Traditional Chinese medicine intervention on fatty liver curative effect image evaluation method and system

Abstract

The present application relates to the technical field of image analysis, in particular to a traditional Chinese medicine intervention on fatty liver curative effect image evaluation method and system, through obtaining the ultrasonic radio frequency data with timestamp, probe pose and abdominal audio stream; extracting the intestinal sound envelope signal to determine the prior trigger time window of intestinal gas wandering and intercepting the dynamic frame set; dividing the dynamic acoustic shadow and the non-shielding echo area, combining the probe pose to preliminarily project the candidate echo matrix; extracting the intrahepatic blood vessel topological feature to calculate the spatial topological deviation value, generating the respiratory compensation vector based on the closed loop feedback and updating the probe pose until the deviation is up to standard; fusing the up-to-standard matrix to the acoustic shadow area to reconstruct the unobstructed liver three-dimensional echo matrix, and then calculating the fat attenuation coefficient to generate the evaluation result; through multi-modal acoustic triggering and bottom layer adaptive respiratory displacement compensation, the interference of intestinal gas shielding and respiratory deformation is avoided, and high-precision and objective fatty liver imaging quantitative evaluation is realized.

Description

Technical Field

[0001] This invention relates to the field of image analysis technology, specifically to an image evaluation method and system for the efficacy of traditional Chinese medicine intervention in fatty liver. Background Technology

[0002] In the field of evaluating the treatment efficacy of fatty liver disease, existing non-invasive assessment methods mainly rely on ultrasound imaging technology and attenuation coefficient measurement logic. By capturing the energy loss characteristics of the ultrasound beam in liver tissue, the degree of fat infiltration in the liver parenchyma can be quantitatively characterized. This can, to a certain extent, avoid the invasive risks of liver biopsy, provide a digital reference for clinical monitoring of the dynamic evolution of liver steatosis, and enable medical staff to conduct preliminary objective analysis and dose-effect assessment of the patient's pathological status based on acoustic physical parameters.

[0003] However, existing technologies cannot achieve precise alignment of the dynamic spatial pose between the detection terminal and internal organs in complex clinical application environments. Because the liver is a flexible organ that undergoes nonlinear displacement with respiration, and the acoustic shadowing generated by gas flowing within the abdominal cavity frequently obscures the effective echo area, traditional static sampling or open-loop reconstruction methods struggle to maintain consistent sampling benchmarks during dynamic respiratory cycles. This spatial instability results in extracted physical features often being mixed with a large number of motion artifacts and noise errors, making it impossible for the final attenuation parameters to accurately reproduce the data evolution of the actual anatomical site, thus affecting the accuracy of determining the efficacy of traditional Chinese medicine intervention for fatty liver. Summary of the Invention

[0004] To address the problems in related technologies, this invention provides an image evaluation method and system for the efficacy of traditional Chinese medicine intervention in fatty liver, thereby overcoming the aforementioned technical problems in existing related technologies.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a method for image evaluation of the efficacy of traditional Chinese medicine intervention in fatty liver, comprising the following steps:

[0006] Acquire timestamped ultrasound radio frequency data streams, probe spatial pose data, and abdominal audio streams;

[0007] The bowel sound envelope signal in the preset frequency band is extracted from the abdominal audio stream, and the a priori triggering time window characterizing intestinal gas movement is determined based on the bowel sound envelope signal.

[0008] Extract a set of target ultrasonic dynamic frames from the ultrasonic radio frequency data stream that are aligned with the a priori trigger time window;

[0009] Dynamic acoustic shadow regions and corresponding unmasked echo regions are divided into the target ultrasound dynamic frame set.

[0010] Based on the probe spatial pose data, a preliminary three-dimensional spatial projection is performed on the unmasked echo region at different time phases to obtain the candidate echo matrix;

[0011] Extract the intrahepatic vascular topological features from the candidate echo matrix, and calculate the spatial topological deviation values ​​between the intrahepatic vascular topological features at different time phases;

[0012] Determine whether the spatial topology deviation value is greater than a preset deviation threshold; if so, generate a respiratory displacement compensation vector based on the intrahepatic vascular topology features and update the probe spatial pose data, and re-execute the preliminary three-dimensional spatial projection based on the updated probe spatial pose data until the spatial topology deviation value is less than or equal to the preset deviation threshold.

[0013] Candidate echo matrices that meet the preset deviation threshold constraint are used as target echo matrices. The target echo matrices are fused into the dynamic sound and shadow region to reconstruct an unobstructed three-dimensional echo matrix of the liver.

[0014] The target fat attenuation coefficient is calculated based on the unobstructed three-dimensional echo matrix of the liver, and the corresponding fatty liver treatment efficacy assessment results are generated.

[0015] Preferably, extracting the bowel sound envelope signal in a preset frequency band from the abdominal audio stream includes the following steps:

[0016] The abdominal audio stream is subjected to time-frequency conversion processing to obtain audio spectrum data;

[0017] Based on a preset blind source separation algorithm, physiological background noise components in the audio spectrum data are separated and filtered out, and bowel sound audio spectrum data in the preset frequency band are extracted.

[0018] Envelope extraction processing is performed on the bowel sound frequency spectrum data to obtain the bowel sound envelope signal.

[0019] Preferably, determining the prior triggering time window characterizing intestinal gas migration includes the following steps:

[0020] The first-order derivative of the bowel sound envelope signal is performed to obtain the energy change rate sequence;

[0021] Extract the start and end sampling time points from the energy change rate sequence whose values ​​are greater than a preset change rate threshold.

[0022] The time interval between the start sampling time point and the end sampling time point is used as the prior trigger time window; wherein, the prior trigger time window is used to define the high-risk time period in which the intense movement of intestinal gas causes acoustic shadowing of ultrasound images.

[0023] Preferably, the step of dividing the dynamic acoustic shadow region and the corresponding unmasked echo region in the target ultrasound dynamic frame set includes the following steps:

[0024] Calculate the gradient data of the radio frequency signal intensity of each pixel in each frame of the target ultrasound dynamic frame set as a function of scanning depth;

[0025] Extract the pixels whose change gradient data is greater than the preset signal attenuation gradient from each frame of the target ultrasound dynamic frame set, and mark the spatial region formed by the pixels as the dynamic sound and shadow region.

[0026] Extract the remaining pixels from the target ultrasound dynamic frame set, excluding the dynamic sound shadow region, and mark the spatial region formed by the remaining pixels as the unmasked echo region under the corresponding time phase.

[0027] Preferably, calculating the spatial topological deviation value between the intrahepatic vascular topological features at different time phases includes the following steps:

[0028] The centroid coordinates of the blood vessel bifurcation are identified from the three-dimensional spatial data of different time phases in the candidate echo matrix.

[0029] Based on a preset feature matching algorithm, the centroid coordinates of the bifurcation of blood vessels that match at different time phases are determined as the centroid coordinates of the bifurcation of the same blood vessels, and the centroid coordinates of the bifurcation of the same blood vessels are used as the topological features of the intrahepatic blood vessels.

[0030] The end-expiratory time phase among the different time phases is determined as the reference phase, and the spatial displacement vector data of the centroid of each homologous vessel bifurcation under each non-reference time phase is calculated relative to the corresponding centroid under the reference phase.

[0031] The spatial displacement vector data is weighted and synthesized, and the vector magnitude of the weighted synthesis result is used as the spatial topology deviation value; wherein, the spatial topology deviation value is used to characterize the nonlinear spatial deformation physical error of the vascular anatomy structure inside the liver due to the patient's respiratory motion between different time phases.

[0032] Preferably, the step of generating a respiratory displacement compensation vector based on the intrahepatic vascular topological features and updating the probe spatial pose data includes the following steps:

[0033] Based on the direction vector and vector magnitude of the spatial displacement vector data under each non-reference time phase, spatial translation matrix and spatial rotation matrix are calculated to quantify the linear offset and angular deflection of the liver anatomical structure under each non-reference time phase.

[0034] Based on the spatial translation matrix and the spatial rotation matrix, a respiratory displacement compensation vector is generated for each non-reference time phase.

[0035] The probe spatial pose data is superimposed with the respiratory displacement compensation vector to obtain the updated probe spatial pose data.

[0036] Preferably, the reconstruction to obtain an unobstructed three-dimensional echo matrix of the liver includes the following steps:

[0037] Identify the overlapping spatial regions of the target echo matrix at different time phases;

[0038] Calculate the local signal-to-noise ratio data of the overlapping spatial region and generate a dynamic fusion weight distribution;

[0039] Based on the dynamic fusion weight distribution, the unmasked echo data in the target echo matrix is ​​weighted and calculated, and the weighted echo data is filled into the three-dimensional spatial coordinates corresponding to the dynamic sound shadow region to reconstruct and generate a physically continuous unmasked three-dimensional echo matrix of the liver.

[0040] Preferably, generating the corresponding fatty liver treatment efficacy assessment result includes the following steps:

[0041] Extract the radio frequency envelope signal sequence along the scanning depth direction from the unobstructed liver three-dimensional echo matrix;

[0042] Calculate the energy attenuation gradient of the radio frequency envelope signal sequence to obtain the target fat attenuation coefficient;

[0043] Acquire TCM syndrome image data corresponding to the consultation cycle of the ultrasound radio frequency data stream; wherein, the TCM syndrome image data includes at least facial image data and tongue image data;

[0044] Color and texture features are extracted from the TCM syndrome image data, and the extraction results are input into a preset feature mapping model for processing to obtain the TCM syndrome quantitative integral.

[0045] Calculate the temporal decline slope of the target fat attenuation coefficient and the temporal improvement slope of the quantitative score of the TCM syndrome.

[0046] Calculate the correlation index between the time series decline slope and the time series improvement slope;

[0047] The correlation index is numerically compared with the preset correlation threshold range in the preset efficacy evaluation standard, and the efficacy level label corresponding to the preset correlation threshold range that matches the correlation index is output to obtain the fatty liver efficacy evaluation result; wherein, the preset efficacy evaluation standard includes at least the preset correlation threshold range corresponding to different efficacy level labels.

[0048] The present invention also includes an image evaluation system for the efficacy of traditional Chinese medicine intervention in fatty liver, comprising a data acquisition module, a priori trigger time window determination module, a region division module, a spatial topology deviation measurement module, a probe spatial pose update module, and a fatty liver efficacy evaluation module.

[0049] The data acquisition module is used to acquire time-stamped ultrasound radio frequency data streams, probe spatial pose data, and abdominal audio streams.

[0050] The prior trigger time window determination module is used to extract the bowel sound envelope signal in a preset frequency band from the abdominal audio stream, and determine the prior trigger time window characterizing the movement of intestinal gas based on the bowel sound envelope signal.

[0051] The region segmentation module is used to extract a set of target ultrasound dynamic frames aligned with the prior trigger time window from the ultrasound radio frequency data stream; and to segment the target ultrasound dynamic frame set into a dynamic acoustic shadow region and a corresponding unmasked echo region.

[0052] The spatial topology deviation measurement module is used to perform preliminary three-dimensional spatial projection of the unmasked echo region at different time phases based on the probe spatial pose data to obtain a candidate echo matrix; extract the intrahepatic vascular topological features in the candidate echo matrix, and calculate the spatial topology deviation value between the intrahepatic vascular topological features at different time phases.

[0053] The probe spatial pose update module is used to determine whether the spatial topology deviation value is greater than a preset deviation threshold; if so, a respiratory displacement compensation vector is generated based on the intrahepatic vascular topology features and the probe spatial pose data is updated. The preliminary three-dimensional spatial projection is re-executed based on the updated probe spatial pose data until the spatial topology deviation value is less than or equal to the preset deviation threshold.

[0054] The fatty liver treatment efficacy assessment module is used to take the candidate echo matrix that meets the preset deviation threshold constraint as the target echo matrix, fuse the target echo matrix into the dynamic sound shadow region, and reconstruct an unobstructed three-dimensional echo matrix of the liver; calculate the target fat attenuation coefficient based on the unobstructed three-dimensional echo matrix of the liver, and generate the corresponding fatty liver treatment efficacy assessment result.

[0055] By employing the above technical solution, the present invention provides a method and system for image evaluation of the efficacy of traditional Chinese medicine intervention in fatty liver, which has at least the following beneficial effects:

[0056] 1. This invention constructs a complete technical chain that integrates abdominal audio signals, ultrasound radio frequency data, probe spatial pose data, and TCM syndrome image data through multimodal fusion processing. This chain covers sound-shadow adaptive trigger acquisition, respiratory displacement closed-loop compensation, three-dimensional echo physical reconstruction, and quantitative evaluation of efficacy in both TCM and Western medicine. It achieves fully automated processing from data acquisition triggering and three-dimensional spatial alignment to quantitative output of efficacy, enabling objective, multidimensional, and longitudinal evaluation of the efficacy of TCM intervention for fatty liver.

[0057] 2. This invention extracts the bowel sound envelope signal by performing time-frequency conversion and blind source separation on the abdominal audio stream and performs first-order differentiation on it. It predicts the high-risk period of violent intestinal gas movement based on the time interval where the energy change rate exceeds a preset threshold, and then selectively extracts the dynamic ultrasound frame set within this time window. This achieves active identification and targeted extraction of high-risk periods of sound and shadow interference, effectively avoiding the technical problems of unpredictable sound and shadow interference and low effective data utilization in the prior art, and providing a reliable data foundation for subsequent operations.

[0058] 3. This invention identifies the centroid coordinates of vascular bifurcations at different time phases from candidate echo matrices. Using the end-expiratory phase as the reference phase, it calculates the spatial displacement vectors of each non-reference phase and weights them to form a spatial topological deviation value. This deviation is then decomposed into spatial translation and rotation matrices to generate a respiratory displacement compensation vector, which is then inversely superimposed onto the probe's spatial pose data for iterative correction until the deviation converges to within a preset threshold. This mechanism uses the inherent anatomical structure within the liver as a self-reference benchmark, requiring no additional sensors, and achieves coordinated and precise compensation for linear offsets and angular deflections induced by respiration.

[0059] 4. This invention extracts the radio frequency envelope signal sequence from the unobstructed three-dimensional echo matrix of the liver and calculates its energy attenuation gradient along the scanning depth direction to obtain the target fat attenuation coefficient. On the other hand, it extracts color and texture features from facial and tongue images of the same treatment period and quantifies them into TCM syndrome integrals through a feature mapping model. Finally, it calculates the correlation index of the temporal slope of the two and maps it with a preset correlation threshold interval to output the efficacy level label. This objectively couples the longitudinal change trend of acoustic physical quantities with the improvement trajectory of TCM macroscopic manifestations on the same time axis, providing multimodal quantitative basis for evidence-based evaluation of the efficacy of TCM intervention. Attached Figure Description

[0060] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0061] Figure 1 A flowchart of the image evaluation method for the efficacy of traditional Chinese medicine intervention in fatty liver provided by the present invention;

[0062] Figure 2 This is a schematic diagram of the modules of the image evaluation system for the efficacy of traditional Chinese medicine intervention in fatty liver provided by the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Exemplary method:

[0065] In this embodiment, the patient received a modified Xiao Chai Hu Tang (Minor Bupleurum Decoction), which consisted of: Bupleurum 12g, Scutellaria baicalensis 9g, Pinellia ternata 9g, Ginseng 6g, Salvia miltiorrhiza 18g, Rheum palmatum 6g, Artemisia capillaris 18g, Curcuma longa 15g, Glycyrrhiza uralensis 6g, Ziziphus jujuba 12g, and Zingiber officinale 9g. This formula is based on Xiao Chai Hu Tang, with the addition of Salvia miltiorrhiza, Rheum palmatum, Artemisia capillaris, and Curcuma longa. It is designed for non-alcoholic fatty liver disease, which is often attributed to liver stagnation and spleen deficiency, with turbidity and blood stasis obstructing the collaterals. The treatment principle is to soothe the liver, strengthen the spleen, and resolve turbidity and blood stasis. Bupleurum, bitter and neutral in nature, soothes the liver and relieves stagnation, clearing away stagnant qi in the heart, abdomen, intestines, and stomach, and promoting metabolism; Scutellaria and Rhubarb, bitter and cold, clear heat; Pinellia, pungent and bitter, ascends when combined with Bupleurum, and descends when combined with Scutellaria and Rhubarb, and together with Artemisia capillaris, drains dampness and clears turbidity, achieving the effect of clearing the mind and clearing the body; Salvia miltiorrhiza invigorates blood and removes blood stasis, while Curcuma longa promotes qi circulation, relieves stagnation, cools the blood, and breaks up blood stasis. Together, they can relieve stagnation in the liver meridian; Ginger, when combined with Pinellia, disperses and clears qi stagnation, assists Bupleurum in relieving stagnation, and can also work with Artemisia capillaris to drain dampness and clear turbidity, while also harmonizing the stomach, suppressing nausea and vomiting; Ginseng, Jujube, and Licorice tonify the middle qi and harmonize the yin and yang, strengthen the spleen and nourish the original qi, so that the body's resistance can overcome the pathogenic factors, and prevent the pathogenic factors of the Shaoyang meridian from spreading to the Taiyin meridian. Modern pharmacological studies have shown that Xiao Chai Hu Tang can protect hepatocytes, prevent liver damage, inhibit liver fibrosis, and regulate immunity through multiple pathways. Herbs such as Danshen, Da Huang, Yin Chen, and Yu Jin all have good hepatoprotective, lipid-lowering, and anti-liver fibrosis effects. The combined effects of these herbs work synergistically to soothe the liver, strengthen the spleen, and resolve turbidity and blood stasis. The evaluation method described in this invention is used to quantitatively assess the objective efficacy of Xiao Chai Hu Tang with added ingredients during each treatment cycle after patients take the medication according to the above formula, in order to verify the actual degree of improvement of non-alcoholic fatty liver disease.

[0066] The existing technology cannot achieve precise alignment of the dynamic spatial pose between the detection terminal and internal organs in complex clinical application environments. As a result, the extracted physical features are often mixed with a large number of motion artifacts and noise errors. This makes it impossible for the final generated attenuation parameters to accurately reproduce the data evolution of the real anatomical site, which in turn affects the accuracy of the evaluation of the efficacy of traditional Chinese medicine intervention for fatty liver.

[0067] This embodiment proposes an image evaluation method for the efficacy of traditional Chinese medicine intervention in fatty liver. A typical application scenario is a fatty liver patient receiving traditional Chinese medicine treatment: the patient is equipped with an ultrasound scanner, a probe pose sensor, and an abdominal audio acquisition device. Each device synchronously acquires data and carries a timestamp from a unified clock source. The system executes the evaluation process according to the following steps. In this specification, the time phase uniformly refers to a specific acquisition time within the target ultrasound dynamic frame set. This time serves as an index to uniquely correspond to a frame of ultrasound data and the corresponding probe spatial pose record; the candidate echo matrix refers to the intermediate result after preliminary three-dimensional spatial projection but before respiratory displacement compensation verification; the target echo matrix refers to the final effective three-dimensional dataset after spatial topological deviation verification meets the accuracy requirements. For example... Figure 1 As shown, the method includes the following steps:

[0068] Acquire timestamped ultrasound radio frequency data streams, probe spatial pose data, and abdominal audio streams;

[0069] The bowel sound envelope signal in the preset frequency band is extracted from the abdominal audio stream, and the a priori triggering time window characterizing intestinal gas movement is determined based on the bowel sound envelope signal.

[0070] Extract a set of target ultrasonic dynamic frames from the ultrasonic radio frequency data stream that are aligned with the a priori trigger time window;

[0071] Dynamic acoustic shadow regions and corresponding unmasked echo regions are divided into the target ultrasound dynamic frame set.

[0072] Based on the probe spatial pose data, a preliminary three-dimensional spatial projection is performed on the unmasked echo region at different time phases to obtain the candidate echo matrix;

[0073] Extract the intrahepatic vascular topological features from the candidate echo matrix, and calculate the spatial topological deviation values ​​between the intrahepatic vascular topological features at different time phases;

[0074] Determine whether the spatial topology deviation value is greater than a preset deviation threshold; if so, generate a respiratory displacement compensation vector based on the intrahepatic vascular topology features and update the probe spatial pose data, and re-execute the preliminary three-dimensional spatial projection based on the updated probe spatial pose data until the spatial topology deviation value is less than or equal to the preset deviation threshold.

[0075] Candidate echo matrices that meet the preset deviation threshold constraint are used as target echo matrices. The target echo matrices are fused into the dynamic sound and shadow region to reconstruct an unobstructed three-dimensional echo matrix of the liver.

[0076] The target fat attenuation coefficient is calculated based on the unobstructed three-dimensional echo matrix of the liver, and the corresponding fatty liver treatment efficacy assessment results are generated.

[0077] The acquisition of timestamped ultrasound radiofrequency data streams, probe spatial pose data, and abdominal audio streams forms the data foundation preparation stage for the entire evaluation method. Its core function is to achieve synchronous acquisition and timestamping of multimodal signals, providing a prerequisite for subsequent time-domain correlation processing between different data streams. The ultrasound radiofrequency data stream is acquired by an ultrasound diagnostic instrument with a convex array probe. The probe's operating frequency is typically set in the range of 2.5MHz to 5MHz to ensure effective penetration of deep liver tissue. The probe spatial pose data is recorded in real-time by an electromagnetic positioning sensor or optical tracking sensor mounted on the probe handle, including the probe's position coordinates (x, y, z) and attitude angles (θ_pitch, θ_roll, θ_yaw) in a three-dimensional Cartesian coordinate system, with a sampling frequency typically not lower than 20Hz. The abdominal audio stream is acquired by a high-sensitivity capacitive microphone attached to the patient's right hypochondrium, with a sampling rate set to 44100Hz to meet the Nyquist sampling conditions for subsequent bowel sound frequency analysis. Each data stream carries a timestamp generated from the same clock source, with a timestamp accuracy of no less than 1ms, to ensure accurate time alignment of subsequent cross-modal data. Taking a typical acquisition scenario as an example: with the patient in a supine position, during a 30-second scan, the ultrasound system can acquire approximately 900 frames of raw radiofrequency data (30fps), while the pose sensor records approximately 600 six-degree-of-freedom pose samples simultaneously, and the audio system acquires approximately 1,323,000 audio sample points. The timestamps of all three data streams are locked to the same reference clock, with a deviation of less than 1ms.

[0078] The preset frequency band refers to the concentrated distribution range of acoustic frequencies that characterize the alternating movement of gas and fluid within the intestinal lumen. Specifically, it can be set between 100Hz and 500Hz based on clinical physiological acoustic statistics or the spectral analysis results of abdominal auscultation audio samples. This is used to trigger the blind source separation algorithm to accurately filter out non-target physiological background noise such as heart sounds and breath sounds. The basis for setting this specific frequency band is that the hydrodynamic acoustic resonance energy generated by the gas-fluid mixture in the intestine under the peristaltic squeezing action of smooth muscle is highly concentrated in this frequency window, and this range can form a significant frequency domain misalignment with low-frequency heart sounds and broadband breath sounds. This ensures that the extracted intestinal gas movement prior trigger signal has extremely high physical specificity and temporal positioning accuracy while eliminating non-stationary background interference to the maximum extent.

[0079] The preset deviation threshold represents the maximum allowable spatial registration error limit of the intrahepatic vascular anatomy in multi-temporal three-dimensional reconstruction, and serves as the convergence criterion for the probe spatial pose inverse closed-loop calibration algorithm. The preset deviation threshold is comprehensively calibrated by combining the hardware physical resolution limit of the clinical ultrasound probe and the clinical error tolerance range of the target fat attenuation coefficient calculation accuracy.

[0080] The step of extracting the bowel sound envelope signal in a preset frequency band from the abdominal audio stream and determining the prior triggering time window characterizing intestinal gas movement based on the bowel sound envelope signal includes the following steps:

[0081] The abdominal audio stream is subjected to time-frequency conversion processing to obtain audio spectrum data;

[0082] Based on a preset blind source separation algorithm, physiological background noise components in the audio spectrum data are separated and filtered out, and bowel sound audio spectrum data in the preset frequency band are extracted.

[0083] Envelope extraction processing is performed on the bowel sound spectrum data to obtain the bowel sound envelope signal;

[0084] The first-order derivative of the bowel sound envelope signal is performed to obtain the energy change rate sequence;

[0085] Extract the start and end sampling time points from the energy change rate sequence whose values ​​are greater than a preset change rate threshold.

[0086] The time interval between the start sampling time point and the end sampling time point is used as the prior trigger time window; wherein, the prior trigger time window is used to define the high-risk time period in which the intense movement of intestinal gas causes acoustic shadowing of ultrasound images.

[0087] The time-frequency conversion process refers to mapping the one-dimensional time-domain audio stream to a two-dimensional spectral feature that characterizes the dynamic change of signal frequency over time. Specifically, it can be implemented using short-time Fourier transform or continuous wavelet transform, which is used to convert the aliased time-domain audio into audio spectral data that is easy to decouple and separate from background noise in the subsequent process.

[0088] Among them, the preset blind source separation algorithm can decouple the superimposed composite physiological audio into multiple statistically independent components in the data processing process under the premise of unknown abdominal multi-source mixing mechanism. Specifically, independent component analysis algorithm, non-negative matrix factorization algorithm or sparse component analysis algorithm can be used to strip and filter out the low-frequency periodic heart sounds and respiratory sounds and other physiological background noise components mixed in the audio spectrum data, so as to accurately extract the bowel sound audio spectrum data that retains only non-stationary burst energy and is within the preset frequency band.

[0089] The envelope extraction process refers to extracting the outer boundary contour that characterizes the macroscopic fluctuation trend of signal energy from the high-frequency oscillating bowel sound spectrum data. Specifically, it can be implemented using Hilbert transform, square-law detection with low-pass filtering, or sliding window absolute value algorithm to eliminate redundant high-frequency fluctuations in the local signal and obtain a smooth bowel sound envelope signal that is convenient for subsequent calculation of abrupt change points.

[0090] Extracting the target ultrasonic dynamic frame set aligned with the prior triggering time window from the ultrasonic radio frequency data stream includes the following steps:

[0091] The ultrasonic radio frequency data stream is analyzed to extract the acquisition timestamps corresponding to each consecutive ultrasonic data frame;

[0092] Map the start and end sampling times of the prior trigger time window to the time-domain coordinate axis corresponding to the acquisition timestamp, and determine the target interception start node and target interception end node aligned with the prior trigger time window;

[0093] Extract all ultrasound data frames located between the target interception start node and the target interception end node from the ultrasound radio frequency data stream in a time sequence, and combine them to generate the target ultrasound dynamic frame set.

[0094] Specifically, when determining the target capture start node and target capture end node, adaptive windowing processing can be performed by introducing hardware transmission delay compensation parameters and preset time margins (e.g., extending forward and backward by 0.5 seconds to 1.0 seconds) to eliminate the data stream timestamp synchronization error between the heterogeneous physical devices of the audio sensor and the ultrasound probe, ensuring that the generated target ultrasound dynamic frame set can completely cover the entire occurrence, development and dissipation cycle of intestinal gas moving in the ultrasound image and generating dynamic sound shadows.

[0095] The step of dividing the dynamic acoustic shadow region and the corresponding unmasked echo region in the target ultrasound dynamic frame set includes the following steps:

[0096] Calculate the gradient data of the radio frequency signal intensity of each pixel in each frame of the target ultrasound dynamic frame set as a function of scanning depth;

[0097] Extract the pixels whose change gradient data is greater than the preset signal attenuation gradient from each frame of the target ultrasound dynamic frame set, and mark the spatial region formed by the pixels as the dynamic sound and shadow region.

[0098] Extract the remaining pixels from the target ultrasound dynamic frame set, excluding the dynamic sound shadow region, and mark the spatial region formed by the remaining pixels as the unmasked echo region under the corresponding time phase.

[0099] Specifically, this step involves performing semantic segmentation on the target ultrasound dynamic frame set to distinguish between the dynamic acoustic shadow region generated by gas and the unmasked normal echo region, providing effective echo data range definition for subsequent three-dimensional spatial projection. Intestinal gas exhibits strong reflection and attenuation characteristics to ultrasound waves, resulting in an abnormally steep attenuation gradient of the radio frequency signal intensity behind the gas along the depth direction. By calculating the gradient of radio frequency signal intensity with spatial depth pixel by pixel and comparing it with a preset signal attenuation gradient threshold, the acoustic shadow boundary can be automatically identified: pixels with a gradient exceeding the threshold are assigned to the first pixel set (dynamic acoustic shadow region), and the remaining pixels are assigned to the second pixel set (unmasked echo region). For example, assuming the preset signal attenuation gradient threshold is −20 dB / cm, if the signal intensity of a pixel column drops sharply from −30 dBm to −80 dBm (spanning 1 cm) in the depth direction, its gradient is −50 dB / cm, exceeding the threshold in absolute value; therefore, this pixel is marked as a dynamic acoustic shadow region.

[0100] The preset signal attenuation gradient refers to the gradient threshold used to distinguish the change of radio frequency signal intensity with spatial depth between dynamic acoustic shadow regions and unmasked echo regions. Specifically, it can be set according to the acoustic attenuation characteristics of different tissue types or the statistical analysis results of historical ultrasound data, and is used to trigger the automatic marking and segmentation conditions of dynamic acoustic shadow regions.

[0101] The step of performing preliminary three-dimensional spatial projection of the unmasked echo region at different time phases based on the probe spatial pose data to obtain the candidate echo matrix includes the following steps:

[0102] Extract the two-dimensional image coordinates of each pixel in the unmasked echo region at the corresponding time phase;

[0103] Based on the probe spatial pose data, a coordinate transformation matrix is ​​constructed to map the two-dimensional image coordinate system corresponding to the time phase to the global three-dimensional spatial coordinate system;

[0104] The coordinate transformation matrix is ​​used to perform spatial transformation on the coordinates of the two-dimensional image to obtain the initial three-dimensional spatial coordinates of each pixel in the global three-dimensional spatial coordinate system.

[0105] The echo intensity values ​​corresponding to each pixel are mapped to a preset three-dimensional voxel grid according to the initial three-dimensional spatial coordinates to generate the candidate echo matrix.

[0106] Spatial transformation processing refers to the process of mapping two-dimensional image coordinates to a three-dimensional physical coordinate system; specifically, it can be implemented using rigid body transformation or affine transformation algorithms to restore two-dimensional ultrasound slices at each time phase to their true three-dimensional anatomical positions.

[0107] The preset three-dimensional voxel mesh refers to a spatially discretized data structure composed of several tiny volume units; specifically, it can be implemented by initializing a three-dimensional array, which is used to structure discrete ultrasonic spatial coordinate points into a continuous three-dimensional echo matrix.

[0108] The calculation of the spatial topological deviation value between the topological features of intrahepatic vessels at different time phases includes the following steps:

[0109] The centroid coordinates of the blood vessel bifurcation are identified from the three-dimensional spatial data of different time phases in the candidate echo matrix.

[0110] Based on a preset feature matching algorithm, the centroid coordinates of the bifurcation of blood vessels that match at different time phases are determined as the centroid coordinates of the bifurcation of the same blood vessels, and the centroid coordinates of the bifurcation of the same blood vessels are used as the topological features of the intrahepatic blood vessels.

[0111] The end-expiratory time phase among the different time phases is determined as the reference phase, and the spatial displacement vector data of the centroid of each homologous vessel bifurcation under each non-reference time phase is calculated relative to the corresponding centroid under the reference phase.

[0112] The spatial displacement vector data is weighted and synthesized, and the vector magnitude of the weighted synthesis result is used as the spatial topology deviation value; wherein, the spatial topology deviation value is used to characterize the nonlinear spatial deformation physical error of the vascular anatomy structure inside the liver due to the patient's respiratory motion between different time phases.

[0113] The end-expiratory time phase refers to the physiological pause between the completion of a natural exhalation and the start of the next inhalation. At this specific time phase, the patient's lungs are at functional residual capacity, and the diaphragm is in its highest anatomical position of natural relaxation. At this time, the diaphragmatic compression stress on the liver is minimal, and its three-dimensional spatial coordinates are in a relatively resting state with a velocity approaching zero. Selecting this specific physiological node as the reference phase for three-dimensional spatial alignment aims to establish an absolute anatomical reference system with the highest spatial stability. This ensures that the subsequently calculated spatial topological deviation value accurately and purely reflects the physical errors caused by respiratory deformation, avoiding overcompensation or ineffective compensation caused by the drift of the reference frame itself.

[0114] Preferably, the temporal respiratory motion curve is obtained by tracking the axial displacement of liver anatomical markers (such as the diaphragm) in the ultrasound dynamic frame. Then, the first derivative of the curve is performed to calculate the instantaneous motion velocity. When it is determined that the absolute value of the instantaneous velocity is less than the preset zero threshold and the displacement is in the local minimum range, it is considered that the liver is in a relatively resting state with minimal compression from the diaphragm. Thus, the extreme moment is automatically locked and output as the end-expiratory time phase.

[0115] Specifically, this step uses the intrahepatic vascular topology as an intrinsic reference benchmark to quantify the registration error between multi-temporal projection data, providing a quantitative basis for subsequent respiratory displacement compensation. Intrahepatic vessels are highly stable in three-dimensional structure, and their bifurcation point coordinates do not change with time in real three-dimensional space. Therefore, if the multi-temporal projections are correctly registered, the set of centroid coordinates of the vascular bifurcations identified in each temporal phase should highly coincide in the world coordinate system. Conversely, if there is a displacement error caused by respiratory motion, the projected position of the same bifurcation point in different temporal phases will show spatial shift.

[0116] The preset feature matching algorithm is used to establish point-to-point mapping relationships in dynamically deformed image sequences. Specifically, a feature comparison algorithm based on three-dimensional fast point feature histogram or local geometric descriptor can be used to eliminate the interference of absolute coordinate displacement caused by breathing. By comparing local echo gradient features, the same physical node of blood vessel bifurcation at different time phases can be accurately located.

[0117] The step of generating a respiratory displacement compensation vector based on the intrahepatic vascular topological features and updating the probe spatial pose data includes the following steps:

[0118] Based on the direction vector and vector magnitude of the spatial displacement vector data under each non-reference time phase, spatial translation matrix and spatial rotation matrix are calculated to quantify the linear offset and angular deflection of the liver anatomical structure under each non-reference time phase.

[0119] The respiratory displacement compensation vector for each non-reference time phase is generated based on the spatial translation matrix and the spatial rotation matrix; the formula for generating the respiratory displacement compensation vector is as follows:

[0120] ,

[0121] in, Indicates the first A respiratory displacement compensation vector under a non-reference time phase;

[0122] The probe spatial pose data is superimposed with the respiratory displacement compensation vector to obtain updated probe spatial pose data; the data superposition formula is as follows:

[0123] ,

[0124] in, This represents the updated probe spatial pose data. This represents the probe's spatial pose data before the update.

[0125] Furthermore, the calculation of the spatial translation and spatial rotation matrices used to quantify the linear offset and angular deflection of the liver anatomical structure at each non-reference time phase includes the following steps:

[0126] Calculate the spatial mean of the centroid coordinates of all homologous vessel bifurcations under the reference time phase and each non-reference time phase to obtain the reference centroid mean and the mean of each non-reference centroid;

[0127] Based on the reference centroid mean and the mean of each non-reference centroid, centroid removal processing is performed on the centroid coordinates of all homologous vessel bifurcations at each non-reference time phase to obtain the centroid-removed coordinate set at each non-reference time phase, and the covariance matrix at each non-reference time phase is constructed; the formula for constructing the covariance matrix is ​​as follows:

[0128] ,

[0129] in, Represents the covariance matrix. Indicates the first phase under the reference time. The column vector of coordinates of the centroid. Indicates the first non-reference time phase The transpose of the column vector of the centroid-free coordinates. This represents the total number of centroid pairs of bifurcation points of homologous vessels;

[0130] Singular value decomposition is performed on the covariance matrix to obtain the left and right singular vector matrices for each non-reference time phase.

[0131] Based on the left singular vector matrix and the right singular vector matrix, calculate the spatial rotation matrix for each non-reference time phase; the calculation formula is as follows:

[0132] ,

[0133] in, Indicates the first Spatial rotation matrices of non-reference time phases relative to reference time phases Describes a right singular vector matrix. Represents a left singular vector matrix;

[0134] Based on the spatial rotation matrix, the mean value of the reference centroid, and the mean values ​​of each non-reference centroid, calculate the spatial translation matrix for each non-reference time phase; the calculation formula is as follows:

[0135] ,

[0136] Among them, the Spatial translation matrices of non-reference time phases relative to reference time phases and They represent the first The centroid mean of each non-reference time phase and the reference time phase.

[0137] The reconstruction to obtain the unobstructed three-dimensional echo matrix of the liver includes the following steps:

[0138] Identify the overlapping spatial regions of the target echo matrix at different time phases;

[0139] Calculate the local signal-to-noise ratio data of the overlapping spatial region and generate a dynamic fusion weight distribution; the formula for generating the dynamic fusion weight distribution is as follows:

[0140] ,

[0141] in, Indicates the first Signal-to-noise ratio of each phase, Indicates the first Signal-to-noise ratio of each phase, Indicates the first Dynamic fusion weights for each phase, This indicates the total number of time phases involved in the fusion;

[0142] Based on the dynamic fusion weight distribution, the unmasked echo data in the target echo matrix is ​​weighted and calculated, and the weighted echo data is filled into the three-dimensional spatial coordinates corresponding to the dynamic sound shadow region to reconstruct and generate a physically continuous unmasked three-dimensional echo matrix of the liver.

[0143] The process of generating the corresponding fatty liver treatment evaluation results includes the following steps:

[0144] Extract the radio frequency envelope signal sequence along the scanning depth direction from the unobstructed liver three-dimensional echo matrix;

[0145] Calculate the energy attenuation gradient of the radio frequency envelope signal sequence to obtain the target fat attenuation coefficient;

[0146] Acquire TCM syndrome image data corresponding to the consultation cycle of the ultrasound radio frequency data stream; wherein, the TCM syndrome image data includes at least facial image data and tongue image data;

[0147] Color and texture features are extracted from the TCM syndrome image data, and the extraction results are input into a preset feature mapping model for processing to obtain the TCM syndrome quantitative integral.

[0148] Calculate the temporal decline slope of the target fat attenuation coefficient and the temporal improvement slope of the quantitative score of the TCM syndrome.

[0149] Calculate the correlation index between the time series decline slope and the time series improvement slope; the calculation formula is as follows:

[0150] ,

[0151] in, Represents the correlation index. Indicates the total number of timing windows. and They represent the first Within a time window, the temporal decrease slope of the target fat attenuation coefficient and its corresponding arithmetic mean. and They represent the first Within a time window, the temporal improvement slope of the quantitative integral of TCM syndrome and the corresponding arithmetic mean;

[0152] The correlation index is numerically compared with the preset correlation threshold range in the preset efficacy evaluation criteria, and the efficacy level label corresponding to the preset correlation threshold range that matches the correlation index is output to obtain the fatty liver efficacy evaluation result; wherein, the preset efficacy evaluation criteria include at least the preset correlation threshold ranges corresponding to different efficacy level labels.

[0153] Specifically, the extraction of color and texture features involves introducing multidimensional independent color space transformation and gray-level co-occurrence matrix operators to quantify the abnormal spatial distribution of local pathological color space shifts and surface moss morphology caused by imbalances in the Qi and blood of the internal organs in the TCM syndrome image data.

[0154] The preset correlation threshold range is typically determined based on large-sample retrospective clinical research data. A typical grading scheme is as follows: [0.75, 1.00] corresponds to a "significantly effective" level (highly synergistic improvement in both traditional Chinese and Western medicine); [0.50, 0.75] corresponds to a "effective" level (moderate improvement in both traditional Chinese and Western medicine); [0.25, 0.50] corresponds to a "mildly effective" level (slight improvement in both traditional Chinese and Western medicine); and (-∞, 0.25) corresponds to a "not effective" level (insufficient synergy in improvement between the two dimensions, or inconsistent improvement directions). Taking a correlation index of 0.94 as an example, 0.94 ∈ [0.75, 1.00], the system outputs the efficacy grade label "significantly effective," completing the full fatty liver efficacy evaluation process of the method of this invention. In addition to grade labels, the efficacy evaluation results can also include the target fat attenuation coefficient change curve for each treatment cycle and the quantitative score change curve of TCM syndrome, providing clinicians with multidimensional and intuitive evidence of the synergistic efficacy of TCM and Western medicine. At the same time, it can serve as a quantitative reference for the dynamic adjustment of Xiao Chai Hu Tang modified prescriptions (such as adjusting the dosage of Artemisia capillaris and Salvia miltiorrhiza).

[0155] Furthermore, the feature mapping model can employ a multimodal cross-attention visual converter; the model construction process is as follows:

[0156] By using a TCM clinical diagnosis and treatment information system and an objective data acquisition device for the four diagnostic methods, several sets of sample feature data that characterize the mapping relationship between the facial and tongue appearance features of patients and the evolution of TCM syndromes of fatty liver are collected to obtain a sample dataset. The sample data in the sample dataset includes at least historical facial image data, historical tongue image data, and corresponding expert-calibrated quantitative score data of TCM syndromes.

[0157] Construct an initial feature mapping model and set the training data ratio, such as 8:2 or 7.5:2.5, which can be adjusted reasonably according to the actual situation.

[0158] The sample dataset is divided according to the training data ratio to obtain a training dataset and a test dataset.

[0159] Set a training error threshold, such as 5%-10%, which can be adjusted reasonably according to the actual situation; input the training data in the training dataset into the initial feature mapping model for training, and continuously adjust the parameters of the initial feature mapping model according to the training results until the training error is less than the training error threshold or the number of training times is greater than the maximum number of training times, and obtain a well-trained feature mapping model.

[0160] Set the test precision, such as 90%-95%, which can be adjusted reasonably according to the actual situation; input the test data in the test dataset into the trained feature mapping model for testing, calculate the accuracy of the test results, and if the accuracy of the test results is greater than the test precision, the final feature mapping model is obtained; otherwise, retrain until the accuracy of the test results is greater than the test precision.

[0161] The structure of the initial feature mapping model can be seen in Table 1 below:

[0162] Model Name Model type Model Structure Initial feature mapping model Multimodal cross-attention visual converter Multimodal Independent Encoding Module: Input Layer: Receives facial and tongue image data; Deep Encoding Layer: Contains 4-6 layers of independent parallel visual converter encoding blocks, each layer configured with patch embedding and multi-head self-attention mechanism, employing the GELU activation function to enhance the nonlinear expressive power of high-dimensional features, outputting high-dimensional facial and tongue semantic vector sequences. Cross-Attention Collaboration Module: Cascades 2-4 bidirectional cross-attention fusion blocks, with internal sub-networks using a scaled dot product attention mechanism to calculate the query and key-value mapping weighting coefficients of cross-modal features, configuring LayerNorm normalization layers to ensure the stability of deep gradient propagation, constructing a deep nonlinear collaborative mapping relationship between the facial complexion appearance space and the tongue micro-feature space. Syndrome Integral Regression Module: Performs global average pooling based on the spliced ​​and aggregated global multimodal collaborative features, combined with a fully connected multilayer perceptron network layer, and outputs continuous TCM syndrome quantitative integral data via forward mapping link regression. Parameter settings: Loss function: composed of a weighted composite of smooth L1 loss and TCM prior regularization penalty term to enhance robustness to abnormal clinical outlier samples; Optimizer: AdamW (initial learning rate 1e-4) combined with cosine annealing decay strategy; Initialization: truncated normal distribution initialization is adopted.

[0163] Table 1

[0164] It should be understood that the evaluation results of this application only represent the quantitative calculation outputs such as the target fat attenuation coefficient, the quantitative score of TCM syndrome, and the correlation index between the two obtained after processing the ultrasound radiofrequency data, probe spatial pose data, abdominal audio stream, and TCM syndrome image data through the above steps, as well as the efficacy level label output by comparing it with the preset efficacy evaluation standard. It does not constitute a diagnostic conclusion for any disease, nor can it replace the professional medical diagnosis and treatment decision made by a licensed physician based on the patient's overall clinical information.

[0165] Exemplary system:

[0166] Please see Figure 2 A traditional Chinese medicine intervention for fatty liver efficacy image evaluation system includes a data acquisition module, a priori trigger time window determination module, a region division module, a spatial topology deviation measurement module, a probe spatial pose update module, and a fatty liver efficacy evaluation module.

[0167] The data acquisition module is used to acquire time-stamped ultrasound radio frequency data streams, probe spatial pose data, and abdominal audio streams.

[0168] The prior trigger time window determination module is used to extract the bowel sound envelope signal in a preset frequency band from the abdominal audio stream, and determine the prior trigger time window characterizing the movement of intestinal gas based on the bowel sound envelope signal.

[0169] The region segmentation module is used to extract a set of target ultrasound dynamic frames aligned with the prior trigger time window from the ultrasound radio frequency data stream; and to segment the target ultrasound dynamic frame set into a dynamic acoustic shadow region and a corresponding unmasked echo region.

[0170] The spatial topology deviation measurement module is used to perform preliminary three-dimensional spatial projection of the unmasked echo region at different time phases based on the probe spatial pose data to obtain a candidate echo matrix; extract the intrahepatic vascular topological features in the candidate echo matrix, and calculate the spatial topology deviation value between the intrahepatic vascular topological features at different time phases.

[0171] The probe spatial pose update module is used to determine whether the spatial topology deviation value is greater than a preset deviation threshold; if so, a respiratory displacement compensation vector is generated based on the intrahepatic vascular topology features and the probe spatial pose data is updated. The preliminary three-dimensional spatial projection is re-executed based on the updated probe spatial pose data until the spatial topology deviation value is less than or equal to the preset deviation threshold.

[0172] The fatty liver treatment efficacy assessment module is used to take the candidate echo matrix that meets the preset deviation threshold constraint as the target echo matrix, fuse the target echo matrix into the dynamic sound shadow region, and reconstruct an unobstructed three-dimensional echo matrix of the liver; calculate the target fat attenuation coefficient based on the unobstructed three-dimensional echo matrix of the liver, and generate the corresponding fatty liver treatment efficacy assessment result.

[0173] Exemplary computer-readable media:

[0174] Embodiments of this application may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps described in the "Exemplary Methods" section above according to the various embodiments of this application.

[0175] Computer-readable storage media may take the form of any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0176] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0177] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0178] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0179] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0180] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for image evaluation of the efficacy of traditional Chinese medicine intervention in fatty liver, characterized in that, Includes the following steps: Acquire timestamped ultrasound radio frequency data streams, probe spatial pose data, and abdominal audio streams; The bowel sound envelope signal in the preset frequency band is extracted from the abdominal audio stream, and the a priori triggering time window characterizing intestinal gas movement is determined based on the bowel sound envelope signal. Extract a set of target ultrasonic dynamic frames from the ultrasonic radio frequency data stream that are aligned with the a priori trigger time window; Dynamic acoustic shadow regions and corresponding unmasked echo regions are divided into the target ultrasound dynamic frame set. Based on the probe spatial pose data, a preliminary three-dimensional spatial projection is performed on the unmasked echo region at different time phases to obtain the candidate echo matrix; Extract the intrahepatic vascular topological features from the candidate echo matrix, and calculate the spatial topological deviation values ​​between the intrahepatic vascular topological features at different time phases; Determine whether the spatial topology deviation value is greater than a preset deviation threshold; if so, generate a respiratory displacement compensation vector based on the intrahepatic vascular topology features and update the probe spatial pose data, and re-execute the preliminary three-dimensional spatial projection based on the updated probe spatial pose data until the spatial topology deviation value is less than or equal to the preset deviation threshold. Candidate echo matrices that meet the preset deviation threshold constraint are used as target echo matrices. The target echo matrices are fused into the dynamic sound and shadow region to reconstruct an unobstructed three-dimensional echo matrix of the liver. The target fat attenuation coefficient is calculated based on the unobstructed three-dimensional echo matrix of the liver, and the corresponding fatty liver treatment efficacy assessment results are generated.

2. The image evaluation method for the efficacy of traditional Chinese medicine intervention in fatty liver according to claim 1, characterized in that, Extracting the bowel sound envelope signal in a preset frequency band from the abdominal audio stream includes the following steps: The abdominal audio stream is subjected to time-frequency conversion processing to obtain audio spectrum data; Based on a preset blind source separation algorithm, physiological background noise components in the audio spectrum data are separated and filtered out, and bowel sound audio spectrum data in the preset frequency band are extracted. Envelope extraction processing is performed on the bowel sound frequency spectrum data to obtain the bowel sound envelope signal.

3. The image evaluation method for the efficacy of traditional Chinese medicine intervention in fatty liver according to claim 1, characterized in that, The determination of the prior triggering time window characterizing intestinal gas migration includes the following steps: The first-order derivative of the bowel sound envelope signal is performed to obtain the energy change rate sequence; Extract the start and end sampling time points from the energy change rate sequence whose values ​​are greater than a preset change rate threshold. The time interval between the start sampling time point and the end sampling time point is used as the prior trigger time window; wherein, the prior trigger time window is used to define the high-risk time period in which the intense movement of intestinal gas causes acoustic shadowing of ultrasound images.

4. The image evaluation method for the efficacy of traditional Chinese medicine intervention in fatty liver according to claim 1, characterized in that, The step of dividing the dynamic acoustic shadow region and the corresponding unmasked echo region in the target ultrasound dynamic frame set includes the following steps: Calculate the gradient data of the radio frequency signal intensity of each pixel in each frame of the target ultrasound dynamic frame set as a function of scanning depth; Extract the pixels whose change gradient data is greater than the preset signal attenuation gradient from each frame of the target ultrasound dynamic frame set, and mark the spatial region formed by the pixels as the dynamic sound and shadow region. Extract the remaining pixels from the target ultrasound dynamic frame set, excluding the dynamic sound shadow region, and mark the spatial region formed by the remaining pixels as the unmasked echo region under the corresponding time phase.

5. The method for image evaluation of the efficacy of traditional Chinese medicine intervention in fatty liver according to claim 1, characterized in that, The calculation of the spatial topological deviation value between the topological features of intrahepatic vessels at different time phases includes the following steps: The centroid coordinates of the blood vessel bifurcation are identified from the three-dimensional spatial data of different time phases in the candidate echo matrix. Based on a preset feature matching algorithm, the centroid coordinates of the bifurcation of blood vessels that match at different time phases are determined as the centroid coordinates of the bifurcation of the same blood vessels, and the centroid coordinates of the bifurcation of the same blood vessels are used as the topological features of the intrahepatic blood vessels. The end-expiratory time phase among the different time phases is determined as the reference phase, and the spatial displacement vector data of the centroid of each homologous vessel bifurcation under each non-reference time phase is calculated relative to the corresponding centroid under the reference phase. The spatial displacement vector data is weighted and synthesized, and the vector magnitude of the weighted synthesis result is used as the spatial topology deviation value; wherein, the spatial topology deviation value is used to characterize the nonlinear spatial deformation physical error of the vascular anatomy structure inside the liver due to the patient's respiratory motion between different time phases.

6. The image evaluation method for the efficacy of traditional Chinese medicine intervention in fatty liver according to claim 1, characterized in that, The step of generating a respiratory displacement compensation vector based on the intrahepatic vascular topological features and updating the probe spatial pose data includes the following steps: Based on the direction vector and vector magnitude of the spatial displacement vector data under each non-reference time phase, spatial translation matrix and spatial rotation matrix are calculated to quantify the linear offset and angular deflection of the liver anatomical structure under each non-reference time phase. Based on the spatial translation matrix and the spatial rotation matrix, a respiratory displacement compensation vector is generated for each non-reference time phase. The probe spatial pose data is superimposed with the respiratory displacement compensation vector to obtain the updated probe spatial pose data.

7. The method for image evaluation of the efficacy of traditional Chinese medicine intervention in fatty liver according to claim 1, characterized in that, The reconstruction to obtain the unobstructed three-dimensional echo matrix of the liver includes the following steps: Identify the overlapping spatial regions of the target echo matrix at different time phases; Calculate the local signal-to-noise ratio data of the overlapping spatial region and generate a dynamic fusion weight distribution; Based on the dynamic fusion weight distribution, the unmasked echo data in the target echo matrix is ​​weighted and calculated, and the weighted echo data is filled into the three-dimensional spatial coordinates corresponding to the dynamic sound shadow region to reconstruct and generate a physically continuous unmasked three-dimensional echo matrix of the liver.

8. The image evaluation method for the efficacy of traditional Chinese medicine intervention in fatty liver according to claim 1, characterized in that, The process of generating the corresponding fatty liver treatment evaluation results includes the following steps: Extract the radio frequency envelope signal sequence along the scanning depth direction from the unobstructed liver three-dimensional echo matrix; Calculate the energy attenuation gradient of the radio frequency envelope signal sequence to obtain the target fat attenuation coefficient; Acquire TCM syndrome image data corresponding to the consultation cycle of the ultrasound radio frequency data stream; wherein, the TCM syndrome image data includes at least facial image data and tongue image data; Color and texture features are extracted from the TCM syndrome image data, and the extraction results are input into a preset feature mapping model for processing to obtain the TCM syndrome quantitative integral. Calculate the temporal decline slope of the target fat attenuation coefficient and the temporal improvement slope of the quantitative score of the TCM syndrome. Calculate the correlation index between the time series decline slope and the time series improvement slope; The correlation index is numerically compared with the preset correlation threshold range in the preset efficacy evaluation criteria, and the efficacy level label corresponding to the preset correlation threshold range that matches the correlation index is output to obtain the fatty liver efficacy evaluation result; wherein, the preset efficacy evaluation criteria include at least the preset correlation threshold ranges corresponding to different efficacy level labels.

9. A system for implementing the image evaluation method for the efficacy of traditional Chinese medicine intervention in fatty liver according to any one of claims 1-8, characterized in that, include: The data acquisition module is used to acquire time-stamped ultrasound radio frequency data streams, probe spatial pose data, and abdominal audio streams. The prior trigger time window determination module is used to extract the bowel sound envelope signal in a preset frequency band from the abdominal audio stream, and determine the prior trigger time window characterizing the movement of intestinal gas based on the bowel sound envelope signal. The region segmentation module is used to extract a set of target ultrasound dynamic frames aligned with the prior trigger time window from the ultrasound radio frequency data stream; and to segment the target ultrasound dynamic frame set into a dynamic acoustic shadow region and a corresponding unmasked echo region. The spatial topology deviation measurement module is used to perform preliminary three-dimensional spatial projection of the unmasked echo region at different time phases based on the probe spatial pose data to obtain a candidate echo matrix; extract the intrahepatic vascular topological features in the candidate echo matrix, and calculate the spatial topology deviation value between the intrahepatic vascular topological features at different time phases. The probe spatial pose update module is used to determine whether the spatial topology deviation value is greater than a preset deviation threshold; if so, a respiratory displacement compensation vector is generated based on the intrahepatic vascular topology features and the probe spatial pose data is updated. The preliminary three-dimensional spatial projection is re-executed based on the updated probe spatial pose data until the spatial topology deviation value is less than or equal to the preset deviation threshold. The fatty liver treatment efficacy assessment module is used to take the candidate echo matrix that meets the preset deviation threshold constraint as the target echo matrix, fuse the target echo matrix into the dynamic sound shadow region, and reconstruct an unobstructed three-dimensional echo matrix of the liver. The target fat attenuation coefficient is calculated based on the unobstructed three-dimensional echo matrix of the liver, and the corresponding fatty liver treatment efficacy assessment results are generated.

10. An electronic device comprising a memory and a processor, characterized in that: The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the method as described in any one of claims 1-8.

Images