Ultrasound imaging method, system and computer readable storage medium

Abstract

Embodiments of the present application disclose an ultrasonic imaging method, system and computer readable storage medium. The method comprises controlling a probe to emit ultrasonic waves to a uterus of a subject, and receiving ultrasonic echoes returned by the subject; controlling to obtain ultrasonic echo data after processing the ultrasonic echoes; obtaining endometrial image data of the uterus of the subject based on the ultrasonic echo data; automatically determining state information of the uterus of the subject based on the endometrial image data; and controlling to output prompt information corresponding to the state information of the uterus of the subject. Embodiments of the present application can reduce the problem that the accuracy of examination is affected by subjectivity of medical staff.

Description

Technical Field

[0001] This application relates to the field of medical testing technology, and in particular to an ultrasound imaging method, system and computer-readable storage medium. Background Technology

[0002] In modern medical imaging, ultrasound technology has become a widely used and frequently employed examination method due to its advantages such as high reliability, speed, convenience, real-time imaging, and repeatability. For example, in gynecological ultrasound, after examining the uterus and its adnexa, medical staff use their clinical experience to examine or diagnose based on the coronal section of the uterus displayed on the monitor. However, relying on clinical experience introduces a degree of subjectivity, and for medical staff with less clinical experience, there is a possibility of missed diagnoses or misinterpretations, leading to a decrease in the accuracy of the examination. Summary of the Invention

[0003] This application provides an ultrasound imaging method, system, and computer-readable storage medium that can improve the accuracy and consistency of gynecological ultrasound.

[0004] The first aspect of this application provides an ultrasound imaging method, including:

[0005] The probe is controlled to emit ultrasound waves toward the subject's uterus and to receive the ultrasound echoes returned by the subject.

[0006] The ultrasonic echo is processed by the controller to obtain ultrasonic echo data;

[0007] Based on the ultrasound echo data, three-dimensional volume data of the endometrial image of the subject's uterus were obtained;

[0008] The state information of the subject's uterus is automatically determined based on the three-dimensional volume data of the endometrial image.

[0009] The control output provides prompts corresponding to the state information of the subject's uterus.

[0010] A second aspect of this application provides an ultrasound imaging system, comprising:

[0011] The probe is used to emit ultrasonic waves to the subject and receive the ultrasonic echoes returned by the subject to obtain ultrasonic echo data.

[0012] A processor, connected to the probe, is used to acquire three-dimensional volume data of the endometrial image of the subject's uterus based on the ultrasound echo data; the processor also automatically determines the state information of the subject's uterus based on the three-dimensional volume data of the endometrial image, and controls the output of prompt information corresponding to the state information of the subject's uterus.

[0013] A third aspect of this application provides a computer-readable storage medium for storing a computer program for electronic data interchange, wherein the computer program causes a computer to perform some or all of the steps described in any method of the first aspect of this application.

[0014] This application embodiment automatically determines the state information of the subject's uterus based on endometrial imaging data and controls the output of prompt information corresponding to the state information of the subject's uterus. By automatically examining and determining the state information of the subject's uterus after obtaining the ultrasound data of the subject's uterus, the inaccuracy of the examination due to the subject's subjectivity can be reduced. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the hardware structure of an ultrasound imaging system in one embodiment of this application.

[0017] Figure 2 This is a flowchart of the steps of the ultrasound imaging method in the embodiments of this application.

[0018] Figure 3 This is a schematic diagram of the coronal plane of the uterus in an embodiment of this application.

[0019] Figure 4 This is a schematic diagram of a two-dimensional ultrasound image with a preset thickness in one embodiment of this application.

[0020] Figure 5 This is a schematic diagram of a volumetric ultrasound image with a preset thickness in another embodiment of this application.

[0021] Figure 6 This is a schematic diagram of the cross-sectional image and trajectory curve of the sagittal plane in an embodiment of this application.

[0022] Figure 7 yes Figure 6 A schematic diagram of the intima in the mid-coronal plane.

[0023] Figure 8 This is a schematic diagram of acquiring intima images based on trajectory curves in an embodiment of this application.

[0024] Figure 9 This is a schematic diagram of the state information of the subject's uterus in one embodiment of this application.

[0025] Figure 10 This is a schematic diagram of the state information of the subject's uterus in another embodiment of this application.

[0026] Figure 11 This is a schematic diagram of the state information of the subject's uterus in another embodiment of this application.

[0027] Figure 12 This is a block diagram of an ultrasound imaging system according to another embodiment of this application. Detailed Implementation

[0028] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0029] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0030] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0031] Please see Figure 1The diagram shows a hardware structure schematic of an ultrasound imaging system according to an embodiment of this application. The ultrasound imaging system 10 may include a probe 100, a transmitting circuit 102 connected to the probe 100, a receiving circuit 104 connected to the probe 100, a beamforming module 106, a signal processing module 108, an imaging processing module 110, and a display 112. The receiving circuit 104, beamforming module 106, signal processing module 108, imaging processing module 110, and display 112 are electrically connected sequentially. In this embodiment, the ultrasound imaging system 10 automatically examines and determines the state information of the subject's uterus after acquiring ultrasound data, and outputs corresponding prompts. This reduces the inaccuracy of the examination due to the subjectivity of medical personnel.

[0032] Please refer to the following: Figure 2 The diagram shows a flowchart of the ultrasound imaging method in an embodiment of this application. The ultrasound imaging method includes the following steps:

[0033] Step 200: Control the probe to emit ultrasound waves toward the subject's uterus and receive the ultrasound echoes returned by the subject.

[0034] In this embodiment, the transmitting circuit 102 can transmit ultrasound waves to the uterus of the subject via the probe 100 to obtain ultrasound echo data of the subject's uterus. After a certain delay, the probe 100 can receive ultrasound echoes reflecting back from the tested tissue, carrying information about the object being tested. The probe 100 may include transducers arranged in a linear array or a matrix. A probe 100 with a linear array of transducers can acquire two-dimensional ultrasound echo data, such as a linear array probe. A probe 100 with a linear array of transducers can also acquire three-dimensional ultrasound echo data, such as a convex array probe with an electric motor. A probe 100 with a matrix of transducers can acquire three-dimensional ultrasound echo data (or three-dimensional volume data), such as an area array probe.

[0035] Step 202: Control the processing of the ultrasonic echo to obtain ultrasonic echo data.

[0036] After the probe 100 acquires two-dimensional or three-dimensional ultrasound echo data of the subject's uterus, the probe 100 converts this ultrasound echo into an electrical signal. The receiving circuit 104 receives the electrical signal generated by the probe 100, obtains the ultrasound echo data, and sends this ultrasound echo data to the beamforming module 106. The beamforming module 106 performs focusing delay, weighting, and channel summation on the ultrasound echo data, and then sends the beam-synthesized ultrasound echo data to the signal processing module 108 for relevant signal processing. The ultrasound echo data processed by the signal processing module 108 is sent to the imaging processing module 110, which performs different processing on the ultrasound echo data according to the different imaging modes required by the user to obtain tissue image data of different modes. Then, after processing such as logarithmic compression, dynamic range adjustment, and digital scan transformation, different modes of ultrasound tissue images are formed and displayed on the display 112. The different modes of ultrasound tissue images may include M-images, B-images, C-images, etc.

[0037] Step 204: Obtain endometrial image data of the subject's uterus based on the ultrasound echo data.

[0038] In this embodiment, the endometrial imaging data includes one or both of two-dimensional ultrasound imaging data and three-dimensional ultrasound echo imaging data. Two-dimensional ultrasound imaging data refers to ultrasound echo data obtained under two-dimensional ultrasound imaging, such as two-dimensional ultrasound images acquired by a probe 100 arranged in a linear array of elements; or, two-dimensional ultrasound imaging data refers to two-dimensional ultrasound echo data corresponding to a preset cross-section (or section) in three-dimensional ultrasound imaging. Three-dimensional ultrasound echo imaging data can be three-dimensional data acquired by a probe 100 arranged in a matrix of elements, or three-dimensional data acquired by a probe 100 arranged in a linear array of elements, such as a convex array probe with an electric motor. The three-dimensional ultrasound echo imaging data refers to all or part of the three-dimensional ultrasound echo data or volumetric data obtained by three-dimensional ultrasound imaging.

[0039] In this application, the three-dimensional ultrasound echo type image data mainly refers to the three-dimensional volumetric data of the endometrial image. In some embodiments, the endometrial image data of the subject's uterus mainly includes the three-dimensional volumetric data of the endometrial image.

[0040] The display 112 can be used for direct two-dimensional ultrasound images. When displaying three-dimensional ultrasound images on the display 112, the imaging processing module 110 can display the images obtained from the three-dimensional ultrasound echo data at different preset cross-sectional positions in the three-dimensional ultrasound imaging on the display 112. In this way, the imaging processing module 110 can obtain two-dimensional and / or three-dimensional ultrasound endometrial image data of the subject's uterus based on the imaging mode of the ultrasound echo data.

[0041] In one embodiment, the imaging processing module 110 can acquire two-dimensional ultrasound images at different cross-sectional positions in the three-dimensional ultrasound echo data. Thus, the endometrial image data includes two-dimensional ultrasound images at one or more cross-sectional positions, wherein the cross-sectional positions include the positions corresponding to one or more cross-sections of the sagittal, coronal, and transverse sections of the subject's uterus.

[0042] Please refer to the following: Figure 3 The diagram shows a schematic representation of the coronal plane of the uterus in an embodiment of this application. In one embodiment, the ultrasound imaging system 10 can receive adjustment signals input by a user. These adjustment signals may include adjustments to the image corresponding to the three-dimensional ultrasound echo data displayed by the imaging processing module 110. For example, the user can input adjustment signals through a human-computer interaction interface to determine the endometrial image 600 of the coronal plane of the uterus of the corresponding subject. The endometrial image 600 may include one or more anatomical structures among the uterine body, endometrium, vagina, urethra, and cervix. Adjustment signals may be input via, but are not limited to, a keyboard, scroll wheel, touch-enabled display, mouse, or a transceiver module for gesture control signals.

[0043] In one embodiment, the image processing module 110 can be based on an image automatic segmentation processing method that is automatically run by the system. For example, after performing automatic image segmentation processing on the obtained three-dimensional ultrasound echo data, the two-dimensional ultrasound images at the corresponding cross-sectional positions of the sagittal, coronal, and transverse sections of the subject's uterus are identified.

[0044] In one embodiment, the cut position can also be obtained by the user through a human-computer interaction interface by inputting adjustment signals to rotate and translate the image of the three-dimensional ultrasound echo data. In this way, the endometrial image data can include two-dimensional ultrasound images of cuts at different positions and directions obtained after rotation and translation operations.

[0045] Please refer to the following: Figure 4The diagram illustrates a two-dimensional ultrasound image of a preset thickness in one embodiment of this application. The two-dimensional ultrasound image may contain volumetric data (or ultrasound echo data) of a preset thickness to improve the contrast resolution of the displayed image and enhance the display of the anatomical structure and features of the uterus. The two-dimensional ultrasound image 620 of the preset thickness may contain ultrasound echo data within the region formed by a first plane 624 and a cross-sectional ultrasound image 628. The first plane 624 is parallel to the cross-sectional ultrasound image 628, and the distance 622 between the first plane 624 and the cross-sectional ultrasound image 628 is the preset thickness. The cross-sectional ultrasound image 628 is the ultrasound image at a cross-sectional position. The imaging processing module 110 can perform volumetric rendering on the two-dimensional ultrasound image 620 of the preset thickness and display the volumetrically rendered two-dimensional ultrasound image. Volumetric rendering includes, but is not limited to, surface mode, X-ray mode, or a fusion of both. Thus, the endometrial image data includes the volumetrically rendered two-dimensional ultrasound image 620 of the preset thickness.

[0046] Please refer to the following: Figure 5 The diagram shows a two-dimensional ultrasound image with a preset thickness in another embodiment of this application. The two-dimensional ultrasound image 620 with the preset thickness can be ultrasound echo data within the region encompassed by the second plane 634 and the third plane 636. The cross-sectional ultrasound image 638 is located between and parallel to the second plane 634 and the third plane 636. The distance 632 between the second plane 634 and the third plane 636 is the preset thickness. The cross-sectional ultrasound image 638 is the ultrasound image at a cross-sectional position.

[0047] The preset thickness can be set to a fixed value (e.g., 2.0 mm) according to actual clinical needs, or the thickness parameter can be adaptively adjusted according to the anatomical structure and characteristics of the uterus. The imaging processing module 110 can set the preset thickness parameter options within the interface corresponding to the endometrial image 600. Users can manually set the preset thickness parameter options based on their personal needs and operating habits.

[0048] In this embodiment, the volumetric ultrasound image with a preset thickness can be displayed by the imaging processing module 110 through one or more volumetric rendering modes, including surface mode, maximum value mode, light and shadow rendering mode, and X-ray mode.

[0049] In this embodiment, the imaging processing module 110 can perform volumetric rendering on the ultrasound echo data at the location of the cross-sectional ultrasound image 628, including but not limited to one or more volumetric rendering modes such as surface mode, maximum value mode, lighting and shadow rendering mode, and X-ray mode for volume rendering. Thus, the endometrial image data can be a two-dimensional ultrasound image obtained by volumetric rendering of the ultrasound echo data at the location of the cross-sectional ultrasound image 628. Figure 3As shown, it is a two-dimensional ultrasound image of the coronal plane of the subject's uterus obtained after volume rendering.

[0050] Please see Figure 6 The diagram shows a cross-sectional image and trajectory curve in the sagittal plane according to an embodiment of this application. The display 112 can display cross-sectional images (or section images) at different cross-sectional positions or directions. However, cross-sectional images at some positions or directions may not contain specific anatomical structures, potentially leading to a less comprehensive or accurate examination of the subject's uterine condition. Therefore, the imaging processing module 110 can acquire a cross-sectional image corresponding to a target location based on the displayed cross-sectional image to obtain an ultrasound image containing specific anatomical structures. The imaging processing module 110 can determine a trajectory curve on a cross-sectional image of the three-dimensional ultrasound echo data and acquire endometrial image data of the subject's uterus based on this trajectory curve.

[0051] For example, the coronal endometrial image of the uterus contains more anatomical structures and is easier to identify. Therefore, when the user needs to display the coronal endometrial image of the subject, the imaging processing module 110 obtains the coronal endometrial image from the non-coronal endometrial image and displays it. For example, when the display 112 displays the sagittal cross-sectional image 602 of the subject's uterus (the cross-sectional image 602 is also an endometrial image), the imaging processing module 110 can automatically determine the trajectory curve 601 on the sagittal cross-sectional image 602 based on image recognition or machine learning, and can also obtain the ultrasound echo data of the trajectory curve 601 on the cross-sectional image 602. In one embodiment, the imaging processing module 110 can receive the user's input operation in the cross-sectional image 602 and determine that the trajectory corresponding to the input operation is the trajectory curve 601. For example, when displaying the sagittal cross-sectional image 602 of the subject's uterus, the user can draw the trajectory curve 601 on the sagittal cross-sectional image 602 using a mouse. In one embodiment, when displaying a sagittal cross-sectional image 602 of the subject's uterus, the user can draw a region of interest on the sagittal cross-sectional image 602 using a mouse. The imaging processing module 110 can determine the trajectory curve 601 by performing image segmentation or machine learning on the endometrial image within the region of interest. In other embodiments, the trajectory curve 601 can also be a straight line or a curve.

[0052] Please refer to the following: Figure 7 As shown Figure 6 A schematic diagram of the endometrial image in the coronal plane. The imaging processing module 110 can acquire the endometrial image 600 of the subject's uterus in the coronal plane based on this trajectory curve 601.

[0053] Please refer to the following: Figure 8The diagram illustrates the acquisition of endometrial images based on a trajectory curve in this embodiment of the application. The imaging processing module 110 can acquire multiple parallel section images parallel to the profile image 602, such as a first parallel section image 604, a second parallel section image 606, and a third parallel section image 608, wherein the distances between each parallel section image are equal or unequal. The imaging processing module 110 can determine a trajectory plane 610 perpendicular to the profile image 602 based on the trajectory curve 601. The imaging processing module 110 determines a two-dimensional ultrasound image composed of ultrasound echo data intersecting the trajectory plane 610 and each of the multiple parallel section images, and can identify this two-dimensional ultrasound image as the endometrial image data of the subject.

[0054] For example, the ultrasound echo data where trajectory plane 610 intersects with the first parallel section 604 includes the ultrasound echo data corresponding to curve 603; the ultrasound echo data where trajectory plane 610 intersects with the second parallel section 606 includes the ultrasound echo data corresponding to curve 605; and the ultrasound echo data where trajectory plane 610 intersects with the third parallel section 608 includes the ultrasound echo data corresponding to curve 607. Therefore, the endometrial image data includes a two-dimensional ultrasound image composed of the ultrasound echo data corresponding to trajectory curves 601, 603, 605, and 607.

[0055] In one embodiment, the trajectory surface 610 may also have a preset thickness. The imaging processing module 110 can display the ultrasound echo image data within the trajectory surface 610, which includes the thickness, using one or more volumetric rendering modes among surface mode, maximum value mode, light and shadow rendering mode, and X-ray mode. Thus, the endometrial image data may include a two-dimensional ultrasound image obtained after volumetric rendering. When the trajectory curve 601 is a curve, the imaging processing module 110 can project the three-dimensional ultrasound data within the trajectory surface 610 onto a plane for display on the display 112.

[0056] In this embodiment, the imaging processing module 110 can perform volumetric rendering on the ultrasound echo data at the location of the trajectory surface 610, including but not limited to one or more volumetric rendering modes such as surface mode, maximum value mode, lighting and shadow rendering mode, and X-ray mode for volume rendering. Thus, the endometrial image data can be a two-dimensional ultrasound image obtained by volumetric rendering of the ultrasound echo data at the location of the trajectory surface 610. Figure 7 As shown, it is a two-dimensional ultrasound image of the coronal plane of the subject's uterus obtained after volume rendering.

[0057] Please refer to the following: Figure 8The imaging processing module 110 can acquire two-dimensional ultrasound images of multiple parallel sections from the three-dimensional ultrasound echo data, such as a first parallel section image 604, a second parallel section image 606, a third parallel section image 608, and a cross-sectional image 602 (which can also be represented as a fourth parallel section image 602). It can also determine that the endometrial image data includes one or more of these parallel two-dimensional ultrasound images. Each parallel section's two-dimensional ultrasound image can also have a preset thickness. The imaging processing module 110 can display each parallel section's two-dimensional ultrasound image using one or more volumetric rendering modes, including surface mode, maximum value mode, lighting and shadow rendering mode, and X-ray mode. Thus, the endometrial image data can include two-dimensional ultrasound images obtained through volumetric rendering.

[0058] In this embodiment, the imaging processing module 110 can perform volumetric rendering on the ultrasound echo data at the location of the fourth parallel section image 602, including but not limited to one or more volumetric rendering modes such as surface mode, maximum value mode, lighting and shadow rendering mode, and X-ray mode for volume rendering. Thus, the endometrial image data can be a two-dimensional ultrasound image obtained by volumetric rendering of the ultrasound echo data at the location of the fourth parallel section image 602. Figure 7 As shown, it is a two-dimensional ultrasound image of the coronal plane of the subject's uterus obtained after volume rendering.

[0059] Step 206: Automatically determine the state information of the subject's uterus based on the endometrial imaging data.

[0060] Wherein, when the endometrial image data includes the three-dimensional volume data of the endometrial image, step 206 includes: automatically determining the state information of the subject's uterus based on the three-dimensional volume data of the endometrial image.

[0061] After acquiring endometrial imaging data of the subject's uterus, the imaging processing module 110 can automatically determine the state information of the subject's uterus based on the endometrial imaging data. The state information of the subject's uterus includes the orientation and type of malformation. Clinically, uterine malformations may include, but are not limited to, bicornuate uterus, septate uterus, and arcuate uterus. The orientation of the uterus can be represented by its position within the pelvic cavity, namely: anteverted uterus, mid-position uterus, and retroverted uterus.

[0062] In this embodiment, the imaging processing module 140 can determine the state information of the subject's uterus based on different detection modes, which may include a first detection mode, a second detection mode, and a third detection mode.

[0063] For the first detection mode:

[0064] The imaging processing module 110 can determine the state information of the subject's uterus corresponding to the endometrial image data based on machine learning. That is, in the first detection mode, the imaging processing module 110 can classify the endometrial image data (such as two-dimensional ultrasound image data or three-dimensional ultrasound image data) as a whole or directly by the first learning model obtained through machine learning to obtain the state information of the subject's uterus.

[0065] For example, the database contains a first preset sample image set labeled with different state information. This first preset sample image set includes several first-type sample images labeled as normal uterus, several second-type sample images labeled with different uterine orientations (e.g., images labeled with anteverted, mid-position, and retroverted uterus), and several third-type sample images labeled with different types of uterine malformations (e.g., images labeled with bicornuate, septate, and arcuate uterus). The imaging processing module 110 can extract features from the first preset sample image set, generate an image feature set corresponding to the first preset sample image set, and control machine learning to perform machine learning on the image feature set of the first preset sample image set to obtain a first learning model.

[0066] In this embodiment, the imaging processing module 110 can control the extraction of features from the first preset sample image set using one or more feature extraction operations, such as PCA (Principal Component Analysis), LDA (Linear Discriminant Analysis), Haar features, texture features, wavelet features, and deep neural networks (e.g., CNN, ResNet, VGG, Inception, MobileNet, etc.). The imaging processing module 110 can also control the learning of the image feature set of the first preset sample image set using one or more learning models, such as nearest neighbor classification, support vector machine, random forest, and neural networks, to obtain the first learning model. The various feature extraction methods and learning modules can refer to relevant technologies. In other embodiments, the first learning model can also be obtained by other hosts or processing modules learning the image feature set of the first preset sample image set, and the imaging processing module 110 only needs to load the first learning model.

[0067] After obtaining endometrial image data of the subject's uterus, the imaging processing module 110 can extract features from the endometrial image data to generate image features corresponding to the endometrial image data. The aforementioned feature extraction method can be used when extracting features from the endometrial image data. The imaging processing module 110 can classify the image features of the endometrial image data based on the first learning model to obtain the state information of the subject's uterus.

[0068] The uterine condition information of the test subject includes at least one of the following: probability information of the test subject's uterus being normal, probability information of each positional state in different orientations, and probability information corresponding to each malformation type in different uterine malformation types.

[0069] In this embodiment, since the first preset sample image set includes several first-type sample images indicating a normal uterus, several second-type sample images indicating different uterine orientations, and several third-type sample images indicating different types of uterine malformations, after classifying the image features of the endometrial image data through the first learning model, at least one of the following can be obtained: probability information of a normal uterus, probability information of each orientation state, and probability information corresponding to each malformation type. Specifically, the probability information of each orientation state may include one or more of the following: a first probability information indicating the subject's uterus is anteverted, a second probability information indicating the subject's uterus is mid-positioned, and a third probability information indicating the subject's uterus is retroverted. The probability information of each malformation type may include a fourth probability information indicating the subject's uterus is bicornuate, a fifth probability information indicating the subject's uterus is normal, a sixth probability information indicating the subject's uterus is bicornuate, a seventh probability information indicating the subject's uterus is septate, and an eighth probability information indicating the subject's uterus is arcuate.

[0070] For the second detection mode:

[0071] Various uterine malformations exhibit certain differences in characteristics across different anatomical structures. For example, in a bicornuate uterus, a longitudinal section reveals two uterine bodies, with endometrial echoes visible within each body. Each body has its own cervix and vagina, or two cervixes and one vagina, but the vagina contains a complete septum. In a bicornuate uterus, the two uterine bodies are separated, each with its own independent endometrium. However, the two endometrium fused together at the cervix or the lower segment of the uterine body, connecting to a single cervix. In a septate uterus, the uterine body is intact, but the endometrial lining within the uterine cavity exhibits a certain degree of separation. Therefore, the imaging processing module 110 can determine the type of uterine malformation based on the relative position, morphology, or differences between characteristic anatomical structures when detecting the uterine malformation type in the subject. The imaging processing module 100 can determine the orientation of the uterus by determining the relative positional relationship between the uterine body and characteristic anatomical structures such as the vagina, urethra, bladder, and anal canal when detecting the orientation information of the subject's uterus. The characteristic anatomical structures may be one or more of the following: uterine body, endometrium, vagina, urethra, cervix, etc. The endometrial imaging data may include one or more of the aforementioned characteristic anatomical structures.

[0072] When diagnosing or examining the uterus of a test subject, the user may need to determine whether the anatomical structure in the region of interest is abnormal, and if it is abnormal, what is the probability that the anatomical structure in the region of interest is abnormal; if it is normal, what is the probability that the anatomical structure in the region of interest is normal.

[0073] Therefore, the imaging processing module 110 can perform feature anatomical structure detection on the endometrial image data to determine the target area containing the feature anatomical structure in the endometrial image data; the imaging processing module 110 can determine the state information of the subject's uterus based on the target area, wherein the target area is the area in the endometrial image data that contains one or more of the aforementioned feature anatomical structures.

[0074] In one embodiment, the identification method for characteristic anatomical structures such as the endometrium can be manual or automatic. When manually acquiring regions of interest containing characteristic anatomical structures, the user marks specific anatomical structures within the image displayed on the monitor 112 using tools such as a keyboard and mouse. The imaging processing module 110 can determine the type and location of the characteristic anatomical structures based on the user's markings to obtain the target region containing the characteristic anatomical structures in the endometrial image data. When automatically acquiring target regions containing one or more characteristic anatomical structures, the imaging processing module 110 can binarize the endometrial image data to obtain one or more candidate regions of the subject's endometrium. The imaging processing module 110 can determine the probability value that each candidate region contains the subject's endometrium based on the morphology of the endometrium. Then, the imaging processing module 110 can determine the candidate region with the highest probability value as the target region. In other embodiments, other characteristic anatomical structures (such as the uterine body, vagina, urethra, cervix, bladder, anal canal, etc.) can also be identified as target regions of characteristic anatomical structures contained in the endometrial image data through manual or automatic methods.

[0075] In one embodiment, the imaging processing module 110 may acquire target regions containing characteristic tissue structures in the endometrial imaging data based on a learning model.

[0076] For example, the database may include a second preset sample image set with calibration information. This second preset sample image set includes several fourth-type sample images calibrated with target regions containing one or more characteristic tissue structures, or several fifth-type sample images calibrated with masks corresponding to one or more characteristic tissue structures. In the fourth-type sample images, regions without target regions are considered non-target regions. In the fifth-type sample images, the masks corresponding to the target regions of the characteristic tissue structures have a first value, and the masks in the fifth-type sample images for regions other than the target regions of the characteristic tissue structures have a second value. The imaging processing module 110 controls feature extraction from the second preset sample image set to obtain an image feature set of the second preset sample image set, and learns from this image feature set to obtain a second learning model. The second learning model may also be a pre-modeled model, and the imaging processing module 110 only loads this second learning model. The imaging processing module 110 can determine, based on the second learning model, the target regions containing the characteristic tissue structures of the subject's uterus in the endometrial imaging data. In one embodiment, after the second learning model determines the target region in the endometrial image data, the imaging processing module 110 can simultaneously classify the target region based on the second learning model to obtain the state information of the subject's uterus.

[0077] For example, in the fourth type of sample images in the second preset sample image set, the target region is further labeled with information such as "normal uterus," the corresponding uterine orientation, and / or the type of uterine malformation. Similarly, in the fifth type of sample images, the mask corresponding to the target region is further labeled with information such as "normal uterus," the corresponding uterine orientation, and / or the type of uterine malformation. Thus, in the second learning model, the imaging processing module 110 can directly and automatically determine the state information of the subject's uterus based on the target region in the endometrial image data.

[0078] In one embodiment, the imaging processing module 110 extracts features from the endometrial image data and a second preset sample image set using a sliding window method. For example, the imaging processing module 110 can extract features from the endometrial image data located within the sliding window during the sliding window operation. The feature extraction using the sliding window method includes, but is not limited to, extraction algorithms such as PCA (Principal Component Analysis), LDA (Linear Discriminant Analysis), Haar features, texture features, and neural networks. Subsequently, the imaging processing module 110 can classify the image features of the endometrial image data within the sliding window based on a second learning model to obtain the state information of the subject's uterus. The second learning model may include, but is not limited to, classifiers such as KNN (k-Nearest Neighbor), SVM (Support Vector Machine), random forest, and neural networks.

[0079] In one embodiment, the imaging processing module 110 can extract features from the endometrial image data and the second preset sample image set based on the bounding box regression method of a deep learning network. The second learning model can be obtained when extracting features from the second preset sample image set. For example, the imaging processing module 110 can learn features and regress parameters on the constructed database by stacking convolutional layers and fully connected layers. For the endometrial image data, the imaging processing module 110 can directly regress the bounding box of the corresponding target region using a deep learning network. The imaging processing module 110 can obtain the bounding box corresponding to the target region in the endometrial image data based on the second learning model, and determine the state information of the subject's uterus corresponding to the endometrial image data within the bounding box. The deep learning network includes, but is not limited to, learning models such as R-CNN, Fast R-CNN, Faster R-CNN, SSD, and YOLO.

[0080] In one embodiment, after determining the target region in the endometrial image data based on the aforementioned second learning model, the imaging processing module 110 further classifies it using a third learning model (or the first learning model). For example, after determining the target region containing characteristic tissue structures in the endometrial image data, the imaging processing module 110 can determine the state information of the subject's uterus based on the target region. The imaging processing module 110 controls feature extraction of the target region in the endometrial image data, generates image features corresponding to the target region, and classifies the image features of the target region based on the third learning model to obtain the state information of the subject's uterus. The imaging processing module 110 can extract features from a third preset sample image set that calibrates different state information, generating an image feature set of the third preset sample image set. The third preset sample image set includes several sixth-type sample images calibrating a normal uterus, several seventh-type sample images calibrating different uterine orientations, and several eighth-type sample images calibrating different types of uterine malformations. The imaging processing module 110 can control the extraction of features from a third preset sample image set to obtain an image feature set of the third preset sample image set, and perform machine learning on the image feature set of the third preset sample image set to obtain a third learning model. In this embodiment, the third learning model is similar to the aforementioned first learning model, and specific details can be found in the aforementioned first learning model.

[0081] In the second detection mode, the imaging processing module 110 has determined the target regions containing characteristic tissue structures in the endometrial image data. Therefore, when classifying the endometrial image data through the second or third learning model, the imaging processing module 110 can obtain the state information of each target region in the endometrial image data, such as obtaining one or more target regions in the endometrial image data and the probability information corresponding to each target region.

[0082] As mentioned above, the state information of the subject's uterus includes at least one of the following: probability information of a normal uterus, probability information of each orientation state, and probability information corresponding to each malformation type among different uterine malformation types. In some embodiments, when the imaging processing module 110 determines the state information of the subject's uterus based on the characteristic anatomical structures in the three-dimensional volume data of the endometrial image, it may specifically include: determining the relative positional relationship between different characteristic anatomical structures, and determining the orientation state of the subject's uterus based on the relative positional relationship; and / or, determining the morphological information of each characteristic anatomical structure, and controlling the comparison of the morphological information of the characteristic anatomical structure with the morphological information of the corresponding standard tissue structure to determine the malformation type of the subject's uterus.

[0083] That is, in some embodiments, the various orientations of the subject's uterus are determined by establishing the relative positional relationships between different characteristic anatomical structures, and then based on those relative positional relationships. The various malformation types of the subject's uterus are determined by first establishing the morphological information of each characteristic anatomical structure, then controlling the comparison of the morphological information of the characteristic anatomical structure with the morphological information of the corresponding standard tissue structure, and finally determining the type of uterine malformation based on the comparison results.

[0084] For the third detection mode:

[0085] In clinical diagnosis, determining the position of the uterus requires assessing its relative location to the vagina, urethra, or bladder. However, diagnosing uterine malformations necessitates precise segmentation of key anatomical structures and accurate measurement of relevant parameters. For example, in an arcuate uterus, the angle between the lines connecting the apex of the endometrium at both uterine horns and the lowest point of the uterine cavity floor is obtuse, and the depth of the uterine cavity floor depression is <10mm; in a septate uterus, the angle between the lines connecting the apex of the endometrium at both uterine horns and the lowest point of the uterine cavity floor is acute, and the depth of the uterine cavity floor depression is ≥10mm. Therefore, in the third detection mode, the imaging processing module 110 can identify one or more characteristic anatomical structures of the subject within the endometrial image data to determine the state information of the subject's uterus based on these specific characteristic anatomical structures.

[0086] In this embodiment, the imaging processing module 110 can determine one or more characteristic anatomical structures of the subject contained in the endometrial image data based on an image segmentation model. The image segmentation method includes one or more of the following: Snake model, Graph Cut model, LevelSet model, and RandomWalker model, to obtain characteristic tissue structures such as the uterine body, vagina, urethra, cervix, bladder, anal canal, or endometrium in the endometrial image data.

[0087] In one embodiment, the imaging processing module 110 extracts features from a fourth preset sample image set calibrated with different anatomical structures to generate an image feature set for the fourth preset sample image set. The fourth preset sample image set includes several ninth-type sample images calibrated with different anatomical structure types. When extracting features from a third preset sample image set, the imaging processing module 110 can acquire image blocks of each pixel and its periodic region in the ninth-type sample images, and extract features from each image block to obtain the image feature set for the fourth preset sample image set. The feature extraction method can be PCA, LDA, Haar features, texture features, etc., or a deep neural network can be used for feature extraction. Subsequently, the image processing module 110 can obtain a fourth learning model based on machine learning (such as KNN, SVM, random forest, neural networks, etc.). When acquiring endometrial image data, the imaging processing module 110 can also acquire data blocks of each pixel in the endometrial image data, wherein each pixel data block is ultrasound echo data containing a preset range of each pixel. The imaging processing module 110 extracts features from the data blocks of each pixel to obtain the image features of the data blocks of each pixel in the corresponding ultrasound echo data. The imaging processing module 110 also classifies the image features of the data blocks of each pixel based on a fourth learning model to obtain the characteristic anatomical structure corresponding to each data block. For example, when the fourth learning model determines that the data block of the current pixel corresponds to a characteristic tissue structure, the fourth learning model can mark the data block of that pixel as a target pixel data block; when the fourth learning model determines that the data block of the current pixel corresponds to a non-characteristic tissue structure, the fourth learning model can mark the data block of that pixel as a background pixel data block. Thus, after identifying the data blocks of each pixel in the endometrial image data, the imaging processing module 110 can determine a portion of the characteristic tissue structure in the endometrial image data marked as a target pixel data block.

[0088] In one embodiment, the imaging processing module 110 can extract features from a fifth preset sample image set labeled with different anatomical structures to generate an image feature set for the fifth preset sample image set. The fifth preset sample image set includes several tenth-type sample images labeled with different anatomical structure types. The imaging processing module 110 applies a deep learning network to the image feature set of the fifth preset sample image set to obtain a fifth learning model; and based on the fifth learning model, it obtains each anatomical structure feature contained in the endometrial imaging data. For example, the imaging processing module 110 uses an end-to-end semantic segmentation network method based on deep learning: it learns features from the image feature set of the fifth preset sample image set by stacking basic convolutional layers and fully connected layers, and adds upsampling or deconvolutional layers to make the input and output sizes the same, thereby directly obtaining the feature tissue structures and their corresponding categories in the endometrial imaging data. The deep learning network may include, but is not limited to, FCN, U-Net, and Mask R-CNN.

[0089] In one embodiment, the third detection mode may be based on the target region determined by the second detection mode, i.e., the target region determined by the second detection mode, and determine one or more feature anatomical structures contained within the target region based on the aforementioned third detection mode. Alternatively, the third detection mode may determine the feature anatomical structures contained within the region of interest (ROI) based on a user-defined ROI bounding box. Alternatively, the third detection mode may acquire points or lines drawn by the user's input operation, obtain the corresponding target region based on an image segmentation algorithm (such as Graph Cut, RandomWalker), and then detect the feature anatomical structures contained within the target region using the third detection mode.

[0090] In this embodiment, the imaging processing module 110 can determine the state information of the subject's uterus based on the characteristic anatomical structures in the endometrial imaging data. For example, the imaging processing module 110 can determine the relative positional relationship between different characteristic anatomical structures, and determine the orientation and malformation type of the subject's uterus based on the relative positions. The imaging processing module 110 can also determine the morphological information of each characteristic anatomical structure, and control the comparison of the morphological information of the characteristic anatomical structure with the morphological information of the corresponding standard tissue structure to determine the malformation type of the subject's uterus.

[0091] For example, each characteristic anatomical structure may have one or more measurement items, such as the distance between the endometrial apexes of the bilateral uterine horns and the angle at the bottom of the uterine cavity. Therefore, after determining the characteristic anatomical structures in the endometrial imaging data, the imaging processing module 110 can obtain the values ​​of the measurement items for each characteristic anatomical structure to determine the type of uterine malformation of the subject based on the data of the measurement items of the anatomical structure. For septate uterus and arcuate uterus, the uterine body contour is displayed normally in both two-dimensional and three-dimensional ultrasound scans. The septate uterus and arcuate uterus can be regionalized by determining the separation angle and depth of the uterine cavity. For example, in arcuate uterus, the angle between the lines connecting the endometrial apexes of the bilateral uterine horns and the lowest point of the uterine cavity is obtuse, and the depth of the uterine cavity depression is <10mm, while in septate uterus, the angle between the lines connecting the endometrial apexes of the bilateral uterine horns and the lowest point of the uterine cavity is acute, and the depth of the uterine cavity depression is ≥10mm. The imaging processing module 110 can compare the values ​​of the measurement items of each characteristic anatomical structure with the standard values ​​of the corresponding standard tissue structure to determine the type of uterine malformation of the subject.

[0092] In one embodiment, since the morphology of characteristic anatomical structures (such as their contour information) can be used to determine the type of uterine malformation, the imaging processing module 110 can acquire the contour information of each characteristic anatomical structure and determine the probability information of normal or abnormal contour information of each characteristic anatomical structure based on a sixth learning model. The sixth learning model is used to extract features from a sixth preset sample image set, generating an image feature set of the sixth preset sample image set. The sixth preset sample image set includes several eleventh-type sample images indicating a normal uterus, several twelfth-type sample images indicating different uterine orientations, and several thirteenth-type sample images indicating different types of uterine malformations. In other embodiments, the target region containing characteristic anatomical structures in the endometrial imaging data can be obtained through the second detection mode described above, and the probability information of normal or abnormal characteristic anatomical structures in the target region can be determined based on the sixth learning model, thus determining the final malformation type.

[0093] In other embodiments, when determining the state information of the subject's uterus, the imaging processing module 110 may combine the probability corresponding to the state information of the endometrial image data determined by the first detection mode, the probability corresponding to the target area of ​​the endometrial image data determined by the second detection mode, and the probability corresponding to the characteristic anatomical structure determined by the third detection mode (which may be a weighted average of the aforementioned probabilities).

[0094] Step 208: Control the output of prompt information corresponding to the state information of the subject's uterus.

[0095] Specifically, step 208 includes: controlling the display of at least one of the following: probability information of the subject's uterus being normal, probability information of each position in different orientations, and probability information of each malformation type among different types of uterine malformations.

[0096] In different detection modes, the imaging processing module 110 may output different prompts corresponding to the state information of the subject's uterus.

[0097] Please refer to the following: Figure 9 The diagram illustrates the state information of the subject's uterus in one embodiment of this application. In the first detection mode, the imaging processing module 110 can display the probability information of each position of the subject's uterus in different orientations and the probability information corresponding to each malformation type in different uterine malformation types in the prompt display area 680. For example, the probability information of each position in different orientations includes the probability information of the subject's uterus being an anteverted uterus (e.g., 0.97), a mid-positioned uterus (e.g., 0.01), and a retroverted uterus (e.g., 0.02); the probability information of each malformation type in different uterine malformation types includes the probability information of the subject's uterus being a bicornuate uterus (e.g., 0.96), a normal uterus (e.g., 0.01), a bicornuate uterus (e.g., 0.02), and a septate uterus (e.g., 0.01).

[0098] Please refer to the following: Figure 10 The diagram illustrates the state information of the subject's uterus in another embodiment of this application. In the second detection mode, the imaging processing module 110 can display the probability information of each position of the subject's uterus in different orientations and the probability information corresponding to each type of uterine malformation in the prompt display area 680. For example, the probability information of each position of the subject's uterus includes the probability information of the subject's uterus being anteverted (e.g., 0.97), mid-position (e.g., 0.01), and retroverted (e.g., 0.02); the probability information of the subject's uterus being a unilateral uterus (e.g., 0.90), a normal uterus (e.g., 0.05), and a bicornuate uterus (e.g., 0.02). When controlling the display of the probability information of each position of the subject's uterus in different orientations, the imaging processing module 110 can mark the region of interest corresponding to the probability information of each position in the displayed endometrial image data. When controlling the display of probability information for each type of uterine malformation in the subject's uterus, the imaging processing module 110 can mark the region of interest corresponding to the probability information of each malformation type in the displayed endometrial image data. For example, if the subject's uterine body has an abnormal shape (only one uterine body, not displayed symmetrically as an inverted triangle), the imaging processing module 110 will mark the abnormal area with a red rectangle and display the probability of a single uterus as 0.9. Correspondingly, if the cervix of the subject's uterus displays normally, the imaging processing module 110 will mark the area with a green rectangle and display the probability of a normal uterus as 0.05.

[0099] Please refer to the following: Figure 11 The diagram illustrates the state information of the subject's uterus in another embodiment of this application. In the third detection mode, the imaging processing module 110 accurately detects and segments key anatomical structures, then performs precise morphological analysis and quantitative measurement to determine the uterine type. The imaging processing module 110 can display the probability information of each position of the subject's uterus in different orientations in the prompt display area 680, as well as the probability information corresponding to each malformation type among different uterine malformation types with different anatomical features. For example, the probability information of each position includes the probability information of the subject's uterus being an anteverted uterus (e.g., 0.97), a mid-positioned uterus (e.g., 0.01), and a retroverted uterus (e.g., 0.02); the uterine malformation type is determined to be a septate uterus, and the value of each measurement of the septate uterus is displayed. The imaging processing module 110 also displays the measured length of the endometrial horn line (AB) as 16 mm, the internal fundus contour angle (MPN) as 75 degrees, the distance (CP) between the bottom of the internal fundus contour and the lines connecting the two endometrial horns as 12 mm, and the ratio of CP to AB as 0.75. Since CP ≥ 10 mm and the angle MPN is an acute angle (i.e., < 90°), the imaging processing module 110 determines that the subject's uterine malformation type is septate uterus. Furthermore, if the ratio of CP to AB is greater than 10%, it will have a certain impact on fertility, and the imaging processing module 110 can also display corresponding prompts.

[0100] In one embodiment, the imaging processing module 110 can also generate a diagnostic report. For example, after completing an automatic diagnosis, the imaging processing module 110 can automatically generate a diagnostic report, reducing the tedious operation of manually filling in the report by the user. The content of the report can be consistent with the automatic diagnosis content displayed on the interface, or it can add content based on the content displayed on the interface. For example, the report can add basic information such as the patient's name, age, gestational age, and medical history, and can also add some descriptive terms to further explain the content displayed on the interface.

[0101] The aforementioned ultrasound imaging method automatically determines the state information of the subject's uterus based on endometrial imaging data and controls the output of prompt information corresponding to the state information of the subject's uterus. By automatically examining and determining the state information of the subject's uterus after acquiring ultrasound data of the subject's uterus, the accuracy of the examination can be reduced due to the subjectivity of medical staff.

[0102] Please see Figure 12 The diagram shown is a block diagram of another embodiment of an ultrasound imaging system of this application. Figure 12As shown, the ultrasound imaging system 80 can be applied to the above embodiments. The ultrasound imaging system 80 provided in this application will be described below. The ultrasound imaging system 80 may include a processor 800, a storage device 802, a probe 100, a control circuit 804, and a display 112, as well as a computer program (instructions) stored in the storage device 802 and executable on the processor 800. The ultrasound imaging system 80 may also include other hardware components, such as communication devices, buttons, keyboards, etc., which will not be elaborated here. The processor 800 can exchange data with the probe 100, the control circuit 804, the storage device 802, and the display 112 via signal line 808.

[0103] The processor 800 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the ultrasound imaging system 80, connecting various parts of the ultrasound imaging system 80 through various interfaces and lines. In this embodiment, the processor 800 can be used to implement all the functions of the image processing module 110, and can also integrate the functions of the beamforming module 106 and the signal processing module 108, as detailed in the foregoing embodiments. The probe 100 can integrate the transmitting circuit 102 and the receiving circuit 104 from the above embodiments. That is, in some embodiments, the ultrasound imaging system 80 may only include the processor 800, the storage device 802, the probe 100, and the display 112.

[0104] The control circuit 804 may also include the functions of the transmitting circuit 102, receiving circuit 104, beamforming module 106 and / or signal processing module 108 in the above embodiments, and can be referred to the foregoing embodiments for details.

[0105] The storage device 802 can be used to store the computer program and / or modules. The processor 800 implements various functions of the ultrasound imaging method described above by running or executing the computer program and / or modules stored in the storage device 802 and calling the data stored in the storage device 802. The storage device 802 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc. In addition, the storage device 802 may include a high-speed random access storage device, and may also include a non-volatile storage device, such as a hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0106] The display 112 can display a user interface (UI), a graphical user interface (GUI), a screen corresponding to the three-dimensional volume data of the subject's head, or a midsagittal plane. The ultrasound imaging system 804 can also be used as an input device and an output device. The display 112 can include at least one of liquid crystal display (LCD), thin film transistor LCD (TFT-LCD), organic light-emitting diode (OLED) touch display, flexible touch display, three-dimensional (3D) touch display, etc.

[0107] The processor 800 runs a program corresponding to the executable program code stored in the storage device 802 to execute the ultrasound imaging method in any of the preceding embodiments.

[0108] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0109] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0110] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0111] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. An ultrasound imaging system, characterized in that, The ultrasound imaging system includes: The probe is used to emit ultrasonic waves to the subject and receive the ultrasonic echoes returned by the subject to obtain ultrasonic echo data. A processor, connected to the probe, is used to acquire three-dimensional volume data of the endometrial image of the subject's uterus based on the ultrasound echo data. The processor also automatically determines the state information of the subject's uterus based on the three-dimensional volume data of the endometrial image and controls the output of prompt information corresponding to the state information of the subject's uterus. Specifically, when automatically determining the state information of the subject's uterus based on the three-dimensional volume data of the endometrial image, the processor determines the state information of the subject's uterus corresponding to the three-dimensional volume data of the endometrial image based on machine learning; or, the processor detects the subject's characteristic anatomical structures contained in the three-dimensional volume data of the endometrial image and determines the state information of the subject's uterus based on the characteristic anatomical structures in the three-dimensional volume data of the endometrial image. Specifically, when determining the state information of the subject's uterus based on the characteristic anatomical structures in the three-dimensional volume data of the endometrial image, the processor is used to: determine the relative positional relationship between different characteristic anatomical structures and determine the orientation state of the subject's uterus based on the relative positional relationship. The state information of the subject's uterus includes at least one of the following: the probability information of the subject's uterus being normal, the probability information of each positional state in different orientations, and the probability information of each malformation type in different uterine malformation types.

2. The ultrasound imaging system as described in claim 1, characterized in that, When determining the state information of the subject's uterus corresponding to the three-dimensional volume data of the endometrial image based on machine learning, the processor is used to control the feature extraction of the three-dimensional volume data of the endometrial image, generate image features corresponding to the three-dimensional volume data of the endometrial image, and classify the image features of the three-dimensional volume data of the endometrial image based on the first learning model to obtain the state information of the subject's uterus.

3. The ultrasound imaging system as described in claim 2, characterized in that, When classifying the image features of the three-dimensional volume data of the endometrial image based on the first learning model to obtain the state information of the subject's uterus, the processor is also used to obtain at least one of the following based on the first learning model: the probability information of the subject's uterus being normal, the probability information of each positional state in different orientations, and the probability information of each malformation type in different uterine malformation types.

4. The ultrasound imaging system as described in claim 3, characterized in that, When the processor controls the output of prompt information corresponding to the state information of the subject's uterus, it controls the display of at least one of the following: the probability information of the subject's uterus being normal, the probability information of each position in different orientations, and the probability information of each malformation type among different types of uterine malformations.

5. The ultrasound imaging system as described in claim 1, characterized in that, When detecting the subject's characteristic anatomical structures contained in the three-dimensional volume data of the endometrial image, the processor controls the display of the three-dimensional volume data of the subject's uterine endometrial image on the display, and determines the characteristic anatomical structures based on the user's input operations on the display.

6. The ultrasound imaging system as described in claim 1, characterized in that, When detecting that the subject's characteristic anatomical structure is contained in the three-dimensional volume data of the endometrial image, the processor controls the detection of characteristic anatomical structure in the three-dimensional volume data of the endometrial image to determine the target region containing the characteristic anatomical structure in the three-dimensional volume data of the endometrial image; the processor also determines the state information of the subject's uterus based on the target region.

7. The ultrasound imaging system as described in claim 6, characterized in that, When controlling the detection of characteristic anatomical structures in the three-dimensional volume data of the endometrial image to determine the target region containing the characteristic anatomical structure in the three-dimensional volume data of the endometrial image, the processor controls the binarization processing of the three-dimensional volume data of the endometrial image to obtain one or more candidate regions of the characteristic tissue structure of the corresponding subject; the processor also determines the probability value of each candidate region containing the characteristic tissue structure of the subject, and determines the candidate region with the highest probability value as the target region.

8. The ultrasound imaging system as described in claim 6, characterized in that, When controlling the detection of characteristic anatomical structures in the three-dimensional volume data of the endometrial image to determine the target region containing the characteristic anatomical structure in the three-dimensional volume data of the endometrial image, the processor determines the target region containing the characteristic tissue structure of the subject's uterus in the three-dimensional volume data of the endometrial image based on the second learning model.

9. The ultrasound imaging system as described in claim 8, characterized in that, While identifying the target region containing the characteristic tissue structure of the subject's uterus in the three-dimensional volumetric data of the endometrial image, the processor also determines the state information of the subject's uterus based on the second learning model; or When the target region containing the characteristic tissue structure of the subject's uterus is determined in the three-dimensional volume data of the endometrial image, the processor controls the feature extraction of the target region and classifies the image features of the target region based on the third learning model to obtain the state information of the subject's uterus.

10. The ultrasound imaging system as claimed in claim 1, characterized in that, When determining the state information of the subject's uterus based on the characteristic anatomical structures in the three-dimensional volume data of the endometrial image, the processor is further configured to: determine the morphological information of each characteristic anatomical structure, and control the comparison of the morphological information of the characteristic anatomical structure with the morphological information of the corresponding standard tissue structure to determine the malformation type of the subject's uterus.

11. The ultrasound imaging system as described in claim 6, characterized in that, The status information includes probability information of different orientation states of the subject's uterus corresponding to the target area and probability information of different malformation types of the subject's uterus; When the processor controls the output of prompt information corresponding to the state information of the subject's uterus, it controls the display of probability information of different orientation states of the subject's uterus, and marks the probability information of orientation states and the corresponding regions of interest in the displayed three-dimensional volume data of the endometrial image. The processor also controls the display of probability information for different types of uterine malformations of the subject, and marks the probability information of the malformation type and the corresponding region of interest in the displayed three-dimensional volume data of the endometrial image.

12. The ultrasound imaging system as claimed in claim 1, characterized in that, When detecting that the subject's characteristic anatomical structures are included in the three-dimensional volumetric data of the endometrial image, the processor determines one or more characteristic anatomical structures of the subject included in the three-dimensional volumetric data of the endometrial image based on an image segmentation model.

13. The ultrasound imaging system as described in claim 1, characterized in that, When detecting that the subject's characteristic anatomical structure is included in the three-dimensional volume data of the endometrial image, the processor is configured to: acquire a data block of each pixel in the three-dimensional volume data of the endometrial image, wherein the data block of each pixel includes ultrasound echo data of a preset range of each pixel; Feature extraction is performed on the data block of each pixel to obtain the image features of the data block of each pixel; and The image features of each pixel data block are classified based on the fourth learning model to obtain the feature anatomy structure corresponding to each pixel data block.

14. The ultrasound imaging system as claimed in claim 1, characterized in that, When detecting the subject's characteristic anatomical structures contained in the three-dimensional volumetric data of the endometrial image, the processor acquires each characteristic anatomical structure contained in the endometrial image data based on the fifth learning model.

15. The ultrasound imaging system as claimed in claim 1, characterized in that, When detecting that the endometrial imaging data contains the subject's characteristic anatomical structures, the processor determines a target region in the endometrial imaging data that contains the characteristic anatomical structures, and determines one or more characteristic anatomical structures of the subject based on the target region.

16. The ultrasound imaging system according to any one of claims 12 to 15, characterized in that, When determining the state information of the subject's uterus based on the characteristic anatomical structures in the endometrial imaging data, the processor is used to determine the relative positional relationship between different characteristic anatomical structures, and determine the orientation state of the subject's uterus based on the relative positions; the processor also determines the morphological information of each characteristic anatomical structure, and controls the comparison of the morphological information of the characteristic anatomical structure with the morphological information of the corresponding standard tissue structure to determine the malformation type of the subject's uterus.

17. The ultrasound imaging system as described in claim 16, characterized in that, Each characteristic anatomical structure includes one or more measurement items. When determining the malformation type of the subject's uterus, the processor acquires the value of the measurement item of each characteristic anatomical structure and controls the comparison of the value of the measurement item of each characteristic anatomical structure with the standard value of the corresponding standard tissue structure to determine the malformation type of the subject's uterus; or, the processor acquires the contour information of each characteristic anatomical structure and determines the probability information of normal or abnormal contour information of each characteristic anatomical structure based on the sixth learning model.

18. The ultrasound imaging system as claimed in claim 17, characterized in that, The status information includes probability information corresponding to different orientation states of the subject's uterus and probability information corresponding to different malformation types of the subject's uterus; when controlling the output of prompt information corresponding to the status information of the subject's uterus, the processor is used to control the display of probability information of different orientation states of the subject's uterus; the processor is also used to control the display of the values ​​of the measurement items of the characteristic anatomical structure and the display of the malformation type of the subject's uterus determined based on the values ​​of the measurement items of the characteristic anatomical structure.

19. The ultrasound imaging system as claimed in claim 1, characterized in that, When acquiring three-dimensional volume data of the endometrial image of the subject's uterus based on the ultrasound echo data, the processor is also used to acquire two-dimensional ultrasound images at at least one preset cross-sectional position in the three-dimensional volume data of the endometrial image.

20. The ultrasound imaging system as described in claim 19, characterized in that, The preset section positions include the positions corresponding to one or more sections of the sagittal, coronal, and transverse sections of the subject's uterus.

21. An ultrasound imaging system, characterized in that, The ultrasound imaging system includes: A probe is used to emit ultrasonic waves to the subject and receive the ultrasonic echoes returned by the subject. A processor, connected to the probe, controls the processing of the ultrasound echoes to obtain ultrasound echo data, and acquires endometrial image data of the subject's uterus based on the ultrasound echo data; the processor also automatically determines the state information of the subject's uterus based on the endometrial image data, and controls the output of prompt information corresponding to the state information of the subject's uterus. The endometrial imaging data includes three-dimensional volumetric data of the endometrial image. When automatically determining the state information of the subject's uterus based on the endometrial imaging data, the processor determines the state information of the subject's uterus corresponding to the three-dimensional volumetric data of the endometrial image based on machine learning. Alternatively, the processor detects the subject's characteristic anatomical structures contained in the three-dimensional volumetric data of the endometrial image and determines the state information of the subject's uterus based on the characteristic anatomical structures in the three-dimensional volumetric data of the endometrial image. When determining the state information of the subject's uterus based on the characteristic anatomical structures in the three-dimensional volumetric data of the endometrial image, the processor is used to: determine the relative positional relationship between different characteristic anatomical structures and determine the orientation state of the subject's uterus based on the relative positional relationship. The state information of the subject's uterus includes at least one of the following: the probability information of the subject's uterus being normal, the probability information of each positional state in different orientations, and the probability information of each malformation type in different uterine malformation types.

22. The ultrasound imaging system as described in claim 21, characterized in that, The three-dimensional volumetric intima image data refers to all or part of the three-dimensional ultrasound echo data obtained by three-dimensional ultrasound imaging.

23. An ultrasound imaging method, characterized in that, The ultrasound imaging method includes: The probe is controlled to emit ultrasound waves toward the subject's uterus and to receive the ultrasound echoes returned by the subject. The ultrasonic echo is processed by the controller to obtain ultrasonic echo data; Based on the ultrasound echo data, three-dimensional volume data of the endometrial image of the subject's uterus were obtained; The state information of the subject's uterus is automatically determined based on the three-dimensional volume data of the endometrial image. The control output provides prompt information corresponding to the state information of the subject's uterus; The automatic determination of the subject's uterine condition information based on the three-dimensional volume data of the endometrial image includes: The state information of the subject's uterus is determined based on the three-dimensional volume data of the endometrial image using machine learning; or The characteristic anatomical structures of the subject contained in the three-dimensional volumetric image data of the endometrial image are detected; Determining the state information of the subject's uterus based on the characteristic anatomical structures in the three-dimensional volume data of the endometrial image includes: determining the relative positional relationship between different characteristic anatomical structures, and determining the orientation state of the subject's uterus based on the relative positional relationship; The state information of the subject's uterus includes at least one of the following: the probability information of the subject's uterus being normal, the probability information of each positional state in different orientations, and the probability information of each malformation type in different uterine malformation types.

24. The ultrasound imaging method as described in claim 23, characterized in that, The process of determining the state information of the subject's uterus corresponding to the three-dimensional volume data of the endometrial image based on machine learning includes: Based on the first learning model, at least one of the following is obtained: the probability information of a normal uterus, the probability information of each position in different orientations, and the probability information of the malformation type among different types of uterine malformations.

25. The ultrasound imaging method as described in claim 24, characterized in that, The control output provides prompts corresponding to the state information of the subject's uterus, including: The control displays at least one of the following: the probability information of the subject's uterus being normal, the probability information of each position in different orientations, and the probability information of each type of uterine malformation.

26. A computer-readable storage medium storing computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the ultrasound imaging method as described in any one of claims 23 to 25.

Images