Cardiopulmonary joint movement image reconstruction method and system and storage medium

Abstract

The invention belongs to the technical field of medical imaging, and particularly relates to a cardiopulmonary joint movement image reconstruction method and system and a storage medium. The method comprises the following steps: acquiring data; carrying out period detection and phase normalization on the electrocardiosignals and the respiratory signals; performing multi-dimensional binning and grouping on the projection data according to heart and respiration double-gating phases; and performing three-dimensional image reconstruction on the projection data of each group of combined time phases. The invention further provides a system applying the method. The method is particularly suitable for patients with unequal-period heartbeat and breathing movement coupling states such as arrhythmia patients, high-quality image data and theoretical basis can be provided for work such as clinical radiotherapy, movement management and scientific research, and the method has good application prospects.

Description

Technical Field

[0001] This invention belongs to the field of medical imaging technology, specifically relating to a method, system, and storage medium for cardiopulmonary combined motion imaging reconstruction. Background Technology

[0002] In clinical practice, the periodic and non-periodic motions of the heart and lungs have a significant impact on the imaging diagnosis, radiotherapy planning, and treatment implementation of thoracic tumors. Conventional four-dimensional computed tomography (4DCT) can effectively reconstruct changes in thoracic anatomy at different respiratory phases by acquiring external respiratory signals, aiding in tumor target delineation and dose calculation. However, the spontaneous beating of the heart, especially in patients with arrhythmias, exhibits unstable and highly variable spatiotemporal patterns. Existing single-gated imaging methods (such as respiratory gating or ECG gating alone) struggle to accurately distinguish and anatomically reconstruct the combined motion of the heart and lungs.

[0003] This technology has limitations in various clinical scenarios, leading to blurred images of the heart and its substructures, mixed cardiopulmonary motion signals, and misregistration of key phases. These limitations affect target area and organ delineation, motion feature analysis, and dose assessment accuracy. In actual radiotherapy, the inability to accurately separate and spatiotemporally quantify the combined motion of the heart and lungs can easily result in misestimation of the actual motion amplitude of the treatment target or the dose distribution in normal tissues, thus impacting treatment planning and accurate efficacy assessment. Particularly in applications such as gated therapy or target-tracking therapy, real-time capture and modeling of the complex coupled motion of the heart and lungs places higher demands on the dynamic adaptation of target contours, optimization of dose delivery timing, and precision of field control. Existing imaging and analysis methods are still insufficient in the synchronous identification and localization of irregular heart rhythms and multi-source motion signals, making it difficult to meet the needs of high-precision clinical radiotherapy.

[0004] Therefore, there is an urgent need in this field to develop novel imaging methods capable of simultaneously acquiring electrocardiogram and respiratory signals and realizing cardiopulmonary motion image reconstruction. This will not only help improve the accuracy of radiotherapy target delineation and organ at risk delineation in patients with arrhythmias, as well as the reliability of dosimetric assessment, but also provide a solid imaging data foundation and theoretical support for the application and optimization of precision radiotherapy technologies such as clinical gated radiotherapy, motion management, and real-time tracking treatment. Summary of the Invention

[0005] To address the problems of existing technologies, this invention provides a method, system, and storage medium for cardiopulmonary combined motion imaging reconstruction.

[0006] A method for cardiopulmonary motion imaging reconstruction includes the following steps: Step 1: Simultaneously collect electrocardiogram (ECG) and respiratory signals, and record the collection time; Step 2: The ECG signal and respiratory signal are synchronously subjected to periodic detection and phase normalization processing to obtain the normalized cardiac phase and normalized respiratory phase corresponding to each acquisition time. Then, the normalized cardiac phase and respiratory phase are jointly binned to characterize the cardiac-respiratory dual-phase state at each acquisition time. Step 3, according to the collection time Binning can be performed either by sorting bins or by using a weighted allocation strategy. i Bins representing normalized cardiac phase, j Bins representing normalized respiratory phase; Step 4: Incorporate the projection data into each bin according to the binning results; Step 5: Perform image reconstruction on the projection data in each sub-box to obtain multiple sets of three-dimensional volume data of cardiopulmonary combined exercise, and integrate the three-dimensional volume data to obtain multi-temporal images of cardiopulmonary combined exercise.

[0007] Preferably, the projection data is CT scan data or CBCT data; And / or, the projection data is real-time acquired data (i.e., the method of the present invention enables forward-looking real-time acquisition of projection data) or historical data used for retrospective image reconstruction (i.e., the method of the present invention also enables retrospective reconstruction of historical data). And / or, the method for acquiring the projection data is selected from at least one of the following: All projection data are collected in real time at each acquisition moment, and then the collected projection data is incorporated into each bin according to the binning results. The system presets the target bins for which projection data needs to be collected, monitors the normalized cardiac phase and / or normalized respiratory phase in real time, and collects projection data only when the collection time belongs to the target bin. The collected projection data is then incorporated into the target bin.

[0008] Preferably, in step 2, the step of obtaining the normalized cardiac phase corresponding to each acquisition time includes: detecting the heartbeat start point of the acquired electrocardiogram signal and calculating the length of each heartbeat cycle; for any acquisition time, determining the heartbeat cycle in which it is located, and linearly normalizing the position of the acquisition time in the cycle to the interval [0,1) according to the start time and duration of the cycle, so as to obtain the normalized cardiac phase, so as to reflect the periodic motion state of the heart at different times; The steps to obtain the normalized respiratory phase corresponding to each acquisition time include: detecting the respiratory initiation point of the acquired respiratory signal and calculating the length of each respiratory cycle; for any acquisition time, determining the respiratory cycle in which it is located, and linearly normalizing the position of the acquisition time within the cycle to the interval [0,1) according to the start time and duration of the cycle, so as to obtain the normalized respiratory phase, which reflects the relative temporal changes of respiratory motion.

[0009] Preferably, in step 2, for patients with arrhythmia, abnormal cardiac cycles are further identified based on the variability of the RR interval, and the projection data of the abnormal cycles are removed. At the same time, the number and width of the cardiac phase bins are adaptively adjusted according to the degree of variability of the RR interval. The adjustment method includes the following steps: RR interval sequences were obtained by QRS wave detection: in, For the actual RR interval, To indicate the first The timing of the R-wave peak occurring with each heartbeat. k Here, K is the heartbeat index number, and K is the total number of R waves detected. Abnormal cardiac cycles are identified based on the coefficient of variation or a preset threshold, and these abnormal cycles are then removed. The coefficient of variation is: in, Indicates the first k The actual RR interval per heartbeat cycle, To represent the standard deviation of all RR intervals, To represent the mean of all RR intervals; After calculating the coefficient of variation of the RR interval, it is further determined whether each period is an abnormal period based on the relative deviation of each period; the period deviation rate is defined as follows: when > If a period is identified as an abnormal RR period, the corresponding projection data for that period is directly removed. The preset threshold; when Less than the preset threshold At that time, the number of boxes is taken as the initial value. ;when Greater than At that time, the number of bins is adaptively adjusted according to the degree of variation, and the adjustment formula is as follows: in, N To adjust the number of boxes, Let be the initial number of boxes, where This is an empirical adjustment coefficient used to control the adjustment range of the number of boxes.

[0010] Preferably, in abnormal cardiac cycles, the normalized cardiac phase is calculated using the following formula: in, For normalized cardiac phase, t The time point for acquiring the projection signal. For the actual RR interval, To indicate the first The timing of the R-wave peak occurring with each heartbeat. The variability correction weight is defined as: in, The mean RR interval is the global interval.

[0011] Preferably, when the signal contains noise, phase boundary discontinuities, or uneven sampling timing, a joint phase pairing strategy is adopted to normalize the cardiac phase. With normalized respiratory phase Combined into a two-dimensional joint phase vector: And based on the center phase of each cardiac chamber and respiratory chamber Based on this, the projected data is allocated to the optimal bins using distance minimization or weighted allocation methods; in, for Center phase coordinates of the sub-box For the center phase of all candidate bins, To assign smoothing coefficients, which are used to control the weighting range, , These represent all possible indices for the cardiac and respiratory phase bins, respectively. Weight Reflection Moment t The degree of matching between the projected data and the phase of the target bin center, when the actual phase With a certain distribution center The closer the bins are, the larger their corresponding weights are; the sum of all bin weights after normalization is 1. Based on the above weight definition, the projected data within each bin is weighted and aggregated to obtain the first... The projected set of bins.

[0012] in, Indicates time t The collected raw projection data, To be allocated to the first Weighted projection results of binning.

[0013] Preferably, in step 4, for bins with insufficient projection data, image completion is performed using deep learning-based interpolation or artificial intelligence models to ensure the continuity and integrity of full-time cardiopulmonary exercise data. The criterion for insufficient projection data is: in any cardiac-respiratory phase sub-bin... If the number of projections collected is less than 70% of the number of projections required for complete 3D reconstruction, or if the angle coverage is less than 75% of the full scanning angle, it is considered that the projection data is insufficient.

[0014] Preferably, the algorithm of the artificial intelligence model is selected from convolutional neural networks or Transformer networks; The input data for the artificial intelligence model includes adjacent temporal data and historical signal features, while also incorporating normalized codes for cardiac and respiratory phases as conditional inputs. These normalized codes are: in, For the heart phase, This is the respiratory phase; The artificial intelligence model incorporates kinematic constraints during training and inference, including displacement field smoothness, anatomical structure consistency, and Jacobian positive constraints. The artificial intelligence model incorporates projection consistency, adjacent temporal continuity, and kinematic regularization terms into the loss function.

[0015] The present invention also provides a cardiopulmonary motion imaging reconstruction system for implementing the above-mentioned cardiopulmonary motion imaging reconstruction method, including a signal acquisition module and an intelligent box-type system; The signal acquisition module is configured to acquire projection data, electrocardiogram (ECG) signals, and respiratory signals, and record the acquisition time of each projection data, ECG signal, and respiratory signal. The intelligent container sorting system includes: The binning module is configured to synchronously perform periodic detection and phase normalization processing on the electrocardiogram signal and respiratory signal to obtain the normalized cardiac phase and normalized respiratory phase corresponding to each acquisition time; then, the normalized cardiac phase and respiratory phase are jointly binned to characterize the cardiac-respiratory dual-phase state at each acquisition time. The data grouping module groups the data collection times according to... Binning can be performed either by sorting bins or by using a weighted allocation strategy. i Bins representing normalized cardiac phase, j Binning is used to represent the normalized respiratory phase; the projected data is then incorporated into each bin according to the binning results. The three-dimensional reconstruction module is configured to perform image reconstruction on the projection data in each sub-box to obtain multiple sets of three-dimensional volume data of cardiopulmonary combined exercise, and integrate the three-dimensional volume data to obtain multi-temporal images of cardiopulmonary combined exercise.

[0016] The present invention also provides a computer-readable storage medium having stored thereon a computer program for implementing the above-described cardiopulmonary motion imaging reconstruction method.

[0017] This invention provides a method for cardiopulmonary motion imaging reconstruction, capable of accurately reconstructing cardiopulmonary motion images of patients with arrhythmias under non-isocyclic heartbeat and respiratory motion coupling conditions. This provides high-quality image data and theoretical basis for automatic target delineation, dose calculation, motion uncertainty analysis, and individualized gated radiotherapy planning in complex motion scenarios. It reduces dose assessment errors caused by motion uncertainty, improves radiotherapy efficacy, and lowers the risk of complications. Furthermore, this method supports integration with various clinical radiotherapy workflows, motion management technologies, and intelligent auxiliary analysis platforms, facilitating the application of precision radiotherapy and individualized treatment in complex cases such as arrhythmias, and further enhancing the radiotherapy benefits for cancer patients.

[0018] Obviously, based on the above description of the present invention, and according to common technical knowledge and conventional methods in the field, various other modifications, substitutions or alterations can be made without departing from the basic technical concept of the present invention.

[0019] The following detailed embodiments further illustrate the above-described content of the present invention. However, this should not be construed as limiting the scope of the present invention to the following examples. All technologies implemented based on the above-described content of the present invention fall within the scope of the present invention. Attached Figure Description

[0020] Figure 1 The figures are application scenario diagrams of Embodiments 2 to 4 of the present invention. In the figures, PQRST refer to the P, Q, R, S, T, and U waves in the electrocardiogram, respectively; M represents dividing the entire respiratory cycle into M respiratory compartments; and N represents dividing the complete cardiac cycle into N cardiac compartments. Represents the i-th heart compartment. Represents the j-th respiration chamber. Representing time t, Projection data at the angle and position s. Detailed Implementation

[0021] It should be noted that the algorithms for data acquisition, transmission, storage and processing steps not specifically described in the embodiments, as well as the hardware structures and circuit connections not specifically described, can all be implemented using content already disclosed in the prior art.

[0022] Example 1: Cardiopulmonary Combined Motion Imaging Reconstruction Method and System This embodiment provides a method and system for cardiopulmonary motion imaging reconstruction.

[0023] The system includes a signal acquisition module and an intelligent sorting system.

[0024] The signal acquisition module is configured to acquire projection data, electrocardiogram (ECG) signals, and respiratory signals, and record the acquisition time of each projection data, ECG signal, and respiratory signal. In the technical solution of the present invention, the image data can be selected to be applicable to CT scan data or CBCT data. The data processing steps of the intelligent binning are the same. This embodiment uses CT scan data as an example for explanation.

[0025] The intelligent container sorting system includes: The binning module is configured to synchronously perform periodic detection and phase normalization processing on the electrocardiogram signal and respiratory signal to obtain the normalized cardiac phase and normalized respiratory phase corresponding to each acquisition time; then, the normalized cardiac phase and respiratory phase are jointly binned to characterize the cardiac-respiratory dual-phase state at each acquisition time. The data grouping module groups the data collection times according to... Binning can be performed either by sorting bins or by using a weighted allocation strategy. i Bins representing normalized cardiac phase, j Binning is used to represent the normalized respiratory phase; the projected data is then incorporated into each bin according to the binning results. The three-dimensional reconstruction module is configured to perform image reconstruction on the projection data in each sub-box to obtain multiple sets of three-dimensional volume data of cardiopulmonary combined exercise, and integrate the three-dimensional volume data to obtain multi-temporal images of cardiopulmonary combined exercise.

[0026] This system enables three-dimensional reconstruction of cardiopulmonary combined motion using CT scan projection data, electrocardiogram signals, and respiratory signals as input data. The resulting three-dimensional dataset can be output in the DICOM standard format, seamlessly integrating with the clinical radiotherapy planning system (TPS) and supporting various applications such as gated irradiation, motion management, dynamic target delineation, dose accumulation, and kinematic analysis.

[0027] The method for performing three-dimensional reconstruction using this system includes the following steps: Step 1, throughout the entire acquisition period of the CT scan In the middle, synchronous acquisition at angle ,time and detector position Projection data values ​​below ECG signal respiratory signals Record the acquisition time of each projection data.

[0028] The acquisition of the electrocardiogram and respiratory signals uses a medically certified synchronous gating system, which can ensure the temporal consistency and traceability accuracy of the signals with the original CT projection data, and is suitable for clinical radiotherapy positioning and medical image acquisition.

[0029] Step 2: The ECG signal and respiratory signal are synchronously subjected to periodic detection and phase normalization processing to obtain the normalized cardiac phase and normalized respiratory phase corresponding to each acquisition time. Then, the normalized cardiac phase and respiratory phase are jointly binned to characterize the cardiac-respiratory dual-phase state at each acquisition time.

[0030] The cycle detection and phase normalization employ clinically common methods such as QRS wave detection (for electrocardiogram signals) and respiratory signal extreme value detection (for respiratory signals) to normalize each exercise cycle, ensuring that the variability of exercise cycles in patients with arrhythmia is accurately reflected.

[0031] The heartbeat start point is detected for the collected electrocardiogram signal, and the length of each heartbeat cycle is calculated. For any acquisition time, the heartbeat cycle in which it is located is determined, and the position of the acquisition time in the cycle is linearly normalized to the interval [0,1) according to the start time and duration of the cycle, so as to obtain the normalized cardiac phase, which reflects the periodic motion state of the heart at different times.

[0032] Cardiac motion normalized phase: in The starting time of the R wave of the nth heartbeat is [time value missing], and the nth heartbeat cycle is [time value missing]. , .

[0033] The steps to obtain the normalized respiratory phase corresponding to each acquisition time include: detecting the respiratory initiation point of the acquired respiratory signal and calculating the length of each respiratory cycle; for any acquisition time, determining the respiratory cycle in which it is located, and linearly normalizing the position of the acquisition time within the cycle to the interval [0,1) according to the start time and duration of the cycle, so as to obtain the normalized respiratory phase, which reflects the relative temporal changes of respiratory motion.

[0034] Normalized phase of respiratory movement: in The start time of the m-th respiratory cycle is [time value missing]. The m-th respiratory cycle is [time value missing]. , .

[0035] This embodiment is particularly suitable for patients with arrhythmias and other conditions exhibiting non-isocyclic heartbeats and respiration. For patients with arrhythmias, abnormal cardiac cycles are further identified based on the variability of the RR interval (the time interval between two consecutive R wave peaks on an electrocardiogram), and the projected data of abnormal cycles are removed. Simultaneously, the number and width of the cardiac phase bins are adaptively adjusted according to the degree of RR interval variability to ensure the stability and accuracy of phase division. The adjustment method includes the following steps: RR interval sequences were obtained by QRS wave detection: in, For the actual RR interval, To indicate the first The timing of the R-wave peak occurring with each heartbeat. k Here, K is the heartbeat index number, and K is the total number of R waves detected. Abnormal cardiac cycles are identified based on the coefficient of variation or a preset threshold, and these abnormal cycles are removed to avoid interference from unstable cardiac cycles with normalized phase and binning accuracy. The coefficient of variation is: in, Indicates the first k The actual RR interval per heartbeat cycle, To represent the standard deviation of all RR intervals, To represent the mean of all RR intervals; After calculating the coefficient of variation of the RR interval, it is further determined whether each period is an abnormal period based on the relative deviation of each period. The period deviation rate is defined as follows: when > If a period is identified as an abnormal RR period, the corresponding projection data for that period is directly removed. This is a preset threshold. As a preferred method, =0.2~0.25.

[0036] when Less than the preset threshold At that time, the number of boxes is taken as the initial value. ;when Greater than In this case, adaptively adjusting the number of bins according to the degree of variation is a preferred approach. =0.15. The adjustment formula is: number of boxes. N Based on the coefficient of variation of patients with arrhythmia Adaptive adjustment: in, Let be the initial number of boxes, where This is an empirical adjustment coefficient used to control the adjustment range of the number of boxes.

[0037] During abnormal cardiac cycles, the normalized cardiac phase is calculated using the following formula: in, For normalized cardiac phase, t The time point for acquiring the projection signal. For the actual RR interval, To indicate the first The timing of the R-wave peak occurring with each heartbeat. The variability correction weight is defined as: in, To determine the global average RR interval, a variability weighting coefficient is introduced. When significant fluctuations occur in the RR interval, the heartbeat cycles of different lengths are dynamically stretched or compressed to ensure a continuous transition of the normalized phase on the time axis, thereby eliminating the interference of abnormal RR intervals on phase mapping.

[0038] Step 3, according to the collection time The boxes are then sorted, among which... i Bins representing normalized cardiac phase, j The bins representing the normalized respiratory phase and the total number of bins representing the normalized cardiac phase are: N The total number of bins for normalized respiratory phase is M ,therefore, i=1,…,N,j=1,…,M The projected data at each acquisition time is assigned to the corresponding... Binning is an operation that aggregates all projected data located in the cardiac and respiratory phases within each bin.

[0039] In multidimensional binning, the division of cardiac and respiratory phases can be dynamically adjusted based on the patient's actual heart rate and respiratory rate using clinically common phase-segmentation algorithms such as QRS wave detection (for ECG signals) and respiratory extreme value detection (for respiratory signals). For example, the cardiac phase can be divided into 8–20 bins, and the respiratory phase into 5–20 bins, to achieve the optimal balance between spatial and temporal resolution. The specific binning formula is as follows: Let N be the number of cardiac cycle bins and M be the number of respiratory cycle bins, then the bin numbering of the cardiac motion projection data is: Respiratory motion projection data bin number: Projection data set: As a preferred approach to address mismatches caused by arrhythmias and irregular breathing, this invention introduces a joint phase pairing strategy. This strategy dynamically allocates projected data based on the real-time cardiac and respiratory phase relationships, and normalizes the cardiac phase... With normalized respiratory phase Combined into a two-dimensional joint phase vector: And based on the center phase of each cardiac chamber and respiratory chamber Based on this, the projected data is allocated to the optimal bins using distance minimization or weighted allocation methods.

[0040] in, for Center phase coordinates of the sub-box For the center phase of all candidate bins, To assign smoothing coefficients, which are used to control the weighting range, , These represent the candidate indices for the cardiac and respiratory phase bins, respectively.

[0041] Weight Reflection Moment t The degree of matching between the projected data and the phase of the target bin center, when the actual phase With a certain distribution center The closer the bins are, the larger their corresponding weights. The sum of all bin weights after normalization is 1.

[0042] Based on the above weight definition, the projected data within each bin is weighted and aggregated to obtain the first... The projected set of bins.

[0043] in, Indicates time t The collected raw projection data, To be allocated to the first Weighted projection results of binning.

[0044] This weighted allocation mechanism enables smooth allocation and robust pairing of projection data under conditions of arrhythmia or irregular breathing, effectively reducing mismatch errors caused by phase boundary discontinuities and uneven signal sampling, improving the stability of joint phase binning and the spatiotemporal continuity of subsequent image reconstruction, thereby avoiding mismatches and artifacts caused by arrhythmia and irregular breathing.

[0045] Step 4, for each Image reconstruction is performed on the projection data within the sub-bins to obtain multiple sets of three-dimensional volume data of cardiopulmonary combined exercise. The three-dimensional volume data is then integrated to obtain multi-temporal images of cardiopulmonary combined exercise (multi-temporal three-dimensional volume data).

[0046] The 3D image reconstruction uses a mature CT reconstruction algorithm from the field of medical imaging. These techniques include filtered back projection (FBP), statistical iterative reconstruction (SIR), or medical image sparse reconstruction techniques to ensure imaging quality for low-dose and sparse projection data.

[0047] Formula for 3D reconstruction of CT images: in, For the sake of the first projection data set The volume function of the 3D image obtained by the 3D reconstruction algorithm represents the voxel gray-level distribution at the corresponding moments of the cardiac and respiratory phases. for The set of projected data in the bins.

[0048] Some bins may have sparse or missing projection data due to arrhythmias and irregular breathing. To address this issue, as a preferred solution, deep learning-based interpolation or artificial intelligence models are used to complete the images for bins with insufficient projection data, ensuring the continuity and integrity of full-time cardiopulmonary motion data. Insufficient projection data is defined as when the number of projections acquired in any combined cardiac and respiratory phase bin is less than 70% of the number of projections required for complete CT reconstruction, or when its angular coverage is less than 75% of the full scan angle.

[0049] AI-based data completion formula: Where G is a deep learning generative model. express( i , j The set of adjacent or related time phases.

[0050] The algorithm of the artificial intelligence model is selected from convolutional neural networks or Transformer networks; The input data for the artificial intelligence model includes adjacent temporal data and historical signal features. It also incorporates normalized encoding of cardiac and respiratory phases as conditional input to ensure that the generated result conforms to the cardiopulmonary exercise rhythm. The normalized encoding is as follows: in, For the heart phase, This is the respiratory phase; The artificial intelligence model incorporates kinematic constraints during training and inference, including displacement field smoothness, anatomical structure consistency, and Jacobian positive constraint, thereby avoiding the generation of false motion patterns that do not conform to physiological laws.

[0051] The artificial intelligence model incorporates projection consistency, adjacent temporal continuity, and kinematic regularization terms into the loss function to obtain more stable and realistic multi-temporal completion data.

[0052] Multi-temporal 3D volumetric data integration: The multi-temporal three-dimensional volumetric data obtained in this embodiment can be used in clinical scenarios such as radiotherapy motion management, gating, tracking irradiation planning, motion characteristic analysis, and quantitative assessment of radiotherapy dosimetric uncertainty. It can significantly improve the safety and efficacy assessment accuracy of radiotherapy for patients with arrhythmias and complex cardiopulmonary movements.

[0053] Example 2 Retrospective Reconstruction Scheme This embodiment illustrates an application scenario for the method and system described in Embodiment 1. Specifically, the application scenario for this embodiment is retrospective reconstruction. For example... Figure 1 As shown, this method is not limited to a specific gating phase during CT scanning, but employs long-term continuous scanning to cover multiple complete cardiac and respiratory cycles. Subsequently, three-dimensional reconstruction of the data was retrospectively performed using the method and system of Example 1.

[0054] Example 3: Proactive Reconstruction Scheme This embodiment illustrates an application scenario for the method and system described in Embodiment 1. Specifically, the application scenario in this embodiment is forward-looking reconstruction. Figure 1As shown, this method involves real-time monitoring of ECG and respiratory signals during CT scans. The system only activates when both signals simultaneously enter a preset target time window—that is, when the ECG signal is in the target cardiac sub-box and the respiratory signal is in the target respiratory sub-box (i.e., when the preset time window is reached). Projection data is collected only during binning. The acquisition method in this embodiment can effectively avoid redundant scanning and ensure that the acquired data accurately corresponds to the timing phase of the dual-gated target.

[0055] Example 4: Respiratory Phase Triggering Acquisition Scheme This embodiment illustrates an application scenario of the method and system described in Embodiment 1. Specifically, the application scenario of this embodiment is a respiratory phase-triggered acquisition scheme. This method divides the entire respiratory cycle into... M Each box (i.e.) M (Each sub-phase box) monitors the ECG signal in real time within each sub-phase box, and initiates projection data acquisition only when the ECG signal is in the preset target cardiac box phase.

[0056] As can be seen from the above embodiments, this invention constructs a novel method and system for cardiopulmonary motion imaging reconstruction, which is particularly suitable for patients with arrhythmias and other conditions involving non-isocyclic heartbeat and respiratory motion coupling. This invention can provide high-quality imaging data and theoretical basis for clinical radiotherapy, exercise management, and scientific research, and has excellent application prospects.

Claims

1. A method for cardiopulmonary motion imaging reconstruction, characterized in that, Includes the following steps: Step 1: Simultaneously collect electrocardiogram (ECG) and respiratory signals, and record the collection time; Step 2: The ECG signal and respiratory signal are synchronously subjected to periodic detection and phase normalization processing to obtain the normalized cardiac phase and normalized respiratory phase corresponding to each acquisition time. Subsequently, the normalized cardiac and respiratory phases were binned together to characterize the cardiac-respiratory dual-phase state at each acquisition time. Step 3, according to the collection time Binning can be performed either by sorting bins or by using a weighted allocation strategy. i Bins representing normalized cardiac phase, j Bins representing normalized respiratory phase; Step 4: Incorporate the projection data into each bin according to the binning results; Step 5: Perform image reconstruction on the projection data in each sub-box to obtain multiple sets of three-dimensional volume data of cardiopulmonary combined exercise, and integrate the three-dimensional volume data to obtain multi-temporal images of cardiopulmonary combined exercise.

2. The cardiopulmonary motion imaging reconstruction method according to claim 1, characterized in that, The projection data is CT scan data or CBCT data; And / or, the projection data is real-time acquired data or historical data used for retrospective image reconstruction; And / or, the method for acquiring the projection data is selected from at least one of the following: All projection data are collected in real time at each acquisition moment, and then the collected projection data is incorporated into each bin according to the binning results. The system presets the target bins for which projection data needs to be collected, monitors the normalized cardiac phase and / or normalized respiratory phase in real time, and collects projection data only when the collection time belongs to the target bin. The collected projection data is then incorporated into the target bin.

3. The cardiopulmonary motion imaging reconstruction method according to claim 1, characterized in that, In step 2, the steps to obtain the normalized cardiac phase corresponding to each acquisition time include: detecting the heartbeat start point of the acquired electrocardiogram signal and calculating the length of each heartbeat cycle; for any acquisition time, determining the heartbeat cycle in which it is located, and linearly normalizing the position of the acquisition time in the cycle to the interval [0,1) according to the start time and duration of the cycle, so as to obtain the normalized cardiac phase, which reflects the periodic motion state of the heart at different times. The steps to obtain the normalized respiratory phase corresponding to each acquisition time include: detecting the respiratory initiation point of the acquired respiratory signal and calculating the length of each respiratory cycle; for any acquisition time, determining the respiratory cycle in which it is located, and linearly normalizing the position of the acquisition time within the cycle to the interval [0,1) according to the start time and duration of the cycle, so as to obtain the normalized respiratory phase, which reflects the relative temporal changes of respiratory motion.

4. The cardiopulmonary motion imaging reconstruction method according to claim 3, characterized in that, In step 2, for patients with arrhythmia, abnormal cardiac cycles are further identified based on the variability of the RR interval, and the projection data of abnormal cycles are removed. At the same time, the number and width of the cardiac phase bins are adaptively adjusted according to the degree of variability of the RR interval. The adjustment method includes the following steps: RR interval sequences were obtained by QRS wave detection: in, For the actual RR interval, To indicate the first The timing of the R-wave peak occurring with each heartbeat. k Here, K is the heartbeat index number, and K is the total number of R waves detected. Abnormal cardiac cycles are identified based on the coefficient of variation or a preset threshold, and these abnormal cycles are then removed. The coefficient of variation is: in, Indicates the first k The actual RR interval per heartbeat cycle, To represent the standard deviation of all RR intervals, To represent the mean of all RR intervals; After calculating the coefficient of variation of the RR interval, it is further determined whether each period is an abnormal period based on the relative deviation of each period; the period deviation rate is defined as follows: when > If a period is identified as an abnormal RR period, the corresponding projection data for that period is directly removed. The preset threshold; when Less than the preset threshold At that time, the number of boxes is taken as the initial value. ;when Greater than At that time, the number of bins is adaptively adjusted according to the degree of variation, and the adjustment formula is as follows: in, N To adjust the number of boxes, The initial number of boxes is , where This is an empirical adjustment coefficient used to control the adjustment range of the number of boxes.

5. The cardiopulmonary motion imaging reconstruction method according to claim 4, characterized in that, During abnormal cardiac cycles, the normalized cardiac phase is calculated using the following formula: in, For normalized cardiac phase, t The time point for acquiring the projection signal. For the actual RR interval, To indicate the first The timing of the R-wave peak occurring with each heartbeat. The variability correction weight is defined as: in, The mean RR interval is the global interval.

6. The cardiopulmonary motion imaging reconstruction method according to claim 1, characterized in that, When the signal contains noise, phase boundary discontinuities, or uneven sampling timing, a joint phase pairing strategy is adopted to normalize the cardiac phase. With normalized respiratory phase Combined into a two-dimensional joint phase vector: And based on the center phase of each cardiac chamber and respiratory chamber Based on this, the projected data is allocated to the optimal bins using distance minimization or weighted allocation methods; in, for Center phase coordinates of the sub-box For the center phase of all candidate bins, To assign smoothing coefficients, which are used to control the weighting range, , These represent all possible indices for the cardiac and respiratory phase bins, respectively. Weight Reflection Moment t The degree of matching between the projected data and the phase of the target bin center, when the actual phase With a certain distribution center The closer the bins are, the larger their corresponding weights are; the sum of all bin weights after normalization is 1. Based on the above weight definition, the projected data within each bin is weighted and aggregated to obtain the first... The projected set of bins; in, Indicates time t The collected raw projection data, To be allocated to the first Weighted projection results of binning.

7. The cardiopulmonary motion imaging reconstruction method according to claim 1, characterized in that: In step 4, for bins with insufficient projection data, deep learning-based interpolation methods or artificial intelligence models are used to complete the images, ensuring the continuity and integrity of the full-time cardiopulmonary exercise data. The criterion for insufficient projection data is: in any cardiac-respiratory phase sub-bin... If the number of projections collected is less than 70% of the number of projections required for complete 3D reconstruction, or if the angle coverage is less than 75% of the full scanning angle, it is considered that the projection data is insufficient.

8. The cardiopulmonary motion imaging reconstruction method according to claim 7, characterized in that: The algorithm of the artificial intelligence model is selected from convolutional neural networks or Transformer networks; The input data for the artificial intelligence model includes adjacent temporal data and historical signal features, while also incorporating normalized codes for cardiac and respiratory phases as conditional inputs. These normalized codes are: in, For the heart phase, This is the respiratory phase; The artificial intelligence model incorporates kinematic constraints during training and inference, including displacement field smoothness, anatomical structure consistency, and Jacobian positive constraints. The artificial intelligence model incorporates projection consistency, adjacent temporal continuity, and kinematic regularization terms into the loss function.

9. A cardiopulmonary motion imaging reconstruction system, characterized in that, The method for implementing the cardiopulmonary combined motion imaging reconstruction method according to any one of claims 1-8 includes a signal acquisition module and an intelligent box-type system; The signal acquisition module is configured to acquire projection data, electrocardiogram (ECG) signals, and respiratory signals, and record the acquisition time of each projection data, ECG signal, and respiratory signal. The intelligent container sorting system includes: The binning module is configured to synchronously perform periodic detection and phase normalization processing on the electrocardiogram signal and respiratory signal to obtain the normalized cardiac phase and normalized respiratory phase corresponding to each acquisition time; then, the normalized cardiac phase and respiratory phase are jointly binned to characterize the cardiac-respiratory dual-phase state at each acquisition time. The data grouping module groups the data collection times according to... Binning can be performed either by sorting bins or by using a weighted allocation strategy. i Bins representing normalized cardiac phase, j Binning is used to represent the normalized respiratory phase; the projected data is then incorporated into each bin according to the binning results. The three-dimensional reconstruction module is configured to perform image reconstruction on the projection data in each sub-box to obtain multiple sets of three-dimensional volume data of cardiopulmonary combined exercise, and integrate the three-dimensional volume data to obtain multi-temporal images of cardiopulmonary combined exercise.

10. A computer-readable storage medium, characterized in that, It stores a computer program for implementing the cardiopulmonary combined motion imaging reconstruction method according to any one of claims 1-8.

Images