An ultrasound image enhancement and motion analysis system for real-time uterine contraction detection

Abstract

This invention relates to the field of image enhancement technology, specifically to an ultrasound image enhancement and motion analysis system for real-time uterine contraction detection. The system includes: acquiring ultrasound images of different frames for uterine contraction detection; determining uterine contraction assessment feature values ​​based on the gradient direction and position distribution of image corner points; selecting target feature points; determining continuity coefficients based on the similarity analysis of motion vectors in the neighborhood of the target feature points; performing clustering based on the continuity coefficients; analyzing the temporal trajectory of uterine contraction changes within different clusters; obtaining the uterine contraction pattern change value for each target feature point; obtaining the enhancement adjustment coefficient for each target feature point based on the uterine contraction pattern change value and the uterine contraction assessment feature value; and performing image enhancement on the ultrasound images of uterine contraction detection based on the enhancement adjustment coefficients. This invention enables adaptive image enhancement, allowing the enhanced image to more clearly display the motion trajectory of uterine contractions and improving imaging results.

Description

Technical Field

[0001] This invention relates to the field of image enhancement technology, and more specifically to an ultrasound image enhancement and motion analysis system for real-time uterine contraction detection. Background Technology

[0002] Uterine contraction monitoring plays a crucial role in obstetric medicine, especially in monitoring contractions during childbirth. Ultrasound, as a non-invasive and painless imaging technique, is widely used in prenatal examinations and real-time monitoring of uterine contractions during labor. Ultrasound images, by scanning the uterus, can display the dynamic changes in the uterine wall during contractions, thereby determining key information such as the frequency, intensity, and duration of contractions.

[0003] Current methods for detecting uterine contractions require the use of a TOCO probe. However, the TOCO probe is easily affected by the patient's position. When the patient breathes, the up-and-down movement of the abdomen causes the TOCO probe to change position, resulting in motion artifacts in the ultrasound image. The presence of motion artifacts makes the image blurry. Using a uniform image enhancement method will result in areas that need to be focused on not being effectively enhanced, making it impossible to clearly show the trajectory of uterine contractions and resulting in poor imaging quality. Summary of the Invention

[0004] To address the technical problem that the use of a uniform image enhancement method in related technologies leads to the ineffective enhancement of areas requiring special attention, resulting in unclear display of the uterine contraction trajectory and poor imaging quality, this invention provides an ultrasound image enhancement and motion analysis system for real-time uterine contraction detection. The specific technical solution adopted is as follows:

[0005] This invention proposes an ultrasound image enhancement and motion analysis system for real-time uterine contraction detection, the system comprising:

[0006] The acquisition module is used to acquire ultrasound images of different frames for uterine contraction detection; based on the gradient direction and position distribution of other edge pixels in the preset neighborhood of each corner point in the ultrasound image, the uterine contraction evaluation feature value of each corner point for contrast is determined.

[0007] The motion vector analysis module is used to filter out target feature points using uterine contraction assessment feature values; based on the similarity of motion vectors between each target feature point and other pixels in the preset neighborhood, the continuity coefficient of motion change of each target feature point in the neighborhood is obtained;

[0008] The motion trajectory analysis module is used to cluster based on the continuity coefficient to obtain target feature point clusters; compare and analyze the uterine contraction change trajectory of target feature points in different target feature point clusters over time, and obtain the uterine contraction pattern change value of each target feature point;

[0009] The enhancement module is used to obtain the enhancement adjustment coefficient for each target feature point based on the uterine contraction pattern change value and the uterine contraction assessment feature value; and to perform image enhancement on the ultrasound image of uterine contraction detection based on the enhancement adjustment coefficient.

[0010] Furthermore, based on the gradient direction and positional distribution of other edge pixels within a preset neighborhood of each corner point in the ultrasound image, the uterine contraction assessment feature value for each corner point relative to contrast is determined, including:

[0011] The FAST corner detection algorithm is used to identify corners in ultrasound images; Canny edge detection is performed within a preset neighborhood centered on the corner to obtain edge pixels.

[0012] Determine the sine of the angle between the gradient direction of each edge pixel and the horizontal direction to obtain the direction index;

[0013] The contrast significance coefficient of each edge pixel is determined based on the differences in Euclidean distance and orientation indices between edge pixels.

[0014] The proportion of edge pixels with a significance coefficient greater than a preset significance threshold within a preset neighborhood is used as the corner point's uterine contraction assessment feature value.

[0015] Furthermore, based on the differences in Euclidean distance and orientation indices between edge pixels, the contrast significance coefficient of each edge pixel is determined, including:

[0016] The mean of the Euclidean distances between any edge pixel and other edge pixels is used as the edge distance feature value of the edge pixel.

[0017] The mean of the absolute values ​​of the differences between the direction indices of any edge pixel and other edge pixels is taken as the edge direction feature value of the edge pixel.

[0018] Calculate the product of the edge distance feature value and the edge direction feature value of the same edge pixel, and normalize the negative of the product value as the contrast significance coefficient.

[0019] Furthermore, target feature points are selected using uterine contraction assessment feature values, including:

[0020] Corner points whose uterine contraction assessment feature values ​​are greater than the preset assessment threshold are used as target feature points.

[0021] Furthermore, based on the similarity of motion vectors between each target feature point and other pixels within a preset neighborhood, the continuity coefficients of motion changes of each target feature point within the neighborhood are obtained, including:

[0022] The feature matching algorithm is used to determine the matching pixels in the ultrasound image of any frame with the previous frame; the motion distance and direction of the matching pixels in time are used to form the motion vector of the pixel in the corresponding frame.

[0023] Determine the cosine similarity between the matching target feature point and the motion vectors of other pixels in the preset neighborhood;

[0024] The negative of the distance is normalized and used as the distance weight. The product of the distance weight and the cosine similarity is calculated and used as the similarity coefficient between the matched target feature point and other pixels in the preset neighborhood.

[0025] The mean of the similarity coefficients between the target feature point and all other pixels in the preset neighborhood is calculated and normalized to obtain continuous coefficients.

[0026] Furthermore, clustering is performed based on the continuity coefficients to obtain clusters of target feature points, including:

[0027] Based on the DBSCANS clustering algorithm, density clustering of target feature points is performed according to the continuity coefficient to obtain target feature point clusters.

[0028] Furthermore, by comparing and analyzing the temporal trajectories of uterine contraction changes in target feature points within different target feature point clusters, the uterine contraction pattern change values ​​for each target feature point are obtained, including:

[0029] For any target feature point in an ultrasound image, the inverse optical flow method is used to analyze its image coordinate position in the previous preset number of ultrasound images, and the image coordinates are arranged in time sequence to obtain the trajectory sequence of the target feature point.

[0030] Based on the difference in the change trajectory sequence between any target feature point and other target feature points in the same target feature point cluster, the change value of the uterine contraction pattern of the target feature point is determined.

[0031] Furthermore, based on the difference in the change trajectory sequence between any target feature point and other target feature points in the same cluster, the uterine contraction pattern change value of the target feature point is determined, including:

[0032] The Dynamic Time Warping (DTW) value of the trajectory sequence of any target feature point and other target feature points is calculated using the Dynamic Time Warping (DTW) algorithm.

[0033] The average DTW value of all other target feature points in the same target feature point cluster is normalized and used as the change value of uterine contraction pattern.

[0034] Furthermore, based on the uterine contraction pattern change values ​​and uterine contraction assessment feature values ​​of the target feature points, the enhancement adjustment coefficient for each target feature point is obtained, including:

[0035] Calculate the product of the uterine contraction pattern change value and the uterine contraction assessment feature value for the same target feature point, and normalize it as the enhancement adjustment coefficient.

[0036] Furthermore, image enhancement is performed on ultrasound images used for uterine contraction detection based on an enhancement adjustment coefficient, including:

[0037] In the ultrasound image at the current moment, the region growing algorithm is used to grow regions with all target feature points as the center point to obtain different uterine contraction regions;

[0038] The maximum value of the enhancement adjustment coefficient of all target feature points in each uterine contraction region is used as the enhancement factor for each uterine contraction region;

[0039] Using the adaptive Laplacian sharpening algorithm, the sharpening intensity of each pixel not located in the uterine contraction area is set to 0, while the sharpening intensity of pixels located in the uterine contraction area is used as an enhancement factor to perform deblurring enhancement processing on the ultrasound image at the current moment.

[0040] The present invention has the following beneficial effects:

[0041] In this embodiment of the invention, corner point analysis is performed on ultrasound images. Based on the gradient direction and positional distribution of pixels surrounding each corner point, a uterine contraction assessment feature value is determined for each corner point belonging to the uterine contraction region. A higher uterine contraction assessment feature value indicates a stronger match to the characteristics of the uterine contraction region. Then, target feature points are selected based on these uterine contraction assessment feature values. Combined with motion vector analysis, a continuity coefficient is determined. The target feature points are then clustered based on this continuity coefficient. The clustering results are used to analyze the uterine contraction change trajectory within different clusters, yielding a uterine contraction pattern change value. This value represents the difference in the change pattern between the target feature point and all other target feature points in the same cluster. Based on the uterine contraction pattern change value and the uterine contraction assessment feature value, an enhancement adjustment coefficient is obtained for each target feature point. Finally, image enhancement is performed on the ultrasound images used for uterine contraction detection based on this enhancement adjustment coefficient. In summary, by analyzing the differences between target feature points that exhibit the same pattern of uterine contraction changes, this embodiment of the invention can identify abnormalities or changes during uterine contractions and assign them higher enhancement adjustment coefficients to achieve adaptive image enhancement. This prioritizes improving the visibility of these target feature points during image enhancement, enabling the enhanced image to more clearly display the trajectory of uterine contractions and improve imaging results. Attached Figure Description

[0042] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a structural diagram of an ultrasound image enhancement and motion analysis system for real-time uterine contraction detection provided in one embodiment of the present invention;

[0044] Figure 2 This is a schematic diagram illustrating the changes in the uterine contraction region in different frames, provided as an embodiment of the present invention. Detailed Implementation

[0045] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an ultrasound image enhancement and motion analysis system for real-time uterine contraction detection proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0046] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0047] The following description, in conjunction with the accompanying drawings, details the specific scheme of the ultrasound image enhancement and motion analysis system for real-time uterine contraction detection provided by this invention.

[0048] Please see Figure 1 The diagram illustrates a structural diagram of an ultrasound image enhancement and motion analysis system for real-time uterine contraction detection provided by an embodiment of the present invention, including: an acquisition module 101, a motion vector analysis module 102, a motion trajectory analysis module 103, and an enhancement module 104.

[0049] The acquisition module 101 is used to acquire ultrasound images of different frames of uterine contraction detection; based on the gradient direction and position distribution of other edge pixels in the preset neighborhood of each corner point in the ultrasound image, the uterine contraction assessment feature value of each corner point belonging to the uterine contraction region is determined.

[0050] Uterine contraction monitoring plays a crucial role in obstetric medicine, especially in monitoring contractions during childbirth. Ultrasound, as a non-invasive and painless imaging technique, is widely used in prenatal examinations and real-time monitoring of uterine contractions during labor. Ultrasound images, by scanning the uterus, can display the dynamic changes in the uterine wall during contractions, thereby determining key information such as the frequency, intensity, and duration of contractions.

[0051] Current methods for detecting uterine contractions require the use of a TOCO probe. However, the TOCO probe is easily affected by the patient's position. When the patient breathes, the up-and-down movement of the abdomen causes the TOCO probe to change position, resulting in motion artifacts in the ultrasound image. The presence of motion artifacts makes the image blurry. Using a uniform image enhancement method will result in areas that need to be focused on not being effectively enhanced, and the movement trajectory of uterine contractions cannot be clearly displayed, resulting in poor imaging quality.

[0052] By analyzing the differences between target feature points that exhibit the same pattern of uterine contraction changes, we can identify which target feature points are abnormal or change during uterine contractions. We can then adaptively enhance the image, prioritizing the improvement of the visibility of these target feature points during the enhancement process. This allows the enhanced image to more clearly display the trajectory of uterine contractions.

[0053] In this embodiment of the invention, the patient lies flat on the examination bed, and a high-resolution TOCO ultrasound probe is used to perform an M-mode ultrasound scan to acquire ultrasound video during the uterine contraction detection process. The video is then divided into frames to obtain ultrasound images of different frames.

[0054] Because the TOCO probe is easily affected by the patient's position, abdominal movements during breathing can cause motion artifacts in the ultrasound image, making it difficult to accurately identify the contraction area in the ultrasound image. During contractions, key areas of focus typically include local changes within the contraction area and the corners of uterine contractions, as these play a crucial role in assessing uterine contractions, uterine morphology, and related pathological conditions. See also Figure 2 , Figure 2 This is a schematic diagram illustrating the changes in the uterine contraction region in different frames, provided as an embodiment of the present invention.

[0055] In ultrasound images, the ultrasound echo intensity varies greatly in the uterine contraction area due to changes in blood flow and tissue tension. Therefore, target feature points within the uterine contraction area have high contrast or significant texture.

[0056] Furthermore, in some embodiments of the present invention, the uterine contraction assessment feature value of each corner point belonging to the uterine contraction region is determined based on the gradient direction and positional distribution of other edge pixels in a preset neighborhood of each corner point in the ultrasound image. This includes: using the FAST corner detection algorithm to determine the corner points in the ultrasound image; performing Canny edge detection within a preset neighborhood centered on the corner point to obtain edge pixels; determining the sine value of the angle between the gradient direction of each edge pixel and the horizontal direction to obtain a direction index; determining the contrast significance coefficient of each edge pixel based on the Euclidean distance between edge pixels and the difference in the direction index; and using the proportion of edge pixels with a contrast significance coefficient greater than a preset significance threshold in the preset neighborhood as the uterine contraction assessment feature value of the corner point.

[0057] The FAST corner detection algorithm is used to detect corners that change significantly in an image. These corners are usually located in regions of the image with high structural information. Therefore, in this embodiment of the invention, feature analysis can be performed within a preset neighborhood centered on the corner.

[0058] The preset neighborhood can be specifically, for example, a 21*21 window area constructed with the corner point as the center, and the Canny edge detection algorithm is used to obtain all edge pixels within the preset neighborhood of the corner point.

[0059] It should be noted that when the edge direction consistency of edge pixels is poor and the distribution of edge pixels is more discrete, it indicates that there are relatively blurry texture changes in the preset neighborhood of the corner point and the image structure within the preset neighborhood is not dense enough. These changes may reduce the salience of the corner point in the image. Conversely, when the edge direction consistency of edge pixels is strong and the distribution is more concentrated, it indicates that the preset neighborhood has a clearer edge structure, which can highlight the contrast of the corner point.

[0060] Therefore, in this embodiment of the invention, significance analysis is achieved through directional indicators and Euclidean distance to obtain the comparative significance coefficient.

[0061] Furthermore, in some embodiments of the present invention, determining the contrast significance coefficient of each edge pixel based on the differences in Euclidean distance and orientation indices between edge pixels includes: taking the mean of the Euclidean distance between any edge pixel and other edge pixels as the edge distance feature value of the edge pixel; taking the mean of the absolute values ​​of the differences in orientation indices between any edge pixel and other edge pixels as the edge orientation feature value of the edge pixel; calculating the product of the edge distance feature value and the edge orientation feature value of the same edge pixel, and normalizing the negative of the product value as the contrast significance coefficient.

[0062] Among them, the smaller the edge distance feature value, the smaller the distance between edge pixels and the more concentrated the distribution; the smaller the edge direction feature value, the more consistent the gradient direction of edge pixels. Therefore, the smaller the edge distance feature value and the edge direction feature value, the higher the significance. Calculate the product of the edge distance feature value and the edge direction feature value, and normalize the negative of the product value to obtain the comparison significance coefficient.

[0063] Understandably, the larger the contrast significance coefficient of the corner point, the more likely the corner point is located in a region with clear and stable signals in the ultrasound image, thus effectively avoiding interference from low-quality regions (such as blood vessels) on the results.

[0064] The preset saliency threshold is a threshold value for the contrast saliency coefficient. In this embodiment of the invention, the preset saliency threshold can be set to 0.3, and edge pixels with a contrast saliency coefficient greater than 0.3 are regarded as points with both high saliency and high contrast.

[0065] The proportion of all pixels within a preset neighborhood is used as the corner point's contraction assessment feature value. The higher the corner point's contraction assessment feature value, the more obvious the contrast and significance within the preset neighborhood, the more it matches the characteristics of the contraction area, and the more likely the corner point is to be located in a clear and stable contraction area in the ultrasound image.

[0066] The motion vector analysis module 102 is used to filter out target feature points using uterine contraction assessment feature values; and to obtain the continuity coefficient of motion change of each target feature point in the neighborhood based on the similarity of motion vectors between each target feature point and other pixels in the preset neighborhood.

[0067] In conjunction with the above description that "the higher the value of the corner point's uterine contraction assessment feature, the more obvious the contrast and significance within the preset neighborhood, the more it matches the characteristics of the uterine contraction area, and the more likely the corner point is to be located in a clear and stable uterine contraction area in the ultrasound image," it can be concluded that the corner point is likely located in a high signal-to-noise ratio area in the ultrasound image. This area has high contrast or significant texture and forms many stable and significant edge points, which is consistent with the characteristics of the uterine contraction area. Therefore, this embodiment of the invention filters it to obtain target feature points. The target feature points represent points located in the uterine contraction area, which are used to achieve a clearer and more reliable image analysis in the future.

[0068] Furthermore, in some embodiments of the present invention, the selection of target feature points using uterine contraction assessment feature values ​​includes: selecting corner points whose uterine contraction assessment feature values ​​are greater than a preset assessment threshold as target feature points.

[0069] The preset evaluation threshold is a threshold value for the uterine contraction evaluation feature value. In this embodiment of the invention, the preset evaluation threshold can be, for example, 0.5, and there is no limitation thereto.

[0070] Since uterine contractions have a certain regularity, and the target feature points are located in the contraction region, it indicates that the motion changes between these feature points and the pixels in their neighborhood should have a certain regularity and continuity. Therefore, by analyzing the similarity in direction between the motion vectors of each target feature point and other pixels in its neighborhood, the continuity of motion changes in the neighborhood of each target feature point can be obtained.

[0071] Further, in some embodiments of the present invention, based on the similarity of motion vectors between each target feature point and other pixels in a preset neighborhood, the continuous coefficient of motion change of each target feature point in the neighborhood is obtained, including: determining the matching pixels in any frame of the ultrasound image with the previous frame based on a feature matching algorithm; constructing the motion vector of the corresponding pixel in the frame by the temporal motion distance and direction of the matching pixels; determining the cosine similarity between the motion vectors of the matching target feature point and other pixels in the preset neighborhood; normalizing the negative of the Euclidean distance between the target feature point and other pixels in the preset neighborhood as a distance weight, calculating the product of the distance weight and the cosine similarity as the similarity coefficient between the matching target feature point and other pixels in the preset neighborhood; calculating the mean of the similarity coefficients between the target feature point and all other pixels in the preset neighborhood, and normalizing it to obtain the continuous coefficient.

[0072] The feature matching algorithm can be specific to either FLANN or Brute-Force Matcher, both of which are well-known technologies and will not be elaborated upon. The feature matching algorithm can match target feature points with similar descriptors in adjacent ultrasound images. It should be noted that target feature points that cannot be matched are not further analyzed.

[0073] The motion vectors are constructed, and similarity analysis is performed on the motion vectors using cosine similarity. The greater the cosine similarity between the target feature point and the pixel point, the smaller the difference between the motion vectors and the more continuous the motion vectors are.

[0074] The Euclidean distance between the target feature point and other pixels in the preset neighborhood reflects the spatial distance between the two points. The smaller the distance, the closer the two points are and the more continuous the motion changes may be. Therefore, the negative number of the Euclidean distance is normalized and used as the distance weight.

[0075] The product of distance weight and cosine similarity is calculated as the similarity coefficient between the target feature point and other pixels in the preset neighborhood. The numerical value of the similarity coefficient represents the similarity between the target feature point and other pixels in the preset neighborhood in terms of motion vector. The mean of the similarity coefficients between the target feature point and all other pixels in the preset neighborhood is normalized to obtain continuous coefficients.

[0076] The larger the value of the continuity coefficient, the more continuous the movement and changes of the target feature point in the neighborhood, and the more it conforms to the regularity and continuity characteristics of normal uterine contractions.

[0077] The motion trajectory analysis module 103 is used to cluster based on the continuity coefficient to obtain target feature point clusters; compare and analyze the uterine contraction change trajectory of target feature points in different target feature point clusters over time, and obtain the uterine contraction pattern change value of each target feature point.

[0078] Inconsistent continuity coefficients indicate that they may be under different patterns of change. Therefore, cluster analysis is needed for target feature points that are under the same pattern of change.

[0079] Furthermore, in some embodiments of the present invention, clustering based on continuity coefficients to obtain target feature point clusters includes: using the DBSCANS clustering algorithm to perform density clustering on target feature points based on continuity coefficients to obtain target feature point clusters.

[0080] The DBSCANS clustering algorithm is an unsupervised density clustering method that can perform clustering to obtain clusters of target feature points. Each cluster corresponds to a change pattern during uterine contractions, representing the set of target feature points for each change pattern during uterine contractions.

[0081] Since pathological conditions can alter the local physiological state of the uterus and affect the normal pattern of uterine contractions, it is necessary to focus on these uterine contraction areas that have undergone pattern changes. Each target feature point cluster represents a set of target feature points for each pattern of change during uterine contractions. If the trajectory of a target feature point in the ultrasound image differs significantly from that of other target feature points, it indicates that the uterine contraction area to which that target feature point belongs has undergone pattern changes. Therefore, by comparing and analyzing the uterine contraction trajectories among the target feature points in the cluster, the uterine contraction pattern change value of each target feature point can be obtained.

[0082] Furthermore, in some embodiments of the present invention, the uterine contraction change trajectory of target feature points in different target feature point clusters over time is compared and analyzed to obtain the uterine contraction pattern change value of each target feature point, including: for any target feature point in the ultrasound image, the image coordinate position in the previous preset number of ultrasound images is analyzed using the reverse optical flow method, and the image coordinates are arranged in time sequence to obtain the change trajectory sequence of the target feature point; the uterine contraction pattern change value of the target feature point is determined based on the difference between the change trajectory sequence of any target feature point and other target feature points in the same target feature point cluster.

[0083] The preset number is 20. An image sequence consisting of the 20 most recent ultrasound images preceding any given frame is analyzed using inverse optical flow to determine the image coordinates of the target feature points within those 20 frames. This time-series analysis creates a trajectory sequence containing the positional changes of the target feature points within those 20 frames, which is then used for trajectory analysis.

[0084] Furthermore, in some embodiments of the present invention, determining the uterine contraction pattern change value of a target feature point based on the difference in the change trajectory sequence between any target feature point and other target feature points in the same target feature point cluster includes: using a dynamic time warping algorithm to calculate the DTW value of the change trajectory sequence between any target feature point and other target feature points; and normalizing the DTW value of any target feature point with the average of all other target feature points in the same target feature point cluster as the uterine contraction pattern change value.

[0085] The Dynamic Time Warping (DTW) algorithm is used to analyze the similarity between different time series. The smaller the DTW value, the more similar the fluctuations between the two series. Therefore, in this embodiment of the invention, the mean DTW value of all other target feature points is directly calculated and normalized to obtain the contraction pattern change value. The larger the contraction pattern change value, the more inconsistent the change pattern of the corresponding target feature point is with all other target feature points in the same target feature point cluster. The contraction region to which the target feature point belongs may have its contraction pattern changed due to pathological conditions.

[0086] The enhancement module 104 is used to obtain the enhancement adjustment coefficient of each target feature point based on the uterine contraction pattern change value and the uterine contraction assessment feature value; and to perform image enhancement on the ultrasound image of uterine contraction detection based on the enhancement adjustment coefficient.

[0087] In this embodiment of the invention, the uterine contraction assessment feature value characterizes the probability of being located in the uterine contraction area and the salience of the surrounding texture, while the uterine contraction pattern change value represents the change pattern characteristics of the target feature point and all other target feature points in the same target feature point cluster. The adaptive enhancement adjustment coefficient can be analyzed based on these two parameters.

[0088] Furthermore, in some embodiments of the present invention, the enhancement adjustment coefficient of each target feature point is obtained based on the contraction pattern change value and the contraction assessment feature value of the target feature point, including: calculating the product of the contraction pattern change value and the contraction assessment feature value of the same target feature point, and normalizing it as the enhancement adjustment coefficient.

[0089] A higher value for the uterine contraction assessment feature indicates greater contrast and significance within the preset neighborhood, more closely matching the characteristics of the uterine contraction area, and a higher likelihood of being located in a clear and stable uterine contraction area in the ultrasound image. A larger value for the uterine contraction pattern change value indicates a greater inconsistency between the change pattern of the corresponding target feature point and all other target feature points in the same cluster. This suggests that the uterine contraction pattern of the target feature point may have changed due to pathological abnormalities, further indicating a greater possibility that the regularity of the uterine contraction area belonging to the target feature point has been broken. Therefore, the target feature point deserves higher attention.

[0090] The more abnormal the changes are located in the uterine contraction area, the higher the focus, and the greater the enhancement effect is needed to ensure image clarity. Therefore, in this embodiment of the invention, the product of the uterine contraction pattern change value and the uterine contraction assessment feature value of the same target feature point is calculated and normalized as the enhancement adjustment coefficient.

[0091] Furthermore, in some embodiments of the present invention, image enhancement of the ultrasound image for uterine contraction detection based on the enhancement adjustment coefficient includes: in the ultrasound image at the current moment, using a region growing algorithm to grow regions with all target feature points as the center points to obtain different uterine contraction regions; using the maximum value of the enhancement adjustment coefficient of all target feature points in each uterine contraction region as the enhancement factor for each uterine contraction region; and using an adaptive Laplacian sharpening algorithm to set the sharpening intensity of each pixel point not in the uterine contraction region to 0, and the sharpening intensity of pixels in the uterine contraction region as the enhancement factor, to perform deblurring enhancement processing on the ultrasound image at the current moment.

[0092] In this embodiment of the invention, an eight-neighbor Laplacian operator can be used to obtain the Laplacian operator value for each pixel in the ultrasound image. This is used to perform strong sharpening on the contraction region where the contraction pattern has changed, thereby improving the detail information within the contraction region and reducing the impact of motion blur. The grayscale values ​​of all pixels in the ultrasound image of the current contraction detection, after adaptive Laplacian sharpening, are used to construct the sharpened and enhanced image.

[0093] In this embodiment of the invention, corner point analysis is performed on ultrasound images. Based on the gradient direction and positional distribution of pixels surrounding each corner point, a uterine contraction assessment feature value is determined for each corner point belonging to the uterine contraction region. A higher uterine contraction assessment feature value indicates a stronger match to the characteristics of the uterine contraction region. Then, target feature points are selected based on these uterine contraction assessment feature values. Combined with motion vector analysis, a continuity coefficient is determined. The target feature points are then clustered based on this continuity coefficient. The clustering results are used to analyze the uterine contraction change trajectory within different clusters, yielding a uterine contraction pattern change value. This value represents the difference in the change pattern between the target feature point and all other target feature points in the same cluster. Based on the uterine contraction pattern change value and the uterine contraction assessment feature value, an enhancement adjustment coefficient is obtained for each target feature point. Finally, image enhancement is performed on the ultrasound images used for uterine contraction detection based on this enhancement adjustment coefficient. In summary, by analyzing the differences between target feature points that exhibit the same pattern of uterine contraction changes, this embodiment of the invention can identify abnormalities or changes during uterine contractions and assign them higher enhancement adjustment coefficients to achieve adaptive image enhancement. This prioritizes improving the visibility of these target feature points during image enhancement, enabling the enhanced image to more clearly display the trajectory of uterine contractions and improve imaging results.

[0094] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0095] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A real-time uterine contraction detection ultrasound image enhancement and motion analysis system, characterized in that, The system includes: The acquisition module is used to acquire ultrasound images of different frames for uterine contraction detection; based on the gradient direction and position distribution of other edge pixels in the preset neighborhood of each corner point in the ultrasound image, the uterine contraction assessment feature value of each corner point belonging to the uterine contraction region is determined. The motion vector analysis module is used to filter out target feature points using uterine contraction assessment feature values; based on the similarity of motion vectors between each target feature point and other pixels in the preset neighborhood, the continuity coefficient of motion change of each target feature point in the neighborhood is obtained; The motion trajectory analysis module is used to cluster based on the continuity coefficient to obtain target feature point clusters; compare and analyze the uterine contraction change trajectory of target feature points in different target feature point clusters over time, and obtain the uterine contraction pattern change value of each target feature point; The enhancement module is used to obtain the enhancement adjustment coefficient for each target feature point based on the uterine contraction pattern change value and the uterine contraction assessment feature value; and to perform image enhancement on the ultrasound image of uterine contraction detection based on the enhancement adjustment coefficient. The step of determining the uterine contraction assessment feature value of each corner point belonging to the uterine contraction region based on the gradient direction and position distribution of other edge pixels within a preset neighborhood of each corner point in the ultrasound image includes: The FAST corner detection algorithm is used to identify corners in ultrasound images; Canny edge detection is performed within a preset neighborhood centered on the corner to obtain edge pixels. Determine the sine of the angle between the gradient direction of each edge pixel and the horizontal direction to obtain the direction index; The contrast significance coefficient of each edge pixel is determined based on the differences in Euclidean distance and orientation indices between edge pixels. The proportion of edge pixels with a contrast significance coefficient greater than a preset significance threshold within a preset neighborhood is used as the corner point's uterine contraction assessment feature value.

2. The ultrasound image enhancement and motion analysis system for real-time uterine contraction detection as described in claim 1, characterized in that, Based on the differences in Euclidean distance and orientation indices between edge pixels, the contrast significance coefficient of each edge pixel is determined, including: The mean of the Euclidean distances between any edge pixel and other edge pixels is used as the edge distance feature value of the edge pixel. The mean of the absolute values ​​of the differences between the direction indices of any edge pixel and other edge pixels is taken as the edge direction feature value of the edge pixel. Calculate the product of the edge distance feature value and the edge direction feature value of the same edge pixel, and normalize the negative of the product value as the contrast significance coefficient.

3. The ultrasound image enhancement and motion analysis system for real-time uterine contraction detection as described in claim 1, characterized in that, Target feature points were selected using uterine contraction assessment feature values, including: Corner points whose uterine contraction assessment feature values ​​are greater than a preset assessment threshold are used as target feature points.

4. The ultrasound image enhancement and motion analysis system for real-time uterine contraction detection as described in claim 1, characterized in that, Based on the similarity of motion vectors between each target feature point and other pixels in a preset neighborhood, the continuity coefficients of motion changes of each target feature point in the neighborhood are obtained, including: The feature matching algorithm is used to determine the matching pixels in the ultrasound image of any frame with the previous frame; the motion distance and direction of the matching pixels in time are used to form the motion vector of the pixel in the corresponding frame. Determine the cosine similarity between the matching target feature point and the motion vectors of other pixels in the preset neighborhood; The negative of the Euclidean distance between the target feature point and other pixels in the preset neighborhood is normalized and used as the distance weight. Calculate the product of distance weight and cosine similarity, and use it as the similarity coefficient between the matched target feature point and other pixels in the preset neighborhood; The mean of the similarity coefficients between the target feature point and all other pixels in the preset neighborhood is calculated and normalized to obtain continuous coefficients.

5. The ultrasound image enhancement and motion analysis system for real-time uterine contraction detection as described in claim 1, characterized in that, Clustering based on continuous coefficients yields clusters of target feature points, including: Based on the DBSCANS clustering algorithm, density clustering of target feature points is performed according to the continuity coefficient to obtain target feature point clusters.

6. The ultrasound image enhancement and motion analysis system for real-time uterine contraction detection as described in claim 1, characterized in that, By comparing and analyzing the temporal trajectories of uterine contraction changes in target feature points within different target feature point clusters, the contraction pattern change values ​​for each target feature point are obtained, including: For any target feature point in an ultrasound image, the inverse optical flow method is used to analyze its image coordinate position in the previous preset number of ultrasound images, and the image coordinates are arranged in time sequence to obtain the change trajectory sequence of the target feature point; Based on the difference in the change trajectory sequence between any target feature point and other target feature points in the same target feature point cluster, the change value of the uterine contraction pattern of the target feature point is determined.

7. The ultrasound image enhancement and motion analysis system for real-time uterine contraction detection as described in claim 6, characterized in that, Based on the difference in the trajectory sequence between any target feature point and other target feature points in the same cluster, the change value of the uterine contraction pattern of the target feature point is determined, including: The Dynamic Time Warping (DTW) value of the trajectory sequence of any target feature point and other target feature points is calculated using the Dynamic Time Warping (DTW) algorithm. The average DTW value of all other target feature points in the same target feature point cluster is normalized and used as the change value of uterine contraction pattern.

8. The ultrasound image enhancement and motion analysis system for real-time uterine contraction detection as described in claim 1, characterized in that, Based on the changes in uterine contraction patterns and uterine contraction assessment feature values ​​of the target feature points, the enhancement adjustment coefficient for each target feature point is obtained, including: Calculate the product of the uterine contraction pattern change value and the uterine contraction assessment feature value for the same target feature point, and normalize it as the enhancement adjustment coefficient.

9. The ultrasound image enhancement and motion analysis system for real-time uterine contraction detection as described in claim 1, characterized in that, Image enhancement of ultrasound images used for uterine contraction detection based on enhancement adjustment coefficients includes: In the ultrasound image at the current moment, the region growing algorithm is used to grow regions with all target feature points as the center point to obtain different uterine contraction regions; The maximum value of the enhancement adjustment coefficient of all target feature points in each uterine contraction region is used as the enhancement factor for each uterine contraction region; Using the adaptive Laplacian sharpening algorithm, the sharpening intensity of each pixel not located in the uterine contraction area is set to 0, while the sharpening intensity of pixels located in the uterine contraction area is used as an enhancement factor to perform deblurring enhancement processing on the ultrasound image at the current moment.

Images