Traditional Chinese medicine prescription efficacy evaluation system for treating intrauterine adhesion

Abstract

The invention relates to the technical field of medical care information processing, and discloses a traditional Chinese medicine prescription efficacy evaluation system for treating intrauterine adhesion, comprising: a data preprocessing module for acquiring endometrial dynamic response gray time sequence characteristic data before and after a subject takes a prescription; the phase analysis module is used for determining a phase deviation distribution coefficient according to the phase angle distribution of the gray scale data of the reference pixel and the response pixel; the frequency domain inversion module is used for constructing an energy transfer gain spectrum by calculating a power ratio and identifying a characteristic resonant frequency; the evaluation index output module is used for generating an evaluation index for evaluating the viscoelasticity improvement degree of the endometrial tissue according to the deviation value of the characteristic resonant frequency in combination with the variation of the phase deviation distribution coefficient. And the quantitative accuracy of evaluation is improved.

Description

Technical Field

[0001] This invention relates to a system for evaluating the efficacy of traditional Chinese medicine formulas used to treat intrauterine adhesions, belonging to the field of medical and health care information processing technology. Background Technology

[0002] Current assessments of the treatment efficacy for intrauterine adhesions rely on monitoring the endometrial repair status. The mainstream assessment method currently utilizes dynamic multidimensional grayscale mapping feature matrix sequences to measure endometrial thickness, echo intensity, and blood perfusion indicators to construct an evaluation system. Traditional Chinese medicine formulas have multi-target, slow-variable intervention effects on endometrial repair, and the physiological response of the endometrial microenvironment usually precedes morphological changes. In actual clinical data collection, the pharmacodynamic signals in the multidimensional grayscale mapping feature matrix sequences exhibit non-stationary characteristics and are deeply coupled with displacement noise generated by respiratory movements and intestinal peristalsis. Existing information processing methods focus on extracting static geometric features, making it difficult to extract the dynamic features characterizing orderly tissue repair in environments with strong background noise.

[0003] Previous studies have shown that the diagnosis of intrauterine adhesions has been clearly established in AI-assisted hysteroscopic surgery, and corresponding surgical procedures have been completed. The AI ​​system has demonstrated its value in lesion identification, risk prediction, and personalized treatment, covering the entire process of preoperative assessment, intraoperative assistance, and postoperative follow-up, significantly improving the accuracy and personalization of intrauterine adhesion diagnosis and treatment. Based on this technology, how to further quantify and evaluate the effect of traditional Chinese medicine formulas on the dynamic improvement of postoperative endometrial repair has become the focus of this invention.

[0004] The industry is seeking breakthroughs in hardware aspects such as image acquisition, or attempting to assist in judgment through software aspects such as constructing biological models. However, there are limitations in real-time clinical feedback and adaptability to the in vivo environment. For example, Chinese invention patent CN113430162B discloses a model of intrauterine adhesions, which uses TGF-β to induce endometrial stromal cells to construct an in vitro fibrosis model, providing an experimental basis for drug screening. In clinical evaluation scenarios, the ex vivo cell model cannot reproduce the complex dynamic transmission characteristics of living tissues, and the evaluation logic relies on destructive molecular biological detection, resulting in long feedback cycles. It cannot provide real-time quantitative discrimination criteria for individual differences such as the subject's baseline heart rate and respiratory perturbations, making it difficult to decouple physiological pulsation signals from physical displacement noise. The evaluation benchmark is biased due to individual differences. Conventionally improving image resolution or applying motion compensation algorithms can easily lead to over-smoothing of effective perfusion information, making it impossible to quantify the mechanical properties during tissue repair. Although the industry has tried to apply deep learning models, due to the lack of modeling of physiological rhythm coherence, it is difficult to provide an objective efficacy evaluation index in the early stage of prescription intervention.

[0005] Therefore, the technical problem to be solved by this invention is how to extract coherent features that characterize the orderly repair of tissues from non-stationary dynamic ultrasound image streams and construct a dynamic evaluation system with self-calibration properties. Summary of the Invention

[0006] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: A system for evaluating the efficacy of traditional Chinese medicine formulas for treating intrauterine adhesions, comprising a data preprocessing module, a phase analysis module, a frequency domain inversion module, and an evaluation index output module: The data preprocessing module is used to acquire gray-scale temporal feature data of endometrial dynamic response of subjects before and after taking traditional Chinese medicine prescriptions. The gray-scale temporal feature data of dynamic response includes gray-scale data of reference pixels corresponding to the feature vector of the reference anatomical domain and gray-scale data of response pixels corresponding to the gray-scale flow of the response analysis domain. The phase analysis module is used to perform cross-correlation operations on the grayscale data of the reference pixel and the grayscale data of the response pixel to determine the time delay, convert the time delay into phase angle data, and determine the phase offset distribution coefficient based on the variance of the phase angle data in the spatial domain. The frequency domain inversion module is used to perform power spectral density calculations on the reference pixel grayscale data and the response pixel grayscale data respectively. It constructs the energy transfer gain spectrum by calculating the power ratio of the response pixel power to the reference pixel power, and identifies the characteristic resonant frequency corresponding to the peak power in the energy transfer gain spectrum. The evaluation index output module is used to calculate the offset of the characteristic resonant frequency after drug administration relative to the characteristic resonant frequency before drug administration, and generate an evaluation index for evaluating the degree of improvement in the viscoelasticity of endometrial tissue based on the offset value combined with the change in the phase offset distribution coefficient.

[0007] Preferably, the phase analysis module is also used to construct a local motion synchronization matrix; the phase analysis module identifies abnormal motion stagnation regions in the dynamic response gray-scale temporal characteristic data by calculating the phase response consistency index between adjacent pixel blocks, and performs coordinate projection transformation using the motion vector of the non-perfusion area to filter out the displacement component caused by respiration in the frequency domain and correct the phase offset distribution coefficient.

[0008] Preferably, the frequency domain inversion module is also used to identify high-frequency cutoff points in the energy transfer gain spectrum; the evaluation index output module maps the incremental value of the high-frequency cutoff point to a tissue compliance improvement parameter, which is used to eliminate local congestion interference signals caused by acute inflammation during the evaluation process.

[0009] Preferably, the phase analysis module is also used to obtain the baseline grayscale features of the subject at preset physiological cycle nodes, and to subtract the periodic background fluctuations affected by hormone levels from the grayscale temporal feature data of the dynamic response, so as to extract the efficacy component data belonging to the intervention of traditional Chinese medicine prescription.

[0010] Preferably, when constructing the energy transfer gain spectrum, the frequency domain inversion module treats the endometrial tissue as a viscoelastic damped load driven by the uterine artery pulse pressure, and uses a dynamic impedance analysis algorithm to calculate the energy transfer efficiency in order to determine the characteristic resonant frequency.

[0011] Preferably, the data preprocessing module is also used to perform pixel tracking on the gray-scale temporal feature data of the dynamic response; the data preprocessing module generates motion vector fields of different sub-regions of the endometrium by calculating the gray-scale cross-correlation coefficient between two adjacent frames of images, and uses the motion vector fields as the input basis for the frequency domain inversion module.

[0012] Preferably, the evaluation index output module is also used to generate an internal environment repair trend view; the internal environment repair trend view includes a microcirculation remodeling curve composed of phase shift distribution coefficients and a tissue stiffness evolution line composed of characteristic resonant frequencies.

[0013] Preferably, the system also includes a storage module; the storage module is used to store the original data of the gray-scale time-series characteristics of the kinetic response of different subjects, as well as the corresponding historical efficacy evaluation indicators, so as to construct a efficacy comparison database for different Chinese medicine prescriptions.

[0014] Preferably, the evaluation index output module is also used to calculate the repair stability parameter; the repair stability parameter is determined by calculating the numerical variance of the characteristic resonant frequency within three consecutive physiological cycles, and is used to assess the long-term recurrence probability of the endometrial tissue status.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In the efficacy evaluation of traditional Chinese medicine formulas, the uterine artery pulsation frequency is used as an endogenous reference clock to establish a spatiotemporal coherence mapping with the gray-scale fluctuation sequence of the endometrial area to be evaluated, and the phase difference distribution entropy is extracted. This mechanism logically constitutes a closed-loop self-calibrating evaluation path, eliminating the interference of individual heart rate benchmarks of subjects and differences in ultrasound equipment gain on efficacy determination. This allows the output evaluation parameters to directly point to the dynamic repair order of the endometrial microcirculation. Since this evaluation logic is based on relative phase shift analysis and does not depend on absolute blood flow velocity or static imaging parameters, it ensures the physical consistency of the evaluation process under different clinical environments, realizing early capture of the endometrial microenvironment repair quality. Thus, it provides a traditional Chinese medicine formula efficacy evaluation system for the treatment of intrauterine adhesions.

[0016] 2. Combining the endometrial dynamic transfer function and resonance feature extraction logic, by identifying the increasing trend of cutoff frequency in the amplitude-frequency characteristic curve, the evolution process of tissue viscoelastic damping is quantified. A linear mapping relationship between the rightward shift of the cutoff frequency and the decrease in the degree of tissue fibrosis is established. This can effectively distinguish between pathological congestion caused by inflammation and functional reconstruction guided by traditional Chinese medicine. Since the shift of the cutoff frequency directly reflects the improvement of tissue physical properties, the system realizes the inversion of tissue mechanical properties using digital image sequences without increasing the elasticity imaging hardware, providing physical evidence for the evaluation of the efficacy of traditional Chinese medicine formulas in softening and dispersing nodules.

[0017] 3. The system utilizes motion displacement residual data generated during image processing, combined with the response delay time under respiratory micro-perturbations, to simultaneously calibrate the myometrial irritation index and microvascular compliance. The pelvic pressure perturbation generated by respiratory commands serves as an active detection window, and its response delay time... By providing nonlinear correction weights to the phase difference distribution entropy and eliminating interference from pseudo-perfusion during the inflammatory period, this method of transforming motion residuals, which originally belong to redundant information, into environmental constraint factors improves the systematic closed loop of drug efficacy evaluation, enabling the system to simultaneously provide feedback on endometrial repair quality and the stability of the physical growth environment. Attached Figure Description

[0018] Figure 1 This is a logic diagram of the system data acquisition and multidimensional pharmacodynamic feature inversion of the present invention; Figure 2 This is a schematic diagram of the clinical physician interaction use cases and functions that integrate the core computing logic of this invention. Detailed Implementation

[0019] The following detailed embodiments are used to explain and illustrate the present invention, and are not intended to limit the scope of protection of the present invention. The technical features disclosed in the following embodiments can be combined with each other without conflict.

[0020] A system for evaluating the efficacy of traditional Chinese medicine formulas for treating intrauterine adhesions includes a data preprocessing module, a phase analysis module, a frequency domain inversion module, and an evaluation index output module. This system processes dynamic ultrasound images of the endometrium before and after administration of the traditional Chinese medicine formula, outputting quantified endometrial microenvironment repair indicators, thus achieving efficient processing of healthcare information. The data preprocessing module acquires a sequence of dynamic multidimensional grayscale mapping feature matrices of the endometrium during a preset time period within the treatment cycle. Due to individual differences in baseline heart rate, blood pressure, and ultrasound equipment gain settings among different subjects, directly extracted physical quantities such as absolute blood flow velocity are difficult to use as a unified standard for cross-individual evaluation. The phase analysis module identifies the uterine artery reference area and the endometrial area to be evaluated in the image sequence. Based on the autocorrelation features of pixel gradients, the phase analysis module determines the pulsation center of the reference area and extracts the pulsation frequency curve of the uterine artery reference area. Using the uterine artery pulsation as an intrinsic clock reference source, errors caused by differences in the subject's heart rate baseline and ultrasound equipment gain are reduced. In clinical scenarios involving heterogeneous imaging terminals, due to differences in hardware gain settings and background levels, the phase analysis module executes an initialization procedure, establishing a background noise baseline using the gray-scale mean and gray-scale variance of the non-perfusion reference area, and calculating the normalized gain coefficient. ,in, To align the dimensionless scaling factors for different device dynamic ranges, the phase analysis module utilizes normalized gain coefficients. Perform grayscale field mapping on the acquired raw image to extract the discrete grayscale time series. The values ​​are within a consistent range, thereby reducing the deviation in physical dimensions caused by differences in equipment.

[0021] The frequency domain inversion module includes a virtual sensor subunit and a transform subunit. The virtual sensor subunit is used to perform temporal signal sampling on the reference area and the area to be evaluated in the image sequence to generate a discrete gray-scale time series. The transformation subunit is used to process discrete grayscale time series. Perform frequency domain transformation based on Fast Fourier Transform to extract the phase information of each pixel at the dominant frequency position. The frequency domain inversion module calculates the phase offset of all pixels in the global domain. The probability distribution function is given, and the phase difference distribution entropy is calculated according to the following formula. : ,in, Let N be the frequency of the phase shift occurring in the i-th phase sub-interval; N is the total number of preset phase interval divisions. Since endometrial repair exhibits spatial synchronicity, the frequency domain inversion module uses spatial domain phase angle data to calculate the phase shift distribution coefficient to quantify the spatial consistency of endometrial repair. It calculates the phase angle of each pixel within the endometrial evaluation area. Spatial distribution variance The phase offset distribution coefficient is established according to the following formula. : ,in, It is a dimensionless parameter; Let Variance be the spatial distribution variance of the phase angle within the region of interest, and let Entropy be the phase difference distribution. The phase shift distribution coefficient showed a decreasing trend after drug administration. When elevated, the microcirculatory response of the subject's endometrial tissue tends to become more ordered; the frequency domain inversion module is also used to construct the tissue dynamics transfer function. By inverting tissue mechanical properties, the technical problem of distinguishing between inflammatory congestion and functional repair through morphological observation alone is solved. The frequency domain inversion module treats endometrial tissue as a viscoelastic damped load driven by uterine artery pulse pressure, and calculates the energy transfer efficiency using the following formula: ,in, The cross-power spectrum of the grayscale sequence of the endometrial region response pixels and the grayscale sequence of the uterine artery reference pixels; To obtain the self-power spectrum of the reference pixel grayscale sequence, the frequency domain inversion module identifies the characteristic resonant frequency corresponding to the peak power in the energy transfer gain spectrum. When traditional Chinese medicine prescriptions exert their effects of softening and dispersing nodules, the viscoelastic damping of the tissue decreases, leading to a change in the characteristic resonant frequency. Shift towards higher frequency bands.

[0022] The evaluation index output module initiates the kinetic regression procedure to determine the transformation coefficient k of the linear mapping model, establishes a mapping using tissue biomechanical test data and image spectrum data from the sample library, and calculates the characteristic resonant frequency. With the damping coefficient of the tissue The changing first derivative establishes the parameter, which follows the mathematical constraints: ,in, is the response value for drug efficacy assessment; k is the tissue kinetic conversion coefficient, in units of ; This is the characteristic resonant frequency after drug administration; The characteristic resonant frequency before drug administration is evaluated by the index output module. The incremental values ​​are mapped to tissue compliance improvement parameters to quantify the degree of improvement in tissue physical properties. For low-frequency displacement noise introduced by respiratory motion, the frequency domain inversion module performs phase decoupling. The module identifies non-perfusion reference areas in the image sequence, extracts the motion displacement frequencies of these areas, and generates a noise mask in the frequency domain. The system uses the noise mask to perform orthogonal projection denoising on the phase shift. The system identifies the uterine serosal layer, which lacks blood flow pulsation signals, as a non-perfusion reference area in the image sequence and extracts its... to The principal components of the displacement vector within the breathing frequency band are used to generate the corresponding orthogonal mask matrix in the complex frequency domain. During orthogonal projection denoising, the system projects the original phase vector of the region to be evaluated onto the normal space of this mask matrix. The projection step size is determined by calculating the displacement cross-correlation coefficient between adjacent frames. Dynamic adjustment is employed. If the phase shift caused by breathing noise accounts for more than 30% of the total phase, the system will initiate secondary smoothing until the residual frequency energy is lower than the background baseline. This process removes spurious phase shifts caused by breathing motion and obtains calibrated phase characteristics. This method utilizes data from non-perfusion areas as an environmental constraint factor to improve the stability of evaluation indicators under non-ideal acquisition environments. The frequency domain inversion module is also used to extract global motion displacement residual data generated during image processing. Through the Power spectral density analysis was performed to determine the irritability index, which reflects the intensity of microcontractions in the uterine myometrium. ,in, The data consists of displacement residuals in pixels. The evaluation index output module is based on the agitation index. Weighted adjustments were applied to the response values ​​of the drug efficacy assessment, when the irritability index... When the level continues to rise, the system outputs a high-sensitivity warning for the repair environment, indicating that the traditional Chinese medicine formula needs to be more effective in addressing spasmodic smooth muscle. The evaluation index output module is used to generate an internal environment repair trend view and calculate repair stability parameters. The internal environment repair trend view includes parameters such as phase difference distribution entropy. The microcirculation reconstruction curve formed by the characteristic resonant frequency, and the characteristic resonant frequency The evolution line of tissue stiffness is formed, and the repair stability parameters are calculated by using characteristic resonant frequencies over three consecutive physiological cycles. The numerical variance was established to assess the probability of recurrence of endometrial tissue condition.

[0023] Example 1: In a specific clinical assessment scenario, the system faces a subject with sinus arrhythmia and significant fluctuations in respiratory rate. In this situation, the endometrial blood flow signal in the acquired dynamic multidimensional grayscale mapping feature matrix sequence is coupled with non-stationary physiological baselines and motion displacement artifacts, making it difficult to extract the tissue repair signal induced by the traditional Chinese medicine intervention based solely on grayscale feature analysis. The data preprocessing module acquires a 20-second dynamic multidimensional grayscale mapping feature matrix sequence of the endometrium on day 7 of treatment. This sequence uses a format conforming to medical digital imaging and communication. The standard medical image format contains multiple consecutive sampled frames. The phase analysis module uses the autocorrelation features of pixel gradients to identify the uterine artery reference area and extracts the frequency of... pulsation frequency curve ,in, The pulsation frequency of the uterine artery reference area, in units of Simultaneously, the frequency domain inversion module delineates a non-perfusion reference zone in the uterine serosal layer, using a synchronous acquisition frequency of [frequency value missing]. The respiratory displacement component, whose frequency is used as an environmental constraint factor for subsequent signal calibration; the frequency domain inversion module executes the aforementioned decoupling procedure, by measuring the uterine artery pulsation frequency curve. As an intrinsic reference source, the system extracts gray-level fluctuation sequences related to the cardiac cycle from a non-stationary background. ,in, The discrete grayscale value varies with the sampling time t.

[0024] During this process, the frequency domain inversion module uses a noise mask generated in the non-infusion reference region to perform orthogonal projection on the original phase data of the intima evaluation area, eliminating frequencies of [missing information]. The low-frequency displacement offset ensures the phase difference distribution entropy. The computational input only includes the effective phase changes caused by microcirculation perfusion, where, A dimensionless parameter characterizing the orderliness of phase distribution; this collaborative mechanism utilizes a deterministic reference source to provide a temporal benchmark for the grayscale sequence, enabling the frequency domain inversion module to extract tissue repair features in a background noise environment, thus solving the technical problem of feature extraction difficulties caused by the deep coupling of physiological fluctuations and drug efficacy response; the frequency domain inversion module calculates the phase difference distribution entropy at this time. The value was 1.95, showing a decreasing trend compared to 2.62 before drug administration. The transformation subunit analyzed the grayscale fluctuation sequence. Perform power spectral density calculations and construct the energy transfer gain spectrum. ,in, The transfer function describes the energy transfer efficiency; experimental results show that the characteristic resonant frequency... From before treatment Offset to ,in, The characteristic resonant frequency, in units of Because the degree of fibrosis in the tissue decreases after the Chinese medicine prescription takes effect, the high-frequency attenuation characteristics in the transfer function are changed, resulting in an improvement in energy transfer efficiency at higher frequencies. The system calculates the tissue's ability to transmit pulse pressure, transforming the evaluation process from morphological observation to kinetic parameter inversion. The final evaluation results show that the subject's tissue compliance improvement parameter shows a positive increment, and the repair stability parameter is maintained within the preset steady-state threshold range. Thus, it is determined that the endometrial repair state has entered the steady-state stage, and the output includes an evaluation view containing microcirculation remodeling curves and tissue stiffness evolution lines.

[0025] Example 2: In the experiment verifying the system's performance in extracting the efficacy signals of traditional Chinese medicine formulas under strong interference, the system faced compound noise interference consisting of Gaussian white noise and involuntary respiratory displacement of the subjects; the experiment used a digital medical image post-processing platform, and the data source was the clinically acquired endometrial medical digital imaging and communication specifications. The format is a sequence of images, and the frame rate of these images is set to [value missing]. The gray level measurement level is no less than 256, and the sampling time is 30 seconds; the sampling frequency is... The settings depend on the bandwidth of the high-frequency component of the uterine artery pulsation. To satisfy the Nyquist sampling theorem and reserve an oversampling margin of more than 2 times to suppress aliasing errors, the system will... Set as This setting balances the computational load during frequency domain transformation while ensuring the capture of transient phase changes in microcirculation. The experiment introduces a gradient of uterine adhesion severity as the core variable, categorized into levels 1 to 4, and establishes experimental, control group A, and control group B. The experimental group employs a complete scheme including arterial phase anchoring and orthogonal projection noise denoising, while control group A has its uterine artery reference frequency removed during processing. In the phase anchoring stage, control group B removed the noise orthogonal projection mask stage based on the non-perfusion reference region, and all groups were injected with background interference with a signal-to-noise ratio of 20dB; the system identified the frequency as... The characteristic waveform of uterine artery pulsation is used as an intrinsic clock reference source, and then the phase difference distribution entropy is extracted. With characteristic resonant frequency See Table 1.

[0026] Table 1: Comparison of performance indicators of different test groups under various working conditions

[0027] As the degree of tissue damage increases, the phase difference distribution entropy... With characteristic resonant frequency The measured values ​​show a monotonic evolution pattern. Under the same moderate adhesion (Level 2) condition, the phase difference distribution entropy of the experimental group is... The value was 2.15, while the control group A lacked [a certain value]. Temporal anchoring caused the phase calculation to lose its reference, with the value rising to 2.65. Control group B, due to the lack of denoising masking, had its respiratory displacement misjudged as perfusion fluctuation, resulting in a value deviation to 2.42. This data difference confirms the cooperative effect between uterine artery pulsation tracking and respiratory noise decoupling mechanisms; that is, the reference source provides synchronous input for phase analysis, enabling the system to maintain the stability of evaluation parameters even in low signal-to-noise ratio environments, achieving system-level gain. In the verification of adhesion levels with a large range, the nonlinear effect of tissue dynamics response was captured. When the tissue is in a state of extremely severe adhesion, the characteristic resonant frequency... The value is from Reduced to the boundary state At this point, the rate of decline in the evaluated response value tends to flatten out. This phenomenon reflects that the viscoelastic damping of the tissue is approaching the physical saturation region. Within this region, the improvement in tissue compliance is unlikely to generate a characteristic drift in the frequency domain gain spectrum. This is achieved by analyzing the discrete grayscale values ​​that change with sampling time t, i.e., the discrete grayscale time series. The process of generating the transfer function, i.e., the energy transfer gain spectrum, describing the energy transfer efficiency through power spectral transformation is analyzed, and the characteristic resonant frequency is determined. As traditional Chinese medicine prescriptions exert their effects of softening and dispersing nodules, they shift to the right in the high-frequency band.

[0028] Example 3: This example combines Figures 1 to 2 This paper describes a system for evaluating the efficacy of a traditional Chinese medicine formula used to treat intrauterine adhesions. Figure 1 As shown, the system obtains gray-scale temporal characteristic data of endometrial dynamic response through the data preprocessing module. This sequence is decomposed into reference anatomical domain feature vectors, reference pixel gray-scale data, and response stream response pixel gray-scale data in the response analysis domain. The data stream then enters the subsequent processing unit. The frequency domain inversion module is responsible for performing power spectral density calculation and energy transfer analysis, constructing an energy transfer gain spectrum, identifying peak power to determine the characteristic resonant frequency, and converting it into phase angle data by determining the time delay. The phase analysis module not only receives the original gray-scale data, but also performs cross-correlation calculation and phase angle distribution analysis in combination with the aforementioned generated characteristic resonant frequency to output the phase shift distribution coefficient. Finally, the evaluation index output module receives the characteristic resonant frequency and the phase shift distribution coefficient, generates and outputs an evaluation index of the degree of improvement in endometrial tissue viscoelasticity.

[0029] like Figure 2As shown, in the interactive functional architecture of the system, clinicians, as the main participants, establish connections with various functional units of the system. They log in and authenticate as users and have the authority to manage patient clinical records. In the data entry stage, clinicians are responsible for entering the prescription composition information. This process is supported by an external TCM knowledge base. At the same time, clinicians complete the task of acquiring multi-dimensional grayscale mapping feature matrix sequences by connecting to ultrasound imaging equipment. The core calculation process is triggered by the clinician to execute intelligent drug efficacy evaluation. The internal logic of this use case includes three key steps in sequence: calculating the phase shift coefficient, constructing the energy transfer gain spectrum, and calculating the drug efficacy response score. Finally, the system summarizes the processing results and feeds them back to the clinician through the function of generating an evaluation report, completing the complete workflow from information collection to decision support.

[0030] Example 4: In the post-drug administration assessment scenario for subjects with recurrent intrauterine adhesions, the system faces acoustic attenuation interference introduced by increased tissue thickness. It is necessary to determine whether the endometrial tissue has transitioned from early functional repair to a stable tissue remodeling phase. The data preprocessing module acquires a 40-second dynamic image sequence, and the phase analysis module performs region recognition, using a spatial difference operator of a three-row, three-column pixel matrix to calculate the grayscale gradient G, where G is the grayscale gradient value. The system sets the difference step size to the physical size of a single pixel and uses a 5-frame moving average window to smooth the original gradient flow, thereby removing instantaneous gradient jumps caused by transducer jitter and providing a stable spatial field input for subsequent feature extraction. The frequency domain inversion module performs a quantization process for the phase distribution order, converting 0 to... The phase angle space is discretized, and the specific execution logic includes determining the total number N of phase intervals according to the following formula: Where N is the total number of phase interval divisions; The preset phase resolution step size is set to 0.17 rad in this embodiment; The function is rounded up; the frequency domain inversion module performs histogram statistics on the phase offset of each pixel in the inner membrane evaluation area, and calculates the normalized frequency of the pixel in each sub-interval. Furthermore, through the phase difference distribution entropy Quantitative indicators are obtained; by using a fixed angular resolution step size, the calculation deviation caused by the difference in the setting of the statistical histogram group interval is eliminated, so that the values ​​of different measurement periods have physical comparability.

[0031] The evaluation indicator output module establishes evaluation conclusions by executing a stability calibration procedure and retrieving the characteristic resonant frequencies of the ovulation period for three consecutive physiological cycles of the subject. Data, through calculation of characteristic resonant frequency Sample variance Characterize the stability of the repair process; introduce a calibration method based on adaptive signal-to-noise ratio to determine the stability threshold. ,in, The value of is positively correlated with the grayscale background variance of the non-perfusion reference area; in the acquisition environment of this embodiment, the measured is... The frequency was 0.045 Hz², lower than the stability threshold of 0.060 Hz² calibrated for environmental noise, indicating that the subject's endometrial repair had reached a kinetically steady state. During this process, the system utilized the tissue damping coefficient... The changing trend can be used to verify the efficacy of the medicine. After the traditional Chinese medicine prescription reduces the tissue viscoelastic damping, the characteristic resonant frequency... offset and The changes in these parameters are inversely proportional, providing physical quantitative evidence for drug efficacy evaluation.

[0032] Example 5: In a clinical setting where the evaluation system is deployed on a heterogeneous ultrasound imaging terminal, inconsistent baseline grayscale distributions arise due to differences in hardware gain settings and background levels. To mitigate device-specific interference, an initialization procedure is executed. A 5-second resting-state ultrasound sequence is acquired during the first menstrual cycle before the subject takes the medication. The mean and variance of grayscale values ​​in the non-perfusion reference area are used to establish a background noise baseline, and the normalized gain coefficient is calculated. ,in, For dimensionless scaling factors used to align the dynamic ranges of different devices; the phase analysis module utilizes normalized gain coefficients. The raw images acquired by the data preprocessing module are subjected to grayscale field mapping, which enables the discrete grayscale time series extracted by the subsequent frequency domain inversion module to be processed. The values ​​are within a consistent range, thereby reducing the deviation of physical dimensions caused by equipment differences.

[0033] The evaluation index output module determines the transformation coefficient k of the linear mapping model by executing a kinetic regression procedure; it establishes a mapping relationship between tissue biomechanical test data and image spectrum data in the sample library, and calculates the characteristic resonant frequency. With the damping coefficient of the tissue The changing first derivative establishes the parameter, which follows the mathematical constraints: ,in, This is the response value for drug efficacy evaluation, which is a dimensionless parameter, and k is the tissue kinetic conversion coefficient, with units of... , The characteristic resonant frequency after drug administration, in units of , The characteristic resonant frequency before drug administration, in units of The system uses the regression residuals as input for adaptive weights, automatically adjusting the weight of the conversion coefficient k when the image signal-to-noise ratio is below a preset threshold to maintain the stability of the evaluation parameters under varying operating conditions; tissue dynamics conversion coefficient The method for determining this is to construct a calibration field using an in vitro biomimetic model with known physical properties, and to set the tissue damping coefficient of the in vitro biomimetic model. The gradient is varied in steps of 0.05 within the range of 0.1 to 0.8, and... to Recording the corresponding characteristic resonant frequencies under simulated pulse pressure drive Through calculation Follow The first derivative of the change is established The baseline value is associated with an index table consisting of the center frequency of the ultrasonic transducer and the detection depth, and is used by the evaluation index output module to calculate the efficacy evaluation response value. Output quantitative alignment data on the degree of tissue biomechanical improvement among different individuals.

[0034] Example 6: In a clinical scenario involving pre-parameter calibration of the subject system, the spatial domain pixel coordinate shift caused by variations in image depth and individual differences in uterine anatomy is encountered. The phase analysis module utilizes an automatic region of interest (ROI) identification procedure to establish the physical analysis boundary. Specifically, the system sets an initial search window of 100×100 pixels within a detection depth range of 40mm to 80mm. It identifies the uterine artery wall boundary using the periodic characteristics of the pixel grayscale gradient G in the time domain. Following the logic of determining the reference pixel center coordinates: the gradient flow within the search window is smoothed using a median filter for spatial noise reduction. The 10×10 pixel array with the highest gradient mean is selected as the physical core of the uterine artery reference area. Using this core as the origin, a 30×30 pixel endometrial evaluation area is delineated 15 pixels off the anatomical ventral direction. This spatial location calibration procedure eliminates the bias caused by manual area selection, ensuring that the grayscale fluctuation sequence acquired by the frequency domain inversion module... It has a unique physical location.

[0035] The frequency domain inversion module uses cross-correlation calculation procedures to determine the system time delay caused by the energy transfer distance. This provides synchronous calibration parameters for phase analysis, utilizing the reference pixel grayscale sequence of the uterine artery reference region. grayscale sequence of response pixels in the inner membrane evaluation area As input, the system searches for the cross-correlation coefficient within a sliding time window from -0.5s to 0.5s. The time delay before reaching the maximum value: ,in, Let be the dimensionless cross-correlation coefficient. For reference pixel grayscale sequence, In response to pixel grayscale sequences, The time delay is measured in seconds; the system measures the time delay. After 0.12s, this value is injected into the subsequent transformation sub-unit to calculate the phase angle. By eliminating phase lag components caused by physical distance, this calibration mechanism based on cross-correlation maxima search transforms the initial calibration process into a mathematical optimization procedure, ensuring the phase difference distribution entropy is effectively eliminated. Maintaining numerical stability at different acquisition depths; the frequency domain transformation of the transformation sub-unit is performed by processing the acquired discrete grayscale time series. Hanning window weighting is used to suppress spectral leakage, and the principal frequency position and phase angle are extracted by obtaining complex domain spectral data through fast Fourier transform. The time delay was measured using cross-correlation calculations. Combined with main frequency Phase compensation, the calculation method for the compensated phase angle is as follows: The phase lag caused by the physical distance between the uterine artery reference region and the endometrial region to be evaluated is eliminated, thus affecting the output phase shift distribution coefficient. Characterizes the spatial synchronicity of tissue microcirculation perfusion.

[0036] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0037] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A system for evaluating the efficacy of a traditional Chinese medicine formula for treating intrauterine adhesions, characterized in that, It includes a data preprocessing module, a phase analysis module, a frequency domain inversion module, and an evaluation index output module: The data preprocessing module is used to acquire gray-scale temporal feature data of endometrial dynamic response of subjects before and after taking traditional Chinese medicine prescriptions. The gray-scale temporal feature data of dynamic response includes gray-scale data of reference pixels corresponding to the feature vector of the reference anatomical domain and gray-scale data of response pixels corresponding to the gray-scale flow of the response analysis domain. The phase analysis module is used to perform cross-correlation operations on the grayscale data of the reference pixel and the grayscale data of the response pixel to determine the time delay, convert the time delay into phase angle data, and determine the phase offset distribution coefficient based on the variance of the phase angle data in the spatial domain. The frequency domain inversion module is used to perform power spectral density calculations on the reference pixel grayscale data and the response pixel grayscale data respectively. It constructs the energy transfer gain spectrum by calculating the power ratio of the response pixel power to the reference pixel power, and identifies the characteristic resonant frequency corresponding to the peak power in the energy transfer gain spectrum. The evaluation index output module is used to calculate the offset of the characteristic resonant frequency after drug administration relative to the characteristic resonant frequency before drug administration, and generate an evaluation index for evaluating the degree of improvement in the viscoelasticity of endometrial tissue based on the offset value combined with the change in the phase offset distribution coefficient.

2. The efficacy evaluation system for a traditional Chinese medicine formula used to treat intrauterine adhesions according to claim 1, characterized in that, The phase analysis module is also used to construct a local motion synchronization matrix. By calculating the phase response consistency index between adjacent pixel blocks, the phase analysis module identifies abnormal motion stagnation regions in the gray-scale temporal characteristic data of dynamic response, and uses the motion vector of the non-perfusion area to perform coordinate projection transformation to filter out the displacement component caused by respiration in the frequency domain and correct the phase offset distribution coefficient.

3. The efficacy evaluation system for a traditional Chinese medicine formula used to treat intrauterine adhesions according to claim 1, characterized in that, The frequency domain inversion module is also used to identify high-frequency cutoff points in the energy transfer gain spectrum; the evaluation index output module maps the incremental values ​​of the high-frequency cutoff points to tissue compliance improvement parameters, which are used to eliminate local congestion interference signals caused by acute inflammation during the evaluation process.

4. The efficacy evaluation system for a traditional Chinese medicine formula used to treat intrauterine adhesions according to claim 1, characterized in that, The phase analysis module is also used to obtain the baseline grayscale features of the subjects at preset physiological cycle nodes, and to subtract the periodic background fluctuations affected by hormone levels from the grayscale temporal feature data of the dynamic response in order to extract the efficacy component data belonging to the intervention of traditional Chinese medicine prescriptions.

5. The efficacy evaluation system for a traditional Chinese medicine formula used to treat intrauterine adhesions according to claim 1, characterized in that, When constructing the energy transfer gain spectrum, the frequency domain inversion module treats the endometrial tissue as a viscoelastic damped load driven by the uterine artery pulse pressure and uses a dynamic impedance analysis algorithm to calculate the energy transfer efficiency in order to determine the characteristic resonant frequency.

6. The efficacy evaluation system for a traditional Chinese medicine formula used to treat intrauterine adhesions according to claim 1, characterized in that, The data preprocessing module is also used to perform pixel tracking on the gray-scale temporal feature data of the dynamic response; the data preprocessing module generates motion vector fields of different sub-regions of the endometrium by calculating the gray-scale cross-correlation coefficient between two adjacent frames of images, and uses the motion vector fields as the input basis for the frequency domain inversion module.

7. The efficacy evaluation system for a traditional Chinese medicine formula used to treat intrauterine adhesions according to claim 1, characterized in that, The evaluation index output module is also used to generate an internal environment repair trend view; the internal environment repair trend view includes a microcirculation remodeling curve composed of phase shift distribution coefficients and a tissue stiffness evolution line composed of characteristic resonant frequencies.

8. The efficacy evaluation system for a traditional Chinese medicine formula for treating intrauterine adhesions according to claim 1, characterized in that, The system also includes a storage module; the storage module is used to store the raw data of the gray-scale time-series characteristics of the kinetic response of different subjects, as well as the corresponding historical efficacy evaluation indicators, in order to build a comparative database of efficacy for different Chinese medicine prescriptions.

9. The efficacy evaluation system for a traditional Chinese medicine formula for treating intrauterine adhesions according to claim 1, characterized in that, The evaluation index output module is also used to calculate repair stability parameters; Repair stability parameters are determined by calculating the numerical variance of characteristic resonant frequencies over three consecutive menstrual cycles, and are used to assess the long-term recurrence probability of endometrial tissue status.

Images