Method, system, medium and device for generating echocardiography imaging recommendations

Abstract

The present application belongs to the technical field of heart ultrasound image processing, and provides a heart ultrasound image shooting suggestion generation method, system, medium and equipment. The technical scheme is to extract the multi-scale spatial features of the heart ultrasound image, fuse the multi-scale spatial features to obtain multi-scale spatial fusion features, perform spatial attention enhancement and channel attention enhancement on the multi-scale spatial fusion features to obtain the spatial attention enhanced and channel attention enhanced multi-scale spatial fusion features, calculate the heart contour definition and echo uniformity based on the spatial attention enhanced multi-scale spatial fusion features, calculate the heart valve definition based on the channel attention enhanced multi-scale spatial fusion features, extract the features of different time points of the heart ultrasound image, calculate the left ventricular wall motion consistency based on the features of different time points of the heart ultrasound image, perform image quality evaluation, generate shooting suggestions according to the evaluation results, and improve the extraction accuracy of important medical information.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of heart ultrasound image processing, and particularly relates to a heart ultrasound image shooting suggestion generation method, system, medium and equipment. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute prior art.

[0003] Heart ultrasound images are widely used in the diagnosis and treatment of cardiovascular diseases. Through ultrasound images, the shape of the heart cavity, the movement of the ventricular wall, and the function of the valve can be observed. However, the current ultrasound image quality evaluation method mainly relies on the experience of doctors for subjective judgment, which is easily affected by personal level and subjective factors, resulting in unstable image quality.

[0004] Heart shooting suggestions are crucial for obtaining key features of heart ultrasound images. The existing evaluation method lacks automatic analysis of heart ultrasound images and generation of optimized shooting suggestions, which may lead to omission of important medical information. SUMMARY

[0005] In order to solve at least one technical problem in the background art, the present application provides a heart ultrasound image shooting suggestion generation method, system, medium and equipment, which automatically evaluates the quality of heart ultrasound images and provides real-time shooting optimization suggestions, thereby improving the accuracy of extracting important medical information.

[0006] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0007] The first aspect of the present application provides a heart ultrasound image shooting suggestion generation method, comprising the following steps:

[0008] Obtaining heart ultrasound image data;

[0009] Extracting multi-scale spatial features of the heart ultrasound image, and obtaining multi-scale spatial fusion features by fusing the multi-scale spatial features;

[0010] Performing spatial attention enhancement and channel attention enhancement on the multi-scale spatial fusion features to obtain spatial attention enhanced and channel attention enhanced multi-scale spatial fusion features;

[0011] Calculating the heart contour sharpness and echo uniformity based on the spatial attention enhanced multi-scale spatial fusion features, and calculating the heart valve sharpness based on the channel attention enhanced multi-scale spatial fusion features;

[0012] Extracting features of different time points of the heart ultrasound image, and calculating the left ventricular wall motion consistency based on the features of different time points of the heart ultrasound image;

[0013] extracting a signal-to-noise ratio of the cardiac ultrasound image;

[0014] performing image quality evaluation based on cardiac contour definition, echo uniformity, cardiac valve definition, left ventricular wall motion consistency and signal-to-noise ratio, and generating a shooting suggestion according to the evaluation result.

[0015] Further, the multi-scale spatial features of the cardiac ultrasound image are extracted, and the multi-scale spatial fusion features are obtained after the multi-scale spatial features are fused, including:

[0016] The spatial features of the cardiac ultrasound image are extracted by using a multi-scale convolutional neural network, different feature maps are obtained by using a plurality of different sizes of convolution kernels, and then the fused feature maps are obtained by fusing the feature maps of each scale.

[0017] Further, the calculation formula for the spatial attention enhancement and the channel attention enhancement of the multi-scale spatial fusion features is:

[0018] ,

[0019] ,

[0020] wherein, represents the importance degree of each position of the pixel, represents the multi-scale spatial fusion features, represents the weight distribution of the pixels at different positions when calculating the importance, represents a convolution operation, which captures the correlation information between different positions in the image by convolution with the weight matrix , and is a bias term for adjusting the offset of the calculation result; represents the importance degree of each channel corresponding to the pixel, is a weight vector of the channel attention, represents a dot product operation, is an activation function.

[0021] Further, the spatial attention enhanced multi-scale spatial fusion features are detected by using a Canny edge detection method to identify the heart cavity boundary, and the cardiac contour definition is obtained.

[0022] Further, the spatial attention enhanced multi-scale spatial fusion features are divided into a plurality of regions, the standard deviation and the average value of the pixel intensity in each region are calculated respectively, and the echo uniformity is obtained based on the standard deviation and the average value of the pixel intensity in each region.

[0023] Further, the method further comprises maximizing the similarity of each feature of similar samples by contrastive learning to obtain an optimized heart contour definition, echo uniformity, heart valve definition, left ventricular wall motion consistency and signal-to-noise ratio.

[0024] Further, the loss function for maximizing the similarity of each feature of similar samples is:

[0025] ,

[0026] wherein, and represent different sample indexes, represent the feature vectors of samples and , and are the inner products of the feature vectors and,

[0027] The second aspect of the present application provides a heart ultrasound image shooting suggestion generation system, comprising:

[0028] a data acquisition module for acquiring heart ultrasound image data;

[0029] a multi-scale feature extraction module for extracting multi-scale spatial features of the heart ultrasound image, and obtaining multi-scale spatial fusion features by fusing the multi-scale spatial features;

[0030] a multi-scale feature enhancement module for performing spatial attention enhancement and channel attention enhancement on the multi-scale spatial fusion features to obtain spatial attention enhanced and channel attention enhanced multi-scale spatial fusion features;

[0031] a feature extraction module for calculating heart contour definition and echo uniformity based on the spatial attention enhanced multi-scale spatial fusion features, calculating heart valve definition based on the channel attention enhanced multi-scale spatial fusion features, extracting features of different time points of the heart ultrasound image, calculating left ventricular wall motion consistency based on the features of different time points of the heart ultrasound image, and extracting signal-to-noise ratio of the heart ultrasound image;

[0032] a shooting suggestion generation module for performing image quality evaluation based on the heart contour definition, echo uniformity, heart valve definition, left ventricular wall motion consistency and signal-to-noise ratio, and generating a shooting suggestion according to the evaluation result.

[0033] The third aspect of the present application provides a computer readable storage medium.

[0034] A computer readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the steps of the heart ultrasound image shooting suggestion generation method described above.

[0035] A fourth aspect of the present application provides a computer device.

[0036] A computer device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in the method for generating a heart ultrasound image shooting suggestion as described above when executing the program.

[0037] Compared with the prior art, the present application has the following beneficial effects:

[0038] The present application extracts a plurality of key heart features according to the set image quality evaluation standard, evaluates the quality of the heart ultrasound image based on the extracted plurality of key heart features, and generates a shooting suggestion according to the quality evaluation result, which can automatically evaluate the quality of the heart ultrasound image and provide a shooting optimization suggestion in real time, realizing efficient and accurate heart ultrasound image processing.

[0039] The advantages of the additional aspects of the present application will be partially given in the following description, partially become obvious from the following description, or be learned through the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0040] The drawings constituting a part of the specification of the present application are used to provide a further understanding of the present application, the illustrative embodiments of the present application and the description thereof are used to explain the present application, and do not constitute an improper limitation on the present application.

[0041] Figure 1 is a flow chart of the method for generating a heart ultrasound image shooting suggestion provided by the embodiments of the present application;

[0042] Figure 2 is a pre-processing result of an input heart valve image provided by the embodiments of the present application;

[0043] Figure 3 is a two-dimensional heart ultrasound image before adjustment provided by the embodiments of the present application;

[0044] Figure 4 is a two-dimensional heart ultrasound image after adjustment provided by the embodiments of the present application. DETAILED DESCRIPTION

[0045] The present application will be further described below in combination with the drawings and embodiments.

[0046] It should be pointed out that the following detailed description is all exemplary, and is intended to provide further description of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as generally understood by those skilled in the art to which the present application belongs.

[0047] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments in accordance with the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, steps, operations, elements, components, and / or groups thereof, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and / or groups thereof.

[0048] The existing evaluation methods mentioned in the background art lack automatic analysis of cardiac ultrasound images and generation of optimized shooting suggestions, which can lead to the omission of important medical information.

[0049] The present application obtains cardiac ultrasound image data, extracts multi-scale spatial features of the cardiac ultrasound image, fuses the multi-scale spatial features to obtain multi-scale spatial fusion features, performs spatial attention enhancement and channel attention enhancement on the multi-scale spatial fusion features to obtain spatial attention enhanced and channel attention enhanced multi-scale spatial fusion features, calculates cardiac contour clarity and echo uniformity based on the spatial attention enhanced multi-scale spatial fusion features, calculates cardiac valve clarity based on the channel attention enhanced multi-scale spatial fusion features, extracts features of different time points of the cardiac ultrasound image, calculates left ventricular wall motion consistency based on the features of different time points of the cardiac ultrasound image, extracts signal-to-noise ratio of the cardiac ultrasound image, performs image quality evaluation based on the optimized cardiac contour clarity, echo uniformity, cardiac valve clarity, left ventricular wall motion consistency and signal-to-noise ratio, and generates shooting suggestions according to the evaluation results.

[0050] Embodiment one

[0051] As shown in Figure 1 , the present embodiment provides a cardiac ultrasound image shooting suggestion generation method, including the following steps:

[0052] Step 1: Obtain cardiac ultrasound image data , the obtained cardiac ultrasound image data is preprocessed to obtain preprocessed cardiac ultrasound image data; as shown in Figure 2 , the preprocessed cardiac ultrasound image is shown;

[0053] Step 2: Extract multi-scale spatial features of the cardiac ultrasound image, and fuse the multi-scale spatial features to obtain multi-scale spatial fusion features.

[0054] In the present embodiment, a multi-scale convolutional neural network is used to extract spatial features of the cardiac ultrasound image. By using multiple convolution kernels of different sizes, feature maps extracted by different convolution kernels are obtained, and then the feature maps of different scales are fused to obtain a fused feature map, which is represented as:

[0055] ,

[0056] wherein, is the feature map extracted by the i-th convolution kernel, is the number of convolution kernels, K is the fusion weight used to adjust the importance of each feature map;

[0057] Step 3: performing spatial attention enhancement and channel attention enhancement on the multi-scale spatial fusion feature to obtain a spatial attention enhanced and channel attention enhanced multi-scale spatial fusion feature;

[0058] Specifically, the method comprises the following steps:

[0059] Step 301: performing spatial attention enhancement on the multi-scale spatial fusion feature to obtain a spatial attention enhanced multi-scale spatial fusion feature;

[0060] In this embodiment, the importance of different positions in the image is calculated to focus on the edge region of the heart, and in particular when evaluating the clarity of the heart cavity contour, spatial attention can help the model to accurately focus on the edge part. The specific formula is:

[0061] ,

[0062] wherein, represents the importance degree of the pixel at each position, represents the weight distribution of the pixels at different positions when calculating the importance, represents a convolution operation, which captures the correlation information between different positions in the image by convolution of and the weight matrix , is a bias term used to adjust the offset of the calculation result, is an activation function which maps the result of convolution and bias calculation to a set interval, so that the output can represent the importance degree of each position.

[0063] According to the importance degree of each position, the final spatial attention enhanced multi-scale spatial fusion feature map is obtained;

[0064] Step 302: performing channel attention enhancement on the multi-scale spatial fusion feature to obtain a channel attention enhanced multi-scale spatial fusion feature;

[0065] By channel weighting, the features related to the heart structure are strengthened, and the noise and irrelevant parts are suppressed.

[0066] The importance degree score calculation formula of each channel corresponding to the pixel is: ​

[0067] ,

[0068] wherein, represents the importance degree of the pixel corresponding to each channel, is a weight vector of channel attention, which weights the data of different channels, represents a dot product operation, which can adjust the data according to the importance of different channels through the dot product of and the weight vector , is a bias term for adjusting the offset of the calculation result, is an activation function for mapping the calculation result to a set interval.

[0069] According to the importance degree of each channel, the final channel attention enhanced multi-scale spatial fusion feature map is obtained;

[0070] Step 4: Calculate the heart contour clarity and echo uniformity based on the spatial attention enhanced multi-scale spatial fusion feature, and calculate the heart valve clarity based on the channel attention enhanced multi-scale spatial fusion feature;

[0071] Specifically, the following steps are included:

[0072] Step 401, calculate the heart contour clarity and echo uniformity based on the spatial attention enhanced multi-scale spatial fusion feature ;

[0073] wherein, the spatial attention enhanced multi-scale spatial fusion feature is detected by using the Canny edge detection method to identify the heart chamber boundary, and the heart contour clarity is obtained;

[0074] When evaluating the heart chamber contour clarity, the model needs to accurately identify the edge of the heart chamber. The spatial attention mechanism can highlight the importance of the edge region in the image through the above calculation.

[0075] For example, for the pixels of the heart chamber edge, their corresponding values may be higher, while the pixels far from the edge have lower values. In this way, the model can pay more attention to the features of the edge region, so as to more accurately quantify the clarity of the heart chamber contour.

[0076] wherein, the calculation process of the echo uniformity includes:

[0077] The spatial attention enhanced multi-scale spatial fusion feature is divided into multiple regions, and the standard deviation and mean value of the pixel intensity in each region are calculated respectively, and the echo uniformity is obtained based on the standard deviation and mean value of the pixel intensity in each region;​

[0078] ,

[0079] wherein, M represents the number of regions, and respectively represent the standard deviation and the average value of the pixel intensity in the first local region.

[0080] Step 402, calculating the heart valve clarity based on the channel attention enhanced multi-scale spatial fusion features and the texture features of the image ;

[0081] Specifically, the following steps are included:

[0082] Step 4021, extracting global features and local features in the image based on the channel attention enhanced multi-scale spatial fusion features;

[0083] In this embodiment, CNN is used to extract features of the valve region, capturing global and local features in the image;

[0084] Step 4022, extracting texture features of the cardiac ultrasound image;

[0085] In this embodiment, first, the cardiac ultrasound image is preprocessed, including denoising and enhancement, to ensure the image quality suitable for feature extraction;

[0086] Then, the gray level co-occurrence matrix (GLCM) method is used to extract texture features such as contrast, correlation, and uniformity, and in addition, the local binary pattern (LBP) is used to analyze the texture features of the local region of the image; by selecting these texture features, the most representative features are selected for subsequent analysis.

[0087] Step 4033, fusing the texture features and the global and local features in the image to evaluate the clarity of the valve region.

[0088] When evaluating the clarity of the heart valve, different channels of data may contain different information, some channels may be more relevant to the features of the heart valve structure, while some may contain more noise or irrelevant information. The channel attention mechanism, through the above calculation, can increase the weight of the channels related to the heart valve structure, so that the model can pay more attention to the data of these important channels when extracting features, thus more accurately quantifying the clarity of the heart valve.

[0089] Step 5: Extracting features of the cardiac ultrasound image at different time points, and calculating the left ventricular wall motion consistency based on the features of the cardiac ultrasound image at different time points;

[0090] Specifically, the following steps are included:

[0091] Step 501, extracting features of different time points of the heart ultrasound image;

[0092] In this embodiment, a TCN (Temporal Convolutional Network) model is used for analysis. The TCN performs convolution operation on data of each time step, which is represented as:

[0093] ,

[0094] wherein, is the weight of the time convolution kernel, is the length of the convolution kernel, is the input data sequence The value at time step is the historical data on which the convolution operation at the current time step depends.

[0095] Through this convolution operation, the TCN can capture the dependency relationship between data at different time points. For example, in a cardiac cycle, the state of the ventricular wall at different times is related, and the state at an early time will affect the state at a later time. The TCN can mine this inherent time-dependent feature through convolution processing of these time series data.

[0096] Step 502, calculating left ventricular wall motion consistency based on features of different time points of the heart ultrasound image;

[0097] In this embodiment, when extracting the cardiac wall motion consistency feature in a cardiac cycle, a TCN (Temporal Convolutional Network) model is used for analysis. The TCN is used to analyze the temporal features of the image, especially the dynamic changes of the left ventricular wall. The motion consistency of the left ventricular wall is evaluated by extracting the dependent features in time.

[0098] The calculation formula of the left ventricular wall motion consistency feature is:

[0099] ,

[0100] wherein, represents the motion consistency of the left ventricular wall, and represent the ventricular wall feature vectors at time points and .

[0101] Step 6: Extracting the signal-to-noise ratio of the heart ultrasound image;

[0102] Signal-to-noise ratio (SNR) quantifies the ratio of signal to noise in an image, reflecting the clarity of the effective information.

[0103] In this embodiment, the signal-to-noise ratio of the cardiac ultrasound image is calculated by calculating the signal and the background noise of the selected region, and the calculation formula is:

[0104] ,

[0105] Among them, is the average value of the signal, is the standard deviation of the noise.

[0106] Step 7: By comparing the similarities of the features of the similar samples, the optimized cardiac contour clarity, echo uniformity, cardiac valve clarity, left ventricular wall motion consistency and signal-to-noise ratio are obtained;

[0107] Specifically, the following steps are included:

[0108] Step 701, defining similar samples;

[0109] For a pair of samples, if they are considered similar under some predefined similarity criteria, Here it can be a similarity label determined based on some properties of the data itself or through some algorithm. For example, in cardiac ultrasound image data, whether two samples are similar can be determined according to some basic features of the image (such as imaging angle, approximate similarity of cardiac structure, etc.);

[0110] Step 702, constructing a loss function that maximizes the feature similarity of each feature of the similar samples:

[0111] ,

[0112] Among them, and represent different sample indexes, represents the inner product of the feature vector sum of sample and sample . The value of the inner product reflects the similarity of the two feature vectors in the direction. and are the norms (lengths) of the feature vector sum. By calculating, the cosine similarity of the two feature vectors is obtained, which is between [-1, 1], and the closer to 1 indicates that the two feature vectors are more similar.

[0113] When , it is hoped that the feature vectors of the two similar samples are as similar as possible, that is, the cosine similarity is as close to 1 as possible. Therefore, in the loss function, to measure the degree of difference between the features of similar samples. By summing all the satisfied sample pairs, the loss function of contrast learning is obtained .

[0114] Step 703, by continuously adjusting the parameters of the model (such as the weights of the convolution kernel), the loss function is minimized, and the model gradually learns the feature expression that can better distinguish similar samples and dissimilar samples. In this way, under unsupervised conditions, the model can learn an effective feature expression for cardiac ultrasound image data, providing a more accurate feature basis for subsequent image quality assessment and shooting suggestion generation.

[0115] Step 8: Perform image quality assessment based on the optimized cardiac contour clarity, echo uniformity, cardiac valve clarity, left ventricular wall motion consistency, and signal-to-noise ratio, and generate shooting suggestions according to the assessment results;

[0116] Specifically, the following steps are included:

[0117] Step 801, perform image quality assessment based on the optimized cardiac contour clarity, echo uniformity, cardiac valve clarity, and left ventricular wall motion consistency, and obtain the assessment results;

[0118] In this embodiment, each image quality indicator is scored according to the set quantitative scoring standard, and the score is 0-5;

[0119] Step 802, generate shooting suggestions according to the obtained results;

[0120] In this embodiment, the specific suggestions include:

[0121] For cardiac contour clarity:

[0122] If the score of the heart chamber contour clarity is less than 2, it is recommended to adjust the probe angle by at least 15 degrees to improve the visualization of the heart chamber edge;

[0123] If the score of the heart chamber contour clarity is between 2-3, it is recommended to fine-tune the probe angle by 5-10 degrees, or increase the image contrast, to improve the edge sharpness;

[0124] If the score of the heart chamber contour clarity is greater than 3, the current setting is sufficient and no adjustment is needed.

[0125] For left ventricular wall motion consistency:

[0126] If the score of the left ventricular wall motion consistency is less than 2, it is recommended to adjust the probe position to ensure full capture of the cardiac cycle, or change the imaging mode;

[0127] If the score of the left ventricular wall motion consistency is between 2-3, it is recommended to fine-tune the probe position, or slightly adjust the capture timing of the cardiac cycle;

[0128] If the score of left ventricular wall motion consistency is greater than 3, the current setting is sufficient, and no adjustment is needed.

[0129] For echo uniformity;

[0130] If the score of echo uniformity is less than 2, it is recommended to adjust the probe focus setting or change the scanning depth to a more uniform area.

[0131] If the score of echo uniformity is between 2-3, it is recommended to fine-tune the transmit power to optimize tissue reflection at different depths.

[0132] If the score of echo uniformity is greater than 3, the current setting is sufficient, and no adjustment is needed.

[0133] For heart valve clarity:

[0134] If the score of echo uniformity is less than 2, it is recommended to adjust the probe angle by at least 20 degrees to better align the valve area, or change the imaging frequency.

[0135] If the score of echo uniformity is between 2-3, it is recommended to fine-tune the probe angle by about 10 degrees, or slightly adjust the imaging frequency.

[0136] If the score of echo uniformity is greater than 3, the current setting is sufficient, and no adjustment is needed.

[0137] The system will monitor the scores of these indicators in real time and provide specific adjustment suggestions. The operator can gradually adjust the device settings according to the system's prompts until the desired image quality is achieved.

[0138] As shown in Figure 3 and Figure 4 , Figure 3 is the two-dimensional heart ultrasound image before adjustment, and the parameters set are probe angle 0°, focus depth 10 cm, Figure 4 is the two-dimensional heart ultrasound image after adjustment, and the parameters set are probe angle 10°, focus depth 12 cm, and it can be seen through comparison that the ultrasound image obtained by the method of the present application is better than the image obtained without adjustment.

[0139] Example Two

[0140] The present embodiment provides a heart ultrasound image shooting suggestion generation system, which comprises:

[0141] A data acquisition module for acquiring heart ultrasound image data;

[0142] A multi-scale feature extraction module for extracting multi-scale spatial features of the heart ultrasound image, and obtaining multi-scale spatial fusion features after fusing the multi-scale spatial features.

[0143] a multi-scale feature enhancement module, configured to perform spatial attention enhancement and channel attention enhancement on the multi-scale spatial fusion features, to obtain spatial attention enhanced and channel attention enhanced multi-scale spatial fusion features;

[0144] a feature extraction module, configured to calculate cardiac contour definition and echo uniformity based on the spatial attention enhanced multi-scale spatial fusion features, to calculate cardiac valve definition based on the channel attention enhanced multi-scale spatial fusion features, to extract features of different time points of the cardiac ultrasound image, and to calculate left ventricular wall motion consistency based on the features of different time points of the cardiac ultrasound image, and to extract a signal-to-noise ratio of the cardiac ultrasound image;

[0145] a shooting suggestion generation module, configured to perform image quality evaluation based on the cardiac contour definition, the echo uniformity, the cardiac valve definition, the left ventricular wall motion consistency and the signal-to-noise ratio, and to generate a shooting suggestion according to the evaluation result.

[0146] It should be noted that the specific implementation of the cardiac ultrasound image shooting suggestion generation system of the embodiment of the present application is similar to that of the cardiac ultrasound image shooting suggestion generation method of the embodiment of the present application, and specific reference can be made to the description in the method part. In order to reduce redundancy, this part will not be repeated.

[0147] Embodiment three

[0148] The embodiment provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to realize the steps in the cardiac ultrasound image shooting suggestion generation method.

[0149] Embodiment four

[0150] The embodiment provides a computer device, which includes a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor executes the program to realize the steps in the cardiac ultrasound image shooting suggestion generation method.

[0151] The above only describes the preferred embodiments of the present application and is not used to limit the present application. For those skilled in the art, the present application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method of generating a cardiac ultrasound imaging recommendation, the method comprising: The method comprises the following steps: obtaining cardiac ultrasound image data; extracting multi-scale spatial features of the cardiac ultrasound image, and obtaining multi-scale spatial fusion features by fusing the multi-scale spatial features; performing spatial attention enhancement and channel attention enhancement on the multi-scale spatial fusion features to obtain spatial attention enhanced and channel attention enhanced multi-scale spatial fusion features; calculating cardiac contour clarity and echo uniformity based on the spatial attention enhanced multi-scale spatial fusion features, and calculating cardiac valve clarity based on the channel attention enhanced multi-scale spatial fusion features; extracting features of different time points of the cardiac ultrasound image, and calculating left ventricular wall motion consistency based on the features of different time points of the cardiac ultrasound image; extracting the signal-to-noise ratio of the cardiac ultrasound image; maximizing the similarities of the features of similar samples through contrastive learning to obtain optimized cardiac contour clarity, echo uniformity, cardiac valve clarity, left ventricular wall motion consistency and signal-to-noise ratio; performing image quality evaluation based on the optimized cardiac contour clarity, echo uniformity, cardiac valve clarity, left ventricular wall motion consistency and signal-to-noise ratio, and generating a shooting suggestion according to the evaluation result; wherein the loss function for maximizing the similarities of the features of similar samples is: ， wherein, and denote different sample indices, denote a sample and a sample the inner product of the feature vectors and, and are the norms of the feature vectors and, respectively.

2. The cardiac ultrasound imaging recommendation generation method of claim 1, wherein, The extraction of the multi-scale spatial features of the cardiac ultrasound image and the obtaining of the multi-scale spatial fusion features by fusing the multi-scale spatial features comprise: using a multi-scale convolutional neural network to extract spatial features of the cardiac ultrasound image, using multiple different size convolution kernels to obtain feature maps extracted by different convolution kernels, and then fusing the feature maps of different scales to obtain fused feature maps.

3. The cardiac ultrasound imaging recommendation generation method of claim 1, wherein, The calculation formula for performing spatial attention enhancement and channel attention enhancement on the multi-scale spatial fusion features is: ， ， wherein, denotes the importance degree of the pixel at each position, denotes the multi-scale spatial fusion feature, denotes the weight distribution of the pixels at different positions when calculating the importance, denotes the convolution operation, which captures the correlation information between different positions in the image by convolution with the weight matrix , and is a bias term for adjusting the offset of the calculation result; denotes the importance degree of the pixel corresponding to each channel, is a weight vector of channel attention, denotes the dot product operation, is an activation function.

4. The cardiac ultrasound imaging recommendation generation method of claim 1, wherein, using a Canny edge detection method to detect the spatial attention enhanced multi-scale spatial fusion features to identify the heart cavity boundary, and obtaining the cardiac contour clarity.

5. The cardiac ultrasound imaging recommendation generation method of claim 1, wherein, The spatial attention enhanced multi-scale spatial fusion features are divided into multiple regions, the standard deviation and the average value of the pixel intensity in each region are calculated respectively, and the echo uniformity is obtained based on the standard deviation and the average value of the pixel intensity in each region.

6. A cardiac ultrasound imaging recommendation generation system, characterized in that, comprise: a data acquisition module configured to obtain cardiac ultrasound image data; a multi-scale feature extraction module configured to extract multi-scale spatial features of the cardiac ultrasound image, and obtain multi-scale spatial fusion features by fusing the multi-scale spatial features; a multi-scale feature enhancement module configured to perform spatial attention enhancement and channel attention enhancement on the multi-scale spatial fusion features to obtain spatial attention enhanced and channel attention enhanced multi-scale spatial fusion features; The feature extraction module is configured to calculate cardiac contour sharpness and echo uniformity based on multi-scale spatial fusion features enhanced by spatial attention, and calculate cardiac valve sharpness based on multi-scale spatial fusion features enhanced by channel attention; extract features of different time points of the cardiac ultrasound image, and calculate left ventricular wall motion consistency based on the features of different time points of the cardiac ultrasound image; extract a signal-to-noise ratio of the cardiac ultrasound image; and maximize similarities of respective features of similar samples through contrast learning to obtain optimized cardiac contour sharpness, echo uniformity, cardiac valve sharpness, left ventricular wall motion consistency, and signal-to-noise ratio. The loss function for maximizing the similarities of respective features of similar samples is as follows: ， wherein, and denote different sample indices, denote sample and sample the inner product of the feature vectors and, and are the norms of the feature vectors and, respectively; The shooting suggestion generation module is configured to perform image quality evaluation based on the optimized cardiac contour sharpness, echo uniformity, cardiac valve sharpness, left ventricular wall motion consistency, and signal-to-noise ratio, and generate a shooting suggestion according to an evaluation result.

7. A computer readable storage medium having stored thereon a computer program, characterized in that The program, when executed by the processor, implements the steps in the cardiac ultrasound image shooting suggestion generation method according to any one of claims 1-5.

8. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor, when executing the program, implements the steps in the cardiac ultrasound image shooting suggestion generation method according to any one of claims 1-5.

Images