Method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging

Abstract

This invention discloses a diagnostic method for stress urinary incontinence based on dual-plane ultrasound multimodal imaging, belonging to the field of medical ultrasound imaging technology. The method first establishes a dual-plane imaging environment, using high-frame-rate grayscale image sequences of a geometric reference plane to calculate the instantaneous motion vectors of anatomical structures, real-time correcting the coordinates of the region of interest within the functional parameter plane, and simultaneously extracting tissue elastic modulus and microblood flow data over the entire time period. Subsequently, confidence weights are generated based on physical motion rates, and stiffness-blood flow coupling model parameters are identified within a static learning window. For motion artifact regions, adaptive reconstruction of microblood flow signals is performed using physical deduction or numerical preservation strategies. Finally, a multidimensional risk assessment model is constructed by combining the hysteresis loop area, the microcirculation functional ischemia index, and the patient's physiological characteristics. This invention effectively solves the problems of functional imaging localization loss and signal distortion under dynamic load, achieving synchronous dynamic quantitative assessment of pelvic floor tissue structure and function.

Description

Technical Field

[0001] This invention relates to the field of medical ultrasound image processing and computer-aided diagnostic technology, specifically a method for diagnosing stress urinary incontinence based on biplane ultrasound multimodal imaging. Background Technology

[0002] Stress urinary incontinence is a common pelvic floor dysfunction in women. Its pathological mechanism mainly involves defects in the urethral support structures and abnormalities in the function of the urethral sphincter. In order to accurately assess the condition and develop a personalized treatment plan, it is necessary to dynamically assess the morphology and functional status of the pelvic floor tissues in patients at rest and under stress loads such as the Valsalva maneuver or coughing, in order to capture the pathophysiological changes at the moment of urinary incontinence.

[0003] Currently, ultrasound imaging has become an important imaging tool for assessing pelvic floor function due to its advantages such as being radiation-free and having high real-time performance. Existing techniques typically use two-dimensional high-frequency ultrasound or three-dimensional / four-dimensional volumetric ultrasound to observe anatomical morphological indicators such as bladder neck mobility and urethral rotation angle. In addition, with the development of ultrasound technology, functional modalities such as shear wave elastography and high-sensitivity microvascular flow imaging have been gradually introduced, aiming to assist in analyzing the biomechanical properties and microcirculatory function of pelvic floor muscles and connective tissues by measuring the stiffness parameters of periurethral tissues and the perfusion level of microvessels.

[0004] While existing technologies have played a significant role in static or quasi-static assessment of the pelvic floor, several limitations remain. The characteristic pathological changes in stress urinary incontinence often occur during dynamic moments of rapid increases in abdominal pressure. Existing functional ultrasound imaging modalities typically require long pulse sequences and complex post-processing to acquire high-quality parametric images, resulting in a significantly lower frame rate compared to conventional B-mode grayscale imaging. This low temporal resolution makes it impossible for a single probe or single-plane imaging to accurately track large displacement anatomical structures in real-time when faced with rapid pressure loads from the patient. This leads to deviations of the region of interest from pre-defined anatomical landmarks, causing spatiotemporal misalignment of measurement data. Furthermore, microflow imaging relies on the detection of motion signals. During strenuous exercise, the macroscopic motion speed of surrounding soft tissues often far exceeds the low-speed blood flow within microvessels. This high-intensity Doppler shift, or scintillation artifact, generated by tissue motion is difficult to filter out with conventional filters, directly masking effective blood flow signals or causing severe signal distortion. This makes it difficult for clinicians to obtain accurate tissue perfusion data at critical stress moments and to establish a quantitative correlation between structural deformation and functional impairment. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging. This method solves the technical problems of anatomical structure tracking loss due to frame rate limitations and motion artifact interference with microblood flow signal measurement in dynamic assessment of stress load using existing functional ultrasound imaging.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging, comprising the following steps: S1. Establish a dual-plane imaging environment, define the long axis section of the probe as the geometric reference plane and the short axis section of the probe as the functional parameter plane, and synchronously trigger the acquisition of grayscale image sequences in the geometric reference plane and the original elastic modulus sequence and original microblood flow sequence in the functional parameter plane in the two planes to establish a unified time reference. S2, using grayscale image sequences to calculate the instantaneous motion vector of anatomical structures, mapping the instantaneous motion vector to the functional parameter plane to correct the coordinate position of the region of interest in real time, and extracting one-to-one corresponding full-time tissue elastic modulus data and microblood perfusion data based on the corrected coordinate position; S3, calculate the magnitude of the instantaneous motion vector and convert it into physical motion rate, generate dynamically changing confidence weights based on the physical motion rate, and identify continuous time periods with confidence weights higher than a preset high confidence threshold as learning windows; S4, within the learning window, analyze the correlation between tissue elastic modulus data and microvascular perfusion data, and determine the signal reconstruction strategy as physical deduction or numerical preservation based on whether the correlation meets the negative correlation condition. S5, for full-time microblood flow perfusion data, uses confidence weights to perform weighted fusion of the original measurement data and the alternative reference signal generated according to the signal reconstruction strategy, and outputs the reconstructed microblood flow signal after motion artifact repair. S6. The reconstructed microblood flow signal and tissue elastic modulus data of the urethral orifice region are mapped to a two-dimensional coordinate system to construct a hysteresis loop. The area of ​​the hysteresis loop is combined with the SUI risk index calculated based on the patient's physiological characteristics and dynamic imaging characteristics to generate an auxiliary assessment conclusion.

[0007] Preferably, in step S1, establishing the dual-plane imaging environment specifically includes: A dual-plane ultrasound probe integrating two independent piezoelectric transducer arrays is used, covering a frequency range of 3 MHz to 13 MHz. The first piezoelectric transducer array is linearly arranged along the long axis of the probe, corresponding to the geometric reference plane, while the second piezoelectric transducer array is linearly arranged along the short axis of the probe, corresponding to the functional parameter plane. The two arrays are fixed at a 90-degree orthogonal angle in physical space. A pre-calibrated spatial transformation matrix is ​​used to store the spatial mapping relationship between the geometric reference plane and the functional parameter plane. The spatial transformation matrix consists of a rotation matrix component describing angular deviation and a translation vector component describing the distance deviation of the physical center point. The imaging depth is set to 2 cm to 4 cm, and the focusing area is set to 1 cm to 2 cm from the probe surface. Patient physiological characteristic data, including age, height, and weight, are recorded before data acquisition.

[0008] Preferably, in step S1, the specific logic for synchronously triggering data acquisition is as follows: A B-mode ultrasound pulse sequence with a transmission center frequency of 3 MHz to 12 MHz and a frame rate of 40 to 100 frames per second is configured in the geometric reference plane. A shear wave elastography pulse sequence and a micro-blood flow imaging pulse sequence are simultaneously configured in the functional parameter plane. The ensemble length of the micro-blood flow imaging pulse sequence is set to 10 to 20 pulses, and the combined frame rate of the two is set to 2 to 10 frames per second. Asymmetric time-triggered logic is executed, continuously triggering 5 to 20 acquisitions in the geometric reference plane within one acquisition cycle, followed by one acquisition in the functional parameter plane. A unified discrete time series is established, and for each moment, the data acquisition timestamp of the functional parameter plane is recorded.

[0009] Preferably, step S2 specifically includes: The pixel displacement of feature tracking points in the geometric reference plane is tracked using a multi-scale pyramid optical flow algorithm to obtain the original two-dimensional instantaneous motion vector. A 2x2 cross-plane projection matrix is ​​constructed, and the original two-dimensional instantaneous motion vector is multiplied by the cross-plane projection matrix to obtain the corrected displacement vector in the functional parameter plane. The center coordinates of the region of interest in the functional parameter plane are iteratively updated using the corrected displacement vector and the anisotropic scaling factor, where the anisotropic scaling factor is the ratio of the pixel physical resolution of the geometric reference plane to that of the functional parameter plane. In the original elastic modulus sequence and the original microblood flow sequence in the functional parameter plane, a sampling window of size 10 pixels by 10 pixels is defined, and the arithmetic mean of the data in the sampling window is calculated using a bilinear interpolation algorithm as the extracted data.

[0010] Preferably, step S3 specifically includes: By combining the average pixel physical resolution of the geometric reference plane with the time interval between the acquisition of two adjacent frames, the pixel displacement distance of the instantaneous motion vector is converted into a physical motion rate in millimeters per second. A confidence scoring model based on the Gaussian decay function is constructed, and the confidence weight is calculated. The weight is equal to the exponent of the natural constant, and the base of the exponent is the square of the physical motion rate divided by the negative value of twice the square of the motion sensitivity control parameter. The motion sensitivity control parameter is set to 2 mm / s to 5 mm / s. The high confidence threshold is set to 0.85 to 0.95. Only when the duration of time when the confidence weight is continuously marked as higher than the threshold exceeds 0.5 seconds to 1.0 seconds is the continuous time period confirmed as the learning window.

[0011] Preferably, step S4 specifically includes: Calculate the Pearson correlation coefficient between tissue elastic modulus data and microvascular perfusion data within the learning window; If the Pearson correlation coefficient is less than the negative correlation validity threshold set between -0.4 and -0.6, the physical coupling is deemed effective, and the signal reconstruction strategy is determined to be physical deduction. If the physical coupling is deemed effective, a nonlinear physical model is established, in which the microvascular perfusion density value is equal to the perfusion volume coefficient multiplied by the tissue elastic modulus value raised to the power of the negative vascular compliance exponent. The natural logarithm of the model is taken to transform it into a linear regression form, and the perfusion volume coefficient and vascular compliance exponent are identified using the least squares method. If the Pearson correlation coefficient is greater than or equal to the negative correlation validity threshold, the physical coupling is deemed ineffective, the signal reconstruction strategy is determined to be numerical preservation, and the arithmetic mean of the microvascular perfusion data within the learning window is calculated as the baseline perfusion constant.

[0012] Preferably, step S5 specifically includes: Generate alternative reference signals for the entire time period: When the strategy is physical deduction, the original tissue elastic modulus data at each time step are substituted into the identified nonlinear physical model to calculate the alternative reference microblood flow value; when the strategy is value preservation, the baseline perfusion constant is directly taken as the alternative reference microblood flow value; perform linear weighted fusion: for each time step, the reconstructed microblood flow signal is equal to the confidence weight at that time step multiplied by the original microblood flow perfusion data, plus 1 minus the difference of the confidence weight at that time step multiplied by the alternative reference microblood flow value; perform Savitzky-Golay filtering on the generated reconstructed microblood flow signal sequence, setting the filtering window length to an odd number between 5 and 9, and the polynomial fitting order to be between 2nd and 3rd order.

[0013] Preferably, when calculating the SUI risk index in step S6, the extracted dynamic image features include cumulative displacement and displacement velocity variance. The cumulative displacement is the sum of the Euclidean norms of the coordinate vector of the center of the region of interest up to the current time relative to the coordinate vector at the initial time; the displacement velocity variance is the root mean square error of the instantaneous velocity magnitude relative to the average velocity magnitude within a sliding window of 5 to 10 sampling points.

[0014] Preferably, in step S6, generating auxiliary assessment conclusions specifically includes calculating the microcirculation functional ischemia index and the urethral mobility-stiffness ratio: The full-time data was divided into a resting baseline interval and a pressure peak interval. The microcirculation functional ischemia index was calculated as follows: the arithmetic mean of the reconstructed microblood flow signals within the resting baseline interval minus the instantaneous value of the reconstructed microblood flow signals at the maximum load time, and then divided by the arithmetic mean of the reconstructed microblood flow signals within the resting baseline interval. The result was expressed as a percentage. The urethral mobility-rigidity ratio was calculated as follows: the maximum cumulative displacement at the end of the pressure load phase divided by the tissue elastic modulus value at the maximum load time.

[0015] Preferably, in step S6, the SUI risk index is calculated using a multi-parameter linear weighted evaluation model: The model multiplies the ratio of the functional ischemia index of microcirculation to the preset physiological reference ischemia index by a microcirculation dimension weighting coefficient, and adds the ratio of the urethral mobility-rigidity ratio to the preset physiological reference mobility-rigidity ratio by a structural mechanics dimension weighting coefficient. The microcirculation dimension weighting coefficient and the structural mechanics dimension weighting coefficient are set based on the contribution of clinical pathophysiology, with values ​​ranging from 0.4 to 0.6 and a sum of 1. The preset physiological reference ischemia index is set based on statistical data from healthy individuals, with values ​​ranging from 30% to 40%. The preset physiological reference mobility-rigidity ratio is set based on statistical data from healthy individuals, with values ​​ranging from 0.5 mm / kPa to 1.0 mm / kPa. Finally, the calculated SUI risk index is compared with a preset risk grading threshold to generate a corresponding low-risk, medium-risk, or high-risk classification result.

[0016] This invention provides a diagnostic method for stress urinary incontinence based on dual-plane ultrasound multimodal imaging. It has the following beneficial effects: 1. This invention solves the problem of region of interest loss under dynamic load in traditional functional imaging by establishing an orthogonal dual-plane imaging system and utilizing cross-plane motion tracking technology. It captures the instantaneous motion vectors of anatomical structures using a high-frame-rate geometric reference plane and corrects the coordinate position of the functional parameter plane in real time through a spatial projection matrix. This ensures that even when the patient undergoes macroscopic body movement, the system can still continuously and accurately lock onto key anatomical sites such as the internal urethral orifice and mid-urethra, achieving precise sampling of the pelvic floor tissues from rest to maximum load.

[0017] 2. This invention proposes a signal adaptive reconstruction strategy based on a stiffness-blood flow coupling physical model, effectively suppressing scintillation artifacts commonly found in Doppler ultrasound during motion. Utilizing the motion-insensitive nature of tissue elastic modulus, when low-confidence motion artifact regions are detected, the true microvascular perfusion signal is reconstructed using patient-specific stiffness-blood flow coupling parameters and elastic modulus data. This method avoids information gaps caused by simply removing artifact data, providing continuous and physiologically sound data support for assessing the dynamic response of pelvic floor microcirculation under pressure.

[0018] 3. This invention constructs a multidimensional quantitative assessment system that includes the microcirculation functional ischemia index, hysteresis loop area, and SUI risk index, overcoming the limitations of single morphological diagnosis. It performs spatiotemporal alignment and fusion analysis of the macroscopic mobility of anatomical structures, tissue biomechanical stiffness, and microvascular bed blood perfusion function. By calculating the hysteresis loop area, which characterizes the degree of hysteresis loop recovery lag, and the degree of ischemia under pressure, it can comprehensively quantify the pathological risk of stress urinary incontinence from both structural damage and functional compensation dimensions, providing objective auxiliary diagnostic basis for developing personalized treatment plans in clinical practice. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging according to the present invention; Figure 2 This is a schematic diagram illustrating the orthogonal dual-plane imaging environment construction and data flow timing control of the present invention; Figure 3 This is a schematic diagram of the cross-plane motion tracking and region of interest dynamic correction process of the present invention; Figure 4 This is a schematic diagram of the motion-dependent confidence weight generation and learning window recognition process of the present invention; Figure 5 This is a schematic diagram of the stiffness-blood flow coupling feature analysis and parameter identification process of the present invention; Figure 6 This is a schematic diagram of the confidence-weighted adaptive reconstruction process of micro-blood flow signals according to the present invention; Figure 7 This is a schematic diagram of the multidimensional quantitative feature extraction and stress urinary incontinence risk assessment process of the present invention; Figure 8 This is a comparison diagram of the temporal variation of multimodal signals and the microblood flow reconstruction effect of the present invention; Figure 9 This is a schematic diagram of the tissue stiffness-microblood flow physical coupling model fitting of the present invention; Figure 10 This is a diagram illustrating the risk grading assessment and clinical sample distribution validation of stress urinary incontinence according to the present invention. Detailed Implementation

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] See attached document Figure 1 , Figure 1 This is a flowchart of a method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging according to an embodiment of the present invention. The present invention provides a method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging, comprising the following steps: S1. Establish a dual-plane imaging environment, input or obtain the patient's physiological characteristic data from the electronic medical record system, define the long axis section of the probe as the geometric reference plane, define the short axis section of the probe as the functional parameter plane, and lock the scanning center of the functional parameter plane to the internal urethral orifice and the mid-urethral region based on clinical pathological characteristics. Simultaneously trigger the acquisition of grayscale image sequences in the geometric reference plane and the original elastic modulus sequence and original microblood flow sequence in the functional parameter plane in both planes to establish a unified time reference. S2 uses the grayscale image sequence obtained in S1 to calculate the instantaneous motion vector of the anatomical structure, and maps the vector to the functional parameter plane to correct the coordinate position of the region of interest in real time. Based on the corrected coordinates at each moment, one-to-one corresponding full-time tissue elastic modulus data and microblood flow perfusion data are extracted from the original elastic modulus sequence and the original microblood flow sequence, which naturally covers the patient's static and dynamic states. Dynamic image features are calculated in this process.

[0022] S3, calculate the magnitude of the instantaneous motion vector obtained from S2, and convert it into a physical motion rate by combining the time interval and physical resolution; based on this motion rate, quantify the micro-blood flow data to generate dynamically changing confidence weights; identify continuous time periods with confidence weights higher than a preset high confidence threshold as learning windows for physical model learning; S4. Within the learning window defined in S3, select the tissue elastic modulus data and microblood flow perfusion data extracted in S2 as the sample set; analyze the correlation between the two in the sample set, and determine the signal reconstruction strategy based on whether the correlation meets the negative correlation condition: if the negative correlation condition is met, the strategy is determined to be physical deduction, and the stiffness-blood flow coupling model parameters that characterize the patient specificity are identified and passed to the next step; if the negative correlation condition is not met, the strategy is determined to be numerical preservation. S5, for the full-time microblood flow perfusion data including static and dynamic states extracted in S2, check the confidence weight at each time point: in the time interval where the confidence weight is higher than the preset threshold, retain the original measurement data; in the motion artifact interval where the confidence weight is lower than the preset threshold, execute the signal reconstruction strategy determined in step S4, and finally output a reconstructed microblood flow signal repaired by motion artifacts. In step S6, the reconstructed microblood flow signal output from step S5 in the urethral orifice region is selected and mapped to a two-dimensional coordinate system with the tissue elastic modulus data extracted in step S2 to construct a hysteresis loop. The geometric area representing the degree of hysteresis in vascular bed recovery is calculated. Simultaneously, a multidimensional feature vector is constructed by combining the patient's physiological characteristic data obtained in step S1 and the dynamic image features extracted in step S2. This vector is then substituted into a pre-set LASSO logistic regression model to calculate the SUI risk index. Finally, the hysteresis loop area and the risk index are combined to generate a pelvic floor microcirculation function impairment index as auxiliary electronic assessment data.

[0023] See attached document Figure 2 , Figure 2 This is a schematic diagram of orthogonal dual-plane imaging environment construction and data flow timing control according to an embodiment of the present invention. To achieve synchronous dynamic assessment of the pelvic floor tissue structure and function in patients with stress urinary incontinence, step S1, orthogonal dual-plane initialization and spatiotemporal alignment, specifically includes the following steps: S101. Connect the dual-plane ultrasound probe to the ultrasound data acquisition device, load the preset parameters for pelvic floor scanning, set the imaging depth to 2 cm to 4 cm, and set the focusing area to 1 cm to 52 cm from the probe surface. Instruct the patient to assume the lithotomy position to fully expose the perineum. Apply ultrasound coupling gel to the probe surface and place the probe on the perineal body surface between the labia minora and below the external urethral orifice. Adjust the probe angle and pressure until the real-time image clearly shows the lower edge of the pubic symphysis and the bladder neck structure.

[0024] The system is configured with a data entry interface that connects to the hospital's electronic medical record system via the DICOM protocol or HL7 standard interface. This interface automatically reads or manually enters the patient's physiological characteristic data, which includes at least the patient's age, height, and weight. This data is then stored in the memory unit for subsequent evaluation. A dual-plane ultrasound probe integrating two independent piezoelectric transducer arrays is used as the acquisition front end, covering a frequency range of 3 MHz to 13 MHz. The first piezoelectric transducer array is linearly arranged along the long axis of the probe, and the second piezoelectric transducer array is linearly arranged along the short axis of the probe. The two arrays are physically orthogonal and fixed at 90 degrees.

[0025] The imaging plane corresponding to the first piezoelectric transducer array is defined as the geometric reference plane, and this plane is configured to display the two-dimensional grayscale anatomical structure of the pubic symphysis, bladder neck, and the entire length of the urethra in a sagittal section. The imaging plane corresponding to the second piezoelectric transducer array is defined as the functional parameter plane, and this plane is configured to display the urethral sphincter and periurethral vascular plexus in a transverse section.

[0026] Since the relative positions of the two arrays are fixed, a definite spatial mapping relationship exists between the geometric reference plane and the functional parameter plane. This relationship is stored using a pre-calibrated spatial transformation matrix. This spatial transformation matrix is ​​a 4x4 homogeneous transformation matrix, consisting of a 3x3 rotation matrix component and a 3x1 translation vector component. The rotation matrix component describes the minute angular deviation between the two imaging planes due to manufacturing processes and the theoretical 90-degree orthogonality; the translation vector component describes the straight-line distance deviation between the physical center points of the two piezoelectric transducer arrays in three-dimensional space, typically on the order of millimeters. This spatial transformation matrix is ​​used to convert the pixel coordinates in the geometric reference plane into spatial coordinates in the functional parameter plane. The specific algorithms for homogeneous transformation matrix calibration and coordinate transformation can be implemented using existing stereo geometric calibration techniques, which are well-known in the field and will not be elaborated upon here.

[0027] S102, on the geometric reference plane, is configured with a high frame rate B-mode ultrasound pulse sequence, with the transmission center frequency set to 3 MHz to 12 MHz and the frame rate set to 40 to 100 frames per second to capture rapid anatomical movements. On the functional parameter plane, shear wave elastography pulse sequences and microflow imaging pulse sequences are simultaneously configured. The shear wave elastography pulse sequence uses acoustic radiation force pulses to excite tissue to generate shear waves and uses ultrafast plane waves for tracking and detection; the microflow imaging pulse sequence uses multi-angle plane wave transmission and repeatedly acquires data at the same location to form a Doppler ensemble data packet, with the ensemble length set to 10 to 20 pulses, and the combined frame rate of the two is set to 2 to 10 frames per second.

[0028] The asymmetric time-triggered logic is executed, that is, within one acquisition cycle, the acquisition of the geometric reference plane is triggered continuously an integer number of times, followed by one acquisition of the functional parameter plane. The specific value of the preset integer is 5 to 20, in order to balance the high temporal resolution requirements of anatomical tracking and the high energy processing requirements of functional imaging.

[0029] S103. A unified discrete time series is established based on the acquisition time of the functional parameter plane with a lower frame rate. For each moment in the discrete time series, the data acquisition timestamp of the functional parameter plane is recorded, and the image frame with the smallest absolute difference from that timestamp is retrieved from the high frame rate data stream of the geometric reference plane as the corresponding anatomical reference frame. The maximum allowable time deviation is set to 20 milliseconds.

[0030] After initialization and alignment are completed, a set of original data containing three dimensions is output at each time step, defined as follows: ; In the formula, :express A time-synchronized multimodal data set at any given moment; : Represents a two-dimensional pixel coordinate vector within the geometric reference plane, including horizontal and depth coordinates; :express The grayscale pixel intensity value at a specific coordinate on the geometric reference plane at any given time, with a value range of 0 to 255; : Represents a two-dimensional pixel coordinate vector within the functional parameter plane; :express The original Young's modulus value measured at a specific coordinate on the functional parameter plane at all times is calculated based on the product of the square of the measured shear wave propagation velocity and the preset soft tissue density constant, usually taken as 1000 kg per cubic meter, and the unit is kilopascal. :express The raw micro-blood flow perfusion density value is measured at a specific coordinate on the functional parameter plane at any time. This value is obtained by autocorrelation processing of the Doppler ensemble signal after removing tissue clutter, and it characterizes the blood flow energy intensity per unit volume.

[0031] Through the above steps, physical spatial registration and temporal synchronization of the dual-plane data were achieved, providing standardized data input for subsequent processing.

[0032] See attached document Figure 3 , Figure 3 This is a schematic diagram of a cross-plane motion tracking and region-of-interest dynamic correction process according to an embodiment of the present invention. Step S2 calculates the instantaneous motion vector of the anatomical structure using the grayscale image sequence acquired in S1, and performs cross-plane correction and data extraction, specifically including the following steps: S201, calculate the instantaneous motion vector of the anatomical structure in the grayscale image sequence of the geometric reference plane. Select the first frame image in the geometric reference plane as the initial reference frame, and use a gradient-based edge detection algorithm to identify the junction between the internal urethral orifice and the posterior lip of the bladder neck, or mark this location as a feature tracking point through human-computer interaction.

[0033] A multi-scale pyramid Lucas-Kanade optical flow algorithm is employed to track the positional changes of feature tracking points in a continuous time series. A Gaussian image pyramid with 3 to 5 layers is constructed, and optical flow is calculated layer by layer from coarse to fine to capture large displacement motions. For the current image and the previous image, the pixel displacement components of the feature tracking points in the horizontal and vertical directions are calculated and combined to obtain the original two-dimensional instantaneous motion vector in the geometric reference plane. The specific iterative process and constraint solution of the multi-scale optical flow algorithm can be implemented using existing computer vision motion estimation algorithms, which are well-known techniques in the field and will not be elaborated upon here.

[0034] S202, construct a spatial projection matrix to map the original two-dimensional instantaneous motion vectors in the geometric reference plane to the functional parameter plane. Based on the biplane orthogonal spatial relationship established in step S1, construct a cross-plane projection matrix with a dimension of 2 rows and 2 columns. The main diagonal elements of this matrix represent the coupling coefficients of the same physical axis, and the off-diagonal elements represent the projection coefficients of the orthogonal axes. In the orthogonal configuration of this embodiment, the first row and first column element of the matrix are set to 0, indicating that the horizontal motion of the geometric plane is not mapped to the horizontal axis of the functional plane; the second row and second column element of the matrix are set to 1, indicating that the vertical depth motion of the geometric plane is mapped to the vertical depth axis of the functional plane at a 1:1 ratio.

[0035] Multiplying the original two-dimensional instantaneous motion vector calculated by S201 by the transplane projection matrix yields the corrected displacement vector in the functional parameter plane. This corrected displacement vector characterizes the physical offset of the region of interest in the functional parameter plane caused by patient breathing, Valsalva maneuvers, or probe jitter.

[0036] S203 utilizes the correction displacement vector to correct the coordinate position of the region of interest (ROI) within the functional parameter plane in real time, and extracts multimodal data across all time periods. The initial coordinates of the ROI center within the functional parameter plane are set to a preset fixed value. For each subsequent time step, the ROI center coordinates are iteratively updated using the following formula: ; In the formula, Indicates the current time The corrected coordinate vector of the center of the region of interest in the functional parameter plane, containing the x-coordinate and y-coordinate values, in pixels; : Indicates the previous moment The corrected coordinate vector of the center of the region of interest within the functional parameter plane; : Represents a 2x2 cross-plane projection matrix, containing coefficients describing the physical motion coupling relationship between two planes; : Indicates the current time The original two-dimensional instantaneous motion vector calculated within the geometric reference plane includes horizontal and vertical components; : Represents the anisotropic scaling factor, which is the ratio of the pixel physical resolution of the geometric reference plane to the pixel physical resolution of the functional parameter plane. It is used to convert the pixel displacement dimension of the geometric plane to the pixel displacement dimension of the functional plane.

[0037] Based on the corrected center coordinates at each time step, a sampling window of 10 pixels by 10 pixels is defined within the original elastic modulus sequence and the original microblood flow sequence in the functional parameter plane. A bilinear interpolation algorithm is used to calculate the values ​​at all sub-pixel positions within this sampling window, and their arithmetic mean is calculated. These values ​​are then used as the tissue elastic modulus data and microblood flow perfusion data for that time step, respectively. The data extracted in this step fully preserves the original signal fluctuations, covering all time periods from the patient's resting baseline state to the stress load state.

[0038] S204, based on the corrected coordinate trajectory and the extracted data sequence, synchronously calculates dynamic image features. These dynamic image features are used to quantify the movement characteristics of anatomical structures, including cumulative displacement and displacement velocity variance.

[0039] The cumulative displacement represents the total distance the urethral structures move under pressure, and the calculation formula is as follows: ; In the formula, : Indicates the time up to the current moment. The cumulative displacement of the center of the region of interest, in pixels; : indicates the first The coordinate vector of the center of the region of interest at time t; : Represents the center coordinate vector of the region of interest at the initial moment; : Represents the Euclidean norm operation of a vector, that is, to find the magnitude of the vector.

[0040] The variance of displacement velocity characterizes the stability of the motion of anatomical structures, and the calculation formula is as follows: ; In the formula, : Indicates the current time The displacement velocity variance, in pixels squared per square second; : Indicates the length of the time sliding window used to calculate the variance, with a value ranging from 5 to 10 sampling points; : indicates the first The magnitude of the instantaneous moving velocity of the center of the region of interest at any given time; : Represents the arithmetic mean of the instantaneous movement speed moduli within the current sliding window.

[0041] Through the above steps, it is possible to accurately locate the target tissue and extract the corresponding functional parameters from the functional imaging plane even when the anatomical structure is moving, while generating auxiliary features that describe the tissue's motion characteristics.

[0042] See attached document Figure 4 , Figure 4 This is a schematic diagram of a motion-dependent confidence weight generation and learning window identification process according to an embodiment of the present invention. Step S3 calculates the physical motion rate and generates confidence weights, and then identifies the window used for physical model learning, specifically including the following steps: S301, convert pixel-level displacement into physical motion rate. Obtain the instantaneous motion vector of the anatomical structure calculated in step S2, which consists of horizontal and vertical components. Calculate the Euclidean norm of this vector, i.e., the pixel displacement distance of the anatomical structure in the image plane.

[0043] Based on the physical parameters of the imaging system, the physical motion rate of the anatomical structure at the current moment is calculated using the following formula: ; In the formula, : Represents the physical motion rate of the anatomical structure at the current moment, in millimeters per second; : Represents the horizontal pixel displacement component of the instantaneous motion vector at the current moment; the unit is pixels; : Represents the vertical pixel displacement component of the instantaneous motion vector at the current moment, in pixels; : Represents the average pixel physical resolution of the geometric reference plane. This value is equal to the physical width of the imaging area of ​​the geometric reference plane divided by the total number of pixels in the horizontal direction of the image. The physical width is typically 40 mm to 60 mm, and the total number of pixels is typically 256 pixels to 512 pixels, in millimeters per pixel. : This represents the time interval between the acquisition of two adjacent image frames. This value is the reciprocal of the geometric parameter surface imaging rate and ranges from 0.01 seconds to 0.025 seconds.

[0044] S302, Quantifying the Confidence Weights of Microblood Flow Data Based on Physical Motion Rate. According to the principles of Doppler ultrasound imaging, the extraction of microblood flow signals relies on wall filters to remove high-amplitude, low-frequency tissue motion signals. When the physical motion rate of the anatomical structure exceeds the cutoff velocity of the wall filter, the Doppler frequency shift generated by tissue motion cannot be filtered out, thus mixing into the blood flow signal and forming high-intensity scintillation artifacts, leading to distortion of microblood flow perfusion measurements. Therefore, a confidence scoring model based on a Gaussian decay function is constructed, mapping the physical motion rate to continuous weight values ​​between 0 and 1.

[0045] For each time step, the confidence weight is calculated using the following formula: ; In the formula, : Represents the confidence weight of the microblood flow data at the current moment, with a value ranging from 0 to 1. The closer the value is to 1, the more reliable the data is, and the closer the value is to 0, the more severely the data is affected by motion artifacts. : Represents the physical motion rate calculated by step S301; : This represents the motion sensitivity control parameter, which characterizes the rate inflection point at which confidence begins to decrease significantly. The value of this parameter is set equal to the physical velocity corresponding to the cutoff frequency of the ultrasound system's wall filter, typically ranging from 2 mm / s to 5 mm / s.

[0046] S303 identifies the learning window used for physical model learning. The learning window refers to the time period during which the patient is in a relatively static state, the stress-strain relationship and microcirculation perfusion status within the tissue are not disturbed by violent macroscopic motion, and the microblood flow data are not contaminated by motion artifacts.

[0047] Set a high confidence threshold, with a value ranging from 0.85 to 0.95. Iterate through the confidence weight sequence for the entire time period to generate a binary mask sequence: when the confidence weight at a certain moment is greater than the high confidence threshold, mark it as a candidate point; otherwise, mark it as a non-candidate point.

[0048] A continuity test is performed on the candidate point sequence, and a counter is set to record the number of consecutive candidate points. Only when the duration of consecutive points marked as candidate points exceeds the minimum window duration is the consecutive time period confirmed as a learning window. The minimum window duration is set to 0.5 to 1.0 seconds, corresponding to the continuous acquisition of 5 to 10 frames of valid low frame rate functional plane data. Finally, one or more discrete time intervals are output as learning windows for the identification of physical model parameters in subsequent steps. This step filters out high-quality data segments, eliminating low-quality data interference caused by respiratory movements or involuntary muscle contractions.

[0049] See attached document Figure 5 , Figure 5 This is a schematic diagram of the stiffness-blood flow coupling feature analysis and parameter identification process according to an embodiment of the present invention. Step S4 analyzes the correlation between tissue elastic modulus and microblood flow perfusion data within the learning window and identifies physical model parameters, specifically including the following steps: S401, Construct a training sample set and verify the effectiveness of physical coupling. From the learning window determined in step S3, extract the tissue elastic modulus data sequence and microblood flow perfusion data sequence within the corresponding time period. These two sequences represent the changes in stiffness and blood flow response of the pelvic floor tissues under the influence of basic physiological activities in a quasi-resting state.

[0050] Calculate the Pearson correlation coefficient between tissue elastic modulus and microvascular perfusion data within this sample set. This coefficient measures the strength and direction of the linear correlation between the two sets of data. A threshold for negative correlation validity is set, ranging from -0.4 to -0.6. If the calculated Pearson correlation coefficient is less than this threshold, indicating a stronger negative correlation, it suggests a significant mechanovascular coupling effect within the tissue. Specifically, as the tissue elastic modulus increases, internal tissue pressure increases, leading to compression and narrowing of the microvascular bed diameter, increased blood flow resistance, and consequently, a decrease in perfusion volume. This aligns with the biomechanical characteristics of soft tissue, thus determining that the physical coupling is effective. If the Pearson correlation coefficient is greater than or equal to this threshold, it indicates a lack of clear physical causal relationship between the data or severe noise interference, thus determining that the physical coupling is ineffective.

[0051] S402, when physical coupling is deemed effective, a stiffness-blood flow physical model is constructed based on a power-law function, and the model parameters are identified. Based on Poiseuille's law in vascular fluid dynamics and the nonlinear elastic characteristics of the vessel wall, the flow resistance of the microcirculation system exhibits a nonlinear decay trend in response to external pressure. Therefore, the following nonlinear physical model is established to describe the response relationship of microvascular perfusion with changes in tissue elastic modulus: ; In the formula, : Represents the microvascular perfusion density value estimated based on the model, a dimensionless normalized value; : Represents the measured elastic modulus of the tissue, in kilopascals; : Represents the perfusion volume coefficient, which characterizes the theoretical maximum perfusion level when the tissue elastic modulus is assumed to be a unit value of 1 kPa, reflecting the baseline vascular density within the region of interest. : Represents the vascular compliance index. This parameter is a dimensionless positive real number that reflects the sensitive decay rate of microvascular perfusion to changes in tissue stiffness. This value depends on the stiffness of the vessel wall and the topology of the vascular bed.

[0052] To solve for the unknown parameters in the above nonlinear model, we take the natural logarithm of both sides of the equation to transform it into a linear regression form: ; In the formula, : Represents the actual microvascular perfusion data measured within the learning window; : Represents the fitting residual, which is the deviation between the measured value and the theoretical value of the model.

[0053] The transformed linear equations were solved using the least squares method to obtain the intercept and slope terms, which were then used to inversely solve for the patient-specific perfusion volume coefficient and vascular compliance index. This set of parameters was then designated as the execution parameters for the physical deduction model.

[0054] S403, based on the determination result of S401, execute the branch logic: if the physical coupling is determined to be effective, then determine the signal reconstruction strategy as physical deduction, and record the perfusion volume coefficient and vascular compliance index identified in step S402 as input parameters for subsequent calculations.

[0055] If physical coupling is deemed ineffective, the signal reconstruction strategy is set to numerical preservation. Under this strategy, the arithmetic mean of all microvascular perfusion data within the learning window is calculated and defined as the baseline perfusion constant. The baseline perfusion constant, as a statistical estimate, is used to replace missing motion-state data with the patient's average resting level in the absence of clear physical laws, thereby avoiding the introduction of larger artificial artifacts due to model bias.

[0056] Through the above steps, the method can adaptively select the optimal signal correlation model based on the patient's real-time physiological data characteristics, providing personalized parameter basis for the subsequent recovery of lost blood flow signals during motion artifacts.

[0057] See attached document Figure 6 , Figure 6 This is a schematic diagram of a confidence-weighted adaptive reconstruction process for micro-blood flow signals according to an embodiment of the present invention. Step S5 performs signal repair and reconstruction on the full-time micro-blood flow perfusion data, including both static and dynamic states, extracted in step S2, specifically including the following steps: S501, Generate a substitute reference signal for the entire time period. Based on the signal reconstruction strategy determined in step S4, calculate the substitute reference microflow value at each moment in the discrete time series.

[0058] When the determined signal reconstruction strategy is physical deduction, the original tissue elastic modulus data extracted in step S2 at the same time is read and substituted into the stiffness-blood flow coupling model identified in step S4 for reverse calculation. The physical basis for using tissue elastic modulus as the repair benchmark here is that shear wave elastography is calculated based on the time-of-flight shift of ultrasound waves, and its sensitivity to low-speed axial jitter of the probe is significantly lower than that of micro-blood flow imaging based on the Doppler frequency shift principle. When the patient experiences slight body movement, Doppler signals are prone to spectral aliasing, while the elastic modulus data can still maintain relatively stable numerical accuracy. The calculation formula for the alternative reference signal is as follows: ; In the formula, : Indicates the current time Alternative reference microblood flow values; : Indicates the current time The original tissue elastic modulus data, in kilopascals; : Represents the perfusion capacity coefficient, which is a fixed constant taken from the parameter identification result in step S4. : Represents the vascular compliance index. This parameter is a fixed constant and is taken from the parameter identification result of step S4.

[0059] When the determined signal reconstruction strategy is value preservation, the baseline perfusion constant calculated in step S4 is directly used as the alternative reference microflow value at each time point throughout the entire time period, that is: ; In the formula, : Represents the baseline perfusion constant, which is the arithmetic mean of the effective data within the learning window.

[0060] S502, perform linear weighted fusion based on confidence weights. In order to achieve a smooth transition between the original measurement data and the alternative reference signal and avoid signal jumps caused by binarization switching, the confidence weight sequence calculated in step S3 is used to perform weighted fusion on the two.

[0061] For each time step, the reconstructed micro-blood flow signal is calculated: ; In the formula, : Indicates the current time Reconstructed microblood flow signal after motion artifact repair; : Indicates the current time The confidence weight ranges from 0 to 1. A higher weight value indicates that the original measurement data is more reliable. : Indicates the current time Raw microvascular perfusion data; : Indicates the current time Alternative reference microblood flow values.

[0062] Using this weighted fusion formula, in the static or low-speed motion range where the confidence weight is close to 1, the reconstructed signal is mainly dominated by the original measurement data, preserving the true physiological fluctuation details; in the high-speed motion range where the confidence weight is close to 0, the reconstructed signal is mainly dominated by the substitute reference signal, and the signal distortion caused by motion artifacts is corrected by using the stiffness-blood flow physical model or statistical baseline value.

[0063] S503 performs time-series smoothing on the reconstructed signal. To eliminate high-frequency noise that may be generated during weighted fusion, Savitzky-Golay filtering is applied to the generated reconstructed microblood flow signal sequence.

[0064] The filter window length is set to an odd number between 5 and 9 data points, and the polynomial fitting order is between 2nd and 3rd. In practice, this step is implemented through convolution operations, specifically by performing a sliding convolution with the reconstructed microblood flow signal sequence using a pre-calculated convolution coefficient vector. This convolution coefficient vector is derived in advance using the least squares method for the selected window length and polynomial order. This processing effectively removes random noise while preserving the peak and trough characteristics of the signal, avoiding excessive smoothing of the crucial transient response features in stress urinary incontinence assessment. The final output is a continuous, complete, motion artifact-free reconstructed microblood flow signal sequence for subsequent multidimensional feature analysis.

[0065] See attached document Figure 7 , Figure 7 This is a schematic diagram of a multidimensional quantitative feature extraction and stress urinary incontinence risk assessment process according to an embodiment of the present invention. Step S6 calculates multidimensional assessment indicators and generates assessment results based on time-aligned anatomical displacement data, original tissue elastic modulus data, and reconstructed microvascular perfusion data, specifically including the following steps: S601, read the cumulative displacement sequence calculated in step S2 and calculate its first derivative with respect to time, i.e., the instantaneous displacement velocity. Set a displacement velocity threshold, ranging from 2 mm / s to 5 mm / s. When the instantaneous displacement velocity is less than this threshold and the duration exceeds 2 seconds, it is determined to be the resting phase; when the instantaneous displacement velocity is greater than this threshold and the cumulative displacement shows a monotonically increasing trend, it is determined to be the pressure loading phase; the interval from 0.5 seconds to 1.0 seconds after the cumulative displacement reaches its maximum value is defined as the maximum load moment. Through this temporal segmentation, the entire time period data is divided into the resting baseline interval and the pressure peak interval.

[0066] S602, calculate the functional ischemia index of microcirculation and the urethral mobility-stiffness ratio. The functional ischemia index of microcirculation is used to quantify the degree of occlusion of the microvascular bed in the pelvic floor tissues under pressure. Using the reconstructed microblood flow signal sequence output in step S5, extract the average value within the resting baseline interval and the instantaneous value at the maximum load time, and calculate using the following formula: ; In the formula, : Indicates the functional ischemia index of microcirculation, expressed as a percentage; : Represents the arithmetic mean of the reconstructed microflow signal within the resting baseline interval, which reflects the baseline perfusion level when the tissue is not compressed; : This represents the instantaneous value of the reconstructed microflow signal at the moment of maximum load. This value reflects the actual residual blood flow perfusion after motion artifact repair under maximum abdominal pressure.

[0067] The urethral mobility-stiffness ratio is used to assess the relaxation and compensatory capacity of the urethral support structure. It is calculated using the maximum cumulative displacement obtained in step S2 and the original tissue elastic modulus obtained in step S1 at the maximum load, using the following formula: ; In the formula, : Indicates the ratio of urethral mobility to stiffness, with units of millimeters per kilopascal. : Represents the maximum cumulative displacement recorded at the end of the pressure load phase, characterizing the maximum downward displacement distance of the urethra; : Represents the tissue elastic modulus value recorded at the moment of maximum load, characterizing the antagonistic stiffness between the pelvic floor muscles and connective tissue at this time.

[0068] S603, construct a multi-parameter linear weighted assessment model and output graded assessment results. To comprehensively evaluate the potential risk of stress urinary incontinence, the following weighted scoring formula is established: ; In the formula, : Represents the comprehensive risk score for stress urinary incontinence, a dimensionless numerical value; : Represents the weighting coefficient of the microcirculation dimension, with a value ranging from 0.4 to 0.6, set according to the contribution of ischemia to tissue function degeneration in clinical pathophysiology; : Represents the weighting coefficient of the structural mechanics dimension, with a value ranging from 0.4 to 0.6, and the sum of the two weighting coefficients equals 1; : Indicates the preset physiological reference ischemia index, which is the average ischemia index of healthy people under the maximum Valsalva maneuver, usually set to 30% to 40%; : This represents the preset physiological reference mobility-hardness ratio, which is the statistical average value of healthy people, usually set to 0.5 mm / kPa to 1.0 mm / kPa.

[0069] Based on the calculated comprehensive risk score, a grading judgment logic is applied: if the comprehensive risk score is less than 1.0, it is judged as negative or low risk, indicating that the function of the pelvic floor support structure is within the normal physiological range; if the comprehensive risk score is greater than or equal to 1.0 and less than 2.0, it is judged as medium risk, indicating the presence of compensatory functional decline; if the comprehensive risk score is greater than or equal to 2.0, it is judged as high risk, indicating the presence of overt stress urinary incontinence pathological changes.

[0070] Through the above steps, the method transforms complex orthogonal biplane ultrasound data into intuitive quantitative clinical indicators, and in particular, provides electronic assessment data that combines structure and function by utilizing the restored blood flow data.

[0071] Specific application examples: In this embodiment, a 45-year-old female subject was selected as the evaluation subject. The subject's clinical complaint was occasional urinary leakage when coughing or laughing, and her body mass index was 26.5 kg / m². 2 .

[0072] In step S1, an orthogonal dual-plane imaging environment is constructed. A custom dual-array probe with a center frequency of 5.0 MHz is used. The geometric reference plane is set to B-mode imaging, the depth is set to 70 mm, and the frame rate is set to 60 frames / second; the functional parameter plane is configured as shear wave elastography and energy Doppler microflow imaging, with a combined frame rate set to 4 frames / second.

[0073] The system reads the patient's electronic medical record via a DICOM interface and establishes a time zero point on both planes. Subjects perform a standard Valsalva maneuver under system instructions, lasting 5 seconds, followed by a 5-second relaxation period, repeated 3 times.

[0074] In steps S2 to S3, the system captures a significant posterior-inferior displacement of the bladder neck region during the subject's peak movement.

[0075] The instantaneous maximum pixel displacement velocity detected by the geometric reference plane is converted to physical velocity as follows: =8.2mm / s. At this time, It exceeded the preset wall filter cutoff speed (set to 3.0 mm / s).

[0076] The confidence weight at that moment is calculated according to the formula in step S302. The value is approximately 0.15, indicating that the original microblood flow signal at this time is mainly composed of motion artifacts, with extremely low confidence.

[0077] In step S4, the system automatically identifies the resting phase before the action, with a time interval of 0.5s to 1.5s, as a learning window. Within this window, the tissue elastic modulus is extracted. With microblood perfusion density The sequences. The Pearson correlation coefficient between the two was calculated. The negative correlation validity condition is met, and the threshold is set to -0.5. The system enters physical deduction mode, performs log-linear regression on the interval data, and identifies patient-specific parameters: Infusion capacity coefficient Vascular compliance index In step S5, for the peak moment of motion ( ≈0.15), using the above parameters and the measured value of elastic modulus during the same period ( Alternative reference values ​​were calculated and weighted and fused with the original signal to output the reconstructed microblood flow signal after artifact removal.

[0078] In step S6, features are extracted and a risk score is calculated: the functional ischemia index of microcirculation (... The calculated value is 38.5%. (Urethra mobility-rigidity ratio) Maximum cumulative displacement mm, peak elastic modulus kPa, calculated mm / kPa. Comprehensive risk score. Set weights Reference value mm / kPa.

[0079] ; According to the grading logic, since the score is less than 1.0, the system automatically outputs an assessment conclusion of low risk / negative, indicating that although urethral migration exists, the pelvic floor muscles are well compensated for and the degree of ischemia has not reached the pathological threshold. This conclusion is consistent with the results of subsequent urodynamic examinations.

[0080] Experimental verification and effect comparison To verify the effectiveness of this method, a comparative experiment was conducted with 30 clinical subjects, including a healthy control group of 10 and a group of 20 patients diagnosed with stress urinary incontinence. The comparison protocol was set as follows: Method A (Prior Art): Only raw microflow data without motion artifact correction are used, and no elastic modulus is introduced for weighted evaluation.

[0081] Method B (the method of this invention): Microblood flow data based on dual-plane motion correction and physical model reconstruction are used, combined with a multidimensional risk scoring model.

[0082] The experimental data are recorded in Table 1. The table shows the measurement results of key parameters and the consistency of the final diagnosis for some typical samples at the moment of maximum Valsalva action.

[0083] Table 1: Comparison of accuracy and diagnostic consistency of pelvic floor hemodynamic parameter reconstruction under different treatment methods Conclusion Analysis Based on the data in Table 1 and the operating results of the above embodiments, the following analysis is performed on this technical solution: Suppression of motion artifacts and restoration of signal authenticity: Observation Table 1 shows that for samples with peak motion velocities exceeding 5.0 mm / s (such as P-007, P-015, P-30), the raw ischemia index obtained by Method A exhibits significant randomness and physical paradoxes. For example, in sample P-015, the raw data shows an ischemia index of -42.8%, meaning a 42.8% increase in blood flow signal. This contradicts the physiological mechanism of ischemia due to pelvic floor tissue compression, and is clearly caused by scintillation artifacts resulting from Doppler signal aliasing caused by vigorous movement. In contrast, the reconstructed ischemia index calculated by Method B is 55.4%, returning to a reasonable physiological range. This indicates that the confidence-weighted fusion strategy in step S5 effectively eliminates motion artifacts and fills the signal gap in the high dynamic range using the stiffness-blood flow coupling model, making microcirculation assessment under high-velocity motion possible.

[0084] The analysis of samples P-012 and P-22, both patients with mild to moderate SUI, aimed to reduce false negatives and misdiagnosis rates. In Method A, due to the failure to remove motion noise, the superimposed artifact signal masked the true ischemia condition (the original indices were only 12.3% and 8.5%), leading to negative results, i.e., false negatives. Method B reconstructed the true ischemia indices (36.5% and 41.2%) and combined them with indicators reflecting structural relaxation. Parameters, calculated The values ​​were 1.15 and 1.38 respectively, accurately classifying them into the medium-risk group, consistent with the clinical gold standard diagnosis. This demonstrates that introducing multidimensional parameters, combining microcirculatory function and structural mechanics, can significantly improve diagnostic sensitivity.

[0085] The P-30 sample, used to circumvent specificity and overdiagnosis, illustrates another scenario: the patient moved rapidly (8.9 mm / s), exhibiting motion artifacts, but... A lower score (0.75) indicates relatively good connective tissue rigidity. Method A was unable to make a judgment (or resulted in a misjudgment) due to artifact interference and data confusion. Method B, by correcting the blood flow data (ischemia index 39.8%) and combining it with a lower structural risk weight, achieved a final score of 0.98, which was determined as low risk / negative, consistent with the clinical diagnosis. This demonstrates that step S603 of the weighted scoring model in this protocol can balance functional and structural indicators, avoiding overdiagnosis caused by abnormalities in a single indicator.

[0086] In summary, this invention solves the spatial alignment problem between geometric tracking and functional measurement through dual-plane collaborative imaging, and addresses signal distortion under high dynamic conditions by utilizing physical model constraints. Experimental data show that this method can still provide physiologically accurate quantitative indicators even when subjects are performing vigorous valsalva maneuvers, significantly improving the accuracy and robustness of early risk assessment for stress urinary incontinence.

[0087] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for diagnosing stress urinary incontinence based on dual plane ultrasound multimodal imaging, characterized in that, Includes the following steps: S1. Establish a dual-plane imaging environment, define the long axis section of the probe as the geometric reference plane and the short axis section of the probe as the functional parameter plane, and synchronously trigger the acquisition of grayscale image sequences in the geometric reference plane and the original elastic modulus sequence and original microblood flow sequence in the functional parameter plane in the two planes to establish a unified time reference. S2, using grayscale image sequences to calculate the instantaneous motion vector of anatomical structures, mapping the instantaneous motion vector to the functional parameter plane to correct the coordinate position of the region of interest in real time, and extracting one-to-one corresponding full-time tissue elastic modulus data and microblood perfusion data based on the corrected coordinate position; S3, calculate the magnitude of the instantaneous motion vector and convert it into physical motion rate, generate dynamically changing confidence weights based on the physical motion rate, and identify continuous time periods with confidence weights higher than a preset high confidence threshold as learning windows; S4, within the learning window, analyze the correlation between tissue elastic modulus data and microvascular perfusion data, and determine the signal reconstruction strategy as physical deduction or numerical preservation based on whether the correlation meets the negative correlation condition. S5, for full-time microblood flow perfusion data, uses confidence weights to perform weighted fusion of the original measurement data and the alternative reference signal generated according to the signal reconstruction strategy, and outputs the reconstructed microblood flow signal after motion artifact repair. S6. The reconstructed microblood flow signal and tissue elastic modulus data of the urethral orifice region are mapped to a two-dimensional coordinate system to construct a hysteresis loop. The area of ​​the hysteresis loop is combined with the SUI risk index calculated based on the patient's physiological characteristics and dynamic imaging characteristics to generate an auxiliary assessment conclusion.

2. The method for diagnosis of stress urinary incontinence based on dual plane ultrasound multimodality imaging as claimed in claim 1, wherein, In step S1, establishing the dual-plane imaging environment specifically includes: A dual-plane ultrasound probe integrating two independent piezoelectric transducer arrays is used, covering a frequency range of 3 MHz to 13 MHz. The first piezoelectric transducer array is linearly arranged along the long axis of the probe, corresponding to the geometric reference plane, while the second piezoelectric transducer array is linearly arranged along the short axis of the probe, corresponding to the functional parameter plane. The two arrays are fixed at a 90-degree orthogonal angle in physical space. A pre-calibrated spatial transformation matrix is ​​used to store the spatial mapping relationship between the geometric reference plane and the functional parameter plane. The spatial transformation matrix consists of a rotation matrix component describing angular deviation and a translation vector component describing the distance deviation of the physical center point. The imaging depth is set to 2 cm to 4 cm, and the focusing area is set to 1 cm to 2 cm from the probe surface. Patient physiological characteristic data, including age, height, and weight, are recorded before data acquisition.

3. The method for diagnosis of stress urinary incontinence based on dual plane ultrasound multimodality imaging as claimed in claim 1, wherein, In step S1, the specific logic for synchronously triggering data acquisition is as follows: A B-mode ultrasound pulse sequence with a transmission center frequency of 3 MHz to 12 MHz and a frame rate of 40 to 100 frames per second is configured in the geometric reference plane; Shear wave elastography pulse sequence and microblood flow imaging pulse sequence are configured simultaneously in the functional parameter plane. The ensemble length of the microblood flow imaging pulse sequence is set to 10 to 20 pulses, and the combined frame rate of the two is set to 2 to 10 frames per second. Execute asymmetric time-triggered logic to continuously trigger 5 to 20 acquisitions of the geometric reference plane within one acquisition cycle, followed by one acquisition of the functional parameter plane; establish a unified discrete time series, and record the data acquisition timestamp of the functional parameter plane at each moment.

4. The method of diagnosing stress urinary incontinence based on dual plane ultrasound multimodality imaging as claimed in claim 1, wherein, Step S2 specifically includes: The pixel displacement of feature tracking points in the geometric reference plane is tracked using a multi-scale pyramid optical flow algorithm to obtain the original two-dimensional instantaneous motion vector. A 2x2 cross-plane projection matrix is ​​constructed, and the original two-dimensional instantaneous motion vector is multiplied by the cross-plane projection matrix to obtain the corrected displacement vector in the functional parameter plane. The center coordinates of the region of interest in the functional parameter plane are iteratively updated using the corrected displacement vector and the anisotropic scaling factor, where the anisotropic scaling factor is the ratio of the pixel physical resolution of the geometric reference plane to that of the functional parameter plane. In the original elastic modulus sequence and the original microblood flow sequence in the functional parameter plane, a sampling window of size 10 pixels by 10 pixels is defined, and the arithmetic mean of the data in the sampling window is calculated using a bilinear interpolation algorithm as the extracted data.

5. The method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging according to claim 1, characterized in that, Step S3 specifically includes: By combining the average pixel physical resolution of the geometric reference plane with the time interval between the acquisition of two adjacent frames, the pixel displacement distance of the instantaneous motion vector is converted into a physical motion rate in millimeters per second. A confidence scoring model based on the Gaussian decay function is constructed, and the confidence weight is calculated. The weight is equal to the exponent of the natural constant, and the base of the exponent is the square of the physical motion rate divided by the negative value of twice the square of the motion sensitivity control parameter. The motion sensitivity control parameter is set to 2 mm / s to 5 mm / s. The high confidence threshold is set to 0.85 to 0.

95. Only when the duration of time when the confidence weight is continuously marked as higher than the threshold exceeds 0.5 seconds to 1.0 seconds is the continuous time period confirmed as the learning window.

6. The method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging according to claim 5, characterized in that, Step S4 specifically includes: Calculate the Pearson correlation coefficient between tissue elastic modulus data and microvascular perfusion data within the learning window; If the Pearson correlation coefficient is less than the negative correlation validity threshold set between -0.4 and -0.6, the physical coupling is deemed effective, and the signal reconstruction strategy is determined to be physical deduction. If the physical coupling is deemed effective, a nonlinear physical model is established, in which the microvascular perfusion density value is equal to the perfusion volume coefficient multiplied by the tissue elastic modulus value raised to the power of the negative vascular compliance exponent. The natural logarithm of the model is taken to transform it into a linear regression form, and the perfusion volume coefficient and vascular compliance exponent are identified using the least squares method. If the Pearson correlation coefficient is greater than or equal to the negative correlation validity threshold, the physical coupling is deemed ineffective, the signal reconstruction strategy is determined to be numerical preservation, and the arithmetic mean of the microvascular perfusion data within the learning window is calculated as the baseline perfusion constant.

7. The method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging according to claim 1, characterized in that, Step S5 specifically includes: Generate alternative reference signals for the entire time period: When the strategy is physical deduction, the original tissue elastic modulus data at each time step are substituted into the identified nonlinear physical model to calculate the alternative reference microblood flow value; when the strategy is value preservation, the baseline perfusion constant is directly taken as the alternative reference microblood flow value; perform linear weighted fusion: for each time step, the reconstructed microblood flow signal is equal to the confidence weight at that time step multiplied by the original microblood flow perfusion data, plus 1 minus the difference of the confidence weight at that time step multiplied by the alternative reference microblood flow value; perform Savitzky-Golay filtering on the generated reconstructed microblood flow signal sequence, setting the filtering window length to an odd number between 5 and 9, and the polynomial fitting order to be between 2nd and 3rd order.

8. The method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging according to claim 1, characterized in that, When calculating the SUI risk index in step S6, the extracted dynamic image features include cumulative displacement and displacement velocity variance. The cumulative displacement is the sum of the Euclidean norms of the coordinate vector of the center of the region of interest up to the current time relative to the coordinate vector at the initial time; the displacement velocity variance is the root mean square error of the instantaneous velocity magnitude relative to the average velocity magnitude within a sliding window of 5 to 10 sampling points.

9. The method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging according to claim 1, characterized in that, In step S6, generating auxiliary assessment conclusions specifically includes calculating the microcirculation functional ischemia index and the urethral mobility-stiffness ratio: The full-time data was divided into a resting baseline interval and a pressure peak interval. The microcirculation functional ischemia index was calculated as follows: the arithmetic mean of the reconstructed microblood flow signals within the resting baseline interval minus the instantaneous value of the reconstructed microblood flow signals at the maximum load time, and then divided by the arithmetic mean of the reconstructed microblood flow signals within the resting baseline interval. The result was expressed as a percentage. The urethral mobility-rigidity ratio was calculated as follows: the maximum cumulative displacement at the end of the pressure load phase divided by the tissue elastic modulus value at the maximum load time.

10. The method for diagnosing stress urinary incontinence based on dual-plane ultrasound multimodal imaging according to claim 1, characterized in that, In step S6, the SUI risk index is calculated using a multi-parameter linear weighted evaluation model: The model multiplies the ratio of the functional ischemia index of microcirculation to the preset physiological reference ischemia index by a microcirculation dimension weighting coefficient, and adds the ratio of the urethral mobility-rigidity ratio to the preset physiological reference mobility-rigidity ratio by a structural mechanics dimension weighting coefficient. The microcirculation dimension weighting coefficient and the structural mechanics dimension weighting coefficient are set based on the contribution of clinical pathophysiology, with values ​​ranging from 0.4 to 0.6 and a sum of 1. The preset physiological reference ischemia index is set based on statistical data from healthy individuals, with values ​​ranging from 30% to 40%. The preset physiological reference mobility-rigidity ratio is set based on statistical data from healthy individuals, with values ​​ranging from 0.5 mm / kPa to 1.0 mm / kPa. Finally, the calculated SUI risk index is compared with a preset risk grading threshold to generate a corresponding low-risk, medium-risk, or high-risk classification result.

Images