A method for magnetic resonance imaging pulse and respiration composite gating based on respiratory guidance

Abstract

The present invention discloses a method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance, comprising: acquiring facial video data of a subject; performing skin region recognition to obtain a corresponding region of interest; calculating the pixel mean value to obtain an original IPPG signal; removing head artifacts; determining the respiration period T1 and the pulse period T2, and calculating the ratio N of the respiration period T1 to the pulse period T2; determining the respiration time interval corresponding to each pulse period; determining the delay time d of the IPPG signal peak relative to the R-wave peak of the electrocardiogram signal; calculating the composite gating time interval for magnetic resonance imaging in each pulse period; and generating a composite gating signal for triggering the MRI device to perform magnetic resonance imaging. Through skin region recognition, effective removal of head motion artifacts, and fine synchronous control of the pulse and respiration periods, it reduces image artifacts and significantly improves the quality and efficiency of MRI imaging.

Description

Technical Field

[0001] The present invention relates to the field of magnetic resonance imaging, and more particularly to a method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance. Background Art

[0002] Magnetic resonance imaging (MRI), as an advanced medical imaging technology, plays a crucial role in clinical medical diagnosis. It uses a strong magnetic field and radiofrequency waves to non-invasively image the human body, and can provide high-resolution anatomical structure and functional information, which is of great significance for the early detection, diagnosis, and formulation of treatment plans for diseases.

[0003] However, in existing magnetic resonance imaging technologies, the use of gating methods is crucial for reducing motion artifacts and improving image quality. Traditional gating methods mainly rely on external devices, such as respiration gating belts or electrocardiogram monitors, but these methods often have problems such as complex operation, discomfort for subjects, and poor signal synchronization. In addition, these methods usually can only monitor respiration or heartbeat singly, and it is difficult to achieve precise synchronization between the two.

[0004] In addition, although in recent years some researchers have tried to use image photoplethysmography (IPPG) technology to extract physiological signals from video data, due to the fact that facial video data is easily interfered by various factors such as illumination changes and head movements, the quality of the extracted IPPG signals is unstable, which in turn affects the triggering accuracy and imaging effect of magnetic resonance imaging.

[0005] Therefore, how to design a method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance, effectively remove motion artifacts, and achieve fine control of the respiration and pulse cycles to improve the quality and diagnostic efficiency of MRI images is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0006] In view of this, the present invention provides a method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance, which realizes synchronous control of the pulse and respiration cycles through precise facial video analysis and advanced signal processing algorithms, thereby reducing motion artifacts during magnetic resonance imaging and improving image quality and diagnostic accuracy.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] A method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance, comprising the following steps:

[0009] S1. Obtain the facial video data of the subject;

[0010] S2. Combine the SLICO algorithm and the pre-trained SVM model to identify the skin regions in each frame of the facial video data, and obtain the corresponding regions of interest;

[0011] S3. Calculate the pixel mean of the regions of interest to obtain the original IPPG signals of multiple frames of images;

[0012] S4. Remove the head artifacts from the original IPPG signals to obtain the IPPG signals after removing the head artifacts;

[0013] S5. Based on the IPPG signals after removing the head artifacts, determine the respiratory cycle T1 and the pulse cycle T2, and calculate the ratio N of the respiratory cycle T1 to the pulse cycle T2;

[0014] S6. Based on the ratio N of the respiratory cycle T1 to the pulse cycle T2, determine the respiratory time intervals corresponding to each pulse cycle;

[0015] S7. Combine the electrocardiogram signals of the subject, and determine the delay time d of the IPPG signal peak relative to the R-wave peak of the electrocardiogram signal through the pre-trained regression model;

[0016] S8. Based on the respiratory time intervals corresponding to each pulse cycle and the delay time d, calculate the composite gating time intervals for magnetic resonance imaging in each pulse cycle;

[0017] S9. Generate a composite gating signal for triggering the MRI device to perform magnetic resonance imaging according to the composite gating time intervals.

[0018] Further, the S2 includes:

[0019] S21. Segment each frame of image through the SLICO algorithm to obtain multiple superpixel blocks with similar color and texture features;

[0020] S22. For each superpixel block, use the pre-trained SVM model to classify the skin regions and non-skin regions;

[0021] S23. Based on the classification results, determine the skin regions as the regions of interest.

[0022] Further, the S3 includes:

[0023] S31. Calculate the light intensity value of each pixel in the region of interest at the time series t;

[0024] C(x,y)=I×(ρ s (t)+ρ d (t))+V n

[0025] Among them, C(x, y) represents the light intensity value of the pixel with coordinates (x, y), I represents the light intensity of the light source, and ρ s (t) represents the specular reflection coefficient, and ρ d (t) represents the specular diffuse reflection coefficient, and V n represents the quantization noise of the image sensor;

[0026] S32. Remove the quantization noise V n and the specular diffuse reflection coefficient ρ d (t), and calculate the pixel mean value ;

[0027]

[0028] Among them, represents the light intensity value of the pixel with coordinates (x, y) after removing the quantization noise V n and the specular diffuse reflection coefficient ρ d (t); N represents the number of pixels.

[0029] Furthermore, the S4 includes:

[0030] S41. Calculate the change information of the four corner spatial average positions of the region of interest over time to obtain the corresponding time series data;

[0031] S42. Use the principal component analysis algorithm to extract the eigenvector v i from the time series data to construct the noise subspace; among them, the eigenvector v i represents the main direction of head movement;

[0032] S43. Project the original IPPG signal x onto the noise subspace to obtain the head artifact noise p;

[0033]

[0034] S44. Remove the head artifact noise p to obtain the IPPG signal after removing the head artifact.

[0035] Furthermore, in the S5, determining the respiratory cycle T1 includes:

[0036] S511. Input the IPPG signal after removing the head artifact into a band-pass filter, and remove the high-frequency components above 0.45 Hz and the low-frequency components below 0.13 Hz through the band-pass filter to obtain the filtered respiratory signal;

[0037] S512. Perform time series analysis on the filtered respiratory signal to identify the respiratory wave peaks or troughs;

[0038] S513. Determine the respiratory cycle T1 based on the time interval between the respiratory peaks or troughs.

[0039] Further, in step S5, determining the pulse cycle T2 includes:

[0040] S521. Set the width of the initial sliding window to L, where L is an even number greater than 2.

[0041] S522. Define the IPPG signal after removing the head artifact as , and the sub - interval of discrete signal points covered by the initial sliding window is ;

[0042] S523. Select as the starting reference point to identify peak feature points or trough feature points. The identification of peak feature points or trough feature points includes: if , and

[0043] , then is a peak feature point; if , and

[0044] , then is a trough feature point.

[0045] S524. Move the sliding window to traverse the IPPG signal after removing the head artifact until all peak feature points or trough feature points in the signal are identified.

[0046] S525. Determine the pulse cycle T2 based on the peak feature points or trough feature points.

[0047] Further, step S6 includes:

[0048] S61. Based on the ratio N of the respiratory cycle T1 to the pulse cycle T2, determine that the length of the respiratory time interval corresponding to each pulse cycle T2 is ;

[0049] S62. Determine the starting time point of each pulse cycle T2, and based on the starting time point and the length of the respiratory time interval, divide the exhalation time interval and inhalation time interval corresponding to each pulse cycle.

[0050] Further, step S7 includes:

[0051] S71. Synchronously acquire the electrocardiogram signal of the subject and the IPPG signal after removing the head artifact.

[0052] S72. Input the electrocardiogram signal of the subject and the IPPG signal after removing head artifacts into the pre-trained regression model, and determine the time difference between the peak of the IPPG signal and the vertex of the electrocardiogram R wave as the delay time d. Among them, the pre-trained regression model uses the least squares method to minimize the prediction error.

[0053] Further, the said S8 includes:

[0054] S81. Define the starting time point of the m-th pulse cycle as Tm_start and the ending time point as Tm_end;

[0055] S82. Based on the starting time point Tm_start, determine that the starting time point of the exhalation phase corresponding to the m-th pulse cycle is , and the ending time point of the exhalation phase is ;

[0056] S83. Combine the delay time d to determine that the starting time point of the composite gating time window is Tm_start , and the ending time point is .

[0057] Further, the said S9 includes:

[0058] S91. Based on the composite gating time interval, determine the time point for triggering the MRI device to perform magnetic resonance imaging;

[0059] S92. Convert the said time point into a signal format matching the MRI device through a preset coding rule and output it to the MRI device for magnetic resonance imaging.

[0060] It can be seen from the above technical solutions that compared with the prior art, the technical solutions of the present invention have the following beneficial effects:

[0061] 1. By combining the SLICO algorithm with the pre-trained SVM model for skin region recognition, this method can more accurately locate the region of interest on the face, thereby improving the quality of the original IPPG signal. And by combining superpixel block segmentation and machine learning classification, the skin region can be stably and reliably extracted in different environments, providing more accurate basic data for subsequent signal processing.

[0062] 2. Principal component analysis (PCA) is introduced to construct a noise subspace, and based on this, head motion artifacts are removed, improving the purity of the IPPG signal. By calculating the time series change information of the angular position of the region of interest and using PCA to extract feature vectors, the main direction of head motion can be accurately captured, and then these artifacts can be effectively separated and removed from the original signal, ensuring the authenticity and accuracy of the obtained respiration and pulse cycles.

[0063] 3. By determining the breathing time interval corresponding to each pulse cycle and the delay time d between the IPPG peak and the R-wave peak of the electrocardiogram signal, the composite gating time interval most suitable for MRI scanning can be accurately calculated. This enables the MRI device to trigger imaging at the optimal timing, reducing image artifacts caused by breathing and heartbeat, improving image quality and diagnostic efficiency, while reducing unnecessary radiation exposure and scanning time, bringing great convenience to clinical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0065] Figure 1 It is a flowchart of a method for pulse and respiration composite gating in magnetic resonance imaging based on respiration guidance provided by an embodiment of the present invention;

[0066] Figure 2 It is a schematic diagram of the process of obtaining facial video data of a subject provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0067] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0068] As Figure 1 shown, this embodiment provides a method for pulse and respiration composite gating in magnetic resonance imaging based on respiration guidance, including the following steps:

[0069] S1. Obtain facial video data of the subject;

[0070] S2. Combine the SLICO algorithm and the pre-trained SVM model to identify the skin region in each frame of the facial video data and obtain the corresponding region of interest;

[0071] S3. Calculate the pixel mean of the region of interest to obtain the original IPPG signal of multiple frames of images;

[0072] S4. Remove the head artifacts from the original IPPG signal to obtain the IPPG signal after removing the head artifacts;

[0073] S5. Based on the IPPG signal after removing the head artifact, determine the respiratory cycle T1 and the pulse cycle T2, and calculate the ratio N of the respiratory cycle T1 to the pulse cycle T2;

[0074] S6. Based on the ratio N of the respiratory cycle T1 to the pulse cycle T2, determine the respiratory time interval corresponding to each pulse cycle;

[0075] S7. Combine the electrocardiogram (ECG) signal of the subject, and determine the delay time d of the peak of the IPPG signal relative to the R-wave peak of the ECG signal through a pre-trained regression model;

[0076] S8. Based on the respiratory time interval corresponding to each pulse cycle and the delay time d, calculate the composite gating time interval for magnetic resonance imaging in each pulse cycle;

[0077] S9. Generate a composite gating signal for triggering the MRI device to perform magnetic resonance imaging according to the composite gating time interval.

[0078] This method not only achieves precise recognition of the skin area through advanced algorithms, but also optimizes the signal processing flow, effectively reducing noise interference, thereby ensuring higher accuracy of the extracted physiological signals. In addition, this method combines the ECG signal and the IPPG signal, and through precise time synchronization and calculation, greatly enhances the stability during the imaging process. Under the combined action of these comprehensive improvement measures, this method is more reliable in the application of magnetic resonance imaging, providing more accurate medical image support for users.

[0079] The following further elaborates on each step in the above technical solution:

[0080] In step S1 of this embodiment, obtain the facial video data of the subject;

[0081] As Figure 2 shown, in the specific process of obtaining the facial video data of the subject, first use the visible light source 1 to irradiate the facial area of the subject, and reflect the light to the acquisition camera 3 through the reflector 2. The acquisition camera 3 is located outside the MRI device and can capture the facial image reflected by the reflector 2. This setting ensures that the facial video data of the subject is recorded in real time without affecting the MRI imaging quality, providing basic information for subsequent skin area recognition and signal processing.

[0082] In step S2 of this embodiment, combine the SLICO algorithm and the pre-trained SVM model to identify the skin area in each frame of the facial video data and obtain the corresponding region of interest; including:

[0083] S21. Segment each frame of the image through the SLICO algorithm to obtain multiple superpixel blocks with similar color and texture features;

[0084] S22. For each superpixel block, use the pre-trained SVM model to classify the skin area and non-skin area;

[0085] S23. Based on the classification results, determine the skin area as the region of interest.

[0086] In step S3 of this embodiment, calculate the pixel mean value of the region of interest to obtain the original IPPG signal of multiple frames of images, including:

[0087] S31. Calculate the light intensity value of each pixel in the region of interest at time series t;

[0088] C(x,y)=I×(ρ s (t)+ρ d (t))+V n

[0089] where C(x,y) represents the light intensity value of the pixel with coordinates (x,y), I represents the light intensity of the light source, ρ s (t) represents the specular reflection coefficient, ρ d (t) represents the specular diffuse reflection coefficient, and V n represents the quantization noise of the image sensor;

[0090] S32. Remove the quantization noise V n and the specular diffuse reflection coefficient ρ d (t), and calculate the pixel mean value ;

[0091]

[0092] where represents the light intensity value of the pixel with coordinates (x,y) after removing the quantization noise V n and the specular diffuse reflection coefficient ρ d (t); N represents the number of pixels. In step S4 of this embodiment, remove the head artifact from the original IPPG signal to obtain the IPPG signal after removing the head artifact, including:

[0093] S41. Calculate the change information of the spatial average positions of the four corners of the region of interest over time to obtain the corresponding time series data;

[0094] S42. Use the principal component analysis algorithm to extract the eigenvector v i from the time series data and construct the noise subspace; where the eigenvector v i represents the main direction of head movement;

[0095] S43. Project the original IPPG signal \(x\) onto the noise subspace to obtain the head artifact noise \(p\).

[0096]

[0097] S44. Remove the head artifact noise \(p\) to obtain the IPPG signal after removing the head artifact.

[0098] In this embodiment S5, based on the IPPG signal after removing the head artifact, determine the respiratory cycle \(T1\) and the pulse cycle \(T2\), and calculate the ratio \(N\) of the respiratory cycle \(T1\) to the pulse cycle \(T2\).

[0099] Specifically, determining the respiratory cycle \(T1\) includes:

[0100] S511. Input the IPPG signal after removing the head artifact into a band - pass filter, and remove the high - frequency components above 0.45 Hz and the low - frequency components below 0.13 Hz through the band - pass filter to obtain the filtered respiratory signal.

[0101] S512. Perform time - series analysis on the filtered respiratory signal to identify respiratory wave peaks or valleys.

[0102] S513. Determine the respiratory cycle \(T1\) based on the time interval between respiratory wave peaks or valleys.

[0103] Further, determining the pulse cycle \(T2\) includes:

[0104] S521. Set the width of the initial sliding window to \(L\); where \(L\) is an even number greater than 2.

[0105] S522. Define the IPPG signal after removing the head artifact as , and the discrete signal point sub - interval covered by the initial sliding window is ;

[0106] S523. Select as the starting reference point to perform peak feature point or valley feature point recognition; the peak feature point or valley feature point recognition includes: if , and

[0107] , then is a peak feature point; if , and

[0108] , then is a valley feature point.

[0109] S524. Move the sliding window and traverse the IPPG signal after removing the head artifact until all peak feature points or valley feature points in the signal are identified;

[0110] S525. Determine the pulse period T2 based on the peak feature points or valley feature points.

[0111] In this embodiment S6, based on the ratio N of the respiratory cycle T1 to the pulse cycle T2, determine the respiratory time interval corresponding to each pulse cycle; including:

[0112] S61. Based on the ratio N of the respiratory cycle T1 to the pulse cycle T2, determine that the length of the respiratory time interval corresponding to each pulse cycle T2 is ;

[0113] S62. Determine the starting time point of each pulse cycle T2, and based on the starting time point and the length of the respiratory time interval, divide the exhalation time interval and the inhalation time interval corresponding to each pulse cycle.

[0114] Furthermore, based on the exhalation time interval and the inhalation time interval corresponding to each pulse cycle, respiratory guidance can be performed to guide the subject to breathe according to a specific breathing pattern. For example, when N = 3, it means that one respiratory cycle is three times as long as the pulse cycle. Therefore, within each pulse cycle T2, time is respectively allocated to the exhalation and inhalation processes.

[0115] In actual operation, whenever a feature point (such as a peak or a valley) in the pulse waveform is detected, it represents the start of a new pulse cycle. At this time, the respiratory guidance program will be started. Taking the indicator light as an example, when the mth pulse feature point is detected, the indicator light shows "green", prompting the subject to start inhaling. This process lasts time, and then the indicator light becomes "yellow", indicating the start of the exhalation phase. The subject should exhale according to the indication, also lasting time. After completing this cycle, if the scan is not over, the indicator light will turn back to "green" again, guiding the subject to repeat the above process until the entire magnetic resonance imaging process ends.

[0116] It can not only ensure that the breathing rhythm of the subject is consistent with the pulse cycle, but also effectively reduce the image artifact problem caused by the incoordination of breathing and cardiac movement, improving the quality and efficiency of magnetic resonance imaging.

[0117] In this embodiment S7, in combination with the subject's electrocardiogram signal, determine the delay time d of the IPPG signal peak relative to the R-wave vertex of the electrocardiogram signal through a pre-trained regression model; including:

[0118] S71. Synchronously acquire the electrocardiogram signal of the subject and the IPPG signal after removing head artifacts;

[0119] S72. Input the electrocardiogram signal of the subject and the IPPG signal after removing head artifacts into the pre-trained regression model, and determine the time difference between the peak of the IPPG signal and the R-wave peak of the electrocardiogram as the delay time d; wherein, the pre-trained regression model uses the least squares method to minimize the prediction error.

[0120] In this embodiment S8, based on the respiratory time interval corresponding to each pulse cycle and the delay time d, calculate the composite gating time interval for magnetic resonance imaging in each pulse cycle; including:

[0121] S81. Define the start time point of the m-th pulse cycle as Tm_start and the end time point as Tm_end;

[0122] S82. Based on the start time point Tm_start, determine that the start time point of the exhalation phase corresponding to the m-th pulse cycle is , and the end time point of the exhalation phase is ;

[0123] S83. Combining the delay time d, determine that the start time point of the composite gating time window is Tm_start , and the end time point is .

[0124] In this step, the start time point of the composite gating time window is set to Tm_start , which means starting the magnetic resonance imaging process at the delay time d after the start of the pulse cycle to ensure that the heart is in a relatively stable period. Similarly, the end time point of the composite gating time window is also adjusted accordingly to , so as to avoid the period when the heart movement is the most intense and reduce the influence of motion artifacts on the image quality.

[0125] In this embodiment S9, generate a composite gating signal for triggering the MRI device to perform magnetic resonance imaging according to the composite gating time interval; including:

[0126] S91. Based on the composite gating time interval, determine the time point for triggering the MRI device to perform magnetic resonance imaging;

[0127] S92. Convert the time point into a signal format matching the MRI device through a preset coding rule and output it to the MRI device for magnetic resonance imaging.

[0128] This embodiment provides a method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance. By acquiring the facial video data of the subject, combining the SLICO algorithm with the pre-trained SVM model to accurately identify the skin area, then calculating the original IPPG signal and removing the head artifacts, determining the respiration and pulse periods and their ratio based on the optimization algorithm, combining the electrocardiogram signal to determine the delay time between the peak of the IPPG signal and the vertex of the electrocardiogram R wave, and finally calculating the composite gating time interval and generating a composite gating signal for triggering the MRI device. This method can accurately monitor the pulse and respiration simultaneously, effectively reduce the noise interference, improve the accuracy of the physiological signal, enhance the imaging stability by combining the electrocardiogram signal, and provide more reliable and accurate medical image support for magnetic resonance imaging.

[0129] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the system disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0130] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance, characterized in that It includes the following steps: S1. Obtain the facial video data of the subject; S2. Combine the SLICO algorithm and the pre-trained SVM model to identify the skin region for each frame image in the facial video data, and obtain the corresponding region of interest; S3. Calculate the pixel mean value of the region of interest to obtain the original IPPG signal of multiple frames of images; S4. Remove the head artifact from the original IPPG signal to obtain the IPPG signal after removing the head artifact; S5. Based on the IPPG signal after removing the head artifact, determine the respiratory cycle T1 and the pulse cycle T2, and calculate the ratio N of the respiratory cycle T1 to the pulse cycle T2; wherein, determining the pulse cycle T2 includes: S521. Set the width of the initial sliding window to L; wherein, L is an even number greater than 2; Define the IPPG signal after removing the head artifact as , and the sub-interval of discrete signal points covered by the initial sliding window is ; S523. Select as the starting reference point to identify peak feature points or valley feature points; the identification of the peak feature points or valley feature points includes: if , and , then is the peak feature point; if , and , then is a valley feature point; S524. Move the sliding window to traverse the IPPG signal after removing the head artifact until all peak feature points or valley feature points in the signal are identified; S525. Determine the pulse cycle T2 based on the peak feature points or valley feature points; S6. Based on the ratio N of the respiratory cycle T1 to the pulse cycle T2, determine the respiratory time interval corresponding to each pulse cycle; and combine the color change of the indicator light to prompt the subject to complete the inhalation and exhalation actions within the corresponding interval; S7. Combine the electrocardiogram signal of the subject, and determine the delay time d of the IPPG signal peak relative to the R wave peak of the electrocardiogram signal through the pre-trained regression model; S8. Based on the respiratory time interval corresponding to each pulse cycle and the delay time d, calculate the composite gating time interval for magnetic resonance imaging in each pulse cycle; S9. Generate a composite gating signal for triggering the MRI device to perform magnetic resonance imaging according to the composite gating time interval.

2. A method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance according to claim 1, characterized in that, The S2 includes: S21. Segment each frame of image through the SLICO algorithm to obtain multiple superpixel blocks with similar color and texture features; S22. For each superpixel block, use the pre-trained SVM model to classify the skin region and the non-skin region; S23. Based on the classification result, determine the skin region as the region of interest.

3. A method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance according to claim 1, characterized in that The S3 includes: S31. Calculate the light intensity value of each pixel in the region of interest at the time series t; C(x,y)=I×(ρ s (t)+ρ d (t))+V n Among them, C(x, y) represents the light intensity value of the pixel with coordinates (x, y), I represents the light intensity of the light source, and ρ s (t) represents the specular reflection coefficient, and ρ d (t) represents the specular diffuse reflection coefficient, and V n represents the quantization noise of the image sensor; S32. Perform quantization noise V n and specular diffuse reflection coefficient ρ d (t) is removed, and the pixel mean value is calculated ; Among them, represents the removal of quantization noise V n and the specular diffuse reflection coefficient ρ d (t), the light intensity value of the pixel with coordinates (x, y); N represents the number of pixels.

4. A method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance according to claim 1, characterized in that The S4 includes: S41. Calculate the change information of the spatial average position of the four corners of the region of interest over time to obtain the corresponding time series data; S42. Extract the eigenvector v from the time series data using the principal component analysis algorithm i , and construct a noise subspace; where the eigenvector v i represents the main direction of head movement; S43. Project the original IPPG signal x onto the noise subspace to obtain the head artifact noise p; ； S44. Remove the head artifact noise p to obtain the IPPG signal after removing the head artifact.

5. A method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance according to claim 1, characterized in that, In the S5, determining the respiratory cycle T1 includes: S511. Input the IPPG signal after removing the head artifact into a band-pass filter, and remove the high-frequency components above 0.45 Hz and the low-frequency components below 0.13 Hz through the band-pass filter to obtain the filtered respiratory signal; S512. Perform time series analysis on the filtered respiratory signal to identify respiratory peaks or valleys; S513. Determine the respiratory cycle T1 based on the time interval between the respiratory peaks or troughs.

6. A method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance according to claim 1, characterized in that, The said S6 includes: S61. Determine that the length of the respiratory time interval corresponding to each pulse cycle T2 is based on the ratio N of the respiratory cycle T1 to the pulse cycle T2 as ; S62. Determine the starting time point of each pulse cycle T2, and based on the starting time point and the length of the respiratory time interval, divide the exhalation time interval and inhalation time interval corresponding to each pulse cycle.

7. A method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance according to claim 1, characterized in that The said S7 includes: S71. Synchronously acquire the electrocardiogram signal of the subject and the IPPG signal after removing the head artifact; S72. Input the electrocardiogram signal of the subject and the IPPG signal after removing the head artifact into the pre-trained regression model, and determine the time difference between the peak of the IPPG signal and the vertex of the electrocardiogram R wave as the delay time d; wherein, the pre-trained regression model uses the least squares method to minimize the prediction error.

8. A method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance according to claim 1, wherein The said S8 includes: S81. Define the starting time point of the m-th pulse cycle as Tm_start and the ending time point as Tm_end; Based on the starting time point Tm_start, determine that the starting time point of the exhalation phase corresponding to the m-th pulse cycle is , and the ending time point of the exhalation phase is ; S83. Determine that the start time point of the composite gating time window is Tm_start in combination with the delay time d , and the end time point is .

9. A method for magnetic resonance imaging pulse and respiration composite gating based on respiration guidance according to claim 1, wherein The said S9 includes: S91. Based on the composite gating time interval, determine the time point for triggering the MRI device to perform magnetic resonance imaging; S92. Convert the said time point into a signal format matching the MRI device through a preset coding rule, and output it to the MRI device for magnetic resonance imaging.

Images