CN117192458B

CN117192458B - Magnetic resonance image generation method for eliminating motion artifact based on dynamic tracking technology

Info

Publication number: CN117192458B
Application number: CN202311297285.5A
Authority: CN
Inventors: 张海燕; 姚瑜; 周倩
Original assignee: Wuxi Fifth Peoples Hospital
Current assignee: Wuxi Fifth Peoples Hospital
Priority date: 2023-10-09
Filing date: 2023-10-09
Publication date: 2024-03-26
Anticipated expiration: 2043-10-09
Also published as: CN117192458A

Abstract

The invention relates to the technical field of intelligent image processing, in particular to a magnetic resonance image generation method for eliminating motion artifacts based on a dynamic tracking technology. The method comprises the steps of acquiring an initial magnetic resonance image and user motion data; performing motion periodic feature analysis on the initial magnetic resonance image by using the user motion data to generate user motion time sequence data; comparing the user motion time sequence data with a preset standard time sequence to generate periodic motion data and non-periodic motion data; generating a feedback signal based on the periodic motion data and the non-periodic motion data to obtain a real-time motion feedback signal; performing model region segmentation on the user motion data and the motion enhancement image by a three-dimensional modeling technology to generate user human body three-dimensional region data; according to the invention, the quality and reliability of the magnetic resonance image are improved by carrying out periodic characteristic analysis and visual angle tracking compensation on the artifacts.

Description

Magnetic resonance image generation method for eliminating motion artifact based on dynamic tracking technology

Technical Field

The invention relates to the technical field of intelligent image processing, in particular to a magnetic resonance image generation method for eliminating motion artifacts based on a dynamic tracking technology.

Background

Magnetic Resonance Imaging (MRI) is a non-invasive medical imaging technique for generating high-resolution human tissue images by using nuclear magnetic resonance phenomenon, however, motion artifacts often occur during MRI image generation, which are caused by image blurring and deformation due to physiological motion or respiration of a user, and early MRI techniques did not consider the problem of motion artifacts, so that image quality was severely limited, and image reconstruction algorithms were improved with the development of computer technology, which provides new possibilities for overcoming motion artifacts. With advanced motion tracking techniques, MRI scanning can monitor patient motion in real-time and correct for during image reconstruction. The method obviously reduces the influence of motion artifact and improves the image quality, however, the traditional motion artifact eliminating method generally cannot effectively distinguish the physiological periodic artifact and the aperiodic artifact, only depends on a static correction method, ignores the dynamic characteristic of the image, and further leads to lower quality of the magnetic resonance image.

Disclosure of Invention

Based on the foregoing, there is a need for a magnetic resonance image generation method based on dynamic tracking technology to avoid motion artifacts, so as to solve at least one of the above-mentioned problems.

In order to achieve the above object, a magnetic resonance image generation method for eliminating motion artifacts based on a dynamic tracking technology, the method comprises the following steps:

step S1: acquiring an initial magnetic resonance image and user motion data; performing motion periodic feature analysis on the initial magnetic resonance image by using the user motion data to generate user motion time sequence data; comparing the user motion time sequence data with a preset standard time sequence to generate periodic motion data and non-periodic motion data; generating a feedback signal based on the periodic motion data and the non-periodic motion data to obtain a real-time motion feedback signal;

step S2: performing model region segmentation on the user motion data and the motion enhancement image by a three-dimensional modeling technology to generate user human body three-dimensional region data; carrying out artifact interval measurement on the three-dimensional region data of the human body of the user by utilizing the real-time feedback signal to generate a physiological period artifact image; performing image level separation on the three-dimensional region data of the human body of the user to generate streak-shaped shadow level data; comparing the streak-shaped shadow level data with a preset standard level to generate a non-physiological periodic artifact image;

step S3: signal marking is carried out on the physiological periodic artifact image and the non-physiological periodic artifact image according to the real-time motion feedback signal, and a first feedback signal and a second feedback signal are generated; classifying the breathing phases of the autonomous movement region data based on the first feedback signal, and generating breathing phase classification data; performing image visual angle compensation on the respiratory phase classification data according to a gating compensation mechanism, so as to generate a standard period artifact correction image;

Step S4: performing image edge detection on the passive motion region data through a second feedback signal, and periodically moving an edge image set and a periodically moving time stamp; calculating an image difference value of the periodic motion edge image set to obtain a periodic motion difference influence coefficient; performing image subtraction on the periodic motion edge image set according to the periodic motion difference influence coefficient and the periodic motion time stamp to generate a motion edge difference image;

step S5: performing view locking according to the motion edge difference image to generate a first locking view; resolving pixel point resolution of the motion edge difference image to obtain edge pixel resolution data; model prediction is carried out on the edge pixel resolution data through a support vector machine algorithm, and the coincident images are corrected in a motion mode; performing view locking based on the motion correction coincident images to generate a second locking view; performing local masking on the motion edge difference image according to the first locking visual angle and the second locking visual angle, so as to generate a standard non-physiological artifact correction image;

step S6: carrying out multi-mode image fusion on the standard physiological artifact correction image and the standard non-physiological artifact correction image to generate an artifact fusion correction image; and carrying out data transfer on the artifact fusion correction image by using an HTTP protocol to generate a magnetic resonance image of the user.

According to the invention, the motion data of the user is utilized for feedback in real time, and the parameters of magnetic resonance can be adjusted according to the motion state of the user, so that a more stable and higher-quality image is obtained, more personalized motion advice and feedback can be provided for each user according to the comparison of the motion time sequence data of each user and a preset standard time sequence, thereby helping the user to perform motion more effectively, and more accurate motion advice can be provided for the user through the comparison of the periodic motion data and the non-periodic motion data, thereby helping the user to improve the motion effect and efficiency; the three-dimensional region of the human body of the user can be segmented more accurately through the three-dimensional modeling and region segmentation technology, so that clearer and accurate magnetic resonance images can be generated, artifact distance measurement is carried out on the three-dimensional region data of the human body of the user through real-time feedback signals, artifacts in the magnetic resonance images can be reduced, the accuracy of the images is improved, the possibility of misdiagnosis is reduced, and non-physiological periodic artifacts can be detected and identified through comparing streak-shaped image level data with preset standard levels, so that the image quality is further improved, and interference factors are reduced; by signal marking and processing of physiological periodic artifacts and non-physiological periodic artifact images, it may be helpful to identify and distinguish between different types of artifacts. The method is beneficial to reducing the influence of the artifact on a final image, improving the quality and usability of the image, classifying the breathing phases of the autonomous motion region data based on the first feedback signal, dividing the motion into different breathing phases more accurately, improving the spatial resolution and accuracy of the image, compensating the image visual angle of the breathing phase classification data by using a gating compensation mechanism, and correcting the image offset caused by the breathing motion. This helps to generate a more accurate standard period artifact corrected image, reducing motion-related image blur; by performing image edge detection on the passive motion region data, motion edges in the image can be more clearly identified. This helps to improve the resolution and accuracy of the image, and periodic motion can be analyzed by periodic motion edge image sets and periodic motion time stamps. The method can be used for researching and knowing the characteristics and modes of periodic motion, calculating the difference value between periodic motion edge image sets by using an image difference influence discrimination formula, and quantifying the intensity and change of motion, thereby helping understand and process motion interference in images and improving the quality and usability of the images; and performing view locking through the motion edge difference image, generating a first locking view, and performing model prediction on edge pixel resolution data by using a support vector machine algorithm, so as to realize motion correction and generating a coincident image. The process is helpful for eliminating image blurring and distortion caused by motion, improving the definition and accuracy of images, and realizing motion correction by carrying out model prediction on edge pixel resolution data through a support vector machine algorithm. The support vector machine is a powerful machine learning algorithm and can be used for modeling and predicting complex data, so that the accuracy and efficiency of motion correction can be improved; the standard physiological artifact corrected image and the standard non-physiological artifact corrected image are subjected to multi-modal image fusion, which means that the images from the different correction methods are combined into one single image. The method can increase the information content of the image, improve the correction effect on physiological and non-physiological artifacts, help to improve the quality and usability of medical images, and generate an artifact fusion correction image through fusion processing. This image is typically of higher quality and accuracy because it integrates information from multiple sources, helping to improve the readability and diagnostic accuracy of the medical image. Therefore, the invention comprehensively processes different types of motion artifacts through multiple technologies such as real-time feedback, periodic feature analysis, visual angle compensation, artifact correction and the like, and improves the quality and reliability of the magnetic resonance image.

The method has the advantage that the periodic motion and the aperiodic motion in the image are identified by analyzing the motion data of the user. This helps to eliminate or mitigate artifacts introduced by user motion, thereby improving the quality and accuracy of the image, using the user's three-dimensional region data in combination with real-time feedback signals to measure artifact spacing and generate physiological periodic artifact images. At the same time, non-physiological periodic artifact images are generated by image-level separation. This helps to distinguish and correct different types of artifacts, thereby improving the quality of the image, classifying the respiratory phase with the first feedback signal, and applying the gating compensation mechanism for view compensation, generating a standard period artifact corrected image. This helps to mitigate artifacts introduced by respiratory motion, improving the sharpness and readability of the image. And detecting the motion edge through the second feedback signal, and generating a motion edge difference image by utilizing image difference value calculation and image subtraction. This helps to identify and correct motion artifacts in the image, improving the quality of the image. And generating a locked view angle according to the motion edge difference image, and performing motion correction by applying pixel resolution analysis and a support vector machine algorithm. Through the local mask, a standard non-physiological artifact correction image can be generated, and the quality of the image is further improved. And carrying out multi-mode image fusion on the standard physiological artifact correction image and the standard non-physiological artifact correction image. This may provide more comprehensive information, improving image quality and visualization. The finally generated artifact fusion corrected image can be delivered to the user via the HTTP protocol, which helps to obtain high quality magnetic resonance images. Therefore, the invention comprehensively processes different types of motion artifacts through multiple technologies such as real-time feedback, periodic feature analysis, visual angle compensation, artifact correction and the like, and improves the quality and reliability of the magnetic resonance image.

Drawings

FIG. 1 is a flow chart showing steps of a method for generating a magnetic resonance image based on a dynamic tracking technique to eliminate motion artifacts;

FIG. 2 is a flowchart illustrating the detailed implementation of step S2 in FIG. 1;

FIG. 3 is a flowchart illustrating the detailed implementation of step S3 in FIG. 1;

FIG. 4 is a flowchart illustrating the detailed implementation of step S35 in FIG. 3;

the achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.

Detailed Description

The following is a clear and complete description of the technical method of the present patent in conjunction with the accompanying drawings, and it is evident that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.

Furthermore, the drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. The functional entities may be implemented in software or in one or more hardware modules or integrated circuits or in different networks and/or processor methods and/or microcontroller methods.

It will be understood that, although the terms "first," "second," etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.

To achieve the above objective, referring to fig. 1 to 4, a magnetic resonance image generating method for eliminating motion artifacts based on a dynamic tracking technique, the method comprises the following steps:

In the embodiment of the present invention, as described with reference to fig. 1, a schematic flow chart of steps of a magnetic resonance image generating method for eliminating motion artifacts based on a dynamic tracking technique according to the present invention is provided, in this example, the magnetic resonance image generating method for eliminating motion artifacts based on a dynamic tracking technique includes the following steps:

in an embodiment of the invention, the magnetic resonance image of the user is acquired by using a medical device such as an MRI scanner. These images are typically two-dimensional or three-dimensional, while data relating to patient motion is collected. This may include physiological information such as patient motion trajectory, posture, respiratory rate, heart beat, etc., and motion periodicity signature analysis of the initial magnetic resonance image. This includes detecting motion artifacts that may be present in the image due to patient motion during scanning, using image processing and analysis techniques, such as image registration, motion artifact correction algorithms, and the like, to identify and quantify the nature and intensity of the motion artifacts. Based on the motion feature analysis, motion time series data of the user is generated. Such data includes information about the type, magnitude and frequency of the motion. The time series of movements may be continuous, recording the change in movement, or discrete, recording the state of movement at a particular point in time. And comparing the generated user motion time sequence data with a preset standard time sequence. The standard time sequence may be a motion pattern in an ideal case without motion. Based on the comparison result, the motion data is divided into periodic motion data and non-periodic motion data. Periodic motion data represents regular motion in the image, while non-periodic motion data represents irregular or random motion. Using the periodic motion data and the non-periodic motion data, a real-time motion feedback signal may be generated. This feedback signal may be used for various purposes, such as dynamic image correction, respiratory synchronous scanning, etc. The specific application depends on the design and requirements of the medical image processing system.

in the embodiment of the invention, the motion data and the motion enhancement image of the user are used for region segmentation of the human body model by using a three-dimensional modeling technology. This may include using computer vision and image processing techniques to identify and segment different parts of the human body, such as the head, chest, abdomen, etc. Based on the result of the model region segmentation, three-dimensional region data of the user's body are generated, which data can be used for subsequent processing steps. And measuring the three-dimensional region data of the human body of the user by using the real-time feedback signals so as to identify and quantify the intervals of the artifacts. The artifact spacing measurements may help determine periodic features of the artifact to further distinguish between physiological and non-physiological artifacts. Based on the results of the artifact pitch measurements, a physiological periodic artifact image is generated. These images may represent artifacts related to the physiological motion cycle. And (3) performing image level separation on the three-dimensional region data of the human body of the user, and decomposing the image into different levels for further analysis and processing. And comparing the striped image level data obtained by separation with a preset standard level to detect and identify the non-physiological periodic artifact. Based on the result of the comparison, a non-physiological periodic artifact image is generated. These images may represent artifacts related to non-physiological motion or other factors.

in the embodiment of the invention, physiological activities such as respiration, heartbeat and the like are detected from real-time motion feedback signals, physiological cycle artifact images are marked, a first feedback signal is generated, non-physiological cycle artifact images are marked, a second feedback signal is generated, autonomous motion area data is classified according to the first feedback signal (namely signals related to the physiological cycle artifact), the respiratory activities are divided into different phases such as inspiration, expiration, apnea and the like, respiratory phase classification data is generated, and a so-called gating technology is used, which is a method for acquiring images in a specific respiratory phase (such as in the case of apnea) so as to reduce motion artifacts caused by respiration, and according to a gating compensation mechanism, adjustment and compensation of image viewing angles are performed on the respiratory phase classification data, which can be realized through a software algorithm, so that image distortion caused by respiration is reduced or eliminated. In connection with the above processing steps, a standard periodic artifact corrected image is generated that should minimize or eliminate image distortions and artifacts due to respiratory or other physiological activities.

in the embodiment of the invention, the edge detection is performed on the data of the passive motion area by using an image processing technology, such as a Canny edge detection algorithm or a Sobel operator, the edge detection can help to identify the motion edge in the image, namely the object boundary or the outline, a periodic motion edge image set is determined based on a second feedback signal, the images represent the periodic motion condition of different time points, the corresponding periodic motion time stamp of each image is recorded so as to facilitate the subsequent calculation and analysis, an image difference influence discrimination formula is formulated, which can be a mathematical equation or algorithm for calculating the difference value between the images in the periodic motion edge image set, the formula can consider factors such as the change of the pixel value, the color difference and the like so as to evaluate the similarity or the difference between the images, the periodic motion difference influence coefficient can be calculated based on the values after the difference value is obtained, the calculated difference value is used, the periodic motion difference influence coefficient is calculated according to the formula, the influence degree of the periodic motion on the image is recorded, the image can be used for correcting the image in the subsequent steps, the periodic motion is influenced by the periodic motion, the difference coefficient is calculated according to the formula, the periodic motion difference coefficient is calculated, the periodic motion influence coefficient is more accurately, the periodic motion is subtracted, the periodic motion is influenced by the motion edge is more accurately, or the motion is more influenced by the motion edge is more greatly reduced, the motion is influenced by the motion region.

in the embodiment of the invention, an initial locking view angle is firstly determined by utilizing a motion edge difference image, the view angle is usually the view angle with the smallest difference with the motion edge, the view angle can be realized by calculating the pixel value or the characteristic in the motion edge difference image, the view angle with the smallest difference is selected as a first locking view angle, the motion edge difference image is subjected to pixel resolution analysis, which means that the actual size or the physical size of each pixel point is determined, the actual size or the physical size of each pixel point can be calculated by known camera parameters and imaging conditions, so that subsequent motion correction and correction image generation can be realized, edge pixel resolution data is used as a training set, a model is established by using a Support Vector Machine (SVM) or other suitable machine learning algorithm, the aim of the model is to predict the resolution of each pixel point in the image, the pixel size in the correction image can be helped, predicting edge pixel resolution using a model and applying it to a motion corrected coincident image, motion correcting the motion edge difference image using the predicted edge pixel resolution data, the motion correcting being aimed at correcting pixel positions in the image to eliminate motion effects and thereby obtain a more accurate image, determining a second lock view based on the motion corrected image, a similar method to the first lock view selection but this time the selected view should be more accurately locked to the actual position of the object under test, performing a partial masking operation on the motion edge difference image using the information of the first lock view and the second lock view, the mask may be a binary image containing which parts need to be preserved and which need to be masked or corrected, and finally, by applying the mask to the motion edge difference image, standard non-physiological artifact corrected images can be generated in which problems of motion and resolution have been corrected.

In the embodiment of the invention, images from different modes are combined into a comprehensive image through multi-mode image fusion, so that richer information is obtained or the image quality is improved. In this case, to fuse the standard physiological artifact correction image and the standard non-physiological artifact correction image, common fusion methods include weighted averaging, image fusion algorithm (such as laplacian pyramid, wavelet transform, etc.), a method that needs to be selected to be suitable for application, inputting two correction images into the fusion algorithm, performing image fusion operation, and the result will be an artifact fusion correction image, in which correction information containing physiological and non-physiological artifacts is included, HTTP (Hypertext Transfer Protocol) is a protocol for transmitting data on the Web, an HTTP server needs to be set up to receive, store and transfer the artifact fusion correction image, at the server, a Web framework (such as Django, flashk, etc.) may be used to create an endpoint that accepts image upload, a user may communicate with the HTTP server through a client application or Web interface, the user may upload related data or provide necessary information to request to generate a magnetic resonance image, the server receives the request, process the artifact fusion correction image, apply any necessary post-processing steps (such as image re-sampling or conversion), and generate a magnetic resonance image that may be returned to the user through the user via a browser or via a client application program that may access the magnetic resonance image.

Preferably, step S1 comprises the steps of:

step S11: acquiring an initial magnetic resonance image by using an image acquisition device; acquiring user motion data by using a motion detection sensor;

step S12: carrying out data denoising on the user motion data based on a digital signal processing technology to generate user motion denoising data; performing Gaussian filtering on the user motion denoising data to generate user motion filtering data; performing data standardization on the user motion filtering data according to a maximum-minimum standardization method to generate user motion standard data;

step S13: performing motion feedback on the initial magnetic resonance image according to the user motion standard data to generate a motion feedback image; performing image enhancement on the motion feedback image to generate a motion enhanced image;

step S14: performing motion periodic characteristic analysis on the motion enhanced image and the user motion standard data by a signal processing method to generate a user motion periodic variation graph; extracting time sequence change characteristics of the user motion cycle change graph to obtain user motion time sequence data; comparing the user motion time sequence data with a preset standard time sequence, and marking the user motion time sequence data as periodic motion data when the user motion time sequence data is greater than or equal to the standard time sequence; when the user motion time sequence data is smaller than the standard time sequence, marking the user motion time sequence data as non-periodic motion data;

Step S15: constructing a user motion trail according to the periodic motion data and the non-periodic motion data, and generating user motion trail data; and generating a feedback signal based on the user motion trail data and the user motion period change graph, so as to obtain a real-time motion feedback signal.

The invention can reduce motion artifact, improve image quality and help to more accurately identify image abnormality by integrating the motion data and the magnetic resonance image, can be used for guiding a user to adjust the motion during the magnetic resonance imaging by the real-time motion feedback signal so as to reduce the problem caused by unnecessary motion, can better understand the motion mode of the user by the periodic feature analysis and the time sequence change feature extraction, and can help to study the physiological process related to the motion and improve the quality and the accuracy of the magnetic resonance imaging.

In an embodiment of the invention, the initial magnetic resonance image is acquired by using an image acquisition device. This may be a magnetic resonance imaging device for capturing image information of the interior of a human body, using a motion detection sensor (such as an accelerometer or gyroscope) to acquire motion data of the user. These sensors may be placed on the body of the user to measure their movements and gestures, and digital signal processing is performed on the user movement data to remove noise and unwanted fluctuations, generating user movement denoising data. This may include gaussian filtering the denoised user motion data using a filtering technique or smoothing algorithm to further reduce noise, generating user motion filtered data, normalizing the user motion filtered data using a max-min normalization method to ensure that the data is within a certain range, and generating user motion standard data that is applied to the initial magnetic resonance image based on the user motion standard data to generate a motion feedback image. This may involve registering (registration) the motion standard data with the image to correct the image to eliminate motion-induced artifacts, and image enhancement processing the generated motion feedback image to improve the quality and sharpness of the image to generate a motion enhanced image. The image enhancement may include contrast enhancement, noise reduction, and other techniques, and the signal processing method is used to analyze the motion enhanced image and the user motion standard data to capture the motion periodic characteristics of the user, generate a motion periodic variation graph of the user, which may include information of frequency, amplitude and other periodic characteristics of the motion, and perform time series analysis on the motion periodic variation graph of the user to extract motion time series data of the user, which may be used to describe the motion variation trend, and construct a motion track of the user according to the periodic motion data and the non-periodic motion data. This may be a path or trajectory of the user over a period of time, generating a real-time motion feedback signal based on the user motion trajectory data and the user motion cycle profile. This signal may be used to guide the user's movements, providing real-time movement advice or improvement.

Preferably, step S2 comprises the steps of:

step S21: performing three-dimensional modeling on the user motion data and the motion enhancement image through a three-dimensional modeling technology to generate a user human body three-dimensional model; performing model region segmentation on the user human body three-dimensional model based on the real-time motion feedback signal to generate user human body three-dimensional region data, wherein the user human body three-dimensional region data comprises autonomous motion region data and passive motion region data;

step S22: the real-time feedback signals are utilized to carry out phase coding direction analysis on the autonomous motion area data and the passive motion area data, and motion phase coding direction data is generated;

step S23: performing artifact distance measurement on the motion phase encoding direction data to obtain motion artifact distance data; performing interval measurement on the motion artifact interval data to obtain artifact interval measurement data; comparing the data of the artifact distance measurement data, and generating a physiological period artifact image when the artifact distance measurement data are equal;

step S24: carrying out artifact level segmentation on the three-dimensional region data of the human body of the user according to the motion phase coding direction data to generate a streak-shaped shadow image; performing image level separation on the streak image to generate streak image level data; comparing the streak-like image level data with a preset standard level, and generating a non-physiological periodic artifact image when the streak-like image level data is greater than or equal to the standard level; and when the streak image level data is smaller than the standard level, eliminating the streak image.

According to the invention, the three-dimensional modeling technology is used for fusing the motion data and the motion enhancement image of the user to generate the three-dimensional model of the human body of the user, so that more accurate surface shape and posture information of the human body of the user can be provided, and the three-dimensional model of the human body of the user is divided into different areas including an autonomous motion area and a passive motion area through model area segmentation. The method is helpful for identifying and analyzing the motion behaviors of different parts, and the phase coding direction analysis is carried out on the autonomous motion region data and the passive motion region data by utilizing the real-time motion feedback signals. This may help to determine the direction and nature of the motion, e.g. the distinction between autonomous and passive motion, and by phase encoding the user's motion pattern and motion may be better understood, artifacts being common problems in magnetic resonance images, typically due to motion or other factors. Artifact spacing measurement is a method for measuring the spacing and nature of artifacts, which when the artifact spacing measurement data is equal, indicates that physiological cycle artifacts may be present, which may be related to the physiological motion cycle of the user. Generating physiological periodic artifact images is helpful for identifying and distinguishing real signals and artifacts, and artifact level segmentation is carried out on three-dimensional region data of a user human body by utilizing motion phase encoding direction data. This helps to separate the artifact from the real signal for further analysis and processing, and the streak level data is compared to a pre-set standard level, which can be used to detect non-physiological periodic artifacts. This helps to identify artifacts and take action to reject or correct.

As an example of the present invention, referring to fig. 2, the step S2 in this example includes:

in the embodiment of the invention, the motion data of the user can be collected and can comprise a motion sensor, a video captured by a camera or other related data sources, a motion enhanced image is acquired, possibly through Magnetic Resonance Imaging (MRI) or other medical imaging technologies, the motion data and the image are preprocessed to remove noise and unnecessary information, the quality and accuracy of the data are ensured, and the motion data and the motion enhanced image are fused by utilizing a three-dimensional modeling technology to generate a three-dimensional model of the human body of the user. This may be achieved by computer vision and image processing techniques such as structured light scanning, stereo matching, point cloud reconstruction, etc. to obtain motion feedback signals from motion sensors or other real-time data sources. These signals may include information of joint angle, muscle activity, movement speed, etc. for tracking the actual movement of the user, and based on real-time movement feedback signals, the three-dimensional model of the user's body is segmented into regions. This may be achieved by identifying and segmenting different body parts, such as the head, chest, arms, legs, etc., region segmentation may require the use of computer vision and image segmentation algorithms, combined with motion feedback data to determine the location and shape of each region, generating three-dimensional region data of the user's body on the basis of model region segmentation. This includes autonomous and passive movement zone data, classified and recorded according to the real-time movement of the different zones.

in embodiments of the present invention, the feedback signal is obtained from a motion sensor or other real-time data source. These signals may include movement related data such as joint angle, muscle activity, movement speed, etc., and the feedback signal is correlated to the autonomous movement region data and the passive movement region data. This may involve data alignment and calibration work, ensuring that the feedback signals correspond one-to-one to the corresponding motion zone data, and analysis of the autonomous and passive motion zone data using phase encoding techniques. Phase encoding is a method for decomposing a signal into different frequency components, and is commonly used for analyzing periodic motion or vibration, and further analyzing the motion direction based on the phase encoding. This may include determining information such as direction, velocity and acceleration of the motion, and combining the results of the phase encoding and direction analysis to generate motion phase encoding direction data. These data will provide information about the direction and speed of movement, etc.

in the embodiment of the invention, the artifact distance measurement is performed on the motion phase encoding direction data by using a proper algorithm or method. This may involve signal processing and mathematical computation techniques to determine periodic artifacts in the data, and further pitch measurements are made on the artifact pitch data. This may include measuring the period, frequency, or other physiological cycle related characteristics of the artifact. Common methods include fourier transform or wavelet transform, etc., spectral analysis techniques to compare the obtained artifact pitch measurement data. This may include comparing the measured data to reference data or making a decision using a predetermined threshold. When the artifact pitch measurement data are within a certain range, which indicates that a physiological cycle artifact is detected, the system should trigger the process of generating a physiological cycle artifact image when the artifact pitch measurement data are equal. This may include converting the artifact data to an image format and visually rendering. The generation of the image can adopt different image processing and visualization technologies according to specific requirements, so that the result is ensured to be clearly visible.

In the embodiment of the invention, the motion phase encoding direction data is used for carrying out artifact level segmentation on the three-dimensional region data of the human body of the user. This may involve image processing techniques such as thresholding, edge detection, region growing, etc. to separate the artifact levels in the image, generate a streak image, and further image level separation of the generated streak image. This may employ image processing methods such as multi-layer segmentation or filter operation to extract information at different levels, generate streak-like image level data, and compare the generated streak-like image level data with a preset standard level. The standard level is typically determined based on previous studies or experience to distinguish between physiological periodic artifacts and non-physiological periodic artifacts, indicating the presence of non-physiological periodic artifacts in the image when the streak level data is greater than or equal to a preset standard level. At this time, a process of generating a non-physiological periodic artifact image may be triggered, artifact level data is converted into an image, and visualized, and when the streak-like artifact level data is smaller than a preset standard level, it is indicated that no non-physiological periodic artifact exists in the image, which means that the image quality may be better. These images may be selected for rejection or retention, depending on the application requirements.

Preferably, step S3 comprises the steps of:

step S31: signal marking is carried out on the physiological periodic artifact image and the non-physiological periodic artifact image according to the real-time motion feedback signal, and a first feedback signal and a second feedback signal are generated;

step S32: performing pressure sensitive detection on the autonomous movement area data based on the first feedback signal to generate user diaphragmatic movement data; acquiring a motion signal by using a diaphragmatic motion detection device, so as to obtain a diaphragmatic motion signal;

step S33: respiratory cycle detection is carried out through the diaphragmatic movement signals and diaphragmatic movement data of a user, and diaphragmatic respiratory cycle data are generated; performing frequency spectrum conversion on the diaphragmatic respiratory cycle data according to a fast Fourier transform algorithm to generate a diaphragmatic respiratory spectrogram; carrying out frequency domain peak analysis on the diaphragm respiratory spectrogram to obtain respiratory rhythm data;

step S34: carrying out respiratory phase classification on the respiratory rhythm data to generate respiratory phase classification data; performing motion artifact correction on the physiological cycle artifact data according to the respiratory stage classification data to generate a physiological cycle artifact correction image;

step S35: and performing image view compensation on the physiological cycle artifact correction image according to a gating compensation mechanism, so as to generate a standard cycle artifact correction image.

The invention marks the physiological periodic artifact image and the non-physiological periodic artifact image according to the real-time motion feedback signal. This may help to distinguish between physiological and non-physiological artefacts in the image, the generated first and second feedback signals possibly containing information about the nature of the artefacts, the user's diaphragmatic movement data being generated by pressure-sensitive detection of the autonomous movement region data. This may be a method for detecting respiratory movement and the acquisition of diaphragmatic movement data may utilize diaphragmatic movement detection means to obtain a signal related to respiration. This helps track the respiratory activity of the user, and uses the diaphragmatic motion signal and the user diaphragmatic motion data for respiratory cycle detection, generating diaphragmatic respiratory cycle data. These data are then spectrally transformed by a fast fourier transform algorithm to generate a diaphragm respiratory spectrogram. Frequency domain peak analysis may be used to obtain respiratory rhythm data that helps to understand the frequency and regularity of respiration, and to classify respiratory phase data to generate respiratory phase classification data. This helps to divide the respiratory activity into different phases. According to the respiratory stage classification data, motion artifact correction can be performed on the physiological cycle artifact data to generate a physiological cycle artifact corrected image. This may improve the visualization quality of the physiological cycle artifact, using a gating compensation mechanism to image view compensation of the physiological cycle artifact corrected image. This may be to correct for image offset or artifacts due to breathing or other factors. A standard periodic artifact corrected image is ultimately generated, which will be a more accurate and reliable image.

As an example of the present invention, referring to fig. 3, the step S3 in this example includes:

in the embodiment of the invention, the image data to be processed is required to be acquired, wherein the image data comprises physiological periodic artifact images and non-physiological periodic artifact images. These images may be obtained by medical imaging equipment (such as X-ray, CT scan or MRI) and require real-time motion feedback signals for signal tagging. This may be achieved by different sensors or devices, such as a motion tracking instrument, an Inertial Measurement Unit (IMU), or other biological sensor. These sensors can monitor the body motion of a patient or subject, such as respiration, heartbeat, or other physiological parameters, and mark motion-related portions of each image for physiological periodic artifact images and non-physiological periodic artifact images based on real-time motion feedback signals. The markers may be pixel-level for indicating an artifact region in the image and the first feedback signal may be position and intensity information of the artifact. For example, marking the position of the artifacts and their relative intensity or density in the image, the second feedback signal may contain more detailed motion information such as the motion speed, direction, etc. of the artifacts, and for each image, a data structure of the first and second feedback signals is generated from the result of the marking, which may involve encoding the marked information into numbers or vectors and correlating it with the image data, which may be further used in subsequent data processing, image correction or analysis steps to correct the artifacts, improve the image quality, or perform other tasks related to the biomedical image.

in embodiments of the present invention, the autonomous movement area data, such as diaphragmatic movement, is detected by selecting an appropriate sensor. Common sensors include pressure sensors, strain sensors, capacitive sensors, etc. that can be used to monitor muscle activity, the sensors being mounted in the autonomous movement region of the user, typically in the chest region, to monitor diaphragmatic muscle movement. The sensor is ensured to be in proper contact with the body of a user, the movement data can be accurately recorded, and signals such as pressure or strain of an autonomous movement area are processed in real time by using the data acquired by the sensor. This may involve signal amplification, filtering and denoising to ensure accurate data recording, extracting features relating to diaphragmatic muscle movement from the pressure sensitive detected data. The characteristics may include amplitude, frequency, amplitude, etc. information of the movement, and based on the extracted characteristics and the first feedback signal, user diaphragmatic movement data is generated. These data will reflect the user's diaphragmatic motor activity, be associated with the first feedback signal, use specialized equipment or devices such as electrode strips, surface electrodes, chest sensors, etc. for detecting and recording the user's diaphragmatic motor signals, install diaphragmatic motor detection means, typically on the user's chest or abdomen, to monitor diaphragmatic motor. Ensuring correct installation position of the equipment so as to obtain accurate signals, and acquiring diaphragmatic motion signals by using a diaphragmatic motion detection device. These signals may be Electromyography (EMG) signals, recording muscle electrical activity, or other biological signals related to diaphragmatic movement, processing the acquired signals, including amplification, filtering, denoising, etc., to obtain high quality diaphragmatic movement data, which may include amplitude, frequency, time and frequency domain characteristics of movement, etc., of the user via the signal acquisition device.

in the embodiment of the present invention, the user' S diaphragmatic movement data and diaphragmatic movement signals obtained by using the aforementioned step S32. Ensuring that the data has been sampled and processed to make it suitable for subsequent analysis, a threshold or feature extraction technique is used to detect respiratory cycles in the diaphragmatic motor signal. This may include identifying peaks or feature points of diaphragmatic movement to determine the start and end of a breathing cycle, once the start and end of a breathing cycle is detected, diaphragmatic breathing cycle data may be generated, including information on duration, frequency, etc. of the breathing cycle, using the breathing cycle data generated by the preceding steps, performing a fast fourier transform on the breathing cycle data, converting time domain data into frequency domain data, and generating a diaphragmatic breathing spectrogram from the FFT transform results. The spectrogram shows the intensities of different frequency components in the respiratory signal, and the diaphragmatic respiratory spectrogram is analyzed to identify peaks in the frequency domain. These peaks correspond to different respiratory frequency components, and frequency and intensity information of the peaks is extracted. These frequencies may correspond to different breathing rhythms or breathing frequencies, which translate the extracted frequency information into breathing rhythm data, which may include information of breathing frequency, breathing depth, etc.

in the embodiment of the present invention, by using the respiratory rhythm data generated in the previous step S33 as input, features such as respiratory rate, respiratory depth, respiratory duration, etc. may be extracted from the respiratory rhythm data, and classification of the respiratory rhythm data may be performed using an appropriate machine learning algorithm (such as a support vector machine, a random forest, a deep learning neural network, etc.). The algorithms may divide the respiratory rhythm data into different respiratory phases, such as deep breath, shallow breath, normal breath, etc., assign a corresponding respiratory phase label to each time period based on the results of the classification algorithm, generate respiratory phase classification data, use the respiratory phase classification data and previously acquired physiological cycle artifact data, and design an artifact correction algorithm using the respiratory phase classification data. This algorithm may take into account different characteristics of the respiratory phases, such as respiratory motion amplitude and frequency of the different phases. The goal of the correction algorithm is to eliminate or reduce respiratory related variations in the artifact data, and the artifact correction algorithm is used to correct the physiological periodic artifact data to generate a physiological periodic artifact corrected image. These corrected images should reflect physiological signals more accurately, without being affected by respiration.

In the embodiment of the present invention, by using the physiological periodic artifact correction image generated in the previous step S34 as input data, the gating compensation mechanism is a method for adjusting the image according to a specific condition. In this context, gating compensation may be a technique of adjusting an image according to a specific phase or characteristic of a physiological cycle, determining specific factors of the physiological cycle to be compensated, such as a specific phase or a specific frequency range of the cycle, designing a compensation function based on the selected gating factors, the input of the function being a physiological cycle artifact corrected image, and the output being an image adjusted according to the gating factors. This may involve pixel level adjustment or a transformation of the entire image, applying a designed gating compensation function to the physiological periodic artifact corrected image. This may be achieved by pixel-by-pixel processing or by applying some image processing technique, by applying gating compensation, generating a standard periodic artifact corrected image that is view-angle-adjusted. The image should more accord with standardized visual angle, reduce the influence from specific physiological periodic stage, adjust the parameter in the gating compensation mechanism, in order to optimize the quality and accuracy of the image, evaluate the standard periodic artifact correction image produced, ensure the gating compensation mechanism can carry on the compensation of the visual angle of the image effectively under various conditions.

Preferably, step S35 includes the steps of:

step S351: performing frequency analysis on the physiological periodic artifact correction image to generate a first type of motion artifact correction image; detecting signal intensity areas of the physiological periodic artifact correction images by using a rapid imaging technology to generate second-type motion artifact correction images; performing sequential pulse processing on the second type of motion artifact correction image based on a pulse sequence technology, so as to determine a third type of motion artifact correction image;

step S352: setting a first gating signal according to the first type of motion artifact correction image; setting a second gating signal according to the second type of motion artifact correction image; setting a third gating signal according to the third type of motion artifact correction image;

step S353: calculating contribution degrees of the first gating signal, the second gating signal and the third gating signal through diaphragm motion signals by using a gating contribution degree analysis formula, so as to generate compensation gating factor data;

step S354: performing single excitation sequence scanning on the first type of motion artifact correction image through compensation gating factor data to generate a first compensation visual angle; adding a saturation band to the second type of motion artifact correction image through compensation gating factor data to generate a second compensation visual angle; performing balanced free-standing precession sequence scanning on the third type of motion artifact correction image according to the compensation gating factor data, so as to generate a third compensation visual angle;

Step S355: performing view integration on the first compensation view angle, the second compensation view angle and the third compensation view angle to generate compensation view angle data; the physiological periodic artifact corrected image is dynamically tracked using the compensated view angle data to generate a standard physiological artifact corrected image.

According to the invention, frequency analysis is carried out on the physiological cycle artifact correction image, namely Fourier transformation and other technologies are possibly used for generating the motion artifact correction image related to specific frequency components, a rapid imaging technology is used for detecting a signal intensity area in the physiological cycle artifact correction image, so that the detection and identification of specific types of artifacts can be facilitated, then a second type of motion artifact correction image is generated, a pulse sequence technology is applied to the second type of motion artifact correction image, namely a pulse gradient image acquisition method is possibly used for further determining a third type of motion artifact correction image, the accuracy and efficiency of physiological cycle artifact correction are improved, different types of artifacts are better distinguished and processed through different types of motion artifact correction images and gating signals, the accuracy and adaptability of correction can be improved based on the gating factor calculation of the contribution analysis, and the generated standard correction image can reflect the physiological state of a patient more accurately, so that the image is facilitated to remove the artifacts.

As an example of the present invention, referring to fig. 4, the step S35 in this example includes:

in embodiments of the present invention, by requiring acquisition of physiological periodic artifact corrected images, this can be a common problem in medical imaging (e.g., MRI, CT scanning, ultrasound, etc.). These artifacts may be caused by physiological movements within the living being (e.g., respiration, heartbeat, etc.). Physiological periodic artifacts typically appear in the image at a frequency, and the physiological periodic artifact corrected image is frequency analyzed, which may be accomplished by fourier transform or other frequency domain analysis techniques. The purpose of the frequency analysis is to determine the frequency and characteristics of the artifact and, based on the results of the frequency analysis, to generate a first type of motion artifact correction image that may be an artifact that is generated for respiratory motion. This step may employ filtering or other digital signal processing techniques to suppress or mitigate components whose frequencies match the physiological cycle artifacts, thereby generating a first type of motion artifact corrected image that is used for signal strength region detection using rapid imaging techniques. This may include local contrast enhancement, edge detection or other image segmentation techniques to better identify areas of signal strength variation in the image, and based on the results of the signal strength area detection, a second type of motion artifact correction image is generated, which may be an artifact generated for blood flow or large vessel pulsations. The purpose of this step is to accurately mark and locate areas of physiological periodic artifacts, and to sequence pulse the second type of motion artifact corrected image based on a pulse sequence technique. The pulse sequence technique may be a time-series signal processing method for further reducing or correcting artifacts. This may include applying a special pulse sequence to improve image quality, and after completion of the sequence pulse processing, determining a third type of motion artifact correction image, which may be an artifact for cerebrospinal fluid flow. The image should be fully corrected to eliminate physiological cycle artifacts as much as possible.

in the embodiment of the invention, the image is corrected according to the first type of motion artifact. This may be one of the corrected images generated in step S351 for suppressing or correcting components related to physiological cycle artifacts, and particular embodiments may involve selecting a particular image processing algorithm or filter to identify and emphasize features of a first type of motion artifact, and setting the gating signal may include determining a threshold or rule to separate regions of the first type of motion artifact from other regions, and setting a second gating signal based on a second type of motion artifact corrected image. This is based on the further corrected image generated in step S351 for identifying and marking signal strength areas of physiological periodic artifacts, embodiments may include locating and marking areas of a second type of motion artifact using image segmentation techniques, the setting of the second gating signal may require a threshold or feature detection algorithm to determine which areas are affected by gating, and finally setting a third gating signal based on a third type of motion artifact corrected image. This is the image generated in step S351, after the sequential pulsing process, the artifacts should be corrected as much as possible, and embodiments may involve further signal processing or filtering to ensure maximization of the quality of the third type motion artifact corrected image, and the setting of the third gating signal may be coordinated with the setting of the first and second gating signals to take into account all types of artifact correction in combination.

embodiments of the present invention collect diaphragmatic motor signals by using a suitable physiological monitoring device, such as an electromyography. This may be done on the user by attaching electrodes or other sensors to ensure that the diaphragmatic motion signal is time synchronized with the first, second and third gating signals. This may be accomplished by time stamping or other synchronization mechanism, designing a gating contribution analysis formula that should take into account the correlation between the diaphragmatic motion signal and each gating signal. This may involve some mathematical model or statistical method to quantify the relationship between them, using the data of the diaphragmatic motion signal and the gating signals, applying a gating contribution analysis formula to calculate the contribution of each gating signal. This may require analysis of the signal over a range of times, generating compensation gating factor data based on the contribution calculation. These factors may be used to adjust the corresponding gating signals to minimize the effect of diaphragm movement on the image, and it may be desirable to establish a feedback mechanism to monitor the improvement in image quality and adjust the calculation of the gating factors as needed. This can be achieved by cooperation with a medical professional to ensure that the adjustment is medical standard compliant, performing a verification experiment, evaluating the effect of the compensation gating factor. Optimizing according to experimental results, and ensuring that calculation of the gating factors can inhibit artifacts caused by diaphragm movement to the greatest extent on the premise of not influencing image quality.

in the embodiment of the invention, the original data of the first type motion artifact correction image is utilized, the compensation gating factor data obtained by calculation in advance is applied, the single excitation sequence scanning technology is utilized, the first type motion artifact is corrected according to the compensation gating factor data, the image of the first compensation viewing angle is generated, the compensation gating factor data is applied to the original data of the second type motion artifact correction image, the saturated band adding technology is utilized, the second type motion artifact is corrected by combining the compensation gating factor data, the image of the second compensation viewing angle is generated, the original data of the third type motion artifact correction image is utilized, the correction is carried out according to the compensation gating factor data, the balanced free steady-state precession sequence scanning technology is utilized, the third type motion artifact is corrected by utilizing the compensation gating factor data, the image of the third compensation viewing angle is generated, the quality evaluation is carried out on the generated first, second and third compensation viewing angle images, the application of artifact correction and compensation gating factor is ensured not to influence the image quality, and parameters of the compensation gating factor data are required to be adjusted according to the actual situation, so that the optimal correction effect is achieved.

In the embodiment of the invention, the image data of the first compensation viewing angle, the second compensation viewing angle and the third compensation viewing angle are integrated. This may involve processing of registration of the images, transformation of the coordinate system, etc. to ensure that they are in the same view and space, with the integrated image data, generating compensated view data. This may be achieved by means of image fusion, interpolation, etc. to obtain smooth and consistent compensated view angle information, and the generated compensated view angle data is used to dynamically track physiological periodic artifacts. This may require consideration of periodic variations in the physiological cycle, such as heartbeat, respiration, etc., and their effects at different viewing angles, employing an appropriate tracking algorithm, possibly model-based tracking, optical flow, etc., to ensure accurate tracking of the physiological cycle artifacts, correcting the physiological cycle artifacts in combination with the compensated viewing angle data obtained by dynamic tracking, generating a standard physiological artifact correction image, which may include steps of phase adjustment, deformation correction, etc., of the image to minimize artifact effects caused by the physiological cycle, performing quality assessment on the generated standard physiological artifact correction image, ensuring that the artifact correction effect meets the requirements of medical imaging, adjusting algorithm parameters according to actual conditions, and optimizing image quality.

Preferably, the gating contribution analysis formula in step S353 is specifically as follows:

where G (T) is expressed as a gating contribution, T is expressed as a continuous range of time, typically from 0 to T, S ₁ (t ') is expressed as the value of the first gating signal at time t', S ₂ (t ') is expressed as the value of the second gating signal at time t', S ₃ (t ') is represented as the value of the third gating signal at time t', beta ₁ Represented as decay rate coefficient, beta, of the first gating signal ₂ Represented as the decay rate coefficient, beta, of the second gating signal ₃ Represented as the decay rate coefficient, alpha, of the third gating signal ₁ Represented as a contribution importance weight, alpha, of the first gating signal ₂ Represented as a contribution importance weight, alpha, of the second gating signal ₃ Denoted as contribution importance weight of the third gating signal, μ denoted as gating contribution analysis anomaly adjustment value.

The invention constructs a gating contribution analysis formula, S in the formula ₁ (t′)、S ₂ (t') and S ₃ (t ') represents the values of the first, second and third gating signals at time t'. By associating these signal values with corresponding weighting coefficients alpha ₁ ，α ₂ And alpha ₃ Multiplication can adjust the importance of each gating signal in contribution calculation by beta ₁ 、β ₂ And beta ₃ Representing the decay rate coefficients of the first, second and third gating signals. The decay rate coefficient controls the degree of decay of the gating signal over time. By adjusting these parameters, the sensitivity of the gating contribution to the gating signal values at different time points can be influenced by α ₁ 、α ₂ And alpha ₃ Represents the contribution importance weights of the first, second and third gating signals. By adjusting these weights, the overall contribution of each gating signal to the gating contribution can be controlled, a larger weight representing that the corresponding gating signal contributes more importantly to the gating contribution, a functional relationship being formed from the correlation between the value of the first gating signal at time t' and the above parameters:

the outlier μ is analyzed by gating contribution for correction of errors and deviations due to complexity and non-idealities of the actual system. The method can correct the difference between theoretical assumption in the formula and an actual system, improves the accuracy and reliability of the gating contribution analysis, generates the gating contribution G (t) more accurately, and simultaneously adjusts parameters such as the attenuation rate coefficient of the first gating signal and the contribution importance weight of the second gating signal in the formula according to actual conditions, thereby adapting to different gating contribution analysis scenes and improving the applicability and flexibility of the algorithm. The formula comprehensively considers the contribution degrees of the gating signals at different time points through weighting and attenuation processing of different gating signals, so that the gating contribution degrees are calculated. By analyzing these effects, information about the importance and contribution of the gating signal can be obtained, providing an important reference for understanding the behavior of the system and optimizing the gating strategy.

Preferably, step S4 comprises the steps of:

step S41: the passive motion area data are subjected to periodic image acquisition through a second feedback signal, so that a periodic motion image set is obtained;

step S42: performing image edge detection on the periodic motion image set to generate a periodic motion edge image set; performing time stamp marking on the periodic motion edge image set to generate a periodic motion time stamp;

step S43: calculating an image difference value of the periodic motion edge image set by using an image difference influence discrimination formula to obtain a periodic motion difference influence coefficient;

step S44: performing image screening on the periodic motion edge image set according to the periodic motion difference influence coefficient and the periodic motion time stamp, so as to generate a high-influence motion edge image and a low-influence motion edge image;

step S45: and carrying out image subtraction on the high-influence moving edge image and the low-influence moving edge image to generate a moving edge difference image.

According to the invention, through the second feedback signal, the system can acquire periodic images aiming at passive movement region data. This helps to capture periodically moving image data, such as heart beat or respiratory motion within a living being. The method has the advantages that the periodic image set is obtained and used for subsequent analysis, and the periodic moving image set is subjected to edge detection and time stamping. This helps determine the location of the edges of motion in the images and creates a time stamp for each image to track the temporal change in motion. The information is critical to subsequent analysis and screening, and the difference value between the periodic motion edge image sets is calculated by using an image difference influence discrimination formula so as to obtain a difference influence coefficient of the periodic motion. The method can help the system to quantify the intensity and influence degree of the motion, is beneficial to subsequent screening and processing, and screens the periodic motion edge image set according to the periodic motion difference influence coefficient and the time stamp to generate a high-influence motion edge image and a low-influence motion edge image. This may help the system focus on high impact moving edges for more accurate analysis and processing, subtracting high impact moving edge images from low impact moving edge images, generating moving edge difference images. This difference image may be useful for visualizing and analyzing the characteristics of the motion, and possibly for correction or processing.

In the embodiment of the invention, by capturing the image data of the passive motion area by using a camera or a sensor, the frequency and duration of capturing the image are dependent on the application requirements, for example, how many frames of images are captured per second, and for video data, the extraction and storage of the image frames may need to be performed for subsequent analysis, and the periodic moving image set needs to be subjected to edge detection and time stamping. The specific steps may include: edge detection is performed on each of the periodic moving images using an image processing algorithm (e.g., sobel, canny, laplacian, etc.) to highlight edge structures in the images; for each processed image, in order to track time, an acquisition time stamp needs to be recorded for subsequent time analysis. This is typically a time stamp in milliseconds or microseconds. For each pair of adjacent periodic motion edge images, a specific image difference calculation method is used to calculate the difference between the two images, the image difference calculation can comprise calculating the difference at the pixel level, using a specific similarity index (such as a structural similarity index-SSIM), and the like, the calculation of a difference value can help to quantify the motion difference between the images, which is very important for the subsequent steps, a threshold is defined based on the difference influence coefficient and the timestamp, so as to divide the images into two types of high influence and low influence, the high influence motion edge images possibly comprise images with significant motion difference, the low influence images comprise images with small change, the aim of this step is to reduce the calculation complexity of the subsequent processing, only focus on the high influence images, and carry out pixel level subtraction operation on each high influence and low influence motion edge image, and the result is one motion edge difference image, wherein the motion difference of the high influence motion edge image relative to the low influence image is reflected, and the important information related to the periodic motion can be used for further analysis or visualization.

Preferably, the image difference influence discrimination formula in step S43 is specifically as follows:

wherein D is expressed as an image difference influence coefficient, G is expressed as a time range of the periodic motion, N is expressed as the number of images in the edge image set, I _i (x, y) is expressed as the pixel value of the ith image at coordinates (x, y), M is expressed as the number of gradient modulus values in the edge image set, P is expressed as the number of time partial derivatives in the edge image set, B _k The weight coefficient expressed as the partial derivative of time,denoted as the time partial derivative of the kth image at coordinates (x, y), dg denoted as the time variable of integration, ω denoted as the image difference influence discrimination anomaly correction.

The invention constructs an image difference influenceDiscrimination formula, first term in formulaThe average of the sum of the absolute values of the second spatial derivatives of each image in the edge image set is measured. It reflects the curvature change of the image edge, is used for capturing the texture and shape difference of the image edge, the second term in the formula +.>A weighted average of the time partial derivatives of each image in the set of edge images is measured, which is used to describe the trend of the edge images in the time dimension, and a functional relationship is formed according to the correlation between the pixel value of the ith image at the coordinates (x, y) and the parameters:

The number of images N in the edge image set and the number of gradient modulus values M can control the sampling and resolution of the image differences. The feature of the edge image can be more comprehensively captured by increasing the number of images and the number of gradient modulus values, the accuracy of difference evaluation is improved, the number of time partial derivatives P and the weight coefficient B _k The sensitivity to time variation can be adjusted. By selecting proper quantity and weight time partial derivatives, the dynamic change of the periodic motion edge image can be better captured, the time range G and the integral time variable dg are used for defining the integral range and step length, the integral summation precision and coverage range of the time dimension are influenced, the abnormal correction quantity omega can be judged through image difference influence, the result of difference evaluation can be corrected and adjusted to adapt to the requirement of a specific application scene, the accuracy and reliability of image difference influence judgment are improved, the image difference influence coefficient D is generated more accurately, and simultaneously the weight coefficient of the time partial derivative in the formula, the quantity of the time partial derivatives in the edge image set and other parameters can be adjusted according to actual conditions, so that the method is suitable for different image difference influence judgment scenes, and the algorithm adaptation is improved Usability and flexibility. By adjusting the combination of parameters and formulas, the image difference influence discrimination formulas can quantify the difference degree of the periodic motion edge image set, and by calculating the image difference coefficient, quantitative evaluation of the image difference is provided, which is helpful for understanding and analyzing the characteristic change of the periodic motion edge, and providing beneficial image processing and analysis results in various applications.

Preferably, step S5 comprises the steps of:

step S51: performing view locking according to the motion edge difference image to generate a first locking view;

step S52: resolving pixel point resolution of the motion edge difference image to obtain edge pixel resolution data; model training is carried out on the edge pixel resolution data through a support vector machine algorithm, and a motion artifact correction model is generated;

step S53: inputting the motion edge difference image into a motion artifact correction model to correct the image artifact, and generating an artifact correction image; carrying out image superposition on the artifact correction image and the motion edge difference image to generate a motion correction superposition image; performing view locking based on the motion correction coincident images to generate a second locking view;

step S54: performing view angle movement path confirmation through the first locking view angle and the second locking view angle to generate movement view angle path data; and carrying out local masking on the motion edge difference image according to the moving view angle path data, thereby generating a standard non-physiological artifact correction image.

The invention generates a first locking visual angle by performing visual angle locking on the motion edge difference image. The purpose of this step is to ensure accurate observation and processing of the specific part of the image, and to resolve the pixel resolution of the motion edge difference image to obtain edge pixel resolution data. Subsequently, the edge pixel resolution data is model trained by a Support Vector Machine (SVM) algorithm to generate a motion artifact correction model. The step involves processing the image resolution and training a machine learning model, and aims to establish a motion artifact correction model according to pixel-level data, input a motion edge difference image into the motion artifact correction model for image artifact correction, and generate an artifact correction image. Then, the artifact correction image and the motion edge difference image are subjected to image superposition to generate a motion correction superposition image. Then, the view angle locking is performed based on the motion corrected coincident image, and a second locked view angle is generated. This step involves performing a plurality of processes on the image to eliminate or attenuate the artifact and to ensure the quality of the image, performing view movement path confirmation through the first lock view and the second lock view, and generating movement view path data. Then, the motion edge difference image is locally masked according to the moving view path data, thereby generating a standard non-physiological artifact correction image. This step involves path validation and local masking of the image to ultimately produce a standard, physiological artifact free corrected image, thereby improving the quality and accuracy of the image.

In the embodiment of the invention, the motion edge difference image is acquired, which may be captured by a camera or other sensors. The image is preprocessed, such as denoising, smoothing, etc., and features of the moving edge are extracted using image processing techniques. This may include determining the first lock-in view by analyzing the motion edge differences using an edge detection algorithm, such as Canny edge detection. This may involve calculating the offset or rotation of the view angles to minimize the edge differences, generating a first lock view angle, performing pixel resolution on the motion edge difference image, obtaining edge pixel resolution data, training the edge pixel resolution data using a Support Vector Machine (SVM) algorithm, and generating a motion artifact correction model. The method may include steps of feature selection, data labeling, model training and the like, a motion artifact correction model is generated, and a motion edge difference image is input into the motion artifact correction model to perform image artifact correction. This may be an image processing process based on a machine learning model, where the artifact correction image and the original motion edge difference image are superimposed to generate a motion correction superimposed image, and the view angle lock is performed based on the motion correction superimposed image to generate a second lock view angle. This may involve further analysis of the image features, confirming the movement path of the view angles by analyzing the first and second lock views, generating movement view angle path data from the confirmed movement path, performing a local masking operation on the movement edge difference image from the movement view angle path data, and generating a standard non-physiological artifact correction image by the local masking operation.

Preferably, step S6 comprises the steps of:

step S61: carrying out multi-mode image fusion on the standard physiological artifact correction image and the standard non-physiological artifact correction image to generate an artifact fusion correction image;

step S62: performing parallel task distribution on the artifact fusion correction image through the cloud computing node to generate parallel task distribution nodes; and carrying out data transfer on the parallel task distribution nodes by using an HTTP protocol to generate a magnetic resonance image of the user.

The present invention fuses the standard physiological artifact correction image and the standard non-physiological artifact correction image to generate an artifact fusion correction image. The multi-modal image fusion may employ various techniques such as image fusion algorithms, deep learning models, and the like. The purpose of this step is to combine the image information of the two different correction modes to improve the image quality and the accuracy of the information, the cloud computing node being a computing resource for processing large-scale data and performing computationally intensive tasks. In this step, the artifact fusion correction image is transmitted to the cloud computing node for processing, and the cloud computing node can process multiple tasks in parallel, which helps to improve processing efficiency and speed. In this step, the artifact fusion corrected image may be split into multiple subtasks and distributed to multiple cloud computing nodes for parallel processing. This can be achieved by task scheduling and parallel computing frameworks, the HTTP protocol being commonly used to transfer data over the internet. Here, the artifact fusion correction image and the task data may be transferred between different computing nodes via the HTTP protocol. This ensures reliable transmission and communication of data, once all sub-tasks have completed processing on the cloud computing node, their results can be collected and combined, generating the final user magnetic resonance image. The image may be a high-quality image containing multi-mode information and subjected to artifact correction and fusion, the data processing speed can be remarkably improved by using cloud computing nodes and parallel task distribution, particularly for large-scale image data, the image quality can be improved by multi-mode image fusion, the artifacts are reduced, more accurate medical diagnosis information is provided, and the safe transmission and reception of the data can be ensured by using the HTTP protocol for data transmission.

In the embodiment of the invention, the standard physiological artifact correction image and the standard non-physiological artifact correction image are acquired. These images are typically stored in digital form in a computer system, and two different artifact-corrected images are image registered to ensure that they have the same spatial coordinate system. This may be accomplished using image processing software and algorithms that incorporate registered physiological and non-physiological artifact correction images into an artifact fusion correction image using an appropriate image fusion algorithm. This may be pixel level fusion, weight based fusion or deep learning methods, depending on the application requirements and image properties, saving the artifact fusion corrected image for subsequent processing, establishing a cloud computing environment, including configuration and settings of cloud computing nodes. These nodes may be virtual machine instances or physical servers, dividing the artifact fusion corrected image into a plurality of subtasks, depending on the computational requirements and scalability decisions. Each subtask contains processing a portion of the image. Parallel computing frameworks (e.g., apache Spark or Hadoop) are used to distribute these tasks to different cloud computing nodes, with image data being transmitted to each cloud computing node via the HTTP protocol. The method can be realized through HTTP POST or PUT requests, reliable transmission and safety of data are ensured, and after each cloud computing node receives a task, the processing of artifact fusion correction images is executed. The nodes can process a plurality of subtasks simultaneously to improve processing speed, the cloud computing nodes store processed data in proper storage positions and return processing results to the main control node, and the main control node collects the processing results from all the cloud computing nodes and combines the processing results to generate a magnetic resonance image of a user. This image is a high quality image containing multimodal information, artifact correction and fusion has been accomplished.

The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

The foregoing is only a specific embodiment of the invention to enable those skilled in the art to understand or practice the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A magnetic resonance image generation method for eliminating motion artifacts based on a dynamic tracking technology is characterized by comprising the following steps:

step S4: performing image edge detection on the passive motion region data through a second feedback signal to generate a periodic motion edge image set and a periodic motion time stamp; calculating an image difference value of the periodic motion edge image set to obtain a periodic motion difference influence coefficient; performing image subtraction on the periodic motion edge image set according to the periodic motion difference influence coefficient and the periodic motion time stamp to generate a motion edge difference image;

2. The method for generating a magnetic resonance image free from motion artifacts based on the dynamic tracking technique according to claim 1, wherein the step S1 comprises the steps of:

3. The method for generating a magnetic resonance image based on motion artifact avoidance by dynamic tracking technique according to claim 2, wherein step S2 comprises the steps of:

4. A method for generating a magnetic resonance image based on dynamic tracking technique free from motion artifacts as claimed in claim 3, characterized in that step S3 comprises the steps of:

5. The method for generating a magnetic resonance image free from motion artifacts based on the dynamic tracking technique as recited in claim 4, wherein step S35 includes the steps of:

6. The method for generating a magnetic resonance image free of motion artifacts based on the dynamic tracking technique as recited in claim 5, wherein the gating contribution analysis formula in step S353 is as follows:

In (1) the->Expressed as gating contribution,/->Expressed as a continuous range of time, typically from 0 to +.>，/>Represented as first gating signal at time +.>Value of (I) at (I)>Represented as second gating signal at time +.>Value of (I) at (I)>Represented as third gating signal at time +.>Value of (I) at (I)>Represented as a decay of the first gating signalCoefficient of deceleration>Represented as the decay rate coefficient of the second gating signal, ">Attenuation rate coefficient expressed as third gating signal, ">Represented as contribution importance of the first gating signal,/o>Represented as a contribution importance of the second gating signal,/o>The contribution importance weight denoted as third gating signal +.>The outlier adjustment is analyzed for gating contribution.

7. The method for generating a magnetic resonance image free from motion artifacts based on the dynamic tracking technique according to claim 6, wherein the step S4 comprises the steps of:

8. The method for generating a magnetic resonance image free from motion artifacts based on the dynamic tracking technique as recited in claim 7, wherein the image difference influence discrimination formula in step S43 is as follows:

in (1) the->Expressed as an image difference influence coefficient->Time range expressed as periodic motion, +.>Expressed as the number of images in the edge image set, +.>Denoted as +.>The sheet image is in coordinates->Pixel value at +.>Expressed as the number of gradient modes in the edge image set,/->Represented as edge image concentrationNumber of intermediate derivatives, +.>Weight coefficient expressed as time partial derivative, < ->Denoted as +.>The individual images are in coordinates- >Time partial derivative of->Time variable expressed as integral,/->Represented as an image difference influence discrimination abnormality correction amount.

9. The method for generating a magnetic resonance image free from motion artifacts based on the dynamic tracking technique according to claim 7, wherein the step S5 comprises the steps of:

10. The method for generating a magnetic resonance image free from motion artifacts based on the dynamic tracking technique according to claim 9, characterized in that step S6 comprises the steps of: