Placenta implantation analysis method based on ultrasonic image

Abstract

The invention discloses a placenta implantation analysis method based on an ultrasonic image, particularly relates to the technical field of medical image analysis, and is used for solving the problem that the placenta implantation condition is difficult to accurately diagnose due to artifact interference caused by complex reflection of an existing placenta area and adjacent blood vessels. The method comprises the following steps: acquiring ultrasonic radio frequency data, generating a two-dimensional gray scale image and a Doppler blood flow image, constructing a dynamic interference distribution diagram, extracting artifact time sequence characteristics, realizing frequency domain separation of an artifact region and generation of a multi-reflection interference pixel set in combination with wavelet transform, and identifying a real blood flow signal through an adaptive threshold algorithm. According to the method, artifacts and real signals are fused to generate an artifact-removed image, morphological processing based on structural elements is carried out on the artifact-removed image, continuous boundary information of the placenta and the uterine wall is extracted, suspicious blood vessel invasion paths are marked in a partitioned mode, and the accuracy of image analysis is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image analysis, and more particularly to a placenta implantation analysis method based on ultrasound images. Background Art

[0002] Placenta accreta is a high-risk pregnancy complication, and its diagnosis and evaluation are crucial for obstetric management. Ultrasound imaging technology is widely used for structural analysis and blood flow assessment in the placental region due to its non-invasive and high-resolution characteristics. However, in cases where the placenta and adjacent large blood vessels (such as the iliac vessels) are complexly distributed, multiple reflections and refractions of the ultrasound signal can easily interfere with the imaging quality, resulting in artifacts or ghosting in the image that are difficult to eliminate with conventional filters. These artifacts not only affect the clear presentation of the relationship between the placental structure and blood vessels, but can also lead to misjudgment of blood flow distribution, increasing diagnostic uncertainty.

[0003] In the existing technology, the methods for processing artifact interference in the placenta area mostly use fixed filters or threshold segmentation technology, but these methods are difficult to effectively distinguish artifact signals from real signals, which will lead to inaccurate ultrasound image analysis results of the placenta area and adjacent blood vessels. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a placenta accreta analysis method based on ultrasound imaging to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: The placenta accreta analysis method based on ultrasound imaging includes the following steps: Acquire ultrasound radiofrequency data covering the placenta area and adjacent blood vessels, and generate two-dimensional grayscale images and Doppler blood flow images; Construct a dynamic interference distribution map of multiple frames of ultrasound radiofrequency data, extract the temporal characteristics of artifact signals in the placenta region, and mark the artifact dynamic characteristic areas; Wavelet transform is used to separate the dynamic characteristic area of ​​artifacts in the frequency domain of two-dimensional grayscale images, and a pixel set of multiple reflection interference areas is generated; Adaptive threshold estimation algorithm is used for Doppler blood flow images to identify real blood flow signals using gray intensity information and blood flow velocity information; The pixel set of the multi-reflection interference area and the real blood flow signal are fused, and the pixels generated by the multi-reflection interference are set as invalid data to generate a de-artifacted image; Morphological processing based on structural elements was performed on the de-artifacted images to extract the continuous boundary information between the placenta and the uterine wall, and to identify the suspicious vascular invasion paths.

[0006] In a preferred embodiment, obtaining ultrasound radiofrequency data including the placenta region and adjacent blood vessels and generating a two-dimensional grayscale image and a Doppler blood flow image specifically includes: An ultrasound device is used to collect radio frequency signals from the placenta and adjacent blood vessels. The ultrasound probe scans the target area at a fixed frequency and angle, covering the placenta and its adjacent blood vessels. A two-dimensional grayscale image is generated by processing the amplitude and phase of the radio frequency signal. The two-dimensional grayscale image is used to display the structural distribution of the placenta tissue and its adjacent blood vessels. A Doppler blood flow image is generated based on the Doppler frequency shift information in the radio frequency signal. The Doppler blood flow image is used to reflect the blood flow velocity and direction in the placenta area and adjacent blood vessels.

[0007] In a preferred embodiment, a dynamic interference distribution map of multiple frames of ultrasound radio frequency data is constructed, the time sequence characteristics of the artifact signal in the placenta region are extracted, and the artifact dynamic characteristic region is marked, specifically including: Acquiring multiple frames of continuous ultrasound radio frequency data including the placenta region and adjacent blood vessels, where the ultrasound radio frequency data includes amplitude information and phase information; Perform time-series demodulation on multiple frames of ultrasound radio frequency data, extract the amplitude and phase change curves of the signal in the placenta area, and form initial time-series data reflecting the dynamic characteristics of the artifact signal; A dynamic interference distribution map is constructed based on the initial time series data reflecting the dynamic characteristics of the artifact signal. The dynamic interference distribution map records the time series characteristics of the artifact signal in the placenta region and adjacent blood vessels, including the changing frequency, intensity, and spatial distribution of the artifact signal. The timing regularity of the artifact signal is analyzed through the dynamic interference distribution diagram, and the dynamic characteristic area of ​​the artifact is marked.

[0008] In a preferred embodiment, wavelet transform is used to perform frequency domain separation on the two-dimensional grayscale image to separate the dynamic characteristic region of the artifact and generate a pixel set of the multi-reflection interference region, specifically including: Based on the marked artifact dynamic characteristic region, pixel data of the corresponding two-dimensional grayscale image is intercepted; Wavelet transform is used to decompose the pixel data of the two-dimensional grayscale image, and the two-dimensional grayscale image signal is decomposed into multiple scale and frequency sub-bands to extract the high-frequency components related to multiple reflection interference; The high-frequency subbands of wavelet decomposition are screened by threshold denoising method; The filtered high-frequency sub-bands are inversely transformed to generate a pixel set containing the multi-reflection interference area, which is used to record the spatial distribution and intensity information of the multi-reflection interference area.

[0009] In a preferred embodiment, an adaptive threshold estimation algorithm is used for the Doppler blood flow image to identify the real blood flow signal using grayscale intensity information and blood flow velocity information, specifically including: Extract the grayscale intensity value and corresponding blood flow velocity value of each pixel from the Doppler blood flow image. The grayscale intensity value reflects the echo signal intensity of the pixel point, and the blood flow velocity value represents the blood flow velocity of the pixel point. Calculating the grayscale intensity distribution of the entire image based on the extracted grayscale intensity values, including the mean and standard deviation of the grayscale intensity values; Calculating velocity distribution characteristics of the entire image based on the extracted blood flow velocity values, including the mean and standard deviation of the blood flow velocity values; Combining the gray intensity distribution and velocity distribution characteristics, an adaptive threshold estimation algorithm is used to calculate the gray intensity threshold and velocity threshold respectively. The pixels in the Doppler blood flow image are screened, and only the pixels with grayscale intensity values ​​higher than the grayscale intensity threshold and blood flow velocity values ​​higher than the velocity threshold are retained and identified as pixels with real blood flow signals.

[0010] In a preferred embodiment, the pixel set of the multi-reflection interference area and the real blood flow signal are fused, and the pixels generated by the multi-reflection interference are set as invalid data to generate a de-artifacted image, specifically including: Based on the pixel set of the multi-reflection interference area, the spatial position coordinates and interference signal intensity of the corresponding pixels are extracted; Based on the pixel set of the real blood flow signal, the spatial position coordinates and blood flow signal intensity of the corresponding pixels are extracted; Matching analysis is performed on the pixel set of the multi-reflection interference area and the pixel set of the real blood flow signal. The matching rules include the degree of overlap of spatial position coordinates and the relative difference in signal intensity. Pixels belonging to the multiple reflection interference area in the matching analysis results are set as invalid data. Invalid data is defined as the signal strength of which is mainly caused by multiple reflection interference. The filtered invalid data is removed from the fused pixel set, and only the pixel data of the real blood flow signal is retained; A de-artifacted image is generated based on the retained pixel data of the real blood flow signal and its corresponding spatial distribution. The de-artifacted image fully records the real blood flow distribution and signal intensity in the placenta area.

[0011] In a preferred embodiment, the de-artifacted image is subjected to morphological processing based on structural elements to extract the continuous boundary information between the placenta and the uterine wall, and to identify the suspicious vascular invasion path by partition, specifically including: The de-artifacted image was binarized, and the grayscale threshold was set to separate the placenta area and the background area to generate a preliminary binary image; The size and shape of the structure element are designed according to the characteristics of the binary image, and the structure element is used in morphological operations to eliminate residual artifacts; Morphological opening operation is used to remove isolated small areas at the boundary of the placenta region and enhance the boundary continuity between the placenta and the uterine wall; Morphological closing operations are used to fill the gaps inside the placenta region and optimize the boundary connectivity between the placenta region and the uterine wall. Apply edge detection algorithm to extract continuous boundary information between the placenta and uterine wall. The continuous boundary information includes the outer contour of the placenta area and its contact interface with the uterine wall. Based on the continuous boundary information, the suspicious vascular invasion path is partitioned and identified, which specifically includes analyzing the irregularity of the boundary and the overlapping area with the vascular signal, and recording the spatial position and boundary characteristics of each partition to form the identification result.

[0012] The technical effects and advantages of the placenta implantation analysis method based on ultrasound imaging of the present invention are as follows: 1. By constructing a dynamic interference distribution map of multi-frame ultrasound radiofrequency data, the temporal characteristics of artifact signals in the placenta region are extracted. Combined with wavelet transforms, the frequency domain separation of the artifact dynamic characteristic region is achieved, effectively generating a pixel set in the multi-reflection interference region. The Doppler blood flow image is analyzed using an adaptive threshold estimation algorithm, integrating grayscale intensity information and blood flow velocity information to accurately identify the true blood flow signal. This is then fused with the artifact signal, and pixels generated by multi-reflection interference are set as invalid data, ultimately generating a de-artifacted image. This method can significantly reduce the impact of artifact interference, optimize the signal quality of ultrasound images, and provide clear and reliable basic data for analyzing the relationship between the placenta region and adjacent blood vessels.

[0013] 2. By performing morphological processing based on structural elements on the de-artifacted images, the continuity and integrity of the boundary between the placenta and the uterine wall are enhanced through opening and closing operations, and the continuous boundary information of the placenta and the uterine wall is extracted in combination with the edge detection algorithm. On this basis, the suspicious vascular invasion path is partitioned and identified, and the overlapping areas of boundary irregularities and vascular signals are accurately located. This can improve the identification accuracy of the risk areas of placenta implantation, provide a clearer reference basis for clinical diagnosis, help reduce misdiagnosis and missed diagnosis, and improve the efficiency and accuracy of early assessment of placenta implantation. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 Schematic diagram of the placenta accreta analysis method based on ultrasound imaging of the present invention. DETAILED DESCRIPTION

[0015] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0016] Example: Figure 1 The present invention provides a method for analyzing placenta accreta based on ultrasound imaging, which comprises the following steps: Ultrasound radiofrequency data covering the placenta and adjacent blood vessels are acquired to generate two-dimensional grayscale images and Doppler blood flow images.

[0017] A dynamic interference distribution map of multi-frame ultrasound radio frequency data was constructed, the temporal characteristics of the artifact signal in the placenta region were extracted, and the artifact dynamic characteristic area was marked.

[0018] Wavelet transform is used to separate the dynamic characteristic area of ​​artifacts in the frequency domain of two-dimensional grayscale images, and a pixel set of multiple reflection interference areas is generated.

[0019] An adaptive threshold estimation algorithm is used for Doppler blood flow images, and the grayscale intensity information and blood flow velocity information are used to identify the real blood flow signal.

[0020] The pixel set of the multi-reflection interference area and the real blood flow signal are fused, and the pixels generated by the multi-reflection interference are set as invalid data to generate a de-artifacted image.

[0021] Morphological processing based on structural elements was performed on the de-artifacted images to extract the continuous boundary information between the placenta and the uterine wall, and to identify the suspicious vascular invasion paths.

[0022] Acquires ultrasound radiofrequency data covering the placenta and adjacent blood vessels, generating two-dimensional grayscale images and Doppler blood flow images, including: Ultrasound equipment is used to collect radio frequency signals from the placenta and adjacent blood vessels. The ultrasound probe scans the target area at a fixed frequency and angle, covering the placenta and its adjacent blood vessels: Configure the ultrasound probe's operating parameters, including scanning frequency, angle, and scanning mode. Set the scanning frequency to a range with high resolution for placental tissue and vascular signals, such as 2-10 MHz. Adjust the scanning angle to cover the entire placental region and its adjacent blood vessels.

[0023] The pregnant woman's position is fixed to ensure adequate exposure of the placenta. The ultrasound probe is moved at a constant speed, scanning the target area with the ultrasound beam to collect radiofrequency signals from the placenta and adjacent blood vessels. The radiofrequency signal includes the amplitude and phase information of the reflected signal, which is used for subsequent signal processing.

[0024] A two-dimensional grayscale image is generated by processing the amplitude and phase of the radiofrequency signal. The two-dimensional grayscale image is used to display the structural distribution of the placenta tissue and its adjacent blood vessels: The signal processing module of the ultrasound device is used to demodulate the amplitude of the collected radio frequency signal to obtain amplitude data related to the echo intensity of the placental tissue and adjacent blood vessels.

[0025] The demodulated amplitude data is mapped to grayscale values ​​to generate a two-dimensional grayscale image. The grayscale image is used to display the tissue structure of the placenta region and the location and distribution of adjacent blood vessels.

[0026] The generated grayscale images were calibrated to ensure that the image resolution met the requirements of subsequent analysis, especially in the area where the placenta and adjacent blood vessels meet the junction, to maintain clear structural boundaries.

[0027] A Doppler blood flow image is generated based on the Doppler frequency shift information in the radio frequency signal. The Doppler blood flow image is used to reflect the blood flow velocity and direction in the placenta area and adjacent blood vessels: The frequency shift component is extracted from the acquired radio frequency signal, and the blood flow velocity and direction in the placenta area and adjacent blood vessels are calculated using the Doppler frequency shift formula.

[0028] The calculated blood flow velocity and direction information is encoded as color markers to generate a Doppler blood flow image. Different colors in the image represent different directions of blood flow, and the brightness indicates the magnitude of blood flow velocity.

[0029] The generated Doppler blood flow images are optimized and the image contrast and color distribution are adjusted to ensure that the blood flow distribution information of the placenta area and adjacent blood vessels is clearly presented.

[0030] The structural information of the grayscale image is aligned with the blood flow velocity and direction information of the Doppler blood flow image to ensure the spatial consistency of the two imaging modes.

[0031] Construct a dynamic interference distribution map of multiple frames of ultrasound radiofrequency data, extract the temporal characteristics of the artifact signal in the placenta area, and mark the artifact dynamic characteristic area, including: Acquire multiple frames of continuous ultrasound radiofrequency data covering the placenta area and adjacent blood vessels. The ultrasound radiofrequency data includes amplitude and phase information: Configure the ultrasound device's probe parameters, including transmit frequency, detection depth, and frame rate. Set the transmit frequency between 2 MHz and 10 MHz to ensure high-resolution imaging of the placenta and adjacent vessels. Set the frame rate to 50 frames / second or higher to ensure temporal resolution of the dynamic characteristics of the artifact signal.

[0032] The placenta area and adjacent blood vessels were selected as the scanning target, ensuring that the ultrasound beam completely covered the target area, and the probe angle was adjusted to avoid missing key areas.

[0033] Acquires multiple frames of continuous RF data, each containing both amplitude and phase information. This data is obtained through ultrasonic reflections. Amplitude reflects the echo intensity of the target tissue, while phase describes the phase variation of the echo signal.

[0034] Perform time series demodulation on multiple frames of ultrasound radio frequency data, extract the amplitude change curve and phase change curve of the signal in the placenta area, and form initial time series data reflecting the dynamic characteristics of the artifact signal: Perform envelope detection on the amplitude part of the RF signal and extract the amplitude change curve of each pixel in the continuous frame, which is recorded as ,in For time.

[0035] The phase part of the RF signal is analyzed and the phase change curve of each pixel in the continuous frame is calculated, which is recorded as ,in For time The phase value on is expressed as: ;in, represents the complex representation of the RF signal, is the real part of the RF signal, is the imaginary part of the RF signal.

[0036] The amplitude change curve and phase change curve of each pixel are combined to generate initial time series data reflecting the dynamic characteristics of the artifact signal, which serves as the basis for subsequent processing.

[0037] A dynamic interference distribution map is constructed based on the initial time series data reflecting the dynamic characteristics of the artifact signal. The dynamic interference distribution map records the time series characteristics of the artifact signal in the placenta region and adjacent blood vessels, including the changing frequency, intensity, and spatial distribution of the artifact signal: Perform Fourier transform on the amplitude change curve and phase change curve of each pixel point to calculate the change frequency of the artifact signal. The formula is expressed as: ;in, Indicates the frequency of change of the artifact signal.

[0038] The mean and variance of the amplitude of each pixel are calculated to characterize the intensity distribution of the artifact signal.

[0039] The changing frequency of the artifact signal, the mean value and variance of the pixel amplitude are recorded in a two-dimensional spatial distribution map to generate a dynamic interference distribution map to display the artifact signal characteristics of the placenta area and adjacent blood vessels.

[0040] Analyze the timing pattern of the artifact signal through the dynamic interference distribution diagram and mark the dynamic characteristic area of ​​the artifact: Based on the change frequency of the artifact signal and the variance of the pixel amplitude, pixels with a high change frequency of the artifact signal and a large variance of the pixel amplitude are screened out as candidate areas of the artifact signal.

[0041] Among them, screening out pixels with high artifact signal change frequency and large pixel amplitude variance can be achieved by setting a threshold. The specific method is to first count the artifact signal change frequency and amplitude variance of all pixels, and calculate their mean and standard deviation; then set the threshold to a value higher than the mean plus several times the standard deviation. Pixels whose frequency and variance simultaneously exceed the corresponding threshold are identified as candidate areas for artifact signals. The adjustment multiple parameter can be flexibly set according to the characteristics of the actual artifact signal to optimize the screening accuracy and adapt to the data distribution of different scenarios.

[0042] Using spatial clustering algorithms (such as the DBSCAN algorithm), adjacent artifact pixels are clustered into continuous regions to form artifact dynamic characteristic regions.

[0043] Wavelet transform is used to separate the dynamic characteristic area of ​​the artifact in the frequency domain of the two-dimensional grayscale image, and a pixel set of the multi-reflection interference area is generated, which includes: Based on the marked artifact dynamic characteristic area, the pixel data of the corresponding two-dimensional grayscale image is intercepted: From the artifact dynamic characteristic region generated in the previous processing step, the spatial coordinate information of the marked region is extracted to ensure that the marked region covers all possible multi-reflection interference regions.

[0044] Based on the spatial coordinates of the marked area, the corresponding pixel data is extracted from the 2D grayscale image to form a sub-image block data set. The size of the sub-image block is determined by the range of the marked area to ensure that the extracted data can fully contain the dynamic characteristics of the artifact area.

[0045] The captured sub-image block data is normalized and the grayscale value is mapped to a fixed range (such as 0 to 1) to eliminate the influence of different image acquisition conditions (such as probe gain and equipment parameters) on the data and provide consistent input for subsequent wavelet decomposition.

[0046] Wavelet transform is used to decompose the pixel data of the two-dimensional grayscale image, and the two-dimensional grayscale image signal is decomposed into multiple scale and frequency sub-bands to extract the high-frequency components related to multiple reflection interference: A wavelet basis function suitable for two-dimensional image analysis, such as Daubechies wavelet or Haar wavelet, is selected to effectively separate high-frequency and low-frequency components. The specific choice of wavelet basis function is adjusted according to the characteristics of the image and the signal characteristics of multiple reflection interference.

[0047] A two-dimensional discrete wavelet transform is performed on the normalized sub-image block data, decomposing the signal into multiple scale and frequency subbands. Each level of wavelet decomposition divides the image into four subbands: a low-frequency subband (LL) and three high-frequency subbands (LH, HL, and HH). LH represents the horizontal high-frequency component, HL represents the vertical high-frequency component, and HH represents the diagonal high-frequency component.

[0048] The high-frequency sub-bands (LH, HL, and HH) associated with multi-reflection interference are extracted from the decomposition results. These high-frequency sub-bands contain the main features of the artifact signal and are the focus of subsequent screening and processing.

[0049] The high-frequency subbands of wavelet decomposition are screened by threshold denoising method: Calculate the amplitude distribution of each high-frequency subband, calculate the mean and standard deviation, and set the threshold to the mean plus a multiple of the standard deviation. This threshold is used to filter out significant high-frequency signals related to multi-reflection interference.

[0050] The value of each pixel in the high-frequency subband is screened, pixel values ​​below the threshold are set to zero, and only pixel values ​​above the threshold are retained to highlight the interference characteristics related to the artifact.

[0051] The high-frequency sub-band data after threshold screening is saved as a high-frequency signal data set reflecting the characteristics of the multiple reflection interference area, providing input for subsequent inverse transformation.

[0052] Perform inverse transformation on the filtered high-frequency sub-band to generate a pixel set containing the multi-reflection interference area, which is used to record the spatial distribution and intensity information of the multi-reflection interference area: The filtered high-frequency subband data is subjected to a two-dimensional inverse wavelet transform, reconstructing the decomposed signal into a set of pixels containing the multi-reflection interference region. The reconstruction process uses only the high-frequency subband data, and the low-frequency subband data is set to zero to exclude low-frequency information unrelated to the interference.

[0053] The reconstructed pixel set represents the spatial distribution and intensity information of the multi-reflection interference area. The value of each pixel reflects the intensity of the interference signal, and the position of the pixel represents the spatial distribution of the interference signal.

[0054] The generated pixel set is stored in a standard image format or array for subsequent artifact removal and image optimization processing.

[0055] An adaptive threshold estimation algorithm is used for Doppler blood flow images to identify real blood flow signals using grayscale intensity information and blood flow velocity information. Specifically, the algorithm includes: Extract the grayscale intensity value and corresponding blood flow velocity value of each pixel from the Doppler blood flow image. The grayscale intensity value reflects the echo signal intensity of the pixel point, and the blood flow velocity value represents the blood flow velocity of the pixel point: Load Doppler blood flow images to ensure image integrity and quality, remove possible noise and artifacts in the image, and provide stable data input for subsequent processing.

[0056] Traverse each pixel in the Doppler blood flow image and read its grayscale intensity value. The grayscale intensity value directly reflects the echo signal strength of the pixel and is an important indicator of blood flow signal quality, used to distinguish valid from invalid signals.

[0057] For each pixel in the Doppler blood flow image, the corresponding blood flow velocity value is extracted. The blood flow velocity value represents the blood flow speed at the pixel location and is important dynamic information for identifying the true blood flow signal.

[0058] The extracted grayscale intensity values ​​and blood flow velocity values ​​are stored as two independent matrix data structures, respectively, to ensure that subsequent processing can directly call these data for analysis.

[0059] Calculate the grayscale intensity distribution of the entire image based on the extracted grayscale intensity values, including the mean and standard deviation of the grayscale intensity values: For the complete Doppler blood flow image matrix data, the pixel range containing the target blood flow area is selected for gray value distribution calculation to exclude the noise influence that may be caused by the boundary area.

[0060] The grayscale intensity values ​​of all pixels within the target blood flow area are counted and the average grayscale value is calculated, which serves as an important basis for subsequent threshold estimation. The average grayscale value is used to reflect the overall intensity level of the target area signal.

[0061] The standard deviation of the grayscale intensity values ​​within the target blood flow region is calculated to characterize the dispersion of grayscale values ​​within the target region. A higher standard deviation indicates greater signal volatility in the target region, and subsequent threshold settings need to be adjusted based on this characteristic.

[0062] The velocity distribution characteristics of the entire image are calculated based on the extracted blood flow velocity values, including the mean and standard deviation of the blood flow velocity values: The target area of ​​the blood flow velocity value matrix is ​​selected for analysis, which is consistent with the calculation range of the gray intensity value to ensure the consistency of the grayscale and velocity characteristic distribution.

[0063] The blood flow velocity values ​​of all pixels within the target area are counted and the average value is calculated as an important reference for subsequent threshold estimation. The average velocity value reflects the overall trend of blood flow movement in the target area.

[0064] The standard deviation of blood velocity values ​​within the target area is calculated to characterize the volatility and magnitude of blood flow. The size of the standard deviation directly determines the sensitivity of the threshold setting; a larger standard deviation requires increasing the threshold range.

[0065] Combining the grayscale intensity distribution and velocity distribution characteristics, an adaptive threshold estimation algorithm is used to calculate the grayscale intensity threshold and velocity threshold respectively: Combining the mean and standard deviation of grayscale intensity and blood flow velocity distribution, an adaptive threshold estimation algorithm is selected to adapt to the changing characteristics of image data by dynamically adjusting the threshold range.

[0066] The grayscale intensity threshold is set based on the mean and standard deviation of the grayscale intensity values. The threshold is set as the mean plus a multiple of the standard deviation to effectively filter out pixels with high grayscale intensity.

[0067] The blood velocity threshold is set based on the mean and standard deviation of the blood velocity values. The threshold is set as the mean plus a multiple of the standard deviation to filter out pixels with high blood flow velocity.

[0068] The grayscale intensity threshold and the blood flow velocity threshold are fused together to retain only the pixels that meet both conditions, ensuring that the screened signals have both strong echo signal intensity and significant blood flow motion characteristics.

[0069] Filter the pixels in the Doppler blood flow image and only retain the pixels whose grayscale intensity value is higher than the grayscale intensity threshold and whose blood flow velocity value is higher than the velocity threshold. These pixels are identified as pixels with real blood flow signals: Traverse all pixels in the Doppler blood flow image, filter the grayscale intensity value and blood flow velocity value of each pixel, and only retain the pixels that meet both threshold conditions.

[0070] The pixels that meet the conditions are marked as real blood flow signals, and the spatial position coordinates, grayscale intensity value and blood flow velocity value of each pixel are recorded to form a complete data set containing spatial and signal characteristics.

[0071] The marked true blood flow signal pixel set is exported into a standardized data format for subsequent artifact removal and image optimization processing.

[0072] The pixel set of the multi-reflection interference area and the real blood flow signal are fused, and the pixels generated by the multi-reflection interference are set as invalid data to generate a de-artifacted image. Specifically, the following steps are performed: Based on the pixel set of the multi-reflection interference area, the spatial position coordinates and interference signal strength of the corresponding pixels are extracted: From the previous processing step, a pixel set in the multi-reflection interference area is obtained, including the spatial coordinates and interference signal strength of the multi-reflection interference pixels. This pixel set is derived from the results of the aforementioned wavelet transform separation and high-frequency signal screening, ensuring that the interference characteristics of each pixel are accurate.

[0073] The pixel set in the multi-reflection interference area is traversed, and the spatial coordinates of each pixel, including horizontal and vertical positions, are extracted one by one for subsequent matching analysis with the true blood flow signal. The interference signal strength of each pixel is read from the pixel set. This strength reflects the interference characteristics of the pixel, such as the strength of the artifact signal or the severity of the interference effect.

[0074] Based on the pixel set of the real blood flow signal, the spatial position coordinates and blood flow signal intensity of the corresponding pixels are extracted: A pixel set of a real blood flow signal is obtained. The pixel set is generated by the aforementioned adaptive threshold estimation algorithm of the Doppler blood flow image and includes the spatial position coordinates and blood flow signal intensity of the real blood flow signal pixel points.

[0075] Traverse the pixel set of the real blood flow signal and extract the spatial position coordinates of each pixel, including horizontal and vertical positions, for subsequent matching analysis with multiple reflection interference pixels.

[0076] The blood flow signal intensity of each pixel is read from the pixel set of the real blood flow signal. The intensity represents the real blood flow characteristics corresponding to the pixel, including a comprehensive reflection of the blood flow velocity and flow direction.

[0077] Matching analysis is performed on the pixel set of the multi-reflection interference area and the pixel set of the real blood flow signal. The matching rules include the degree of overlap of spatial position coordinates and the relative difference in signal intensity: The pixel set of the multi-reflection interference area and the pixel set of the real blood flow signal are combined to read the spatial position coordinates and signal intensity of each pixel respectively.

[0078] The matching rule consists of two parts: The degree of coincidence of spatial position coordinates: Determine whether the spatial position coordinates of the multi-reflection interference pixels and the true blood flow signal pixels are the same or close (for example, the position difference is within one pixel).

[0079] Relative difference in signal strength: Calculate the difference between the interference signal strength and the true blood flow signal strength. If the interference signal strength ratio exceeds a certain threshold, it is determined to be an artifact signal.

[0080] Generate matching analysis results based on matching rules, including the matching status and signal characteristics of each pixel, for subsequent screening and elimination of invalid data.

[0081] Pixels in the matching analysis results that belong to the multi-reflection interference area are set as invalid data. Invalid data is defined as the signal strength of which is mainly caused by multi-reflection interference: In the matching analysis results, if a pixel meets any of the following conditions, it will be set as invalid data: The spatial position of the pixel only appears in the multi-reflection interference pixel set and does not appear in the real blood flow signal pixel set; The signal strength of the pixel point mainly comes from the interference signal, that is, the interference signal strength ratio exceeds the preset threshold (for example, 70%).

[0082] Pixels that meet the definition of invalid data are marked as invalid, and their spatial position coordinates and signal characteristics are recorded to ensure that these pixels can be accurately eliminated in the subsequent fusion process.

[0083] Remove the filtered invalid data from the fused pixel set and only retain the pixel data of the real blood flow signal: The fused pixel set of the multi-reflection interference area and the real blood flow signal is updated, all pixels marked as invalid data are removed, and only pixels not marked as invalid are retained.

[0084] The updated pixel set only contains pixels of real blood flow signals, and retains the spatial position coordinates and signal intensity information of these pixels.

[0085] The de-artifacted image is generated based on the pixel data of the retained true blood flow signal and its corresponding spatial distribution. The de-artifacted image fully records the true blood flow distribution and signal intensity in the placenta area: Based on the retained true blood flow signal pixel data, the pixel distribution in the image is reconstructed according to its spatial coordinates to generate a complete de-artifacted image. The generated de-artifacted image is stored in a standard image format, including the signal intensity and spatial distribution information of each pixel in the image, for subsequent in-depth analysis of the placenta image.

[0086] Morphological processing based on structural elements is performed on the de-artifacted images to extract the continuous boundary information between the placenta and the uterine wall, and to identify suspicious vascular invasion paths. Specifically, the following steps are performed: Perform binary processing on the de-artifacted image, set the grayscale threshold to separate the placenta area and the background area, and generate a preliminary binary image: The de-artifacted image is input. This image is obtained by removing the pixels of the multi-reflection interference area in the previous step, and only retaining the true signal of the placenta area and its adjacent blood vessels.

[0087] Analyze the grayscale histogram of the de-artifacted image and select an appropriate grayscale threshold to separate the placenta region from the background region. The grayscale threshold is dynamically set based on the mean and standard deviation of the image grayscale distribution to accurately distinguish the grayscale characteristics of the placenta region from the background region.

[0088] For each pixel in the de-artifacted image, its grayscale value is compared with a set threshold. If the grayscale value is higher than the threshold, the pixel is set as the foreground (placenta region); if it is lower than the threshold, it is set as the background. The resulting preliminary binary image is used for subsequent morphological processing.

[0089] The size and shape of the structural element are designed according to the characteristics of the binary image. The structural element is used in morphological operations to eliminate residual artifacts: The size of the structuring element is selected based on the pixel distribution characteristics of the placenta region in the binary image. The size of the structuring element should cover the small-sized features in the artifact residual area while avoiding affecting the normal structure of the placenta region.

[0090] Depending on the morphological characteristics of the placental region boundary, a rectangular, elliptical, or cross-shaped structuring element was selected. For regions with complex boundaries, a structuring element shape with smooth edges was preferred to optimize the processing effect.

[0091] Use the image processing library to generate the required structuring element, ensuring that its size and shape meet the processing requirements of binary images. The structuring element is used in subsequent opening and closing operations.

[0092] The morphological opening operation is used to remove isolated small areas at the boundary of the placenta region and enhance the boundary continuity between the placenta and the uterine wall: The morphological opening operation involves performing an erosion operation followed by a dilation operation on the binary image. The erosion operation removes isolated small regions at the boundary, while the dilation operation restores the overall shape of the placenta region, ultimately enhancing the continuity of the boundary.

[0093] An opening operation is applied to the binary image, processing the image pixel by pixel using the previously designed structuring element. This operation eliminates noise and residual artifacts at the placenta's boundaries, preserving the primary structural contours. The resulting binary image, processed by the opening operation, exhibits smoother boundaries and effectively removes small, isolated regions.

[0094] Morphological closing operations are used to fill the gaps inside the placenta region and optimize the boundary connectivity between the placenta region and the uterine wall: The morphological closing operation involves dilation followed by erosion of the binary image. The dilation operation fills the gaps inside the placenta region, while the erosion operation restores the precise outline of the boundary.

[0095] A closing operation is applied to the binary image after the opening operation, using the previously designed structuring element. This closing operation further enhances the connectivity between the placenta region and the uterine wall boundary and fills small gaps within the placenta region. The resulting closed binary image has a more complete placenta region boundary, making it suitable for subsequent edge detection operations.

[0096] The edge detection algorithm is used to extract the continuous boundary information between the placenta and the uterine wall. The continuous boundary information includes the outer contour of the placenta area and its contact interface with the uterine wall: Select a gradient-based edge detection algorithm, such as the Canny edge detection algorithm. This algorithm extracts edges by detecting the maximum value of the gradient of grayscale changes.

[0097] An edge detection algorithm was applied to the binary image after closing operation to extract the continuous boundary information between the placenta area and the uterine wall, including the outer contour of the placenta and the interface with the uterine wall.

[0098] The extracted boundary information is recorded as a set containing the spatial positions of pixel points for subsequent partition identification and analysis.

[0099] Based on the continuous boundary information, suspicious vascular invasion paths are identified by partitioning. Specifically, the irregularities of the boundaries and the overlapping areas with the vascular signals are analyzed, and the spatial position and boundary characteristics of each partition are recorded to form the identification results: Based on the extracted continuous boundary information, the interface between the placenta region and the uterine wall is partitioned. Partitioning rules include boundary irregularities and overlap with vascular signals.

[0100] The continuous boundary is divided into several regions, and the boundary characteristics of each region are analyzed one by one. For example, if a region's boundary has a large protrusion or depression that overlaps with a blood vessel signal, the region is marked as a suspected vascular invasion path.

[0101] The spatial position, boundary morphological characteristics and relationship with vascular signals of each partition are recorded, and identification results are generated for subsequent analysis and evaluation.

[0102] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0103] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in the embodiments of this application are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0104] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0105] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0106] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0107] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.

[0108] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0109] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0110] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

[0111] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A placenta accreta analysis method based on ultrasound imaging, characterized in that: The steps include: Acquire ultrasound radiofrequency data covering the placenta area and adjacent blood vessels, and generate two-dimensional grayscale images and Doppler blood flow images; Construct a dynamic interference distribution map of multiple frames of ultrasound radiofrequency data, extract the temporal characteristics of artifact signals in the placenta region, and mark the artifact dynamic characteristic areas; Wavelet transform is used to separate the dynamic characteristic areas of artifacts in the frequency domain on the two-dimensional grayscale image, and a pixel set of the multi-reflection interference area is generated; Adaptive threshold estimation algorithm is used for Doppler blood flow images to identify real blood flow signals using gray intensity information and blood flow velocity information; The pixel set of the multi-reflection interference area and the real blood flow signal are fused, and the pixels generated by the multi-reflection interference are set as invalid data to generate a de-artifacted image; Morphological processing based on structural elements was performed on the de-artifacted images to extract the continuous boundary information between the placenta and the uterine wall, and to identify the suspicious vascular invasion paths.

2. The method for analyzing placenta accreta based on ultrasound imaging according to claim 1, wherein: Acquires ultrasound radiofrequency data covering the placenta and adjacent blood vessels, generating two-dimensional grayscale images and Doppler blood flow images, including: An ultrasound device is used to collect radio frequency signals from the placenta and adjacent blood vessels. The ultrasound probe scans the target area at a fixed frequency and angle, covering the placenta and its adjacent blood vessels. A two-dimensional grayscale image is generated by processing the amplitude and phase of the radio frequency signal. The two-dimensional grayscale image is used to display the structural distribution of the placenta tissue and its adjacent blood vessels. A Doppler blood flow image is generated based on the Doppler frequency shift information in the radio frequency signal. The Doppler blood flow image is used to reflect the blood flow velocity and direction in the placenta area and adjacent blood vessels.

3. The method for analyzing placenta accreta based on ultrasound imaging according to claim 1, wherein: Construct a dynamic interference distribution map of multiple frames of ultrasound radiofrequency data, extract the temporal characteristics of the artifact signal in the placenta area, and mark the artifact dynamic characteristic area, including: Acquiring multiple frames of continuous ultrasound radio frequency data including the placenta region and adjacent blood vessels, where the ultrasound radio frequency data includes amplitude information and phase information; Perform time-series demodulation on multiple frames of ultrasound radio frequency data, extract the amplitude and phase change curves of the signal in the placenta area, and form initial time-series data reflecting the dynamic characteristics of the artifact signal; A dynamic interference distribution map is constructed based on the initial time series data reflecting the dynamic characteristics of the artifact signal. The dynamic interference distribution map records the time series characteristics of the artifact signal in the placenta region and adjacent blood vessels, including the changing frequency, intensity, and spatial distribution of the artifact signal. The timing regularity of the artifact signal is analyzed through the dynamic interference distribution diagram, and the dynamic characteristic area of ​​the artifact is marked.

4. The method for analyzing placenta accreta based on ultrasound imaging according to claim 1, wherein: Wavelet transform is used to separate the dynamic characteristic area of ​​the artifact in the frequency domain of the two-dimensional grayscale image, and a pixel set of the multi-reflection interference area is generated, which includes: Based on the marked artifact dynamic characteristic region, pixel data of the corresponding two-dimensional grayscale image is intercepted; Wavelet transform is used to decompose the pixel data of the two-dimensional grayscale image, and the two-dimensional grayscale image signal is decomposed into multiple scale and frequency sub-bands to extract the high-frequency components related to multiple reflection interference; The high-frequency subbands of wavelet decomposition are screened by threshold denoising method; The filtered high-frequency sub-bands are inversely transformed to generate a pixel set containing the multi-reflection interference area, which is used to record the spatial distribution and intensity information of the multi-reflection interference area.

5. The method for analyzing placenta accreta based on ultrasound imaging according to claim 1, wherein: An adaptive threshold estimation algorithm is used for Doppler blood flow images to identify real blood flow signals using grayscale intensity information and blood flow velocity information. Specifically, the algorithm includes: Extract the grayscale intensity value and corresponding blood flow velocity value of each pixel from the Doppler blood flow image. The grayscale intensity value reflects the echo signal intensity of the pixel point, and the blood flow velocity value represents the blood flow velocity of the pixel point. Calculating the grayscale intensity distribution of the entire image based on the extracted grayscale intensity values, including the mean and standard deviation of the grayscale intensity values; Calculating velocity distribution characteristics of the entire image based on the extracted blood flow velocity values, including the mean and standard deviation of the blood flow velocity values; Combining the gray intensity distribution and velocity distribution characteristics, an adaptive threshold estimation algorithm is used to calculate the gray intensity threshold and velocity threshold respectively. The pixels in the Doppler blood flow image are screened, and only the pixels with grayscale intensity values ​​higher than the grayscale intensity threshold and blood flow velocity values ​​higher than the velocity threshold are retained and identified as pixels with real blood flow signals.

6. The method for analyzing placenta accreta based on ultrasound imaging according to claim 1, wherein: The pixel set of the multi-reflection interference area and the real blood flow signal are fused, and the pixels generated by the multi-reflection interference are set as invalid data to generate a de-artifacted image. Specifically, the following steps are performed: Based on the pixel set of the multi-reflection interference area, the spatial position coordinates and interference signal intensity of the corresponding pixels are extracted; Based on the pixel set of the real blood flow signal, the spatial position coordinates and blood flow signal intensity of the corresponding pixels are extracted; Matching analysis is performed on the pixel set of the multi-reflection interference area and the pixel set of the real blood flow signal. The matching rules include the degree of overlap of spatial position coordinates and the relative difference in signal intensity. Pixels belonging to the multiple reflection interference area in the matching analysis results are set as invalid data. Invalid data is defined as the signal strength of which is mainly caused by multiple reflection interference. The filtered invalid data is removed from the fused pixel set, and only the pixel data of the real blood flow signal is retained; A de-artifacted image is generated based on the retained pixel data of the real blood flow signal and its corresponding spatial distribution. The de-artifacted image fully records the real blood flow distribution and signal intensity in the placenta area.

7. The method for analyzing placenta accreta based on ultrasound imaging according to claim 1, wherein: Morphological processing based on structural elements is performed on the de-artifacted images to extract the continuous boundary information between the placenta and the uterine wall, and to identify suspicious vascular invasion paths. Specifically, the following steps are performed: The de-artifacted image was binarized, and the grayscale threshold was set to separate the placenta area and the background area to generate a preliminary binary image; The size and shape of the structure element are designed according to the characteristics of the binary image, and the structure element is used in morphological operations to eliminate residual artifacts; Morphological opening operation is used to remove isolated small areas at the boundary of the placenta region and enhance the boundary continuity between the placenta and the uterine wall; Morphological closing operations are used to fill the gaps inside the placenta region and optimize the boundary connectivity between the placenta region and the uterine wall. Apply edge detection algorithm to extract continuous boundary information between the placenta and uterine wall. The continuous boundary information includes the outer contour of the placenta area and its contact interface with the uterine wall. Based on the continuous boundary information, the suspicious vascular invasion path is partitioned and identified, which specifically includes analyzing the irregularity of the boundary and the overlapping area with the vascular signal, and recording the spatial position and boundary characteristics of each partition to form the identification result.

Images