CN115736980B - Emergency ultrasonic imaging system and construction method thereof - Google Patents

Abstract

The present invention discloses an ultrasonic imaging system for emergency use, the system comprising a user terminal, a wireless ultrasonic probe, a computing platform, and an electrocardiogram monitor; the user terminal is installed with corresponding ultrasonic information display software, which can display ultrasonic images and receive and display auxiliary diagnosis results transmitted by the computing platform; the wireless ultrasonic probe is used to perform ultrasonic scanning on the part to be inspected, and send the image file formed by the ultrasonic scanning to the user terminal via wireless transmission; the computing platform is deployed on a cloud server, and is used to detect and calculate the ultrasonic video uploaded by the user terminal and/or the electrocardiogram signal transmitted by the electrocardiogram monitor, obtain auxiliary information, and transmit it back to the user terminal for display; the electrocardiogram monitor has at least one lead, which can wirelessly transmit the electrocardiogram data to the computing platform. The present invention also proposes the application of the above system in rapid emergency ultrasonic examination of abdominal injury or abnormal abdominal bleeding.

Description

An emergency ultrasonic imaging system and construction method thereof

Technical Field

The present invention belongs to the technical field of ultrasonic imaging, and relates to an ultrasonic imaging system for emergency use and a construction method thereof. By using a wireless ultrasonic probe for ultrasonic imaging, combined with the help of a cloud computing platform and an electrocardiogram monitor, emergency inspection of abdominal injuries and abnormal abdominal bleeding can be achieved.

Background Art

There are a series of important organs in the abdominal cavity of the human body, such as the liver and spleen. These organs are very fragile and are easily damaged by external gravity impact, causing bleeding. If not treated in time, it is very easy to cause the patient's death. Therefore, early diagnosis of abdominal organ injury plays a key role in timely treatment of patients and improving survival rate.

For abdominal injuries, a variety of examinations can be performed in hospitals, such as ultrasound examinations and CT examinations. However, in pre-hospital emergency rescue situations and remote areas, only ultrasound can achieve on-site diagnosis, and it is low-priced and easy to operate, so it can be used as the preferred method for emergency diagnosis. In addition, monitoring ECG changes can assist in determining the condition of abdominal injuries. For example, ECG will be abnormal when the patient is bleeding.

Combining ultrasound with ECG information to examine abdominal injuries and abnormal bleeding, the current examinations rely heavily on the experience of doctors, and the lack of experienced doctors has increasingly failed to meet the growing needs of examinations. In addition, thanks to the rapid development of artificial intelligence in recent years, many diseases have begun to use artificial intelligence methods for examinations. The use of computer cloud processing technology trained with large amounts of data can fully achieve the results of artificial intelligence diagnosis for imaging diagnosis of abdominal injuries. Unfortunately, however, artificial intelligence detection applications for combined ultrasound and ECG diagnosis are still scarce.

Summary of the invention

The purpose of the present invention is to propose an emergency ultrasonic imaging system and a construction method thereof in response to the problems existing in existing ultrasonic inspection equipment and inspections, which can realize emergency inspection and prevention of abdominal injuries and abnormal abdominal bleeding, and detect diseases as early as possible and treat them in time.

The technical solution of the present invention is:

The present invention provides an ultrasonic imaging system for emergency use, which mainly includes a user terminal, a wireless ultrasonic probe, a computing platform and an electrocardiogram monitor, wherein:

The user terminal is a portable user terminal that is matched with the wireless ultrasound probe and installed with the corresponding ultrasound information display software, including a mobile phone or a tablet, etc. It is used to display two-dimensional ultrasound images, integrate continuous two-dimensional ultrasound images into ultrasound videos, and receive and display the auxiliary diagnosis results transmitted by the computing platform. The user terminal may further include: an ultrasound image display processing module, a wireless ultrasound probe communication module, and a computing platform result receiving module.

The ultrasonic image display processing module is used to display the ultrasonic image of the abdominal cavity generated by the wireless ultrasonic probe. In addition, it can also integrate the ultrasonic image frame by frame into an ultrasonic video under the control of the user, and provide the ultrasonic video information required by the computing platform result receiving module. Finally, it can also display the diagnosis results transmitted back from the computing platform.

The wireless ultrasound probe communication module is used to realize the communication between the user terminal and the wireless ultrasound probe, and receives the ultrasound detection image information sent back by the wireless ultrasound probe in the form of wireless wifi.

The computing platform result receiving module is used to realize the communication between the user terminal and the computing platform. It transmits the abdominal ultrasound video information synthesized by the ultrasound image display processing module to the computing platform through the form of wireless Internet network information transmission, and obtains the auxiliary information of abdominal injury or abnormal abdominal bleeding returned from the computing platform after the computing platform completes the calculation. The auxiliary information can be used for auxiliary diagnosis of abdominal injury or abnormal abdominal bleeding.

The wireless ultrasound probe is a low-frequency convex array abdominal wireless ultrasound probe, which is used to perform corresponding ultrasound scanning on the relevant parts of the abdominal cavity to be examined, and send the image file formed by the ultrasound scanning to the user terminal connected to the wireless ultrasound probe through wireless wifi transmission. The wireless ultrasound probe includes: an ultrasound imaging module, and a first communication module that communicates with the user terminal.

The ultrasonic imaging module is used to perform ultrasonic scanning on the relevant parts to be detected and run the imaging algorithm built into the probe chip according to the relevant data files obtained by the scanning to generate ultrasonic images; the process of converting the data file into an ultrasonic image is shown in Figure 6, which is as follows: the wireless ultrasonic probe is connected to the integrated circuit A for pulse transmission and echo reception switch. The integrated circuit B used for echo signal processing and image encoding and decoding sends a control signal to control the switch of the integrated circuit A to switch between the transmission signal driving the wireless ultrasonic probe and the echo input signal of the wireless ultrasonic probe to achieve beam synthesis. The echo signal enters the integrated circuit B to complete the analog front-end amplification and analog-to-digital signal conversion, and the digital signal is further encoded into a grayscale image. The microcontroller C is responsible for controlling the WIFI module or USB to communicate with the user terminal; the FLASH memory is used to store the control program of the microcontroller C and the control program of the WiFi module. The structures including integrated circuits A, B, microcontroller C, etc. in Figure 6 are all installed in the wireless ultrasonic probe.

The first communication module that communicates with the user terminal is used to transmit the generated ultrasonic detection image file to the user terminal via wireless wifi transmission.

The computing platform is deployed on a cloud server and has a series of ultrasound image and ECG signal processing models. It can detect and calculate the ultrasound video uploaded by the user terminal and/or the ECG signal transmitted by the ECG monitor, obtain auxiliary diagnosis results, and transmit them back to the user terminal for display in the form of wireless Internet network information transmission. Its information processing diagram is shown in Figure 7. The computing platform mainly includes: an ultrasound image intelligent detection module based on manifold compression, a multimodal fusion intelligent detection module based on graph neural network, a second communication module for communicating with the user terminal, and a communication module for communicating with the ECG monitor.

The ultrasonic image intelligent detection module based on manifold compression is used for the detection of simple abdominal injuries, and the multimodal fusion intelligent detection module based on graph neural network is used for the detection of abnormal abdominal bleeding.

The ultrasonic image intelligent detection module based on manifold compression uses an ultrasonic image intelligent algorithm based on manifold compression. Its input is an ultrasonic video of the abdomen, and its output is the diagnosis result of abdominal organ damage. Since the dimensionality of ultrasonic images is extremely high at this stage, the present invention proposes a targeted high-dimensional medical deep learning method based on ultrasonic images. By reducing the dimensionality of ultrasonic images, low-dimensional key information of ultrasonic images including clustering and classification can be obtained, and based on this information, auxiliary information on whether the abdominal organs are damaged can be obtained in an interpretable manner, which can be used to assist in the diagnosis results. Its main principles are:

1. Introduce the ideas of information theory and compression into the basic theory of deep learning, optimize the deep learning model in medical image processing based on the maximum coding rate attenuation, and use the rate-distortion theory to measure the effective measurement of the code length. In the present invention, the input ultrasound image is compressed, and the rate-distortion theory is used to measure the effective measurement of the code length.

The rate-distortion theory is a theory that uses the basic viewpoints and methods of information theory to study data compression problems. It is mainly used to illustrate the value that the minimum description code rate can be set to if a certain distortion limit is met. Under the guidance of this theory, the present invention proposes an effective measure of the overall spatial coding rate.

The maximum coding rate attenuation is an optimization algorithm for the model. The goal is to use this algorithm to achieve the optimal construction of the model. The optimization principle is to minimize the coding length required after the data of mixed categories is divided, that is, to allow samples of different structures to be close to each other.

The calculation method of the overall spatial coding rate of ultrasound images is as follows:

Where R is the overall spatial coding rate, det is the determinant calculation, I is the unit matrix, ∈ ² is the square error allowed for encoding each vector, Z is the feature of all data, N is the number of vectors contained in the feature Z, and d is the dimension of the data.

2. Use manifold compression, a goal with a solid mathematical foundation, as the final optimization function. Since the storage overhead should be minimal after the new samples are classified into appropriate categories, the optimal representation efficiency can be obtained through correct classification. The decision boundary of this algorithm is closer to the manifold structure characteristics of the data itself. The optimization objectives are as follows:

Where ^Rc is the sum of the spatial coding rates of each category, det() is the determinant calculation, tr() is the trace of the matrix, I is the identity matrix, ∈ ² is the square error allowed for encoding each vector, Z is the feature of all data, N is the number of vectors contained in the feature Z, d is the dimension of the data, _Aj is the distribution matrix of the true label of the jth category, and k is the total number of categories.

Using this optimization method, the manifold compression network model structure diagram shown in Figure 1 is generated, where Z _l is the input of the lth layer, I is the unit matrix, which means no transformation is performed. is the linear transformation layer of each classification, σ is the activation function layer, η is the proportional coefficient, is the normalization layer, and Z _l+1 is the input of the l+1th layer.

Subsequently, by performing gradient descent optimization on the above optimization targets, the present invention uses this principle to construct an interpretable deep ultrasound medical image processing network, by performing gradient descent optimization on the input ultrasound image and its corresponding label value. After the training is completed, it is deployed to the cloud to construct an ultrasound image intelligent detection module based on manifold compression.

The multimodal fusion intelligent detection module based on graph neural network takes abdominal ultrasound video and electrocardiogram signal as model input, and uses the intelligent algorithm of multimodal fusion to provide auxiliary information on whether there is abnormal bleeding in the abdomen. The main principle is:

1. Under the premise of maintaining the local relationship structure of ultrasound image frames and electrocardiogram neighborhoods, the multimodal high-dimensional data is nonlinearly transformed layer by layer into a lower-dimensional feature space in an unsupervised learning manner, and the new neighborhood structure relationship of the latent space features is learned at the same time, ultimately achieving the purpose of manifold learning; in fact, the overall model is shown in Figure 2.

2. The basic GTAE network based on unsupervised learning adopts the standard AutoEncoder (AE) structure, which is divided into two parts: encoder and decoder. The data of each layer is a topological map with attributes.

3. In the encoding stage, the encoder layer node information X ^l is first used to obtain the graph structure matrix D ^l through the KNN graph, and then the next layer node information X ^l+1 is calculated under the conditions of graph structure D ^l , node information X ^l and trainable parameters W ^l .

Where A ^l is the adjacency matrix, σ is the nonlinear activation function, is the normalized adjacency matrix, which is a diagonal matrix. The calculation method of the diagonal elements is in I is the identity matrix, and F _gcn represents the encoding operation.

Finally, we get the latent variable Z = ^XL .

4. In the decoding stage, the logic is similar to that of AE. The multilayer perceptron directly maps the latent variables back to the original multimodal high-dimensional space.

in, Indicates the mapped node information. Represents the graph structure matrix after mapping, represents the trainable parameters after mapping, and F represents the decoding operation.

5. The above GTAE model can thoroughly improve the multimodal signal manifold relationship structure of ECG signals and ultrasound images. It generates a graph of ECG information and ultrasound image information, uses the adjacency matrix of the graph to describe the relationship between the two, and performs dimensionality reduction transformation. In actual operation, the input is the multimodal signal of ECG signal and ultrasound and its corresponding adjacency matrix table, and the intelligent diagnosis of intravascular volume is made through the graph neural network formed by a series of GCN layers and RELU activation function layers stacked based on the GTAE network layer principle as shown in Figure 3.

The model can then judge the bleeding situation based on the output results and make a diagnosis of whether the bleeding is abnormal.

The second communication module that communicates with the user terminal is used to receive video information from the user terminal in the form of wireless Internet network information transmission, and return auxiliary information of abdominal injury or abnormal abdominal bleeding to the user terminal for subsequent auxiliary diagnosis.

The communication module that communicates with the ECG monitor is used to receive the ECG data transmitted by the ECG monitor in the form of wireless Internet network information transmission.

The ECG monitor is used to provide ECG information in real time during diagnosis. In the present invention, it is required to have at least one lead and the ability to wirelessly transmit ECG data to a computing platform.

The present invention also provides a method for intelligent detection of ultrasonic images based on manifold compression using the above ultrasonic imaging system, which can be used for auxiliary detection of abdominal cavity injuries. The method comprises the following steps:

Step 1: Transmit the ultrasound image information generated by the ultrasound wireless probe to the user terminal via wireless wifi;

Step 2: The user terminal integrates the ultrasound images frame by frame into an ultrasound video, and the user terminal transmits the ultrasound video to the computing platform via the wireless Internet;

Step 3: The computing platform processes the ultrasound video, obtains auxiliary information and transmits it to the user terminal via the wireless Internet. The user terminal displays the two-dimensional ultrasound image information and the auxiliary information transmitted by the computing platform.

The present invention also provides a method for performing multimodal fusion intelligent detection based on graph neural network using the above-mentioned ultrasound imaging system, which can be used for auxiliary detection of abnormal abdominal bleeding. The method comprises the following steps:

Step (1), transmitting the ultrasound image information generated by the ultrasound wireless probe to the user terminal via wireless wifi;

Step (2), the user terminal integrates the ultrasound images frame by frame into an ultrasound video, and the user terminal transmits the ultrasound video to the computing platform via the wireless Internet;

Step (3), the ECG information collected by the ECG monitor is transmitted to the computing platform via the wireless Internet;

Step (4): The computing platform processes the ultrasound video and ECG information to obtain auxiliary information, and transmits the auxiliary information to the user terminal via the wireless Internet. The user terminal displays the two-dimensional ultrasound image information and the auxiliary information transmitted by the computing platform.

The present invention also provides application of the ultrasonic imaging system in rapid emergency ultrasonic examination of abdominal injury or abnormal abdominal bleeding.

The beneficial effects of the present invention include: the present invention can realize rapid emergency ultrasonic examination of abdominal cavity injury and abnormal abdominal cavity bleeding.

Compared with the existing examination methods for abdominal injuries and abnormal abdominal bleeding, the existing methods are limited to using only one of ultrasound images or electrocardiograms for diagnosis, and the accuracy of most diagnoses needs to rely on the doctor's experience to ensure. The present invention creatively proposes a series of algorithms that integrate ultrasound images, electrocardiogram information and deep learning to diagnose diseases.

The present invention has low requirements on the application environment and can realize very convenient emergency detection. In addition, the present invention uses the auxiliary diagnosis of the computing platform to reduce the ability requirements of the detection personnel, so that ordinary medical staff can also use this equipment to perform relevant ultrasonic emergency examinations.

The promotion of the present invention can effectively save the lives of people with abdominal injuries and abnormal abdominal bleeding caused by accidents, can help them discover the cause of the disease as early as possible, and take relevant treatment plans in time to ensure that they are treated within the golden treatment time.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a manifold compression network module structure diagram of an ultrasonic image intelligent detection module based on manifold compression in the computing platform of the present invention.

Figure 2 is a structural diagram of the GTAE relationship extraction model of the multimodal fusion intelligent detection module based on graph neural network in the computing platform of the present invention.

Figure 3 is a graph neural network structure diagram of the multimodal fusion intelligent detection module based on graph neural network in the computing platform of the present invention.

FIG. 4 is a schematic diagram of the system structure of an abdominal injury detection example of the present invention.

FIG. 5 is a schematic diagram of the system structure of an abnormal abdominal bleeding detection example of the present invention.

FIG. 6 is a diagram of a wireless ultrasound image information processing component of the present invention.

FIG. 7 is an information processing diagram of the computing platform of the present invention.

DETAILED DESCRIPTION

The present invention is further described in detail with reference to the following specific examples and drawings. The process, conditions, experimental methods, etc. for implementing the present invention, except for the contents specifically mentioned below, are all common knowledge and common common sense in the art and are not particularly limited by the present invention.

The present invention provides an emergency ultrasonic imaging system, which mainly includes a user terminal, a wireless ultrasonic probe, a computing platform, and an electrocardiogram monitor;

The user terminal is matched with a wireless ultrasound probe and is installed with corresponding ultrasound information display software, which can display ultrasound images, integrate continuous two-dimensional ultrasound images into ultrasound videos, and receive and display auxiliary diagnosis results transmitted by the computing platform;

The wireless ultrasound probe is used to perform ultrasound scanning on the part to be examined, and to send the image file formed by the ultrasound scanning to the user terminal connected to the wireless ultrasound probe via wireless transmission;

The computing platform is deployed on a cloud server and is used to detect and calculate the ultrasound video uploaded by the user terminal and/or the electrocardiogram signal transmitted by the electrocardiogram monitor, obtain auxiliary diagnosis results, and transmit them back to the user terminal for display;

The ECG monitor has at least one lead, which can provide ECG information in real time and transmit the ECG data wirelessly to a computing platform.

The emergency ultrasonic imaging system of the present invention can be used for ultrasonic imaging operations under emergency conditions.

Example 1

Embodiment 1 is shown in FIG4 , which is an ultrasound imaging system for abdominal injury detection, and is mainly divided into three parts, namely, an abdominal ultrasound wireless probe, a user terminal, and a computing platform, wherein the abdominal ultrasound wireless probe is used for emergency ultrasound scanning of the abdominal area and for generating ultrasound images; the user terminal is used for displaying the generated ultrasound images, integrating continuous two-dimensional ultrasound images into ultrasound videos and displaying the diagnosis results transmitted back from the computing platform; the computing platform is used for calculating the detected ultrasound image videos and returning the diagnosis results to the user terminal.

The specific implementation process of Example 1 is as follows: the user first connects the wireless ultrasound probe to the user terminal through a wireless wifi signal. After the connection is stable, the user selects the abdominal injury detection mode, and the user uses the probe to scan the abdomen-related detection area of the person to be inspected. The wireless ultrasound probe is connected to the integrated circuit A for pulse transmission and echo reception switch. The integrated circuit B for echo signal processing and image encoding and decoding sends a control signal to control the switch of the integrated circuit A to switch between the transmission signal driving the wireless ultrasound probe and the echo input signal of the wireless ultrasound probe to achieve beam synthesis. The echo signal enters the integrated circuit B to complete the analog front-end amplification and analog-to-digital signal conversion, and the digital signal is further encoded into a grayscale image; the probe will send the abdominal ultrasound image in Figure 4 to the user terminal, and the user terminal will display the real-time ultrasound image. Then the user terminal will send the abdominal ultrasound video information to the computing platform. After the computing platform is calculated by the ultrasonic image intelligent detection module based on manifold compression, the diagnosis result is returned to the user terminal and displayed by the user terminal.

Example 2

Embodiment 2 As shown in FIG5 , it is an ultrasound imaging system for detecting abnormal abdominal bleeding, which is mainly divided into four parts, namely, an abdominal ultrasound wireless probe, a user terminal, an electrocardiogram monitor, and a computing platform, wherein the abdominal ultrasound wireless probe is used to perform emergency ultrasound scanning of the abdominal area to be detected and to generate ultrasound images; the electrocardiogram monitor performs feature monitoring, obtains electrocardiogram information, and transmits the electrocardiogram signal to the computing platform; the computing platform receives the abdominal ultrasound video information transmitted by the user terminal, receives the electrocardiogram information transmitted by the electrocardiogram monitor, starts the multimodal fusion intelligent detection module based on the graph neural network, performs corresponding intelligent auxiliary diagnosis of abnormal abdominal bleeding, and sends the diagnosis result back to the user terminal. The user terminal is used to display ultrasound images, integrate continuous two-dimensional ultrasound images into ultrasound videos, and the results of the auxiliary diagnosis of the computing platform.

The specific implementation process of Example 2 is as follows: the user first connects the wireless ultrasound probe to the user terminal through a wireless wifi signal. After the connection is stable, the user selects the abnormal abdominal bleeding detection mode. After that, the user uses the probe to scan the area to be inspected of the person to be inspected. The probe will send the abdominal ultrasound image in Figure 5 to the user terminal. The user terminal will send the generated abdominal ultrasound video data to the computing platform, and the ECG monitor will send the generated ECG information to the computing platform. The computing platform integrates the abdominal ultrasound video information and ECG information, and after calculation by the multimodal fusion intelligent detection module based on the graph neural network, it will return the auxiliary diagnosis results to the user terminal, and the user terminal will display the ultrasound image and its corresponding auxiliary diagnosis results.

The protection content of the present invention is not limited to the above embodiments. Without departing from the spirit and scope of the present invention, changes and advantages that can be thought of by those skilled in the art are included in the present invention and are protected by the attached claims.

Claims (7)

1. An emergency ultrasonic imaging system is characterized by comprising a user terminal, a wireless ultrasonic probe, a computing platform and an electrocardiograph monitor;

The user terminal is matched with the wireless ultrasonic probe, is provided with corresponding ultrasonic information display software, and can display ultrasonic pictures, integrate continuous two-dimensional ultrasonic pictures into an ultrasonic video, receive auxiliary diagnosis results transmitted by the computing platform and display the auxiliary diagnosis results;

The wireless ultrasonic probe is used for carrying out ultrasonic scanning on a part to be inspected, and sending an image file formed by ultrasonic scanning to the user terminal connected with the wireless ultrasonic probe in a wireless transmission mode;

the computing platform is deployed on the cloud server and is used for detecting and computing the ultrasonic video uploaded by the user terminal and the electrocardiosignals transmitted by the electrocardiograph monitor to obtain an auxiliary diagnosis result, and transmitting the auxiliary diagnosis result back to the user terminal for display;

the computing platform further includes:

The system comprises an ultrasonic image intelligent detection module based on manifold compression, a multi-mode fusion intelligent detection module based on a graph neural network, a second communication module communicated with a user terminal and a communication module communicated with an electrocardiograph monitor;

The intelligent ultrasonic image detection module based on manifold compression applies an intelligent ultrasonic image algorithm based on manifold compression to obtain low-dimensional key information including clustering and classification after dimension reduction by dimension reduction of an ultrasonic image, wherein an ultrasonic video of an abdomen is input, and auxiliary information for damage of an abdominal organ is output;

the model of the multi-modal fusion intelligent detection module based on the graph neural network inputs an abdomen ultrasonic video and an electrocardiosignal, and auxiliary information of whether the abdomen is abnormally bleeding or not is given by utilizing an intelligent algorithm of multi-modal fusion;

the second communication module is used for receiving video information transmitted by the user terminal in a wireless internet network information transmission mode and transmitting auxiliary information of abdominal cavity injury or abdominal cavity abnormal bleeding back to the user terminal;

The communication module is used for receiving the electrocardiograph data transmitted by the electrocardiograph monitor in the form of wireless internet network information transmission;

in the intelligent ultrasonic image detection process based on manifold compression, an input ultrasonic image is compressed, and the effective measurement of the coding length is measured by using a rate distortion theory;

The overall spatial encoding rate of the ultrasound image is calculated as follows:

Wherein R is the total spatial coding rate, det is determinant calculation, I is an identity matrix, epsilon ² is the square error allowed by coding each vector, Z is the characteristic of all data, N is the number of vectors contained in the characteristic Z, and d is the dimension of the data;

The objectives of optimization with manifold compression are:

Wherein R ^c is the sum of the spatial encoding rates of the various classifications, det () is a determinant calculation, tr () is the trace of a matrix, I is an identity matrix, E ² is the square error allowed by encoding each vector, Z is the characteristics of all data, N is the number of vectors contained in the characteristic Z, d is the dimension of the data, A _j is the distribution matrix of the j-th class real labels, and k is the total number of classifications;

In the multi-mode fusion intelligent detection process based on the image neural network, on the premise of maintaining the local relation structure of an ultrasonic image frame and an electrocardiogram neighborhood, multi-mode high-dimensional data are subjected to non-linear transformation layer by layer to a lower-dimensional feature space in an unsupervised learning mode, and meanwhile, new neighborhood structural relations of hidden space features are learned;

The non-supervised learning network is divided into an encoder and a decoder, in the encoding stage, firstly, a graph structure matrix D ^l is obtained through a KNN graph by using encoder layer node information X ^l, and then, under the conditions of a graph structure D ^l, node information X ^l and trainable parameters W ^l, next layer node information X ^l+1 is calculated; the calculation method comprises the following steps:

Wherein A ^l is an adjacency matrix, sigma is a nonlinear activation function, The normalized adjacent matrix is a diagonal matrix, and the calculation method of the elements on the diagonal is as followsWherein the method comprises the steps ofI is a unit array, and F _gcn represents a coding operation;

Finally obtaining hidden variable Z=X ^L;

in the decoding stage, the multi-layer perceptron directly maps hidden variables back to the original multi-mode high-dimensional space, and the calculation method is as follows:

wherein, Representing the information of the node after the mapping,Representing the mapped matrix of the graph structure,Representing the mapped trainable parameters, F representing the decoding operation;

the electrocardiograph monitor is provided with at least one lead, and can provide electrocardiograph information in real time and wirelessly transmit electrocardiograph data to the computing platform.

2. The ultrasound imaging system of claim 1, wherein the user terminal further comprises an ultrasound image display processing module, a wireless ultrasound probe communication module, a computing platform result receiving module;

The ultrasonic image display processing module is used for displaying ultrasonic pictures of the abdominal cavity generated by the wireless ultrasonic probe, integrating the ultrasonic pictures into ultrasonic videos frame by frame under the control of a user, providing the ultrasonic videos to the calculation platform result receiving module, and sending back diagnostic results from the calculation platform;

The wireless ultrasonic probe communication module is used for realizing communication between the user terminal and the wireless ultrasonic probe, and receiving ultrasonic detection picture information transmitted back by the wireless ultrasonic probe in a wireless wifi mode;

The computing platform result receiving module is used for realizing communication between the user terminal and the computing platform, transmitting the abdominal ultrasonic video information to the computing platform in a wireless internet network information transmission mode, and obtaining auxiliary information of abdominal injury or abdominal abnormal bleeding returned from the computing platform after the computing platform is calculated.

3. The ultrasound imaging system of claim 1, wherein the wireless ultrasound probe is a celiac ultrasound wireless probe in the form of a low frequency convex array;

The structure of the wireless ultrasonic probe further comprises an ultrasonic imaging module and a first communication module communicated with the user terminal;

the ultrasonic imaging module is used for carrying out ultrasonic scanning on a part to be detected, obtaining an ultrasonic imaging data file and generating an ultrasonic picture by utilizing an imaging algorithm built in the probe chip;

The first communication module is used for transmitting the generated ultrasonic detection picture file to the user terminal.

4. The ultrasound imaging system of claim 3, wherein the process of converting the data file into an ultrasound image is as follows: the wireless ultrasonic probe is connected with an integrated circuit A for a pulse transmitting and echo receiving switch; an integrated circuit B for echo signal processing and image encoding and decoding sends a control signal to control a switch of the integrated circuit A to switch between a transmitting signal for driving the wireless ultrasonic probe and an echo input signal of the wireless ultrasonic probe, so that beam synthesis is realized; the echo signal enters an integrated circuit B to finish analog front end amplification and analog-to-digital signal conversion, and the digital signal is further encoded into a gray level image; the microcontroller C is responsible for controlling the WIFI module or the USB to communicate with the user terminal; the FLASH memory is used for storing a control program of the microcontroller C and a control program of the WiFi module; integrated circuit A, B, microcontroller C is mounted in the wireless ultrasound probe.

5. A method of manifold compression based intelligent ultrasound image detection using the ultrasound imaging system of any of claims 1-4, the method comprising the steps of:

Step one, transmitting ultrasonic picture information generated by an ultrasonic wireless probe to a user terminal in a wireless wifi mode;

step two, the user terminal integrates the ultrasonic pictures into ultrasonic videos frame by frame, and the ultrasonic videos are transmitted to the computing platform through the wireless internet by the user terminal;

And thirdly, the computing platform processes the ultrasonic video, obtains auxiliary information and transmits the auxiliary information to the user terminal through the wireless internet, and the user terminal displays the two-dimensional ultrasonic picture information and the auxiliary information transmitted by the computing platform.

6. A method for multi-modal fusion intelligent detection based on a graph neural network using the ultrasound imaging system of any one of claims 1-4, the method comprising the steps of:

Step (1), transmitting ultrasonic picture information generated by an ultrasonic wireless probe to a user terminal in a wireless wifi mode;

step (2), the user terminal integrates the ultrasonic pictures into ultrasonic videos frame by frame, and the ultrasonic videos are transmitted to a computing platform through the wireless internet by the user terminal;

Step (3), the electrocardiograph monitor transmits the electrocardiograph information collected by the electrocardiograph monitor to the computing platform through the wireless internet;

And (4) processing the ultrasonic video and the electrocardiographic information by the computing platform to obtain auxiliary information, transmitting the auxiliary information to the user terminal through the wireless internet, and displaying the two-dimensional ultrasonic picture information and the auxiliary information transmitted by the computing platform by the user terminal.

7. Use of the ultrasound imaging system of any of claims 1-4 in rapid emergency ultrasound inspection of abdominal injuries or abnormal bleeding from the abdominal cavity.