CN120542477B - Shear wave physical information inversion method based on external excitation driving and related device - Google Patents
Shear wave physical information inversion method based on external excitation driving and related deviceInfo
- Publication number
- CN120542477B CN120542477B CN202511036700.0A CN202511036700A CN120542477B CN 120542477 B CN120542477 B CN 120542477B CN 202511036700 A CN202511036700 A CN 202511036700A CN 120542477 B CN120542477 B CN 120542477B
- Authority
- CN
- China
- Prior art keywords
- physical information
- shear wave
- data
- inversion
- shear
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/08—Clinical applications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/48—Diagnostic techniques
- A61B8/485—Diagnostic techniques involving measuring strain or elastic properties
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/5207—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of raw data to produce diagnostic data, e.g. for generating an image
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/5215—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data
- A61B8/5223—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data for extracting a diagnostic or physiological parameter from medical diagnostic data
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/042—Knowledge-based neural networks; Logical representations of neural networks
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Public Health (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Surgery (AREA)
- Medical Informatics (AREA)
- Heart & Thoracic Surgery (AREA)
- Radiology & Medical Imaging (AREA)
- Pathology (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- Physiology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Ultra Sonic Daignosis Equipment (AREA)
Abstract
The application discloses a shear wave physical information inversion method based on external excitation driving and a related device, and relates to the technical field of physical information reaction, wherein the method comprises the following steps: and triggering low-frequency mechanical vibration on the surface of the detected object by means of an external excitation device, acquiring a shear wave echo signal based on an ultrafast ultrasonic acquisition sequence, and performing vibration detection. And then, inputting discrete space-time coordinates corresponding to the measured data of the pure velocity field into a shear wave physical information inversion model for iterative inversion, so that a two-dimensional inversion result of the shear modulus can be obtained. The shear wave physical information inversion model constructed here contains a physical information neural network that is spatially and spatially correlated with space-time correlation, respectively. According to the application, a two-dimensional inversion model is constructed by introducing the physical information neural network, so that the limit that the traditional instantaneous elastography depends on one-dimensional sampling is successfully broken through, the influence of manual operation on a result is obviously reduced, and the equipment cost and the operation threshold are also greatly reduced.
Description
Technical Field
The application relates to the technical field of physical information inversion, in particular to a shear wave physical information inversion method based on external excitation driving and a related device.
Background
Ultrasound elastography (Ultrasound Elastography) is a medical imaging technique that is specifically used to evaluate the elastic properties of biological tissue. The basic principle is based on different biological tissues (including normal tissues and pathological tissues) with different elastic coefficients. Under the action of external force or alternating vibration, tissues with different elastic coefficients can generate different degrees of strain, and the strain is morphologically represented as deformation of the tissues. The hardness distribution of the tissue is intuitively displayed by capturing signal segments of the region of interest over a specific time period and converting the signals into gray-scale or color-coded images using comprehensive analysis techniques such as autocorrelation.
In the field of research of computational mechanics, conventional numerical methods such as finite-difference method (FDM) and finite-element method (FEM) have been widely used for numerical solution of wave equation and positive problem modeling. However, when solving nonlinear and high-dimensional inverse problems, these methods suffer from the double problems that, on one hand, the inversion process needs to solve the problem repeatedly, resulting in an exponential increase in computational complexity, and on the other hand, the conventional numerical methods are prone to fall into a pathological solution space due to factors such as noise interference, spatial discrete sampling, uncertainty of boundary conditions and the like which are commonly present in actual measurement data, which severely limit the value of the conventional numerical methods in clinical application.
In recent years, deep learning methods have become an effective means of extracting feature representations from high-dimensional observation data. Nevertheless, the model based on pure data driving still has significant limitations that firstly, the labeling of medical image data depends on an ex-vivo tissue mechanical test or numerical simulation, which leads to high data acquisition cost and risk of domain offset, and secondly, the existing feature extraction method lacks physical constraints, and when the distribution of test data exceeds the range of a training set, the generalization capability of the model is drastically reduced, and even a prediction result violating the physical law is possibly generated. Therefore, there is an urgent need to deeply fuse the physical conservation law with the deep learning architecture, and guide the network learning to conform to the feature representation of the physical law by embedding prior knowledge, so as to improve the robustness and extrapolation capability of the model under the condition of limited data.
Disclosure of Invention
The application aims to provide a shear wave physical information inversion method based on external excitation driving and a related device, which can improve the robustness and extrapolation capability of an inversion model under the condition of limited data.
In order to achieve the above object, the present application provides the following solutions:
In a first aspect, the application provides a shear wave physical information inversion method based on external excitation driving, which comprises the following steps:
The method comprises the steps of generating low-frequency mechanical vibration on the surface of a measured object through an external excitation device, acquiring a shear wave echo signal based on an ultrafast ultrasonic acquisition sequence, and performing vibration detection on the shear wave echo signal to obtain actual measurement data of a pure velocity field of the measured object, wherein the external excitation device comprises a signal generator, a power amplifier and a customized vibrator.
The method comprises the steps of converting a linear elastic fluctuation control equation into a physical information driving loss item of a physical information neural network, and constructing a shear wave physical information inversion model, wherein the shear wave physical information inversion model comprises a space-time related physical information neural network and a space-time related physical information neural network, the space-time related physical information neural network is used for inputting discrete time coordinates and discrete space coordinates serving as models, outputting a flow function prediction result and a Lagrangian multiplier prediction result, solving a bias guide on the flow function prediction result to obtain an axial velocity field prediction result, the space-related physical information neural network is used for inputting the discrete space coordinates serving as models, outputting a shear modulus prediction result, and a loss function of the shear wave physical information inversion model is used for calculating total loss by taking the flow function prediction result, pure velocity field actual measurement data and the shear modulus prediction result.
The method comprises the steps of taking discrete space coordinates and discrete time coordinates corresponding to actual measurement data of a pure velocity field as first input data, taking discrete space coordinates corresponding to actual measurement data of the pure velocity field as second input data, inputting the discrete space coordinates into a shear wave physical information inversion model together, and carrying out iterative inversion on the actual measurement data of the pure velocity field as labels of a space-time related physical information neural network to obtain a shear wave physical information inversion result, wherein a loss function of the shear wave physical information inversion model comprises a data driving loss item and a physical information driving loss item, the shear wave physical information inversion result is a shear modulus inversion result corresponding to each discrete space coordinate, and the shear modulus inversion result is a shear modulus prediction result output when the model meets an iterative exit condition after a plurality of iterations.
Optionally, the external excitation device is used for generating low-frequency mechanical vibration on the surface of the measured object, meanwhile, the shear wave echo signal is acquired based on the ultrafast ultrasonic acquisition sequence, and vibration detection is carried out on the shear wave echo signal to obtain the actual measurement data of the pure velocity field of the measured object, and the method specifically comprises the following steps:
the method comprises the steps of generating low-frequency mechanical vibration on the surface of a measured object through an external excitation device, and carrying out displacement field estimation on a shear wave echo signal by adopting a displacement field estimation method based on the phase change of an in-phase component and a quadrature component of an ultrasonic echo signal to obtain tissue micro-displacement field data.
And performing directional filtering on the tissue micro-displacement field data by adopting a frequency domain 2D directional filtering technology to obtain the actual measurement data of the pure velocity field of the measured object, wherein the actual measurement data of the pure velocity field is used as a data driving loss term of the physical information neural network.
Optionally, the displacement field estimation is performed according to the following equation:
。
Wherein, the C is the propagation speed of the ultrasonic wave in the measured object,For the center frequency of the transducer,For pulse repetition frequency, M is the number of sampling points of a single frame, N is the number of continuous acquisition frames,AndRespectively representing the in-phase component and the quadrature component of the mth sampling point of the nth frame.
Optionally, a frequency domain 2D directional filtering technology is adopted to perform directional filtering on the tissue micro displacement field data to obtain the actual measurement number of the pure velocity field of the measured object, which specifically comprises:
and performing fast Fourier transform on the displacement data in the tissue micro-displacement field data to obtain frequency domain displacement data.
Multiplying the frequency domain displacement data with a mask function, and reserving frequency domain components conforming to the shear wave velocity characteristics to obtain filtered frequency domain displacement data.
And recovering the filtered frequency domain displacement data to a time domain by utilizing inverse Fourier transform to obtain the actual measurement number of the pure velocity field of the measured object.
Optionally, the mask function is as follows:
。
Wherein, the As a function of the mask,For the number of waves in space,In order to be of an angular frequency,For the target speed to be the same,To allow for deviations.
Restoring the filtered frequency domain displacement data to the time domain according to the following formula:
。
Wherein, the For the measured number of the pure velocity field of the measured object,And i is an imaginary unit, and t is a discrete time coordinate.
Optionally, the loss function of the shear wave physical information inversion model is as follows:
。
Wherein, the To invert the value of the loss function of the model for shear wave physical information,The loss term is driven for the physical information,For the data-driven loss term,AndThe weights of the physical information drive loss term and the data drive loss term are respectively.
The physical information drive loss term is calculated by:
。
。
Wherein μ is the shear modulus, v x and v y are the velocity data in the x-direction and y-direction, respectively, x and y are the discrete spatial coordinates, t is the discrete time coordinates, p is the partial derivative of Lagrangian multiplier p 0 with respect to time, Ρ is the density of the measured object.
The data drive loss term is calculated by:
。
Wherein, the Is velocity data in the y direction and is equivalent to the measured data of the pure velocity field of the measured object。
In a second aspect, the present application provides a shear wave physical information inversion system based on external excitation driving, comprising:
The ultrasonic application and vibration detection module is used for generating low-frequency mechanical vibration on the surface of the detected object through an external excitation device, acquiring a shear wave echo signal based on an ultrafast ultrasonic acquisition sequence, and performing vibration detection on the shear wave echo signal to obtain actual measurement data of a pure velocity field of the detected object, wherein the external excitation device comprises a signal generator, a power amplifier and a customized vibrator.
The shear wave inversion model construction module is used for converting a linear elastic fluctuation control equation into a physical information driving loss item of a physical information neural network to construct a shear wave physical information inversion model, wherein the shear wave physical information inversion model comprises a space-time related physical information neural network and a space-time related physical information neural network, the space-time related physical information neural network is used for inputting discrete time coordinates and discrete space coordinates serving as models and outputting a flow function prediction result and a Lagrangian multiplier prediction result, the flow function prediction result is biased to obtain an axial velocity field prediction result, the space-related physical information neural network is used for inputting discrete space coordinates serving as models and outputting a shear modulus prediction result, and a loss function of the shear wave physical information inversion model is used for calculating total loss according to the flow function prediction result, pure velocity field actual measurement data and the shear modulus prediction result.
The shear wave physical information inversion module is used for taking the discrete space coordinates and the discrete time coordinates corresponding to the actual measurement data of the pure velocity field as first input data, taking the discrete space coordinates corresponding to the actual measurement data of the pure velocity field as second input data, jointly inputting the discrete space coordinates and the actual measurement data of the pure velocity field into the shear wave physical information inversion model, and carrying out iterative inversion on the actual measurement data of the pure velocity field as labels of a space-time related physical information neural network to obtain a shear wave physical information inversion result, wherein a loss function of the shear wave physical information inversion model comprises a data driving loss item and a physical information driving loss item, the shear wave physical information inversion result is a shear modulus inversion result corresponding to each discrete space coordinate, and the shear modulus inversion result is a shear modulus prediction result output when the model meets an iterative exit condition after a plurality of iterations.
In a third aspect, the application provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor executing the computer program to perform the steps of the external stimulus driven based shear wave physical information inversion method described hereinbefore.
In a fourth aspect, the present application provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the external stimulus driven based shear wave physical information inversion method described hereinbefore.
In a fifth aspect, the application provides a computer program product comprising a computer program which, when executed by a processor, implements the steps of the externally stimulated driven shear wave physical information inversion method described hereinbefore.
According to the specific embodiment provided by the application, the application discloses the following technical effects:
The application provides a shear wave physical information inversion method and a related device based on external excitation driving, wherein the method is characterized in that low-frequency mechanical vibration is generated on the surface of a measured object through the external excitation device, meanwhile, a shear wave echo signal is acquired based on an ultra-fast ultrasonic acquisition sequence and vibration detection is carried out, then discrete space-time coordinates corresponding to actual measurement data of a pure velocity field are input into a shear wave physical information inversion model for iterative inversion, so that a shear wave physical information inversion result, namely a two-dimensional inversion result of shear modulus, is obtained. Compared with the traditional instantaneous elastography (TE) technology, the application constructs a two-dimensional elastic modulus inversion model by introducing the Physical Information Neural Network (PINN), and thoroughly breaks through the limitation that TE only depends on one-dimensional sampling. Moreover, the traditional TE relies on an operator to manually position the ROI, so that subjective deviation risks exist, and the influence of manual operation on the result is remarkably reduced by automatically learning the wave propagation track through PINN, and the method has stronger robustness in detection of fat patients (the failure rate of traditional TE measurement is 20% due to sound attenuation) and deep organs. In addition, the application also has obvious advantages in terms of hardware compatibility and imaging performance, such as the Shear Wave Elastography (SWE) based on acoustic radiation force pulse depends on a high-frequency focusing ultrasonic probe, the imaging depth is generally less than 6cm, the pressure control requirement on the probe is severe, and the application of the probe in deep liver and pelvic organs is severely limited, while the application adopts a low-frequency probe, the penetration depth can reach more than 8cm, and can excite stable shear waves without a special transducer in cooperation with external vibrator excitation, so that the equipment cost and the operation threshold are greatly reduced.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a shear wave physical information inversion method based on external excitation driving according to an embodiment of the present application.
Fig. 2 is a flowchart of step A1 in a shear wave physical information inversion method based on external excitation driving according to an embodiment of the present application.
Fig. 3 is a flowchart of step a12 in a shear wave physical information inversion method based on external excitation driving according to an embodiment of the present application.
Fig. 4 is a schematic diagram of a technical route for performing inversion iteration on a shear wave physical information inversion model in a shear wave physical information inversion method based on external excitation driving according to an embodiment of the present application.
Fig. 5 is a schematic diagram of a functional module of a shear wave physical information inversion system based on external excitation driving according to an embodiment of the present application.
Fig. 6 is a schematic structural diagram of a computer device according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
The biomechanical properties of tissue are widely recognized as important parameters for tissue characterization, and many pathological and physiological changes in the clinic result in changes in the mechanical properties of tissue. Thus, assessing changes in the stiffness of biological soft tissue can help in the diagnosis and treatment of many diseases. Whereas ultrasound elastography has been widely adopted clinically as a non-invasive and reliable technique.
The current ultrasonic shear wave elastography technology has several problems, including:
(1) The imaging depth is insufficient, although the high-frequency probe (more than or equal to 5 MHz) improves the resolution of superficial tissues, the ultrasonic attenuation is obviously increased along with the depth, so that the signal-to-noise ratio of shear wave signals of deep tissues (such as deep liver and pelvic organs with the depth of more than 8 cm) is reduced, the measurement accuracy of elastic modulus is reduced, and the elastic characteristics of deep focuses are difficult to accurately evaluate.
(2) The probe hardware has high requirements that a high-precision ultrasonic probe (such as a broadband sensor) is required to be equipped to rapidly capture the time-space information of the propagation of the shear wave, and the hardware has complex design and high cost.
(3) The force, angle and stability of the pressure applied by the probe directly influence the excitation and propagation of shear waves, and when irregular organs are scanned, the position of the probe needs to be manually adjusted to avoid structures such as blood vessels, bones and the like, and the fluctuation of the elastic value of the same focus is obvious due to the difference of the methods.
(4) The resolution is not high, the spatial resolution is reduced along with the depth, the elastic boundary recognition capability of the micro focus (< 5 mm) is insufficient, the elastic image is generated to be in a salt-pepper effect due to ultrasonic speckle noise and tissue heterogeneity (such as fat particles and fiber intervals), the elastic image is required to be processed smoothly by an algorithm, and local elastic details can be covered.
Ultrasound transient elastography is currently listed by the authorities of the European institute of liver research (EASL), the American society of liver disease research (AASLD) and the like as the first non-invasive method for assessing liver fibrosis, and it is clearly recommended to replace liver biopsy for non-contraindicated patients, but only to provide average elasticity values in one-dimensional ROI, and it is impossible to locate or assess focal lesions, if there is heterogeneity in the tissue such as inflammation, necrosis or tumor, which may lead to distortion of the elasticity values. Although having irreplaceable advantages in the evaluation of diffuse lesions of the liver, the evaluation capability of focal lesions, extrahepatic organs and complex pathological scenes is obviously short, and the evaluation capability is supplemented by multi-mode imaging technology (such as SWE and MRI elastography) and clinical comprehensive judgment.
In view of the problems of the current ultrasonic elastography technology, the application aims to provide an inversion model fused with a physical information network, so that the inversion model can finish accurate inversion of mechanical properties of biological tissues by utilizing limited data driving on the premise of not building a data set.
In order that the above-recited objects, features and advantages of the present application will become more readily apparent, a more particular description of the application will be rendered by reference to the appended drawings and appended detailed description.
In an exemplary embodiment, as shown in fig. 1, the shear wave physical information inversion method based on external excitation driving provided in the embodiment of the present application includes the following steps:
a1, generating low-frequency mechanical vibration on the surface of a measured object through an external excitation device, acquiring a shear wave echo signal based on an ultrafast ultrasonic acquisition sequence, and performing vibration detection on the shear wave echo signal to obtain actual measurement data of a clean speed field of the measured object, wherein the external excitation device comprises a signal generator, a power amplifier and a customized vibrator.
As an exemplary embodiment, the external excitation device constructs a controllable shear wave excitation platform based on the electrical connection architecture of the customized vibrator, the signal generator and the power amplifier. The signal generator is used as an excitation source to output a single-period sine pulse electric signal, the frequency of the single-period sine pulse electric signal can be adjusted in a multi-frequency range, and typical working frequency comprises key frequency ranges such as 200 Hz. After the amplitude amplification treatment is carried out on the electric signal through the power amplifier, the customized vibrator is driven to generate low-frequency mechanical vibration on the surface of the measured object. The vibrations propagate through the interface coupling effect into the interior of the object, thereby inducing shear waves. The external excitation device has high controllability, and can accurately generate shear waves with specific wavelengths according to the requirements of different detection scenes by flexibly setting the frequency, the output amplitude and the waveform parameters of the signal generator and combining a gain matching mechanism of the power amplifier.
The matched acquisition device adopts Versonics programmable ultrasonic platform, and high-precision acquisition of shear wave IQ signals is realized based on an ultrafast ultrasonic acquisition sequence. The platform can be optimally configured according to actual detection requirements through dynamic adjustment of imaging frame rate and resolution parameters, so that dynamic monitoring and accurate analysis of shear wave propagation characteristics are realized.
Compared with the shear wave elastography technology based on acoustic radiation force pulse (Acoustic Radiation Force Impulse, ARFI) excitation, the scheme does not need to rely on special hardware equipment such as a high-end ultrasonic probe and the like. The imaging depth is optimally regulated and controlled through PINN algorithm, so that the dependence degree of the system on the professional experience of an operator is remarkably reduced, and the clinical applicability and popularity of the technology are greatly improved.
A2, converting the linear elastic fluctuation control equation into a physical information driving loss item of a physical information neural network, and constructing to obtain a shear wave physical information inversion model, wherein the shear wave physical information inversion model comprises a space-time related physical information neural network and a space-time related physical information neural network, the space-time related physical information neural network is used for taking discrete time coordinates and discrete space coordinates as model inputs to output a flow function prediction result and a Lagrangian multiplier prediction result, the flow function prediction result is polarized to obtain an axial velocity field prediction result, the space-related physical information neural network is used for taking the discrete space coordinates as model inputs to output a shear modulus prediction result, and a loss function of the shear wave physical information inversion model is used for calculating total loss by taking the flow function prediction result, the pure velocity field actual measurement data and the shear modulus prediction result.
The scheme remarkably enhances the robustness of the system to noise interference in the ultrasonic signal processing process by means of PINN unique physical constraint mechanisms. Experiments show that the method can realize higher spatial resolution in the detection of micro foreign matters (such as early tumor focus), and the performance of the method is obviously superior to that of the traditional numerical calculation method.
A3, taking a discrete space coordinate and a discrete time coordinate corresponding to the measured data of the pure velocity field as first input data, taking a discrete space coordinate corresponding to the measured data of the pure velocity field as second input data, inputting the discrete space coordinate into a shear wave physical information inversion model together, and carrying out iterative inversion on the measured data of the pure velocity field as a label of a space-time related physical information neural network to obtain a shear wave physical information inversion result, wherein a loss function of the shear wave physical information inversion model comprises a data driving loss item and a physical information driving loss item, the shear wave physical information inversion result is a shear modulus inversion result corresponding to each discrete space coordinate, and the shear modulus inversion result is a shear modulus prediction result output when the model meets an iterative exit condition after a plurality of iterations.
In one exemplary embodiment, as shown in fig. 2, step A1 includes the steps of:
A11, generating low-frequency mechanical vibration on the surface of the measured object through an external excitation device, and estimating a displacement field of the shear wave echo signal by adopting a displacement field estimation method based on the phase change of the in-phase component and the quadrature component of the ultrasonic echo signal to obtain tissue micro-displacement field data.
In order to realize the ultra-fast ultrasonic reconstruction speed in the shear wave elastography, the embodiment adopts a displacement field estimation method based on the phase change of the in-phase component I and the quadrature component Q of the ultrasonic echo signal. The method realizes high-precision reconstruction of tissue micro-displacement by analyzing the phase characteristics of IQ signals between adjacent ultrasonic frames. Specifically, a one-dimensional autocorrelation algorithm is employed to determine the relative velocity between reference frames by calculating the average phase shift amount relative to the center frequency. In particular, in this embodiment, the displacement field estimation is performed according to the following equation:
。
Wherein, the C is the propagation speed of the ultrasonic wave in the measured object,For the center frequency of the transducer,For pulse repetition frequency, M is the number of sampling points of a single frame, N is the number of continuous acquisition frames,AndRespectively representing the in-phase component and the quadrature component of the mth sampling point of the nth frame.
The step effectively suppresses random noise interference by carrying out statistical average processing on the phase difference of the shear wave IQ signals, can realize micron-level displacement resolution and kilohertz Cheng Xiangzhen-level transmission rate, is particularly suitable for monitoring the propagation characteristics of shear waves in biological soft tissues, and provides a data basis for high-precision reconstruction of tissue elastic modulus.
A12, performing directional filtering on the tissue micro-displacement field data by adopting a frequency domain 2D directional filtering technology to obtain the actual measurement data of the pure velocity field of the measured object, wherein the actual measurement data of the pure velocity field is used as a data driving loss item of the physical information neural network.
In this embodiment, as shown in fig. 3, step a12 specifically includes the following steps:
a121, performing fast Fourier transform on displacement data in the tissue micro-displacement field data to obtain frequency domain displacement data. Because the boundary condition of the excitation is very complex, aiming at the problems of reflection and refraction interference caused by the complex excitation boundary, the scheme adopts a frequency domain 2D directional filtering technology, and the two-dimensional directional filtering technology is to perform fast Fourier transform FFT (fast Fourier transform FFT) on all points at a certain specific depth, and the last step is obtained through phase estimation In fact the displacement field is a displacement field over a period of time at all points of the whole two-dimensional plane at a specific depth z, i.e. First for the relative speed between reference framesPerforming a Fast Fourier Transform (FFT) to a frequency domain representationWherein, the method comprises the steps of,Is thatIn the case where y is fixed to a certain value, the displacement sets of all points of different x coordinates.
And A122, multiplying the frequency domain displacement data by a mask function, and reserving frequency domain components conforming to the shear wave velocity characteristics to obtain filtered frequency domain displacement data. Specifically, the mask function is shown as follows:
。
Wherein, the As a function of the mask,For the number of waves in space,In order to be of an angular frequency,For the target speed to be the same,To allow for deviations. By combiningAnd (3) withMultiplying to retain frequency domain components conforming to shear wave velocity characteristics and obtain filtered frequency domain displacement data。
And A123, recovering the filtered frequency domain displacement data to a time domain by utilizing inverse Fourier transform to obtain the actual measurement number of the pure velocity field of the measured object. In this embodiment, the filtered frequency domain displacement data may be recovered to the time domain according to the following equation:
。
Wherein, the For the measured number of the pure velocity field of the measured object,And i is an imaginary unit, and t is a discrete time coordinate. Through inverse Fourier transform operation, the frequency domain 2D directional filtering is processedAnd recovering the frequency-wave number domain to the time domain, and reconstructing to obtain pure velocity field data.
The step A12 utilizes the dispersion characteristic of the shear wave and the selective inhibition capability of the directional filtering operator to accurately regulate and control the energy distribution of the interference wave in the frequency domain, effectively filters the frequency domain components of the interference wave such as linear interference, random noise and the like, and obviously improves the signal-to-noise ratio and the resolution of the speed field data.
Specifically, in this embodiment, before the inversion model is constructed, the derivation of the wave equation conforming to the data needs to be completed first, specifically, the two-dimensional dynamic equilibrium equation is shown in the following formula:
。
。
consider here a linear elastic model that satisfies incompressibility and isotropy in two dimensions, whose constitutive equation is expressed as:
。
。
。
Wherein, the 、Respectively representing positive stress and shear stress, wherein subscripts j and k have values of x, y and z; In order to be strained the material is, In order to achieve a shear modulus, the polymer is,Is a non-compressible constraint-dependent lagrange multiplier. The linear elastic wave control equation under a non-uniform medium represented by displacement can be derived by combining the above formulas:
。
due to incompressible constraints, i.e. the volume strain becomes 0:
。
the mixed partial derivatives in the linear elastic wave equation can be combined and simplified:
。
For this embodiment, particle velocity is often used, so calculating the partial derivative of the simplified wave equation over time results in a linear elastic wave control equation expressed by velocity, organized as:
。
Wherein, the And in order to ensure an incompressible condition, i.e. a velocity field divergence of 0, i.eAnd introducing a flow function for naturally satisfying the conditionSatisfies the following formula:
。
Wherein, the physical meaning represented by v y is consistent with the measured data of the pure velocity field after the direction filtering.
The shear wave physical information inversion model is composed of two sub-models constructed by physical information neural networks and is used for solving parameters of spatial correlation and space-time correlation respectively, and the architecture of the shear wave physical information inversion model is shown in figure 4. Wherein the space-time related physical information neural network (NN 1) takes space-time coordinates as input parameters and output parameters as flow functionsAnd Lagrangian multiplier p, wherein NN1 outputs a stream functionVelocity data in different directions can be obtained by deviatorAnd. The spatial correlation physical information neural network (NN 2) takes only the spatial coordinates as input parameters to output the shear modulus. The outputs of NN1 and NN2 are iteratively learned through inversion to fit the input data and solve for a preset physical constraint (i.e., line elastic wave control equation). The shear modulus spatial distribution of the wave propagation region will be inferred by NN2 while fitting the input data through NN 1.
In order that the shear modulus can be correctly inverted, it must be ensured that the incoming shear wave enters from one side and exits the ROI from the other side, as shown in fig. 4, the shear wave propagates from the left side of the ROI to the right side thereof, and for NN1 and NN2 a fully connected feed forward structure is used, and specific neural network structure parameters are shown in table 1.
TABLE 1 neural network structural parameter table
Wherein the number of layers of NN1 and NN2 is 6 (one input layer, one output layer, and 4 hidden layers), each hidden layer contains 80 neurons. The weights and biases will be updated by the inversion iteration process of the back propagation algorithm to achieve the goal of minimizing the loss function L. In the embodiment, tanh is selected as an activation function, and the characteristics of continuity and smoothness of the Tanh are favorable for accurately simulating the behavior of a physical field, and meanwhile, the numerical stability of the model in the inversion iteration process is ensured. Due to the output parameters of NN2Only positive values are present and therefore the final output at NN2 will activate the function through Softplus to ensure that the output results meet the demand.
As an exemplary embodiment, when updating model parameters in the inversion iteration process, the loss function of the shear wave physical information inversion model is as follows:
。
Wherein, the To invert the value of the loss function of the model for shear wave physical information,The loss term is driven for the physical information,For the data-driven loss term,AndThe weights of the physical information driven penalty term and the data driven penalty term, respectively, are used to balance the effects of the two-part penalty function to ensure that they converge effectively during the inversion iteration, especially in experimental data.
The physical information drive loss term is calculated by:
。
。
Wherein μ is the shear modulus, v x and v y are the velocity data in the x-direction and y-direction, respectively, x and y are the discrete spatial coordinates, t is the discrete time coordinates, p is the partial derivative of Lagrangian multiplier p 0 with respect to time, Ρ is the density of the measured object.
The data drive loss term is calculated by:
。
Wherein, the Is velocity data in the y direction and is equivalent to the actual measurement number of the pure velocity field of the measured object。
In order to improve the computational efficiency and ensure the representativeness of the training data, the model inversion iteration process is set to randomly extract 10000 points from the space-time space every 1000 Epochs for inversion iteration. The sampling process employs a latin hypercube sampling (Latin Hypercube Sampling, LHS) method that helps maximize sample coverage and reduce sampling bias by ensuring uniform sampling in each dimension. The strategy not only improves the utilization efficiency of the data, but also accelerates the inversion iteration process of the network.
To stop iterative inversion when the loss function converges, as shown in FIG. 4, a maximum number of iterations N 0=8×104 and a learning rate are definedThe iteration exit condition is when Epochs number N reaches N 0 or the current total lossAnd initial total lossAnd the ratio is smaller than a preset value epsilon, namely stopping calculation. To ensure comparability of the results, the same weight and bias initialization is loaded in the network, and an Adam optimizer is adopted to update parameters so as to speed up convergence and improve training stability.
The scheme builds a collaborative framework of a frequency domain two-dimensional directional filtering technology and PINN inversion models, and effectively suppresses interference factors in the process of shear wave propagation through a closed loop optimization strategy of signal preprocessing-elastic parameter reconstruction. This mechanism ensures that the imaging system can still maintain high accuracy of the elastic parameter reconstruction capability under complex boundary conditions. Meanwhile, based on PINN physical information neural network, the elastic mechanical physical equation and the data driving learning paradigm are creatively fused. The method effectively solves the problem of larger error in the inversion of the elastic modulus of heterogeneous tissues such as fatty liver, liver fibrosis and the like in the traditional technology, and realizes the high-precision quantitative evaluation of the elastic distribution of a focal lesion area. In addition, by introducing the self-adaptive weight distribution strategy, the PINN algorithm can dynamically adjust imaging parameters according to the characteristics of different tissue areas, further optimize the two-dimensional elastic distribution imaging effect and enable the imaging result to be more fit with the clinical actual requirements.
According to the shear wave physical information inversion method based on external excitation driving provided by the embodiment of the application, PINN technology is creatively applied to ultrasonic instantaneous elastography, and various application limitations of various traditional elastography methods are overcome.
Compared with the traditional instantaneous elastography (TE) technology, the application constructs the two-dimensional elastic modulus inversion model by introducing the Physical Information Neural Network (PINN), and thoroughly breaks through the limitation that TE only depends on one-dimensional sampling. While the traditional TE is bound by a uniform medium hypothesis, the elastic modulus inversion error of heterogeneous tissues such as fatty liver, fibrosis and the like exceeds 20%, and the specific hardness characteristics of focal lesions cannot be positioned, the application can accurately capture the change of shear waves in the heterogeneous tissues by combining a PINN embedded elastic mechanical wave equation through an ultra-fast ultrasonic acquisition space-time domain propagation diagram. In addition, the traditional TE relies on an operator to manually position the ROI (such as the S5/S8 section of the right lobe of the liver), and subjective deviation risks exist, and the influence of human operation on the result is obviously reduced through PINN automatic learning of the wave propagation track, so that stronger robustness is particularly shown in obese patients (the failure rate of traditional TE measurement is 20% due to sound attenuation) and deep organ detection.
The present application exhibits significant advantages in hardware compatibility and imaging performance over Shear Wave Elastography (SWE) based on acoustic radiation force pulses (ARFI). The ARFI technology relies on a high-frequency focusing ultrasonic probe, the imaging depth is usually less than 6cm, the pressure control requirement on the probe is severe (the pressure deviation is more than 10g/cm < 2 >, namely, the fluctuation of an elastic value is more than 15%), the application of the probe in deep liver and pelvic organs is severely limited, the penetration depth can reach more than 8cm by adopting a low-frequency probe and matching with external vibrator excitation (such as 200Hz low-frequency vibration), stable shear waves can be excited without a special transducer, and the equipment cost and the operation threshold are greatly reduced. In the algorithm level, the PINN model of the application naturally inhibits ultrasonic speckle noise and motion artifact through physical equation constraint, and can still keep smaller inversion error when the signal-to-noise ratio is low. The design of hardware weight reduction and algorithm intellectualization not only solves the problems of complexity and insufficient depth of ARFI equipment, but also provides an elastography solution with data driving flexibility and mechanical rule rigor for clinic through physical interpretability inversion.
Based on the same inventive concept, the embodiment of the application also provides a system for realizing the shear wave physical information inversion method based on external excitation driving. The implementation of the solution provided by the system is similar to that described in the method above, and in an exemplary embodiment, as shown in fig. 5, there is provided a shear wave physical information inversion system based on external excitation driving, including:
The ultrasonic application and vibration detection module is used for generating low-frequency mechanical vibration on the surface of the detected object through an external excitation device, acquiring a shear wave echo signal based on an ultrafast ultrasonic acquisition sequence, and performing vibration detection on the shear wave echo signal to obtain actual measurement data of a pure velocity field of the detected object, wherein the external excitation device comprises a signal generator, a power amplifier and a customized vibrator.
The shear wave inversion model construction module is used for converting a linear elastic fluctuation control equation into a physical information driving loss item of a physical information neural network to construct a shear wave physical information inversion model, wherein the shear wave physical information inversion model comprises a space-time related physical information neural network and a space-time related physical information neural network, the space-time related physical information neural network is used for inputting discrete time coordinates and discrete space coordinates serving as models and outputting a flow function prediction result and a Lagrangian multiplier prediction result, the flow function prediction result is biased to obtain an axial velocity field prediction result, the space-related physical information neural network is used for inputting discrete space coordinates serving as models and outputting a shear modulus prediction result, and a loss function of the shear wave physical information inversion model is used for calculating total loss according to the flow function prediction result, pure velocity field actual measurement data and the shear modulus prediction result.
The shear wave physical information inversion module is used for taking the discrete space coordinates and the discrete time coordinates corresponding to the actual measurement data of the pure velocity field as first input data, taking the discrete space coordinates corresponding to the actual measurement data of the pure velocity field as second input data, jointly inputting the discrete space coordinates and the actual measurement data of the pure velocity field into the shear wave physical information inversion model, and carrying out iterative inversion on the actual measurement data of the pure velocity field as labels of a space-time related physical information neural network to obtain a shear wave physical information inversion result, wherein a loss function of the shear wave physical information inversion model comprises a data driving loss item and a physical information driving loss item, the shear wave physical information inversion result is a shear modulus inversion result corresponding to each discrete space coordinate, and the shear modulus inversion result is a shear modulus prediction result output when the model meets an iterative exit condition after a plurality of iterations.
Of course, the architecture shown in fig. 5 is merely exemplary, and one or at least two components of the system shown in fig. 5 may be omitted as actually needed when implementing different functions.
In an exemplary embodiment, a computer device, which may be a server or a terminal, is provided, and an internal structure thereof may be as shown in fig. 6. The computer device includes a processor, a memory, an Input/Output interface (I/O) and a communication interface. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface is connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input/output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used for communicating with an external terminal through a network connection. The computer program, when executed by a processor, can implement a shear wave physical information inversion method based on external excitation driving provided in the above embodiment.
It will be appreciated by those skilled in the art that the structure shown in FIG. 6 is merely a block diagram of some of the structures associated with the present inventive arrangements and is not limiting of the computer device to which the present inventive arrangements may be applied, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
In an exemplary embodiment, a computer device is also provided, comprising a memory and a processor, the memory having stored therein a computer program, the processor implementing the steps of the method embodiments described above when the computer program is executed.
In an exemplary embodiment, a computer-readable storage medium is provided, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method embodiments described above.
In an exemplary embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of the method embodiments described above.
It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are both information and data authorized by the user or sufficiently authorized by each party, and the collection, use and processing of the related data are required to meet the related regulations.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magneto-resistive random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (PHASE CHANGE Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in various forms such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), etc.
The databases referred to in the embodiments provided herein may include at least one of a relational database and a non-relational database. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processor referred to in the embodiments provided in the present application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, or the like, but is not limited thereto.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The principles and embodiments of the present application have been described herein with reference to specific examples, which are intended to facilitate an understanding of the principles and concepts of the application and are to be varied in scope and detail by persons of ordinary skill in the art based on the teachings herein. In view of the foregoing, this description should not be construed as limiting the application.
Claims (10)
1. The shear wave physical information inversion method based on external excitation driving is characterized by comprising the following steps of:
Generating low-frequency mechanical vibration on the surface of a measured object through an external excitation device, acquiring a shear wave echo signal based on an ultrafast ultrasonic acquisition sequence, and performing vibration detection on the shear wave echo signal to obtain actual measurement data of a pure velocity field of the measured object;
Converting a linear elastic fluctuation control equation into a physical information driving loss term of a physical information neural network, and constructing a shear wave physical information inversion model, wherein the shear wave physical information inversion model comprises a space-time related physical information neural network and a space-time related physical information neural network, the space-time related physical information neural network is used for taking discrete time coordinates and discrete space coordinates as model inputs and outputting a flow function prediction result and a Lagrange multiplier prediction result, the flow function prediction result is subjected to partial derivation to obtain an axial velocity field prediction result, the space-related physical information neural network is used for taking the discrete space coordinates as model inputs and outputting a shear modulus prediction result, and a loss function of the shear wave physical information inversion model is used for calculating total loss by taking the flow function prediction result, the pure velocity field actual measurement data and the shear modulus prediction result;
The method comprises the steps of taking a discrete space coordinate and a discrete time coordinate corresponding to the measured data of the pure velocity field as first input data, taking a discrete space coordinate corresponding to the measured data of the pure velocity field as second input data, jointly inputting the discrete space coordinate and the discrete space coordinate into the shear wave physical information inversion model, and carrying out iterative inversion on the measured data of the pure velocity field as a label of a space-time related physical information neural network to obtain a shear wave physical information inversion result, wherein a loss function of the shear wave physical information inversion model comprises a data driving loss item and a physical information driving loss item, the shear wave physical information inversion result is a shear modulus inversion result corresponding to each discrete space coordinate, and the shear modulus inversion result is a shear modulus prediction result output when the model meets an iterative exit condition after a plurality of iterations.
2. The shear wave physical information inversion method based on external excitation driving of claim 1, wherein the external excitation device is used for generating low-frequency mechanical vibration on the surface of the measured object, simultaneously acquiring a shear wave echo signal based on an ultrafast ultrasonic acquisition sequence, and performing vibration detection on the shear wave echo signal to obtain actual measurement data of a pure velocity field of the measured object, and the method specifically comprises the following steps:
Generating low-frequency mechanical vibration on the surface of a measured object through an external excitation device, and carrying out displacement field estimation on the shear wave echo signal by adopting a displacement field estimation method based on the phase change of an in-phase component and a quadrature component of an ultrasonic echo signal to obtain tissue micro-displacement field data;
And performing directional filtering on the tissue micro-displacement field data by adopting a frequency domain 2D directional filtering technology to obtain the actual measurement data of the pure velocity field of the measured object, wherein the actual measurement data of the pure velocity field is used as a data driving loss term of a physical information neural network.
3. The external excitation drive based shear wave physical information inversion method according to claim 2, wherein the displacement field estimation is performed according to the following equation:
;
Wherein, the C is the propagation speed of the ultrasonic wave in the measured object,For the center frequency of the transducer,For pulse repetition frequency, M is the number of sampling points of a single frame, N is the number of continuous acquisition frames,AndRespectively representing the in-phase component and the quadrature component of the mth sampling point of the nth frame.
4. The shear wave physical information inversion method based on external excitation driving of claim 2, wherein the method is characterized in that the method adopts a frequency domain 2D directional filtering technology to perform directional filtering on the tissue micro-displacement field data to obtain pure velocity field actual measurement data of the measured object, and specifically comprises the following steps:
performing fast Fourier transform on displacement data in the tissue micro-displacement field data to obtain frequency domain displacement data;
Multiplying the frequency domain displacement data with a mask function, and reserving frequency domain components conforming to the shear wave speed characteristics to obtain filtered frequency domain displacement data;
and recovering the filtered frequency domain displacement data to a time domain by utilizing inverse Fourier transform to obtain actual measurement data of the pure velocity field of the measured object.
5. The external stimulus driven shear wave physical information inversion method of claim 4 wherein the mask function is as follows:
;
Wherein, the As a function of the mask,For the number of waves in space,In order to be of an angular frequency,For the target speed to be the same,Is an allowable deviation;
Restoring the filtered frequency domain displacement data to the time domain according to the following formula:
;
Wherein, the For the measured data of the pure velocity field of the measured object,And i is an imaginary unit, and t is a discrete time coordinate.
6. The external stimulus driven shear wave physical information inversion method of claim 2, wherein the loss function of the shear wave physical information inversion model is represented by the following formula:
;
Wherein, the To invert the value of the loss function of the model for shear wave physical information,The loss term is driven for the physical information,For the data-driven loss term,AndWeights of the physical information drive loss term and the data drive loss term respectively;
the physical information drive loss term is calculated by:
;
;
Wherein μ is the shear modulus, v x and v y are the velocity data in the x-direction and y-direction, respectively, x and y are the discrete spatial coordinates, t is the discrete time coordinates, p is the partial derivative of Lagrangian multiplier p 0 with respect to time, Ρ is the density of the measured object;
The data drive loss term is calculated by:
;
Wherein, the Is velocity data in the y direction and is equivalent to the measured data of the pure velocity field of the measured object。
7. An external excitation drive-based shear wave physical information inversion system, comprising:
The ultrasonic application and vibration detection module is used for generating low-frequency mechanical vibration on the surface of the detected object through an external excitation device, acquiring a shear wave echo signal based on an ultrafast ultrasonic acquisition sequence, and performing vibration detection on the shear wave echo signal to obtain actual measurement data of a pure velocity field of the detected object;
The shear wave inversion model construction module is used for converting a linear elastic fluctuation control equation into a physical information driving loss item of a physical information neural network to construct a shear wave physical information inversion model, wherein the shear wave physical information inversion model comprises a space-time related physical information neural network and a space-time related physical information neural network, the space-time related physical information neural network is used for taking discrete time coordinates and discrete space coordinates as model input to output a flow function prediction result and a Lagrangian multiplier prediction result, the flow function prediction result is biased to obtain an axial velocity field prediction result, the space-related physical information neural network is used for taking the discrete space coordinates as model input to output a shear modulus prediction result, and a loss function of the shear wave physical information inversion model is used for calculating total loss by taking the flow function prediction result, the pure velocity field actual measurement data and the shear modulus prediction result;
The shear wave physical information inversion module is used for taking the discrete space coordinate and the discrete time coordinate corresponding to the measured data of the pure velocity field as first input data, taking the discrete space coordinate corresponding to the measured data of the pure velocity field as second input data, jointly inputting the discrete space coordinate and the discrete time coordinate into the shear wave physical information inversion model, and taking the measured data of the pure velocity field as a label of a space-time related physical information neural network for iterative inversion to obtain a shear wave physical information inversion result, wherein a loss function of the shear wave physical information inversion model comprises a data driving loss item and a physical information driving loss item, the shear wave physical information inversion result is a shear modulus inversion result corresponding to each discrete space coordinate, and the shear modulus inversion result is a shear modulus prediction result output when the model meets an iterative exit condition after a plurality of iterations.
8. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor executes the computer program to implement the external stimulus driven shear wave physical information inversion method according to any of claims 1-6.
9. A computer readable storage medium having stored thereon a computer program, which when executed by a processor implements the external stimulus driven shear wave physical information inversion method of any of claims 1-6.
10. A computer program product comprising a computer program which, when executed by a processor, implements the external stimulus driven shear wave physical information inversion method of any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202511036700.0A CN120542477B (en) | 2025-07-28 | 2025-07-28 | Shear wave physical information inversion method based on external excitation driving and related device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202511036700.0A CN120542477B (en) | 2025-07-28 | 2025-07-28 | Shear wave physical information inversion method based on external excitation driving and related device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN120542477A CN120542477A (en) | 2025-08-26 |
| CN120542477B true CN120542477B (en) | 2025-09-23 |
Family
ID=96786002
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202511036700.0A Active CN120542477B (en) | 2025-07-28 | 2025-07-28 | Shear wave physical information inversion method based on external excitation driving and related device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN120542477B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119138926A (en) * | 2024-11-20 | 2024-12-17 | 深圳大学 | Ultrasonic viscoelasticity measurement method, device, equipment and medium |
| CN120163018A (en) * | 2025-03-18 | 2025-06-17 | 华东理工大学 | Complex medium inversion imaging method and system guided by physical information neural network |
-
2025
- 2025-07-28 CN CN202511036700.0A patent/CN120542477B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119138926A (en) * | 2024-11-20 | 2024-12-17 | 深圳大学 | Ultrasonic viscoelasticity measurement method, device, equipment and medium |
| CN120163018A (en) * | 2025-03-18 | 2025-06-17 | 华东理工大学 | Complex medium inversion imaging method and system guided by physical information neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN120542477A (en) | 2025-08-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109978778B (en) | Convolutional neural network medical CT image denoising method based on residual learning | |
| CN112771374B (en) | Image reconstruction method based on training nonlinear mapping | |
| Zhou et al. | High spatial–temporal resolution reconstruction of plane-wave ultrasound images with a multichannel multiscale convolutional neural network | |
| Prakash et al. | Fractional regularization to improve photoacoustic tomographic image reconstruction | |
| Yin et al. | SWENet: A physics-informed deep neural network (PINN) for shear wave elastography | |
| CN101902970B (en) | Method and system of strain gain compensation in elasticity imaging | |
| Madasamy et al. | Deep learning methods hold promise for light fluence compensation in three-dimensional optoacoustic imaging | |
| CN115797263B (en) | Three-dimensional ultrasonic craniocerebral imaging method based on full convolution network | |
| Awasthi et al. | Sinogram super-resolution and denoising convolutional neural network (SRCN) for limited data photoacoustic tomography | |
| Delaunay et al. | An unsupervised approach to ultrasound elastography with end-to-end strain regularisation | |
| CN112022215A (en) | Ultrasonic elastography cornea detection method, device and system and storage medium | |
| CN115700106A (en) | Ultrasonic method and system for estimating nonlinear shear wave elasticity of media | |
| Wang et al. | Adaptive machine learning method for photoacoustic computed tomography based on sparse array sensor data | |
| Wang et al. | Ultrasound lung aeration map via physics-aware neural operators | |
| CN108392751B (en) | A kind of method of real-time monitoring High Intensity Focused Ultrasound treatment acoustic cavitation | |
| Zhang et al. | Ultrasound imaging based on the variance of a diffusion restoration model | |
| CN120542477B (en) | Shear wave physical information inversion method based on external excitation driving and related device | |
| Bochud et al. | Axial transmission: Techniques, devices and clinical results | |
| Ahmed et al. | SHEAR-net: An end-to-end deep learning approach for single push ultrasound shear wave elasticity imaging | |
| Csány et al. | A real-time data-based scan conversion method for single element ultrasound transducers | |
| Floquet et al. | Spatially variant ultrasound image restoration with product convolution | |
| Mao et al. | Ultrasound-assisted aberration correction of transcranial photoacoustic imaging based on angular spectrum theory | |
| Kuo | Advancing photoacoustic neuroimaging through deep learning | |
| CN103142216A (en) | Method for calculating multilayer medium sound velocity based on photoacoustic imaging technology | |
| KR20230011856A (en) | Method and apparatus for quantitative imaging in medical ultrasound |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |