Cardiac compression monitoring method, device, equipment and medium

Abstract

The invention relates to the technical field of heart monitoring, and discloses a cardiac compression monitoring method which comprises the following steps: after in-vitro membrane lung oxygenation is established, monitoring a target object in real time through ultrasonic equipment to obtain an initial heart image; obtaining an image analysis result, inputting the image analysis result into the pressing equipment, and enabling the pressing equipment to execute a pressing operation according to the image analysis result; if the interval time reaches a preset interval period, acquiring an updated heart image of the target object through the ultrasonic equipment; and obtaining an updated analysis result, inputting the updated analysis result into the pressing equipment, and enabling the pressing equipment to execute a pressing operation according to the updated analysis result. According to the invention, through the ultrasonic equipment, real-time monitoring of the target object and acquisition of the heart image are realized, and then detection of whether thrombus exists in the heart image is realized. And through pressing equipment, blood circulation is accelerated, so that the risk of thrombosis is reduced, and the workload of medical staff is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of cardiac monitoring, and particularly to a cardiac compression monitoring method, device, equipment and medium. Background Art

[0002] ECMO (Extracorporeal Membrane Oxygenation) is an extracorporeal circulation technology for cardiac surgery. Its principle is to draw venous blood from the body, oxygenate it through a special artificial cardiopulmonary bypass, and then inject it into the arterial or venous system of the patient, playing a role in partially replacing the heart and lungs and maintaining the oxygenated blood supply to the body's organs and tissues.

[0003] However, since the extracorporeal membrane lung replaces the functions of the heart and lungs, and the blood pump replaces the contraction and relaxation of the heart, the blood in the heart cannot effectively participate in the overall blood circulation, with a large residue, and the blood flow in the heart is slow, making it easy for patients to develop pulmonary artery embolism during the use of ECMO. And thrombus will reduce the efficiency of membrane lung oxygenation, and clinicians mainly need to press the heart multiple times during ECMO to ensure the blood fluidity in the heart part and prevent the formation of thrombus. Summary of the Invention

[0004] The present invention provides a cardiac compression monitoring method, device, equipment and medium to solve the problem in the prior art that clinicians need to frequently press the heart during ECMO treatment to prevent the formation of thrombus.

[0005] A cardiac compression monitoring method includes: After the establishment of extracorporeal membrane oxygenation, use an ultrasonic device to perform real-time monitoring on the target object corresponding to the established extracorporeal membrane oxygenation to obtain an initial cardiac image corresponding to the target object; Obtain an image analysis result corresponding to the initial cardiac image, and input the image analysis result into a pressing device, and make the pressing device perform a pressing operation on the target object according to the image analysis result; If the interval time reaches a preset interval period, collect an updated cardiac image of the target object through the ultrasonic device; Obtain an updated analysis result corresponding to the updated cardiac image, and input the updated analysis result into the pressing device, and make the pressing device perform a pressing operation on the target object according to the updated analysis result.

[0006] A cardiac compression monitoring device includes: An initial image acquisition module, configured to, after the establishment of extracorporeal membrane oxygenation, use an ultrasonic device to perform real-time monitoring on the target object corresponding to the established extracorporeal membrane oxygenation to obtain an initial cardiac image corresponding to the target object; An initial pressing operation module, configured to obtain an image analysis result corresponding to the initial heart image, and input the image analysis result into a pressing device, so that the pressing device performs a pressing operation on the target object according to the image analysis result; An updated image acquisition module, configured to collect an updated heart image of the target object through an ultrasonic device if an interval time reaches a preset interval period; An updated pressing operation module, configured to obtain an updated analysis result corresponding to the updated heart image, and input the updated analysis result into a pressing device, so that the pressing device performs a pressing operation on the target object according to the updated analysis result.

[0007] A computer device includes a memory, a controller, and a computer program stored in the memory and executable on the controller. When the controller executes the computer program, the above-mentioned heart pressing monitoring method is implemented.

[0008] A computer-readable storage medium stores a computer program, and when the computer program is executed by a controller, the above-mentioned heart pressing monitoring method is implemented.

[0009] The heart pressing monitoring method, device, equipment and medium provided by the present invention realize real-time monitoring of a target object through an ultrasonic device, and further realize acquisition of a heart image. Through the image analysis result, detection of whether there is a risk of thrombus formation in the heart image is realized, thereby realizing precise control of the pressing device. Through the pressing device, blood fluidity in the heart part is accelerated, thereby reducing the risk of thrombus formation, and further realizing replacement of medical staff and reducing the workload of medical staff. Description of the Drawings

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments of the present invention will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings without creative efforts.

[0011] Figure 1 is a flowchart of the heart pressing monitoring method in an embodiment of the present invention; Figure 2 is a principle block diagram of the heart pressing monitoring device in an embodiment of the present invention. Detailed Embodiments

[0012] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0013] The present invention provides a method for monitoring cardiac compression. In one embodiment, as Figure 1 shown, its technical solution mainly includes the following steps: S10, after extracorporeal membrane oxygenation (ECMO) is established, use an ultrasonic device to perform real-time monitoring on a target object corresponding to the established ECMO to obtain an initial cardiac image corresponding to the target object.

[0014] It can be understood that extracorporeal membrane oxygenation refers to drawing venous blood in the body out of the body, and after oxygenation through an artificial cardiopulmonary bypass of special materials, injecting it into the arterial or venous system of the patient, playing a role in partial cardiopulmonary replacement and maintaining the oxygenated blood supply of human organs and tissues. An ultrasonic device refers to a series of instruments that work based on the principle of ultrasonic waves. The initial cardiac image refers to the cardiac image of the target object collected by the ultrasonic device.

[0015] Specifically, establish extracorporeal membrane oxygenation for the target object, that is, use extracorporeal membrane to replace the functions of the heart and lungs, and use a blood pump to replace the contraction and relaxation of the heart. After extracorporeal membrane oxygenation is established, use an ultrasonic device to perform real-time monitoring on the target object corresponding to the established extracorporeal membrane oxygenation, that is, obtain a preset position, and place an ultrasonic probe at the preset position to detect the preset position, and collect images of each preset position according to ultrasonic waves, so as to obtain an initial cardiac image of the target object including multiple positions.

[0016] S20, obtain an image analysis result corresponding to the initial cardiac image, and input the image analysis result into a pressing device, so that the pressing device performs a pressing operation on the target object according to the image analysis result.

[0017] It can be understood that the image analysis result refers to the result of analyzing the initial cardiac image and is used to characterize the risk of thrombus formation. The pressing device refers to a device used to press the cardiac part of the target object.

[0018] Specifically, an image recognition model is obtained, and the initial cardiac image is input into the image recognition model. The recognition ability learned by the image recognition model during training is used to identify the risks of sectional images at different positions in the initial cardiac image, so as to output whether there is a risk of thrombus formation in the initial cardiac image and the formation probability, and determine whether there is a risk of thrombus formation and the formation probability as the image analysis result corresponding to the initial cardiac image. Among them, a risk threshold is set. When the recognized formation probability exceeds the risk threshold, it is determined that there is a risk of thrombus formation and the image analysis result of the formation probability is output; when the recognized formation probability does not exceed the risk threshold, it is determined that there is no risk of thrombus formation and the image analysis result of the formation probability is output. Then, the image analysis result is input into the pressing device. By recognizing the image analysis result, it is determined whether there is a risk of thrombus formation. If there is no risk of thrombus formation, the pressing device is not controlled to perform a pressing operation on the cardiac part of the target object. If there is a risk of thrombus formation, the formation probability is obtained, and the corresponding pressing depth, pressing frequency, and pressing time are queried through the formation probability, and the pressing device is controlled to perform a pressing operation with the pressing depth, pressing frequency, and pressing time.

[0019] S30. If the interval time reaches the preset interval period, the updated cardiac image of the target object is collected by the ultrasonic device.

[0020] Understandably, the preset interval period refers to the time interval between two pressing operations set in advance. For example, 1 hour. The interval time refers to the time from the current time to the last pressing operation. The updated cardiac image refers to the image of the heart of the target object collected by the ultrasonic device before a new pressing operation.

[0021] Specifically, record the time of the last pressing operation, and continuously count the interval time from the current time to the last pressing operation. Compare the interval time with the preset interval period. If the interval time reaches the preset interval period, control the ultrasonic probe in the ultrasonic device to re-collect the image at the preset position, so as to obtain the updated cardiac image corresponding to the target object.

[0022] S40. Obtain the updated analysis result corresponding to the updated cardiac image, and input the updated analysis result into the pressing device, and make the pressing device perform a pressing operation on the target object according to the updated analysis result.

[0023] Understandably, the updated analysis result refers to the result of analyzing the updated cardiac image, which is used to represent whether there is a risk of thrombus formation.

[0024] Specifically, an image recognition model is obtained, and the updated cardiac image is input into the image recognition model. The recognition ability learned by the image recognition model during training is used to identify the risks of sectional images at different positions in the updated cardiac image, so as to output whether there is a risk of thrombus formation in the updated cardiac image and the formation probability. And whether there is a risk of thrombus formation and the formation probability are determined as the updated analysis result corresponding to the updated cardiac image. Among them, when the recognized updated formation probability exceeds the risk threshold, it is determined that there is a risk of thrombus formation and the updated analysis result of the updated formation probability is output; when the recognized updated formation probability does not exceed the risk threshold, it is determined that there is no risk of thrombus formation and the updated analysis result of the updated formation probability is output. Then, the updated analysis result is input into the pressing device. By recognizing the updated analysis result, it is determined whether there is a risk of thrombus formation. If there is no risk of thrombus formation, the pressing device is not controlled to perform a pressing operation on the cardiac part of the target object. If there is a risk of thrombus formation, the updated formation probability is obtained, and the corresponding updated pressing depth, updated pressing frequency, and updated pressing time are queried through the updated formation probability, and the pressing device is controlled to perform a pressing operation with the updated pressing depth, updated pressing frequency, and updated pressing time. In another embodiment, the updated cardiac image is transmitted to a preset client, and the updated analysis result feedback by the preset client is received.

[0025] In the embodiment of the present invention, through the ultrasonic device, real-time monitoring of the target object is realized, and then acquisition of the cardiac image is realized. Through the image analysis result, detection of whether there is a risk of thrombus formation in the cardiac image is realized, so as to realize precise control of the pressing device. Through the pressing device, the blood fluidity of the cardiac part is accelerated, thereby reducing the risk of thrombus formation, and then replacement of medical staff is realized, reducing the workload of medical staff.

[0026] In one embodiment, in the step S10, that is, real-time monitoring of the target object corresponding to the establishment of extracorporeal membrane oxygenation is performed through an ultrasonic device, and an initial cardiac image corresponding to the target object is obtained, including: S101, controlling the ultrasonic probe in the ultrasonic device to detect each preset position, so as to collect images of different cardiac sections of the target object, and obtaining an initial cardiac image including a plurality of sectional images.

[0027] Understandably, an ultrasonic probe (also called an ultrasonic transducer) is a device that can convert electrical energy into ultrasonic energy and emit it into the human body, and at the same time can convert the ultrasonic signal reflected back in the human body into an electrical signal. The preset position refers to a preset detection position, for example, the 3rd and 4th intercostal spaces on the left margin of the sternum.

[0028] Specifically, at least one preset position is obtained, and the ultrasonic probe in the ultrasonic device is controlled to detect each preset position. That is, information of the position is obtained through the ultrasonic probe placed at the preset position, and image acquisition of the heart condition at each preset position is performed, so as to obtain cross-sectional images corresponding to each preset position, and all the cross-sectional images are determined as the initial heart image.

[0029] In a specific embodiment, when the preset position is the parasternal long-axis section: Place the ultrasonic probe at the 3rd and 4th intercostal spaces on the left edge of the sternum, and adjust the probe angle and direction so that the ultrasonic beam is parallel to the long axis of the heart. Structures such as the left atrium, left ventricle, mitral valve, aortic valve, and interventricular septum can be observed in this section. When the preset position is the parasternal short-axis section: On the basis of the parasternal long-axis section, rotate the ultrasonic probe clockwise by about 90 degrees to obtain short-axis sections at different levels, including short-axis sections at the mitral valve level, papillary muscle level, and apex level, etc. The movement of the ventricular walls of each segment of the left ventricle, the shape and size of the right ventricle, the pulmonary artery valve and other structures can be observed in this part of the section. When the preset position is the apical four-chamber view: Place the ultrasonic probe at the apical impulse and point it in the direction of the right sternoclavicular joint to obtain the apical four-chamber view. This section shows the four cardiac chambers (left atrium, left ventricle, right atrium, right ventricle) and the conditions of the mitral valve and tricuspid valve. When the preset position is the apical two-chamber view: On the basis of the apical four-chamber view, slightly rotate the ultrasonic probe (usually counterclockwise by about 30 degrees) to obtain the apical two-chamber view, mainly observing the left atrium, left ventricle, and mitral valve structures.

[0030] In this embodiment, by moving the ultrasonic probe to different positions, the acquisition of cardiac images of different sections is realized, thereby ensuring the accuracy of subsequent image recognition and the precise control of subsequent pressing operations.

[0031] In one embodiment, in step S20, that is, obtaining the image analysis result corresponding to the initial heart image includes: S201, obtain an image recognition model, and input the initial heart image into the image recognition model.

[0032] S202, perform image recognition on the initial heart image through the image recognition model to obtain at least one cross-sectional image.

[0033] S203, perform thrombus recognition on all the cross-sectional images through the image recognition model respectively to obtain image recognition results corresponding to each cross-sectional image.

[0034] S204, determine the image analysis result corresponding to the initial heart image according to all the image recognition results.

[0035] Understandably, the image recognition model refers to a network structure used to identify whether there is a risk of thrombus formation in a cardiac image, and this model is trained with a large amount of positive and negative sample data. The sectional image refers to an image collected at different preset positions. The image recognition result is used to characterize the probability of thrombus formation in the sectional image. A thrombus refers to a solid mass formed by blood coagulation or the aggregation of certain formed elements in the blood in the living heart and blood vessels.

[0036] Specifically, after obtaining the initial cardiac image, an image recognition model is acquired, and the initial cardiac image is input into the image recognition model. The image recognition model performs image recognition on the initial cardiac image, that is, it recognizes the sectional images included in the initial cardiac image through the image recognition ability learned during the training of the image recognition model, thereby obtaining at least one sectional image. Then, the image recognition model performs thrombus recognition on each sectional image, that is, it recognizes the information in each sectional image through the thrombus recognition ability learned during the training of the image recognition model, thereby obtaining an image recognition result corresponding to each sectional image. Next, according to all the image recognition results, an image analysis result corresponding to the initial cardiac image is determined, that is, a preset risk threshold is obtained, and all the image recognition results are compared with the preset risk threshold one by one. If all the image recognition results do not exceed the preset risk threshold, an image analysis result indicating no risk of thrombus formation and the formation probability is obtained. If there is at least one image recognition result exceeding the preset risk threshold, an image analysis result indicating the risk of thrombus formation and the formation probability is obtained. Among them, when there are multiple image recognition results exceeding the preset risk threshold, weighted processing is performed on all the formation probabilities to obtain the image analysis result.

[0037] In this embodiment, by using the image recognition model to recognize the initial cardiac image, rapid and accurate recognition of the initial cardiac image is achieved, thereby obtaining the image recognition result, and further obtaining the image analysis result, which facilitates the subsequent precise pressing operation on the target object.

[0038] In one embodiment, in step S203, that is, the image recognition model performs thrombus recognition on all the sectional images respectively to obtain an image recognition result corresponding to each sectional image, including: S2031, extracting features from all the sectional images through the convolutional layer in the image recognition model to obtain convolutional features corresponding to each sectional image.

[0039] S2032, performing pooling processing on all the convolutional features through the pooling layer in the image recognition model to obtain pooling features corresponding to each sectional image.

[0040] S2033, perform image recognition on all the pooling features through the fully connected layer in the image recognition model to obtain an image recognition result corresponding to each of the sectional images.

[0041] Understandably, convolutional features refer to the features extracted through a convolutional network. Pooling features refer to the features after being processed by a pooling network.

[0042] Specifically, after obtaining the sectional images, perform feature extraction on all the sectional images through the convolutional layer in the image recognition model, that is, traverse each sectional image with a preset convolutional kernel and a preset stride, that is, the convolutional kernel slides on the sectional image to extract the local features of the image, and process the extracted features through an activation function, so as to obtain convolutional features corresponding to each sectional image. Then, perform pooling processing on all the convolutional features through the pooling layer in the image recognition model, that is, perform a traversal operation on the entire convolutional feature map according to the preset pooling window size and stride, that is, start from the upper left corner of the convolutional feature map, move the pooling window by the stride, and perform corresponding pooling processing on each area covered by the pooling window in turn until all areas of the entire convolutional feature map are traversed, so as to obtain pooling features corresponding to each convolutional feature. Among them, if there are multiple convolutional layers, process the image through multiple convolutional layers one by one and then perform pooling. Or if there are multiple convolutional pooling layers, process the image one by one and then proceed to the next step. Then, perform image recognition on all the pooling features through the fully connected layer in the image recognition model, that is, calculate the pooling features through each unit in the fully connected layer, so as to obtain an image recognition result representing the formation probability of a thrombus. In this way, an image recognition result corresponding to each sectional image can be obtained.

[0043] In this embodiment, through the convolutional layer, the extraction of features in each sectional image is realized, and the acquisition of convolutional features is realized. Through the pooling layer, the reduction of the dimension of each convolutional feature is realized, the acquisition of pooling features is realized, and further the prediction of the probability of thrombus formation is realized, and the acquisition of the image recognition result is realized.

[0044] In one embodiment, in step S20, that is, making the pressing device perform a pressing operation on the target object according to the image analysis result includes: S205, if the image analysis result indicates that the target object has a risk of thrombus formation, extract the formation probability from the image analysis result.

[0045] S206, determine the pressing depth, pressing frequency, and pressing time corresponding to the target object according to the formation probability.

[0046] S207, make the pressing device perform a pressing operation on the target object according to the pressing depth, the pressing frequency, and the pressing time.

[0047] Understandably, the formation probability refers to the probability of thrombus formation. The pressing depth, pressing frequency, and pressing time refer to the squeezing depth, squeezing frequency, and squeezing time corresponding to this probability. For example, the greater the probability, the deeper the pressing depth, the greater the pressing frequency, and the longer the pressing time. Among them, the maximum pressing depth does not exceed 5 cm.

[0048] Specifically, after inputting the image analysis result into the pressing device, if the image analysis result indicates that the target object has a risk of thrombus formation, the formation probability is extracted from the image analysis result. Then, according to the formation probability, the pressing depth, pressing frequency, and pressing time corresponding to the target object are determined, that is, a preset mapping table is obtained, and the corresponding pressing depth, pressing frequency, and pressing time are queried from the preset mapping table based on the formation probability, and are determined as the pressing depth, pressing frequency, and pressing time corresponding to the target object. Next, the pressing depth, pressing frequency, and pressing time are input into the pressing device, and the pressing device is made to perform a pressing operation on the target object according to the pressing depth, pressing frequency, and pressing time, that is, the pressing device is controlled to press the heart position of the target object at the pressing depth and pressing frequency, and after the pressing time is executed, the pressing operation is stopped.

[0049] In this embodiment, through the formation probability, the acquisition of the pressing depth, pressing frequency, and pressing time is realized, and further the precise control of the pressing device is realized, ensuring that the target object will not be damaged during the pressing process. By the pressing device performing the pressing operation, the blood circulation is accelerated, and further the risk of thrombus formation and the workload of medical staff are reduced.

[0050] In one embodiment, after the step S20, that is, after obtaining the image analysis result corresponding to the initial heart image, it further includes: S50, when the image analysis result indicates that the target object has a risk of thrombus formation, obtain a preset pressing depth, a preset pressing frequency, and a preset pressing time, and input the preset pressing depth, the preset pressing frequency, and the preset pressing time into the pressing device.

[0051] S60, make the pressing device perform a pressing operation on the target object according to the preset pressing depth, the preset pressing frequency, and the preset pressing time.

[0052] Understandably, the preset pressing depth refers to the squeezing depth of the pressing device set in advance. The preset pressing frequency refers to the number of pressing operations performed per unit time set in advance. The preset pressing time refers to the pressing duration set in advance, for example, 5 minutes.

[0053] Specifically, after obtaining the image analysis result corresponding to the initial heart image, when the image analysis result indicates that the target object has a risk of forming a thrombus, obtain the preset pressing depth, preset pressing frequency, and preset pressing time. Then, input the preset pressing depth, preset pressing frequency, and preset pressing time into the pressing device to control the pressing device. Make the pressing device perform a pressing operation on the target object according to the preset pressing depth, preset pressing frequency, and preset pressing time, that is, control the pressing device to press the heart position of the target object with the preset pressing depth, preset pressing frequency, and preset pressing time, so as to accelerate the blood circulation in this part.

[0054] In this embodiment, through the preset pressing depth, preset pressing frequency, and preset pressing time, the pressing of the target object is realized, thereby avoiding the change of the pressing depth, pressing frequency, and pressing time, and further facilitating the pressing operation on the target object by the pressing device.

[0055] In one embodiment, before the step S201, that is, before obtaining the image recognition model, it further includes: S701, obtain a training data set, where the training data set includes at least one sample data and a sample label corresponding to each sample data.

[0056] Understandably, the sample data refers to the cross-sectional images of different target objects with a risk of thrombus formation and the cross-sectional images without a risk of thrombus formation. One sample data is associated with one sample label, and the sample label is used to indicate whether the cross-sectional image of the sample data has a risk of thrombus formation. Obtain the sample data and the sample label corresponding to each sample data from different databases or clients, and then construct a training data set according to the obtained sample data and the sample label corresponding to each sample data.

[0057] S702, obtain a preset training model, and perform image recognition on each sample data through the preset training model to obtain a sample recognition result.

[0058] Understandably, the preset training model refers to a pre-set neural network for identifying whether there is a risk of thrombus formation in the cross-sectional image. The convolutional layer of this model adopts the methods of parameter sharing and sparse connection. Parameter sharing means that the same convolutional kernel shares the same weight at different positions of the image, which enables the model to learn the same features at different positions in the image; sparse connection means that the convolutional kernel is only connected to a part of the pixels in the image, rather than all pixels. The sample recognition result refers to the prediction result of whether there is a risk of thrombus formation in the sample data.

[0059] Specifically, obtain a preset training model and input all sample data into the preset training model. The preset training model preprocesses all sample data to process the sample data into images that meet the preset requirements. Then, the preprocessed sample data is subjected to convolution processing through the convolutional layer in the preset training model, that is, the convolution kernel traverses each preprocessed sample data through a preset stride, thereby obtaining convolution features corresponding to each sample data. Next, the convolution features corresponding to each sample data are subjected to pooling processing through the pooling layer in the preset training model, that is, the pooling kernel traverses each convolution feature through a preset stride, and the pooling features corresponding to each sample data can be obtained. It can be understood that the sample data can be processed through multiple convolutional layers, or processed through multiple convolutional layers and pooling layers. Finally, risk prediction is performed on the pooling features corresponding to each sample data through the fully connected layer, thereby obtaining a sample recognition result indicating whether there is a risk of thrombus formation.

[0060] S703. Determine the prediction loss value of the preset training model according to the sample label and the sample recognition result corresponding to the same sample data.

[0061] It can be understood that the prediction loss value is the loss value generated during the prediction process of the sample data.

[0062] Specifically, arrange the sample recognition results corresponding to each sample data in the order of the sample data in the training dataset, and then compare the sample label associated with the sample data with the sample recognition result corresponding to the sample data with the same sequence; that is, sort according to the sample data, compare the sample label associated with the first sample data with the sample recognition result corresponding to the first sample data, and calculate the loss value between the sample label and the sample recognition result; furthermore, compare the sample label associated with the second sample data with the sample recognition result corresponding to the second sample data, and calculate the loss value between the sample label and the sample recognition result, until the loss values of all sample labels and all sample recognition results are compared, and the prediction loss value can be obtained.

[0063] S704. When the prediction loss value reaches the preset convergence condition, determine the preset training model after convergence as the image recognition model.

[0064] It can be understood that the convergence condition can be the condition that the prediction loss value is less than the set threshold, or the condition that the prediction loss value is very small and will not decrease after 50,000 calculations, and stop training.

[0065] Specifically, when the predicted loss value does not reach the preset convergence condition, the initial parameters of the preset training model are adjusted through the predicted loss value, and the sample data and sample labels are re-input into the preset training model with adjusted initial parameters, and the preset training model with adjusted initial parameters is retrained, so as to obtain the predicted loss value corresponding to the preset training model with adjusted initial parameters. In this way, the prediction result continuously approaches the correct result, and the accuracy of the preset training model becomes higher and higher. Until the predicted loss value of the preset training model reaches the preset convergence condition, the preset training model after convergence is determined as the image recognition model.

[0066] In this embodiment, by iteratively training the preset training model and calculating the overall loss value of the preset training model, the determination of the predicted loss value is realized, and then the determination of the image recognition model is realized, ensuring that the image recognition model has a high accuracy.

[0067] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The order of execution of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0068] In one embodiment, a cardiac compression monitoring device is provided, and the cardiac compression monitoring device corresponds to the cardiac compression monitoring method in the above embodiment one by one. As Figure 2 shown, the cardiac compression monitoring device includes an initial image acquisition module 10, an initial compression operation module 20, an updated image acquisition module 30, and an updated compression operation module 40. The detailed description of each functional module is as follows: The initial image acquisition module 10 is configured to, after extracorporeal membrane oxygenation is established, perform real-time monitoring on a target object corresponding to the established extracorporeal membrane oxygenation through an ultrasonic device to obtain an initial cardiac image corresponding to the target object; The initial compression operation module 20 is configured to obtain an image analysis result corresponding to the initial cardiac image, and input the image analysis result into a compression device, so that the compression device performs a compression operation on the target object according to the image analysis result; The updated image acquisition module 30 is configured to collect an updated cardiac image of the target object through an ultrasonic device if an interval time reaches a preset interval period; The updated compression operation module 40 is configured to obtain an updated analysis result corresponding to the updated cardiac image, and input the updated analysis result into a compression device, so that the compression device performs a compression operation on the target object according to the updated analysis result.

[0069] In one embodiment, the initial compression operation module 20 includes: A model acquisition unit, configured to acquire an image recognition model and input the initial heart image into the image recognition model; An image recognition unit, configured to perform image recognition on the initial heart image through the image recognition model to obtain at least one sectional image; A thrombus recognition unit, configured to perform thrombus recognition on all the sectional images respectively through the image recognition model to obtain an image recognition result corresponding to each sectional image; An image analysis unit, configured to determine an image analysis result corresponding to the initial heart image according to all the image recognition results.

[0070] In one embodiment, the thrombus recognition unit includes: A convolution subunit, configured to perform feature extraction on all the sectional images through a convolution layer in the image recognition model to obtain convolution features corresponding to each sectional image; A pooling subunit, configured to perform pooling processing on all the convolution features through a pooling layer in the image recognition model to obtain pooling features corresponding to each sectional image; A recognition subunit, configured to perform image recognition on all the pooling features through a fully connected layer in the image recognition model to obtain an image recognition result corresponding to each sectional image.

[0071] In one embodiment, the initial pressing operation module 20 includes: An information extraction unit, configured to extract a formation probability from the image analysis result if the image analysis result indicates that the target object has a risk of forming a thrombus; A parameter determination unit, configured to determine a pressing depth, a pressing frequency, and a pressing time corresponding to the target object according to the formation probability; A pressing operation unit, configured to instruct the pressing device to perform a pressing operation on the target object according to the pressing depth, the pressing frequency, and the pressing time.

[0072] In one embodiment, the initial image acquisition module 10 includes: An image acquisition unit, configured to control an ultrasonic probe in the ultrasonic device to detect each preset position, so as to perform image acquisition on different heart sections of the target object to obtain an initial heart image including a plurality of sectional images.

[0073] In one embodiment, the device further includes: A preset parameter acquisition unit, configured to acquire a preset pressing depth, a preset pressing frequency, and a preset pressing time when the image parsing result indicates that the target object has a risk of forming a thrombus, and input the preset pressing depth, the preset pressing frequency, and the preset pressing time into a pressing device; A preset parameter execution unit, configured to instruct the pressing device to perform a pressing operation on the target object according to the preset pressing depth, the preset pressing frequency, and the preset pressing time.

[0074] In one embodiment, the device further includes: A sample acquisition module, configured to acquire a training data set, where the training data set includes at least one sample data and a sample label corresponding to each sample data; An image recognition module, configured to acquire a preset training model, perform image recognition on each sample data through the preset training model, and obtain a sample recognition result; A loss prediction module, configured to determine a prediction loss value of the preset training model according to the sample label and the sample recognition result corresponding to the same sample data; A model convergence module, configured to determine the preset training model after convergence as an image recognition model when the prediction loss value reaches a preset convergence condition.

[0075] For the specific limitations of the cardiac compression monitoring device, reference may be made to the limitations on the cardiac compression monitoring method in the foregoing text, which will not be elaborated herein. Each module in the foregoing cardiac compression monitoring device can be implemented in whole or in part by software, hardware, and their combination. The foregoing modules can be embedded in or independent of a controller in a computer device in the form of hardware, or stored in a memory in the computer device in the form of software, so as to facilitate the controller to call and execute the operations corresponding to the foregoing modules.

[0076] In one embodiment, a computer device is provided. The computer device includes a controller, a memory, a network interface, and a database connected through a system bus. The controller of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a readable storage medium and an internal memory. The readable storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the readable storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the controller, it implements a cardiac compression monitoring method.

[0077] In one embodiment, a computer device is provided, including a memory, a controller, and a computer program stored in the memory and executable on the controller. When the controller executes the computer program, the above-mentioned cardiac compression monitoring method is implemented.

[0078] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a controller, the cardiac compression monitoring method in the above embodiment is implemented.

[0079] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided by the present invention can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0080] Those skilled in the art can clearly understand that for the convenience and brevity of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

[0081] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A cardiac compression monitoring method, characterized in that: include: After the extracorporeal membrane oxygenation is established, a target object corresponding to the establishment of the extracorporeal membrane oxygenation is monitored in real time by an ultrasound device to obtain an initial heart image corresponding to the target object; Acquire an image analysis result corresponding to the initial heart image, and input the image analysis result into a pressing device, so that the pressing device performs a pressing operation on the target object according to the image analysis result; If the interval time reaches the preset interval period, acquiring an updated cardiac image of the target object by means of an ultrasound device; An updated analysis result corresponding to the updated cardiac image is obtained, and the updated analysis result is input into a pressing device, so that the pressing device performs a pressing operation on the target object according to the updated analysis result.

2. The cardiac compression monitoring method according to claim 1, characterized in that: The obtaining of an image analysis result corresponding to the initial cardiac image includes: Acquire an image recognition model, and input the initial heart image into the image recognition model; Performing image recognition on the initial heart image by using the image recognition model to obtain at least one cross-sectional image; Performing thrombus recognition on all the section images respectively by using the image recognition model to obtain image recognition results corresponding to each section image; An image analysis result corresponding to the initial heart image is determined according to all the image recognition results.

3. The cardiac compression monitoring method according to claim 2, characterized in that: The step of performing thrombus recognition on all the section images respectively by using the image recognition model to obtain image recognition results corresponding to each section image includes: Performing feature extraction on all the section images through the convolution layer in the image recognition model to obtain convolution features corresponding to each section image; Performing pooling processing on all the convolutional features through the pooling layer in the image recognition model to obtain pooling features corresponding to each of the section images; Image recognition is performed on all the pooled features through the fully connected layer in the image recognition model to obtain image recognition results corresponding to each of the section images.

4. The cardiac compression monitoring method according to claim 1, characterized in that: The step of causing the pressing device to perform a pressing operation on the target object according to the image analysis result includes: If the image analysis result indicates that the target object has a risk of thrombus formation, extracting a formation probability from the image analysis result; Determining, according to the formation probability, a pressing depth, a pressing frequency, and a pressing time corresponding to the target object; The pressing device is enabled to perform a pressing operation on the target object according to the pressing depth, the pressing frequency and the pressing time.

5. The cardiac compression monitoring method according to claim 1, wherein: The step of monitoring a target object corresponding to establishing extracorporeal membrane oxygenation in real time by using an ultrasonic device to obtain an initial cardiac image corresponding to the target object includes: The ultrasonic probe in the ultrasonic device is controlled to detect each preset position to collect images of different cardiac sections of the target object to obtain an initial cardiac image containing multiple section images.

6. The cardiac compression monitoring method according to claim 1, characterized in that: After obtaining the image analysis result corresponding to the initial cardiac image, the method further includes: When the image analysis result indicates that the target object has a risk of thrombosis, obtaining a preset compression depth, a preset compression frequency, and a preset compression time, and inputting the preset compression depth, the preset compression frequency, and the preset compression time into a compression device; The pressing device is enabled to perform a pressing operation on the target object according to the preset pressing depth, the preset pressing frequency, and the preset pressing time.

7. The cardiac compression monitoring method according to claim 2, characterized in that: Before acquiring the image recognition model, the method further includes: Acquire a training data set, wherein the training data set includes at least one sample data and a sample label corresponding to each sample data; Obtain a preset training model, and perform image recognition on each of the sample data using the preset training model to obtain a sample recognition result; Determining a prediction loss value of the preset training model according to the sample label and the sample recognition result corresponding to the same sample data; When the predicted loss value reaches a preset convergence condition, the preset training model after convergence is determined as the image recognition model.

8. A cardiac compression monitoring device, characterized in that: include: An initial image acquisition module is used to monitor the target object corresponding to the establishment of extracorporeal membrane oxygenation in real time through an ultrasonic device after the establishment of extracorporeal membrane oxygenation, and obtain an initial cardiac image corresponding to the target object; an initial pressing operation module, used for acquiring an image analysis result corresponding to the initial cardiac image, and inputting the image analysis result into a pressing device, so that the pressing device performs a pressing operation on the target object according to the image analysis result; An updated image acquisition module, configured to acquire an updated cardiac image of the target object through an ultrasound device if the interval time reaches a preset interval period; The update pressing operation module is used to obtain the update analysis result corresponding to the updated cardiac image, and input the update analysis result into the pressing device, so that the pressing device performs a pressing operation on the target object according to the update analysis result.

9. A computer device comprising a memory, a controller, and a computer program stored in the memory and executable on the controller, wherein: When the controller executes the computer program, the cardiac compression monitoring method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the controller, the cardiac compression monitoring method according to any one of claims 1 to 7 is implemented.

Images