Ultrasonic imaging method based on image fusion

Abstract

The invention discloses an ultrasonic imaging method based on image fusion, and belongs to the technical field of assisted reproduction, and the method comprises the steps: respectively collecting image data of a uterus through uterus vagina ultrasound and abdominal ultrasound; preprocessing the acquired uterus and vagina ultrasonic image and the abdominal ultrasonic image; performing feature extraction and feature matching on the two preprocessed images; and carrying out transformation alignment on the two images after feature extraction and feature matching, carrying out weight distribution on the two aligned images according to the fusion weight, and fusing the two aligned images into a comprehensive image. According to the scheme, the two ultrasonic images are fused, the high-resolution characteristic of uterus and vagina ultrasound is utilized, the limitation of a single imaging technology is overcome, the wide view field of abdominal ultrasound is combined, and more comprehensive anatomical information is provided. By providing a clearer and more comprehensive ultrasonic imaging scheme, the definition of the abdominal ultrasonic image is improved, so that the evaluation of the endometrium and embryo implantation position is more accurate.

Description

Technical Field

[0001] This application relates to the field of assisted reproductive technology, and particularly to an ultrasonic imaging method based on image fusion. Background Art

[0002] In assisted reproductive technology, the success rate of embryo transfer is affected by various factors, and the application of ultrasonic imaging technology is crucial. Although traditional abdominal ultrasound images can provide basic information about the uterus, they are often insufficient in terms of details and clarity. Especially when evaluating the state of the endometrium and the location of embryo implantation, misjudgment or missed judgment may occur. Transvaginal ultrasound (ultrasound through the vagina) is widely used in gynecological examinations and can provide higher-resolution images, clearly showing the uterine structure and its surrounding tissues. However, there are also certain limitations in using only transvaginal ultrasound images (uterine images obtained by transvaginal ultrasound). For example, in some cases, the comfort of the patient and the acceptability of the examination may be affected. Therefore, how to effectively combine these two ultrasonic imaging technologies to improve the image quality and diagnostic accuracy during embryo transfer has become a major challenge in the current technical field. Summary of the Invention

[0003] This application aims to overcome the defects of the above-mentioned prior art and provides an ultrasonic imaging method based on image fusion, which can effectively combine the two ultrasonic imaging technologies of abdominal ultrasound and transvaginal ultrasound to improve the image quality and diagnostic accuracy during embryo transfer.

[0004] To achieve the above object, the technical solution of this application is: An ultrasonic imaging method based on image fusion, including: Collecting image data of the uterus through transvaginal ultrasound and abdominal ultrasound respectively; Preprocessing the collected transvaginal ultrasound image and abdominal ultrasound image; Performing feature extraction and feature matching on the preprocessed transvaginal ultrasound image and abdominal ultrasound image; Performing transformation alignment on the transvaginal ultrasound image and abdominal ultrasound image after feature extraction and feature matching, and allocating weights to the aligned transvaginal ultrasound image and abdominal ultrasound image according to the fusion weights, and fusing them into a comprehensive image.

[0005] Optionally, the collecting image data of the uterus through transvaginal ultrasound and abdominal ultrasound respectively includes: performing the transvaginal ultrasound examination using a high-frequency probe and performing the abdominal ultrasound examination using a low-frequency probe to obtain the transvaginal ultrasound image and abdominal ultrasound image with high resolution and wide field of view.

[0006] Optionally, the preprocessing of the collected transvaginal ultrasound image and the abdominal ultrasound image includes: denoising, contrast enhancement, and image correction; the denoising includes, Performing wavelet decomposition on the collected transvaginal ultrasound image and the abdominal ultrasound image to obtain sub-images of different frequencies; For the high-frequency sub-images, applying threshold processing to remove noise; Performing wavelet reconstruction on the sub-images of different frequencies after removing noise to obtain the denoised image.

[0007] Optionally, the contrast enhancement includes, Dividing the transvaginal ultrasound image and the abdominal ultrasound image into multiple small blocks, and performing histogram equalization on each small block; Calculating the equalization results of each small block and merging them into a complete image; After the contrast enhancement, it further includes: smoothing the merged area using the linear interpolation method.

[0008] Optionally, the image correction includes, Selecting a set of feature points and then performing correction using an affine transformation model; Estimating the parameters of the affine transformation model using the least squares method.

[0009] Optionally, after the preprocessing of the collected transvaginal ultrasound image and the abdominal ultrasound image, it further includes: adjusting the sizes of the transvaginal ultrasound image and the abdominal ultrasound image using the interpolation method so that the transvaginal ultrasound image and the abdominal ultrasound image have the same resolution in the same coordinate system.

[0010] Optionally, the feature extraction includes, Performing Gaussian blur processing on the transvaginal ultrasound image and the abdominal ultrasound image; Constructing a pyramid structure of the transvaginal ultrasound image and the abdominal ultrasound image, and extracting feature points at different scales; Calculating the direction and descriptor of each feature point to form a set of feature points.

[0011] Optionally, the feature matching includes, Matching the extracted descriptors, and calculating the Euclidean distance between the descriptors; Setting a distance threshold to screen out feature points with high matching degrees; Using the random sample consensus algorithm to further eliminate incorrect matches.

[0012] Optionally, after performing feature extraction and feature matching on the transvaginal ultrasound image and the abdominal ultrasound image, the transvaginal ultrasound image and the abdominal ultrasound image are transformed and aligned, and weight distribution is performed on the aligned transvaginal ultrasound image and abdominal ultrasound image according to the fusion weights to fuse them into a comprehensive image, including: According to the selected feature points with high matching degree, a transformation matrix between the transvaginal ultrasound image and the abdominal ultrasound image is calculated; Using the calculated transformation matrix to transform any one of the transvaginal ultrasound image and the abdominal ultrasound image to align it with the other image in the transvaginal ultrasound image and the abdominal ultrasound image, obtaining two aligned ultrasound images; According to the fusion weights, the two aligned ultrasound images are fused to obtain the comprehensive image; the fusion weights are adjusted according to the clarity and details of the comprehensive image, and the higher the value of the fusion weights, the higher the clarity and the more details of the comprehensive image.

[0013] Optionally, after fusing into a comprehensive image, it further includes: post-processing the comprehensive image to further improve the image quality; the post-processing includes: using image sharpening technology to enhance the edge details of the image. The present application provides an ultrasonic imaging scheme that combines transvaginal ultrasound images and abdominal ultrasound images into a clear and comprehensive one to assist the embryo transfer process.

[0014] The present application provides an ultrasonic imaging scheme that combines transvaginal ultrasound images and abdominal ultrasound images into a clear and comprehensive one to assist the embryo transfer process. The solution of the present application improves the clarity of the abdominal ultrasound image, making the evaluation of the endometrium and the embryo implantation position more accurate. Utilizing the high-resolution characteristics of transvaginal ultrasound, it overcomes the limitations of a single imaging technology and combines the wide field of view of abdominal ultrasound to provide more comprehensive anatomical information. By fusing two ultrasonic images, a new ultrasonic imaging technology is formed, providing a clearer and more comprehensive ultrasonic imaging scheme, enhancing the doctor's decision-making ability during the embryo transfer process, increasing the transplantation success rate, reducing the psychological and physiological burden of patients, and providing more reliable support for clinical practice.

[0015] To make the above features and advantages of the application more obvious and understandable, specific embodiments are hereinafter given and described in detail in conjunction with the accompanying drawings as follows. Brief Description of the Drawings

[0016] Figure 1 It is a flowchart of the ultrasonic imaging method based on image fusion of the present application. Detailed Embodiments

[0017] To make the objectives and technical solutions of the embodiments of this application clearer, the following will clearly and completely describe the technical solutions of the embodiments of this application in conjunction with the accompanying drawings of the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the described embodiments of this application without creative efforts shall fall within the scope of protection of this application.

[0018] In one embodiment, please refer to Figure 1 , this application provides an ultrasound imaging method based on image fusion to provide more accurate and reliable decision-making support for clinicians. The ultrasound imaging method based on image fusion includes the following steps: S10 to S30.

[0019] S10: Collect image data of the uterus through transvaginal ultrasound and abdominal ultrasound respectively.

[0020] S20: Preprocess the collected transvaginal ultrasound image of the uterus and the abdominal ultrasound image.

[0021] S30: Extract features and match features for the preprocessed transvaginal ultrasound image of the uterus and the abdominal ultrasound image.

[0022] S40: Transform and align the transvaginal ultrasound image of the uterus and the abdominal ultrasound image after feature extraction and feature matching, and allocate weights to the aligned transvaginal ultrasound image of the uterus and the abdominal ultrasound image according to the fusion weights to fuse them into a comprehensive image.

[0023] This application provides an ultrasound imaging method based on image fusion to assist in the embryo transfer process. The solution of this application improves the clarity of abdominal ultrasound images, making the evaluation of the endometrium and the embryo implantation position more accurate. Utilizing the high-resolution characteristics of transvaginal ultrasound, it overcomes the limitations of single imaging techniques and combines the wide field of view of abdominal ultrasound to provide more comprehensive anatomical information. By fusing two ultrasound images, a new ultrasound imaging technique is formed, providing a clearer and more comprehensive ultrasound imaging solution, enhancing the doctor's decision-making ability during the embryo transfer process, improving the transplantation success rate, reducing the psychological and physiological burden on patients, and providing more reliable support for clinical practice.

[0024] In step S10, please refer to Figure 1 in step S10, and collect image data of the uterus through transvaginal ultrasound and abdominal ultrasound respectively.

[0025] As an example, transvaginal ultrasound acquisition uses a high-frequency probe for transvaginal ultrasound examination. The high-frequency probe is usually 5 - 7.5 MHz, which can provide higher image resolution and is suitable for observing the details of the endometrium and the structure of the ovaries. During the examination, the operator needs to gently insert the probe into the vagina and adjust the probe angle to obtain the best view. At this time, the system will display the cross-sectional and longitudinal-sectional images of the uterus in real time, enabling clear observation of the shape of the uterus, endometrial thickness, and any abnormal structures.

[0026] As an example, abdominal ultrasound acquisition uses a low-frequency probe for abdominal ultrasound examination. The low-frequency probe is usually 2 - 5 MHz, which is suitable for observing a larger range of anatomical structures and can provide an overall view of the uterus and its surrounding tissues. When performing abdominal ultrasound, the operator needs to place the probe on the patient's abdomen, adjust the probe position and angle to obtain the overall view of the uterus and the relative positions of the surrounding organs. At this time, the system will display the ultrasound image of the abdomen to help the doctor evaluate the overall condition of the uterus. In step S20, refer to Figure 1 step S20 in, and preprocess the collected transvaginal ultrasound image of the uterus and the abdominal ultrasound image. Through image preprocessing, the quality during subsequent further operations on the images is ensured, providing high-quality input for subsequent image combination and analysis. Among them, the preprocessing includes: removing noise, enhancing contrast, and image correction.

[0027] As an example, wavelet transform is used to remove noise from the transvaginal ultrasound image of the uterus and the abdominal ultrasound image. The image is wavelet decomposed to obtain sub-images of different frequencies, effectively separating the high-frequency noise and low-frequency signals of the image. For the obtained low-frequency sub-images, which are characterized as the general background framework in the image, no excessive processing is performed; for the obtained high-frequency sub-images, since they contain rich details such as edges and texture features, the soft threshold method is used for processing to remove noise, and the formula is as follows: Among them, is the wavelet coefficient, is the threshold, is the sign function, max is the maximum value function.

[0028] Furthermore, after wavelet decomposition, wavelet reconstruction is performed to reconstruct the sub-images of different frequencies after removing noise to obtain the denoised image.

[0029] As an example, adaptive histogram equalization is used to enhance the contrast of an image. The image is divided into multiple small blocks, such as 8x8 pixel blocks. Histogram equalization is then performed on each small block, and the equalization result of each small block is calculated. By adjusting the contrast in the local area and merging them into a complete image, the detail performance of the complete image is improved.

[0030] Furthermore, in order to avoid edge effects, bilinear interpolation can be used to smooth the merged area, thereby improving the detail performance of the image by adjusting the contrast in the local area.

[0031] As an example, geometric correction is used to correct the uterine vaginal ultrasound image and the abdominal ultrasound image to eliminate image distortion caused by changes in probe position or angle. The geometric correction includes: first selecting a set of feature points, such as corner points or feature edges in the image, and recording their coordinates in the original image, and then using an affine transformation model for correction. The image is rotated, translated, scaled, and sheared by the affine transformation model to achieve conversion from one two-dimensional coordinate system to another two-dimensional coordinate system, keeping the parallel lines in the image parallel and the angle relationship unchanged. Specifically, after obtaining a reliable set of matching points, the calculation stage of the transformation matrix parameters is entered. The transformation matrix includes a 2×3 parameter, which includes information such as rotation, scaling, and translation; the perspective transformation requires the calculation of a 3×3 homography matrix. Taking the perspective transformation as an example, the perspective transformation requires the calculation of a 3×3 homography matrix, which is obtained based on the direct linear transformation algorithm combined with the least squares method.

[0032] As an example, the feature points selected on the uterine ultrasound image are , the feature points selected on the abdominal ultrasound image are , taking the two selected feature points as a pair of matching points, according to the mathematical model of perspective transformation, for each pair of matching points, two linear equations can be established: in, is the element of the perspective transformation matrix H. Since the system of equations is nonlinear, it is solved by converting it into a linear system of equations. Multiplying both sides of the above equation by the denominator, we can get: The equations corresponding to multiple pairs of matching points are combined to form an overdetermined linear equation system Ah = b, where A is the coefficient matrix, h is the column vector containing the elements of the perspective transformation matrix H, and b is the constant vector. At this time, the least squares method is used to solve the equation system, that is, to find the error vector The value h with the smallest second norm is used as the parameter estimation of the perspective transformation matrix H. Through this method, in the presence of noise and errors, the transformation matrix can be calculated as accurately as possible to achieve precise alignment between images, providing a strong guarantee for the effective integration of multi-source image information.

[0033] As an example, after the preprocessing of the transvaginal ultrasound image and the abdominal ultrasound image, it further includes: using the interpolation method to prepare for the fusion of the transvaginal ultrasound image and the abdominal ultrasound image. The interpolation method includes bilinear interpolation. By adjusting the image size, the two images have the same resolution in the same coordinate system.

[0034] In step S30, please refer to Figure 1 step S30 in, and perform feature extraction and feature matching on the preprocessed transvaginal ultrasound image and abdominal ultrasound image.

[0035] As an example, using the Scale-Invariant Feature Transform (SIFT) or Speeded-Up Robust Features (SURF) algorithm, extract the key points and their descriptors in the transvaginal ultrasound image and the abdominal ultrasound image to perform feature extraction on the transvaginal ultrasound image and the abdominal ultrasound image. First, perform Gaussian blur processing on the transvaginal ultrasound image and the abdominal ultrasound image to reduce the influence of residual noise on feature extraction; then construct the Gaussian pyramid structure of the transvaginal ultrasound image and the abdominal ultrasound image to extract feature points at different scales; starting from the preprocessed transvaginal ultrasound image and abdominal ultrasound image, repeatedly apply Gaussian kernel convolution and downsampling operations to generate an image sequence with decreasing resolution at multiple levels. During this process, the high-frequency details of the transvaginal ultrasound image and the abdominal ultrasound image are gradually stripped, and the low-frequency contour information is highlighted, laying a foundation for subsequent hierarchical fusion; finally, calculate the direction and descriptor of each feature point to form a set of feature points.

[0036] As an example, after the feature extraction, it further includes: using the nearest neighbor matching or FLANN algorithm to perform feature matching on the features extracted from the transvaginal ultrasound image and the abdominal ultrasound image. First, match the extracted feature descriptors and calculate the Euclidean distance between the descriptors. Set a distance threshold, and the value range of the distance threshold is between the absolute value of the high-frequency wave frequency and the absolute value of the low-frequency wave frequency. Screen out the feature points with high matching degree to ensure the accuracy of the matching. Finally, further eliminate the wrong matches through the Random Sample Consensus (RANSAC) algorithm to ensure the reliability of the matching result.

[0037] In step S40, please refer to Figure 1 step S40 in, perform transformation alignment on the transvaginal ultrasound image and the abdominal ultrasound image after feature extraction and feature matching, and allocate weights to the aligned transvaginal ultrasound image and abdominal ultrasound image according to the fusion weights to fuse them into a comprehensive image.

[0038] As an example, according to the feature points matched by features, an affine transformation or a perspective transformation is used to calculate the transformation matrix between the two images; further, the calculated transformation matrix is used to transform any one of the transvaginal uterine ultrasound image and the abdominal ultrasound image so that it is aligned with the other one of the transvaginal uterine ultrasound image and the abdominal ultrasound image, and two aligned ultrasound images are obtained. Further, the weighted average method is used to perform image fusion on the transvaginal uterine ultrasound image and the abdominal ultrasound image according to the fusion weight. When the doctor operates to control the instrument to collect images, at this time, the highest priority is positioning, and then details. Therefore, weight allocation is adopted, and the fusion weight at this time can be system-regulated or adjusted by the doctor's operation.

[0039] As an example, at the beginning of abdominal ultrasound acquisition, is the aligned transvaginal uterine ultrasound image, is the aligned abdominal ultrasound image. At this time, taking the transvaginal uterine ultrasound image as the main basis, the weight is correspondingly higher, and the weight of the abdominal ultrasound image is lower, mainly for positioning. When accurate placement points are needed after positioning to the approximate position, the abdominal ultrasound image is taken as the main basis, and the corresponding weight is higher. The formula for image fusion of the transvaginal uterine ultrasound image and the abdominal ultrasound image is as follows: where, is the fused image, and are the two aligned images respectively, is the aligned coordinate, is the fusion weight, and its value range is between 0 and 1.

[0040] Further, by adjusting the value of the fusion weight , the clarity and detail performance of the fused image are controlled. The higher the value of the fusion weight , the higher the clarity of the image and the clearer the detail performance.

[0041] As an example, the present application adopts an image fusion algorithm based on feature extraction and matching, combines the preprocessed transvaginal uterine ultrasound image and the abdominal ultrasound image, identifies the matching features and fuses them into a clearer and more information-rich comprehensive image.

[0042] In some embodiments, the aligned transvaginal ultrasound image and the abdominal color Doppler ultrasound image can be fused. The fusion methods can include, but are not limited to, pixel-level fusion, feature-level fusion, and decision-level fusion. For example, in some embodiments, the operation of applying the pixel-level fusion method can refer to the spatial domain fusion method: specifically, it includes simply processing each corresponding pixel in the transvaginal ultrasound image and the abdominal color Doppler ultrasound image, and then fusing them into a clear image for doctors to use. For another example, in some embodiments, the operation of applying the pixel-level fusion method can also refer to the frequency domain fusion method: specifically, it includes using methods such as pyramid decomposition and wavelet transform to process the high-frequency and low-frequency information of the transvaginal ultrasound image and the abdominal color Doppler ultrasound image respectively.

[0043] It should be noted that the specific methods for fusing the aligned transvaginal ultrasound image and the abdominal color Doppler ultrasound image can include, but are not limited to, the above fusion methods.

[0044] As an example, after the fusion into a comprehensive image, it further includes: performing post-processing to further improve the image quality. Image sharpening techniques, such as Laplacian sharpening, are used to enhance the edge details of the image, making the fused image clearer.

[0045] In another example, after the image post-processing, it further includes: feeding back the processed image and the analysis results to the clinician to assist the clinician in making more accurate decisions during the embryo transfer process, and collecting feedback information in clinical applications to continuously optimize the processing and fusion algorithms based on the combination of transvaginal uterine ultrasound images and abdominal ultrasound images, including the following steps S501~S504.

[0046] In step S501, after completing the image fusion algorithm of feature extraction and matching, a result feedback mechanism is adopted for the usage results of the embryo transfer application method based on the combination of transvaginal uterine ultrasound images and abdominal ultrasound images. The system will automatically generate a detailed report, which includes the processed fused image, the thickness of the endometrium, morphological features, and other key indicators. This information will be transmitted to the clinician in real time through the electronic medical record system or a dedicated medical software platform to ensure that the doctor can obtain the latest analysis results in a timely manner.

[0047] In step S502, after the system generates a detailed report, a decision support system is adopted. To assist the doctor in making more accurate decisions, the system integrates decision support algorithms. These algorithms are based on historical data and clinical guidelines and can provide personalized suggestions for the doctor. For example, the system can evaluate the success rate of embryo transfer based on the thickness and morphological features of the endometrium, combined with the patient's historical medical records, and give corresponding suggestions. This process can be achieved through a machine learning model, and the model needs to be trained on a large amount of clinical data to improve its prediction accuracy.

[0048] In step S503, after the doctor makes a decision and applies it, the system collects feedback information. During the clinical application process, the doctor's feedback information is crucial. The system designs a feedback information collection module that allows the doctor to evaluate the effects of the image processing and fusion algorithms. The doctor can provide information on aspects such as image clarity, feature extraction accuracy, and the effectiveness of decision support through a simple questionnaire or scoring system. This feedback will be recorded and analyzed by the system to identify potential areas for improvement.

[0049] In step S504, after collecting the feedback information, the collected feedback information will be used to continuously optimize the image processing and fusion algorithms. Specifically, the development team can regularly analyze the feedback data, identify the reasons why the algorithm performs poorly under specific circumstances, and adjust the algorithm parameters or improve the processing flow accordingly. For example, if the doctor feedbacks that the image clarity is insufficient in certain cases, the development team can consider introducing more advanced noise reduction techniques or image enhancement methods to improve the image quality.

[0050] In another embodiment, during a clinical application, the doctor needs to combine a transvaginal uterine ultrasound image with an abdominal ultrasound image. First, the system uses the Scale-Invariant Feature Transform (SIFT) algorithm to extract the feature points in the two images, obtaining their respective feature descriptors. Then, the system performs feature matching through the Fast Library for Approximate Nearest Neighbors (FLANN) algorithm, selects the feature point pairs with high matching degrees, and uses the Random Sample Consensus (RANSAC) algorithm to eliminate the incorrect matches. Subsequently, the matched feature point pairs calculate the transformation matrix between the two images, and through the calculated transformation matrix, the transvaginal uterine ultrasound image is aligned to the abdominal ultrasound image. Finally, the two images are fused into a clear composite image through the weighted average method. After image post-processing, the fused image obtained by the doctor is not only clearly visible but also contains the key information of the two images, providing a more reliable basis for embryo transfer. Through this image fusion algorithm based on feature extraction and matching, the usability and accuracy of the ultrasound images are improved. After the doctor receives the processed fused image and analysis results, it is found that the thickness of the endometrium is 8 mm and the morphology is normal. The system also provides a success rate prediction based on historical data, showing that the success rate of embryo transfer at this thickness is 75%. Based on this information, the doctor decides to perform embryo transfer and records this decision-making process. Subsequently, the doctor fills out a feedback questionnaire in the system, indicating that the image clarity is good but hopes to add more morphological feature analysis functions in future versions. After collecting this feedback, the development team begins to study how to incorporate more morphological features into the analysis scope to further enhance the usability and accuracy of the system. Through this cycle of feedback and optimization, the system continuously evolves and ultimately provides more accurate and reliable decision support for clinicians.

[0051] The present application provides an ultrasound imaging method based on image fusion to assist in the embryo transfer process. The solution of the present application improves the clarity of abdominal ultrasound images, making the evaluation of the endometrium and the embryo implantation position more accurate. Utilizing the high-resolution characteristics of transvaginal ultrasound, it overcomes the limitations of a single imaging technique and combines the wide field of view of abdominal ultrasound to provide more comprehensive anatomical information. By fusing the two ultrasound images, a new ultrasound imaging technique is formed, providing a clearer and more comprehensive ultrasound imaging solution, enhancing the doctor's decision-making ability during the embryo transfer process, increasing the transplantation success rate, reducing the psychological and physiological burden on patients, and providing more reliable support for clinical practice.

[0052] Although the present application has been disclosed above by way of examples, it is not intended to limit the present application. Any person with ordinary knowledge in the technical field to which the present application pertains may make some modifications and refinements without departing from the spirit and scope of the present application. Therefore, the protection scope of the present application shall be subject to that defined by the appended patent application scope.

Claims

1. An ultrasonic imaging method based on image fusion, characterized in that, including, acquiring image data of the uterus through transvaginal ultrasound and abdominal ultrasound respectively; preprocessing the acquired transvaginal ultrasound images and abdominal ultrasound images of the uterus; performing feature extraction and feature matching on the preprocessed transvaginal ultrasound images and abdominal ultrasound images of the uterus; performing transformation alignment on the transvaginal ultrasound images and abdominal ultrasound images after the feature extraction and feature matching, and performing weight allocation on the aligned transvaginal ultrasound images and abdominal ultrasound images according to the fusion weights, and fusing them into a comprehensive image.

2. The ultrasonic imaging method based on image fusion according to claim 1, characterized in that The acquiring image data of the uterus through transvaginal ultrasound and abdominal ultrasound respectively includes: performing the transvaginal ultrasound examination using a high-frequency probe and performing the abdominal ultrasound examination using a low-frequency probe to obtain the transvaginal ultrasound images and abdominal ultrasound images with high resolution and wide field of view.

3. The ultrasonic imaging method based on image fusion according to claim 1, characterized in that The preprocessing the acquired transvaginal ultrasound images and abdominal ultrasound images of the uterus includes: denoising, enhancing contrast, and image correction; the denoising includes, performing wavelet decomposition on the acquired transvaginal ultrasound images and abdominal ultrasound images to obtain sub-images with different frequencies; applying threshold processing to the high-frequency sub-images to remove noise; performing wavelet reconstruction on the sub-images with different frequencies after removing noise to obtain the denoised images.

4. The ultrasonic imaging method based on image fusion according to claim 3, wherein, The enhancing contrast includes, dividing the transvaginal ultrasound images and abdominal ultrasound images into multiple small blocks, and performing histogram equalization on each small block; calculating the equalization results of each small block and merging them into a complete image; after the enhancing contrast, it further includes: smoothing the merging area by using the linear interpolation method.

5. The ultrasound imaging method based on image fusion according to claim 4, wherein The image correction includes, selecting a group of feature points and then performing correction by using an affine transformation model; estimating the parameters of the affine transformation model by using the least squares method.

6. The ultrasonic imaging method based on image fusion according to claim 3, wherein After the preprocessing the acquired transvaginal ultrasound images and abdominal ultrasound images of the uterus, it further includes: adjusting the sizes of the transvaginal ultrasound images and abdominal ultrasound images by using the interpolation method so that the transvaginal ultrasound images and abdominal ultrasound images have the same resolution in the same coordinate system.

7. The ultrasonic imaging method based on image fusion according to claim 5, characterized in that, The feature extraction includes, performing Gaussian blur processing on the transvaginal ultrasound images and abdominal ultrasound images of the uterus; constructing a pyramid structure of the transvaginal ultrasound images and abdominal ultrasound images of the uterus and extracting feature points at different scales; calculating the direction and descriptor of each feature point to form a set of feature points.

8. The ultrasonic imaging method based on image fusion according to claim 7, wherein, The feature matching includes, matching the extracted descriptors, and calculating the Euclidean distance between the descriptors; setting a distance threshold to screen out the feature points with high matching degree; using the random sample consensus algorithm to further eliminate the wrong matches.

9. The ultrasonic imaging method based on image fusion according to claim 8, characterized in that, The performing transformation alignment on the transvaginal ultrasound images and abdominal ultrasound images after the feature extraction and feature matching, and performing weight allocation on the aligned transvaginal ultrasound images and abdominal ultrasound images according to the fusion weights, and fusing them into a comprehensive image, includes, calculating the transformation matrix between the transvaginal ultrasound images and abdominal ultrasound images according to the screened feature points with high matching degree; Perform transformation on any one of the transvaginal ultrasound image and the abdominal ultrasound image using the calculated transformation matrix to align it with the other one of the transvaginal ultrasound image and the abdominal ultrasound image, and obtain two aligned ultrasound images; Fuse the two aligned ultrasound images according to the fusion weight to obtain the composite image; the fusion weight is adjusted according to the clarity and details of the composite image, and the higher the value of the fusion weight, the higher the clarity and the more details of the composite image.

10. The ultrasonic imaging method based on image fusion according to claim 9, characterized in that, After the fusion into a composite image, it further includes: performing post-processing on the composite image to further improve the image quality; the post-processing includes: using an image sharpening technique to enhance the edge details of the image.

Images