Method for determining misalignment artifact recognition strategy, misalignment artifact recognition method and device

Abstract

Embodiments of the present invention disclose a method for determining a misalignment artifact recognition strategy, a misalignment artifact recognition method, and an apparatus. By presetting multiple combinations of image capture parameters, corresponding misalignment artifact recognition models are respectively trained based on each combination of image capture parameters. The misalignment artifact recognition accuracy of each misalignment artifact recognition model is verified based on misalignment artifact verification samples, and the target misalignment artifact recognition model is selected based on the obtained misalignment artifact recognition accuracy. The target misalignment artifact recognition model and the corresponding combination of image capture parameters form a misalignment artifact recognition strategy. The influencing factors of misalignment artifact recognition accuracy: window size, step size, and misalignment artifact recognition model. The misalignment artifact recognition strategy is obtained as a whole through the screening of misalignment artifact recognition accuracy. Therefore, the above factors all meet the requirements of misalignment artifact recognition accuracy, which is beneficial to ensuring the accuracy of misalignment artifact recognition for medical tomographic images to be processed.

Description

Technical Field

[0001] Embodiments of the present invention relate to image processing technologies, and in particular, to a method for determining a misregistration artifact recognition strategy, a misregistration artifact recognition method, and an apparatus therefor. Background Art

[0002] A misregistration artifact is an artifact caused by the misalignment of the axial plane from its original position due to imaging equipment errors during 3D image acquisition, resulting in a high-contrast edge and showing a sawtooth phenomenon in the coronal and sagittal planes, also known as a step artifact.

[0003] Currently, in the methods for recognizing misregistration artifacts, there is a problem of poor positioning accuracy. Summary of the Invention

[0004] The present invention provides a method for determining a misregistration artifact recognition strategy, a misregistration artifact recognition method, and an apparatus therefor, to determine a misregistration artifact recognition strategy for recognizing misregistration artifacts in medical images, and the misregistration artifact recognition strategy can improve the recognition accuracy of misregistration artifacts.

[0005] According to one aspect of the present invention, there is provided a method for determining a misregistration artifact recognition strategy, including:

[0006] Obtaining misregistration artifact training samples and multiple sets of image cropping parameter combinations including window sizes and strides;

[0007] For any image cropping parameter combination, performing window data cropping on the misregistration artifact training samples based on the window size and stride in the image cropping parameter combination, and training a misregistration artifact recognition model corresponding to the image cropping parameter combination based on the cropped window data and the misregistration artifact labels corresponding to the respective window data;

[0008] Obtaining misregistration artifact verification samples, and verifying the misregistration artifact recognition models corresponding to the respective image cropping parameter combinations based on the misregistration artifact verification samples to determine the artifact recognition accuracy rates of the respective misregistration artifact recognition models;

[0009] Determining the misregistration artifact recognition models that meet the artifact recognition accuracy rate screening conditions, and the image cropping parameter combinations corresponding to the selected misregistration artifact recognition models, as the misregistration artifact recognition strategy.

[0010] According to another aspect of the present invention, there is provided a misregistration artifact recognition method, including:

[0011] Obtaining an image to be processed;

[0012] Performing window data cropping on the image to be processed based on the window size and stride in the misregistration artifact recognition strategy to obtain a plurality of window data;

[0013] Determine the misalignment artifact description parameters corresponding to the window data based on the misalignment artifact recognition model in the misalignment artifact recognition strategy;

[0014] Determine the misalignment artifact window based on each of the misalignment artifact description parameters and the corresponding threshold, and determine the misalignment artifact position based on the misalignment artifact window.

[0015] According to another aspect of the present invention, there is provided a misalignment artifact recognition strategy determination device, including:

[0016] A data acquisition module, configured to acquire misalignment artifact training samples and multiple sets of image cropping parameter combinations including window size and stride;

[0017] A model training module, for any image cropping parameter combination, based on the window size and stride in the image cropping parameter combination, perform window data cropping on the misalignment artifact training samples, and based on the cropped window data and the misalignment artifact labels corresponding to each window data, train to obtain the misalignment artifact recognition model corresponding to the image cropping parameter combination;

[0018] A model verification module, configured to acquire misalignment artifact verification samples, verify the misalignment artifact recognition models corresponding to each image cropping parameter combination based on the misalignment artifact verification samples, and determine the artifact recognition accuracy rates of the misalignment artifact recognition models;

[0019] An identification strategy determination module, configured to determine the misalignment artifact recognition model that meets the artifact recognition accuracy rate screening condition and the image cropping parameter combination corresponding to the selected misalignment artifact recognition model as the misalignment artifact recognition strategy.

[0020] According to another aspect of the present invention, there is provided a misalignment artifact recognition device, including:

[0021] An image acquisition module, configured to acquire an image to be processed;

[0022] A window data cropping module, configured to perform window data cropping on the image to be processed based on the window size and stride in the misalignment artifact recognition strategy to obtain multiple window data;

[0023] A description parameter determination module, configured to determine the misalignment artifact description parameters corresponding to each window data based on the misalignment artifact recognition model in the misalignment artifact recognition strategy;

[0024] A misalignment artifact recognition module, configured to determine the misalignment artifact window based on each of the misalignment artifact description parameters and the corresponding threshold, and determine the misalignment artifact position based on the misalignment artifact window.

[0025] According to another aspect of the present invention, there is provided an electronic device, the electronic device includes:

[0026] At least one processor; and

[0027] A memory communicatively connected to the at least one processor; wherein

[0028] The memory stores a computer program executable by the at least one processor, and when the computer program is executed by the at least one processor, the at least one processor is enabled to execute the misalignment artifact recognition strategy determination method and / or the misalignment artifact recognition method according to any embodiment of the present invention.

[0029] According to another aspect of the present invention, there is provided a computer-readable storage medium storing computer instructions for causing a processor to implement the misalignment artifact recognition strategy determination method and / or the misalignment artifact recognition method according to any embodiment of the present invention when executed.

[0030] In the technical solution of the embodiment of the present invention, by presetting multiple combinations of image capture parameters, different combinations of image capture parameters include different window sizes and step lengths. Corresponding misalignment artifact recognition models are respectively trained based on each combination of image capture parameters. The misalignment artifact recognition accuracy of each misalignment artifact recognition model is verified based on misalignment artifact verification samples, and a target misalignment artifact recognition model is selected based on the obtained misalignment artifact recognition accuracy. The target misalignment artifact recognition model and the corresponding combination of image capture parameters form a misalignment artifact recognition strategy for recognizing misalignment artifacts in medical tomographic images to be processed. As the window size, step length, and misalignment artifact recognition model that affect the misalignment artifact recognition accuracy are obtained through the screening of the misalignment artifact recognition accuracy as a whole, the above factors all meet the requirements of the misalignment artifact recognition accuracy, which is beneficial to ensuring the accuracy of recognizing misalignment artifacts in medical tomographic images to be processed.

[0031] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0033] Figure 1 It is a flowchart showing a method for determining a misalignment artifact recognition strategy provided by an embodiment of the present invention;

[0034] Figure 2 It is a schematic diagram of a tomographic image provided by an embodiment of the present invention;

[0035] Figure 3 It is a schematic diagram of a window data interception provided by an embodiment of the present invention;

[0036] Figure 4 It is a schematic flow diagram of a misregistration artifact recognition method provided by an embodiment of the present invention;

[0037] Figure 5 It is a schematic structural diagram of a misregistration artifact recognition strategy determination device provided by an embodiment of the present invention;

[0038] Figure 6 It is a schematic structural diagram of a misregistration artifact recognition device provided by an embodiment of the present invention;

[0039] Figure 7 It is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0040] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0041] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0042] Figure 1The flowchart of a method for determining a misregistration artifact recognition strategy provided by an embodiment of the present invention. This embodiment is applicable to the situation of determining a misregistration artifact recognition strategy suitable for recognizing misregistration artifacts in medical images. This method can be executed by a misregistration artifact recognition strategy determination device provided by an embodiment of the present invention. The misregistration artifact recognition strategy determination device can be implemented by software and / or hardware and can be configured on an electronic computing device. The specific steps are as follows:

[0043] S110. Obtain misregistration artifact training samples and multiple groups of image cropping parameter combinations including window sizes and strides.

[0044] S120. For any one of the image cropping parameter combinations, perform window data cropping on the misregistration artifact training samples based on the window size and stride in the image cropping parameter combination, and train a misregistration artifact recognition model corresponding to the image cropping parameter combination based on the cropped window data and the misregistration artifact labels corresponding to each window data.

[0045] S130. Obtain misregistration artifact verification samples, verify the misregistration artifact recognition models corresponding to each image cropping parameter combination based on the misregistration artifact verification samples, and determine the artifact recognition accuracy rates of the misregistration artifact recognition models.

[0046] S140. Determine the misregistration artifact recognition models that meet the artifact recognition accuracy rate screening conditions and the image cropping parameter combinations corresponding to the selected misregistration artifact recognition models as the misregistration artifact recognition strategy.

[0047] In this embodiment, misregistration artifacts may exist in the image data collected by tomographic imaging methods. Tomographic scanning is slicing or slice imaging using any kind of penetrating wave, and each slice imaging can form a three-dimensional image through image reconstruction. During tomographic scanning, errors in the imaging device cause adjacent slices to be misaligned, forming misregistration artifacts. Correspondingly, images with misregistration artifacts may include, but are not limited to, CT (Computed Tomography) images and MRI (Magnetic Resonance Imaging) images, etc. In this embodiment, the scanning object of the above tomographic images is not limited. Exemplarily, the scanning object can be a human body, an animal body, or other objects such as rocks. In some embodiments, the image data obtained by the above tomographic scanning is medical images. Exemplarily, see Figure 2 , Figure 2 which is a schematic diagram of a tomographic image provided by an embodiment of the present invention.

[0048] In this embodiment, the recognition of misregistration artifacts in medical tomographic images includes determining whether misregistration artifacts are included in the medical tomographic images, and in the case of the existence of misregistration artifacts, determining the localization of the misregistration artifacts in the medical tomographic images, that is, the position information. By determining the misregistration artifact recognition strategy, the recognition of misregistration artifacts in medical tomographic images is realized. Among them, the misregistration artifact recognition strategy includes the window size, the step size, and the misregistration artifact recognition model. Specifically, during the window sliding process by the window size and the step size, the window data of the tomographic medical image is intercepted, and further, the misregistration artifact recognition model is used to recognize the misregistration artifacts in the intercepted tomographic image. Among them, each window data is part of the data in the medical tomographic image. By processing each window data separately, the misregistration artifact recognition result of the tomographic image is obtained, replacing the way of overall recognition of the medical tomographic image, and improving the pertinence and accuracy of misregistration artifact recognition.

[0049] The window size, the step size, and the misregistration artifact recognition model in the misregistration artifact recognition strategy are all influencing factors in the misregistration artifact recognition process. Among them, the window size is used to reflect the sensitivity of misregistration artifact judgment. If the window size is too small, there is no visual difference in the transverse position. If the window size is too large, the information intensity within the window is insufficient, resulting in poor accuracy in determining the existence of misregistration artifacts. The step size is used to reflect the localization accuracy of misregistration artifacts. If the step size is too small, there is a problem of high computational complexity. If the step size is too large, there is a problem of large localization accuracy error. The model parameters in the misregistration artifact recognition model also affect the recognition accuracy of misregistration artifacts. Based on the above technical problems, in this embodiment, by determining the misregistration artifact recognition strategy, the computational complexity and the recognition accuracy are taken into account to solve the above technical problems.

[0050] The training samples and validation samples for misregistration artifacts can be medical tomographic images obtained by tomographic scanning. Misregistration artifact sample images are acquired, and each misregistration artifact sample image is correspondingly set with a misregistration artifact label. The misregistration artifact sample images are divided into training samples and validation samples. For example, based on a preset ratio, random extraction is performed in the misregistration artifact sample image set to obtain a misregistration artifact training sample set and a misregistration artifact validation sample set. Among them, there may be partial sample overlap between the misregistration artifact training sample set and the misregistration artifact validation sample set.

[0051] Multiple misregistration artifact recognition strategies are trained through the misregistration artifact training sample set, and the multiple misregistration artifact recognition strategies obtained by training are optimized and screened through the misregistration artifact validation sample set to obtain the final misregistration artifact recognition strategy.

[0052] A plurality of image capture parameter combinations are preset to respectively determine the misregistration artifact recognition strategies corresponding to the respective image capture parameter combinations. Among them, the image capture parameter combinations include a window size and a step size, and the window size and / or the step size are different in different image capture parameter combinations. Among them, the window size includes the height, width, and depth of the window. In some embodiments, the height and width in different window sizes may remain unchanged, and the depth changes. Exemplarily, the height and width of the window size may be the size of the coronal plane and the size of the sagittal plane respectively, the depth direction is the vertical direction of the cross-section of the image, the depth is the number of tomographic scan layers, and the depth of the window size is greater than or equal to 2 to avoid the problem that the data depth in the window data is small, resulting in no visual difference in the transverse plane.

[0053] In the process of determining the corresponding misregistration artifact recognition strategy based on each image capture parameter combination, the mouth image capture parameter combination is used to perform data capture on the misregistration artifact training samples to obtain window data for training the misregistration artifact recognition model. Optionally, performing data capture on the misregistration artifact training samples based on the window size and the step size in the image capture parameter combination includes: along the vertical direction of the cross-section of the misregistration artifact training sample, performing data capture on the misregistration artifact training sample based on the window size and the step size to obtain window data corresponding to each window. Specifically, based on the step size, control the window corresponding to the window size to slide in the misregistration artifact training sample along the vertical direction of the cross-section of the misregistration artifact training sample, and during the sliding process, sequentially capture the window data corresponding to each window position to obtain a plurality of window data. Exemplarily, see Figure 3 , Figure 3 FIG. is a schematic diagram of window data capture provided by an embodiment of the present invention.

[0054] In this embodiment, the model type of the misregistration artifact recognition model is not limited. Exemplarily, the misregistration artifact recognition model may include, but is not limited to, machine learning models such as neural network models, relationship models obtained by data fitting, etc., as long as it has the function of recognizing misregistration artifacts in window data. It should be noted that different types of misregistration artifact recognition models can be obtained through different training methods. In some embodiments, for each image capture parameter combination, a misregistration artifact recognition model of the same type is trained to obtain a misregistration artifact recognition model corresponding to each image capture parameter combination, and the recognition accuracy of each misregistration artifact recognition model is verified through the misregistration artifact verification sample set to screen out the optimal misregistration artifact recognition strategy. In some embodiments, for each image capture parameter combination, different types of misregistration artifact recognition models can be trained, and the recognition accuracy of different types of misregistration artifact recognition models corresponding to different image capture parameter combinations respectively is verified through the misregistration artifact verification sample set to screen out the optimal misregistration artifact recognition strategy.

[0055] Among them, to verify the recognition accuracy of multiple misregistration artifact recognition models through a misregistration artifact verification sample set, it can be to intercept window data for multiple misregistration artifact verification samples in the misregistration artifact verification sample set respectively based on corresponding image interception parameter combinations, perform misregistration artifact recognition on the intercepted window data through each misregistration artifact recognition model to obtain misregistration artifact verification results, and compare the misregistration artifact verification results with the misregistration artifact labels of each misregistration artifact verification sample to determine the recognition accuracy.

[0056] To determine a misregistration artifact recognition strategy based on the recognition accuracy of the misregistration artifact recognition model, it can be to screen the misregistration artifact recognition models based on preset misregistration artifact recognition accuracy screening conditions, and determine the misregistration artifact recognition model and the corresponding image interception parameter combination as the misregistration artifact recognition strategy. Optionally, determining the misregistration artifact recognition model that meets the misregistration artifact recognition accuracy screening conditions and the image interception parameter combination corresponding to the selected misregistration artifact recognition model as the misregistration artifact recognition strategy includes: determining the misregistration artifact recognition model with the highest misregistration artifact recognition accuracy and the image interception parameter combination corresponding to the misregistration artifact recognition model with the highest misregistration artifact recognition accuracy as the misregistration artifact recognition strategy. By training different misregistration artifact recognition models based on different image interception parameter combinations and screening the misregistration artifact recognition model with the highest misregistration artifact recognition accuracy, while ensuring the misregistration artifact recognition accuracy, the influence of window size and step size selection on the misregistration artifact recognition accuracy is avoided, and at the same time, the difficulty of selecting window size and step size is reduced.

[0057] Optionally, a misalignment artifact recognition model that meets the misalignment artifact recognition accuracy screening criteria, and the corresponding image cropping parameter combination of the selected misalignment artifact recognition model are determined as the misalignment artifact recognition strategy, including: determining the misalignment artifact recognition model that meets the misalignment artifact recognition accuracy threshold, and the corresponding image cropping parameter combination of the misalignment artifact recognition model that meets the misalignment artifact recognition accuracy threshold as the misalignment artifact recognition strategy. Set the misalignment artifact recognition accuracy threshold according to the accuracy requirements of the misalignment artifact. Exemplarily, the misalignment artifact recognition accuracy threshold can be 95% or 90%, etc., and there is no limitation in comparison. By determining one or more misalignment artifact recognition models with a misalignment artifact recognition accuracy greater than or equal to the misalignment artifact recognition accuracy threshold as candidate misalignment artifact recognition models, correspondingly, the candidate misalignment artifact recognition models and their corresponding image cropping parameter combinations form candidate misalignment artifact recognition strategies. Optionally, any one of the candidate misalignment artifact recognition strategies can be determined as the target misalignment artifact recognition strategy. Optionally, determine the target misalignment artifact recognition strategy among multiple candidate misalignment artifact recognition strategies based on the step size. For example, determine the candidate misalignment artifact recognition strategy corresponding to the maximum step size as the target misalignment artifact recognition strategy. By increasing the step size, the number of window data intercepted during the medical tomography image processing can be reduced, so as to further reduce the computational complexity during the misalignment recognition. The computational complexity is small, and both the misalignment artifact recognition accuracy and the computational complexity are taken into account.

[0058] The technical solution improved in this embodiment is to preset multiple image cropping parameter combinations. Different image cropping parameter combinations include different window sizes and step sizes. Based on each image cropping parameter combination, the corresponding misalignment artifact recognition model is trained respectively. The misalignment artifact recognition accuracy of each misalignment artifact recognition model is verified based on the misalignment artifact verification samples, and the target misalignment artifact recognition model is screened based on the obtained misalignment artifact recognition accuracy. The target misalignment artifact recognition model and the corresponding image cropping parameter combination form the misalignment artifact recognition strategy, which is used to identify the misalignment artifacts in the medical tomography image to be processed. As the window size, step size, and misalignment artifact recognition model that affect the misalignment artifact recognition accuracy are obtained through the screening of the misalignment artifact recognition accuracy as a whole, the above factors all meet the requirements of the misalignment artifact recognition accuracy, which is beneficial to ensuring the accuracy of misalignment artifact recognition for the medical tomography image to be processed.

[0059] Based on the above embodiments, the misalignment artifact recognition model may include a relationship model and a target threshold. Among them, the relationship model can be constructed based on the correspondence between the input information and the output information, and is used to determine the corresponding output information based on the input information. The target threshold is used to compare the output information of the relationship model. When the comparison condition is met (for example, greater than or equal to the target threshold), a first response value is generated. When the comparison condition is not met (for example, less than the target threshold), a second response value is generated. The first response value and the second response value are used to determine the separation and localization of the misalignment artifacts in the input information, that is, the recognition result is obtained. In some embodiments, it may be determined that there are misalignment artifacts in the window data when the response value is a specific response value. For example, the specific response value is the first response value. In some embodiments, the response value changes of multiple window data in the same image are determined, and the window data with the response value change is determined as the window data with misalignment artifacts. The above input information may be window data, and the output information may be feature information for characterizing the window data, such as statistical metrics, etc.

[0060] In some embodiments, training the misalignment artifact recognition model corresponding to the image cropping parameter combination based on the cropped window data and the misalignment artifact labels corresponding to each window data includes: determining a first statistical metric corresponding to the window data, and constructing a relationship model between the first statistical metric and the window size, step size, and window data based on the correspondence between the window size, step size, and the window data and the first statistical metric; optimizing the threshold corresponding to the first statistical metric based on the misalignment artifact labels corresponding to each window data and the training recognition results of each window data, and determining the target threshold corresponding to the first statistical metric. The target threshold is used to compare with the first statistical metric to determine the misalignment artifact recognition results of each window. The relationship model and the target threshold constitute the misalignment artifact recognition model.

[0061] Among them, the first statistical indicator is a single indicator data that can characterize the feature information in the window data. By comparing the target threshold with the first statistical indicator, the identification of misalignment artifacts can be achieved. The first statistical indicator includes, but is not limited to, variance and frequency-domain statistical information, which can be obtained by calculating the window data according to a pre-set algorithm. Exemplarily, taking the first statistical indicator as variance, the variance of any window data is calculated. Taking the first statistical indicator as frequency-domain statistical information as an example, the frequency-domain statistical information can be the information obtained by statistically analyzing the feature information of the window data in the frequency domain. Optionally, the determination method of the frequency-domain statistical information may include: converting the window data into a frequency-domain image, extracting the high-frequency information and low-frequency information in the frequency-domain image, and determining the high-frequency statistical information of the high-frequency information and the low-frequency statistical information of the low-frequency information. Among them, the window data is converted into a frequency-domain image through a frequency-domain transformation method. Exemplarily, the frequency-domain transformation method includes, but is not limited to, discrete cosine transform, discrete Fourier transform, discrete wavelet transform, etc. The frequency-domain image obtained through frequency-domain conversion and ZIG-ZAG sorting includes high-frequency information and low-frequency information. The high-frequency information and low-frequency information can be extracted by dividing the frequency-domain image into regions. Exemplarily, the upper-left corner region of the frequency-domain image is used as the low-frequency region, and the lower-right corner region is used as the high-frequency region. Correspondingly, the low-frequency information is extracted based on the low-frequency region to obtain the low-frequency statistical information, and the high-frequency information is extracted based on the high-frequency region to obtain the high-frequency statistical information. Among them, the division of the low-frequency region and the high-frequency region can be pre-set. Specifically, the pixel data of each pixel point in the low-frequency region can be determined as the low-frequency information, and the sum value or average value of the low-frequency information of each pixel point can be determined as the low-frequency statistical information; the pixel data of each pixel point in the high-frequency region can be determined as the high-frequency information, and the sum value or average value of the high-frequency information of each pixel point can be determined as the high-frequency statistical information.

[0062] For the window data obtained by intercepting through each image intercepting parameter combination, the first statistical indicator of the window data is calculated respectively. Correspondingly, the window size, the step size, and the window data are used as influencing factors of the first statistical indicator. Based on the corresponding relationship between the window size, the step size, the window data, and the first statistical indicator, a relationship model is constructed. Here, the form of the relationship model is defined. The input information of the relationship model is the window size, the step size, and the window data, and the output information is the first statistical indicator. Further, the window size and the step size in the relationship model can be fixed parameters.

[0063] In some embodiments, the relationship model can be a scatter plot model. Taking the first statistical metric as variance as an example, the relationship model can be a variance scatter plot model. Taking the first statistical metric as frequency domain statistical information as an example, the relationship model can be a frequency domain information scatter plot model. Exemplarily, based on the window size, step size, and the corresponding relationship between window data and variance, a variance scatter plot is constructed, and a variance scatter plot model is constructed based on the variance scatter plot. For example, the mapping relationship between variance and window size, step size, and window data is obtained through fitting the variance scatter plot. Exemplarily, based on the corresponding relationships between window size, step size, and window data and high-frequency statistical information and low-frequency statistical information respectively, a high-frequency statistical information scatter plot and a low-frequency information scatter plot are constructed respectively. Further, a frequency domain information scatter plot model is constructed. It should be noted that for the characteristics of misregistration artifacts, the low-frequency information of misregistration artifacts remains basically unchanged, while the high-frequency information changes greatly.

[0064] The relationship model constructed in the above embodiments can perform mapping transformation on the input information to obtain the corresponding first statistical metric, and compare the first statistical metric of each window data based on the target threshold corresponding to the first statistical metric to obtain the response value of the first statistical metric. Different target thresholds are obtained by training different types of first statistical metrics. Exemplarily, when the first statistical metric is variance, the corresponding threshold is the variance threshold; when the first statistical metric is frequency domain statistical information, the corresponding threshold is the frequency domain threshold.

[0065] The target threshold is obtained through iterative training with the misregistration artifact labels of each window data. Optionally, optimizing the threshold corresponding to the first statistical metric based on the misregistration artifact labels corresponding to each window data and the training recognition results of each window data to determine the target threshold corresponding to the first statistical metric includes: comparing the first statistical metric corresponding to each window data based on the current threshold, and determining the training recognition result of the misregistration artifact training sample based on the comparison result; adjusting the current threshold based on the training recognition result of the misregistration artifact training sample and the misregistration artifact label until the current threshold meets the recognition accuracy and is determined as the target threshold.

[0066] Determine the initial threshold, and perform iterative optimization on the initial threshold. In each iteration process, based on the current threshold, compare the first statistical metric corresponding to each window data to determine the training recognition result, and determine the misregistration artifact recognition accuracy corresponding to the current threshold based on the training recognition result and the misregistration artifact label corresponding to each window data. Adjust the current threshold by increasing or decreasing it, and based on the adjusted threshold, determine the training recognition result and the corresponding misregistration artifact recognition accuracy again. Determine the threshold corresponding to the optimal misregistration artifact recognition accuracy as the target threshold.

[0067] Based on the above embodiments, the input information of the misregistration artifact recognition model can be the second statistical metric, and the output information can be the misregistration artifact recognition index. In some embodiments, the misregistration artifact recognition index can be in the range of [0, 1], and this misregistration artifact recognition index can be used as the misregistration artifact response value. This misregistration artifact recognition index is used to represent the probability of the existence of misregistration artifacts in the window data. When the misregistration artifact recognition index is greater than or equal to the index threshold, it is determined that there are misregistration artifacts in the window data. When the misregistration artifact recognition index is less than the index threshold, it is determined that there are no misregistration artifacts in the window data.

[0068] Optionally, based on the intercepted window data and the misregistration artifact labels corresponding to each window data, training the misregistration artifact recognition model corresponding to the image interception parameter combination includes: determining the second statistical metric corresponding to the window data, and based on the corresponding relationship between the window size, stride, second statistical metric, and misregistration artifact labels, constructing a misregistration artifact recognition model of the misregistration artifact recognition index with respect to the window size, stride, and second statistical metric.

[0069] Among them, the second statistical metric includes, but is not limited to, variance, frequency domain statistical information, and texture feature information. The calculation method of the second statistical metric is not limited here. Taking variance as an example, the variance of the window data can be calculated by using the variance calculation formula for the window data, or can be obtained by processing the window data through a variance scatter plot model.

[0070] The misregistration artifact recognition model can be a mapping relationship model between the window size, stride, second statistical metric, and misregistration artifact recognition index. For example, the mapping relationship model can be obtained through data fitting. Taking the second statistical metric as texture feature information as an example, optionally, determining the second statistical metric corresponding to the window data, and based on the corresponding relationship between the second statistical metric and the misregistration artifact labels, constructing a misregistration artifact recognition model of the misregistration artifact recognition index with respect to the second statistical metric includes: determining the texture matrix corresponding to the window data, and extracting texture feature information based on the texture matrix; based on the corresponding relationship between the texture feature information and the misregistration artifact labels, fitting to obtain a texture mapping relationship model of each texture feature information and the misregistration artifact recognition index.

[0071] Among them, the texture matrix is a matrix used to describe the characteristics of window data. Optionally, the texture matrix may include, but is not limited to, Gray-level Co-occurrence Matrix (GLCM), Local Binary Pattern (LBP), Gram matrix (GRAM), etc. A scalar vector is extracted from the texture matrix as texture feature information. Taking the texture matrix as the gray-level co-occurrence matrix as an example, the texture feature vector may include ASM energy, contrast, inverse difference moment, entropy, and autocorrelation value. Among them, the ASM energy (angular second moment) is the sum of the squares of each matrix element in the gray-level co-occurrence matrix, and the contrast is used to reflect the contrast of the brightness of a certain pixel value and its neighboring pixel values. If the elements off the diagonal have large values, that is, the image brightness values change rapidly, the contrast will have a large value. The contrast reflects the clarity of the image and the depth of the texture grooves. The deeper the texture grooves, the greater the contrast and the clearer the visual effect; conversely, the smaller the contrast, the shallower the grooves and the more blurred the effect. Among them Among them, G(i,j) is each matrix element in the gray-level co-occurrence matrix. The inverse difference moment is used to reflect the homogeneity of the image texture and measure the amount of local change in the image texture. A large value indicates that there is little change between different regions of the image texture and it is very uniform locally. Among them, the inverse difference moment The entropy is a measure of the amount of information in the image, indicating the degree of non-uniformity or complexity of the texture in the image. When all elements in the co-occurrence matrix have the greatest randomness and all values in the spatial co-occurrence matrix are almost equal, the entropy is small. When the elements in the co-occurrence matrix are distributed dispersedly, the entropy is large. Among them, the entropy The autocorrelation is used to measure the similarity degree of the elements of the gray-level co-occurrence matrix in the row or column direction. The size of the correlation value reflects the local gray correlation in the image. When the matrix element values are uniformly equal, the correlation value is large; on the contrary, if the matrix pixel values vary greatly, the correlation value is small. Among them, the autocorrelation value

[0072] It should be noted that the texture feature information corresponding to different types of texture matrices can be set according to requirements, and no limitation is made in this regard. By fitting the corresponding relationship between multiple texture feature information (such as ASM energy, contrast, inverse difference moment, entropy, and autocorrelation value) and the misregistration artifact label of the window data, the mapping relationship f between the texture feature information as the input information X and the misregistration artifact label y is obtained, and the texture mapping relationship model y = f(X) is obtained.

[0073] In some embodiments, the input information of the misalignment artifact recognition model may be window data, and the output information may be a misalignment artifact recognition index or a misalignment artifact classification probability. Correspondingly, the misalignment artifact recognition model may be a machine learning model such as a neural network model. Exemplarily, the network structure of the misalignment artifact recognition model may be a convolutional neural network model, a recurrent neural network, or a Transformer model, etc. Exemplarily, the convolutional neural network module includes but is not limited to AlexNet, VGG, Inception, ResNet, DenseNet, etc., and the recurrent neural network model includes but is not limited to LSTM (Long Short Term Memory Network), GRU (Gated Recurrent Unit) model, etc.

[0074] For the window data intercepted for each image interception parameter combination, a neural network model of at least one network structure can be trained to obtain at least one trained misalignment artifact recognition model. For any network type, neural network models with different network depths can be created. Taking the convolutional neural network as an example, neural network models with different numbers of layers can be created, such as three-layer, ten-layer, fifteen-layer, etc., which is not limited herein. By training neural network models with different network structures and / or different network depths, the corresponding misalignment artifact recognition models are obtained.

[0075] Optionally, for any image interception parameter combination, based on the intercepted window data and the misalignment artifact labels corresponding to each window data, training to obtain the misalignment artifact recognition model corresponding to the image interception parameter combination includes: constructing an initial neural network model of at least one network type, and iteratively executing the following training process until the training condition is met to obtain the misalignment artifact recognition models corresponding to each network type: inputting the window data into the neural network model of the current iteration to obtain the misalignment artifact prediction result output by the neural network model, and determining a loss function based on the misalignment artifact prediction result and the corresponding misalignment artifact label to adjust the network parameters in the neural network model of the current iteration.

[0076] It should be noted that before inputting the window data into the neural network model to be trained, preprocessing of the window data may also be included, where the preprocessing includes but is not limited to denoising, image enhancement, image data scaling, etc. In the training process, the network parameters in the neural network model are adjusted iteratively, where the network parameters include but are not limited to weights. When the training process meets the training condition, it is determined that the neural network model of the current iteration is trained and determined as the misalignment artifact recognition model. The training conditions include one or more of the following: the number of iterative training times meets the preset training times, the prediction accuracy of the model meets the preset accuracy threshold, and the training process reaches the minimum convergence state.

[0077] In each iteration training process, the neural network model in the current iteration process is used to predict the input window data, and the misregistration artifact prediction result of the window data is obtained. A loss function is determined based on the misregistration artifact prediction result and the corresponding misregistration artifact label. The loss function includes, but is not limited to, exponential loss function, cross-entropy loss function, hinge loss function, etc., and can be determined according to the training requirements.

[0078] The loss function is reversely input into the neural network model of the current iteration to adjust the network parameters in the neural network model of the current iteration. Specifically, the adjustment of the network parameters of the neural network model can be implemented based on the gradient descent method. The above training process is iteratively executed until a trained misregistration artifact recognition model is obtained, and the misregistration artifact recognition model has the function of recognizing misregistration artifacts in window data.

[0079] In this embodiment, one or more misregistration artifact recognition models are trained for each image cropping parameter combination. Each misregistration artifact recognition model can correspond to one or more misregistration artifact recognition models. Each misregistration artifact recognition model and the corresponding image cropping parameter combination can form a misregistration artifact recognition strategy, that is, multiple candidate misregistration artifact recognition strategies are obtained. By verifying the obtained multiple candidate misregistration artifact recognition strategies, the target misregistration artifact recognition strategy is determined. It should be noted that the determination processes of misregistration artifact recognition strategies corresponding to different image cropping parameter combinations can be implemented in parallel to improve the determination efficiency of the target misregistration artifact recognition strategy.

[0080] Each misregistration artifact verification sample in the misregistration artifact verification sample set is respectively input into the misregistration artifact recognition model in each of the above candidate misregistration artifact recognition strategies to obtain the verification recognition results corresponding to each misregistration artifact verification sample. The misregistration artifact verification sample verification recognition results are compared with the misregistration artifact labels to determine the misregistration artifact recognition accuracy of the misregistration artifact recognition strategy, and the target misregistration artifact recognition strategy is determined based on the misregistration artifact recognition accuracy.

[0081] On the basis of the above embodiment, it further includes statistically processing the processing duration and computational consumption parameters of the misregistration artifact recognition process of each misregistration artifact recognition strategy for the misregistration artifact verification sample, and comprehensively determining the target misregistration artifact recognition strategy based on the misregistration artifact recognition accuracy and the consumption parameters. Exemplarily, the misregistration artifact recognition strategy with the misregistration artifact recognition accuracy meeting the accuracy threshold and the minimum consumption parameter can be determined as the target misregistration artifact recognition strategy.

[0082] The technical solution of this embodiment is to obtain multiple groups of image cropping parameter combinations. For different window sizes and strides, window data of misregistration artifact training samples are respectively cropped, and misregistration artifact recognition models corresponding to each group of image cropping parameter combinations are trained based on the cropped window data, forming multiple candidate misregistration artifact recognition strategies. Further, multiple misregistration artifact recognition strategies are verified by misregistration artifact verification samples, and a target misregistration artifact recognition strategy that meets the misregistration artifact recognition accuracy requirement is obtained. The target misregistration artifact recognition strategy includes the window size and stride for window cropping of tomographic images, and a misregistration artifact recognition model for recognizing misregistration artifacts in the cropped window data. The above window size, stride, and misregistration artifact recognition model all meet the misregistration artifact recognition requirements, and there are no negative factors affecting misregistration artifact recognition. At the same time, the determination process of the above factors does not require manual determination, reducing the consumption of resources such as manpower and time in the manual determination process, and reducing errors caused by human operation.

[0083] Based on the above embodiment, an embodiment of the present invention further provides a misregistration artifact recognition method. Refer to Figure 4 , Figure 4 which is a flowchart of a misregistration artifact recognition method provided by an embodiment of the present invention. This embodiment is applicable to the situation of misregistration artifact recognition of tomographic images. This method can be executed by a misregistration artifact recognition device provided by an embodiment of the present invention. The misregistration artifact recognition device can be implemented by software and / or hardware, and the misregistration artifact recognition device can be configured on an electronic computing device. The specific steps are as follows:

[0084] S210. Obtain an image to be processed.

[0085] S220. Based on the window size and stride in the misregistration artifact recognition strategy, crop window data from the image to be processed to obtain multiple pieces of window data.

[0086] S230. Determine misregistration artifact description parameters corresponding to each piece of window data based on the misregistration artifact recognition model in the misregistration artifact recognition strategy.

[0087] S240. Determine misregistration artifact windows based on each misregistration artifact description parameter and a corresponding threshold, and determine misregistration artifact positions based on the misregistration artifact windows.

[0088] It should be noted that the recognition of misregistration artifacts in the image to be processed in the embodiment of the present invention is implemented based on a pre-set misregistration artifact recognition strategy. Among them, the pre-set misregistration artifact recognition strategy includes a window size, a stride, and a misregistration artifact recognition model. The pre-set misregistration artifact recognition strategy can be determined in advance according to the misregistration artifact recognition strategy determination method provided in any of the above embodiments.

[0089] Based on the window size and step size, control the window to slide vertically along the cross-section in the image to be processed, and intercept the window data during the sliding process to obtain multiple intercepted window data. Process each window data based on the misregistration artifact recognition model in the misregistration artifact recognition strategy, and determine the misregistration artifact recognition result of the image to be processed based on the recognition results corresponding to each window data.

[0090] The misregistration artifact recognition model can be any one of the models provided in the above embodiments, such as a misregistration artifact recognition model based on variance statistics, a misregistration artifact recognition model based on frequency domain statistical information, a misregistration artifact recognition model based on texture feature information, or a neural network model, etc. The above misregistration artifact recognition models are all used to recognize the response values of each window data, and determine the misregistration artifact recognition result based on the response values corresponding to each window data in the image to be processed. Specifically, based on the characteristics of misregistration artifacts, when the response value of the window data is a specific response value, it is determined that there is a misregistration artifact in the window data, that is, the corresponding window is a misregistration artifact window; or, when there is a change in the response value of adjacent window data, the window corresponding to the changed response value is determined as the misregistration artifact window.

[0091] In this embodiment, the misregistration artifact recognition result of the image to be processed includes the position information of the misregistration artifacts in the image to be processed. Determine the position information of the misregistration artifacts in the image to be processed according to the misregistration artifact windows. For example, the position information of each misregistration artifact window can be determined as the position information of the misregistration artifacts, or the center position of the misregistration artifact window can be determined as the position information of the misregistration artifacts. It can also be that, based on each misregistration artifact window, the misregistration artifact region in the image to be processed is determined, and the center position information of the misregistration artifact region is determined as the position information of the misregistration artifacts. Among them, the misregistration artifact region can be a region formed by continuous misregistration artifact windows.

[0092] Optionally, determining the misalignment artifact position based on the misalignment artifact window in the image to be processed includes: determining consecutive misalignment artifact windows according to the adjacent relationship of each window; for any group of consecutive misalignment artifact windows, determining the corresponding misalignment artifact position based on the sum of the window start position and half of the window depth in at least one consecutive misalignment artifact window. Among them, the number of consecutive misalignment artifact windows is greater than or equal to 1. Exemplarily, n window data are intercepted from the image to be processed, and the response values of each window data are 01110010…. The windows with specific response values (for example, 1) are determined as misalignment artifact windows, that is, windows 2, 3, 4, and 7 are misalignment artifact windows. Further, the first group of consecutive misalignment artifact windows includes windows 2, 3, and 4, and the second group of consecutive misalignment artifact windows includes window 7. The first misalignment artifact position is determined based on the sum of the window start position and half of the window depth in the first group of consecutive misalignment artifact windows, and the second misalignment artifact position is determined based on the sum of the window start position and half of the window depth in the second group of consecutive misalignment artifact windows.

[0093] In this embodiment, the image to be processed may include multiple misalignment artifacts. By identifying the positions of the misalignment artifacts in the image to be processed, it is convenient to locate the misalignment artifacts. Optionally, de-artifact processing is performed on the image to be processed based on the position information of the misalignment artifacts.

[0094] In some embodiments, the misalignment artifact recognition model includes a variance scatter plot model and a variance threshold, and the misalignment artifact description parameter includes a variance response value. Correspondingly, determining the misalignment artifact description parameter corresponding to each window data based on the misalignment artifact recognition model in the misalignment artifact recognition strategy includes: inputting the window data into the variance scatter plot model to obtain the variance of the window data, comparing the variance threshold with the variance of the window data, and determining the variance response value corresponding to the window data according to the comparison result.

[0095] Among them, the variance scatter plot model can reflect the mapping relationship between the window data and the variance under the current window size and step length. By inputting the window data into the variance scatter plot model, the variance output by the variance scatter plot model is obtained. In some embodiments, the variance of the window data can also be determined based on the variance calculation method, which is not limited herein.

[0096] Comparing the variance of the window data with the variance threshold, when the variance of the window data is greater than or equal to the variance threshold, determining the variance response value as the first response value, and when the variance of the window data is less than the variance threshold, determining the variance response value as the second response value. Exemplarily, the first response value can be 1, and the second response value can be 0.

[0097] Correspondingly, determining the misregistration artifact recognition result of the image to be processed based on each of the misregistration artifact description parameters includes: determining a window with a variance response value being a first specific value as a misregistration artifact window; determining the misregistration artifact position based on the misregistration artifact window in the image to be processed. Wherein, the first specific value can be 1, that is, determining a window with the variance of the window data being greater than or equal to the variance threshold as a misregistration artifact window. Based on the interception order of each window in the image to be processed, determining consecutive misregistration artifact windows in the image to be processed, and respectively determining the misregistration artifact positions based on each group of consecutive misregistration artifact windows.

[0098] In some embodiments, the misregistration artifact recognition model includes a frequency domain information scatter plot model and a frequency domain threshold, and the misregistration artifact description parameter includes a frequency domain information response value; correspondingly, determining the misregistration artifact description parameter corresponding to each window data based on the misregistration artifact recognition model in the misregistration artifact recognition strategy includes: inputting the window data into the frequency domain information scatter plot model to obtain the high-frequency statistical information and low-frequency statistical information of the window data, comparing the frequency domain threshold with the high-frequency statistical information of the window data, and comparing the low-frequency threshold with the low-frequency statistical information of the window data, and determining the frequency domain information response value corresponding to the window data according to the high-frequency information comparison result and the low-frequency information comparison result.

[0099] Among them, the frequency domain information scatter plot model can reflect the mapping relationship between the window data and the frequency domain statistical information under the current window size and step size. The frequency domain statistical information includes high-frequency statistical information and low-frequency statistical information. Correspondingly, the frequency domain threshold includes a high-frequency threshold and a low-frequency threshold. By inputting the window data into the frequency domain information scatter plot model, the frequency domain statistical information output by the frequency domain information scatter plot model is obtained. In some embodiments, the frequency domain information scatter plot model includes a high-frequency information scatter plot model and a low-frequency information scatter plot model. Correspondingly, the high-frequency information scatter plot model is used to output the corresponding high-frequency statistical information based on the input window data, and the low-frequency information scatter plot model is used to output the corresponding low-frequency statistical information based on the input window data. In some embodiments, the frequency domain statistical information of the window data can also be obtained by converting the window data into a frequency domain image, extracting the high-frequency information and low-frequency information from the frequency domain image, and respectively determining the high-frequency statistical information and low-frequency statistical information based on the extracted high-frequency information and low-frequency information, which is not limited herein.

[0100] Compare the frequency-domain statistical information of the window data with the frequency-domain threshold, and determine the frequency-domain information response value corresponding to the window data according to the comparison result. Optionally, compare the high-frequency statistical information of the window data with the high-frequency threshold to determine the high-frequency response value, and compare the low-frequency statistical information of the window data with the low-frequency threshold to determine the low-frequency response value. Determine the misregistration artifact window based on the high-frequency response value and the low-frequency response value of each window. Specifically, a window with a change in the high-frequency response value and / or the low-frequency response value can be determined as the misregistration artifact window. Specifically, when the high-frequency statistical information is greater than or equal to the high-frequency threshold, the high-frequency response value is the first high-frequency response value, and when the high-frequency statistical information is less than the high-frequency threshold, the high-frequency response value is the second high-frequency response value; when the low-frequency statistical information is greater than or equal to the low-frequency threshold, the low-frequency response value is the first low-frequency response value, and when the low-frequency statistical information is less than the low-frequency threshold, the low-frequency response value is the second low-frequency response value. Determine the misregistration artifact window based on the frequency-domain information response value corresponding to each window. For example, a window with a change in the response value is determined as the misregistration artifact window. Exemplarily, if the high-frequency response values of each window are 00100010… in sequence, then windows 3 and 7 can be determined as the misregistration artifact windows. Exemplarily, if the low-frequency response values of each window are 11011011… in sequence, then windows 3 and 6 can be determined as the misregistration artifact windows.

[0101] Determine the misregistration artifact position based on the misregistration artifact window in the image to be processed. In this embodiment, the misregistration artifact windows are not continuous, that is, only one misregistration artifact window is included in consecutive misregistration artifact windows. The sum of the starting position of the misregistration artifact window and half of the window depth is determined as the position of the misregistration artifact.

[0102] In some embodiments, the misregistration artifact recognition model includes a texture mapping relationship model, and the misregistration artifact description parameter includes a misregistration artifact index; correspondingly, determine the misregistration artifact description parameter corresponding to each window data based on the misregistration artifact recognition model in the misregistration artifact recognition strategy, including: extracting the texture feature information corresponding to the window data, and inputting the texture feature information into the texture mapping relationship model to obtain the misregistration artifact index of the window data output by the texture mapping relationship model.

[0103] The texture mapping relationship model can reflect the mapping relationship between the texture feature information and the misregistration artifact index under the current window size and step size. For each window data, extract the corresponding texture feature information. Exemplarily, the texture feature information includes ASM energy, contrast, inverse difference, entropy, and autocorrelation value. Optionally, determine the texture matrix corresponding to the window data, such as the gray-level co-occurrence matrix, and extract the above texture feature information based on the texture matrix. Input the extracted texture feature information into the texture mapping relationship model to obtain the misregistration artifact index output by the texture mapping relationship model. This misregistration artifact index is used to characterize the recognition probability of the misregistration artifact.

[0104] Determine the misregistration artifact recognition result of the image to be processed based on each of the misregistration artifact description parameters, including: determining the window with a misregistration artifact index greater than the index threshold as the misregistration artifact window; and determining the misregistration artifact position based on the misregistration artifact windows in the image to be processed. Specifically, based on the intercept order of each window in the image to be processed, determine the continuous misregistration artifact windows in the image to be processed, and based on each group of continuous misregistration artifact windows, determine the misregistration artifact positions respectively.

[0105] In some embodiments, the misregistration artifact recognition model includes a neural network model, and the misregistration artifact description parameters include the window recognition type or the misregistration artifact recognition probability; correspondingly, determining the misregistration artifact description parameters corresponding to each window data based on the misregistration artifact recognition model in the misregistration artifact recognition strategy includes: inputting the window data into the neural network model to obtain the window recognition type or the misregistration artifact recognition probability of the window data output by the neural network model.

[0106] The trained neural network model has the function of predicting the misregistration artifact type of the window data based on the window data. The neural network model can output the window recognition type. For example, the window recognition type includes the misregistration artifact type and the non-misregistration artifact type. For example, it can be represented by 0 and 1 for the non-misregistration artifact type and the misregistration artifact window type respectively. Determining the misregistration artifact recognition result of the image to be processed based on each of the misregistration artifact description parameters includes: determining the window with the window recognition type being the misregistration artifact type as the misregistration artifact window. The neural network model can also output the misregistration artifact recognition probability, and determining the window with the misregistration artifact recognition probability greater than the probability threshold as the misregistration artifact window.

[0107] Determine the misregistration artifact position based on the misregistration artifact windows in the image to be processed. Specifically, based on the intercept order of each window in the image to be processed, determine the continuous misregistration artifact windows in the image to be processed, and based on each group of continuous misregistration artifact windows, determine the misregistration artifact positions respectively.

[0108] The technical solution provided in this embodiment intercepts window data of the image to be processed through the window size and step size in the pre-determined misregistration artifact recognition strategy, and uses the misregistration artifact recognition model in the misregistration artifact recognition strategy to recognize each window data, obtaining the response value of each window data. Based on this response value, it is determined whether there are misregistration artifacts in the image to be processed and the located misregistration artifacts are positioned, facilitating accurate subsequent processing of the misregistration artifacts in the image to be processed. The misregistration artifact recognition strategy is pre-trained and verified for recognition accuracy. Among them, the window size, step size, and misregistration artifact recognition model are all factors in the training and verification. The above factors all meet the recognition accuracy of misregistration artifacts, ensuring the recognition accuracy of misregistration artifacts for the image to be processed.

[0109] Based on the above embodiment, an embodiment of the present invention further provides a device for determining a misregistration artifact recognition strategy. Refer to Figure 5 , Figure 5 which is a schematic structural diagram of a device for determining a misregistration artifact recognition strategy provided by an embodiment of the present invention. The device includes:

[0110] A data acquisition module 310, configured to acquire misregistration artifact training samples and multiple groups of image interception parameter combinations including window size and step size;

[0111] A data interception module 320, configured to, for any image interception parameter combination, intercept window data of the misregistration artifact training samples based on the window size and step size in the image interception parameter combination;

[0112] A model training module 330, configured to train a misregistration artifact recognition model corresponding to the image interception parameter combination based on the intercepted window data and the misregistration artifact labels corresponding to each window data;

[0113] A model verification module 340, configured to acquire misregistration artifact verification samples, verify the misregistration artifact recognition models corresponding to each image interception parameter combination based on the misregistration artifact verification samples, and determine the artifact recognition accuracy rates of the misregistration artifact recognition models;

[0114] A recognition strategy determination module 350, configured to determine the misregistration artifact recognition models that meet the misregistration artifact recognition accuracy rate screening conditions and the image interception parameter combinations corresponding to the selected misregistration artifact recognition models as the misregistration artifact recognition strategy.

[0115] Optionally, the misregistration artifact training samples and the misregistration artifact verification samples are medical tomographic images;

[0116] The data truncation module 320 is used to perform data truncation on the misalignment artifact training samples in the vertical direction of the cross-section of the misalignment artifact training samples based on the window size and the step size, so as to obtain window data corresponding to each window.

[0117] Optionally, the window size includes the height, width and depth of the window. Among them, the depth direction is the vertical direction of the cross-section of the image, the depth is the number of tomography layers, and the depth of the window size is greater than or equal to 2.

[0118] Optionally, the model training module 330 includes:

[0119] The relationship module construction unit is used to determine the first statistical index corresponding to the window data, and construct a relationship model between the first statistical index and the window size, step size, and window data based on the corresponding relationship between the window size, step size, window data and the first statistical index;

[0120] The target threshold determination unit is used to optimize the threshold corresponding to the first statistical index based on the misalignment artifact labels corresponding to the window data and the training recognition results of the window data, and determine the target threshold corresponding to the first statistical index. Among them, the target threshold is used to compare with the first statistical index to determine the misalignment artifact recognition results of each window. The relationship model and the target threshold constitute the misalignment artifact recognition model.

[0121] Optionally, the first statistical index includes variance, and the relationship model includes a variance scatter plot model.

[0122] Optionally, the first statistical index includes frequency domain statistical information, and the relationship model includes a frequency domain information scatter plot model;

[0123] The relationship module construction unit is used to: convert the window data into a frequency domain image, extract the high-frequency information and low-frequency information in the frequency domain image, and determine the high-frequency statistical information of the high-frequency information and the low-frequency statistical information of the low-frequency information;

[0124] Based on the corresponding relationship between the window size, step size, window data and the high-frequency statistical information, construct a high-frequency information scatter plot model of the high-frequency statistical information and the window size, step size, window data;

[0125] Based on the corresponding relationship between the window size, step size, window data and the low-frequency statistical information, construct a low-frequency information scatter plot model of the low-frequency statistical information and the window size, step size, window data.

[0126] Optionally, the target threshold determination unit is used to:

[0127] Compare the first statistical metrics corresponding to the window data based on the current threshold, and determine the training recognition result of the misalignment artifact training sample based on the comparison result;

[0128] Adjust the current threshold based on the training recognition result of the misalignment artifact training sample and the misalignment artifact label until the current threshold meets the recognition accuracy and is determined as the target threshold.

[0129] Optionally, the model training module 330 is used to: determine the second statistical metric corresponding to the window data, and construct a misalignment artifact recognition model of the misalignment artifact recognition index with the window size, step size, and second statistical metric based on the corresponding relationship between the window size, step size, second statistical metric, and misalignment artifact label.

[0130] Optionally, the second statistical metric includes texture feature information;

[0131] The model training module 330 is used to: determine the texture matrix corresponding to the window data, and extract texture feature information based on the texture matrix; fit a texture mapping relationship model between each texture feature information and the misalignment artifact recognition index based on the corresponding relationship between the texture feature information and the misalignment artifact label.

[0132] Optionally, the model training module 330 is used to: construct an initial neural network model of at least one network type, and iteratively execute the following training process until the training condition is met to obtain a misalignment artifact recognition model corresponding to each network type:

[0133] Input the window data into the neural network model of the current iteration, obtain the misalignment artifact prediction result output by the neural network model, and determine the loss function based on the misalignment artifact prediction result and the corresponding misalignment artifact label to adjust the network parameters in the neural network model of the current iteration.

[0134] Optionally, the recognition strategy determination module 350 is used to determine the misalignment artifact recognition model with the highest misalignment artifact recognition accuracy and the combination of image capture parameters corresponding to the misalignment artifact recognition model with the highest misalignment artifact recognition accuracy as the misalignment artifact recognition strategy; or,

[0135] Determine the misalignment artifact recognition model that meets the misalignment artifact recognition accuracy threshold and the combination of image capture parameters corresponding to the misalignment artifact recognition model that meets the misalignment artifact recognition accuracy threshold as the misalignment artifact recognition strategy.

[0136] The misalignment artifact recognition strategy determination device provided by the embodiments of the present invention can execute the misalignment artifact recognition strategy determination method provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects for executing the misalignment artifact recognition strategy determination method.

[0137] Based on the above embodiments, an embodiment of the present invention further provides a misregistration artifact recognition device. Refer to Figure 6 , Figure 6 which is a schematic structural diagram of a misregistration artifact recognition device provided by an embodiment of the present invention. The device includes:

[0138] An image acquisition module 410, configured to acquire an image to be processed;

[0139] A window data interception module 420, configured to intercept window data from the image to be processed based on the window size and step length in the misregistration artifact recognition strategy, to obtain a plurality of window data;

[0140] A description parameter determination module 430, configured to determine misregistration artifact description parameters corresponding to each window data based on the misregistration artifact recognition model in the misregistration artifact recognition strategy;

[0141] A misregistration artifact recognition module 440, configured to determine a misregistration artifact window based on each misregistration artifact description parameter and a corresponding threshold, and determine a misregistration artifact position based on the misregistration artifact window.

[0142] Optionally, the misregistration artifact recognition model includes a variance scatter plot model and a variance threshold, and the misregistration artifact description parameter includes a variance response value;

[0143] The description parameter determination module 430 is configured to: input the window data into the variance scatter plot model to obtain the variance of the window data, compare the variance threshold with the variance of the window data, and determine the variance response value corresponding to the window data according to the comparison result.

[0144] Optionally, the misregistration artifact recognition model includes a frequency domain information scatter plot model and a frequency domain threshold, and the misregistration artifact description parameter includes a frequency domain information response value;

[0145] The description parameter determination module 430 is configured to: input the window data into the frequency domain information scatter plot model to obtain the high-frequency statistical information and low-frequency statistical information of the window data, compare the frequency domain threshold with the high-frequency statistical information of the window data, and compare the low-frequency threshold with the low-frequency statistical information of the window data, and determine the frequency domain information response value corresponding to the window data according to the high-frequency information comparison result and the low-frequency information comparison result.

[0146] Optionally, the misregistration artifact recognition model includes a texture mapping relationship model, and the misregistration artifact description parameter includes a misregistration artifact index;

[0147] The description parameter determination module 430 is configured to: extract the texture feature information corresponding to the window data, input the texture feature information into the texture mapping relationship model, and obtain the misregistration artifact index of the window data output by the texture mapping relationship model.

[0148] Optionally, the misregistration artifact recognition model includes a neural network model, and the misregistration artifact description parameter includes a window recognition type;

[0149] The description parameter determination module 430 is configured to: input the window data into the neural network model, and obtain the window recognition type of the window data output by the neural network model.

[0150] Optionally, the misregistration artifact position determination unit is configured to: determine consecutive misregistration artifact windows according to the adjacent relationship of each window; for any group of consecutive misregistration artifact windows, determine the corresponding misregistration artifact position based on the sum of the window start position and half of the window depth in at least one consecutive misregistration artifact window.

[0151] The misregistration artifact recognition device provided by the embodiments of the present invention can execute the misregistration artifact recognition method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the misregistration artifact recognition method.

[0152] Figure 7 FIG. shows a schematic structural diagram of an electronic device 10 that can be used to implement the embodiments of the present invention. The electronic device is intended to represent various forms of digital computers, such as, for example, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, for example, a personal digital processor, a cellular phone, a smart phone, a wearable device (such as a helmet, glasses, a watch, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0153] Such as Figure 7As shown, the electronic device 10 includes at least one processor 11 and a memory communicatively connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc. The memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0154] Multiple components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disc, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0155] The processor 11 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the misalignment artifact recognition strategy determination method, and / or the misalignment artifact recognition method.

[0156] In some embodiments, the misalignment artifact recognition strategy determination method, and / or the misalignment artifact recognition method can be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the misalignment artifact recognition strategy determination method, and / or the misalignment artifact recognition method described above can be executed. Alternatively, in other embodiments, the processor 11 can be configured to execute the misalignment artifact recognition strategy determination method, and / or the misalignment artifact recognition method in any other suitable manner (e.g., by means of firmware).

[0157] The various embodiments of the systems and techniques described above in this specification can be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that receives data and instructions from, and transmits data and instructions to, a storage system, at least one input device, and at least one output device.

[0158] The computer program for implementing the methods of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the computer programs, when executed by the processor, cause the functions / operations specified in the flowchart and / or block diagram to be implemented. The computer program can be executed entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine or entirely on the remote machine or server.

[0159] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0160] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the electronic device. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, speech input, or tactile input).

[0161] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), blockchain network, and the Internet.

[0162] A computing system can include a client and a server. The client and the server are generally far from each other and usually interact through a communication network. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system and solves the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0163] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is imposed herein.

[0164] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0165] Note that the above are only the preferred embodiments of the present invention and the technical principles applied. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein. Various obvious changes, re-adjustments, and substitutions can be made by those skilled in the art without departing from the protection scope of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments. Without departing from the concept of the present invention, more other equivalent embodiments can be included, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for determining a misalignment artifact recognition strategy, characterized in that Including: Obtaining misregistration artifact training samples and multiple groups of image cropping parameter combinations including window sizes and strides; For any image cropping parameter combination, perform window data cropping on the misregistration artifact training samples based on the window size and stride in the image cropping parameter combination, and train a misregistration artifact recognition model corresponding to the image cropping parameter combination based on the cropped window data and the misregistration artifact labels corresponding to each window data; Obtain misregistration artifact validation samples, and validate the misregistration artifact recognition models corresponding to each image cropping parameter combination based on the misregistration artifact validation samples to determine the artifact recognition accuracy rates of the misregistration artifact recognition models; Determine the misregistration artifact recognition models that meet the artifact recognition accuracy rate screening conditions and the image cropping parameter combinations corresponding to the selected misregistration artifact recognition models as the misregistration artifact recognition strategy.

2. The method according to claim 1, characterized in that, The misregistration artifact training samples and the misregistration artifact validation samples are medical tomographic images; The performing data cropping on the misregistration artifact training samples based on the window size and stride in the image cropping parameter combination includes: Along the vertical direction of the cross-section of the misregistration artifact training sample, perform data cropping on the misregistration artifact training sample based on the window size and stride to obtain window data corresponding to each window; wherein, the window size includes the height, width, and depth of the window, wherein the depth direction is the vertical direction of the cross-section of the image, the depth is the number of tomographic scan layers, and the depth of the window size is greater than or equal to 2.

3. The method according to claim 1, wherein The training a misregistration artifact recognition model corresponding to the image cropping parameter combination based on the cropped window data and the misregistration artifact labels corresponding to each window data includes: Determine a first statistical index corresponding to the window data, and construct a relationship model between the first statistical index and the window size, stride, and window data based on the corresponding relationship between the window size, stride, window data, and the first statistical index; Optimize the threshold corresponding to the first statistical index based on the misregistration artifact labels corresponding to each window data and the training recognition results of each window data to determine the target threshold corresponding to the first statistical index, wherein the target threshold is used to compare with the first statistical index to determine the misregistration artifact recognition results of each window, and the relationship model and the target threshold constitute the misregistration artifact recognition model.

4. The method according to claim 3, characterized in that, The first statistical index includes variance, and the relationship model includes a variance scatter plot model.

5. The method according to claim 3, characterized in that The first statistical index includes frequency domain statistical information, and the relationship model includes a frequency domain information scatter plot model; The determining a first statistical index corresponding to the window data, and constructing a relationship model between the first statistical index and the window size, stride, and window data based on the corresponding relationship between the window size, stride, window data, and the first statistical index includes: Convert the window data into a frequency domain image, extract the high-frequency information and low-frequency information in the frequency domain image, and determine the high-frequency statistical information of the high-frequency information and the low-frequency statistical information of the low-frequency information; Based on the correspondence between the window size, step size, window data and high-frequency statistical information, a high-frequency information scatter plot model of the high-frequency statistical information and the window size, step size and window data is constructed; Based on the correspondence between the window size, step length, window data and low-frequency statistical information, a low-frequency information scatter plot model of the low-frequency statistical information and the window size, step length and window data is constructed.

6. The method according to claim 3, characterized in that, The optimizing the threshold corresponding to the first statistical indicator based on the misalignment artifact labels corresponding to the window data and the training recognition results of the window data to determine the target threshold corresponding to the first statistical indicator includes: Comparing the first statistical indicator corresponding to each window data based on the current threshold, and determining the training recognition result of the misalignment artifact training sample based on the comparison result; The current threshold is adjusted based on the misalignment artifact training sample training recognition result and the misalignment artifact label until the current threshold meets the recognition accuracy and is determined as the target threshold.

7. The method according to claim 1, wherein The training of the misalignment artifact recognition model corresponding to the image interception parameter combination based on the intercepted window data and the misalignment artifact labels corresponding to each window data includes: Determine a second statistical indicator corresponding to the window data, and based on the correspondence between the window size, step size, the second statistical indicator and the misalignment artifact label, construct a misalignment artifact recognition model of the misalignment artifact recognition index and the window size, step size and the second statistical indicator.

8. The method according to claim 7, characterized in that, The second statistical indicator includes texture feature information; The determining of the second statistical indicator corresponding to the window data and constructing a misalignment artifact recognition model of a misalignment artifact recognition index and the second statistical indicator based on the corresponding relationship between the second statistical indicator and the misalignment artifact label includes: Determine a texture matrix corresponding to the window data, and extract texture feature information based on the texture matrix; Based on the corresponding relationship between the texture feature information and the misalignment artifact label, a texture mapping relationship model between each of the texture feature information and the misalignment artifact recognition index is obtained by fitting.

9. A method for identifying misregistration artifacts, characterized in that, include: Get the image to be processed; Based on the window size and step size in the misalignment artifact recognition strategy, performing window data interception on the image to be processed to obtain a plurality of window data; Determine the misalignment artifact description parameters corresponding to each window data based on the misalignment artifact recognition model in the misalignment artifact recognition strategy; Determine a misalignment artifact window based on each of the misalignment artifact description parameters and a corresponding threshold value, and determine a misalignment artifact position based on the misalignment artifact window; The misalignment artifact recognition strategy is predetermined based on the misalignment artifact recognition strategy determination method according to any one of claims 1 to 8.

10. The method according to claim 9, wherein The misalignment artifact recognition model includes a variance scatter plot model and a variance threshold, and the misalignment artifact description parameter includes a variance response value; Determining the misalignment artifact description parameters corresponding to each window data based on the misalignment artifact recognition model in the misalignment artifact recognition strategy includes: The window data is input into the variance scatter plot model to obtain the variance of the window data, the variance threshold is compared with the variance of the window data, and the variance response value corresponding to the window data is determined according to the comparison result.

11. The method according to claim 10, characterized in that, The misalignment artifact recognition model includes a frequency-domain information scatter plot model and a frequency-domain threshold, and the misalignment artifact description parameter includes a frequency-domain information response value; Determining the misalignment artifact description parameter corresponding to each window data based on the misalignment artifact recognition model in the misalignment artifact recognition strategy includes: Inputting the window data into the frequency-domain information scatter plot model to obtain the high-frequency statistical information and low-frequency statistical information of the window data, comparing the frequency-domain threshold with the high-frequency statistical information of the window data, and comparing the low-frequency threshold with the low-frequency statistical information of the window data, and determining the frequency-domain information response value corresponding to the window data according to the high-frequency information comparison result and the low-frequency information comparison result.

12. The method according to claim 10, wherein The misalignment artifact recognition model includes a texture mapping relationship model, and the misalignment artifact description parameter includes a misalignment artifact index; Determining the misalignment artifact description parameter corresponding to each window data based on the misalignment artifact recognition model in the misalignment artifact recognition strategy includes: Extracting the texture feature information corresponding to the window data, inputting the texture feature information into the texture mapping relationship model, and obtaining the misalignment artifact index of the window data output by the texture mapping relationship model.

13. The method according to claim 10, wherein The misalignment artifact recognition model includes a neural network model, and the misalignment artifact description parameter includes a window recognition type; Determining the misalignment artifact description parameter corresponding to each window data based on the misalignment artifact recognition model in the misalignment artifact recognition strategy includes: Inputting the window data into the neural network model to obtain the window recognition type of the window data output by the neural network model.

14. The method according to claim 11, wherein Determining the misalignment artifact position based on the misalignment artifact window in the image to be processed includes: Determining continuous misalignment artifact windows according to the adjacent relationship of each window; For any group of continuous misalignment artifact windows, determining the corresponding misalignment artifact position based on the sum of the window start position and half of the window depth in at least one continuous misalignment artifact window.

15. An apparatus for determining a misregistration artifact recognition strategy, characterized in that, Including: A data acquisition module for acquiring misalignment artifact training samples and multiple groups of image cropping parameter combinations including window size and step size; A model training module for, for any image cropping parameter combination, performing window data cropping on the misalignment artifact training samples based on the window size and step size in the image cropping parameter combination, and training to obtain the misalignment artifact recognition model corresponding to the image cropping parameter combination based on the cropped window data and the misalignment artifact label corresponding to each window data; A model verification module for acquiring misalignment artifact verification samples, verifying the misalignment artifact recognition models corresponding to each image cropping parameter combination based on the misalignment artifact verification samples, and determining the artifact recognition accuracy of each misalignment artifact recognition model; A recognition strategy determination module for determining the misalignment artifact recognition model that meets the artifact recognition accuracy screening condition and the image cropping parameter combination corresponding to the selected misalignment artifact recognition model as the misalignment artifact recognition strategy.

16. A misregistration artifact recognition device, characterized in that Including: An image acquisition module for acquiring the image to be processed; A window data interception module, used for intercepting window data of the image to be processed based on the window size and step size in the misalignment artifact recognition strategy to obtain a plurality of window data; A description parameter determination module, used to determine the misalignment artifact description parameters corresponding to each window data based on the misalignment artifact recognition model in the misalignment artifact recognition strategy; A misalignment artifact recognition module, used to determine a misalignment artifact window based on each of the misalignment artifact description parameters and a corresponding threshold value, and determine a misalignment artifact position based on the misalignment artifact window; The misalignment artifact recognition strategy is predetermined based on the misalignment artifact recognition strategy determination method according to any one of claims 1 to 8.

17. An electronic device, characterized in that, The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for determining the misalignment artifact identification strategy described in any one of claims 1-8, and / or the method for identifying the misalignment artifact described in any one of claims 9-14.

18. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method for determining a misalignment artifact identification strategy according to any one of claims 1 to 8, and / or the method for identifying a misalignment artifact according to any one of claims 9 to 14 when executed.

Images