferred using the biological characteristics. Thereby, the problem in the related art that mainly relies on the manual inspection and analysis of doctors, which requires a large amount of labor costs, and the inspection results of different doctors are different, reducing the accuracy of fetal gestational age assessment, is solved.

[0101] And, as Figure 6 shown, it is a block diagram of another fetal gestational age assessment device based on multi-plane ultrasound image measurements according to the embodiments of the present application.

[0102] As Figure 6 shown, the fetal gestational age assessment device 20 based on multi-plane ultrasound image measurements applied to the online assessment stage includes: an acquisition module 400 and an assessment module 500.

[0103] Specifically, the acquisition module 400 is used to acquire the current fetal ultrasound image of the target patient.

[0104] The assessment module 500 is used to input the current fetal ultrasound image of the target patient into the fetal gestational age assessment model to output the gestational age assessment result of the target patient.

[0105] Optionally, in an embodiment of the present application, the assessment module 500 includes: a first assessment unit and a second assessment unit.

[0106] Among them, the first assessment unit is used to input the current fetal ultrasound image into the target measurement model in the fetal gestational age assessment model to output the target segmentation measurement value of the current fetal ultrasound image.

[0107] The second assessment unit is used to input the target segmentation measurement value into the gestational age prediction model in the fetal gestational age assessment model to output the gestational age assessment result of the target patient.

[0108] It should be noted that the foregoing explanation of the embodiments of the fetal gestational age assessment method based on multi-plane ultrasound image measurements also applies to the fetal gestational age assessment device based on multi-plane ultrasound image measurements in this embodiment, and will not be elaborated here.

[0109] The fetal gestational age assessment device based on multi-plane ultrasound image measurement proposed according to the embodiments of the present application can input the current fetal ultrasound image of the target patient collected into the fetal gestational age assessment model to output the gestational age assessment result of the target patient, which can quickly and accurately assess the gestational age, reduce the artificial measurement error, and improve the diagnostic efficiency. Thus, it solves the problems in the related art that mainly rely on the manual examination and analysis of doctors, which requires a large amount of labor costs, and the examination results of different doctors are different, reducing the accuracy of fetal gestational age assessment.

[0110] Figure 7 It is a schematic structural diagram of an electronic device provided by an embodiment of the present application. The electronic device may include:

[0111] A memory 701, a processor 702, and a computer program stored on the memory 701 and executable on the processor 702.

[0112] When the processor 702 executes the program, it implements the fetal gestational age assessment method based on multi-plane ultrasound image measurement provided in the above embodiment.

[0113] Further, the electronic device further includes:

[0114] A communication interface 703 for communication between the memory 701 and the processor 702.

[0115] The memory 701 is used to store a computer program executable on the processor 702.

[0116] The memory 701 may include a high-speed RAM memory, and may also include a non-volatile memory, such as at least one disk memory.

[0117] If the memory 701, the processor 702, and the communication interface 703 are independently implemented, the communication interface 703, the memory 701, and the processor 702 can be connected to each other through a bus and complete the communication between each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of representation, Figure 7 only a thick line is shown in the figure, but it does not mean that there is only one bus or one type of bus.

[0118] Optionally, in a specific implementation, if the memory 701, the processor 702, and the communication interface 703 are integrated on a single chip, the memory 701, the processor 702, and the communication interface 703 can communicate with each other through an internal interface.

[0119] The processor 702 may be a central processing unit (CPU for short), or an application specific integrated circuit (ASIC for short), or one or more integrated circuits configured to implement the embodiments of the present application.

[0120] This embodiment also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the above-mentioned fetal gestational age assessment method based on multi-plane ultrasound image measurement values is implemented.

[0121] This embodiment also provides a computer program product, including a computer program, which is used to implement the above-mentioned fetal gestational age assessment method based on multi-plane ultrasound image measurement values when the computer program is executed.

[0122] In the description of this specification, the descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0123] In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "N" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0124] Any process or method description represented in a flowchart or otherwise described herein can be understood to represent a module, segment, or portion of code including one or N executable instructions for implementing a customized logical function or process, and the scope of the preferred embodiments of the present application includes additional implementations, where the functions may be executed in a substantially simultaneous manner or in a reverse order according to the functions involved, rather than in the order shown or discussed, which should be understood by those skilled in the art to which the embodiments of the present application pertain.

[0125] The logic and / or steps represented in a flowchart or otherwise described herein, for example, can be considered a sequenced list of executable instructions for implementing a logical function, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in conjunction with such instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection (electronic device) having one or N wirings, a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, as the program can be obtained electronically by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory.

[0126] It should be understood that various parts of the present application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), and the like.

[0127] Those of ordinary skill in the art can understand that all or part of the steps carried out in implementing the above-described embodiment methods can be completed by a program instructing related hardware. The program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0128] In addition, each functional unit in various embodiments of the present application can be integrated in a processing module, or each unit can exist physically alone, or two or more units can be integrated in a module. The above-mentioned integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0129] The above-mentioned storage medium can be a read-only memory, a magnetic disk, an optical disc, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A method for evaluating the gestational age of a fetus based on measurements from multi-plane ultrasound images, characterized in that, Applied to the model construction stage, wherein the method includes the following steps: Based on the transverse cranial section, abdominal section, and femoral section in the target fetal ultrasound image, determine the semantic segmentation training dataset, and use the semantic segmentation training dataset to construct a target measurement model that meets the first preset condition; Based on multiple measurement values in the target fetal ultrasound image and the gestational age data corresponding to the multiple measurement values, determine the gestational age prediction dataset, and use the gestational age prediction dataset to construct a gestational age prediction model that meets the second preset condition; Perform splicing processing on the target measurement model and the gestational age prediction model to obtain a fetal gestational age evaluation model for evaluating the fetal gestational age.

2. The method according to claim 1, wherein The determining the semantic segmentation training dataset based on the transverse cranial section, abdominal section, and femoral section in the target fetal ultrasound image includes: Use the transverse cranial section, abdominal section, and femoral section in the target fetal ultrasound image as the target key sections for predicting the gestational age; Determine the target key parts corresponding to the target key sections, wherein the key part of the transverse cranial section is the skull, the key part of the abdominal section is the abdomen, and the key part of the femoral section is the femur; Draw a target area on the corresponding first target section sample for the target key part to generate a mask image containing the target key part, and correspond the first target section sample with the mask image to determine the semantic segmentation training dataset.

3. The method according to claim 1, characterized in that, The using the semantic segmentation training dataset to construct a target measurement model that meets the first preset condition includes: Based on the semantic segmentation training dataset, construct a measurement training dataset, and perform data preprocessing on the measurement training dataset to determine the processed measurement training dataset; Combine the target point regression deep neural network model with the pre-constructed target semantic segmentation network model to construct a measurement neural network; Combine the measurement neural network with the semantic segmentation objective function to determine the measurement hybrid objective function; Use the processed measurement training dataset and the measurement hybrid objective function to train the measurement neural network to obtain the trained measurement neural network; Determine the target measurement model that meets the first preset condition according to the trained measurement neural network.

4. The method according to claim 3, wherein The performing data preprocessing on the measurement training dataset to determine the processed measurement training dataset includes: Use the head circumference and biparietal diameter of the transverse cranial section, the abdominal circumference of the abdominal section, and the femoral length of the femoral section in the target fetal ultrasound image as the target key measurement values for predicting the gestational age; Use the target key measurement values to determine the target measurement key points of the target fetal ultrasound image, wherein the measurement key points of the head circumference and biparietal diameter are the perimeter of the head and the maximum distance between the two parietal bones on both sides of the head, the measurement key point of the abdominal circumference is the perimeter of the abdomen, and the measurement key point of the femoral length is the longest axis length of the femur; Draw a target line segment and target points for the target measurement key points in the corresponding second target section sample to obtain a plotted image, record the data in the second target section sample and the data in the plotted image, and combine the semantic segmentation training data set to determine the processed measurement training data set.

5. The method according to claim 1, wherein The constructing a gestational age prediction model that meets the second preset condition by using the gestational age prediction data set includes: Perform data preprocessing on the gestational age prediction data set to obtain a processed gestational age prediction data set; Construct a target multi-layer perceptron neural network model and determine the objective function of the target multi-layer perceptron neural network model; Use the processed gestational age prediction data set and the objective function to train the target multi-layer perceptron neural network model to obtain a trained multi-layer perceptron neural network model; Determine a gestational age prediction model that meets the second preset condition according to the trained multi-layer perceptron neural network model.

6. A method for evaluating the gestational age of a fetus based on measurements of multi-plane ultrasound images, characterized in that, Applied to the online evaluation stage, wherein, the method includes the following steps: Obtain the current fetal ultrasound image of the target patient; Input the current fetal ultrasound image of the target patient into the fetal gestational age evaluation model to output the gestational age evaluation result of the target patient.

7. The method according to claim 6, characterized in that The inputting the current fetal ultrasound image of the target patient into the fetal gestational age evaluation model to output the gestational age evaluation result of the target patient includes: Input the current fetal ultrasound image into the target measurement model in the fetal gestational age evaluation model to output the target segmentation measurement value of the current fetal ultrasound image; Input the target segmentation measurement value into the gestational age prediction model in the fetal gestational age evaluation model to output the gestational age evaluation result of the target patient.

8. A fetal gestational age assessment device based on multi-plane ultrasound image measurements, characterized in that, Applied to the model construction stage, wherein, the device includes: A first determination module, configured to determine a semantic segmentation training data set based on the transverse cranial section, abdominal section, and femoral section in the target fetal ultrasound image, and use the semantic segmentation training data set to construct a target measurement model that meets the first preset condition; A second determination module, configured to determine a gestational age prediction data set based on multiple measurement values in the target fetal ultrasound image and the gestational age data corresponding to the multiple measurement values, and use the gestational age prediction data set to construct a gestational age prediction model that meets the second preset condition; A model construction module, configured to splice the target measurement model and the gestational age prediction model to obtain a fetal gestational age evaluation model for evaluating the gestational age of a fetus.

9. A fetal gestational age assessment device based on multi-plane ultrasound image measurements, characterized in that, Applied to the online evaluation stage, wherein, the device includes: An acquisition module, configured to acquire the current fetal ultrasound image of the target patient; An evaluation module, configured to input the current fetal ultrasound image of the target patient into the fetal gestational age evaluation model to output the gestational age evaluation result of the target patient.

10. An electronic device, characterized in that, Includes: A memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor executes the program to implement the method for evaluating the gestational age of a fetus based on multi-section ultrasound image measurements as described in any one of claims 1-5 or 6-7.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by a processor to implement the fetal gestational age assessment method based on multi-plane ultrasonic image measurements as described in any one of claims 1-5 or 6-7.

12. A computer program product, comprising a computer program, characterized in that, The computer program is executed by a processor to implement the fetal gestational age assessment method based on multi-plane ultrasonic image measurements as described in any one of claims 1-5 or 6-7.

Images