Myocardial strain assessment method and system based on two-dimensional spot tracking

Abstract

The invention provides a myocardial strain assessment method and system based on two-dimensional spot tracking, and relates to the technical field of medical image processing.The method comprises the steps that a two-dimensional spot image sequence of a target user is obtained through scanning of ultrasonic equipment, interpolation compensation is conducted on the two-dimensional spot image sequence in combination with a generative adversarial network, and a myocardial strain assessment result is obtained; obtaining a compensated spot image sequence; performing myocardial strain analysis according to the compensated spot image sequence to obtain strain evaluation data; based on the physical sign data of the target user, the real-time state monitoring data, the ultrasonic equipment attribute characteristics and the two-dimensional spot tracking parameters, strain measurement error analysis is carried out, and adaptive strain measurement errors are obtained; and expanding the strain evaluation data according to the adaptive strain measurement error, and outputting a strain evaluation data interval as a myocardial strain evaluation result of the target user. The technical problem that in the prior art, accurate assessment of myocardial strain cannot be achieved when the frame frequency of a two-dimensional spot tracking image is low is solved.

Description

Technical Field

[0001] This invention relates to the field of medical image processing, and in particular to a method and system for assessing myocardial strain based on two-dimensional speckle tracking. Background Technology

[0002] Obese patients often exhibit lower image frame rates in two-dimensional speckle tracking ultrasound examinations due to their thick subcutaneous fat layer and pericardial fatty infiltration, leading to insufficient accuracy and reliability in assessment. Furthermore, traditional methods cannot accurately assess myocardial strain at low frame rates in two-dimensional speckle tracking images. Summary of the Invention

[0003] This invention addresses the technical problem in the prior art that it is impossible to accurately assess myocardial strain when the frame rate of two-dimensional speckle tracking images is low, and provides a method and system for assessing myocardial strain based on two-dimensional speckle tracking.

[0004] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, the present invention provides a method for assessing myocardial strain based on two-dimensional speckle tracking, comprising: Two-dimensional speckle image sequences of the target user are obtained by scanning with an ultrasound device, and the two-dimensional speckle image sequences are interpolated and compensated by a generative adversarial network to obtain a compensated speckle image sequence. Myocardial strain analysis was performed based on the compensated speckle image sequence to obtain strain assessment data; Based on the target user's vital signs data, real-time status monitoring data, and the ultrasonic equipment's attribute characteristics and two-dimensional spot tracking parameters, strain measurement error analysis is performed to obtain the appropriate strain measurement error. The strain assessment data is expanded based on the adaptive strain measurement error, and the strain assessment data range is output as the myocardial strain assessment result for the target user.

[0005] Secondly, the present invention provides a myocardial strain assessment system based on two-dimensional speckle tracking, comprising: The data acquisition module is used to acquire a two-dimensional speckle image sequence of the target user through ultrasound scanning, and to perform interpolation compensation on the two-dimensional speckle image sequence by combining a generative adversarial network to obtain a compensated speckle image sequence. The strain assessment module is used to perform myocardial strain analysis based on the compensated speckle image sequence and obtain strain assessment data. The error acquisition module is used to perform strain measurement error analysis based on the target user's vital signs data, real-time status monitoring data, ultrasonic equipment attribute characteristics and two-dimensional spot tracking parameters, and to obtain the appropriate strain measurement error. An extended output module is used to extend the strain assessment data according to the adaptive strain measurement error, and output the strain assessment data range as the myocardial strain assessment result for the target user.

[0006] The beneficial effects of this invention are: Compared to existing technologies, this application first acquires a two-dimensional speckle image sequence of the target user through ultrasound scanning. Then, it uses a generative adversarial network (GAN) to interpolate and compensate this sequence, resulting in a compensated speckle image sequence. For specific populations such as obese patients, the ultrasound scanning parameters are adjusted to ensure image coverage. The GAN is then used to specifically compensate for low frame rate defects, improving the integrity of myocardial motion images in low frame rate scenarios and providing a reliable image data foundation for subsequent strain assessment. Secondly, myocardial strain analysis is performed based on the compensated speckle image sequence to obtain strain assessment data. Image features are transformed into quantitative data indicators. These data reflect the deformation capacity and dynamic characteristics of the myocardium during contraction / diastole from different dimensions, providing objective and sensitive quantitative evidence for clinical assessment of myocardial function. Thirdly, strain measurement error analysis is performed based on the target user's vital sign data, real-time status monitoring data, ultrasound equipment attribute characteristics, and two-dimensional speckle tracking parameters to obtain an adaptive strain measurement error. By integrating user-specific characteristics, equipment parameters, and interpolation process errors, an adaptive strain measurement error that reflects the comprehensive error throughout the target user's examination process is obtained, providing reliable support for subsequent myocardial strain assessment. Finally, the strain assessment data is expanded based on the adaptive strain measurement error, and the strain assessment data range is output as the myocardial strain assessment result for the target user. The single value is expanded into a range, which reflects the personalized error characteristics and makes the myocardial strain assessment result more in line with clinical practice.

[0007] Through the above technical solution, this application uses a generative adversarial network to perform image interpolation compensation for missing frame rates during the examination of obese patients. Based on the compensation result, myocardial strain is assessed, improving the accuracy of strain measurement analysis. Furthermore, it combines the patient's vital signs data, real-time status monitoring data during the examination process, ultrasound equipment attributes, and two-dimensional speckle tracking parameters to perform strain measurement error analysis, obtaining an adaptive strain measurement error. Finally, the measurement analysis results are expanded based on the adaptive strain measurement error to generate a measurement data interval as the myocardial strain assessment result. Thus, the accuracy and reliability of myocardial strain assessment are improved when the frame rate of the two-dimensional speckle tracking image is low. Attached Figure Description

[0008] Figure 1 A schematic flowchart of a myocardial strain assessment method based on two-dimensional speckle tracking provided by the present invention; Figure 2 This is a schematic diagram of the structure of a myocardial strain assessment system based on two-dimensional speckle tracking provided by the present invention.

[0009] In the attached diagram, the components represented by each number are as follows: Data acquisition module 11, strain assessment module 12, error acquisition module 13, and extended output module 14. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0012] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0013] Example 1, as Figure 1 As shown, this embodiment of the invention provides a method for assessing myocardial strain based on two-dimensional speckle tracking, including: S10: Obtain a two-dimensional speckle image sequence of the target user by scanning with an ultrasound device, and perform interpolation compensation on the two-dimensional speckle image sequence using a generative adversarial network to obtain a compensated speckle image sequence.

[0014] When acquiring two-dimensional speckle image sequences of a target user through ultrasound scanning, due to individual differences among users (especially obese patients), it is necessary to specifically increase two-dimensional speckle tracking parameters such as scanning depth and scanning sector width in order to penetrate the thickened subcutaneous fat layer and completely cover the heart area. However, such parameter adjustments will consume more computing resources of the ultrasound equipment, resulting in a decrease in image frame rate, which in turn leads to the omission of details of myocardial motion.

[0015] To address the aforementioned issues, this application obtains a two-dimensional speckle image sequence of the target user through ultrasound scanning, and then performs interpolation compensation on the two-dimensional speckle image sequence using a generative adversarial network to obtain a compensated speckle image sequence.

[0016] Specifically, step S10 in the method includes: Configure two-dimensional speckle tracking parameters based on the target user's vital signs data, wherein the vital signs data include the subcutaneous fat thickness of the heart region, heart size, and heart wall thickness, and the two-dimensional speckle tracking parameters include scan depth, number of scan lines, and scan sector width; Using an ultrasound device, the target user's heart is scanned according to the two-dimensional speckle tracking parameters to obtain a two-dimensional speckle image sequence of the heart region. By combining a generative adversarial network, image interpolation compensation is performed on the low-speed motion phase of the two-dimensional speckle image sequence, and a compensated speckle image sequence is output.

[0017] In this embodiment, two-dimensional speckle tracking parameters are first configured based on the target user's vital signs data. These vital signs data include the thickness of subcutaneous fat in the heart region, heart size, and heart wall thickness. The two-dimensional speckle tracking parameters include scanning depth, number of scan lines, and scanning sector width. For example, obese patients with thick subcutaneous fat (e.g., exceeding 2 cm) require an increased scanning depth to ensure ultrasound waves penetrate the fat layer and reach the heart. Larger heart sizes (e.g., ventricular enlargement due to hypertension) require an increased scanning sector width (e.g., from 60° to 90°) to fully cover the heart region. Increased heart wall thickness (e.g., myocardial hypertrophy) requires adjusting the number of scan lines (e.g., increasing line density) to improve image resolution and capture the motion details of the thickened myocardium. Thus, configuring two-dimensional speckle tracking parameters based on the target user's vital signs data can balance frame rate while ensuring image coverage and penetration.

[0018] Secondly, using ultrasound equipment, a cardiac scan is performed on the target user according to the aforementioned two-dimensional speckle tracking parameters to obtain a two-dimensional speckle image sequence of the cardiac region. For example, using ultrasound equipment, a cardiac scan is performed on the target user according to the configured two-dimensional speckle tracking parameters to obtain a two-dimensional speckle image sequence of the cardiac region. These images record the motion trajectory of the myocardium during the cardiac cycle, such as rapid motion during systole and slow motion during diastole. However, for target users with specific physical characteristics (such as obese patients), adjustments are made to the scan depth and scan sector width during the scan to accommodate their subcutaneous fat thickness, heart size, and other physical characteristics. Although increasing the number of scan lines can improve image resolution, the frame rate performance of the ultrasound equipment is negatively correlated with the scan depth and sector width. Furthermore, increasing the number of scan lines further consumes computational resources, resulting in the frame rate failing to return to the ideal level after parameter adjustments. Ultimately, this leads to excessively large frame intervals in the low-speed motion phase (such as end-diastole) of the two-dimensional speckle image sequence, resulting in incomplete recording of motion details.

[0019] Finally, using a generative adversarial network (GAN), image interpolation compensation is performed on the low-speed motion phase of the two-dimensional speckle image sequence to output a compensated speckle image sequence. For example, images in the low-speed motion phase are prone to losing motion details (such as subtle deformations of the myocardium) due to large frame intervals. The GAN can fill in the missing frames, ultimately outputting a more complete compensated speckle image sequence. Image interpolation compensation is only performed in the low-speed motion phase because forced interpolation in the high-speed motion phase would introduce uncontrollable errors, destroying the biomechanical realism of the strain curve.

[0020] Specifically, the step of "combining a generative adversarial network to perform image interpolation compensation on the low-speed motion phase of the two-dimensional speckle image sequence and outputting a compensated speckle image sequence" includes: Cardiac motion analysis is performed sequentially on adjacent images in the two-dimensional blob image sequence, and two-dimensional blob images with cardiac motion velocity less than the motion velocity threshold are set as images to be compensated, thus dividing the sequence into image sequences that do not need compensation and image sequences to be compensated. Using the ultrasonic equipment attribute characteristics and the two-dimensional speckle tracking parameters as equipment constraints, and the target user's vital sign data as conditional constraints, multiple sample speckle image sequences in the low-speed motion phase are retrieved and obtained. The historical speckle image sequences after interpolation compensation based on the standard image frame rate are also obtained, resulting in multiple sample compensated speckle image sequences. Using the multiple sample blot image sequences as input and the multiple sample compensated blot image sequences as supervision, with the myocardial incompressibility loss function as training constraint, the generative adversarial network is trained until convergence to obtain the image interpolation compensation plugin. Using the image interpolation compensation plugin, image compensation is performed on the image sequence to be compensated to obtain a compensated image sequence. The compensated image sequence and the image sequence without compensation are then stitched together to obtain a compensated speckle image sequence.

[0021] In this embodiment, cardiac motion analysis is first performed sequentially on adjacent images in the two-dimensional speckle image sequence. Two-dimensional speckle images with cardiac motion velocities less than a motion velocity threshold are designated as images to be compensated, thus dividing the sequence into a sequence of images requiring no compensation and a sequence of images requiring compensation. The motion velocity threshold can be set based on information such as the target user's gender, age, health status, and vital signs; for example, 3 cm / s. Those skilled in the art can dynamically adjust this threshold according to actual conditions. For example, cardiac motion analysis is performed sequentially on adjacent images in the two-dimensional speckle image sequence. For instance, the displacement velocity of myocardial pixels can be calculated using optical flow as the cardiac motion velocity. Two-dimensional speckle images with cardiac motion velocities less than the motion velocity threshold (e.g., 3 cm / s) are marked as images requiring compensation, while the remaining two-dimensional speckle images are images requiring no compensation. This results in a sequence of images requiring no compensation and a sequence of images requiring compensation. Dividing the two-dimensional speckle image sequence into these sequences based on the motion velocity threshold avoids blind interpolation of all two-dimensional speckle images, allowing processing only on images requiring compensation where information is easily lost, thereby improving efficiency.

[0022] Secondly, using the ultrasound equipment's attributes (such as years of use, probe frequency, brand and model) and the two-dimensional speckle tracking parameters (including scan depth, number of scan lines, and scan sector width) as equipment constraints, and the target user's vital signs data (including subcutaneous fat thickness in the cardiac region, heart size, and heart wall thickness) as conditional constraints, multiple sample speckle image sequences during the low-speed motion phase are retrieved. Furthermore, historical speckle image sequences are obtained after interpolation compensation (e.g., using traditional optical flow methods) based on a standard image frame rate (the higher image frame rate of the same ultrasound equipment with the same attributes and characteristics under the two-dimensional speckle tracking parameters), resulting in multiple sample-compensated speckle image sequences. The equipment constraints ensure that the acquired sample data matches the target user's scanning conditions, while the conditional constraints ensure that the acquired sample data is similar to the target user's physiological characteristics.

[0023] Next, using the multiple sample blob image sequences as input and the multiple sample compensated blob image sequences as supervision, and with the myocardial incompressibility loss function as the training constraint, the generative adversarial network is trained until convergence to obtain the image interpolation compensation plugin. The myocardial incompressibility loss function is used as the training constraint because the physiological characteristics of myocardial tissue determine that its volume is approximately constant (incompressible) during motion. By calculating the volume deviation between the generated image and the real image, the generator is constrained to output an image sequence that conforms to this rule, avoiding the generation of images with abnormally large or small myocardial volume, and ensuring the physiological rationality of the interpolation results.

[0024] For example, a generative adversarial network mainly includes a generator and a discriminator. The generator takes a low frame rate sample blob image sequence as input and outputs an interpolated high frame rate image sequence. The discriminator takes a sample-compensated blob image sequence or an image sequence output by the generator as input and outputs a probability value of 0 to 1 to determine the authenticity of the input sequence. The generator mainly consists of an input embedding layer, a 3D convolutional layer, a self-attention module, and an upsampling and frame generation layer. The input embedding layer converts a single frame image into a feature vector and incorporates device parameters (such as scan depth and sector width) and vital sign data (such as fat thickness) as constraints to ensure that the generated image is adapted to a specific scene. The 3D convolutional layer extracts the spatiotemporal features of the sequence through 3D convolutional kernels (size such as 3×3×3, with the first two dimensions being spatial and the third dimension being temporal) to capture the dynamic patterns of myocardial motion. The self-attention module focuses on key motion regions (such as the endocardial boundary) through an attention mechanism to improve the accuracy of local detail generation for subtle deformations in the low-speed motion phase (such as the slow stretching of the myocardium at end-diastole). The upsampling and frame generation layer upsamples the feature map in the temporal dimension through transposed convolution (such as expanding frames with an interval of 100ms to an interval of 25ms), and finally generates missing intermediate frames, which are then spliced ​​into a complete high frame rate sequence. The discriminator mainly consists of a multi-scale input layer, a 3D convolution and pooling layer, and a fully connected output layer. The multi-scale input layer downsamples the input sequence to different resolutions and performs discrimination accordingly, ensuring that the generated sequence conforms to the real features at different scales. The 3D convolution and pooling layer extracts multi-scale spatiotemporal features through 3D convolution and combines max pooling to compress the feature dimension, capturing the global motion consistency of the sequence. The fully connected output layer outputs a single-channel probability value to determine the authenticity of the input sequence.

[0025] For example, during training, an adversarial loss-driven game is used between the generator and discriminator, alternately optimizing their parameters to make the generated sequence approximate the true distribution. The myocardial incompressibility loss function is used as a constraint to measure the rate of change of myocardial volume in the generated image, ensuring compliance with the physiological characteristic that the volume of myocardial tissue remains approximately constant during movement. Training terminates when the composite loss function converges. Specifically, training can be performed using the following technical path: 1. Data preparation: Multiple sample blob image sequences and corresponding multiple sample compensated blob image sequences are divided into training, validation, and test sets according to a 7:1.5:1.5 ratio. 2. Model Training: Randomly initialize the network parameters of the generator and discriminator. Fix the generator parameters and minimize the adversarial loss through gradient descent to improve the discriminator's ability to distinguish between real and fake sequences. Fix the discriminator parameters and minimize the composite loss through gradient descent so that the sequence generated by the generator can both fool the discriminator and approximate the pixel distribution and physiological characteristics of the real sequence. Alternately optimize the parameters of the generator and discriminator until the change in the loss function is less than 0.001 for 10 consecutive rounds, which is considered convergence, resulting in the image interpolation compensation plugin. The image interpolation compensation plugin can efficiently interpolate the image sequence to be compensated for the target user, outputting a compensated image sequence that meets high frame rate standards and has medical rationality.

[0026] Finally, the image interpolation compensation plugin is used to perform image compensation on the image sequence to be compensated, resulting in a compensated image sequence. The compensated image sequence and the image sequence without compensation are then concatenated to obtain a compensated speckle image sequence. For example, a pre-trained image interpolation compensation plugin is used to process the target user's image sequence to be compensated, generating missing frames to obtain the compensated image sequence. The compensated image sequence is then concatenated with the image sequence without compensation in chronological order to form a complete compensated speckle image sequence. The compensated speckle image sequence has a higher frame rate, such as increasing it from 30 frames / second to 120 frames / second, which can fully reflect the motion details of the myocardium throughout the cardiac cycle.

[0027] In summary, compared to existing technologies, this application acquires a two-dimensional speckle image sequence of the target user through ultrasound scanning, and then uses a generative adversarial network (GAN) to interpolate and compensate the two-dimensional speckle image sequence to obtain a compensated speckle image sequence. Thus, for the vital signs of special populations such as obese patients, the ultrasound scanning parameters are first adjusted to ensure image coverage, and then the GAN is used to specifically compensate for low frame rate defects, improving the integrity of myocardial motion images in low frame rate scenarios and providing a reliable image data foundation for subsequent strain assessment.

[0028] S20: Perform myocardial strain analysis based on the compensated speckle image sequence to obtain strain assessment data.

[0029] The aforementioned steps yielded a compensated speckle image sequence after interpolation compensation, which can be used to accurately analyze the myocardial strain assessment data of the target user, quantify core myocardial function indicators, and provide objective data support for myocardial function assessment.

[0030] To address the aforementioned issues, this application performs myocardial strain analysis based on the compensated speckle image sequence to obtain strain assessment data including longitudinal strain values, circumferential strain values, radial strain values, and strain rates.

[0031] Specifically, step S20 in the method includes: The strain assessment data includes longitudinal strain values, circumferential strain values, radial strain values, and strain rates.

[0032] In this embodiment, natural spots (texture features formed by echo signals of fine structures within the myocardium) in the compensated speckled image sequence are used as natural markers. By tracking the spatial position changes of these spots in the compensated sequence, the degree of deformation (i.e., strain) of the myocardial tissue in different directions is calculated to obtain strain assessment data, thereby quantifying the assessment of myocardial contractile / diastolic function. The strain assessment data includes longitudinal strain values, circumferential strain values, radial strain values, and strain rates. Specifically: The longitudinal strain value refers to the degree of deformation of myocardial tissue along the long axis of the heart (i.e., from the base of the heart to the apex). It is expressed as a percentage of deformation by tracking the stretching (diastole) and shortening (systole) of the subendocardial myocardial fibers along the long axis. For example, a negative longitudinal strain value during systole represents shortening. The more negative the value, the stronger the contractile ability. The longitudinal strain value can reflect the function of the subendocardial myocardium and is a sensitive indicator for assessing the overall and local function of the left ventricle.

[0033] Among them, the circumferential strain value refers to the degree of deformation of myocardial tissue along the short axis section (the cross section perpendicular to the long axis) around the heart chamber. By tracking the contraction (circumferential shortening) and relaxation (circumferential stretching) of the myocardial circumferential fibers, the ability of the myocardium to squeeze blood around the heart chamber is quantified. The circumferential strain value is more sensitive to the assessment of the function of the myocardial middle layer fibers and can be used to determine the abnormal contraction coordination caused by local myocardial ischemia or fibrosis.

[0034] Radial strain value refers to the degree of radial deformation of myocardial tissue along the short axis of the heart from the endocardium (near the heart chamber) to the epicardium (near the pericardium), that is, the degree of myocardial thickening (systole) or thinning (diastole). Radial strain value can reflect the thickening capacity of myocardium during contraction and is closely related to the ejection function of the left ventricle. A decrease in local radial strain often indicates that the corresponding segment of myocardial function is impaired.

[0035] Among them, strain rate refers to the rate of change of strain over time, that is, the speed of myocardial deformation. It is obtained by calculating the ratio of the difference in strain values ​​between adjacent frames to the time interval. It can more sensitively reflect the instantaneous dynamic changes of myocardial motion (such as the rapid deformation rate in the early stage of systole and the filling rate in the early stage of diastole). Strain rate can distinguish the mechanical characteristics of active contraction and passive stretching of myocardium, which is of great significance for the early detection of myocardial dysfunction (such as diastolic dysfunction).

[0036] In summary, compared to existing technologies, this application performs myocardial strain analysis based on the compensated speckle image sequence to obtain strain assessment data including longitudinal strain values, circumferential strain values, radial strain values, and strain rates. In this way, image features are transformed into quantitative data indicators, which reflect the deformation capacity and dynamic characteristics of the myocardium during contraction / relaxation from different dimensions, providing objective and sensitive quantitative evidence for clinical assessment of myocardial function.

[0037] S30: Based on the target user's vital signs data, real-time status monitoring data, and the characteristics of ultrasonic equipment attributes and two-dimensional spot tracking parameters, strain measurement error analysis is performed to obtain the appropriate strain measurement error.

[0038] When target users perform scanning inspections, factors such as individual characteristics, equipment parameters, and interpolation process errors can all affect the accuracy of the detection data. In order to improve the accuracy of the evaluation, measurement error analysis is required.

[0039] To address the aforementioned issues, this application analyzes strain measurement error based on the target user's vital signs data, real-time status monitoring data, ultrasonic equipment attribute characteristics, and two-dimensional spot tracking parameters to obtain the appropriate strain measurement error.

[0040] Specifically, step S30 in the method includes: While the target user is being scanned, real-time status monitoring data of the target user is simultaneously acquired, including heart rate, blood pressure and respiratory rate. Based on the vital signs data, real-time status monitoring data, ultrasonic equipment attribute characteristics, and two-dimensional spot tracking parameters, strain measurement error is predicted, and the predicted strain measurement error is output. The interpolation compensation error of the image interpolation compensation plugin is tested, and the image measurement error is derived based on the interpolation compensation error. An error correction coefficient is obtained by adding 1 to the image measurement error, and the predicted strain measurement error is adjusted to obtain the adaptive strain measurement error.

[0041] In this embodiment, when the target user is undergoing a scanning examination, real-time status monitoring data of the target user is simultaneously acquired. The status monitoring data includes heart rate, blood pressure, and respiratory rate. This is because these status monitoring data directly affect the accuracy of strain measurement. An excessively fast heart rate may lead to an accelerated rhythm of myocardial movement, increasing motion artifact errors in speckle tracking. Fluctuations in blood pressure may change myocardial load, affecting the stability of strain values. Abnormal respiratory rate may cause the heart to move up and down with respiration, introducing additional displacement measurement errors. Thus, capturing these dynamic factors provides real-time physiological data for subsequent error prediction.

[0042] Secondly, based on the vital signs data, real-time status monitoring data, ultrasound equipment attribute characteristics, and two-dimensional speckle tracking parameters, a pre-trained strain measurement error predictor is used to predict the strain measurement error and output the predicted strain measurement error. The predicted strain measurement error reflects the measurement deviation caused by individual user differences, equipment characteristics, and scanning parameters when there is no interpolation compensation. For example, signal attenuation error caused by thick fat in obese patients and loss of motion details caused by low frame rate of the equipment.

[0043] Next, the interpolation compensation error of the image interpolation compensation plugin is tested and obtained, and the image measurement error is derived based on the interpolation compensation error. For example, the intermediate frames generated by the image interpolation compensation plugin may have pixel-level deviations. Based on the average pixel distance of the real data, the interpolation compensation error of the image interpolation compensation plugin is tested and obtained. For example, if the average pixel distance between the frame generated by the image interpolation compensation plugin and the real frame is 0.3mm, then 0.3mm is taken as the interpolation compensation error, and the image measurement error is then derived based on the interpolation compensation error.

[0044] Finally, an error correction coefficient is obtained by adding 1 to the image measurement error, and the predicted strain measurement error is adjusted to obtain the adapted strain measurement error. Here, the error correction coefficient = 1 + image measurement error, and the adapted strain measurement error = predicted strain measurement error * error correction coefficient. For example, if the calculated image measurement error is 0.5% and the predicted strain measurement error is 2%, then the error correction coefficient = 1 + 0.5% = 1.005, and the adapted strain measurement error = 2% * 1.005 = 2.01%. This comprehensively considers the errors caused by changes in vital signs during the target user's scanning and the errors introduced by the image interpolation compensation plugin, reflecting the overall error level throughout the process.

[0045] Specifically, the step of "predicting strain measurement error based on the vital signs data, real-time status monitoring data, ultrasonic equipment attribute characteristics, and two-dimensional spot tracking parameters, and outputting the predicted strain measurement error" includes: Using the characteristics of ultrasound equipment as equipment constraints and myocardial strain detection as a guide, we obtained sample vital signs dataset, sample state dataset, and sample spot tracking parameter set based on big data retrieval, and analyzed the sample strain measurement error set. Using the sample vital signs dataset, sample state dataset, and sample blot tracking parameter set as inputs, and the sample strain measurement error set as supervision, a deep learning model is trained until convergence to construct a strain measurement error predictor. The strain measurement error predictor is used to analyze the vital signs data, real-time condition monitoring data, and two-dimensional spot tracking parameters to obtain the predicted strain measurement error.

[0046] In this embodiment, the ultrasound equipment's attributes (such as years of use, probe frequency, brand and model) are used as equipment constraints. Guided by myocardial strain detection, a sample vital signs dataset (such as user data on different subcutaneous fat thicknesses, heart sizes and heart wall thicknesses), a sample status dataset (such as heart rate, blood pressure and respiratory rate data when the corresponding sample is scanned), and a sample speckle tracking parameter set (such as the scanning depth, number of scan lines and scanning sector width data of the corresponding sample) are obtained by analyzing the error between the sample strain measurement value and the standard value.

[0047] Secondly, using the sample vital signs dataset, sample state dataset, and sample blot tracking parameter set as inputs, and the sample strain measurement error set as supervision, a deep learning model is trained until convergence to construct a strain measurement error predictor. For example, a deep learning model can employ a regressive neural network. This network learns the nonlinear mapping between input features and the supervised target, outputting the predicted strain measurement error. It mainly consists of an input layer, hidden layers, and an output layer. The input layer receives sample vital signs data, sample state data, and sample blob tracking parameters. These input features are standardized, such as through Z-score standardization, to eliminate the impact of dimensional differences on model training. The number of neurons in the input layer equals the total dimension of the preprocessed features. The hidden layers learn the complex relationship between input features and sample strain measurement errors through multiple nonlinear transformations. Three to five fully connected layers can be used, with the number of layers and neurons adjusted according to the data scale. Shallow layers focus on basic feature combinations, while deeper layers focus on higher-order feature mappings. Nonlinear activation functions such as ReLU and Leaky ReLU are added after each layer. Batch normalization accelerates training convergence and alleviates internal covariate bias. The Dropout layer randomly discards some neurons to prevent the model from overfitting the training samples. The output layer outputs the predicted strain measurement error. Since the strain measurement error is a continuous value, the output layer consists of a single neuron using a linear activation function to directly output the predicted strain measurement error.

[0048] For example, the strain measurement error predictor can be trained using the following technical path: 1. Data preparation: Divide the sample vital signs dataset, sample state dataset, sample blob tracking parameter set, and corresponding sample strain measurement error set into a training set and a validation set in an 8:2 ratio. 2. Model training: Using the sample vital signs dataset, sample state dataset, and sample blob tracking parameter set as input, and the sample strain measurement error set as supervision, calculate the difference between the predicted strain measurement error and the actual sample strain measurement error using mean squared error (MSE), and use this as the loss function. Use the Adam optimization algorithm to dynamically adjust the learning rate to accelerate model convergence. When the validation set loss no longer decreases (e.g., no improvement for 10 consecutive rounds) or reaches the preset number of iterations, it is considered converged, training stops, and the trained strain measurement error predictor is obtained.

[0049] Finally, the strain measurement error predictor is used to analyze the vital signs data, real-time status monitoring data, and two-dimensional speckle tracking parameters to obtain the predicted strain measurement error. For example, a pre-trained strain measurement error predictor is invoked, inputting the target user's vital signs data (e.g., fat thickness 3cm), real-time status monitoring data (e.g., heart rate 95 bpm), and two-dimensional speckle tracking parameters (e.g., scan depth 8cm), and outputting the predicted strain measurement error, such as 2%.

[0050] Furthermore, the phrase "deriving the image measurement error based on the interpolation compensation error" includes: Using error propagation theory, the displacement error is derived from the interpolation compensation error. Based on the displacement error, an influence analysis of strain measurement is performed, and the measurement error of the output image is generated.

[0051] In this embodiment, the displacement error is first derived from the interpolation compensation error using error propagation theory. Error propagation theory is a theoretical method for studying how measurement errors are transmitted from the input to the output through a functional relationship. For example, displacement error = interpolation compensation error * scaling factor. The scaling factor is determined by the image resolution; for example, if 1 pixel = 0.1 mm, then the scaling factor is 0.1 mm. For instance, if the interpolation compensation error is 0.3 mm, then the corresponding displacement error = 0.3 * 0.1 = 0.03 mm.

[0052] Secondly, an impact analysis on strain measurement is performed based on the displacement error, and the image measurement error is output. For example, the essence of myocardial strain is calculating the relative displacement change of the myocardium: myocardial strain = displacement change / original length. Therefore, displacement error directly affects the accuracy of strain measurement. For instance, in longitudinal strain measurement, if the original length of the myocardium is 50mm, and a displacement error of 0.03mm is caused by the image interpolation compensation plugin, then the longitudinal image measurement error = 0.03mm / 50mm = 0.06%. The circumferential and radial image measurement errors are calculated using the same method. Combining the longitudinal, circumferential, and radial image measurement errors, the final image measurement error is 0.5%. The image measurement error reflects the strain error introduced by the interpolation compensation process.

[0053] In summary, compared to existing technologies, this application analyzes strain measurement errors based on the target user's vital signs data, real-time status monitoring data, ultrasound equipment attribute characteristics, and two-dimensional speckle tracking parameters to obtain adaptive strain measurement errors. Thus, by integrating user-specific characteristics, equipment parameters, and interpolation process errors, an adaptive strain measurement error reflecting the comprehensive error throughout the target user's examination process is obtained, providing reliable support for subsequent myocardial strain assessment.

[0054] S40: Expand the strain assessment data according to the adaptive strain measurement error, and output the strain assessment data range as the myocardial strain assessment result for the target user.

[0055] Myocardial strain assessment data is a quantitative indicator calculated through ultrasound speckle tracking. However, its measurement process is affected by various factors, such as fat interference in obese patients, equipment parameter adjustments, and interpolation compensation errors, inevitably introducing a certain degree of uncertainty. Outputting only a single value may mask the fluctuations caused by these errors, leading to clinical misjudgment. Therefore, based on the adaptive strain measurement error obtained from the aforementioned steps, which reflects the comprehensive error throughout the target user's examination process, the single strain assessment data can be expanded into an interval range. This range includes both the core measurement value and the possible fluctuation range caused by errors, more accurately reflecting the actual state of myocardial function.

[0056] In this embodiment, a single strain assessment data point can be expanded into an interval range by using strain assessment data ± adaptive strain measurement error. For example, if the longitudinal strain value of the target user is -18% and the adaptive strain measurement error is 2.01%, the strain assessment data is expanded according to the adaptive strain measurement error, i.e., -18% ± 2.01% = -20.01% to -15.99%, to obtain the longitudinal strain assessment data interval. The same method is used to expand other strain assessment data to obtain the myocardial strain assessment result of the target user.

[0057] In summary, compared to existing technologies, this application expands the strain assessment data based on the adaptive strain measurement error, outputting a strain assessment data range as the myocardial strain assessment result for the target user. This expands a single value into a range, reflecting personalized error characteristics and making the myocardial strain assessment result more closely aligned with clinical practice, providing a rigorous and reliable basis for accurate assessment of myocardial function.

[0058] In summary, the embodiments of this application have at least the following technical effects: Compared to existing technologies, this application first acquires a two-dimensional speckle image sequence of the target user through ultrasound scanning, and then uses a generative adversarial network (GAN) to interpolate and compensate the two-dimensional speckle image sequence to obtain a compensated speckle image sequence. Thus, for the vital signs of special populations such as obese patients, the ultrasound scanning parameters are first adjusted to ensure image coverage, and then the GAN is used to specifically compensate for low frame rate defects, improving the integrity of myocardial motion images in low frame rate scenarios and providing a reliable image data foundation for subsequent strain assessment.

[0059] Secondly, this application performs myocardial strain analysis based on the compensated speckle image sequence to obtain strain assessment data including longitudinal strain value, circumferential strain value, radial strain value, and strain rate. In this way, image features are transformed into quantitative data indicators, which reflect the deformation capacity and dynamic characteristics of the myocardium during contraction / relaxation from different dimensions, providing objective and sensitive quantitative evidence for clinical assessment of myocardial function.

[0060] Furthermore, this application analyzes strain measurement error based on the target user's vital signs data, real-time status monitoring data, ultrasound equipment attributes, and two-dimensional speckle tracking parameters to obtain adaptive strain measurement error. Thus, by integrating user-specific characteristics, equipment parameters, and interpolation process errors, an adaptive strain measurement error reflecting the comprehensive error throughout the target user's examination process is obtained, providing reliable support for subsequent myocardial strain assessment.

[0061] Finally, this application expands the strain assessment data based on the adapted strain measurement error, outputting a strain assessment data range as the myocardial strain assessment result for the target user. In this way, expanding a single value into a range reflects personalized error characteristics, making the myocardial strain assessment result more closely aligned with clinical reality, and providing a rigorous and reliable basis for the accurate assessment of myocardial function.

[0062] Through the above technical solution, this application uses a generative adversarial network to perform image interpolation compensation for missing frame rates during the examination of obese patients. Based on the compensation result, myocardial strain is assessed, improving the accuracy of strain measurement analysis. Furthermore, it combines the patient's vital signs data, real-time status monitoring data during the examination process, ultrasound equipment attributes, and two-dimensional speckle tracking parameters to perform strain measurement error analysis, obtaining an adaptive strain measurement error. Finally, the measurement analysis results are expanded based on the adaptive strain measurement error to generate a measurement data interval as the myocardial strain assessment result. Thus, the accuracy and reliability of myocardial strain assessment are improved when the frame rate of the two-dimensional speckle tracking image is low.

[0063] Example 2, as Figure 2 As shown, based on the same inventive concept as the myocardial strain assessment method based on two-dimensional speckle tracking provided in Embodiment 1, this embodiment of the invention also provides a myocardial strain assessment system based on two-dimensional speckle tracking, comprising: Data acquisition module 11 is used to acquire a two-dimensional speckle image sequence of the target user by scanning with an ultrasound device, and to perform interpolation compensation on the two-dimensional speckle image sequence by combining a generative adversarial network to obtain a compensated speckle image sequence. The strain assessment module 12 is used to perform myocardial strain analysis based on the compensated speckle image sequence and obtain strain assessment data. Error acquisition module 13 is used to perform strain measurement error analysis based on the target user's vital signs data, real-time status monitoring data, ultrasonic equipment attribute characteristics and two-dimensional spot tracking parameters, and to obtain the appropriate strain measurement error. The extended output module 14 is used to extend the strain assessment data according to the adaptive strain measurement error and output the strain assessment data range as the myocardial strain assessment result of the target user.

[0064] Specifically, the data acquisition module 11 is used for: Configure two-dimensional speckle tracking parameters based on the target user's vital signs data, wherein the vital signs data include the subcutaneous fat thickness of the heart region, heart size, and heart wall thickness, and the two-dimensional speckle tracking parameters include scan depth, number of scan lines, and scan sector width; Using an ultrasound device, the target user's heart is scanned according to the two-dimensional speckle tracking parameters to obtain a two-dimensional speckle image sequence of the heart region. By combining a generative adversarial network, image interpolation compensation is performed on the low-speed motion phase of the two-dimensional speckle image sequence, and a compensated speckle image sequence is output.

[0065] Specifically, the step of "combining a generative adversarial network to perform image interpolation compensation on the low-speed motion phase of the two-dimensional speckle image sequence and outputting a compensated speckle image sequence" includes: Cardiac motion analysis is performed sequentially on adjacent images in the two-dimensional blob image sequence, and two-dimensional blob images with cardiac motion velocity less than the motion velocity threshold are set as images to be compensated, thus dividing the sequence into image sequences that do not need compensation and image sequences to be compensated. Using the ultrasonic equipment attribute characteristics and the two-dimensional speckle tracking parameters as equipment constraints, and the target user's vital sign data as conditional constraints, multiple sample speckle image sequences in the low-speed motion phase are retrieved and obtained. The historical speckle image sequences after interpolation compensation based on the standard image frame rate are also obtained, resulting in multiple sample compensated speckle image sequences. Using the multiple sample blot image sequences as input and the multiple sample compensated blot image sequences as supervision, with the myocardial incompressibility loss function as training constraint, the generative adversarial network is trained until convergence to obtain the image interpolation compensation plugin. Using the image interpolation compensation plugin, image compensation is performed on the image sequence to be compensated to obtain a compensated image sequence. The compensated image sequence and the image sequence without compensation are then stitched together to obtain a compensated speckle image sequence.

[0066] Specifically, the strain assessment module 12 is used for: Based on the compensated speckle image sequence, myocardial strain analysis was performed to obtain strain assessment data, which included longitudinal strain value, circumferential strain value, radial strain value, and strain rate.

[0067] Specifically, the error acquisition module 13 is used for: While the target user is being scanned, real-time status monitoring data of the target user is simultaneously acquired, including heart rate, blood pressure and respiratory rate. Based on the vital signs data, real-time status monitoring data, ultrasonic equipment attribute characteristics, and two-dimensional spot tracking parameters, strain measurement error is predicted, and the predicted strain measurement error is output. The interpolation compensation error of the image interpolation compensation plugin is tested, and the image measurement error is derived based on the interpolation compensation error. An error correction coefficient is obtained by adding 1 to the image measurement error, and the predicted strain measurement error is adjusted to obtain the adaptive strain measurement error.

[0068] Specifically, the step of "predicting strain measurement error based on the vital signs data, real-time status monitoring data, ultrasonic equipment attribute characteristics, and two-dimensional spot tracking parameters, and outputting the predicted strain measurement error" includes: Using the characteristics of ultrasound equipment as equipment constraints and myocardial strain detection as a guide, we obtained sample vital signs dataset, sample state dataset, and sample spot tracking parameter set based on big data retrieval, and analyzed the sample strain measurement error set. Using the sample vital signs dataset, sample state dataset, and sample blot tracking parameter set as inputs, and the sample strain measurement error set as supervision, a deep learning model is trained until convergence to construct a strain measurement error predictor. The strain measurement error predictor is used to analyze the vital signs data, real-time condition monitoring data, and two-dimensional spot tracking parameters to obtain the predicted strain measurement error.

[0069] Further, the phrase "deriving the image measurement error based on the interpolation compensation error" includes: Using error propagation theory, the displacement error is derived from the interpolation compensation error. Based on the displacement error, an influence analysis of strain measurement is performed, and the measurement error of the output image is generated.

[0070] Specifically, the extended output module 14 is used for: The strain assessment data is expanded based on the adaptive strain measurement error, and the strain assessment data range is output as the myocardial strain assessment result for the target user.

[0071] In summary, the embodiments of this application have at least the following technical effects: Compared to existing technologies, this application first uses a data acquisition module to acquire a two-dimensional speckle image sequence of the target user using ultrasound equipment. Then, a generative adversarial network (GAN) is used to interpolate and compensate this two-dimensional speckle image sequence, resulting in a compensated speckle image sequence. For specific populations such as obese patients, the ultrasound equipment scanning parameters are first adjusted to ensure image coverage, and then the GAN is used to specifically compensate for low frame rate defects, improving the integrity of myocardial motion images in low frame rate scenarios and providing a reliable image data foundation for subsequent strain assessment. Secondly, through a strain assessment module, myocardial strain analysis is performed based on the compensated speckle image sequence to obtain strain assessment data. This transforms image features into quantitative data indicators. These data reflect the deformation capacity and dynamic characteristics of the myocardium during contraction / relaxation from different dimensions, providing objective and sensitive quantitative evidence for clinical assessment of myocardial function. Secondly, the error acquisition module analyzes strain measurement errors based on the target user's vital signs data, real-time status monitoring data, ultrasound equipment attributes, and two-dimensional speckle tracking parameters to obtain an adaptive strain measurement error. By integrating user-specific characteristics, equipment parameters, and interpolation errors, an adaptive strain measurement error reflecting the comprehensive error throughout the target user's examination process is obtained, providing reliable support for subsequent myocardial strain assessment. Finally, the extended output module expands the strain assessment data based on the adaptive strain measurement error, outputting a strain assessment data range as the target user's myocardial strain assessment result. This expands a single value into a range, reflecting personalized error characteristics and making the myocardial strain assessment result more aligned with clinical reality. This improves the accuracy and reliability of myocardial strain assessment when the frame rate of two-dimensional speckle tracking images is low.

[0072] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0073] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0074] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0075] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0076] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0077] Although preferred embodiments of the invention have been described, those skilled in the art, once they have learned the basic inventive concept, can make other changes and modifications to these embodiments.

[0078] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for assessing myocardial strain based on two-dimensional speckle tracking, characterized in that, The methods include: Two-dimensional speckle image sequences of the target user are obtained by scanning with an ultrasound device, and the two-dimensional speckle image sequences are interpolated and compensated by a generative adversarial network to obtain a compensated speckle image sequence. Myocardial strain analysis was performed based on the compensated speckle image sequence to obtain strain assessment data; Based on the target user's vital signs data, real-time status monitoring data, and the ultrasonic equipment's attribute characteristics and two-dimensional spot tracking parameters, strain measurement error analysis is performed to obtain the appropriate strain measurement error. The strain assessment data is expanded based on the adaptive strain measurement error, and the strain assessment data range is output as the myocardial strain assessment result for the target user.

2. The method for assessing myocardial strain based on two-dimensional speckle tracking according to claim 1, characterized in that, A two-dimensional speckle image sequence of the target user is obtained by scanning with an ultrasound device. This sequence is then interpolated and compensated using a generative adversarial network (GAN) to obtain a compensated speckle image sequence, including: Configure two-dimensional speckle tracking parameters based on the target user's vital signs data, wherein the vital signs data include the subcutaneous fat thickness of the heart region, heart size, and heart wall thickness, and the two-dimensional speckle tracking parameters include scan depth, number of scan lines, and scan sector width; Using an ultrasound device, the target user's heart is scanned according to the two-dimensional speckle tracking parameters to obtain a two-dimensional speckle image sequence of the heart region. By combining a generative adversarial network, image interpolation compensation is performed on the low-speed motion phase of the two-dimensional speckle image sequence, and a compensated speckle image sequence is output.

3. The method for assessing myocardial strain based on two-dimensional speckle tracking according to claim 2, characterized in that, By incorporating a generative adversarial network, image interpolation compensation is performed on the low-speed motion phase of the two-dimensional speckle image sequence to output a compensated speckle image sequence, including: Cardiac motion analysis is performed sequentially on adjacent images in the two-dimensional blob image sequence, and two-dimensional blob images with cardiac motion velocity less than the motion velocity threshold are set as images to be compensated, thus dividing the sequence into images that do not need compensation and images that need compensation. Using the ultrasonic equipment's attribute characteristics and the two-dimensional speckle tracking parameters as equipment constraints, and the target user's vital sign data as conditional constraints, multiple sample speckle image sequences during the low-speed motion phase are retrieved and obtained. Historical speckle image sequences after interpolation compensation based on the standard image frame rate are also obtained, resulting in multiple sample compensated speckle image sequences. Using the multiple sample blot image sequences as input and the multiple sample compensated blot image sequences as supervision, with the myocardial incompressibility loss function as training constraint, the generative adversarial network is trained until convergence to obtain the image interpolation compensation plugin. Using the image interpolation compensation plugin, image compensation is performed on the image sequence to be compensated to obtain a compensated image sequence. The compensated image sequence and the image sequence without compensation are then stitched together to obtain a compensated speckle image sequence.

4. The method for assessing myocardial strain based on two-dimensional speckle tracking according to claim 1, characterized in that, The strain assessment data includes longitudinal strain values, circumferential strain values, radial strain values, and strain rates.

5. The method for assessing myocardial strain based on two-dimensional speckle tracking according to claim 1, characterized in that, Based on the target user's vital sign data, real-time status monitoring data, and the ultrasonic equipment's attribute characteristics and two-dimensional speckle tracking parameters, strain measurement error analysis is performed to obtain the adaptive strain measurement error, including: While the target user is being scanned, real-time status monitoring data of the target user is simultaneously acquired, including heart rate, blood pressure and respiratory rate. Based on the vital signs data, real-time status monitoring data, ultrasonic equipment attribute characteristics, and two-dimensional spot tracking parameters, strain measurement error is predicted, and the predicted strain measurement error is output. The interpolation compensation error of the image interpolation compensation plugin is tested, and the image measurement error is derived based on the interpolation compensation error. An error correction coefficient is obtained by adding 1 to the image measurement error, and the predicted strain measurement error is adjusted to obtain the adaptive strain measurement error.

6. The method for assessing myocardial strain based on two-dimensional speckle tracking according to claim 5, characterized in that, Based on the vital signs data, real-time status monitoring data, ultrasonic equipment attribute characteristics, and two-dimensional speckle tracking parameters, strain measurement error is predicted, and the predicted strain measurement error is output, including: Using the characteristics of ultrasound equipment as equipment constraints and myocardial strain detection as a guide, we obtained sample vital signs dataset, sample state dataset, and sample spot tracking parameter set based on big data retrieval, and analyzed the sample strain measurement error set. Using the sample vital signs dataset, sample state dataset, and sample blot tracking parameter set as inputs, and the sample strain measurement error set as supervision, a deep learning model is trained until convergence to construct a strain measurement error predictor. The strain measurement error predictor is used to analyze the vital signs data, real-time condition monitoring data, and two-dimensional spot tracking parameters to obtain the predicted strain measurement error.

7. The method for assessing myocardial strain based on two-dimensional speckle tracking according to claim 5, characterized in that, The image measurement error is derived based on the interpolation compensation error, including: Using error propagation theory, the displacement error is derived from the interpolation compensation error. Based on the displacement error, an influence analysis of strain measurement is performed, and the measurement error of the output image is generated.

8. A myocardial strain assessment system based on two-dimensional speckle tracking, characterized in that, For performing the method according to any one of claims 1-7, comprising: The data acquisition module is used to acquire a two-dimensional speckle image sequence of the target user through ultrasound scanning, and to perform interpolation compensation on the two-dimensional speckle image sequence by combining a generative adversarial network to obtain a compensated speckle image sequence. The strain assessment module is used to perform myocardial strain analysis based on the compensated speckle image sequence and obtain strain assessment data. The error acquisition module is used to perform strain measurement error analysis based on the target user's vital signs data, real-time status monitoring data, ultrasonic equipment attribute characteristics and two-dimensional spot tracking parameters, and to obtain the appropriate strain measurement error. An extended output module is used to extend the strain assessment data according to the adaptive strain measurement error, and output the strain assessment data range as the myocardial strain assessment result for the target user.

Images