Hysteroscope image auxiliary diagnosis and treatment evaluation system fused with traditional Chinese and western medicine multi-modal representation

Abstract

The invention relates to the technical field of medical care informatics, and discloses a hysteroscope image auxiliary diagnosis and treatment evaluation system fusing traditional Chinese and western medicine multi-mode characterization, which comprises an optical field data acquisition module, a spectral texture orthogonal decoupling module, a heterogeneous flow field dynamic characterization module and a rheological state inversion mapping module, by decoupling specular reflection and diffuse reflection signals, a Qi-blood body fluid operation retardation quantitative index of a fluid drift model relative to an endometrial basement motion model is calculated, a light tissue relative slip vector is generated, a rheological characteristic index is inversed and is mapped to an orthogonal evaluation space in combination with chromaticity characteristics, and the evaluation accuracy is improved. According to the method, a microstate quantification mechanism based on fluid dynamics is constructed, the technical problem that the traditional Chinese medicine body fluid metabolism syndrome lacks objective physical characterization indexes is effectively solved, and the evaluation precision and diagnosis specificity of the pathological and physiological state of the uterine cavity microenvironment are improved.

Description

A Hysteroscopic Image-Assisted Diagnosis and Evaluation System Integrating Multimodal Characterization from Traditional Chinese and Western Medicine Technical Field

[0001] This invention belongs to the field of healthcare informatics technology, and in particular relates to a hysteroscopic image-assisted diagnosis and evaluation system that integrates multimodal representations of traditional Chinese and Western medicine. Background Technology

[0002] Currently, in the existing gynecological clinical diagnosis and treatment system, hysteroscopy is the core means of diagnosing intrauterine lesions. It mainly relies on high-resolution image sensors to collect reflected light signals in the uterine cavity, and performs noise reduction, edge sharpening and color correction through pathological feature analysis units. The core logic of this technology is to construct a clear anatomical view to help doctors identify the geometric morphological characteristics of space-occupying lesions such as polyps, fibroids or adhesions.

[0003] However, in scenarios involving functional assessment of the endometrial microenvironment or objective interpretation of TCM syndromes, the aforementioned imaging methods based on static anatomical morphology have limitations. Such assessments focus not on the overall geometric boundaries of the tissue, but rather on the dynamic changes in the tissue's microcirculation perfusion state and the physical properties of intracavitary secretions. Existing signal processing procedures typically treat weak light and color fluctuations or dynamic drifts in high-light reflection on the mucosal surface as environmental noise and smooth them out. This results in the filtering out of crucial physical information characterizing tissue metabolic activity and fluid rheological properties during the preprocessing stage. This approach forces the diagnostic and treatment system to rely solely on physicians when faced with functional lesions that have not yet formed physical lesions but have already shown abnormal blood and qi circulation or dampness accumulation. The subjective experience used for qualitative description cannot provide objective physical quantitative evidence. In addition to the limitations of hardware imaging methods, existing diagnostic assistance solutions also have shortcomings in signal processing and information dimension extraction. For example, Chinese invention patent application CN112863645A discloses an image diagnostic assistance system that records the entire diagnostic process through a camera, a recorder and an external laser positioning device to achieve diagnostic scene restoration and coarse spatial localization of lesions. The essence of this solution is limited to the overall image recording and physical marking level. It cannot analyze the deep tissue dynamic characteristics from the dynamic observation stream of monocular white light physiological state, and it is difficult to solve the problem of lack of objective physical quantitative indicators for syndromes such as body fluid metabolism in traditional Chinese medicine.

[0004] Therefore, the technical problem to be solved by this invention is how to use standard white light hysteroscopy to dynamically observe the physiological state, analyze the dynamic micro-perturbation and optical flow characteristics, decouple and quantify objective indicators that characterize the tissue microcirculation dynamics and fluid rheological properties, thereby achieving synchronous and accurate assessment of the intrauterine anatomical structure and functional state. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of this invention is as follows: A hysteroscopic image-assisted diagnosis and evaluation system integrating multimodal characterization of traditional Chinese and Western medicine, comprising: an optical field data acquisition module, configured to acquire a continuous dynamic observation stream of the physiological state of a biological cavity containing endometrial tissue under polarized illumination, and perform light intensity normalization processing based on histogram equalization, outputting a dynamic physiological information stream of the intrauterine microenvironment containing specular reflection signals of secretion surface and spectral information of endometrial basal tissue, the dynamic physiological information stream of the intrauterine microenvironment constituting an optical field input characterizing the tissue basal layer and the intrauterine secretions and irrigation fluid medium covering its surface; and a spectral texture orthogonal decoupling module, configured to perform multimodal parametric decoupling based on the saturation gradient in the HSI color space and the spectral continuity constraints of tissue sites, decoupling the dynamic physiological information stream of the intrauterine microenvironment into secretions characterizing the surface tension properties of intrauterine secretions and irrigation fluid medium. The system comprises a physical property characterization layer and a diffuse reflection signal layer characterizing the inherent texture of the tissue substrate; a heterogeneous flow field dynamics characterization module, configured to perform endometrial substrate biodynamic analysis on the diffuse reflection signal layer to establish an endometrial substrate motion model, and simultaneously perform fluid medium drift trajectory capture on the secretion physical property characterization layer to establish a fluid drift model, and generate a light tissue relative slip vector characterizing the state of medium qi, blood and body fluid flow obstruction by calculating the quantitative index of qi, blood and body fluid flow obstruction of the fluid drift model relative to the endometrial substrate motion model; and a rheological state inversion mapping module, configured to invert the rheological characteristic index through a preset fluid dynamics shear model based on the decay rate and phase lag parameter of the light tissue relative slip vector in the time dimension, and map the rheological characteristic index and chromaticity features to the rheological-chromatic orthogonal evaluation space, outputting a comprehensive evaluation index characterizing the metabolic and perfusion state of the intrauterine microenvironment.

[0006] Preferably, the spectral texture orthogonal decoupling module is specifically configured to perform the following operations: calculate the ratio of the brightness component to the saturation component of each microenvironment sampling unit in the dynamic physiological information flow of the uterine cavity microenvironment to generate a specular probability distribution matrix; secondly, compare the specular probability distribution matrix with a preset specular reflection critical threshold to generate a binarized specular mask; finally, use the specular mask to perform signal segmentation on the dynamic physiological information flow of the uterine cavity microenvironment to extract the physical characteristic characterization layer of secretions, and use the texture information of the outer region of the specular mask to reconstruct the diffuse reflection signal layer through the endometrial inherent pathological feature reconstruction logic.

[0007] Preferably, when generating the relative slip vector of the photoorganism, the heterogeneous flow field dynamics characterization module performs the following mathematical operation logic: obtain the fluid velocity vector in the fluid drift model at the same time t. Tissue displacement vector in the endometrial basal motion model ; Calculate the vector difference between the two ; for vector difference Perform a moving average filtering process based on a time window W to eliminate high-frequency noise interference and output a smoothed optical organization relative slip vector.

[0008] Preferably, the rheological state inversion mapping module is configured to calculate the rheological characteristic index according to the following formula. : Where N is the number of sampling points. Let be the relative slip vector of the optical organization at the i-th sampling point. This corresponds to the basal motion vector of the endometrium. The phase difference between the two and The preset weighting coefficients, The damping factor, To prevent non-zero constants with a denominator of zero; rheological characteristic index The numerical value is monotonically mapped to the physical viscosity grade of intrauterine secretions and irrigation fluid.

[0009] Preferably, it also includes: a depth-sensing spectral correction module, connected before the spectral texture orthogonal decoupling module, configured to construct a three-dimensional depth field of the cavity based on the degree of anatomical morphology characteristics of the cavity tissue, and to perform nonlinear compensation for the attenuation of light signals at different depth positions according to the Lambert-Beer law, so as to eliminate the spectral redshift deviation caused by medium absorption and ensure that the chromaticity distribution characteristics of the diffuse reflection signal layer truly reflect the inherent spectral reflectivity of the biological tissue substrate.

[0010] Preferably, the rheological state inversion mapping module further includes: a boundary layer turbulence determination submodule, used to calculate the curl characteristics of the relative slip vector of photoorganism in spatial distribution; when the curl characteristics exceed the preset laminar limit threshold, a flow anomaly marking signal is triggered, and the flow anomaly marking signal is configured to reduce the calculation weight of the rheological characteristic index in the comprehensive evaluation index of microenvironment diagnosis and treatment according to the preset weight decay function, until the curl characteristics recover to below the laminar limit threshold.

[0011] Preferably, the orthogonal state assessment space is defined as follows: the first dimension axis is the medium metabolic state axis generated based on the rheological characteristic index normalization, used to characterize the physical viscosity state of the fluid medium in the cavity; the second dimension axis is the perfusion state axis generated based on the red-green channel ratio and texture entropy value of the diffuse reflection signal layer, used to characterize the microcirculation filling state of the biological tissue substrate; the comprehensive assessment index for microenvironment differentiation and disease identification is the Euclidean distance of the current measurement point relative to the preset health benchmark origin in the orthogonal state assessment space.

[0012] Preferably, it also includes: a micro-motion spectrum feature extraction module, which is connected in parallel to the heterogeneous flow field dynamics characterization module and is configured to perform tissue micro-motion frequency analysis on the endometrial basal motion vector to extract micro-motion spectrum energy values ​​with frequencies in the range of 0.8Hz to 2.5Hz; and a rheological state inversion mapping module that uses the micro-motion spectrum energy values ​​as a correction factor reflecting the local biological activity of the biological tissue basal layer to dynamically compensate for the comprehensive evaluation index of microenvironment syndrome differentiation and disease diagnosis.

[0013] Preferably, the optical field data acquisition module includes: a polarization state time-series modulation submodule, configured to output polarization state control commands to adjust the polarization angle of the incident light, thereby acquiring multiple sets of original image data with different polarization states in a time series; and a spectral texture orthogonal decoupling module that uses the intensity difference information between images with different polarization states to help determine the spatial distribution boundary of the secretion physical property characterization layer.

[0014] Preferably, the comprehensive evaluation index for the diagnosis and treatment of diseases in the microenvironment is configured to drive the visualization interaction module to perform the following operations: construct and display an augmented reality image containing details of the texture of the biological tissue substrate; at the corresponding coordinate position of the augmented reality image, display the spatial distribution of rheological feature index in the form of a semi-transparent pathophysiological feature distribution cloud map layer, wherein the hue of the pathophysiological feature distribution cloud map layer is directly controlled by the numerical value of the rheological feature index, so as to intuitively present the physical heterogeneity of the intracavitary microenvironment.

[0015] Compared to existing technologies, this invention, a hysteroscopic image-assisted diagnostic and assessment system integrating multimodal characterization from both traditional Chinese and Western medicine, has the following advantages: 1. In hysteroscopic image-assisted diagnostic and assessment, based on the non-contact biofluid viscosity quantification achieved through differential mirror optical flow analysis, this invention solves the technical problem that existing monocular endoscopic imaging technology cannot perceive fluid physical viscosity without contacting the tissue by constructing a differential analysis mechanism between the mirror reflection component and the spectral information of the endometrial basal tissue in the time-domain motion field. It utilizes the high viscosity non-Newtonian fluid flowing on the tissue surface to produce… By separating and calculating the slip velocity difference between the specular optical flow vector field characterizing liquid film flow and the texture optical flow vector field characterizing the peristalsis of underlying tissue, the invisible fluid rheological properties are transformed into calculable kinematic differences. This physical parameter inversion logic based on the decoupling of dual optical flow fields enables the system to effectively distinguish between physiological fluids and pathological high-viscosity secretions that have similar visual features but different physical properties, eliminate misjudgments caused by the light and shadow aliasing effect, and improve the physical objectivity of the assessment of abnormal body fluid metabolism syndromes.

[0016] 2. Microcirculation perfusion dynamics identification based on frequency domain phase lag analysis: This invention introduces frequency domain phase analysis logic of time-series signals into the process of quantifying microscopic phenotypes, overcoming the ambiguity of static images in judging the local blood circulation status of tissues. The system extracts photoelectric signal components with the same frequency as the cardiac cycle within the region of interest and calculates the phase lag angle of each microenvironment sampling unit relative to the reference frame, establishing a mapping relationship from optical signal transmission delay to microvascular perfusion resistance. Utilizing the unique hemodynamic transmission characteristics of living tissue, it can accurately identify false color changes caused by external physical pressure from instruments or old blood accumulation, and only identify areas with phase conduction lag as pathological microcirculation disorders. Thus, based on the imaging system's ability to visualize tissue metabolic activity, it ensures the specificity and accuracy of the blood circulation assessment results.

[0017] 3. Based on depth topological prior, this invention utilizes three-dimensional depth data generated by anatomical structure analysis units as physical constraints to construct a media spectral transmission correction model that dynamically changes with optical path. This solves the interference of spectral nonlinear attenuation caused by intraoperative turbidity. Instead of using general global white balance adjustment, this technical solution establishes a functional relationship between anatomical depth and spectral scattering coefficient, and performs differentiated gain compensation on pixels at different depth positions in the image. This cross-channel collaborative mechanism that uses spatial geometric information to correct optical transmission deviations enables the system to restore the true spectral reflectance of deep tissues even during surgery when the medium gradually becomes turbid, avoiding false redshift of tissue color caused by blue light scattering attenuation, and ensuring the stability of syndrome feature quantification under complex conditions. Attached Figure Description

[0018] Figure 1 is a flowchart of the uterine image-assisted diagnosis and evaluation logic of the present invention, which integrates multimodal representations of traditional Chinese and Western medicine; Figure 2 is a schematic diagram of the composition of each functional module and the multimodal signal processing architecture of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] A hysteroscopic image-assisted diagnostic and assessment system integrating multimodal characterization from both traditional Chinese and Western medicine includes: an optical field data acquisition module configured to acquire a dynamic observation stream of the continuous physiological state of the biological cavity containing endometrial tissue under polarized illumination, and to perform intensity normalization processing based on histogram equalization, outputting a dynamic physiological information stream of the intrauterine microenvironment containing specular reflection signals from the secretion surface and spectral information of the endometrial basal tissue. This dynamic physiological information stream of the intrauterine microenvironment constitutes the optical field input characterizing the tissue basal layer and the intrauterine secretions and irrigation fluid medium covering its surface; and a spectral texture orthogonal decoupling module configured to perform multimodal parametric decoupling based on the saturation gradient in the HSI color space and the spectral continuity constraints of tissue sites, decoupling the dynamic physiological information stream of the intrauterine microenvironment into a secretion physical property characterization layer characterizing the surface tension characteristics of the intrauterine secretions and irrigation fluid medium, and a tissue characterization layer. The system comprises: a diffuse reflection signal layer representing the inherent texture of the uterine basal layer; a heterogeneous flow field dynamics characterization module configured to perform endometrial basal biodynamic analysis on the diffuse reflection signal layer to establish an endometrial basal motion model, and simultaneously perform fluid medium drift trajectory capture on the secretion physical property characterization layer to establish a fluid drift model. It also generates a relative slip vector of light tissue representing the state of obstruction in the flow of Qi, blood, and body fluids by calculating the quantitative index of Qi, blood, and body fluid circulation obstruction relative to the endometrial basal motion model of the fluid drift model; and a rheological state inversion mapping module configured to invert rheological characteristic indices based on the decay rate and phase lag parameters of the relative slip vector of light tissues in the time dimension using a preset fluid dynamics shear model. The rheological characteristic indices are then combined with chromatic features and mapped to a rheological-chromatic orthogonal evaluation space, outputting a comprehensive evaluation index of the microenvironment for syndrome differentiation and disease identification, representing the metabolic and perfusion state of the uterine cavity microenvironment.

[0021] Preferably, the spectral texture orthogonal decoupling module is specifically configured to perform the following operations: calculate the ratio of the brightness component to the saturation component of each microenvironment sampling unit in the dynamic physiological information flow of the uterine cavity microenvironment to generate a specular probability distribution matrix; secondly, compare the specular probability distribution matrix with a preset specular reflection critical threshold to generate a binarized specular mask; finally, use the specular mask to perform signal segmentation on the dynamic physiological information flow of the uterine cavity microenvironment to extract the physical characteristic characterization layer of secretions, and use the texture information of the outer region of the specular mask to reconstruct the diffuse reflection signal layer through the endometrial inherent pathological feature reconstruction logic.

[0022] Preferably, when generating the relative slip vector of the photoorganism, the heterogeneous flow field dynamics characterization module performs the following mathematical operation logic: obtain the fluid velocity vector in the fluid drift model at the same time t. Tissue displacement vector in the endometrial basal motion model ; Calculate the vector difference between the two ; for vector difference Perform a moving average filtering process based on a time window W to eliminate high-frequency noise interference and output a smoothed optical organization relative slip vector.

[0023] Preferably, the rheological state inversion mapping module is configured to calculate the rheological characteristic index according to the following formula. : Where N is the number of sampling points. Let be the relative slip vector of the optical organization at the i-th sampling point. This corresponds to the basal motion vector of the endometrium. The phase difference between the two and The preset weighting coefficients, The damping factor, To prevent non-zero constants with a denominator of zero; rheological characteristic index The numerical value is monotonically mapped to the physical viscosity grade of intrauterine secretions and irrigation fluid.

[0024] Preferably, it also includes: a depth-sensing spectral correction module, connected before the spectral texture orthogonal decoupling module, configured to construct a three-dimensional depth field of the cavity based on the degree of anatomical morphology characteristics of the cavity tissue, and to perform nonlinear compensation for the attenuation of light signals at different depth positions according to the Lambert-Beer law, so as to eliminate the spectral redshift deviation caused by medium absorption and ensure that the chromaticity distribution characteristics of the diffuse reflection signal layer truly reflect the inherent spectral reflectivity of the biological tissue substrate.

[0025] Preferably, the rheological state inversion mapping module further includes: a boundary layer turbulence determination submodule, used to calculate the curl characteristics of the relative slip vector of photoorganism in spatial distribution; when the curl characteristics exceed the preset laminar limit threshold, a flow anomaly marking signal is triggered, and the flow anomaly marking signal is configured to reduce the calculation weight of the rheological characteristic index in the comprehensive evaluation index of microenvironment diagnosis and treatment according to the preset weight decay function, until the curl characteristics recover to below the laminar limit threshold.

[0026] Preferably, the orthogonal state assessment space is defined as follows: the first dimension axis is the medium metabolic state axis generated based on the rheological characteristic index normalization, used to characterize the physical viscosity state of the fluid medium in the cavity; the second dimension axis is the perfusion state axis generated based on the red-green channel ratio and texture entropy value of the diffuse reflection signal layer, used to characterize the microcirculation filling state of the biological tissue substrate; the comprehensive assessment index for microenvironment differentiation and disease identification is the Euclidean distance of the current measurement point relative to the preset health benchmark origin in the orthogonal state assessment space.

[0027] Preferably, it also includes: a micro-motion spectrum feature extraction module, which is connected in parallel to the heterogeneous flow field dynamics characterization module and is configured to perform tissue micro-motion frequency analysis on the endometrial basal motion vector to extract micro-motion spectrum energy values ​​with frequencies in the range of 0.8Hz to 2.5Hz; and a rheological state inversion mapping module that uses the micro-motion spectrum energy values ​​as a correction factor reflecting the local biological activity of the biological tissue basal layer to dynamically compensate for the comprehensive evaluation index of microenvironment syndrome differentiation and disease diagnosis.

[0028] Preferably, the optical field data acquisition module includes: a polarization state time-series modulation submodule, configured to output polarization state control commands to adjust the polarization angle of the incident light, thereby acquiring multiple sets of original image data with different polarization states in a time series; and a spectral texture orthogonal decoupling module that uses the intensity difference information between images with different polarization states to help determine the spatial distribution boundary of the secretion physical property characterization layer.

[0029] Preferably, the comprehensive evaluation index for the diagnosis and treatment of diseases in the microenvironment is configured to drive the visualization interaction module to perform the following operations: construct and display an augmented reality image containing details of the texture of the biological tissue substrate; at the corresponding coordinate position of the augmented reality image, display the spatial distribution of rheological feature index in the form of a semi-transparent pathophysiological feature distribution cloud map layer, wherein the hue of the pathophysiological feature distribution cloud map layer is directly controlled by the numerical value of the rheological feature index, so as to intuitively present the physical heterogeneity of the intracavitary microenvironment.

[0030] Example 1: In a typical clinical application scenario for functional assessment of the endometrial microenvironment, the system receives real-time dynamic observation stream data of physiological status from a standard white light hysteroscopy. This scenario aims to quantify the intrauterine state, where no obvious morphological lesions have yet formed but microscopic pathophysiological changes exist. The optical field data acquisition module identifies the characteristic regions of surgical instruments appearing in the field of view of the video frame, namely metal electrosurgical loops or operating rods with known optical reflection properties, and uses them as dynamic reference benchmarks. The system calculates the color shift of this reference region in the current frame, generates compensation coefficients for the global illumination environment, and converts the original RGB format dynamic observation stream of physiological status to device-independent CIELAB uniform color. The spatial model establishes the luminosity basis for subsequent feature extraction. The orthogonal decoupling module for spectral texture, based on the saturation gradient in the HSI (hue-saturation-luminance) color space and the spectral continuity constraints of tissue sites, performs pixel-level signal separation, decoupling a single frame image into two independent signal layers: a secretion physical property characterization layer representing the surface tension characteristics of intrauterine secretions and irrigation fluid, and a diffuse reflection signal layer representing the inherent texture of the tissue substrate. The heterogeneous flow field dynamics characterization module performs differentiated motion analysis in parallel for these two heterogeneous signal layers. For the diffuse reflection signal layer, the system applies endometrial substrate biodynamic analysis methods to track the displacement of tissue surface texture feature points and establish an endometrial substrate motion model. For the physical property characterization layer of secretions, the system executes a fluid medium drift trajectory capture algorithm to establish a fluid drift model. The system performs vector difference operations to obtain quantitative indicators of the obstruction of qi, blood, and body fluid circulation between the fluid drift model and the endometrial basal motion model at the same time t. Its computational logic satisfies .

[0031] When there is highly viscous pathological secretion in the uterine cavity, the flow of qi, blood, and body fluids is obstructed due to the fluid shear thinning effect and surface tension hysteresis. The rheological state inversion mapping module exhibits specific damping decay characteristics over time. The decay rate and phase hysteresis parameters are used to invert the rheological characteristic index using a preset hydrodynamic shear model. The rheological characteristic index The calculation follows the formula below: Where N is the number of sampling points; Let be the relative slip vector of the optical organization at the i-th sampling point; This corresponds to the basal motion vector of the endometrium; The phase difference between the two; and These are preset weighting coefficients; It is the damping factor; To prevent non-zero constants with a denominator of zero, this rheological characteristic index objectively quantifies the physical viscosity grade of the medium, realizing a physical characterization of the rheological state of the intrauterine microenvironment. The micro-motion spectrum feature extraction module performs tissue micro-motion frequency analysis on the green channel component in the diffuse reflection signal layer, locking the weak photoplethysmography pulse wave signal with a frequency in the range of 0.8Hz to 2.5Hz. The system constructs a phase reference system with the high signal-to-noise ratio region at the center of the field of view as the reference, and calculates the phase lag angle of each microenvironment sampling unit in the region of interest relative to this reference. For regions with microcirculation perfusion disorders, this phase lag angle increases, and the system generates a blood circulation index accordingly. Finally, the system introduces Z-score based on historical big data. The standardization module subtracts the preset healthy mean of 0.2 from the rheological characteristic index and divides it by the standard deviation of 0.05 to obtain a dimensionless standardized rheological value. At the same time, it subtracts the preset baseline of 15 degrees from the Qi and Blood circulation index and divides it by the standard deviation of 5 degrees to obtain a dimensionless standardized Qi and Blood value. The square root of the sum of the squares of these two standardized values ​​is calculated. The rheological state inversion mapping module maps the rheological characteristic index and the Qi and Blood circulation index to the rheological-chromatic orthogonal evaluation space, and outputs a comprehensive evaluation index of the microenvironment that includes the state of media metabolism and tissue perfusion. This is presented on the display terminal in the form of a numerical matrix or a pathophysiological feature distribution cloud map, providing clinicians with quantitative evidence of the pathophysiological state of the uterine cavity microenvironment.

[0032] Example 2: In a typical experimental scenario for quantitatively verifying the rheological properties of the intrauterine microenvironment, the system faces the challenge of accurately distinguishing media with different physical viscosities and objectively characterizing their rheological features to verify the effectiveness of this technical solution in addressing the lack of quantitative indicators for TCM body fluid metabolism syndromes. The experimental platform is built on a fluid dynamics testing system that simulates the intrauterine microenvironment, including a biomimetic uterine model with known optical reflection properties. Its inner wall is covered with a biomaterial layer that simulates the texture of the endometrium and is filled with intrauterine secretions and irrigation fluid media with different rheological properties. The experimental data is acquired from a high-resolution industrial-grade CMOS image sensor with a sampling frequency set to 60fps to ensure that the capture of minute fluid displacements satisfies the Nyquist sampling theorem and avoids signal aliasing. To comprehensively evaluate the performance of the heterogeneous flow field dynamics characterization module, a set of control experiments with different viscosity grades were designed. The experimental media were divided into three groups: the first group was low-viscosity physiological saline, simulating the normal state of body fluids; the second group was medium-viscosity simulated mucus (prepared from sodium carboxymethyl cellulose, with a viscosity of approximately 150 mPa·s), simulating a wet and turbid state; and the third group was high-viscosity simulated pus (mixed from high-concentration starch and simulated cell debris, with a viscosity of approximately 800 mPa·s), simulating a sputum state. In each group of experiments, the same pulsed pressure excitation was applied to simulate the transient flow field changes caused by uterine contractions. To simulate interference in the real clinical environment, Gaussian white noise with an intensity of 10 dB was introduced during the experiment, and a baseline drift with a frequency of 0.5 Hz was superimposed to verify the algorithm's anti-interference ability.

[0033] After the experiment started, the optical field data acquisition module captured and output the dynamic physiological information flow of the intrauterine microenvironment in real time, which included the specular reflection signal of the secretion surface and the spectral information of the endometrial basal tissue. The spectral texture orthogonal decoupling module then processed the original dynamic physiological observation flow, separating the secretion physical property characterization layer, which characterizes the surface properties of the fluid, and the diffuse reflection signal layer, which characterizes the endometrial texture. In the first set of low-viscosity media tests, the fluid drift model calculated by the heterogeneous flow field dynamics characterization module... Endometrial basal motion model They exhibit a high degree of synchronicity, with quantitative indicators of the obstruction in the flow of Qi, blood, and body fluids between them. The rheological characteristic index remains at a low level close to zero. Maintaining a value between 0.1 and 0.3 indicates excellent medium flowability. In the tests of the second and third groups of high-viscosity media, rheological differences were observed with increasing medium viscosity, particularly in the third group of high-viscosity media, where pressure excitation was applied. Compared to It exhibits obvious phase lag and amplitude attenuation, and quantitative indicators of qi, blood and body fluid circulation obstruction. It exhibits damped oscillation characteristics in the time series, with its peak value increasing nonlinearly with increasing viscosity. The rheological state inversion mapping module is based on the formula... Calculations were performed, and the results showed the rheological characteristic index of the third group of media. The value jumped to above 0.8 and showed a strong positive correlation with the physical viscosity measured by a standard viscometer (correlation coefficient r>0.95). This data trend clearly confirms the rheological characteristic index. It can effectively quantify the physical viscosity grade of the medium, thereby achieving an objective characterization of high-viscosity syndromes such as phlegm and fluid retention in traditional Chinese medicine; further noise interference tests show that, with the introduction of 10dB white noise and baseline drift, the untreated original... The signal exhibits violent random fluctuations, making it difficult to extract effective rheological features. However, after the spatiotemporal joint filtering processing of this invention, The curve recovered a clear damping decay pattern, and the rheological characteristic index... The fluctuation deviation was controlled within 5%, proving that the scheme can still maintain high stability and measurement accuracy in complex noise environment. Finally, the rheological-chromatic orthogonal evaluation index output by the rheological state inversion mapping module accurately mapped the three different media to the preset normal, damp and turbid and phlegm evaluation intervals, verifying the clinical application value of the system in auxiliary diagnosis and treatment evaluation.

[0034] Example 3: Regarding the calibration of key model parameters and the stability assurance of system boundary conditions in the rheological state inversion mapping module, this example constructs a complete technical closed loop including standardized engineering calibration procedures and an abnormal state circuit breaker mechanism. On a simulated uterine cavity testing platform at a constant temperature of 37℃, the system sequentially pumps in a set of standardized uterine cavity secretions and irrigation fluid media with gradient physical viscosity values, such as 50 mPa·s to 1000 mPa·s, to simulate the continuous rheological state from normal body fluids to pathological phlegm. The optical field data acquisition module captures the dynamic physiological state observation flow of the above-mentioned standard media under preset pressure pulse excitation. The heterogeneous flow field dynamic characterization module synchronously outputs the corresponding optical tissue relative slip vector sequence. The processor, based on the least squares criterion, quantifies the measured qi, blood, and body fluid flow stagnation indicators. and phase difference Substitute the preset fluid dynamics shear model and determine the weighting coefficients through iterative calculations. , and damping factor The optimal solution is thus established during the system initialization phase, thereby establishing the rheological characteristic index. The deterministic monotonic mapping relationship between the physical viscosity grade and the physical viscosity grade eliminates the randomness and empirical dependence of parameter settings.

[0035] Based on this, to avoid non-pathological interference introduced by drastic intraoperative manipulation or sudden environmental changes, the system incorporates confidence-gated logic based on motion vector thresholds. When the endometrial basal motion model... When the instantaneous modulus exceeds the preset physiological tremor threshold, i.e., when an unexpected, violent displacement or large-amplitude scanning motion of the probe is detected, the system triggers a circuit breaker mechanism. This automatically suspends the integration and updating of the rheological characteristic index, maintaining the output state of the previous valid frame until the motion vector falls back to the stable observation range. This mechanism effectively eliminates the misleading influence of motion artifacts on the assessment of microviscosity, ensuring that the output index only reflects the rheological properties of the medium itself. Furthermore, regarding data security and compliance architecture, the system adopts an edge computing strategy, offloading core computing power such as orthogonal decoupling of spectral texture and rheological state inversion to the local execution of the front-end acquisition device. Data output to the upper-level display terminal or storage server is strictly limited to encrypted and de-identified data. Numerical matrices and orthogonal evaluation coordinates, along with the original high-resolution dynamic observation stream of the physiological state of the uterine cavity and endometrial tissue, are discarded or temporarily stored in volatile memory after real-time feature extraction is completed. This achieves physical isolation and minimal utilization of the patient's biological anatomical privacy information at the system's underlying architecture.

[0036] Example 4: In the initial calibration scenario for standardizing the consistency of system optical response and spectral accuracy, the optical field data acquisition module performs a white balance correction process for the specific light source color temperature and sensor spectral response characteristics during the device startup phase. The system guides the probe to align with a standard diffuse white board placed in a sterile environment, acquires multiple frames of static calibration images, and calculates the gain compensation coefficient matrix of the RGB three channels. To correct the photoelectric conversion deviation of the CMOS sensor, the system loads a preset spectral mapping lookup table and linearly transforms the device's unique original color space to the CIE1931 standard colorimetric system. This eliminates the input spectral drift caused by light source aging or batch differences in the device, and establishes the absolute physical benchmark for subsequent colorimetric feature analysis.

[0037] When the system enters the pre-clinical interventional uterine distension steady-state establishment phase, the heterogeneous flow field dynamics characterization module automatically initiates baseline drift suppression and environmental parameter normalization procedures. The system monitors the average modulus of the optical flow vector across the entire field of view in real time. When the time variance of this modulus is within a continuous range... Within seconds, the value falls below the preset steady-state threshold. When the uterine distension pressure is determined to have reached a state of hydrodynamic equilibrium, the system immediately locks the anatomical feature points of the current frame as the geometric reference frame at time zero, and records the current background noise level as the signal-to-noise ratio threshold for subsequent micro-motion signal extraction, ensuring the rheological characteristic index. The calculation is based on a physical environment with stable pressure and corrected geometric distortion, thus enabling objective comparability and statistical significance of pathophysiological indicators among different patients.

[0038] Example 5: In complex application scenarios involving dynamic deformation of the uterine cavity and air bubble interference, the system executes an effective analysis region locking procedure to isolate the effective fluid-tissue interaction region from non-informational background noise. The heterogeneous flow field dynamics characterization module calculates the local information entropy H(x,y) of the diffuse reflection signal layer and the brightness gradient modulus of the secretion physical property characterization layer frame by frame. And generate a binarized validity mask. When the entropy value of the microenvironment sampling unit exceeds the preset texture threshold, And its gradient value falls within the laminar flow range. The microenvironment sampling unit is activated only when the light is within the range, and the spatial filtering logic will organize the relative slip vector of the light. The computational domain is physically constrained at the solid-liquid coupling interface, thereby effectively shielding computational singularities caused by suspended bubbles or floating debris.

[0039] Meanwhile, to address the modulation effect of patient circadian rhythms on microfluidic field measurements, the system implements an adaptive time windowing strategy for signal integration. The micro-motion spectral feature extraction module separates the heart rate component frequency from the green channel signal. And dynamically adjust the width of the sliding filter window. To meet The relationship is defined by n, where n is a positive integer, to synchronize the sampling period with the physiological pulsation period to minimize aliasing error. In the initial stage of system startup, for the first 3 seconds before a valid heart rate component is detected, the sliding filter window width is forcibly locked to a default safety value of 1.5 seconds to cover the lowest physiological frequencies above 0.67 Hz. Once the signal-to-noise ratio of the detected heart rate signal exceeds 6 dB, the system switches to synchronous sampling mode. In addition, the signal quality gating logic calculates the normalized cross-correlation coefficient between the fluid drift vector and the tissue motion vector within each window. and automatically discard Below the coherence threshold The data frame, because this state implies uncoupled chaotic motion, thus ensuring the rheological characteristic exponent of the final output. It converges to a statistical value that reflects the inherent viscosity of the medium, rather than a transient disturbance.

[0040] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.

Claims

1. A hysteroscopic image-assisted diagnosis and assessment system integrating multimodal representations from both traditional Chinese and Western medicine, characterized in that, include: The optical field data acquisition module is configured to acquire a dynamic observation stream of the continuous physiological state of the biological cavity under polarized illumination conditions, and perform light intensity normalization processing based on histogram equalization to output a dynamic physiological information stream of the uterine cavity microenvironment containing specular reflection signals from the secretion surface and spectral information of the endometrial basal tissue. The spectral texture orthogonal decoupling module is configured to decouple the dynamic physiological information stream of the uterine cavity microenvironment into a secretion physical property characterization layer and a diffuse reflection signal layer based on the saturation gradient in the HSI color space and the spectral continuity constraints of tissue sites. The heterogeneous flow field dynamic characterization module is configured to establish an endometrial basal motion model for the diffuse reflection signal layer and a fluid drift model for the secretion physical property characterization layer, and generate a relative slip vector of light tissue by calculating the quantitative index of the qi, blood and body fluid flow stagnation of the fluid drift model relative to the endometrial basal motion model. The rheological state inversion mapping module is configured to invert the rheological characteristic index based on the decay rate and phase hysteresis parameter of the relative slip vector of photoorganism in the time dimension through a preset hydrodynamic shear model, and then map the rheological characteristic index and chromaticity feature to the rheological-chromatic orthogonal evaluation space to output a comprehensive evaluation index for microenvironment diagnosis and treatment.

2. The hysteroscopic image-assisted diagnosis and assessment system integrating multimodal representations of traditional Chinese and Western medicine as described in claim 1, characterized in that, The spectral texture orthogonal decoupling module is specifically configured to perform the following operations: calculate the ratio of the brightness component to the saturation component of each microenvironment sampling unit in the dynamic physiological information flow of the uterine cavity microenvironment to generate a specular probability distribution matrix; secondly, compare the specular probability distribution matrix with a preset specular reflection threshold to generate a binarized specular mask; finally, use the specular mask to segment the dynamic physiological information flow of the uterine cavity microenvironment to extract the physical characteristic characterization layer of secretions, and use the texture information of the outer region of the specular mask to reconstruct the diffuse reflection signal layer through the endometrial inherent pathological feature reconstruction logic.

3. The hysteroscopic image-assisted diagnosis and assessment system integrating multimodal representations of traditional Chinese and Western medicine as described in claim 1, characterized in that, When generating the relative slip vector of the photostructure, the heterogeneous flow field dynamics characterization module performs the following mathematical operation logic: obtains the fluid velocity vector in the fluid drift model at the same time t. Tissue displacement vector in the endometrial basal motion model ; Calculate the vector difference between the two ; for vector difference Perform a moving average filtering process based on a time window W to eliminate high-frequency noise interference and output a smoothed optical organization relative slip vector.

4. The hysteroscopic image-assisted diagnosis and assessment system integrating multimodal representations of traditional Chinese and Western medicine as described in claim 1, characterized in that, The rheological state inversion mapping module is configured to calculate the rheological characteristic index based on the following formula. ： Where N is the number of sampling points. Let be the relative slip vector of the optical organization at the i-th sampling point. This corresponds to the basal motion vector of the endometrium. The phase difference between the two and The preset weighting coefficients, The damping factor, To prevent non-zero constants with a denominator of zero; rheological characteristic index The numerical value is monotonically mapped to the physical viscosity grade of intrauterine secretions and irrigation fluid.

5. The hysteroscopic image-assisted diagnosis and assessment system integrating multimodal representations of traditional Chinese and Western medicine as described in claim 1, characterized in that, Also includes: The depth-sensing spectral correction module, connected before the spectral texture orthogonal decoupling module, is configured to construct a three-dimensional depth field of the cavity based on the degree of anatomical morphology features of the cavity tissue, and to perform nonlinear compensation for the attenuation of light signals at different depth positions according to the Lambert-Beer law, so as to eliminate the spectral redshift deviation caused by medium absorption.

6. The hysteroscopic image-assisted diagnosis and assessment system integrating multimodal representations of traditional Chinese and Western medicine as described in claim 1, characterized in that, The rheological state inversion mapping module also includes a boundary layer turbulence determination submodule, which is used to calculate the curl characteristics of the relative slip vector of photoorganism in spatial distribution; when the curl characteristics exceed the preset laminar limit threshold, a flow anomaly marking signal is triggered. The flow anomaly marking signal is configured to reduce the calculation weight of the rheological characteristic index in the comprehensive evaluation index of microenvironment diagnosis and treatment according to the preset weight decay function, until the curl characteristics recover to below the laminar limit threshold.

7. The hysteroscopic image-assisted diagnosis and assessment system integrating multimodal representations of traditional Chinese and Western medicine as described in claim 1, characterized in that, The orthogonal state assessment space is defined as follows: the first dimension axis is the medium metabolic state axis generated based on the rheological characteristic index normalization, used to characterize the physical viscosity state of the fluid medium in the cavity; the second dimension axis is the perfusion state axis generated based on the red-green channel ratio and texture entropy value of the diffuse reflection signal layer, used to characterize the microcirculation filling state of the biological tissue substrate; the comprehensive assessment index for microenvironment differentiation and disease identification is the Euclidean distance of the current measurement point relative to the preset health benchmark origin in the orthogonal state assessment space.

8. The hysteroscopic image-assisted diagnosis and evaluation system integrating multimodal representations of traditional Chinese and Western medicine as described in claim 1, characterized in that, Also includes: The micro-motion spectrum feature extraction module, connected in parallel to the heterogeneous flow field dynamics characterization module, is configured to perform tissue micro-motion frequency analysis on the endometrial basal motion vector and extract the micro-motion spectrum energy value with a frequency range of 0.8Hz to 2.5Hz. The rheological state inversion mapping module uses the micro-motion spectrum energy value as a correction factor reflecting the local biological activity of the biological tissue basal layer and dynamically compensates the comprehensive evaluation index of the microenvironment for syndrome differentiation and disease diagnosis.

9. The hysteroscopic image-assisted diagnosis and evaluation system integrating multimodal representations of traditional Chinese and Western medicine as described in claim 1, characterized in that, The optical field data acquisition module includes: a polarization state time-series modulation submodule, which is configured to output polarization state control commands to adjust the polarization angle of the incident light, thereby acquiring multiple sets of raw image data with different polarization states in a time series; and a spectral texture orthogonal decoupling module that uses the intensity difference information between images with different polarization states to help determine the spatial distribution boundary of the secretion physical property characterization layer.

10. The hysteroscopic image-assisted diagnosis and assessment system integrating multimodal representations of traditional Chinese and Western medicine as described in claim 1, characterized in that, The comprehensive evaluation index for microenvironment-based diagnosis and treatment is configured to drive the visualization interaction module to perform the following operations: construct and display augmented reality images containing details of the texture of biological tissue substrates; At the corresponding coordinate positions of the augmented reality image, the spatial distribution of rheological feature indices is superimposed and displayed as a semi-transparent pathophysiological feature distribution cloud map layer, wherein the hue of the pathophysiological feature distribution cloud map layer is directly controlled by the numerical value of the rheological feature index.

Images