Cardiac ultrasonic examination auxiliary system based on artificial intelligence

Abstract

The invention relates to the technical field of image processing, in particular to a cardiac ultrasound examination auxiliary system based on artificial intelligence, comprising: an image processing module comprising an acquisition unit for acquiring cardiac ultrasound dynamic images by using an ultrasound probe; the model training module is connected with the image processing module and comprises a model generation unit which is used for training an initial model according to the optimized image so as to output a convolutional neural network model; the feature extraction module is connected with the model training module and comprises a feature extraction unit which is used for extracting key features of the heart structure through a convolutional neural network model so as to output heart ultrasonic dynamic image features; and the control module is respectively connected with the image processing module, the model training module and the feature extraction module and is used for determining the pulse repetition frequency of the ultrasonic probe according to the artifact occurrence rate of the cardiac ultrasonic dynamic image. According to the invention, the processing stability of the cardiac ultrasonic dynamic image is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to an artificial intelligence-based cardiac ultrasound examination auxiliary system. Background Art

[0002] In existing technologies, with the rapid development of medical imaging technology, cardiac ultrasound (echocardiography) has become a core tool for the diagnosis, treatment evaluation and follow-up of cardiovascular diseases. It uses algorithms such as convolutional neural networks and Transformer to achieve automatic segmentation of cardiac structures and disease classification, and combines time series modeling to analyze dynamic blood flow and valve movement. At the same time, it uses multimodal data fusion and adversarial generative networks to enhance data diversity, and realizes real-time analysis by lightweight model deployment on portable devices. It solves the problems of strong operation dependence, high measurement subjectivity and artifact interference in traditional ultrasound examinations, and ultimately improves diagnostic efficiency and accuracy.

[0003] Chinese Patent Publication No. CN112233087A discloses an artificial intelligence-based ophthalmic ultrasound disease diagnosis method and system, comprising: a client and a server; the client is used to monitor and collect images from an ophthalmic ultrasound examination device, upload them to the server via a network, receive and display the current eyeball position, lesion location, and disease type in real time; the server is used to receive ophthalmic ultrasound images collected from the client, perform eyeball segmentation using a convolutional neural network, identify whether the segmented eyeball has a disease, and if so, further identify the disease type and lesion location, and feed back various identification results to the client.

[0004] It can be seen that the existing technology has the following problems: due to the presence of heartbeat motion artifacts and blood flow signal interference in cardiac ultrasound dynamic images, the existing system lacks an effective mechanism to deal with these interference factors. In addition, the training data does not cover enough heart disease scenarios, resulting in weak model generalization ability, which leads to the processing stability of cardiac ultrasound dynamic images not meeting the requirements. Summary of the Invention

[0005] To this end, the present invention provides an artificial intelligence-based cardiac ultrasound examination auxiliary system to overcome the problems in the prior art caused by heartbeat motion artifacts and blood flow signal interference in cardiac ultrasound dynamic images. The existing system lacks an effective processing mechanism for these interference factors. In addition, the training data does not cover enough heart disease scenarios, resulting in weak model generalization ability, which leads to the problem that the processing stability of cardiac ultrasound dynamic images does not meet the requirements.

[0006] To achieve the above objectives, the present invention provides an artificial intelligence-based cardiac ultrasound examination auxiliary system, comprising: An image processing module, comprising an acquisition unit for acquiring dynamic cardiac ultrasound images using an ultrasound probe and a preprocessing unit connected to the acquisition unit for preprocessing the dynamic cardiac ultrasound images to output optimized images; a model training module connected to the image processing module, comprising a model generation unit for training an initial model according to the optimized image to output a convolutional neural network model, and a model update unit connected to the model generation unit for updating the convolutional neural network model using a stochastic gradient descent optimization algorithm; a feature extraction module connected to the model training module, comprising a feature extraction unit for extracting key features of the cardiac structure through a convolutional neural network model to output features of a cardiac ultrasound dynamic image, and an analysis unit connected to the feature extraction unit for analyzing the features of the cardiac ultrasound dynamic image to output analysis results; A control module is respectively connected to the image processing module, the model training module and the feature extraction module, and is used to determine the pulse repetition frequency of the ultrasound probe according to the artifact incidence rate of the cardiac ultrasound dynamic image, or to determine the gradient clipping threshold according to the analysis accuracy of the convolutional neural network model, and to determine the transmission power of the ultrasound probe according to the signal-to-noise ratio of the cardiac ultrasound dynamic image.

[0007] Furthermore, the control module is used to determine that the processing stability of the cardiac ultrasound dynamic image meets the requirements according to the artifact occurrence rate of the cardiac ultrasound dynamic image being less than or equal to a preset first occurrence rate; The control module is used for determining that the processing stability of the cardiac ultrasound dynamic image does not meet the requirements according to the artifact occurrence rate of the cardiac ultrasound dynamic image being greater than a preset first occurrence rate.

[0008] Furthermore, the control module is used to preliminarily determine that the training effectiveness of the convolutional neural network model does not meet the requirements based on the fact that the artifact incidence rate of the cardiac ultrasound dynamic image is greater than the preset first incidence rate and less than or equal to the preset second incidence rate.

[0009] Furthermore, the control module is configured to increase the pulse repetition frequency of the ultrasound probe according to the artifact occurrence rate of the cardiac ultrasound dynamic image being greater than the preset second occurrence rate; The increase range of the pulse repetition frequency of the ultrasound probe is determined by the difference between the artifact occurrence rate of the cardiac ultrasound dynamic image and a preset second occurrence rate.

[0010] Furthermore, the control module is used to determine whether the training effectiveness of the convolutional neural network model meets the requirements based on the analysis accuracy of the convolutional neural network model being greater than or equal to a preset second accuracy rate; The control module is used to determine that the training effectiveness of the convolutional neural network model does not meet the requirements based on the analysis accuracy of the convolutional neural network model being less than a preset second accuracy.

[0011] Furthermore, the control module is used to reduce the gradient clipping threshold according to the analysis accuracy of the convolutional neural network model being greater than a preset first accuracy and less than the preset second accuracy.

[0012] Furthermore, the control module is used to preliminarily determine that the accuracy of extracting cardiac ultrasound dynamic image features does not meet the requirements based on the analysis accuracy of the convolutional neural network model being less than or equal to the preset first accuracy.

[0013] Furthermore, the reduction range of the gradient clipping threshold is determined by the difference between the analysis accuracy of the convolutional neural network model and a preset first accuracy.

[0014] Furthermore, the control module is configured to determine that the accuracy of extracting features of the cardiac ultrasound dynamic image does not meet requirements based on the signal-to-noise ratio of the cardiac ultrasound dynamic image being less than a preset signal-to-noise ratio, and to increase the transmission power of the ultrasound probe.

[0015] Furthermore, the increase range of the transmission power of the ultrasound probe is determined by the difference between a preset signal-to-noise ratio and a signal-to-noise ratio of the cardiac ultrasound dynamic image.

[0016] Compared with the prior art, the beneficial effect of the present invention lies in that the system of the present invention adjusts the pulse repetition rate according to the artifact incidence rate of cardiac ultrasound dynamic images by setting an image acquisition module, a model training module, a feature extraction module and a control module, and uses artificial intelligence image recognition training to quickly identify the morphological changes of the heart under the ultrasonic four-chamber heart section under cardiac arrest, as well as common heart diseases. At the same time, it intelligently identifies images with non-standard sections and prompts the scanner to adjust the scanning angle. Artifacts caused by factors such as air and bone affect the clarity and edge recognition of cardiac structures. By increasing the duty cycle of the pulse repetition frequency, the effective signal emission per unit time can be increased, thereby enhancing the intensity and continuity of the ultrasonic echo signal and reducing image fluctuations caused by signal attenuation. Especially in areas with large detection depth or weak tissue echoes, stable signal output can improve image contrast and resolution and reduce artifacts caused by signal instability. The gradient clipping threshold is adjusted according to the analysis accuracy of the convolutional neural network model. Due to the insufficient generalization ability of the model, the convolutional neural network model performs well on the training set. However, when faced with unseen real-world data, the accuracy and stability of feature extraction significantly decrease. The model cannot effectively identify key patterns in the data and is prone to misinterpreting noise or local features in the training data as universal patterns, resulting in deviations in the contour extraction of fine cardiac structures, which in turn reduces the accuracy of the diagnostic assistance system. By reducing the gradient clipping threshold, excessive gradients can be more strictly limited, suppressing the model's overfitting to noise and local features in the training data, and preventing the model from misinterpreting occasional noise signals as key patterns, thereby reducing deviations in the contour extraction of fine cardiac structures. The ultrasound probe's transmit power is adjusted according to the signal-to-noise ratio of the dynamic cardiac ultrasound image. Due to hardware issues in the ultrasound equipment itself, the image resolution is reduced, and fine cardiac structures such as chordae tendineae and valve edges become blurred, making it impossible to obtain continuous and accurate spatiotemporal features. This makes it difficult for artificial intelligence to accurately extract their contours and morphological features, ultimately significantly reducing the accuracy of the diagnostic assistance system. By increasing the ultrasound probe's transmit power, the initial energy of the ultrasound wave can be increased, enhancing the echo signal strength of deep tissues and far-field structures, and improving image contrast.

[0017] Furthermore, the system of the present invention adjusts the pulse repetition rate by setting a preset first incidence rate and a preset second incidence rate, and uses artificial intelligence image recognition training to quickly identify the morphological changes of the heart under the ultrasonic four-chamber heart section under cardiac arrest, as well as common heart diseases. At the same time, it intelligently identifies images with non-standard sections and prompts the scanner to adjust the scanning angle. For artifacts caused by factors such as air and bones that affect the clarity and edge recognition of the heart structure, by increasing the duty cycle of the pulse repetition frequency, the effective signal emission per unit time can be increased, thereby enhancing the intensity and continuity of the ultrasonic echo signal, and reducing image fluctuations caused by signal attenuation. Especially in areas with large detection depths or weak tissue echoes, stable signal output can improve image contrast and resolution, reduce artifacts caused by signal instability, and further improve the processing stability of cardiac ultrasound images.

[0018] Furthermore, the system of the present invention adjusts the gradient clipping threshold by setting a preset first accuracy rate and a preset second accuracy rate. Due to the insufficient generalization ability of the model, the convolutional neural network model performs well on the training set, but when faced with unseen real data, the accuracy and stability of feature extraction are significantly reduced. The model cannot effectively identify key patterns in the data, and is prone to misjudge the noise or local features of the training data as general rules, resulting in deviations in the contour extraction of the fine structure of the heart, which in turn leads to a reduction in the accuracy of the diagnostic assistance system. By reducing the gradient clipping threshold, excessive gradients can be more strictly limited, suppressing the model's overfitting of noise and local features in the training data, and avoiding the model from misjudging accidental noise signals as key patterns, thereby reducing the deviation in the extraction of the contour of the fine structure of the heart and further improving the processing stability of cardiac ultrasound images.

[0019] Furthermore, the system of the present invention adjusts the transmission power of the ultrasound probe by setting a preset signal-to-noise ratio. Due to hardware problems of the ultrasound equipment itself, the image resolution is reduced, and fine cardiac structures such as chordae tendineae and valve edges become blurred, making it impossible to obtain continuous and accurate spatiotemporal features, making it difficult for artificial intelligence to accurately extract their contours and morphological features, ultimately leading to a significant decrease in the accuracy of the diagnostic assistance system. By increasing the transmission power of the ultrasound probe, the initial energy of the ultrasound wave can be increased, the echo signal intensity of deep tissues and far-field structures can be enhanced, the image contrast can be improved, and the processing stability of cardiac ultrasound images can be further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a block diagram of the overall structure of the artificial intelligence-based cardiac ultrasound examination auxiliary system according to an embodiment of the present invention; Figure 2This is a logic flow chart of a process for determining the pulse repetition frequency of an ultrasound probe in an artificial intelligence-based cardiac ultrasound examination auxiliary system according to an embodiment of the present invention; Figure 3 This is a logic flow chart of a process for determining a gradient clipping threshold in an artificial intelligence-based cardiac ultrasound examination auxiliary system according to an embodiment of the present invention; Figure 4 The present invention provides a logical flow chart of a process for determining the transmission power of an ultrasound probe in an artificial intelligence-based cardiac ultrasound examination auxiliary system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0022] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0023] See also Figure 1 、 Figure 2 、 Figure 3 as well as Figure 4 , which are respectively a block diagram of the overall structure of the artificial intelligence-based cardiac ultrasound examination auxiliary system according to an embodiment of the present invention, a logic flow chart of the process of determining the pulse repetition frequency of the ultrasound probe, a logic flow chart of the process of determining the gradient clipping threshold, and a logic flow chart of the process of determining the transmission power of the ultrasound probe. The artificial intelligence-based cardiac ultrasound examination auxiliary system of the present invention includes: An image processing module, comprising an acquisition unit for acquiring dynamic cardiac ultrasound images using an ultrasound probe and a preprocessing unit connected to the acquisition unit for preprocessing the dynamic cardiac ultrasound images to output optimized images; a model training module connected to the image processing module, comprising a model generation unit for training an initial model according to the optimized image to output a convolutional neural network model, and a model update unit connected to the model generation unit for updating the convolutional neural network model using a stochastic gradient descent optimization algorithm; a feature extraction module connected to the model training module, comprising a feature extraction unit for extracting key features of the cardiac structure through a convolutional neural network model to output features of a cardiac ultrasound dynamic image, and an analysis unit connected to the feature extraction unit for analyzing the features of the cardiac ultrasound dynamic image to output analysis results; A control module is respectively connected to the image processing module, the model training module and the feature extraction module, and is used to determine the pulse repetition frequency of the ultrasound probe according to the artifact incidence rate of the cardiac ultrasound dynamic image, or to determine the gradient clipping threshold according to the analysis accuracy of the convolutional neural network model, and to determine the transmission power of the ultrasound probe according to the signal-to-noise ratio of the cardiac ultrasound dynamic image.

[0024] Specifically, the cardiac ultrasound dynamic image includes a two-dimensional cardiac ultrasound image, a three-dimensional cardiac ultrasound image, and a Doppler blood flow image.

[0025] Specifically, preprocessing includes denoising, image enhancement, and image size unification.

[0026] Specifically, the optimized images include a denoised two-dimensional cardiac ultrasound image, an image-enhanced three-dimensional cardiac ultrasound image, and a standardized Doppler blood flow image.

[0027] Specifically, the stochastic gradient descent optimization algorithm is an algorithm used to minimize the objective function and thus update the model parameters.

[0028] Specifically, cardiac ultrasound dynamic image features include ventricular volume, ejection fraction, and valvular regurgitation velocity.

[0029] Specifically, artifacts are false images in the image that have nothing to do with the actual anatomical structure or blood flow.

[0030] Specifically, the analysis results include cardiac structural assessment, functional parameter measurements, and hemodynamic analysis.

[0031] Specifically, the artifact incidence rate of cardiac ultrasound dynamic images is the ratio of the number of artifact occurrences to the total number of examinations.

[0032] Specifically, the pulse repetition frequency of the ultrasonic probe is the number of times the ultrasonic probe transmits a periodic pulse signal per unit time.

[0033] Specifically, the gradient clipping threshold is a boundary value used to limit the gradient size during the process of optimizing stochastic gradient descent.

[0034] Specifically, the signal-to-noise ratio of a cardiac ultrasound dynamic image is the ratio of the true anatomical structure signal intensity to the noise signal intensity.

[0035] Specifically, the transmission power of an ultrasonic probe is the energy output per unit time when the probe transmits ultrasonic waves.

[0036] During implementation, the system of the present invention adjusts the pulse repetition rate according to the artifact incidence rate of dynamic cardiac ultrasound images by setting an image processing module, a model training module, a feature extraction module, and a control module. It uses artificial intelligence image recognition training to quickly identify morphological changes in the heart under the four-chamber ultrasound view under cardiac arrest, as well as common heart diseases. At the same time, it intelligently identifies images with non-standard sections and prompts the scanner to adjust the scanning angle. Artifacts caused by factors such as air and bone affect the clarity and edge recognition of cardiac structures. By increasing the duty cycle of the pulse repetition frequency, the effective signal emission per unit time can be increased, thereby enhancing the intensity and continuity of the ultrasonic echo signal and reducing image fluctuations caused by signal attenuation. Especially in areas with large detection depths or weak tissue echoes, stable signal output can improve image contrast and resolution, reduce artifacts caused by signal instability, and reduce artifacts caused by signal instability. The gradient clipping threshold is adjusted according to the analysis accuracy of the convolutional neural network model. Due to the insufficient generalization ability of the model, the convolutional neural network model performs well on the training set. However, when faced with unseen real-world data, the accuracy and stability of feature extraction significantly decrease. The model cannot effectively identify key patterns in the data and is prone to misinterpreting noise or local features in the training data as universal patterns, resulting in deviations in the contour extraction of fine cardiac structures, which in turn reduces the accuracy of the diagnostic assistance system. By reducing the gradient clipping threshold, excessive gradients can be more strictly limited, suppressing the model's overfitting to noise and local features in the training data, and preventing the model from misinterpreting occasional noise signals as key patterns, thereby reducing deviations in the contour extraction of fine cardiac structures. The ultrasound probe's transmit power is adjusted according to the signal-to-noise ratio of the dynamic cardiac ultrasound image. Due to hardware issues in the ultrasound equipment itself, the image resolution is reduced, and fine cardiac structures such as chordae tendineae and valve edges become blurred, making it impossible to obtain continuous and accurate spatiotemporal features. This makes it difficult for artificial intelligence to accurately extract their contours and morphological features, ultimately significantly reducing the accuracy of the diagnostic assistance system. By increasing the ultrasound probe's transmit power, the initial energy of the ultrasound wave can be increased, enhancing the echo signal strength of deep tissues and far-field structures, and improving image contrast.

[0037] Specifically, the control module is used to determine that the processing stability of the cardiac ultrasound dynamic image meets the requirements according to the artifact occurrence rate of the cardiac ultrasound dynamic image being less than or equal to a preset first occurrence rate; The control module is used for determining that the processing stability of the cardiac ultrasound dynamic image does not meet the requirements according to the artifact occurrence rate of the cardiac ultrasound dynamic image being greater than a preset first occurrence rate.

[0038] Specifically, the control module is used to preliminarily determine that there is a risk that the training effectiveness of the convolutional neural network model does not meet the requirements based on the fact that the artifact incidence rate of the cardiac ultrasound dynamic image is greater than the preset first incidence rate and less than or equal to the preset second incidence rate, and determine whether the training effectiveness of the convolutional neural network model meets the requirements based on the analysis accuracy of the convolutional neural network model.

[0039] It can be understood that the three intervals divided by the preset first incidence rate and the preset second incidence rate correspond to three situations respectively: The first interval is when the artifact occurrence rate of the cardiac ultrasound dynamic image is less than or equal to the preset first occurrence rate, and the corresponding situation is: determining that the processing stability of the cardiac ultrasound dynamic image meets the requirements; The second interval is when the artifact incidence rate of dynamic cardiac ultrasound images is greater than the preset first incidence rate and less than or equal to the preset second incidence rate. This corresponds to the following situation: due to insufficient generalization ability of the model, the convolutional neural network model performs well on the training set, but when faced with unseen real data, the accuracy and stability of feature extraction decrease significantly. The model cannot effectively identify key patterns in the data and is prone to misinterpreting noise or local features of the training data as universal patterns, resulting in deviations in the contour extraction of subtle cardiac structures, which in turn reduces the accuracy of the diagnostic assistance system. The third interval is when the artifact incidence rate of cardiac ultrasound dynamic images is greater than the preset second incidence rate. The corresponding situation is: using artificial intelligence image recognition training to quickly identify the morphological changes of the heart under the ultrasound four-chamber heart section under cardiac arrest, as well as common heart diseases. At the same time, it intelligently identifies images with non-standard sections and prompts the scanner to adjust the scanning angle. It also identifies artifacts caused by factors such as air and bones that affect the clarity and edge recognition of the heart structure.

[0040] It is understood that the preset first incidence rate and the preset second incidence rate can be set based on actual operating conditions, and the preset first incidence rate and the preset second incidence rate are intended to ensure the accuracy and practicality of the test results. Optionally, the preset first incidence rate and the preset second incidence rate are calculated by extracting a large number of historical examination records from a clinical database, and the actual incidence rates of various artifacts are calculated as a benchmark reference. The incidence rates are then adjusted based on the clinical impact of the artifacts. For example, the preset first incidence rate is generally selected from the range of [6%, 10%], and the preset second incidence rate is generally selected from the range of [11%, 15%].

[0041] Preferably, the preferred embodiment of the preset first occurrence rate is 8%, and the preferred embodiment of the preset second occurrence rate is 13%.

[0042] Specifically, the control module is used to increase the pulse repetition frequency of the ultrasound probe according to the artifact occurrence rate of the cardiac ultrasound dynamic image being greater than the preset second occurrence rate; The increase range of the pulse repetition frequency of the ultrasound probe is determined by the difference between the artifact occurrence rate of the cardiac ultrasound dynamic image and a preset second occurrence rate.

[0043] Specifically, when the difference between the artifact incidence rate of the cardiac ultrasound dynamic image and the preset second incidence rate is within 2%, the pulse repetition frequency of the ultrasound probe is increased to 1.1 times the original rate; when the difference between the artifact incidence rate of the cardiac ultrasound dynamic image and the preset second incidence rate exceeds 2%, on the basis of being increased to 1.1 times the original rate, the pulse repetition frequency of the ultrasound probe is increased by 1kHz for every 1% that exceeds it. For example, the difference between the artifact incidence rate of the cardiac ultrasound dynamic image and the preset second incidence rate is 4%, the current pulse repetition frequency of the ultrasound probe is 10kHz, and the increased pulse repetition frequency of the ultrasound probe is 10×1.1+2×1=13kHz.

[0044] Specifically, kHz is the unit of the pulse repetition frequency of an ultrasound probe, which means kilohertz.

[0045] During implementation, the system of the present invention adjusts the pulse repetition rate by setting a preset first incidence rate and a preset second incidence rate, and uses artificial intelligence image recognition training to quickly identify the morphological changes of the heart under the ultrasonic four-chamber heart section under cardiac arrest, as well as common heart diseases. At the same time, it intelligently identifies images with non-standard sections and prompts the scanner to adjust the scanning angle. For artifacts caused by factors such as air and bones, which affect the clarity and edge recognition of the heart structure, the effective signal emission per unit time can be increased by increasing the duty cycle of the pulse repetition frequency, thereby enhancing the intensity and continuity of the ultrasonic echo signal, and reducing image fluctuations caused by signal attenuation. Especially in areas with large detection depths or weak tissue echoes, stable signal output can improve image contrast and resolution, reduce artifacts caused by signal instability, and further improve the processing stability of dynamic cardiac ultrasound images.

[0046] Specifically, the control module is used to determine whether the training effectiveness of the convolutional neural network model meets the requirements based on the analysis accuracy of the convolutional neural network model being greater than or equal to a preset second accuracy rate; The control module is used to determine that the training effectiveness of the convolutional neural network model does not meet the requirements based on the analysis accuracy of the convolutional neural network model being less than a preset second accuracy.

[0047] Specifically, the control module is used to reduce the gradient clipping threshold according to the analysis accuracy of the convolutional neural network model being greater than a preset first accuracy and less than the preset second accuracy.

[0048] Specifically, the control module is used to preliminarily determine that there is a risk that the accuracy of extracting cardiac ultrasound dynamic image features does not meet the requirements based on the analysis accuracy of the convolutional neural network model being less than or equal to the preset first accuracy, and to determine whether the accuracy of extracting cardiac ultrasound dynamic image features meets the requirements based on the signal-to-noise ratio of the cardiac ultrasound dynamic image.

[0049] It can be understood that the three intervals divided by the preset first accuracy rate and the preset second accuracy rate correspond to three situations respectively: The first interval is when the analysis accuracy of the convolutional neural network model is greater than or equal to the preset second accuracy, corresponding to the situation where it is determined that the training effectiveness of the convolutional neural network model meets the requirements; The second interval is when the convolutional neural network model's analysis accuracy is greater than the preset first accuracy but less than the preset second accuracy. This corresponds to the following situation: Due to insufficient generalization ability of the model, the convolutional neural network model performs well on the training set, but when faced with unseen real data, the accuracy and stability of feature extraction decrease significantly. The model cannot effectively identify key patterns in the data and is prone to misinterpreting noise or local features of the training data as universal patterns, resulting in deviations in the contour extraction of subtle cardiac structures, which in turn reduces the accuracy of the diagnostic assistance system. The third interval is when the analysis accuracy of the convolutional neural network model is less than or equal to the preset first accuracy. The corresponding situation is: due to hardware problems of the ultrasound equipment itself, the image resolution is reduced, and the fine structures of the heart such as chordae tendineae and valve edges become blurred, and continuous and accurate spatiotemporal features cannot be obtained, making it difficult for artificial intelligence to accurately extract their contours and morphological features, ultimately leading to a significant decrease in the accuracy of the diagnostic assistance system.

[0050] It is understandable that the preset first accuracy rate and the preset second accuracy rate can be set according to actual working conditions, and the preset first accuracy rate and the preset second accuracy rate are intended to ensure the accuracy and practicality of the test results. Optionally, the preset first accuracy rate and the preset second accuracy rate are determined by determining the target accuracy range based on task requirements and clinical standards, and then obtaining a reference benchmark through literature review and baseline model testing, and then optimizing the preset value through cross-validation and confusion matrix analysis, and finally iteratively verifying in real clinical data. Exemplarily, the preset first accuracy rate is generally selected in the range of [89%, 93%], and the preset second accuracy rate is generally selected in the range of [94%, 98%].

[0051] Preferably, the preferred embodiment of the preset first accuracy rate is 91%, and the preferred embodiment of the preset second accuracy rate is 95%.

[0052] Specifically, the reduction range of the gradient clipping threshold is determined by the difference between the analysis accuracy of the convolutional neural network model and a preset first accuracy.

[0053] Specifically, when the difference between the analysis accuracy of the convolutional neural network model and the preset first accuracy is within 2%, the gradient clipping threshold is reduced to 0.8 times the original value. When the difference between the analysis accuracy of the convolutional neural network model and the preset first accuracy exceeds 2%, on the basis of being reduced to 0.8 times the original value, the gradient clipping threshold is reduced by 0.5 for every 1% that exceeds it. For example, the difference between the analysis accuracy of the convolutional neural network model and the preset first accuracy is 3%, the current gradient clipping threshold is 5.0, and the reduced gradient clipping threshold is 5.0×0.8-0.5×1=3.5.

[0054] During implementation, the system of the present invention adjusts the gradient clipping threshold by setting a preset first accuracy rate and a preset second accuracy rate. Due to the insufficient generalization ability of the model, the convolutional neural network model performs well on the training set, but when faced with unseen real data, the accuracy and stability of feature extraction are significantly reduced. The model cannot effectively identify key patterns in the data, and is prone to misjudge the noise or local features of the training data as general rules, resulting in deviations in the contour extraction of the fine structure of the heart, which in turn leads to a reduction in the accuracy of the diagnostic assistance system. By reducing the gradient clipping threshold, excessive gradients can be more strictly limited, suppressing the model's overfitting of noise and local features in the training data, and avoiding the model from misjudging accidental noise signals as key patterns, thereby reducing the deviation in the extraction of the contour of the fine structure of the heart and further improving the processing stability of dynamic cardiac ultrasound images.

[0055] Specifically, the control module is used to determine that the accuracy of extracting the features of the cardiac ultrasound dynamic image does not meet the requirements according to the signal-to-noise ratio of the cardiac ultrasound dynamic image being less than a preset signal-to-noise ratio, and to increase the transmission power of the ultrasound probe.

[0056] It is understandable that the two intervals divided by the preset signal-to-noise ratio correspond to two situations: The first interval is when the signal-to-noise ratio of the cardiac ultrasound dynamic image is greater than or equal to the preset signal-to-noise ratio, corresponding to the situation that: it is determined that the accuracy of feature extraction of the cardiac ultrasound dynamic image meets the requirements; The second interval is when the signal-to-noise ratio of the dynamic cardiac ultrasound image is greater than the preset signal-to-noise ratio. The corresponding situation is: due to hardware problems with the ultrasound equipment itself, the image resolution is reduced, and the subtle structures of the heart such as chordae tendineae and valve edges become blurred, making it impossible to obtain continuous and accurate spatiotemporal features, making it difficult for artificial intelligence to accurately extract their contours and morphological features, ultimately leading to a significant decrease in the accuracy of the diagnostic assistance system.

[0057] It is understandable that the preset signal-to-noise ratio can be set according to actual working conditions, and the preset signal-to-noise ratio is intended to ensure the accuracy and practicality of the test results. Optionally, the preset signal-to-noise ratio is obtained by experimentally measuring the baseline signal-to-noise ratio of typical clinical images, calculating the signal mean and noise standard deviation in the myocardial area and the echo-free area respectively, and obtaining the original signal-to-noise ratio distribution. Subsequently, in conjunction with clinical experts, an initial threshold is formulated according to diagnostic needs, and then through simulation testing, controllable Gaussian noise is added to the original image to verify the detection rate of key lesions under the preset signal-to-noise ratio, and finally determined in combination with device performance and patient-specific factors. Exemplarily, the preset signal-to-noise ratio is generally selected in the range of [20dB, 25dB].

[0058] Preferably, the preset signal-to-noise ratio is 23 dB.

[0059] Specifically, the increase range of the transmission power of the ultrasound probe is determined by the difference between a preset signal-to-noise ratio and a signal-to-noise ratio of the cardiac ultrasound dynamic image.

[0060] Specifically, when the difference between the signal-to-noise ratio of the dynamic cardiac ultrasound image and the preset signal-to-noise ratio is within 2dB, the transmission power of the ultrasound probe is increased to 1.2 times the original value; when the difference between the signal-to-noise ratio of the dynamic cardiac ultrasound image and the preset signal-to-noise ratio exceeds 2dB, on the basis of being increased to 1.2 times the original value, the transmission power of the ultrasound probe is increased by 2mW for every 1dB exceeding it. For example, when the difference between the signal-to-noise ratio of the dynamic cardiac ultrasound image and the preset signal-to-noise ratio is 3dB, the current transmission power of the ultrasound probe is 30mW, and the increased transmission power of the ultrasound probe is 30×1.2+2×1=38mW.

[0061] Specifically, mW is the unit of the transmission power of the ultrasound probe, which means milliwatt.

[0062] During implementation, the system of the present invention adjusts the transmission power of the ultrasound probe by setting a preset signal-to-noise ratio. Due to hardware problems of the ultrasound equipment itself, the image resolution is reduced, and fine cardiac structures such as chordae tendineae and valve edges become blurred, making it impossible to obtain continuous and accurate spatiotemporal features, making it difficult for artificial intelligence to accurately extract their contours and morphological features, ultimately leading to a significant decrease in the accuracy of the diagnostic assistance system. By increasing the transmission power of the ultrasound probe, the initial energy of the ultrasound wave can be increased, the echo signal intensity of deep tissues and far-field structures can be enhanced, the image contrast can be improved, and the processing stability of dynamic cardiac ultrasound images can be further improved.

[0063] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. An artificial intelligence-based cardiac ultrasound examination auxiliary system, characterized in that: include: An image processing module, comprising an acquisition unit for acquiring dynamic cardiac ultrasound images using an ultrasound probe and a preprocessing unit connected to the acquisition unit for preprocessing the dynamic cardiac ultrasound images to output optimized images; a model training module connected to the image processing module, comprising a model generation unit for training an initial model according to the optimized image to output a convolutional neural network model, and a model update unit connected to the model generation unit for updating the convolutional neural network model using a stochastic gradient descent optimization algorithm; a feature extraction module connected to the model training module, comprising a feature extraction unit for extracting key features of the cardiac structure through a convolutional neural network model to output features of a cardiac ultrasound dynamic image, and an analysis unit connected to the feature extraction unit for analyzing the features of the cardiac ultrasound dynamic image to output analysis results; A control module is respectively connected to the image processing module, the model training module and the feature extraction module, and is used to determine the pulse repetition frequency of the ultrasound probe according to the artifact incidence rate of the cardiac ultrasound dynamic image, or to determine the gradient clipping threshold according to the analysis accuracy of the convolutional neural network model, and to determine the transmission power of the ultrasound probe according to the signal-to-noise ratio of the cardiac ultrasound dynamic image.

2. The artificial intelligence-based cardiac ultrasound examination auxiliary system according to claim 1, characterized in that: The control module is configured to determine that the processing stability of the cardiac ultrasound dynamic image meets the requirements according to the artifact occurrence rate of the cardiac ultrasound dynamic image being less than or equal to a preset first occurrence rate; The control module is used for determining that the processing stability of the cardiac ultrasound dynamic image does not meet the requirements according to the artifact occurrence rate of the cardiac ultrasound dynamic image being greater than a preset first occurrence rate.

3. The artificial intelligence-based cardiac ultrasound examination auxiliary system according to claim 2, characterized in that: The control module is used to preliminarily determine that the training effectiveness of the convolutional neural network model does not meet the requirements based on the artifact incidence rate of the cardiac ultrasound dynamic image being greater than the preset first incidence rate and less than or equal to the preset second incidence rate.

4. The artificial intelligence-based cardiac ultrasound examination auxiliary system according to claim 3, characterized in that: The control module is configured to increase the pulse repetition frequency of the ultrasound probe according to the artifact occurrence rate of the cardiac ultrasound dynamic image being greater than the preset second occurrence rate; The increase range of the pulse repetition frequency of the ultrasound probe is determined by the difference between the artifact occurrence rate of the cardiac ultrasound dynamic image and a preset second occurrence rate.

5. The artificial intelligence-based cardiac ultrasound examination auxiliary system according to claim 3, characterized in that: The control module is configured to determine whether the training effectiveness of the convolutional neural network model meets the requirements based on the analysis accuracy of the convolutional neural network model being greater than or equal to a preset second accuracy rate; The control module is used to determine that the training effectiveness of the convolutional neural network model does not meet the requirements based on the analysis accuracy of the convolutional neural network model being less than a preset second accuracy.

6. The artificial intelligence-based cardiac ultrasound examination auxiliary system according to claim 5, characterized in that: The control module is used to reduce the gradient clipping threshold according to the analysis accuracy of the convolutional neural network model being greater than a preset first accuracy and less than the preset second accuracy.

7. The artificial intelligence-based cardiac ultrasound examination auxiliary system according to claim 6, characterized in that: The control module is used to preliminarily determine that the extraction accuracy of the cardiac ultrasound dynamic image features does not meet the requirements based on the analysis accuracy of the convolutional neural network model being less than or equal to the preset first accuracy.

8. The artificial intelligence-based cardiac ultrasound examination auxiliary system according to claim 7, characterized in that: The reduction range of the gradient clipping threshold is determined by the difference between the analysis accuracy of the convolutional neural network model and a preset first accuracy.

9. The artificial intelligence-based cardiac ultrasound examination auxiliary system according to claim 8, characterized in that: The control module is used to determine that the extraction accuracy of the cardiac ultrasound dynamic image features does not meet the requirements based on the signal-to-noise ratio of the cardiac ultrasound dynamic image being less than a preset signal-to-noise ratio, and to increase the transmission power of the ultrasound probe.

10. The artificial intelligence-based cardiac ultrasound examination auxiliary system according to claim 9, characterized in that: The increase range of the transmission power of the ultrasonic probe is determined by the difference between a preset signal-to-noise ratio and a signal-to-noise ratio of a dynamic cardiac ultrasonic image.

Images