Fetal heart ultrasound image processing method, device and computer equipment

Abstract

This application relates to a method, apparatus, computer device, storage medium, and computer program product for processing fetal cardiac ultrasound images. By performing target segmentation processing on the acquired target four-chamber ultrasound cross-sectional image, atrial septum contour data, ventricular septum contour data, spinal contour data, and thoracic cavity contour data are obtained. These data are used for fitting the first and second cardiac axes, thereby achieving automatic measurement of the cardiac axis, i.e., the angle between the direction of the fetal heart apex and the line connecting the spine and sternum. Compared with manual tracing measurement, this method is less affected by clinical experience and other subjective factors, greatly improving measurement accuracy. It is suitable for prenatal ultrasound examinations and helps doctors quickly classify and judge fetal four-chamber ultrasound cross-sectional images, improving the efficiency of doctors' examinations.

Description

Technical Field

[0001] This application relates to the field of medical image processing technology, and in particular to a method, apparatus, computer device, storage medium, and computer program product for processing fetal cardiac ultrasound images. Background Technology

[0002] Congenital heart disease is the most common birth defect worldwide, characterized by a high incidence, high disability rate, and high mortality rate. With improved living standards and advancements in medical technology, more and more families are using fetal echocardiography for prenatal screening and diagnosis of congenital heart disease to determine early on whether their fetus has a heart malformation. Current research on fetuses in the mid-to-late stages of pregnancy suggests that excessive rotation of the glomerulus and ventricular loops in early embryonic genetics may lead to abnormalities in the cardiac axis, which is correlated with congenital heart disease in the fetus.

[0003] In clinical medicine, ultrasound is typically used to obtain a cross-sectional image of the fetal chest cavity at the level of the four chambers of the heart. The fetal cardiac axis is then measured manually to aid in the screening and diagnosis of congenital heart disease. However, this method of measuring the cardiac axis is easily affected by the doctor's clinical experience and other subjective factors, leading to significant errors in the manually measured data and resulting in missed or misdiagnosed cases. Summary of the Invention

[0004] Therefore, it is necessary to address the technical problem of large measurement errors caused by manual measurement of the cardiac axis mentioned above, and to provide a fetal cardiac ultrasound image processing method, device, computer equipment, storage medium, and computer program product that can more accurately measure the cardiac axis in the four-chamber view of fetal ultrasound.

[0005] In a first aspect, this application provides a method for processing fetal echocardiogram images, the method comprising:

[0006] Acquire a target four-chamber echocardiogram image, and perform target segmentation processing on the target four-chamber echocardiogram image based on a preset target segmentation model to obtain atrial septum contour data, interventricular septum contour data, spinal contour data, and thoracic cavity contour data.

[0007] A first cardiac axis is obtained by fitting the atrial septal contour data and the ventricular septal contour data to the axis.

[0008] A second cardiac axis is obtained by fitting the spinal contour data and the thoracic contour data together.

[0009] The cardiac axis is obtained based on the angle between the first cardiac axis and the second cardiac axis.

[0010] In one embodiment, the step of performing axis fitting based on the atrial septum contour data and the interventricular septum contour data to obtain a first cardiac axis includes:

[0011] Skeleton extraction is performed on the atrial septum contour data and the ventricular septum contour data respectively to obtain atrial septum skeleton data and ventricular septum skeleton data;

[0012] The first cardiac axis is obtained by performing linear fitting based on the atrial septum skeleton data and the ventricular septum skeleton data.

[0013] In one embodiment, the step of performing axis fitting based on the spinal contour data and the thoracic contour data to obtain the second cardiac axis includes:

[0014] Skeleton extraction is performed on the spinal contour data and the thoracic contour data respectively to obtain spinal skeleton data and thoracic skeleton data;

[0015] The average coordinates of the spinal skeleton are calculated based on the spinal skeleton data to obtain the average coordinate points of the spinal skeleton.

[0016] The average coordinates of the thoracic skeleton are calculated based on the thoracic skeleton data to obtain the average coordinate points of the thoracic skeleton.

[0017] Centroids are extracted from the spinal contour data and the thoracic contour data to obtain the centroids of the spine and the thoracic cavity.

[0018] The second heart axis is obtained by performing a straight line fit based on the average coordinate points of the spinal skeleton, the average coordinate points of the thoracic skeleton, the centroid of the spine, and the centroid of the thoracic cavity.

[0019] In one embodiment, before performing skeleton extraction on the atrial septum contour data and the ventricular septum contour data, and before performing skeleton extraction on the spinal contour data and the thoracic cavity contour data, the method further includes:

[0020] The atrial septum contour data, the ventricular septum contour data, the spinal contour data, and the thoracic cavity contour data are subjected to contour smoothing processing.

[0021] In one embodiment, obtaining the second cardiac axis by performing a straight-line fit based on the average coordinate points of the spinal skeleton, the average coordinate points of the thoracic skeleton, the centroid of the spine, and the centroid of the thoracic cavity includes:

[0022] The second heart axis is obtained by performing a linear fit using the least squares method based on the average coordinate points of the spinal skeleton, the average coordinate points of the thoracic skeleton, the centroid of the spine, and the centroid of the thoracic cavity.

[0023] In one embodiment, the preset target segmentation model is trained based on the U-Net segmentation model.

[0024] Secondly, this application also provides a fetal cardiac ultrasound image processing device, the device comprising:

[0025] The target segmentation module is used to acquire the target four-chamber ultrasound cross-sectional image and perform target segmentation processing on the target four-chamber ultrasound cross-sectional image based on a preset target segmentation model to obtain atrial septal contour data, interventricular septal contour data, spinal contour data and thoracic cavity contour data.

[0026] The first fitting module is used to perform axis fitting based on the atrial septal contour data and the ventricular septal contour data to obtain the first cardiac axis.

[0027] The second fitting module is used to perform axis fitting based on the spinal contour data and the thoracic contour data to obtain the second cardiac axis.

[0028] Angle calculation module is used to obtain the cardiac axis based on the angle between the first cardiac axis and the second cardiac axis.

[0029] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method described above.

[0030] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the above-described method.

[0031] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps of the above-described method.

[0032] The aforementioned fetal cardiac ultrasound image processing method, device, computer equipment, storage medium, and computer program product obtain atrial septum contour data, interventricular septum contour data, spinal contour data, and thoracic cavity contour data by performing target segmentation processing on the acquired target four-chamber cardiac ultrasound cross-sectional image. This data is then used to fit the first cardiac axis to the second cardiac axis, thereby achieving automatic measurement of the cardiac axis, i.e., the angle between the direction of the fetal heart apex and the line connecting the spine and sternum. Compared with manual tracing measurement, this method is less susceptible to the influence of clinical experience and other subjective factors, and the measurement accuracy is greatly improved. Attached Figure Description

[0033] Figure 1This is a diagram illustrating the application environment of a fetal echocardiography image processing method in one embodiment.

[0034] Figure 2 This is a flowchart illustrating a fetal cardiac ultrasound image processing method in one embodiment;

[0035] Figure 3 This is a schematic diagram of the structure of the first cardiac axis, the second cardiac axis, and the cardiac axis in one embodiment;

[0036] Figure 4 This is a flowchart illustrating the step of obtaining the first cardiac axis in one embodiment;

[0037] Figure 5 This is a schematic diagram of the extracted atrial septal skeleton and ventricular septal skeleton in one embodiment;

[0038] Figure 6 This is a flowchart illustrating the step of obtaining the second cardiac axis in one embodiment;

[0039] Figure 7 This is a schematic diagram of the spinal skeleton and thoracic skeleton extracted in one embodiment;

[0040] Figure 8 This is a structural block diagram of a fetal cardiac ultrasound image processing device in one embodiment;

[0041] Figure 9 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0043] The fetal cardiac ultrasound image processing method provided in this application embodiment can be applied to, for example... Figure 1In the application environment shown, the ultrasound imaging device 102 communicates with the server 104 via a network. A data storage system can store the data that the server 104 needs to process. The data storage system can be integrated onto the server 104 or placed in the cloud or on another network server. Specifically, the server 104 acquires the target four-chamber ultrasound cross-sectional image collected by the ultrasound imaging device 102, and performs target segmentation processing on the target four-chamber ultrasound cross-sectional image based on a preset target segmentation model to obtain atrial septal contour data, interventricular septal contour data, spinal contour data, and thoracic cavity contour data; it then performs axis fitting based on the atrial septal contour data and interventricular septal contour data to obtain a first cardiac axis; it further performs axis fitting based on the spinal contour data and thoracic cavity contour data to obtain a second cardiac axis; and finally, it obtains the cardiac axis based on the angle between the first and second cardiac axes. The server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0044] In one embodiment, such as Figure 2 As shown, a method for processing fetal cardiac ultrasound images is provided, which can be applied to... Figure 1 Taking server 104 as an example, the following S200 to S800 are included:

[0045] S200: Acquire the target four-chamber ultrasound cross-sectional image, and perform target segmentation processing on the target four-chamber ultrasound cross-sectional image based on the preset target segmentation model to obtain atrial septal contour data, interventricular septal contour data, spinal contour data and thoracic cavity contour data.

[0046] Fetal echocardiography screening is one type of prenatal ultrasound screening, generally performed in the mid-to-late stages of pregnancy (18-35 weeks). The obtained fetal echocardiography images can be used to show the anteroposterior and non-localized development of fetal internal organs. Four-chamber echocardiography is a type of fetal echocardiography image that can be used to observe the four chambers of the fetal heart: left atrium, left ventricle, right atrium, and right ventricle. It can simultaneously observe the heart's position, cardiac axis, size, and intracardiac structures, and further measure cardiac-related indicators to assist doctors in diagnosing whether the fetus has heart malformations. Measurable cardiac-related indicators may include the cardiothoracic ratio and cardiac axis. The cardiothoracic ratio is the ratio of the area of ​​the heart region to the area of ​​the thoracic cavity, and the cardiac axis is the angle between the apex of the heart and the line connecting the spine and sternum. The embodiments in this application are all explained using the measurement of the cardiac axis from four-chamber echocardiography images as an example.

[0047] Specifically, the target four-chamber echocardiogram image is a standard four-chamber echocardiogram image, that is, an image that conforms to the cross-sectional image required for clinical measurement of cardiac-related indicators. It can be understood that, based on the acquisition principles of different ultrasound acquisition devices, fetal cardiac ultrasound images can be either two-dimensional ultrasound images acquired through two-dimensional ultrasound or cross-sectional images obtained from three-dimensional ultrasound data of the fetus. The standard four-chamber echocardiogram image can be acquired manually or automatically. The method for acquiring the target four-chamber echocardiogram image is not unique. It can be obtained by manually controlling the ultrasound imaging device, viewing and filtering fetal ultrasound images by moving a mouse or trackball, and finding the ultrasound image containing the standard four-chamber echocardiogram image based on its structural characteristics. Alternatively, the ultrasound image sequence acquired by the ultrasound imaging device can be automatically identified using machine learning or deep learning algorithms to obtain the target four-chamber echocardiogram image.

[0048] Furthermore, after obtaining the target four-chamber echocardiogram image, in order to measure the cardiac axis, it is necessary to segment the target region from the target four-chamber echocardiogram image that reflects the direction of the heart apex and the line connecting the spine and sternum, namely the atrial septum region, the interventricular septum region, the spinal region, and the thoracic cavity region. It can be understood that obtaining the target region from the target four-chamber echocardiogram image can be achieved through image processing methods such as thresholding, image segmentation methods based on edge detection, or machine learning algorithms such as target segmentation and pattern recognition. In this embodiment, the process of obtaining the target region from the target four-chamber echocardiogram image is explained using a target segmentation algorithm as an example.

[0049] The preset target segmentation model is a training set constructed from standard four-chamber cardiac images with labeled target regions. This training set is then input into an image segmentation network model for training, parameter tuning, and validation, resulting in a target segmentation model with excellent segmentation performance. Based on this preset target segmentation model, the atrial septum, interventricular septum, spine, and thoracic cavity regions in the target four-chamber cardiac ultrasound image can be accurately identified. The deep learning-based image segmentation network model that can be used is not unique, and includes models such as FCN semantic segmentation, U-Net semantic segmentation, and BiseNet bidirectional segmentation network.

[0050] In one embodiment, the preset target segmentation model of this application is trained based on the U-Net segmentation model. Specifically, the U-Net semantic segmentation model and its improved versions have very accurate recognition performance in the field of medical image segmentation. The U-Net semantic segmentation model is constructed based on the encoder and decoder concept. It first extracts target features from the input image through four downsampling operations, and then decodes the extracted target features through four upsampling operations. The output result is the predicted probability that each pixel in the input image belongs to a preset category. Furthermore, by setting the confidence level for each preset category, target segmentation of the input image can be achieved.

[0051] It is understandable that the training samples in the training set are obtained by annotating the target regions of standard four-chamber echocardiogram images. The annotation can be done by using polygonal bounding boxes to represent the atrial septum, interventricular septum, spine, and thoracic cavity regions in the standard four-chamber echocardiogram image, and then generating a JSON file based on the coordinates of the polygonal bounding boxes corresponding to each target region to form the training set. Correspondingly, the atrial septum contour data represents the coordinates of the boundary pixels of the atrial septum region in the target four-chamber echocardiogram image; the interventricular septum contour data, spine contour data, and thoracic cavity contour data, etc., represent the coordinates of the boundary pixels of the interventricular septum, spine, and thoracic cavity regions, etc., in the target four-chamber echocardiogram image.

[0052] S400: The first cardiac axis is obtained by fitting the atrial septal contour data and the ventricular septal contour data.

[0053] The first cardiac axis is characterized by the central axis of the two structural contours of the atrial septum region and the interventricular septum region extracted from the target four-chamber echocardiogram image. It is also the line connecting the long axis of the atrial septum region and the interventricular septum region, and can be used to reflect the apex orientation in the target four-chamber echocardiogram image.

[0054] Specifically, one could first extract the central axis or center point from the contour data of the atrial septum and ventricular septum regions, respectively, and then fit the first cardiac axis based on the central axes or center points of the two regions. It's understood that there isn't only one way to extract the central axis or center point; one could use image thinning algorithms or skeleton extraction algorithms to repeatedly remove edges, transforming the image contour of a certain width into a skeleton with a width of only one pixel, and then using the skeleton as the central axis of the contour. Alternatively, one could calculate the centroid of the contour and use it as the center point of the contour.

[0055] S600: A second cardiac axis is obtained by fitting the axis based on the spinal contour data and the thoracic contour data.

[0056] The second cardiac axis is characterized as the central axis of the two structural contours of the spinal region and the thoracic region extracted from the target four-chamber ultrasound image, which is also the anterior-posterior axis of the fetal thoracic cavity and can be used to represent the line connecting the spine and sternum in the target four-chamber ultrasound image.

[0057] Specifically, the central axis or center point can be extracted first from the contour data of the spinal region and the thoracic region, respectively, and then the second cardiac axis can be fitted based on the central axis or center point of the two regions. The method for extracting the central axis or center point can be consistent with that described in S400, and will not be repeated here.

[0058] S800: The heart axis is obtained based on the angle between the first heart axis and the second heart axis.

[0059] Specifically, after measuring the first and second cardiac axes, the cardiac axis can be obtained by the angle between these two axes. For example... Figure 3 As shown, A represents the first cardiac axis, B represents the second cardiac axis, and α represents the cardiac axis. The International Society of Ultrasound in Obstetrics and Gynecology (ISUOG) practice guidelines state that the normal reference range for the cardiac axis (CA) of a healthy fetus is (45±20)°. Therefore, after obtaining the cardiac axis using the method described in this application, if the angle of the cardiac axis is less than 25° or greater than 65°, it indicates a possible abnormality in the fetal cardiac axis, requiring further detailed diagnosis by the physician.

[0060] The aforementioned fetal cardiac ultrasound image processing method, by performing target segmentation processing on the acquired target four-chamber ultrasound cross-sectional image, obtains atrial septum contour data, ventricular septum contour data, spinal contour data, and thoracic cavity contour data. This data is then used to fit the first and second cardiac axes, thereby enabling automatic measurement of the cardiac axis, i.e., the angle between the direction of the fetal heart apex and the line connecting the spine and sternum. Compared to manual tracing measurement, this method is less susceptible to the influence of clinical experience and other subjective factors, significantly improving accuracy. It is suitable for prenatal ultrasound examinations and helps doctors quickly classify and judge fetal four-chamber ultrasound cross-sectional images, improving the efficiency of doctor examinations.

[0061] In one embodiment, such as Figure 4 As shown, S400 includes the following S420 to S440, wherein:

[0062] S420: Extract the skeleton from the atrial septum contour data and the interventricular septum contour data respectively to obtain the atrial septum skeleton data and the interventricular septum skeleton data.

[0063] Specifically, such as Figure 5As shown, the atrial septum skeleton 21 is the skeleton line obtained by extracting the skeleton from the atrial septum contour 20, reflecting the geometry of the atrial septum region. Correspondingly, the atrial septum skeleton data is represented by the coordinate positions of the pixels of the atrial septum skeleton in the target four-chamber echocardiogram image. The interventricular septum skeleton 11 is the skeleton line obtained by extracting the skeleton from the interventricular septum contour 10, reflecting the geometry of the interventricular septum region. Correspondingly, the interventricular septum skeleton data is represented by the coordinate positions of the pixels of the interventricular septum skeleton in the target four-chamber echocardiogram image.

[0064] Furthermore, there is no single way to extract the skeleton. It can be done using algorithms based on erosion and opening operations, or by using the skeletonize() function in the morphology submodule of the Python third-party library skimage, or the thinning() function in the ximgproc module integrated in the OpenCV image processing library.

[0065] S440: The first cardiac axis is obtained by linear fitting based on the atrial septum skeleton data and the ventricular septum skeleton data.

[0066] Specifically, the first cardiac axis, i.e., the line connecting the long axis of the atrial septum and the ventricular septum, can be obtained by fitting a straight line to all coordinate points on the atrial septum and ventricular septum skeleton data. The method for performing the straight line fitting is not unique; it can be achieved using the least squares method, the gradient descent method, or linear regression.

[0067] In one embodiment, such as Figure 6 As shown, S600 includes the following S610 to S650, wherein:

[0068] S610: Extract the skeleton from the spinal contour data and the thoracic contour data respectively to obtain the spinal skeleton data and the thoracic skeleton data.

[0069] Specifically, such as Figure 7 As shown, the spinal skeleton 31 is the skeleton line obtained by extracting the spine contour 30, reflecting the geometry of the spinal region. Correspondingly, the spinal skeleton data is represented by the coordinate positions of the pixels of the spinal skeleton in the target four-chamber ultrasound image. The thoracic skeleton 41 is the skeleton line obtained by extracting the thoracic contour 40, reflecting the geometry of the thoracic region. Correspondingly, the thoracic skeleton data is represented by the coordinate positions of the pixels of the thoracic skeleton in the target four-chamber ultrasound image.

[0070] Similarly, the methods for obtaining spinal skeleton data and thoracic skeleton data can be based on algorithms based on erosion and opening operations, or on methods such as the skeletonize() function in the morphology submodule of the Python third-party library skimage, and the thinning() function in the ximgproc module integrated in the OpenCV image processing library.

[0071] Understandable, from Figure 7 As can be seen, the thoracic cavity contour and spinal contour are more complex in shape than the atrial septum contour and ventricular septum contour, and the extracted skeletal lines are not merely straight lines. To make the fitted second cardiac axis more accurate, this embodiment uses fixed points or center points on the spinal skeleton and thoracic skeleton for fitting.

[0072] S620: Calculate the average coordinates based on the spinal skeleton data to obtain the average coordinate points of the spinal skeleton. Specifically, the average coordinate points of the spinal skeleton can be obtained by traversing all coordinate points in the spinal skeleton data. The average coordinates can be calculated by summing the coordinate values ​​in the same direction and dividing by the total number of coordinates to obtain the average coordinate value in that direction. By summing the average coordinate values ​​from multiple directions in the coordinate system, the average coordinate points can be obtained.

[0073] S630: Calculate the average coordinates based on the thoracic skeleton data to obtain the average coordinate points of the thoracic skeleton. Similarly, the average coordinate points of the thoracic skeleton can be obtained by iterating through all coordinate points in the thoracic skeleton data.

[0074] S640: Extract the centroids of the spinal contour data and the thoracic contour data respectively to obtain the centroids of the spine and the thoracic cavity.

[0075] Specifically, the centroid is the average position of an object within a system of mass distributions. For a two-dimensional binary image, its centroid can represent the average of the pixels within the image. Therefore, the centroid of the spine can be used to represent the average of the spine contour data, and the centroid of the thoracic cavity can be used to represent the average of the thoracic cavity contour data. There is no single way to extract the centroid from the spine and thoracic cavity contour data; for example, it can be done using the `numpy.argwhere()` function from the Python NumPy library, or it can be done using the `cv2.moments()` function provided by OpenCV.

[0076] S650: The second heart axis is obtained by fitting a straight line based on the average coordinate points of the spinal skeleton, the average coordinate points of the thoracic skeleton, the centroid of the spine, and the centroid of the thoracic cavity.

[0077] Specifically, the second cardiac axis, or the anterior-posterior axis of the fetal chest cavity, can be obtained by fitting a straight line to four coordinate points: the average coordinate point of the spinal skeleton, the average coordinate point of the thoracic skeleton, the center of mass of the spine, and the center of mass of the chest cavity. The method for performing the straight line fitting is not unique; it can be achieved using the least squares method, the gradient descent method, or the linear regression method. In one embodiment, obtaining the second cardiac axis by fitting a straight line to the average coordinate points of the spinal skeleton, the average coordinate point of the thoracic skeleton, the center of mass of the spine, and the center of mass of the chest cavity includes: using the least squares method to fit a straight line to the average coordinate points of the spinal skeleton, the average coordinate point of the thoracic skeleton, the center of mass of the spine, and the center of mass of the chest cavity to obtain the second cardiac axis.

[0078] In one embodiment, before performing skeleton extraction on the atrial septum contour data and ventricular septum contour data, and before performing skeleton extraction on the spinal contour data and thoracic cavity contour data, the method further includes: performing contour smoothing processing on the atrial septum contour data, ventricular septum contour data, spinal contour data, and thoracic cavity contour data. It can be understood that smoothing the contour data using a Gaussian filtering smoothing algorithm before subsequent skeleton extraction and other processing steps helps the extracted skeleton lines to be closer to actual lines, thereby making the fitted first and second cardiac axes more accurate.

[0079] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0080] Based on the same inventive concept, this application also provides a fetal cardiac ultrasound image processing apparatus for implementing the aforementioned fetal cardiac ultrasound image processing method. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more fetal cardiac ultrasound image processing apparatus embodiments provided below can be found in the limitations of the fetal cardiac ultrasound image processing method described above, and will not be repeated here.

[0081] In one embodiment, such as Figure 8As shown, a fetal cardiac ultrasound image processing device is provided, comprising: a target segmentation module 110, a first fitting module 120, a second fitting module 130, and an angle calculation module 140, wherein:

[0082] The target segmentation module 110 is used to acquire the target four-chamber ultrasound cross-sectional image and perform target segmentation processing on the target four-chamber ultrasound cross-sectional image based on the preset target segmentation model to obtain atrial septal contour data, interventricular septal contour data, spinal contour data and thoracic cavity contour data.

[0083] The first fitting module 120 is used to perform axis fitting based on the atrial septal contour data and the ventricular septal contour data to obtain the first cardiac axis.

[0084] The second fitting module 130 is used to perform axis fitting based on spinal contour data and thoracic contour data to obtain the second cardiac axis.

[0085] Angle calculation module 140 is used to obtain the cardiac axis based on the angle between the first cardiac axis and the second cardiac axis.

[0086] In this embodiment, by performing target segmentation processing on the acquired target four-chamber ultrasound cross-sectional image, atrial septum contour data, interventricular septum contour data, spinal contour data, and thoracic cavity contour data are obtained. These data are used to fit the first and second cardiac axes, thereby achieving automatic measurement of the cardiac axis, i.e., the angle between the direction of the fetal heart apex and the line connecting the spine and sternum. Compared with manual tracing measurement, this method is less susceptible to the influence of clinical experience and other subjective factors, greatly improving measurement accuracy. It is suitable for prenatal ultrasound examinations and helps doctors quickly classify and judge fetal four-chamber ultrasound cross-sectional images, improving the efficiency of doctors' examinations.

[0087] In one embodiment, the first fitting module 120 is further configured to extract the skeleton from the atrial septum contour data and the ventricular septum contour data respectively to obtain atrial septum skeleton data and ventricular septum skeleton data; and to perform linear fitting based on the atrial septum skeleton data and the ventricular septum skeleton data to obtain the first cardiac axis.

[0088] In one embodiment, the second fitting module 130 is further configured to extract the skeleton from the spinal contour data and the thoracic contour data respectively to obtain spinal skeleton data and thoracic skeleton data; calculate the average coordinates based on the spinal skeleton data to obtain the average coordinate points of the spinal skeleton; calculate the average coordinates based on the thoracic skeleton data to obtain the average coordinate points of the thoracic skeleton; extract the centroids from the spinal contour data and the thoracic contour data respectively to obtain the centroid points of the spine and the thoracic cavity; and perform linear fitting based on the average coordinate points of the spinal skeleton, the average coordinate points of the thoracic skeleton, the centroid points of the spine and the thoracic cavity to obtain the second cardiac axis.

[0089] In one embodiment, the apparatus further includes:

[0090] The preprocessing module is used to perform contour smoothing on the atrial septum contour data, interventricular septum contour data, spinal contour data, and thoracic cavity contour data.

[0091] In one embodiment, the second fitting module 130 is further configured to perform linear fitting using the least squares method based on the average coordinate points of the spinal skeleton, the average coordinate points of the thoracic skeleton, the centroid of the spine, and the centroid of the thoracic cavity to obtain the second cardiac axis.

[0092] In one embodiment, the preset target segmentation model in the target segmentation module 110 is trained based on the U-Net segmentation model.

[0093] Each module in the aforementioned fetal echocardiography image processing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0094] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores target four-chamber echocardiographic images and their corresponding contour and skeleton data. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for processing fetal echocardiographic images.

[0095] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0096] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0097] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0098] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the method described above.

[0099] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data shall comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0100] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0101] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0102] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for processing fetal cardiac ultrasound images, characterized in that, The method includes: Acquire a target four-chamber echocardiogram image, and perform target segmentation processing on the target four-chamber echocardiogram image based on a preset target segmentation model to obtain atrial septum contour data, interventricular septum contour data, spinal contour data, and thoracic cavity contour data. Skeleton extraction is performed on the atrial septum contour data and the ventricular septum contour data respectively to obtain atrial septum skeleton data and ventricular septum skeleton data; A first cardiac axis is obtained by performing linear fitting between the atrial septum skeleton data and the ventricular septum skeleton data. Skeleton extraction is performed on the spinal contour data and the thoracic contour data respectively to obtain spinal skeleton data and thoracic skeleton data; The average coordinates of the spinal skeleton are calculated based on the spinal skeleton data to obtain the average coordinate points of the spinal skeleton. The average coordinates of the thoracic skeleton are calculated based on the thoracic skeleton data to obtain the average coordinate points of the thoracic skeleton. Centroids are extracted from the spinal contour data and the thoracic contour data to obtain the centroids of the spine and the thoracic cavity. The second heart axis is obtained by performing a straight line fit based on the average coordinate points of the spinal skeleton, the average coordinate points of the thoracic skeleton, the centroid of the spine, and the centroid of the thoracic cavity. The cardiac axis is obtained based on the angle between the first cardiac axis and the second cardiac axis.

2. The method according to claim 1, characterized in that, The step of extracting skeletons from the atrial septum contour data and the ventricular septum contour data, and the step of extracting skeletons from the spinal contour data and the thoracic cavity contour data, respectively, includes: An algorithm based on erosion and opening operations is used to extract the skeleton from the atrial septum contour data and the ventricular septum contour data, as well as from the spinal contour data and the thoracic cavity contour data.

3. The method according to claim 1, characterized in that, The calculation of average coordinates includes: The average coordinate value in a given direction is obtained by summing the coordinate values ​​in the same direction and dividing by the total number. The average coordinate values ​​in multiple directions are then summed in the coordinate system to obtain the average coordinate point.

4. The method according to claim 1, characterized in that, Before performing skeleton extraction on the atrial septum contour data and the ventricular septum contour data respectively, and before performing skeleton extraction on the spinal contour data and the thoracic cavity contour data respectively, the method further includes: The atrial septum contour data, the ventricular septum contour data, the spinal contour data, and the thoracic cavity contour data are subjected to contour smoothing processing.

5. The method according to claim 1, characterized in that, The step of obtaining the second cardiac axis by performing linear fitting based on the average coordinate points of the spinal skeleton, the average coordinate points of the thoracic skeleton, the centroid of the spine, and the centroid of the thoracic cavity includes: The second heart axis is obtained by performing a linear fit using the least squares method based on the average coordinate points of the spinal skeleton, the average coordinate points of the thoracic skeleton, the centroid of the spine, and the centroid of the thoracic cavity.

6. The method according to claim 1, characterized in that, The preset target segmentation model is trained based on the U-Net segmentation model.

7. A fetal cardiac ultrasound image processing device, characterized in that, The device includes: The target segmentation module is used to acquire the target four-chamber ultrasound cross-sectional image and perform target segmentation processing on the target four-chamber ultrasound cross-sectional image based on a preset target segmentation model to obtain atrial septal contour data, interventricular septal contour data, spinal contour data and thoracic cavity contour data. The first fitting module is used to extract the skeleton from the atrial septum contour data and the ventricular septum contour data respectively to obtain atrial septum skeleton data and ventricular septum skeleton data; and to perform linear fitting based on the atrial septum skeleton data and ventricular septum skeleton data to obtain the first cardiac axis. The second fitting module is used to extract the skeleton from the spinal contour data and the thoracic contour data respectively to obtain spinal skeleton data and thoracic skeleton data; calculate the average coordinates based on the spinal skeleton data to obtain the average coordinate points of the spinal skeleton; calculate the average coordinates based on the thoracic skeleton data to obtain the average coordinate points of the thoracic skeleton; extract the centroids from the spinal contour data and the thoracic contour data respectively to obtain the centroid points of the spine and the thoracic cavity; and perform linear fitting based on the average coordinate points of the spinal skeleton, the average coordinate points of the thoracic skeleton, the centroid points of the spine and the thoracic cavity to obtain the second cardiac axis. Angle calculation module is used to obtain the cardiac axis based on the angle between the first cardiac axis and the second cardiac axis.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images