A transient elastography method and apparatus

Abstract

The application discloses a transient elastography method and device, wherein the method comprises: in a transient elastography mode: based on a transient elastic ultrasound probe, performing real-time ultrasound scanning on a first target tissue of a target object to obtain first ultrasound image data of the first target tissue; processing the first ultrasound image data to obtain a first target feature map; determining a tissue type of the first target tissue based on the first target feature map; automatically determining a first elastography strategy based on the tissue type of the first target tissue; controlling the transient elastic ultrasound probe to apply mechanical vibration to the first target tissue to excite shear waves to propagate in the first target tissue based on the first elastography strategy, and detecting the propagation of the shear waves by the transient elastic ultrasound probe to perform transient elastography on the first target tissue, and obtaining a transient elastic image of the first target tissue.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical imaging, in particular to a transient elastography method and device. BACKGROUND

[0002] Elastography is a medical imaging method that evaluates the mechanical properties of tissue by measuring the propagation speed of shear waves inside the tissue. This technique has the advantages of convenient operation, low cost, and strong repeatability, and has been widely used in clinical practice and gradually expanded to the detection of elasticity of various physiological tissues. However, related elastography methods usually rely on users to manually select the examination method suitable for the current tissue type from a number of preset elastography modes. This fixed mode selection method lacks flexibility and is difficult to adapt to the differences in anatomical structure and physiological characteristics of different tissues, resulting in limited measurement accuracy or reduced operation efficiency. SUMMARY

[0003] To solve the above technical problems, the embodiments of the present application provide a transient elastography method and device.

[0004] The elastography method provided by the embodiments of the present application is applied to a transient elastography device, and includes:

[0005] In response to a first selection instruction of a user, enter a transient elastography mode;

[0006] In the transient elastography mode:

[0007] Perform real-time ultrasound scanning on a first target tissue of a target object based on a transient elastography ultrasound probe to obtain first ultrasound image data about the first target tissue;

[0008] Process the first ultrasound image data to obtain a plurality of first feature maps; splice the plurality of first feature maps in the channel dimension and perform channel compression to obtain a second feature map;

[0009] Extract hierarchical features in the second feature map to obtain a plurality of third feature maps, and extract a fourth feature map in the plurality of third feature maps based on a self-attention mechanism; perform layer-by-layer upsampling on the fourth feature map to the original resolution to obtain a first target feature map; determine the tissue type of the first target tissue based on the first target feature map;

[0010] Automatically determine a first elastography strategy based on the tissue type of the first target tissue, the first elastography strategy including at least one of a first vibration control strategy, a first emission scanning strategy, or a first imaging processing method, wherein the first elastography strategy is an elastography strategy matched with the tissue type of the first target tissue;

[0011] control the transient elastography ultrasound probe to apply mechanical vibration to the first target tissue to excite shear wave to propagate in the first target tissue, and detect the propagation of the shear wave by the transient elastography ultrasound probe to perform transient elastography on the first target tissue, and obtain a transient elastography image of the first target tissue.

[0012] The elastic imaging method provided in the embodiments of the present application is applied to a transient elastography device, and includes the following steps.

[0013] In response to a first selection instruction of a user, a transient elastography mode is entered.

[0014] In the transient elastography mode, the following steps are performed.

[0015] Perform real-time ultrasound scanning on a first target tissue of a target object by a transient elastography ultrasound probe to obtain first ultrasound image data of the first target tissue;

[0016] Determine a tissue type of the first target tissue based on the first ultrasound image data;

[0017] Automatically determine a first elastic imaging strategy based on the tissue type of the first target tissue, wherein the first elastic imaging strategy is an elastic imaging strategy matched with the tissue type of the first target tissue;

[0018] Control the transient elastography ultrasound probe to apply mechanical vibration to the first target tissue to excite shear wave to propagate in the first target tissue, and detect the propagation of the shear wave by the transient elastography ultrasound probe to perform transient elastography on the first target tissue, and obtain a transient elastography image of the first target tissue.

[0019] The transient elastography device provided in the embodiments of the present application includes a transmitting and receiving circuit, a transient elastography ultrasound probe, a vibration excitation device and a processor, and in particular,

[0020] The transmitting and receiving circuit is configured to control the transient elastography ultrasound probe to perform real-time ultrasound scanning.

[0021] The vibration excitation device is configured to control the transient elastography ultrasound probe to apply mechanical vibration to excite shear wave to propagate in a target tissue.

[0022] The transmitting and receiving circuit is further configured to detect the propagation of the shear wave by the transient elastography ultrasound probe.

[0023] The processor is configured to perform any one of the transient elastography methods.

[0024] In the technical solution of the embodiment of the present application, the transient elastography device enters the transient elastography mode in response to a first selection instruction of a user; in the transient elastography mode: based on the transient elastography probe, real-time ultrasound scanning is performed on a first target tissue of a target object to obtain first ultrasound image data about the first target tissue; the first ultrasound image data is processed to obtain a plurality of first feature maps; the plurality of first feature maps are spliced in the channel dimension and compressed in the channel dimension to obtain a second feature map; hierarchical features in the second feature map are extracted to obtain a plurality of third feature maps, and a fourth feature map in the plurality of third feature maps is extracted based on a self-attention mechanism; the fourth feature map is up-sampled layer by layer to the original resolution to obtain a first target feature map; the tissue type of the first target tissue is determined based on the first target feature map; a first elastography strategy is automatically determined based on the tissue type of the first target tissue, the first elastography strategy including at least one of a first vibration control strategy, a first emission scanning strategy, or a first imaging processing method, wherein the first elastography strategy is an elastography strategy matched with the tissue type of the first target tissue; the transient elastography probe is controlled to apply mechanical vibration to the first target tissue to excite the propagation of shear waves in the first target tissue based on the first elastography strategy, and the propagation of the shear waves is detected by the transient elastography probe to perform transient elastography on the first target tissue to obtain a transient elastography image of the first target tissue. In this way, first, the transient elastography mode is entered in response to the user instruction, then the image data of the target tissue is collected by the ultrasound probe, the tissue type is identified by using multi-modal feature fusion and a self-attention mechanism, and the elastography strategy is automatically adjusted according to the identification result, and finally the personalized elastography for different tissues is realized. In this way, on the one hand, by channel splicing and compression of the image data of the target tissue, multi-level features are extracted and combined with the self-attention mechanism, which can improve the identification accuracy of the tissue type; on the other hand, by presetting multiple elastography strategies and automatically switching according to the identification result, the adaptability of the transient elastography device to different tissue types can be improved, thereby avoiding the operation complexity and time waste caused by manual switching, and improving the measurement efficiency and accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 is a flowchart of a transient elastography method provided by the embodiment of the present application Figure One ;

[0026] Figure 2 is a flowchart of a transient elastography method provided by the embodiment of the present application Figure Two ;

[0027] Figure 3 is a frame diagram of a transient elastography system provided by the embodiment of the present application, which can be adjusted in real time

[0028] Figure 4 is a flowchart of a process of target tissue identification performed by the real-time self-adaptive transient elastography system provided in the embodiments of the present application;

[0029] Figure 5 is a flowchart of a process of transient elastography performed by the real-time self-adaptive transient elastography system provided in the embodiments of the present application Figure One ;

[0030] Figure 6 is a flowchart of a process of transient elastography performed by the real-time self-adaptive transient elastography system provided in the embodiments of the present application Figure Two ;

[0031] Figure 7 is a structural diagram of the real-time self-adaptive transient elastography device provided in the embodiments of the present application. DETAILED DESCRIPTION

[0032] The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0033] In order to facilitate the understanding of the technical solutions of the embodiments of the present application, the related technologies of the embodiments of the present application are described below. The following related technologies can be combined with the technical solutions of the embodiments of the present application in any way as optional solutions, which all fall within the protection scope of the embodiments of the present application.

[0034] Transient elastography is a technology for quantitatively evaluating the elasticity or hardness of tissue by detecting the propagation speed of shear waves in the tissue after the shear waves caused by transient vibration of a probe propagate to the tissue through the body surface. Because of the characteristics of non-invasiveness, convenience, low cost, and easy repeatability of the transient elastography technology itself, it has been widely recognized in the clinic and has been included in the clinical diagnosis and treatment guidelines.

[0035] In recent years, the application of transient elastography technology in various physiological tissues has gradually expanded, providing new possibilities for clinical diagnosis. Because the anatomical structures and physiological characteristics of different tissues are different, in order to obtain more accurate measurement results, there are certain differences in the vibration control strategy, ultrasonic emission scanning strategy, and imaging processing method of transient elastography technology applied to different tissues.

[0036] Clinical operators often need to perform transient elastography between different tissues. However, the related transient elastography system usually needs the user to manually select one transient elastography mode from one or more transient elastography modes preset by the system before performing transient elastography measurement. Such a system cannot be automatically adjusted or switched in real time, resulting in limited functions of the system and limiting the adaptability of the system in different clinical applications. Due to the inability of the system to automatically adjust or switch in real time, the operation efficiency of the user is reduced, and the examination time and waiting time of the patient are prolonged.

[0037] In view of the problem that the related transient elastography system cannot automatically switch the corresponding transient elastography mode in real time during the examination process according to the change of the to-be-measured part, the present application provides an elastic imaging method capable of real-time adaptive adjustment and a transient elastography imaging system carrying the elastic imaging method. In the acquisition preparation state, the system identifies the target tissue to be measured in the current scanning section, and correspondingly adjusts the vibration control strategy of transient elasticity, the ultrasonic emission scanning strategy and the imaging processing method, so as to finally ensure that the transient elastography measurement of different tissues can efficiently and accurately obtain reliable measurement results.

[0038] In order to facilitate understanding of the technical solutions of the embodiments of the present application, the technical solutions of the present application are described in detail below through specific embodiments. The above related technologies can be combined with the technical solutions of the embodiments of the present application as optional solutions, which all belong to the protection scope of the embodiments of the present application. The embodiments of the present application include at least part of the following contents.

[0039] The present application provides a transient elastography imaging method, which is applied to a transient elastography imaging device. The transient elastography imaging device can be an ultrasonic diagnostic instrument, an elastic imaging probe or other hardware devices.

[0040] Figure 1 The present application provides a transient elastography imaging method, which is applied to a transient elastography imaging device. The transient elastography imaging device can be an ultrasonic diagnostic instrument, an elastic imaging probe or other hardware devices. Figure One As shown in Figure 1 The method comprises the following steps:

[0041] Step 101: In response to a first selection instruction of a user, enter a transient elastography imaging mode.

[0042] In the embodiments of the present application, the user usually issues a first selection instruction in the form of a touch screen, a button, voice, etc. to activate the transient elastography mode. Thus, upon receiving the first selection instruction from the user, the transient elastography device will respond to the instruction immediately and enter the transient elastography mode. In the transient elastography mode, the transient elastography device will switch to a state dedicated to elasticity measurement and prepare for subsequent scanning and data processing procedures. For example, when the user clicks the elastography button on the interface of an ultrasound diagnostic instrument or presses a specific function key on a handheld device, the device enters the transient elastography mode. After entering the transient elastography mode, the device will automatically load the relevant preset parameters and initialize the software and hardware modules required for elastography.

[0043] Step 102: In the transient elastography mode: performing real-time ultrasound scanning on the first target tissue of the target object based on the transient elastography probe to obtain first ultrasound image data about the first target tissue.

[0044] In the embodiments of the present application, the transient elastography probe is a special ultrasound probe capable of transmitting mechanical vibrations and simultaneously receiving shear wave propagation signals. The transient elastography probe not only has the B-mode imaging capability of an ordinary ultrasound probe, but also integrates a vibration device for exciting shear waves and a high-frame-rate transmission scanning system for detecting shear wave propagation. The first target tissue refers to the biological tissue in the current scanning region, such as the liver, spleen, kidney, etc. When the transient elastography device is in the transient elastography mode, the transient elastography probe in the transient elastography device will perform real-time ultrasound scanning on the first target tissue of the target object to obtain first ultrasound image data about the first target tissue. Real-time ultrasound scanning means continuously acquiring image data without interrupting imaging to ensure dynamic observation of the state of the first target tissue. The first ultrasound image data is multi-modal data, and each modal data is a single-channel image or a multi-channel image.

[0045] For example, when a doctor uses an ultrasound probe to scan the liver region of a patient, the transient elastography probe will continuously send high-frequency ultrasound beams, and the transient elastography probe receives the echo signals to generate B-mode images. At the same time, the transient elastography probe records the morphology and motion information of the first target tissue, which are used as basic data for subsequent elasticity analysis.

[0046] In some embodiments, step 102 specifically includes:

[0047] sending a second ultrasound wave to the first target tissue based on the transient elastography probe, and receiving a second ultrasound echo signal returned by the first target tissue in response to the second ultrasound wave;

[0048] The second ultrasonic echo signal is processed to obtain first ultrasonic image data.

[0049] The second ultrasonic wave is an ultrasonic wave of a specific frequency and amplitude emitted by the transient elastography ultrasonic probe to excite shear waves in the tissue. The second ultrasonic wave is usually a low-frequency sine wave or a triangular wave, and the frequency and amplitude are usually dynamically adjusted according to the tissue type of the first target tissue to adapt to the physical characteristics of the first target tissue. For example, when measuring the liver, the system can use an ultrasonic wave in the form of a sine wave with a frequency of 50 Hz and an amplitude of 1 mm, and when measuring the spleen, the system can use an ultrasonic wave in the form of a sine wave with a frequency of 100 Hz and an amplitude of 0.5 mm.

[0050] Specifically, by emitting the second ultrasonic wave of a specific frequency and amplitude into the first target tissue through the transient elastography ultrasonic probe, when the second ultrasonic wave enters the first target tissue, it will cause micro-vibration in the tissue, thereby forming a shear wave. The shear wave is detected by the transient elastography ultrasonic probe during propagation in the tissue and is forwarded as a second ultrasonic echo signal, which contains the propagation speed information of the shear wave and is the basis for subsequent image reconstruction and elasticity calculation. Because the elasticity of different tissues differs greatly, the content of the second ultrasonic echo signal will also change accordingly, and the hardness of the first target tissue can be accurately reflected by processing the second ultrasonic echo signal.

[0051] After receiving the second ultrasonic echo signal returned by the first target tissue to the second ultrasonic wave, the second ultrasonic echo signal can be filtered, amplified, time of flight (TOF) calculated, cross-correlation analyzed, and other signal processing, to extract the propagation speed and direction of the shear wave, and then convert the second ultrasonic echo signal into visual first ultrasonic image data.

[0052] Step 103: processing the first ultrasonic image data to obtain a plurality of first feature maps; concatenating and channel compressing the plurality of first feature maps in the channel dimension to obtain a second feature map.

[0053] In the embodiments of the present application, by performing convolution, batch normalization processing and activation processing on the first ultrasonic image data, different types of local feature maps, i.e. a plurality of first feature maps, can be extracted, which include edge features, texture features, direction gradient features, etc. Then, the channel information of the plurality of first feature maps is merged into a same multi-channel feature map, and the multi-channel feature map is channel compressed by convolution operation to reduce the feature dimension, and a second feature map can be obtained.

[0054] For example, assuming that the first ultrasound image data includes three modal image data of B-mode images, color blood flow images, and tissue displacement images, the corresponding first feature maps are extracted , each of which has a size of (n is the number of channels), and the three first feature maps are spliced in the channel dimension to form a multi-channel feature map , which is then processed by 1x1 convolution to compress the number of channels to 64, and finally output the second feature map .

[0055] In some embodiments, the first ultrasound image data includes first black-and-white ultrasound image data and / or first color ultrasound image data. The first black-and-white ultrasound image data is gray-scale image data collected by an ultrasound probe and does not contain color blood flow information, and is mainly used to observe the morphological structure of tissues, such as the outline and boundary of organs such as the liver and spleen. The first color ultrasound image data is blood flow signals superimposed on the basis of black-and-white images, used to display blood vessel distribution, blood flow direction and speed, commonly used to assess hemodynamic status. The two kinds of image data can be used alone or in combination to provide richer input information, thereby improving the accuracy and robustness of target tissue recognition.

[0056] Based on this, step 103 specifically includes:

[0057] The first black-and-white ultrasound image data and / or the first color ultrasound image data are input into the first network model, and the first black-and-white ultrasound image data and / or the first color ultrasound image data are convolved, batch normalized, and activated by the first network model to obtain a plurality of first feature maps;

[0058] The plurality of first feature maps are spliced in the channel dimension to obtain a first spliced feature map, and the first spliced feature map is channel compressed to obtain a second feature map.

[0059] Specifically, first, the first black-and-white ultrasound image data and / or the first color ultrasound image data are input into the first network model, and a plurality of filters are applied by the first network model to extract spatial features of the first black-and-white ultrasound image data and / or the first color ultrasound image data, such as edges, corners, and textures, and the extracted features are standardized with a mean of 0 and a variance of 1, and the standardized features are activated with nonlinear characteristics to obtain a plurality of first feature maps; then, the plurality of first feature maps are combined along the channel dimension to form a high-dimensional feature tensor, and the high-dimensional feature tensor is channel compressed to obtain a second feature map. Channel compression processing can reduce computational overhead while retaining key information and avoiding overfitting. ​

[0060] Step 104: extracting hierarchical features in the second feature map to obtain a plurality of third feature maps, and extracting a fourth feature map in the plurality of third feature maps based on a self-attention mechanism; performing layer-by-layer upsampling on the fourth feature map to the original resolution to obtain a first target feature map.

[0061] In the embodiments of the present application, the hierarchical features refer to features of different abstraction levels from low to high, which can be extracted from the second feature map by an encoder structure in a deep neural network to obtain a plurality of third feature maps, and a long-distance dependency relationship in the plurality of third feature maps can be captured based on a self-attention mechanism to obtain a final feature representation after self-attention processing, that is, a fourth feature map, which contains more global and semantic information. Then, the fourth feature map is layer-by-layer upsampled by transposed convolution or the like to gradually restore the resolution of the image (consistent with the size of the first ultrasound image data) to obtain the first target feature map.

[0062] For example, ResNet34 can be used as a convolutional encoder to extract hierarchical features in the second feature map , and four layers of feature maps are output, each of which corresponds to different spatial resolutions and channel numbers; then, the deepest feature map in the four layers of features is flattened into a sequence form and input into a module composed of a six-layer multi-head self-attention mechanism and a feedforward fully connected network to obtain the fourth feature map ; subsequently, the fourth feature map is layer-by-layer upsampled to the original resolution by a transposed convolution operation in the decoder, and weighted addition is performed after alignment at the skip connection with the feature channels of the corresponding encoding layer to obtain the first target feature map .

[0063] In some embodiments, step 104 specifically includes:

[0064] inputting the second feature map into a second network model, extracting hierarchical features in the second feature map by the second network model to obtain a plurality of third feature maps; and inputting the plurality of third feature maps into a third network model, extracting a fourth feature map in the plurality of third feature maps based on a self-attention mechanism by the third network model;

[0065] inputting the fourth feature map into a fourth network model, layer-by-layer upsampled the fourth feature map to the original resolution by transposed convolution in the fourth network model, and performing weighted addition after alignment at the skip connection with the feature channels of the plurality of third feature maps to obtain the first target feature map.

[0066] Specifically, the hierarchical features refer to the abstract features formed from shallow to deep in the deep neural network as the network layers deepen. The first layer may extract edge information, the second layer extracts texture information, the third layer extracts shape information, and so on. By inputting the second feature map into the second network model, the multi-level features in the second feature map can be extracted by the second network model to obtain a plurality of third feature maps. The multi-level feature extraction manner helps the model better understand the image content and improves the accuracy of target tissue classification.

[0067] Then, the deepest level feature map in the plurality of third feature maps can be input into a third network model, and the third network model can capture the dependency relationship between each element in the deepest level feature map based on a self-attention mechanism to more effectively focus on the key positions in the ultrasound image, and obtain a fourth feature map after self-attention processing. The deepest level feature map refers to the feature map in the plurality of third feature maps that contains highly abstract semantic information, which can capture global semantic association information rather than local details.

[0068] Then, the fourth feature map can be input into a fourth network model, and the fourth network model can perform layer-by-layer upsampling on the fourth feature map to the resolution of the original image (the first ultrasound image data) by using a reverse convolution operation (transposed convolution). In this process, the shallow features and the deep features can be fused by performing feature channel alignment and weighted addition on the plurality of third feature maps extracted by the second network model at the jump connection, and finally a first target feature map is obtained.

[0069] Step 105: determining the tissue type of the first target tissue based on the first target feature map.

[0070] In the embodiments of the present application, the tissue type of the first target tissue refers to the biological tissue species in the first target tissue region, such as liver, spleen, kidney, pancreas, blood vessel, muscle, or lesion, etc. After obtaining the first target feature map, global average pooling can be performed on the first target feature map to obtain a one-dimensional vector, and then the one-dimensional vector is mapped to the probability distribution of each tissue type through a fully connected layer, and finally the final tissue type of the first target tissue and its confidence are output by the Softmax function.

[0071] For example, when the first target feature map After global average pooling, a vector with a length of 64 is obtained, which is input into a fully connected layer. The fully connected layer maps the vector to N tissue types (such as liver, spleen, kidney, etc.), and outputs a probability distribution indicating the probability that the first target tissue belongs to a certain tissue type. Finally, the final tissue type of the first target tissue is determined according to the highest probability value in the probability distribution.

[0072] In some embodiments, the first ultrasound image data can also be analyzed by an image processing algorithm to determine the tissue type of the first target tissue. This process can be implemented by a deep learning model, or by traditional image feature extraction methods such as morphological features, texture features, Histogram of Oriented Gradient (HOG), tissue spontaneous displacement, etc., or by a combination of the two methods for fusion recognition. Based on this, the above process specifically includes:

[0073] inputting the first black-and-white ultrasound image data and / or the first color ultrasound image data into a fifth network model, and predicting the tissue type of the first target tissue based on the first black-and-white ultrasound image data and / or the first color ultrasound image data by the fifth network model; or,

[0074] performing feature analysis on the first black-and-white ultrasound image data and / or the first color ultrasound image data to obtain first data features; determining a tissue type matched with the first data features based on one or more preconfigured second matching relationships, and determining the matched tissue type as the tissue type of the first target tissue, wherein the second matching relationship is a matching relationship between data features and tissue types; or,

[0075] performing feature analysis on the first black-and-white ultrasound image data and / or the first color ultrasound image data to obtain first data features; inputting the first data features into a sixth network model, and predicting the tissue type of the first target tissue based on the first data features by the sixth network model.

[0076] Specifically, the first ultrasound image data includes first black-and-white ultrasound image data and / or first color ultrasound image data, and the tissue type recognition of the first ultrasound image data can include three ways:

[0077] The first way is to input the first black-and-white ultrasound image data and / or the first color ultrasound image data into a fifth network model to predict the tissue type of the first target tissue. The fifth network model is a classification model based on a convolutional neural network or other deep learning architecture, which is specifically used to identify different biological tissue types. The fifth network model learns the feature expression form of various tissues in images by training a large amount of labeled ultrasound image data, and can achieve accurate tissue type prediction.

[0078] In the second mode, the first data features including morphological features and / or texture features and / or blood flow direction features and / or blood vessel morphology features of the first target tissue can be extracted by performing feature analysis on the first black-and-white ultrasound image data and / or the first color ultrasound image data. The morphological features refer to macro-structural features such as shapes, boundaries, contours, etc. of the tissue in the image; the texture features refer to micro-structural features such as gray scale distribution, contrast, roughness, etc. inside the tissue; the blood flow direction features refer to information such as flow direction, velocity distribution, etc. of the blood flow in the tissue; and the blood vessel morphology features refer to information such as orientation, density, branching, etc. of the blood vessels in the tissue.

[0079] After obtaining the first data features, the first data features can be compared with one or more groups of second matching relationships pre-configured to determine the tissue type matched with the first data features, and the matched tissue type is determined as the tissue type of the first target tissue. The second matching relationship is a group of pre-defined mapping rules for corresponding the first data features with a specific tissue type. The second matching relationship can be an empirical rule set artificially or a probability model trained by a machine learning method. For example, for a certain specific texture feature, if the value of the texture feature falls within a certain range, it is matched to liver tissue according to the second matching relationship; and if the value of the texture feature falls within another range, it is matched to spleen tissue according to the second matching relationship.

[0080] In the third mode, the first data features including morphological features and / or texture features and / or blood flow direction features and / or blood vessel morphology features of the first target tissue can be extracted by performing feature analysis on the first black-and-white ultrasound image data and / or the first color ultrasound image data, and the first data features are input into a sixth network model for tissue type prediction of the first target tissue. The sixth network model is a classification model based on a convolutional neural network or other deep learning architecture, which can receive morphological features, texture features, blood flow direction features, blood vessel morphology features, etc. as input and output corresponding tissue type prediction results. Compared with the fifth network model, the sixth network model focuses more on processing non-image raw data, i.e., the sixth network model is based on the extracted features for classification.

[0081] Step 106: automatically determining a first elastography strategy based on the tissue type of the first target tissue, the first elastography strategy including at least one of a first vibration control strategy, a first emission scanning strategy or a first imaging processing method.

[0082] The first elastography strategy is an elastography strategy matched with the tissue type of the first target tissue, and different tissue types match different elastography strategies.

[0083] In the embodiments of the present application, the first elastography strategy refers to a set of control parameters and technical methods used in the elastography process according to the tissue type of the first target tissue, including at least one of the first vibration control strategy, the first emission scanning strategy, and the first imaging processing method. The vibration control strategy determines the frequency, amplitude, period, and waveform of the vibration signal emitted by the probe; the emission scanning strategy determines the emission mode, center frequency, and scanning frame rate of the ultrasonic beam; and the imaging processing method determines how to process the collected ultrasonic echo signal to generate an elastographic image.

[0084] Specifically, the first vibration control strategy refers to adjusting the probe vibration parameters (such as frequency, amplitude, period, and waveform) according to the characteristics of the first target tissue to excite shear waves suitable for the current tissue type. For example, in liver tissue, the device can use a 50Hz frequency, 1mm amplitude sine signal as the vibration source; in the spleen tissue, the device can switch to a 100Hz frequency, 0.5mm amplitude sine signal. The first vibration control strategy can improve the excitation efficiency of shear waves and ensure stable shear wave signals in different tissue types.

[0085] The first emission scanning strategy refers to adjusting the emission parameters (such as center frequency, emission frame rate, and emission waveform) of the ultrasonic wave according to the first target tissue. For example, in liver tissue, the device can use an 8000Hz frame rate ultrasonic wave scan; while in the spleen tissue, the device can switch to a 10000Hz frame rate. The flexible adjustment of the first emission scanning strategy helps to improve the detection accuracy of shear waves and improve the quality of elastographic images.

[0086] The first imaging processing method refers to the algorithm used to generate a transient elastographic image according to shear wave propagation information. Common imaging processing methods include TOF algorithm, cross-correlation method, MLE algorithm, etc. Different imaging processing methods are suitable for different tissue types and shear wave propagation characteristics. For example, in liver tissue, the device can use the TOF algorithm; while in kidney tissue, the MLE algorithm can be used. The selection of different imaging processing methods can optimize image contrast and resolution and improve diagnostic accuracy.

[0087] Among them, a plurality of elastography strategies are preset in the transient elastography device, and a default strategy configuration is assigned to each tissue type. When the current scanning area is identified as a certain tissue type, the system will automatically call the corresponding elastography strategy of the tissue type. For example, for liver tissue, the system may select a single-cycle sinusoidal signal with a frequency of 50 Hz and an amplitude of 1 mm as the vibration control strategy, and cooperate with an ultrasonic emission scanning strategy with a frame rate of 8000 Hz and a TOF algorithm as the imaging processing method; while for the spleen tissue, the system may select a single-cycle sinusoidal signal with a frequency of 100 Hz and an amplitude of 0.5 mm, an emission scanning strategy with a frame rate of 10000 Hz, and a maximum likelihood estimation (Maximum Likelihood Estimate, MLE) algorithm as the imaging processing method.

[0088] In some embodiments, one or more sets of first matching relationships between tissue types and elastography strategies can also be preconfigured. Specifically, step 106 includes:

[0089] Based on one or more sets of first matching relationships, a first elastography strategy matching the tissue type of the first target tissue is determined.

[0090] Here, a set of rules can be pre-set in the transient elastography device to correspond a specific type of tissue type to a corresponding elastography strategy to form one or more sets of first matching relationships. For example, liver tissue may correspond to a vibration signal with a frequency of 50 Hz and an amplitude of 1 mm; while spleen tissue may correspond to a vibration signal with a frequency of 100 Hz and an amplitude of 0.5 mm. Among them, these matching relationships can be set by artificial experience, or can be obtained by deep learning algorithm training, and can be adjusted or expanded according to different clinical needs.

[0091] Due to the differences in density, hardness, blood flow state and other factors of different tissues, the excitation parameters and image processing methods used in elastography will also be different. Based on this, one or more sets of first matching relationships between a plurality of preconfigured tissue types and elastography strategies can be used to automatically match the most suitable first elastography strategy in real time according to the tissue type of the first target tissue. Once the first elastography strategy is selected, it will be immediately applied to subsequent transient elastography operations, including vibration control operation, ultrasonic emission scanning operation and image processing operation, to improve the measurement accuracy and examination efficiency of different tissue types.

[0092] It should be noted that the determination of the elastography strategy not only depends on the tissue type itself, but also considers other auxiliary information, such as the size of the region of interest (ROI), the probe pressure value, the user's physiological parameters, etc. The size of the ROI region, the probe pressure value, the user's physiological parameters, etc. can be used as a screening condition to further optimize the accuracy of strategy matching.

[0093] In some embodiments, the first vibration control strategy includes one or more of: a first vibration amplitude, a first vibration period, a first vibration waveform, and a first vibration frequency; and / or,

[0094] The first emission scanning strategy includes one or more of: a first emission waveform, a first center frequency, a first emission period, a first emission amplitude, and a first emission scanning frame rate; and / or,

[0095] The first imaging processing method includes one or more of: a first TOF algorithm, a first cross-correlation algorithm, and a first MLE algorithm.

[0096] The first vibration control strategy includes one or more of: a first vibration amplitude, a first vibration period, a first vibration waveform, and a first vibration frequency. The first vibration amplitude represents the maximum displacement of the vibration, usually in millimeters (mm); the first vibration period represents the time required to complete one complete vibration, usually in milliseconds (ms); the first vibration waveform describes the shape of the vibration signal, common ones are sine wave, triangle wave, square wave, etc.; the first vibration frequency represents the number of vibrations per unit time, usually in hertz (Hz).

[0097] The selection of the first vibration amplitude, the first vibration period, the first vibration waveform, and the first vibration frequency needs to be optimized according to the physical characteristics of the tissue to be measured and the actual requirements. By reasonably configuring the vibration control method, the effective excitation of the shear wave can be ensured, and the sensitivity and accuracy of the elastography can also be improved.

[0098] The first emission scanning strategy includes one or more of: a first emission waveform, a first center frequency, a first emission period, a first emission amplitude, and a first emission scanning frame rate. The first emission waveform determines the shape of the emission signal, such as sine wave, pulse wave, etc.; the first center frequency refers to the main operating frequency of the emission signal, usually in megahertz (MHz); the first emission period represents the time interval between two consecutive emissions; the first emission amplitude represents the intensity of the emission signal, usually in volts (V); the first emission scanning frame rate represents the number of scans completed per second, affecting the update speed and dynamic response capability of the image.

[0099] The first transmission waveform, the first center frequency, the first transmission period, the first transmission amplitude, and the first transmission scan frame rate jointly determine the performance in collecting shear wave data. By reasonably configuring the transmission scanning mode, the quality of the shear wave propagation image can be improved, thereby improving the resolution and stability of the elastic imaging.

[0100] The first imaging processing method includes one or more of a first TOF algorithm, a first cross-correlation algorithm, and a first MLE algorithm. The first TOF algorithm estimates the propagation speed of the shear wave by calculating the time required for the shear wave to propagate from the excitation point to the receiving point; the first cross-correlation algorithm determines the propagation distance of the shear wave by comparing the similarity of signals at different time points; and the first MLE algorithm estimates the propagation parameters of the shear wave by maximizing the likelihood function based on a statistical model.

[0101] The first TOF algorithm, the first cross-correlation algorithm, and the first MLE algorithm are suitable for different application scenarios and tissue types. By appropriately configuring the imaging processing method, the calculation accuracy of the shear wave propagation speed can be effectively improved, thereby enhancing the reliability and applicability of the elastic imaging.

[0102] By subdividing the first elastic imaging strategy into a first vibration control strategy, a first transmission scanning strategy, and a first imaging processing method, and defining and parameterizing each method in detail, the measurement requirements of different tissue types can be met, the instantaneous elastic imaging process can be finely controlled, and the accuracy and consistency of the measurement results can be improved.

[0103] Step 107: Based on the first elastic imaging strategy, the instantaneous elastic ultrasound probe is controlled to apply mechanical vibration to the first target tissue to excite the propagation of shear waves in the first target tissue, and the propagation of the shear waves is detected by the instantaneous elastic ultrasound probe to perform instantaneous elastic imaging on the first target tissue, thereby obtaining an instantaneous elastic image of the first target tissue.

[0104] In the embodiments of the present application, the instantaneous elastic imaging refers to the process of shear wave excitation, detection, and imaging of the first target tissue based on the first elastic imaging strategy. Specifically, the instantaneous elastic imaging device will control the instantaneous elastic ultrasound probe to apply transient mechanical vibration of a specific frequency and amplitude to the first target tissue according to the determined first elastic imaging strategy, thereby exciting the propagation of shear waves in the first target tissue. The propagation speed of the shear wave is closely related to the hardness of the first target tissue. Subsequently, the propagation of the shear wave is detected by the instantaneous elastic ultrasound probe, the elastic properties of the first target tissue can be quantitatively evaluated, and a specific imaging algorithm (such as TOF, cross-correlation, MLE, etc.) is used to perform instantaneous elastic imaging on the first target tissue, thereby obtaining an instantaneous elastic image of the first target tissue.

[0105] The instantaneous elasticity image is an image reflecting the hardness distribution of the target tissue, and the color or gray value of the instantaneous elasticity image represents the elasticity degree of the target tissue at different positions. For example, in liver elasticity imaging, the red area on the image indicates that the tissue is harder, and the green or blue area indicates that the tissue is softer. Therefore, the health status of the target tissue can be evaluated according to the color distribution of the instantaneous elasticity image, such as judging whether there is liver fibrosis or cirrhosis and the like.

[0106] For example, when the tissue type of the first target tissue is identified as kidney, a triangular wave vibration signal with a frequency of 100 Hz and an amplitude of 1 mm can be selected, and the instantaneous elasticity ultrasonic probe is controlled to apply the transient mechanical vibration with the above frequency and amplitude to the first target tissue according to the triangular wave vibration signal with the frequency of 100 Hz and the amplitude of 1 mm, and the propagation path and speed of the shear wave are monitored by high-frame-rate emission scanning, and finally an instantaneous elasticity image reflecting the hardness distribution of the kidney is generated.

[0107] It should be noted that after obtaining the instantaneous elasticity image of the first target tissue, the instantaneous elasticity image of the first target tissue is frozen and saved in response to a second selection instruction of the user; wherein the target feature map includes at least one of morphological features, texture features, gradient features, displacement features, blood flow direction features or blood vessel morphology features.

[0108] In some embodiments, step 107 specifically includes:

[0109] Based on the first vibration control strategy, the instantaneous elasticity ultrasonic probe generates mechanical vibration, which is transmitted to the first target tissue and generates a shear wave at the first target tissue; the hardness or elasticity of the first target tissue is related to the propagation speed of the shear wave in the first target tissue;

[0110] Based on the first emission scanning strategy, the instantaneous elasticity ultrasonic probe emits a first ultrasonic wave to the first target tissue, and receives a first ultrasonic echo signal corresponding to the first ultrasonic wave, the first ultrasonic wave returns to the first ultrasonic echo signal under the action of the shear wave at the first target tissue;

[0111] Based on the first imaging processing method, the first ultrasonic echo signal is subjected to shear wave imaging processing to obtain an elasticity image at the first target tissue.

[0112] Specifically, the transient elastography ultrasound probe is a device with dual functions of vibration excitation and ultrasound detection, which is used to generate and detect shear waves in human tissues. First, a short-time mechanical vibration is applied to the first target tissue by the transient elastography ultrasound probe through a selected first vibration control strategy, so as to excite the shear waves inside the first target tissue. The shear wave is a transverse wave generated in soft tissue by external mechanical vibration, and the propagation speed of the shear wave is closely related to the hardness or elasticity of the tissue. The faster the propagation speed of the shear wave, the harder the tissue; otherwise, the tissue is softer. For example, in liver tissue, when the propagation speed of the shear wave is about 1.5 m / s, it indicates that the liver tissue is in a mild fibrosis state; and when the propagation speed of the shear wave exceeds 3.0 m / s, it indicates that the liver tissue may be in a moderate to severe fibrosis state.

[0113] Secondly, the transient elastography ultrasound probe emits a first ultrasonic wave of a specific frequency and frame rate to the first target tissue according to a selected first emission scanning strategy, and receives a first ultrasonic echo signal modulated by the shear wave. The first ultrasonic wave refers to a high-frequency sound wave used to detect the propagation of the shear wave. Since the shear wave will affect the propagation path and reflection behavior of the first ultrasonic wave in the first target tissue, when the first ultrasonic wave encounters the shear wave region, the reflection signal (first ultrasonic echo signal) of the first ultrasonic wave will contain relevant information of the shear wave. Through the collection and analysis of these echo signals, the propagation information of the first ultrasonic wave can be further extracted. It should be noted that the parameter setting in the first emission scanning strategy needs to be matched with the frequency range of the shear wave, so as to ensure that the parameter setting can effectively capture the echo changes caused by the shear wave. In addition, the first emission scanning strategy should also consider factors such as the depth and thickness of the tissue, so as to optimize the penetration ability and resolution of the ultrasonic wave.

[0114] Then, the first ultrasonic echo signal is analyzed by a selected first imaging processing method to extract the propagation information of the shear wave, so as to convert the propagation information of the shear wave into a visualized elasticity image. Specifically, the first ultrasonic echo signal can be subjected to shear wave imaging processing based on the first imaging processing method to obtain a displacement map of the shear wave at the first target tissue, and then the propagation speed of the shear wave at the first target tissue is determined based on the displacement map of the shear wave at the first target tissue, and the hardness or elasticity of the first target tissue is determined based on the propagation speed of the shear wave at the first target tissue.

[0115] It should be noted that under the action of the shear wave, the micro displacement in the first target tissue will cause a change in the first ultrasound echo signal. By continuously collecting multiple frames of echo signals, the propagation path and speed of the shear wave can be tracked, and thus an elasticity distribution map of the first target tissue can be constructed. For example, in liver tissue, a high-frame-rate ultrasound scan can be used to capture the rapid propagation process of the shear wave to generate a high-resolution elasticity image.

[0116] In some embodiments, the above method further comprises:

[0117] When the target object is switched from the first target tissue to the second target tissue, performing real-time ultrasound scanning on the second target tissue based on the transient elastography ultrasound probe to obtain second ultrasound image data about the second target tissue;

[0118] Extracting a second target feature map in the second ultrasound image data, and comparing the first target feature map and the second target feature map, if the first target feature map and the second target feature map are inconsistent, stopping performing real-time ultrasound scanning on the current target tissue and prompting the user to select other tissues for ultrasound scanning; if the first target feature map and the second target feature map are inconsistent, determining the tissue type of the second target tissue based on the second target feature map; Figure One

[0119] The second elasticity imaging strategy is determined based on the tissue type of the second target tissue, and the first elasticity imaging strategy includes at least one of a second vibration control strategy, a second emission scanning strategy, or a second imaging processing method; wherein the second elasticity imaging strategy is an elasticity imaging strategy matched with the tissue type of the second target tissue; the second elasticity imaging strategy is different from the first elasticity imaging strategy;

[0120] Based on the second elasticity imaging strategy, the transient elastography ultrasound probe is controlled to apply mechanical vibration to the second target tissue to excite the propagation of the shear wave in the second target tissue, and the propagation of the shear wave is detected by the transient elastography ultrasound probe to perform transient elastography imaging on the second target tissue, thereby obtaining a transient elasticity image of the second target tissue.

[0121] Here, the target object refers to a patient or a specific part thereof who is currently receiving an ultrasound examination; the first target tissue and the second target tissue respectively refer to anatomical structures targeted in different scanning stages, for example, switching from the liver to the spleen.

[0122] ​Specifically, when the real-time ultrasound scan is unfrozen and executed again in response to the third selection instruction of the user, the target object is switched from the first target tissue to a second target tissue, and the second target tissue can be subjected to real-time ultrasound scanning by the transient elastography probe in the transient elastography device to continuously collect image information without interrupting the operation process, to ensure instant feedback of the tissue state and obtain second ultrasound image data of the second target tissue, wherein the second ultrasound image data is multi-modal data, and each modal data is a single-channel image or a multi-channel image; then the specific process of extracting the first target feature map in the first ultrasound image data is referred to for extracting a second target feature map in the second ultrasound image data, and the first target feature map and the second target feature map are compared for similarity, if they are the same or the similarity is greater than a similarity threshold, it indicates that the current target tissue is the same as or highly similar to the first target tissue, the real-time ultrasound scanning for the current target tissue is stopped and the user is prompted to reselect other tissues for ultrasound scanning, if they are different (the similarity is less than the similarity threshold), the specific process of determining the tissue type of the first target tissue based on the first target feature map is referred to for determining the tissue type of the second target tissue based on the second target feature map.

[0123] Secondly, a second elastography strategy matched with the tissue type of the second target tissue is selected from one or more groups of preconfigured first matching relationships according to the tissue type of the second target tissue, the second elastography strategy including at least one of a second vibration control strategy, a second emission scanning strategy or a second imaging processing method; then under the selected second elastography strategy, the transient elastography probe is controlled to apply mechanical vibration to the second target tissue to excite the propagation of shear waves in the second target tissue, and the propagation of the shear waves is detected by the transient elastography probe to perform transient elastography on the second target tissue, to obtain a transient elastography image of the second target tissue.

[0124] Wherein, the second elastography strategy has at least one of the following differences relative to the first elastography strategy: the first vibration control strategy and the second vibration control strategy are different; the first emission scanning strategy and the second emission scanning strategy are different; the first imaging processing method and the second imaging processing method are different.

[0125] In the above manner, real-time adaptive adjustment function can be realized, the elasticity measurement of the second target tissue can be completed without manual intervention, and high-quality transient elastography images can be output, to ensure the continuity and consistency of the measurement process, thereby realizing efficient and accurate elasticity evaluation, and further improving the automation level of generating transient elastography images.

[0126] In the technical solution of the embodiment of the application, the transient elastography device enters the transient elastography mode in response to a first selection instruction of a user; in the transient elastography mode: real-time ultrasound scanning is performed on a first target tissue of a target object based on a transient elastography probe to obtain first ultrasound image data about the first target tissue; the first ultrasound image data is processed to obtain a plurality of first feature maps; the plurality of first feature maps are spliced in the channel dimension and compressed in the channel dimension to obtain a second feature map; hierarchical features in the second feature map are extracted to obtain a plurality of third feature maps, and a fourth feature map in the plurality of third feature maps is extracted based on a self-attention mechanism; the fourth feature map is up-sampled layer by layer to the original resolution to obtain a first target feature map; the tissue type of the first target tissue is determined based on the first target feature map; a first elastography strategy is automatically determined based on the tissue type of the first target tissue, the first elastography strategy including at least one of a first vibration control strategy, a first emission scanning strategy, or a first imaging processing method, wherein the first elastography strategy is an elastography strategy matched with the tissue type of the first target tissue; the transient elastography probe is controlled to apply mechanical vibration to the first target tissue to excite the propagation of shear waves in the first target tissue based on the first elastography strategy, and the propagation of the shear waves is detected by the transient elastography probe to perform transient elastography on the first target tissue to obtain a transient elastography image of the first target tissue. In this way, first, the transient elastography mode is entered in response to the user instruction, then the image data of the target tissue is collected by the ultrasound probe, the tissue type is identified by using the multi-modal feature fusion and the self-attention mechanism, and the elastography strategy is automatically adjusted according to the identification result, and finally the personalized elastography for different tissues is realized. In this way, on the one hand, by splicing and compressing the channel of the image data of the target tissue, extracting multi-level features and combining the self-attention mechanism, the identification accuracy of the tissue type can be improved; on the other hand, by presetting multiple elastography strategies and automatically switching according to the identification result, the adaptability of the transient elastography device to different tissue types can be improved, thereby avoiding the operation complexity and time waste caused by manual switching, and improving the measurement efficiency and accuracy.

[0127] The embodiment of the application further provides a transient elastography method, which is applied to a transient elastography device.

[0128] Figure 2 is a flowchart of the transient elastography method provided by the embodiment of the application Figure Two As shown in Figure 2 , the method comprises the following steps:

[0129] Step 201: entering the transient elastography mode in response to a first selection instruction of a user.

[0130] Here, the specific implementation steps can refer to the above method flow, and will not be repeated here.

[0131] Step 202: In the transient elastography mode: based on the transient elasticity ultrasound probe, performing real-time ultrasound scanning on the first target tissue of the target object to obtain first ultrasound image data about the first target tissue.

[0132] Here, the specific implementation steps can refer to the above method flow, and will not be repeated here.

[0133] In some embodiments, step 202 specifically includes:

[0134] Based on the transient elasticity ultrasound probe, sending second ultrasound waves to the first target tissue, and receiving second ultrasound echo signals returned by the first target tissue for the second ultrasound waves;

[0135] Signal processing the second ultrasound echo signals to obtain the first ultrasound image data.

[0136] Here, the specific implementation steps can refer to the above method flow, and will not be repeated here.

[0137] Step 203: determining the tissue type of the first target tissue based on the first ultrasound image data.

[0138] Here, the specific implementation steps can refer to the above method flow, and will not be repeated here.

[0139] Step 204: automatically determining a first elastography strategy based on the tissue type of the first target tissue.

[0140] Wherein, the first elastography strategy is an elastography strategy matched with the tissue type of the first target tissue.

[0141] Here, the specific implementation steps can refer to the above method flow, and will not be repeated here.

[0142] In some embodiments, one or more groups of first matching relationships can be pre-configured, and the first matching relationship is a matching relationship between the tissue type and the elastography strategy; based on this, step 204 specifically includes:

[0143] Based on one or more groups of first matching relationships, determining a first elastography strategy matched with the tissue type of the first target tissue.

[0144] Here, the specific implementation steps can refer to the above method flow, and will not be repeated here.

[0145] In some embodiments, the first vibration control strategy includes one or more of the following: first vibration amplitude, first vibration period, first vibration waveform, and first vibration frequency; and / or,

[0146] The first emission scanning strategy comprises one or more of the following: a first emission waveform, a first center frequency, a first emission period, a first emission amplitude, and a first emission scanning frame rate; and / or,

[0147] The first imaging processing method comprises one or more of the following: a first time-of-flight (TOF) algorithm, a first cross-correlation algorithm, and a first maximum likelihood estimation (MLE) algorithm.

[0148] Step 205: based on the first elastography strategy, controlling the transient elastography probe to apply mechanical vibration to the first target tissue to excite the shear wave to propagate in the first target tissue, and detecting the propagation of the shear wave by the transient elastography probe to perform transient elastography on the first target tissue, to obtain a transient elastography image of the first target tissue.

[0149] Here, the specific implementation steps can refer to the above method flow, and will not be repeated here.

[0150] In some embodiments, step 205 specifically comprises:

[0151] controlling the transient elastography probe to generate mechanical vibration based on the first vibration control strategy, the mechanical vibration being transmitted to the first target tissue and generating a shear wave at the first target tissue; the hardness or elasticity of the first target tissue being related to the propagation speed of the shear wave in the first target tissue;

[0152] controlling the transient elastography probe to emit a first ultrasonic wave to the first target tissue based on the first emission scanning strategy, and receiving a first ultrasonic echo signal corresponding to the first ultrasonic wave, the first ultrasonic wave returning the first ultrasonic echo signal under the action of a shear wave at the first target tissue;

[0153] based on the first imaging processing method, performing shear wave imaging processing on the first ultrasonic echo signal to obtain a transient elastography image at the first target tissue.

[0154] Here, the specific implementation steps can refer to the above method flow, and will not be repeated here.

[0155] In some embodiments, the above method further comprises:

[0156] when the target object is switched from the first target tissue to a second target tissue, performing real-time ultrasonic scanning on the second target tissue based on the transient elastography probe to obtain second ultrasonic image data about the second target tissue;

[0157] determine a tissue type of the second target tissue based on the second ultrasound image data; and determine a second elastography strategy based on the tissue type of the second target tissue; wherein the second elastography strategy is an elastography strategy matched with the tissue type of the second target tissue; and the second elastography strategy is different from the first elastography strategy;

[0158] control the transient elastography ultrasound probe to apply mechanical vibration to the second target tissue to excite the shear wave to propagate in the second target tissue based on the second elastography strategy, and detect the propagation of the shear wave by the transient elastography ultrasound probe to perform transient elastography on the second target tissue, to obtain a transient elastography image of the second target tissue.

[0159] Here, the specific implementation steps can refer to the above method flow, and will not be repeated here.

[0160] In the technical solution of the embodiment of the application, the transient elastography imaging device enters the transient elastography imaging mode in response to the first selection instruction of the user; in the transient elastography imaging mode: based on the transient elastography ultrasound probe performing real-time ultrasound scanning on the first target tissue of the target object, first ultrasound image data about the first target tissue is obtained; based on the first ultrasound image data, the tissue type of the first target tissue is determined; based on the tissue type of the first target tissue, the first elastography strategy is automatically determined, wherein the first elastography strategy is an elastography strategy matched with the tissue type of the first target tissue; based on the first elastography strategy, the transient elastography ultrasound probe is controlled to apply mechanical vibration to the first target tissue to excite the shear wave to propagate in the first target tissue, and the propagation of the shear wave is detected by the transient elastography ultrasound probe to perform transient elastography on the first target tissue, to obtain a transient elastography image of the first target tissue. In this way, by obtaining the first ultrasound image data and identifying the corresponding tissue type, and then selecting the first elastography strategy matched with the identified result, since the physiological characteristics of different tissue types are different, the elastography strategy matched with the characteristics can more accurately control the mechanical vibration and the ultrasound scanning parameters, thereby improving the accuracy of shear wave detection and the quality and reliability of the elastography image. Compared with the mode switching method manually performed by the user in the related art, the method realizes adaptive adjustment, improves the operation efficiency and the consistency of measurement.

[0161] The application also provides a transient elastography imaging system capable of real-time adaptive adjustment, as shown in Figure 3As shown is a framework schematic diagram of the real-time self-adaptive adjustment instant elasticity imaging system provided by the embodiment of the application. Wherein, the system performs imaging and display after acquiring the section scan data by the system execution acquisition preparation imaging processing link, while delivering the current ROI setting, probe pre-pressing and other instant elasticity starting necessary information. The system identifies the target tissue according to the current section scan data, and then selectively adjusts the vibration control strategy, emission scanning strategy and imaging processing method of the instant elasticity imaging in real time according to the tissue identification result. After the system completes the adaptive adjustment, the instant elasticity measurement can be started.

[0162] As can be seen from the above, the most important link in the application is the instant elasticity real-time self-adaptive adjustment module based on target tissue identification. The instant elasticity real-time self-adaptive adjustment module is realized based on real-time target tissue identification. The real-time target tissue identification can be realized by a deep learning algorithm, or by a traditional feature analysis algorithm such as morphology, texture, directional gradient, and tissue spontaneous displacement, or by a combination of two or more of the above algorithms, such as first obtaining an intermediate result by morphology feature and texture feature analysis, and then inputting the intermediate result into a deep learning algorithm to obtain the final target identification result.

[0163] Subsequently, the system automatically selects the corresponding instant elasticity vibration control strategy and ultrasonic signal emission scanning mode and imaging processing method according to the identification result. The system presets multiple basic vibration control strategies (amplitude, period, waveform, frequency, etc.), ultrasonic signal emission scanning modes (emission waveform, center frequency, period, amplitude, scanning frame frequency, etc.) and imaging processing methods (TOF algorithm, cross-correlation algorithm, MLE algorithm, etc.), sets the corresponding default strategy for different tissue types, and automatically switches to the corresponding strategy after the system identifies the corresponding tissue.

[0164] Next, the embodiments of the application will be described in detail to further illustrate the implementation process and technical details of the application.

[0165] As Figure 4 As shown is a flowchart of the target tissue identification process of the real-time self-adaptive adjustment instant elasticity imaging system provided by the embodiment of the application, which includes the following steps:

[0166] Step 401: input the B-mode ultrasound image, color blood flow, tissue displacement and other multi-modal data into the self-attention-based multi-modal feature extraction network architecture.

[0167] Wherein, the self-attention-based multi-modal feature extraction network architecture can output the tissue category label (such as liver, spleen, kidney, pancreas, blood vessel, muscle, lesion, etc.) of the target tissue and its confidence distribution. The network architecture receives multi-modal input, each modality is a single-channel image or a multi-channel image, and the final input image size is 256x256x8.

[0168] Step 402: Perform 3×3 convolution, batch normalization, and activation processing (such as ReLU) on each modality of data to obtain feature maps. .

[0169] Step 403: Put all feature maps Obtained by splicing along the channel dimension .

[0170] Step 404: Convolve using 1×1 convolution The channels are compressed to 64, forming a unified feature map. .

[0171] Step 405: Use ResNet34 as a convolutional encoder to process the feature map. Extract features from multiple levels and output four-layer features. .

[0172] Among them, four layers of features The output dimensions are 128×128×64, 64×64×128, 32×32×256, and 16×16×512, respectively.

[0173] Step 406: Flatten the deepest feature map E4 into a sequence of size 256×512, perform linear mapping and positional encoding, and input it into a six-layer module consisting of a multi-head self-attention mechanism and a feedforward fully connected network to obtain the result. .

[0174] Step 407: Use transposed convolution through the decoder to... Upsample layer by layer to the original resolution, and during this process, after aligning the feature channels of the skip connections with the corresponding coding layers, perform weighted summation to obtain the final feature map. .

[0175] Step 408: Process the feature map Global averaging—fully connected layer—Softmax function, outputting the final target recognition result.

[0176] like Figure 5 The diagram illustrates the process of performing elastic imaging using the real-time adaptive instantaneous elastic imaging system provided in this embodiment of the application. Figure One The process includes the following steps:

[0177] Step 501: Enter the acquisition preparation stage, and use B-mode ultrasound to image and acquire data of the current scanning section.

[0178] Step 502: Apply a deep learning algorithm to identify the current section as liver tissue.

[0179] Step 503: According to the identification result, a single-cycle sinusoidal signal with a frequency of 50Hz and an amplitude of 1mm is used as the vibration signal.

[0180] Step 504: An ultrasonic emission scanning beam with a frame frequency of 8000Hz is used to detect tissue shear waves.

[0181] Step 505: A TOF algorithm is used for shear wave propagation imaging processing.

[0182] Step 506: After the liver measurement is completed, the user changes the scan site to the spleen, and enters the acquisition preparation stage again.

[0183] Step 507: The current scanning tissue is identified as spleen tissue by applying a self-attention-based multi-modal feature extraction network to B-mode ultrasound image section data.

[0184] Step 508: The vibration frequency of the transient elastography is automatically adjusted to a single-cycle sinusoidal signal with a frequency of 100Hz and an amplitude of 0.5mm.

[0185] Step 509: The ultrasonic emission scanning configuration is adjusted to a frame frequency of 10000Hz.

[0186] As Figure 6 shown is a flowchart of the transient elastography system provided by the embodiments of the present application for performing real-time adaptive adjustment of the transient elastography system. Figure Two The flowchart includes the following steps:

[0187] Step 601: Enter the acquisition preparation stage, and the system performs imaging through B-mode ultrasound images + Color images. The system obtains data of the current scanning section.

[0188] Step 602: Extract morphological features, texture features, blood flow direction and vessel morphological features from B-mode ultrasound image data and Color data.

[0189] Step 603: Identify the current section as kidney tissue.

[0190] Step 604: Automatically switch the transient elasticity configuration to a single-cycle triangular signal with a frequency of 100Hz and an amplitude of 1mm.

[0191] Step 605: Configure the ultrasonic emission scanning frame frequency to 8000Hz.

[0192] Step 606: Perform imaging processing of the shear wave propagation graph using a TOF algorithm and segment the data during the imaging process.

[0193] Through the above embodiments, the system can adjust the corresponding instantaneous elasticity measurement parameters in real time according to different tissue types, thereby improving the accuracy and applicability of elasticity measurement. Moreover, by introducing a real-time adaptive adjustment mechanism based on target tissue identification, automatic switching between different tissue types is realized, significantly improving the intelligent level and operation efficiency of the system.

[0194] The embodiments of the present application also provide a real-time adaptive instantaneous elasticity imaging device, as shown in the accompanying drawings: Figure 7 The structure of the real-time adaptive instantaneous elasticity imaging device provided by the embodiments of the present application is shown in the accompanying drawings, and the device comprises a transmitting-receiving circuit 701, an instantaneous elasticity ultrasound probe 702, a vibration excitation device 703 and a processor 704; wherein,

[0195] The transmitting-receiving circuit 701 is configured to control the instantaneous elasticity ultrasound probe 702 to perform real-time ultrasonic scanning.

[0196] The vibration excitation device 703 is configured to control the instantaneous elasticity ultrasound probe 702 to apply mechanical vibration to excite the shear wave to propagate in the target tissue.

[0197] The transmitting-receiving circuit 701 is further configured to detect the propagation of the shear wave through the instantaneous elasticity ultrasound probe 702.

[0198] The processor 704 is configured to execute the steps of any one of the above instantaneous elasticity imaging methods.

[0199] The transmitting-receiving circuit 701 is a core control module in the elasticity imaging device, responsible for coordinating the instantaneous elasticity ultrasound probe 702 to complete the transmission of ultrasonic waves and the acquisition of echo signals. The working process of the transmitting-receiving circuit 701 includes: generating a driving signal according to the preset parameters, so that the instantaneous elasticity ultrasound probe 702 transmits ultrasonic waves with specific frequency and amplitude; after the transmission is completed, the transmitting-receiving circuit 701 starts the receiving mode, and the transmitting-receiving circuit 701 receives the ultrasonic echo signals reflected by the tissue. In addition, the transmitting-receiving circuit 701 also has signal conditioning functions such as amplification, filtering and digitization processing, to ensure that the subsequent processor 704 can obtain high-quality data.

[0200] There is a clear physical connection between the transmit-receive circuit 701 and the transient elastography ultrasound probe 702. The output end of the transmit-receive circuit 701 is connected to the excitation end of the transient elastography ultrasound probe 702, and the receiving end of the transmit-receive circuit 701 is connected to the sensor array of the transient elastography ultrasound probe 702. In actual operation, the transmit-receive circuit 701 will dynamically adjust the transmission parameters according to system instructions. For example, when scanning liver tissue, the transmit-receive circuit 701 uses a vibration frequency of 50 Hz; when scanning the spleen, the transmit-receive circuit 701 switches to a vibration frequency of 100 Hz to adapt to the mechanical properties of different tissues.

[0201] The transient elastography ultrasound probe 702 is a transducer device that converts electrical signals into mechanical vibrations (i.e., ultrasonic waves) and vice versa. The transient elastography ultrasound probe 702 usually contains multiple piezoelectric crystal elements arranged in a linear or phased array form to achieve multi-angle scanning. In the transmission phase, after receiving the driving signal provided by the transmit-receive circuit 701, the transient elastography ultrasound probe 702 generates high-frequency vibrations and emits ultrasonic waves into human tissue; in the receiving phase, when the ultrasonic waves are reflected back after passing through the tissue, the transient elastography ultrasound probe 702 converts the received echo signal into an electrical signal for subsequent processing.

[0202] The transient elastography ultrasound probe 702 and the transmit-receive circuit 701 form a complete signal link. The transmit circuit controls the vibration parameters of the transient elastography ultrasound probe 702, while the transient elastography ultrasound probe 702 serves as the input and output interface for signals. In actual application, the performance of the transient elastography ultrasound probe 702 directly affects the image quality and elasticity measurement accuracy. For example, in high-resolution scanning, using a high-frequency transient elastography ultrasound probe 702 can obtain clearer boundary information, while in deep tissue measurement, a low-frequency transient elastography ultrasound probe 702 may be selected to enhance the penetration ability.

[0203] The vibration excitation device 703 is a hardware component that can provide external excitation signals to the transient elastography ultrasound probe 702. The role of the vibration excitation device 703 is to make the transient elastography ultrasound probe 702 produce controllable mechanical vibrations. The vibration excitation device 703 usually includes a vibration generator and a driving circuit, which are used to control the vibration behavior of the transient elastography ultrasound probe 702 according to the set frequency, amplitude, and waveform parameters. For example, in a transient elastography imaging system, the vibration excitation device 703 can be configured to output a single-cycle sine wave, a triangle wave, or other specific waveforms to excite shear waves within the tissue. Among them, the parameters of mechanical vibration (such as frequency, amplitude, and waveform) have a direct impact on the generation and propagation of shear waves, so under different tissue types, the vibration excitation device 703 needs to adjust these parameters according to actual needs to obtain the best measurement effect.

[0204] The processor 704 is the computing core of the elastography device, responsible for processing, analyzing and imaging the collected ultrasound data. The main tasks of the processor 704 include: the processor 704 receives raw data from the transmit-receive circuit 701 and the transient elastography probe; the processor 704 runs elastography algorithms such as TOF algorithm, cross-correlation method, MLE algorithm, etc.; when the elastography device identifies the tissue type, the processor 704 dynamically adjusts the transient elastography strategy according to the identification result, and finally generates an image reflecting the elasticity of the tissue according to the transient elastography strategy.

[0205] In addition, the processor 704 not only undertakes image processing functions, but also has logical judgment and decision-making capabilities, and can make adaptive adjustments based on real-time data. When the tissue switching is detected, the processor 704 will call the transient elastography strategy that matches the current tissue type for adaptive adjustment, including setting specific vibration frequency, scanning frame rate and imaging algorithm, to optimize the measurement results.

[0206] It should be understood that the processor of the embodiments of the present application can be an integrated circuit chip with signal processing capability. In the implementation process, each step of the method embodiments described above can be completed by integrated logic circuits or instructions in the form of software in the processor. The processor described above can be a general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), a ready-to-program gate array (Field Programmable Gate Array, FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component. The disclosed methods, steps and logic block diagrams in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module can be located in a random access memory, a flash memory, a read-only memory, a programmable read-only memory, an electrically erasable programmable memory, a register or other mature storage medium in the art. The storage medium is located in the memory, and the processor reads the information in the memory, and combines the hardware to complete the steps of the above method.

[0207] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0208] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0209] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the above-described device embodiments are only schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or imaging connection between the units shown or discussed can be indirect coupling or imaging connection through some interface, device or unit, which can be electrical, mechanical or other form.

[0210] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or they can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0211] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit.

[0212] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0213] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application.

Claims

1. A method of transient elastography, characterized in that, The method is applied to a transient elastography device, and the method comprises: in response to a first selection instruction of a user, entering a transient elastography mode; in the transient elastography mode: performing real-time ultrasound scanning on a first target tissue of a target object based on a transient elastography probe to obtain first ultrasound image data about the first target tissue; processing the first ultrasound image data to obtain a plurality of first feature maps, splicing the plurality of first feature maps in a channel dimension and performing channel compression to obtain a second feature map, extracting hierarchical features in the second feature map to obtain a plurality of third feature maps, and extracting a fourth feature map in the plurality of third feature maps based on a self-attention mechanism; and performing layer-by-layer up-sampling on the fourth feature map to an original resolution to obtain a first target feature map; determining a tissue type of the first target tissue based on the first target feature map; automatically determining a first elastography strategy based on the tissue type of the first target tissue, the first elastography strategy comprising at least one of a first vibration control strategy, a first emission scanning strategy, or a first imaging processing method, wherein the first elastography strategy is an elastography strategy matched with the tissue type of the first target tissue; controlling the transient elastography probe to apply mechanical vibration to the first target tissue to excite shear waves to propagate in the first target tissue based on the first elastography strategy, and detecting the propagation of the shear waves by the transient elastography probe to perform transient elastography on the first target tissue to obtain a transient elastography image of the first target tissue.

2. The method of claim 1, wherein, The method further comprises: preconfiguring one or more groups of first matching relationships, the first matching relationship being a matching relationship between a tissue type and an elastography strategy; the determination of the first elastography strategy based on the tissue type of the first target tissue comprises: determining the first elastography strategy matched with the tissue type of the first target tissue based on the one or more groups of first matching relationships.

3. The method of claim 1, wherein, the control of the transient elastography probe to apply mechanical vibration to the first target tissue to excite shear waves to propagate in the first target tissue based on the first elastography strategy, and the detection of the propagation of the shear waves by the transient elastography probe to perform transient elastography on the first target tissue to obtain a transient elastography image of the first target tissue, comprises: controlling the transient elastography probe to generate mechanical vibration based on the first vibration control strategy, the mechanical vibration being transmitted to the first target tissue and generating shear waves at the first target tissue; the hardness or elasticity of the first target tissue being related to the propagation speed of the shear waves in the first target tissue; controlling the transient elastography probe to emit first ultrasound waves to the first target tissue based on the first emission scanning strategy, and receiving first ultrasound echo signals corresponding to the first ultrasound waves, the first ultrasound waves returning the first ultrasound echo signals under the action of shear waves at the first target tissue; Performing shear wave imaging processing on the first ultrasonic echo signal based on the first imaging processing method to obtain an elasticity image at the first target tissue.

4. The method of claim 1, wherein, The first vibration control strategy includes one or more of the following: a first vibration amplitude, a first vibration period, a first vibration waveform, and a first vibration frequency; and / or, The first emission scanning strategy includes one or more of the following: a first emission waveform, a first center frequency, a first emission period, a first emission amplitude, and a first emission scanning frame rate; and / or, The first imaging processing method includes one or more of the following: a first time-of-flight (TOF) algorithm, a first cross-correlation algorithm, and a first maximum likelihood estimation (MLE) algorithm.

5. The method of claim 1, wherein, The first target tissue is scanned in real time by the transient elastography ultrasonic probe to obtain first ultrasonic image data of the first target tissue, including: The transient elastography ultrasonic probe sends a second ultrasonic wave to the first target tissue and receives a second ultrasonic echo signal returned by the first target tissue in response to the second ultrasonic wave; The second ultrasonic echo signal is processed to obtain the first ultrasonic image data.

6. The method of claim 1, wherein, The first ultrasonic image data includes first black-and-white ultrasonic image data and / or first color ultrasonic image data; The first ultrasonic image data is processed to obtain a plurality of first feature maps; The plurality of first feature maps are concatenated in the channel dimension and compressed in the channel dimension to obtain a second feature map, including: The first black-and-white ultrasonic image data and / or the first color ultrasonic image data are input into a first network model, and the first black-and-white ultrasonic image data and / or the first color ultrasonic image data are convolved, batch normalized, and activated by the first network model to obtain the plurality of first feature maps; The plurality of first feature maps are concatenated in the channel dimension to obtain a first concatenated feature map, and the first concatenated feature map is compressed in the channel dimension to obtain the second feature map; The hierarchical features in the second feature map are extracted to obtain a plurality of third feature maps, and a fourth feature map is extracted from the plurality of third feature maps based on a self-attention mechanism; the fourth feature map is upsampled layer by layer to the original resolution to obtain a first target feature map, including: The second feature map is input into a second network model, and hierarchical features in the second feature map are extracted by the second network model to obtain the plurality of third feature maps; and the plurality of third feature maps are input into a third network model, and a fourth feature map is extracted from the plurality of third feature maps based on a self-attention mechanism by the third network model; The fourth feature map is input into a fourth network model, and the fourth feature map is upsampled layer by layer to the original resolution by the fourth network model using transposed convolution, and is weighted and added after aligning the feature channels of the plurality of third feature maps at the jump connection to obtain the first target feature map.

7. The method of claim 1, wherein, The first ultrasonic image data includes first black-and-white ultrasonic image data and / or first color ultrasonic image data; The method further comprises: inputting the first black-and-white ultrasound image data and / or the first color ultrasound image data into a fifth network model, and predicting the tissue type of the first target tissue based on the first black-and-white ultrasound image data and / or the first color ultrasound image data through the fifth network model; or, performing feature analysis on the first black-and-white ultrasound image data and / or the first color ultrasound image data to obtain first data features; determining a tissue type matched with the first data features based on one or more groups of preconfigured second matching relationships, and determining the matched tissue type as the tissue type of the first target tissue, wherein the second matching relationship is a matching relationship between data features and tissue types; or, performing feature analysis on the first black-and-white ultrasound image data and / or the first color ultrasound image data to obtain first data features; inputting the first data features into a sixth network model, and predicting the tissue type of the first target tissue based on the first data features through the sixth network model.

8. The method according to any one of claims 1 to 7, characterized in that, The method further comprises: when the target object switches from the first target tissue to a second target tissue, performing real-time ultrasound scanning on the second target tissue based on the transient elastography ultrasound probe to obtain second ultrasound image data about the second target tissue; extracting a second target feature map in the second ultrasound image data, and comparing the first target feature map with the second target feature map; if the first target feature map is consistent with the second target feature map, stopping performing real-time ultrasound scanning on the current target tissue and prompting a user to reselect other tissues for ultrasound scanning; if the first target feature map is inconsistent with the second target feature map, determining a tissue type of the second target tissue based on the second target feature map; determining a second elastography strategy based on the tissue type of the second target tissue, the second elastography strategy comprising at least one of a second vibration control strategy, a second emission scanning strategy or a second imaging processing method; wherein the second elastography strategy is an elastography strategy matched with the tissue type of the second target tissue; the second elastography strategy is different from the first elastography strategy; controlling the transient elastography ultrasound probe to apply mechanical vibration to the second target tissue to excite shear waves to propagate in the second target tissue based on the second elastography strategy, and detecting the propagation of the shear waves through the transient elastography ultrasound probe to perform transient elastography on the second target tissue, thereby obtaining a transient elastography image of the second target tissue.

9. The method of claim 8, wherein, The second elastography strategy has at least one of the following differences relative to the first elastography strategy: the first vibration control strategy is different from the second vibration control strategy; the first emission scanning strategy is different from the second emission scanning strategy; the first imaging processing method is different from the second imaging processing method.

10. A transient elastography device characterized by, The transient elastography device comprises a transmitting-receiving circuit, a transient elastography ultrasound probe, a vibration excitation device and a processor; wherein, The transmitting-receiving circuit is configured to control the transient elastography probe to perform real-time ultrasonic scanning. The vibration excitation device is configured to control the transient elastography probe to apply mechanical vibration to excite the shear wave to propagate in the target tissue. The transmitting-receiving circuit is further configured to detect the propagation of the shear wave by the transient elastography probe. The processor is configured to perform the steps of the transient elastography method according to any one of claims 1 to 9.

Images