Ultrasonic image-based contraction recognition method, system, terminal and storage medium

Abstract

The present application belongs to the technical field of uterine contraction recognition, and particularly relates to a uterine contraction recognition method and system based on ultrasound images, a terminal and a storage medium. The method comprises: based on comparison and processing of the obtained first-time point ultrasound image and second-time point ultrasound image, a muscle change area is recognized; when the muscle change area meets a preset trigger condition, uterine contraction state determination processing is performed. The uterine contraction state determination processing comprises: calculating a uterine contraction evaluation value based on the muscle change area, and determining the uterine contraction state according to a preset judgment relationship between the uterine contraction evaluation value and a preset uterine contraction judgment threshold. Through multi-level condition judgment, the present application effectively filters out transient noise and non-uterine contraction activity artifacts, and improves the accuracy and robustness of uterine contraction recognition.

Description

Technical Field

[0001] This invention belongs to the field of uterine contraction recognition technology, specifically relating to a method, system, terminal, and storage medium for uterine contraction recognition based on ultrasound images. Background Technology

[0002] Uterine contractions, also known as uterine contractions, are a key physiological indicator of the labor process. Their status can be used to determine the initiation and progress of labor, thereby assessing the progress of labor, deciding on the delivery method, and promptly addressing any abnormalities.

[0003] Current ultrasound image-based uterine contraction recognition technology generally uses past experience parameters or preset manual selection to determine the image evaluation area for analysis. The above methods ignore the physiological differences between different pregnant women or the same pregnant woman at different stages of labor, causing the analysis area to deviate from reality and affecting the evaluation results.

[0004] Existing identification mechanisms mostly rely on simple image feature comparison or fixed thresholds to make judgments. They cannot perform in-depth analysis of subtle dynamic changes in ultrasound images, making it difficult for medical staff to grasp the best clinical intervention window and the timing of delivery preparation. After the initial assessment area is determined, the system cannot dynamically track and optimize according to the actual changes in muscle morphology during the uterine contraction process, thus greatly reducing its adaptability in complex clinical scenarios.

[0005] In view of this, the present invention provides a method, system, terminal and storage medium for uterine contraction recognition based on ultrasound images. Summary of the Invention

[0006] The purpose of this invention is to provide a method, system, terminal, and storage medium for uterine contraction recognition based on ultrasound images, in order to solve the technical problem in the prior art where the assessment of uterine contraction status through ultrasound images is misjudged or delayed due to the arbitrary determination of the assessment area.

[0007] To achieve the above-mentioned objectives, the present invention employs the following technical solution: a method for identifying uterine contractions based on ultrasound images, comprising the following steps:

[0008] Acquire and record an ultrasound image sequence of the target muscle tissue, the ultrasound image sequence including at least a first time point ultrasound image and a second time point ultrasound image;

[0009] The acquired ultrasound image sequences were compared and processed to identify areas of muscle change.

[0010] When the area of ​​muscle change meets the preset triggering conditions, the uterine contraction status determination process is executed.

[0011] The process for determining the state of uterine contractions includes:

[0012] The contraction assessment value is calculated based on the area of ​​muscle change, and the contraction status is determined according to the preset judgment relationship between the contraction assessment value and the preset contraction judgment threshold.

[0013] Among them, the muscle change areas that meet the preset trigger conditions include:

[0014] The area of ​​muscle change is greater than the first preset threshold;

[0015] The calculation of uterine contraction assessment values ​​based on muscle change areas includes:

[0016] Determine the start and end times of the changes corresponding to the areas of muscle change;

[0017] Calculate the second parameter based on the start and end times of the change;

[0018] Based on the second parameter, the uterine contraction assessment value is determined.

[0019] Preferably, the method further includes: comparing the second parameter with a second preset threshold to obtain a final effective parameter; and determining a uterine contraction assessment value based on the final effective parameter.

[0020] Preferably, the method further includes: identifying muscle regions in the ultrasound image at a first time point, determining the edge positions of the muscle regions, and using the edge positions as location points to be tracked for identifying areas of muscle change.

[0021] Preferably, the method further includes: registering and aligning the ultrasound image at the first time point with the standard template image.

[0022] The ultrasound image-based uterine contraction recognition system includes the following modules:

[0023] The image acquisition and processing module is used to acquire at least a first time point ultrasound image and a second time point ultrasound image of muscle tissue, and to identify the muscle change area based on the images;

[0024] The contraction status judgment module is used to calculate the contraction assessment value based on the muscle change area when the muscle change area meets the preset trigger conditions, and to determine the contraction status according to the preset judgment relationship between the contraction assessment value and the preset contraction judgment threshold.

[0025] Preferably, the preset triggering condition is: the area of ​​muscle change is greater than a first preset threshold.

[0026] Preferably, the uterine contraction status judgment module is further used to: determine the start time point and the end time point of the change corresponding to the muscle change area, calculate a second parameter based on the start time point and the end time point of the change, and determine the uterine contraction assessment value based on the second parameter.

[0027] Preferably, the uterine contraction status judgment module is further configured to: compare the second parameter with a second preset threshold to obtain a final effective parameter, and determine a uterine contraction assessment value based on the final effective parameter.

[0028] The terminal includes a memory and a processor. The memory stores a computer program, and the processor is configured to execute the computer program to implement a method for recognizing uterine contractions based on ultrasound images, such as the one described above.

[0029] A storage medium on which a computer program is stored, which, when executed by a processor, implements a method for recognizing uterine contractions based on ultrasound images, such as the one described above.

[0030] Based on the above technical solution, the present invention can achieve the following beneficial effects:

[0031] 1. This invention identifies muscle change areas by comparing ultrasound images at a first time point and ultrasound images at a second time point, and then filters these muscle change areas according to preset triggering conditions. A second parameter is calculated based on the start and end times of the change corresponding to each muscle change area, and a valid parameter marker is generated using this second parameter. Based on the temporal continuity of the valid parameter markers, the second parameter that meets the conditions is determined as the final valid parameter. Through multi-level filtering, not only can artifacts caused by instantaneous image noise or non-uterine contraction physiological activities be effectively filtered out, but it also ensures that only the final valid parameter representing continuous muscle changes enters the uterine contraction state determination process, thereby improving the accuracy of uterine contraction identification.

[0032] 2. After identifying the muscle change area, the present invention dynamically sets a second preset threshold for screening the second parameter based on the degree of change of the muscle change area itself. This enables the present invention to have adaptive capability when performing uterine contraction recognition. It automatically adjusts the judgment benchmark according to the ultrasound image characteristics of different individuals to cope with data fluctuations caused by probe position, imaging quality or physiological differences, and avoids the recognition inaccuracy problem caused by using a fixed threshold.

[0033] 3. This invention obtains a significance score by statistically analyzing the consecutive occurrences of valid parameter markers, and uses a numerical range containing a minimum threshold and a maximum threshold to judge the significance score. Data exceeding the maximum threshold is judged as abnormal data and removed. Through this two-way constraint filtering strategy, the minimum threshold can filter out low-intensity noise that has no clinical significance, while the maximum threshold can exclude signal artifacts caused by non-physiological strenuous movements such as body movements, thus ensuring the purity of the data used to calculate uterine contraction assessment values. Attached Figure Description

[0034] Figure 1 This is a flowchart of the method provided by the present invention;

[0035] Figure 2This is a flowchart of the method for calculating uterine contraction assessment values ​​based on start and end time points provided by the present invention;

[0036] Figure 3 This is a flowchart of the method for outputting assessment results based on uterine contraction assessment values ​​provided by the present invention;

[0037] Figure 4 This is a system module diagram provided by the present invention. Detailed Implementation

[0038] To make the objectives, features, and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0039] Example 1

[0040] like Figures 1-3 As shown, this embodiment discloses a uterine contraction recognition method based on ultrasound images. By objectively and quantitatively analyzing the ultrasound image sequence, it captures subtle physical changes in uterine muscle tissue caused by uterine contractions, such as tissue deformation and changes in internal echo texture, thereby achieving reliable recognition and quantitative assessment of the uterine contraction state.

[0041] This embodiment analyzes the physiological process of uterine contractions, tracks and quantifies the changes in physical characteristics in ultrasound images, and eliminates various interferences through a multi-level screening and verification process to arrive at an accurate judgment.

[0042] Specifically, the steps include the following:

[0043] To acquire and record the ultrasound image sequence of the target muscle tissue, the operator uses an ultrasound probe to continuously and dynamically scan the area of ​​the uterine muscle tissue in the pregnant woman's abdomen. According to the preset frame rate, ultrasound images arranged in chronological order are acquired to form an ultrasound image sequence for subsequent analysis.

[0044] Before the acquisition begins, the target muscle tissue boundary contour is identified and delineated in the ultrasound image based on the gradient difference or texture features of pixel gray values, according to image processing rules, so as to determine the muscle region to be analyzed.

[0045] Furthermore, before acquiring ultrasound images at subsequent time points, the target muscle tissue is preprocessed in a structured manner to establish a stable reference system. The original ultrasound image is then transformed to eliminate geometric distortion and registered with a standard template image to correct the overall image displacement caused by non-uterine contraction physiological activities such as the pregnant woman's breathing and slight movements. This allows subsequent change analysis to focus on local deformation caused by uterine contractions.

[0046] Specifically, a two-dimensional planar coordinate transformation is performed on the determined muscle region to obtain planar data. Through preset geometric transformation calculations, the original ultrasound sector scan image is mapped onto a two-dimensional Cartesian coordinate system to eliminate geometric distortion caused by changes in probe angle. Then, a digital graphic representing the muscle tissue contour under a typical anatomical structure is provided as a standard template image, which is used as a reference benchmark in the registration and alignment process. By performing a series of coordinate transformation operations of rotation, scaling, and translation, the planar data and the standard template image are registered and aligned. An optimal set of transformation parameters is found to minimize the spatial correspondence error between the planar data and the standard template image. This is used to quantify the degree of inconsistency in the spatial position between the planar data and the standard template image, such as the root mean square error of the corresponding point set on their contours. After registration and alignment are completed, the boundary of the planar data can be stably obtained, and this boundary is determined as the edge position.

[0047] Furthermore, after obtaining a stable edge position, a set of tracking points are marked in the ultrasound image based on the edge position as anchor points for subsequent motion analysis to track and quantify the local deformation of muscle tissue. The selection rules can be to evenly distribute points on the boundary contour at preset arc length intervals, or to distribute points at locations where the contour curvature changes significantly and at locations within the muscle tissue with stable and distinguishable echo texture features.

[0048] Further, after preprocessing, the change recognition step begins. This step requires obtaining images at two time points from the ultrasound image sequence: usually, ultrasound images at the initial time and a subsequent time are selected, i.e., the first time point and the second time point.

[0049] Then, by comparing the two images and calculating the displacement of the points to be tracked, the region reflecting the local mechanical deformation of the muscle can be identified.

[0050] The algorithm for calculating the displacement of the tracked location point between two time points is completed through a set of tracking processing steps. It typically involves searching for the pixel block with the highest similarity to the corresponding pixel block in the first time point image within the second time point image. In the ultrasound image at the first time point, a pixel block is defined centered on the tracked location point. Then, within a preset search range in the ultrasound image at the second time point, the pixel block with the highest similarity to this pixel block is found. The similarity can be measured using criteria such as normalized cross-correlation. The center position of the found block is determined as the new position of the tracked location point at the second time point. Connecting the positions at the two time points generates a displacement vector representing its direction and distance of motion. When a group of spatially adjacent tracked location points exhibit directional coordination and significant amplitude displacement, the area enclosed by these points is identified as the muscle change region. This method of identifying physical displacement more directly reflects the mechanical deformation of the muscle and has higher reliability.

[0051] Furthermore, by setting adaptive assessment thresholds, the differences in uterine contraction intensity among different pregnant women and gestational weeks can be accommodated, and the assessment thresholds can be dynamically updated based on the latest information on muscle activity.

[0052] Specifically, by continuously statistically analyzing a recent time window, the average displacement amplitude of all identified muscle change areas is multiplied by a specific coefficient to serve as the current evaluation threshold. This screening criterion can then be automatically adjusted based on actual observation data.

[0053] Furthermore, within the identified muscle change areas, a refined analysis is performed to screen out the same pixel locations within the area between the first and second time points. Then, the portion where the pixel value difference exceeds the first preset threshold is designated as the area to be evaluated. If there is a significant signal change within the area to be evaluated, it indicates that it is caused by changes in the tissue microstructure.

[0054] In ultrasound images, changes in pixel values ​​manifest as changes in speckle patterns, which are directly related to the compression or stretching of the microstructure of muscle tissue. The first preset threshold is a fixed value, and its core function is to initially filter out background noise and avoid irrelevant signals from interfering with subsequent analysis.

[0055] Then, for each region to be evaluated, a first parameter is calculated and output. This first parameter is an indicator that quantifies the degree of local change. It is a numerical indicator used to quantify the degree of change within the region to be evaluated. For example, it can be the average, maximum, or standard deviation of the differences in all pixel values ​​within the region.

[0056] Furthermore, the first parameter of each region to be evaluated is compared with the set adaptive evaluation threshold. If the first parameter is greater than or equal to the adaptive evaluation threshold, that is, the preset judgment relationship is met, the corresponding region to be evaluated is determined as the region to be verified, which facilitates the secondary confirmation of the preliminary screening results. Only regions with sufficiently significant changes are regarded as signals with potential physiological significance, so as to enter the subsequent verification process.

[0057] Furthermore, the continuity of the region to be verified is verified in the time dimension, and transient, non-physiological artifact interference is eliminated for each region to be verified.

[0058] Specifically, by searching frame by frame in the time series image based on the current region, when the change index of the region first falls below the preset resting threshold, it is determined whether the region is in a state of change, so as to determine the start time point and end time point of the change.

[0059] Then, based on the start and end times of the change, a second parameter is calculated. This second parameter is a comprehensive evaluation index for the change event, used to quantify the physical characteristics of a single change event defined by the start and end times of the change. For example, it could be the peak displacement or average strain rate of the region during the duration, or simply the duration of the change event.

[0060] Furthermore, the second parameter is compared with the second preset threshold. The second preset threshold is used to judge the significance of a single change event from the perspective of physical characteristics. If the second parameter satisfies the preset judgment relationship, the change event is determined as a candidate event, indicating that the event has preliminary physical significance and needs to be confirmed for time continuity. In order to further eliminate isolated noise and confirm the rhythm of uterine contractions, the candidate event needs to be confirmed for time continuity in order to finally generate effective parameter labels.

[0061] Specifically, the steps for generating effective parameter labels include: counting the number of frames in a continuous time frame sequence in which a candidate event at a certain spatial location is continuously identified, defining this number of frames as a significance score, which represents the persistence and stability of the changing event.

[0062] Furthermore, a reference threshold is set, and the significance score is compared with the reference threshold. When the significance score of a candidate event falls within a preset numerical range, a valid parameter label is generated. Here, a numerical range is used as a reference threshold to compare with the significance score in order to determine whether the duration of the event is within a reasonable physiological range.

[0063] Specifically, it is determined whether the significance score is within the numerical range. If the significance score is within the numerical range, a valid parameter label is generated, and the corresponding muscle change area is determined as the valid change area to represent the real uterine contraction event. Its duration should be within a reasonable physiological range.

[0064] The numerical range includes a minimum threshold and a maximum threshold. The minimum threshold defines the lower limit of the numerical range to filter out random noise that is too short in duration or artifacts caused by other body movements.

[0065] The highest threshold defines the upper limit of the numerical range to exclude long-term artifacts caused by non-physiological factors such as continuous probe pressure. When the significance score is between the lower and upper limits, a valid parameter label is generated. If the significance score is greater than the highest threshold, the corresponding data is judged as abnormal data and removed from subsequent calculations. This two-way constraint filtering strategy improves the specificity of recognition.

[0066] Furthermore, after completing all the above screening and verification steps, the final uterine contraction status determination process is initiated. By identifying one or more sets of valid parameter markers that appear consecutively in the time dimension, the second parameter corresponding to the consecutive valid parameter markers is determined as the final valid parameter, constituting a fully verified quantitative indicator of a complete uterine contraction event.

[0067] Furthermore, the steps for determining the state of uterine contractions include: summarizing all the second parameters that were determined to be the final effective parameters within the same time period, and obtaining a comprehensive uterine contraction assessment value through calculation rules such as weighted average or taking the maximum value, so as to quantify the overall intensity of the current uterine contractions;

[0068] The contraction assessment value is compared with the preset contraction judgment threshold to determine the contraction status. If the contraction assessment value is greater than the preset contraction judgment threshold, it is determined that a contraction has occurred.

[0069] Furthermore, once the state of uterine contractions is determined, the state of uterine contractions is determined to be either the pre-contraction stage or the mid-contraction stage. This determination is achieved by analyzing the time series trend of the uterine contraction assessment values.

[0070] Specifically, by calculating the rate of change of uterine contraction assessment values ​​over time;

[0071] When the rate of change is greater than the preset positive slope threshold, it is determined to be the stage before the onset of uterine contractions, marking the period of rapid increase in the intensity of uterine contractions, and an early warning can be issued accordingly.

[0072] When the rate of change is close to zero or enters a stable fluctuation range, it can be determined as the middle stage of uterine contractions, indicating that the intensity of uterine contractions has reached its peak or is in a plateau phase.

[0073] Example 2

[0074] like Figure 2 As shown, this embodiment discloses a uterine contraction recognition system based on ultrasound images. The system uses its internal processor to execute instructions stored in its memory to drive the collaborative operation of various functional modules, specifically including the following modules:

[0075] The image acquisition and processing module acquires and records ultrasound image sequences of muscle tissue and identifies areas of muscle change. Specifically, it includes: driving the ultrasound device to perform continuous dynamic scanning to acquire image sequences; executing a set of image processing rules to determine muscle regions; establishing a stable reference system and marking the tracking location points through coordinate transformation and template registration; and calculating the displacement of the tracking location points between two time points by executing a set of tracking processing steps, and identifying areas of muscle change based on the coordinated displacement of a set of adjacent location points.

[0076] The threshold setting module sets adaptive evaluation thresholds and other thresholds required in the management method. It dynamically calculates and updates adaptive evaluation thresholds by continuously statistically analyzing the degree of regional changes in recent muscle change areas. It is also responsible for storing and providing a first preset threshold, a second preset threshold, a reference threshold as a numerical range, and a preset uterine contraction judgment threshold.

[0077] The region to be evaluated identification module identifies the region to be evaluated and calculates its first parameter. By comparing the pixel value difference with a first preset threshold within the muscle change region, the region to be evaluated is selected, and the average or maximum value of the pixel value difference within each region to be evaluated is calculated as the first parameter output.

[0078] The region to be verified module filters regions to be verified based on the significance of changes. It compares the first parameter output by the region to be evaluated module with the adaptive evaluation threshold provided by the threshold setting module, and determines the regions to be evaluated that meet the preset judgment relationship as regions to be verified, and passes them to subsequent modules for processing.

[0079] The second parameter calculation module calculates the time characteristic parameters of the region to be verified. For each region to be verified, it determines its duration by retrieving the start and end points of its change in the time series and calculates a second parameter that can comprehensively evaluate the change event, such as peak displacement or average strain rate.

[0080] The validity judgment module performs the final validity confirmation of the changing event and generates a valid parameter label. By comparing the second parameter with the second preset threshold, events that meet the conditions are identified as candidate events. Then, the significance score is calculated by counting the number of consecutive occurrence frames of the candidate events. The significance score is then compared with a numerical range containing a lower limit and an upper limit. Only when the score falls into the range is a valid parameter label generated, thereby completing the dual verification of the physiological authenticity of the event.

[0081] The contraction status judgment module summarizes all valid information, finally determines the contraction status and its stage, identifies continuous valid parameter markers and extracts the corresponding final valid parameters; by summarizing and calculating these final valid parameters, a contraction assessment value is obtained; then the contraction assessment value is compared with the preset contraction judgment threshold to determine the contraction status, and further, based on the time change rate of the contraction assessment value, it is determined whether it is the pre-contraction stage or the middle stage of contraction.

[0082] The interactive output module presents the analysis results to the user and receives the output from the contraction status judgment module, including the contraction status, contraction assessment value, and contraction stage. The results are displayed in real time through a graphical user interface in the form of charts, numerical values, or status indicator lights, or sound or light warning signals are triggered when the contraction is determined to be in the pre-contraction stage for medical staff to refer to.

[0083] In summary, this embodiment provides an accurate technical solution for intelligent monitoring of uterine contractions through the coordinated operation of the above-mentioned functional modules and the automated processing flow from original image input to final diagnostic conclusion output.

[0084] The above are merely preferred embodiments of this application and are not intended to limit this application. For those skilled in the art, this application can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method of contraction recognition based on an ultrasonic image, characterized by, The method comprises the following steps: obtaining an ultrasound image sequence of a target muscle tissue, the ultrasound image sequence comprising at least a first time point ultrasound image and a second time point ultrasound image; performing a comparison process based on the obtained ultrasound image sequence, and identifying a muscle change region; when the muscle change region meets a preset trigger condition, performing a contraction state determination process; wherein the contraction state determination process comprises: calculating a contraction evaluation value based on the muscle change region, and determining a contraction state according to a preset judgment relationship between the contraction evaluation value and a preset contraction judgment threshold value; wherein the muscle change region meeting the preset trigger condition comprises: the muscle change region being greater than a first preset threshold value; wherein calculating the contraction evaluation value based on the muscle change region comprises: determining a change start time point and a change end time point corresponding to the muscle change region; calculating a second parameter based on the change start time point and the change end time point; determining the contraction evaluation value based on the second parameter; wherein the method further comprises: identifying a muscle region in the first time point ultrasound image, determining an edge position of the muscle region, and taking the edge position as a to-be-tracked position point for identifying the muscle change region.

2. The ultrasonographic contraction identification method according to claim 1, characterized by, The method further comprises: comparing the second parameter with a second preset threshold value to obtain a final effective parameter, and determining the contraction evaluation value based on the final effective parameter. 3.The uterine contraction identification method based on an ultrasound image according to claim 1, characterized by, The method further comprises: aligning the first time point ultrasound image with a standard template image.

4. A contraction identification system based on ultrasound images, characterized in that, The method comprises the following modules: an image acquisition and processing module, configured to obtain at least a first time point ultrasound image and a second time point ultrasound image of a muscle tissue, and identify a muscle change region based on the images; a to-be-evaluated region identification module, configured to identify a to-be-evaluated region and calculate a first parameter thereof, and filter out the to-be-evaluated region by comparing a pixel value difference with a first preset threshold value in the muscle change region; a to-be-verified region determination module, configured to filter a to-be-verified region according to change saliency; a second parameter calculation module, configured to calculate a time characteristic parameter of the to-be-verified region; an effectiveness judgment module, configured to finally confirm the effectiveness of a change event and generate an effective parameter label; a contraction state judgment module, configured to calculate a contraction evaluation value based on the muscle change region when the muscle change region meets a preset trigger condition, and determine a contraction state according to a preset judgment relationship between the contraction evaluation value and a preset contraction judgment threshold value; and determine a change start time point and a change end time point corresponding to the muscle change region, calculate a second parameter based on the change start time point and the change end time point, and determine the contraction evaluation value based on the second parameter; an interactive output module, configured to present an analysis result to a user and receive an output from the contraction state judgment module. The preset trigger condition is that the muscle change region is greater than a first preset threshold value. ​ 5.The uterine contraction identification system based on ultrasound images according to claim 4, characterized in that, ​ 6.The uterine contraction identification system based on ultrasound images according to claim 4, characterized in that, The uterine contraction state judgment module is further configured to compare the second parameter with a second preset threshold to obtain a final effective parameter, and determine a uterine contraction evaluation value based on the final effective parameter.

7. An ultrasound image-based contraction recognition terminal, characterized by, The method comprises a memory and a processor, the memory stores a computer program, and the processor is configured to execute the computer program to implement the method for identifying uterine contraction based on an ultrasound image according to any one of claims 1 to 3.

8. Storage medium, characterized in that The computer program is stored on the memory and is executed by the processor to implement the method for identifying uterine contraction based on an ultrasound image according to any one of claims 1 to 3.

Images