PET / MR (positron emission tomography / magnetic resonance) all-in-one machine-based dynamic attenuation correction and respiratory and cardiac dual-motion correction method

Abstract

The invention discloses a PET / MR (positron emission tomography / magnetic resonance) all-in-one machine-based dynamic attenuation correction and respiratory and cardiac dual motion correction method. The method comprises the following steps of: generating a dynamic mu graph of a respiratory cycle; mR cardiac function movie sequence scanning of the cardiac cycle; accurately segmenting the cardiac muscle; 6, performing tissue segmentation on the dynamic mu graph; clustering the PET data; pET data space-time registration is carried out; merging PET data of the same time phase; performing frame-by-frame dynamic attenuation correction; respiratory movement correction; correcting a partial volume effect; carrying out myocardial MR image space standardization on different time phases of the cardiac cycle; and carrying out spatial standardization on the myocardial PET image of different time phases in the cardiac cycle. According to the method, the mu graph of the respiration and cardiac cycles is reconstructed through time-phase sharing and is accurately registered with the PET data, so that the high-quality attenuation correction of the heart PET image is realized and the motion correction is completed at the same time.

Description

Technical Field

[0001] The present invention belongs to the technical field of nuclear medicine imaging, and in particular relates to a dynamic attenuation correction and respiratory and cardiodynamic dual motion correction method based on a PET / MR all-in-one machine. Background Art

[0002] The photons used in positron emission tomography (PET) imaging experience varying degrees of attenuation as they propagate within tissues, particularly in deep tissues, where they must traverse more tissue, resulting in a weakened signal reaching the detector. Without attenuation correction, the radioactive signal from deep tissues will appear weaker than that from superficial tissues, leading to uneven signal intensity in the image and compromising diagnostic accuracy. Attenuation correction is an essential step in PET imaging. By correcting for signal loss caused by tissue attenuation, it ensures image authenticity and accuracy, providing a reliable basis for clinical diagnosis.

[0003] In traditional cardiac PET imaging, the attenuation correction (AC) technology based on MR or CT has the following defects: the cardiac PET imaging process is in a state of free breathing movement and cardiac beating without breath holding. The existing AC technology uses a single static μ image and does not take into account the organ displacement caused by breathing and cardiac movement, resulting in misclassification of the attenuation values of the myocardium and the surrounding cardiac chambers, large blood vessels, and lung tissue, resulting in artifacts.

[0004] In an integrated PET / MR system, the PET detector is tightly mounted on the MR RF coil, allowing simultaneous data acquisition from both imaging modalities. This synchronizes the patient's breathing, heartbeat, and other motion information. Using phase-matched MR images for PET attenuation correction can minimize respiratory artifacts. "Spatiotemporal consistency" is the primary advantage of using MR images for PET attenuation correction. However, current AC techniques typically use a single static μ image to perform AC on PET data, introducing distortion during this critical AC step.

[0005] MRAC, the most commonly used technique in integrated PET / MR systems, emits less ionizing radiation than CTAC and holds great promise for clinical application in PET imaging, enabling precise dynamic attenuation correction and respiratory-cardiac dual motion correction. This technique combines DIXON water-fat separation technology with UTE (ultra-short echo time) or ZTE (zero echo time) sequences for tissue segmentation, employing a five-tissue classification scheme (including water / soft tissue, fat, air, lung, and bone) for attenuation correction. However, traditional five-tissue segmentation techniques fail to distinguish the myocardium from the blood pool within cardiac chambers and major vessels, leading to errors in attenuation correction for myocardial PET images.

[0006] In order to reduce the negative impact of respiratory motion on PET data processing and reconstruction, some technologies only use PET data from the phase with less motion at the end of exhalation for AC or reconstruction. However, a large amount of PET data outside the end of inspiration is discarded and not included in the final image reconstruction, and the impact of cardiac pulsation on AC is not considered. Therefore, the current attenuation correction technologies for cardiac PET images have not solved the fundamental problem, and the image quality after attenuation correction is not much different between different AC technologies. Existing MRI-based PET motion compensation technologies are mostly limited to single respiratory or ECG gating, and cannot directly couple spatiotemporally synchronized dynamic attenuation correction with respiratory / cardiac motion correction on the original PET data. Summary of the Invention

[0007] The present invention is proposed to address the above-mentioned shortcomings, and its purpose is to provide a method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on a PET / MR integrated device. This method reconstructs the μ map of the respiratory and cardiac cycles by time phase and accurately aligns it with the PET data, thereby achieving high-quality attenuation correction of cardiac PET images while completing motion correction.

[0008] To achieve the above objectives, the present invention provides a method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on a PET / MR integrated machine, comprising the following steps: S1: Synchronously acquire PET and MR data in the free-breathing state while monitoring the respiratory motion amplitude; divide the respiratory cycle into several phases according to the respiratory motion amplitude, generate the corresponding five-tissue segmentation μ map, and form a dynamic μ map of several frames of respiratory cycles; S2: 3DSSFP was used to perform MR cardiac function movie sequence scanning, and each cardiac cycle was divided into several phases based on the RR interval of ECG gating; S3: Based on the MR signal differences between the myocardium, cardiac vessels, and lung tissue, myocardial segmentation is performed at each phase of the cardiac cycle. S4: Based on the five-tissue segmentation μ-map, the myocardium is separately segmented as the sixth tissue type; in several phases of the cardiac cycle in each respiratory cycle, μ-maps of the six-tissue segmentation are separately generated, and dynamic μ-maps of the six-tissue segmentation of several phases of the respiratory cycle and their corresponding several phases of the cardiac cycle are obtained; S5: Acquire 3D PET data, divide each respiratory cycle into several phases according to the respiratory motion amplitude, and further divide the phase of each respiratory cycle into several phases of PET data of the cardiac cycle, thereby obtaining several clusters of raw PET data; S6: Using rigid body transformation, each cluster of PET data is registered with the dynamic μ map of the corresponding phase; S7: superimposing and recombining several clusters of PET data of the same time phase to form several frames of PET images that are not attenuated and arranged in time sequence; S8: First perform scatter correction on several frames of PET images, and then perform attenuation correction based on the corresponding μ map of six tissue segmentations; S9: Using the end-inspiratory phase PET image as a reference, register and superimpose the attenuation-corrected images of the same phase of the cardiac cycle within the respiratory cycle to obtain several frames of attenuation-corrected myocardial PET images that are not affected by respiratory motion; S10: after registering the acquired attenuation-corrected myocardial PET images with the MR structural images of the corresponding time phase, the myocardial PET images are corrected for partial volume effects based on the 3D MR cardiac cine structural images; S11: Using the end-diastolic MR cardiac structural image as a template, perform affine transformation and nonlinear transformation on the non-end-diastolic image to complete spatial standardization and save the transformation parameters; S12: performing spatial normalization transformation on the non-end-diastolic myocardial PET image according to the acquired conversion parameters, and superimposing and recombining the image with the end-diastolic myocardial PET image into a single myocardial PET image.

[0009] Furthermore, in step S1, the List-mode mode is combined with segmented K-space acquisition to reorganize multiple respiratory cycles into respiratory motion movie images that are arranged and played back in time sequence, and the respiratory cycle is divided into 8 phases according to the respiratory motion amplitude, and a five-tissue segmentation μ map corresponding to each phase is generated. The five-tissue segmentation μ map includes water / soft tissue, fat, air, lung, and bone.

[0010] Furthermore, in step S2, compressed sensing 3DSSFP is used to perform MR cardiac function movie sequence scanning, the scanning adopts a single breath-hold or free breathing state, and each cardiac cycle is equally divided into 8 phases based on the RR interval of ECG gating.

[0011] Furthermore, in step S4, a six-tissue segmentation μ map is generated separately in each of the eight phases of the cardiac cycle in each respiratory cycle, thereby obtaining 64 six-tissue segmentation dynamic μ maps of the eight phases of the respiratory cycle and the corresponding eight phases of the cardiac cycle.

[0012] Furthermore, in step S5, 3D PET data is collected in List-mode, each respiratory cycle is divided into 8 phases according to the amplitude of respiratory movement, and each phase of the respiratory cycle is further divided into PET data of 8 phases of the cardiac cycle, obtaining 64 clusters of PET raw data of 8 phases of the respiratory cycle and corresponding 8 phases of the cardiac cycle.

[0013] Furthermore, in step S7, the 64 clusters of PET data of the same phase between different respiratory cycles and cardiac cycles are superimposed and recombined to form 64 frames of PET images of the same phase between different respiratory cycles and cardiac cycles, which are arranged in time sequence and are not attenuated and arranged in time sequence; In step S8, after scatter correction is performed on the PET images arranged in time sequence, attenuation correction is performed based on the 64 μ maps of six tissue segmentations to obtain 64 frames of PET images of different respiratory and cardiac cycles that have been attenuated and arranged in time sequence.

[0014] Furthermore, in step S9, a rigid body transformation is used to align the myocardial images of each frame of the PET image in the same phase of the cardiac cycle after attenuation correction within the respiratory motion cycle with the PET image of the end-inspiratory phase of the respiratory motion cycle as a reference, and the data is superimposed and reorganized, thereby obtaining 8 frames of myocardial PET images that are attenuated and corrected and arranged in time sequence and are not affected by respiratory motion factors.

[0015] Furthermore, in step S10, a rigid body transformation is used to register the acquired PET cardiac images of eight phases within the cardiac cycle with the MR structural images of the corresponding phases to achieve spatial standardization and consistency between the myocardial PET images and the MR images. Based on the structural relationship between the myocardium and the surrounding cardiac chambers, great vessels, and lung tissue in the 3D MR cardiac cine structural images, the eight frames of attenuation-corrected myocardial PET images are corrected for partial volume effects.

[0016] Furthermore, in step S11, the 3D MR cardiac movie structural image at the non-end-diastolic phase of the cardiac cycle is spatially normalized with reference to the MR cardiac structural image at the end-diastolic phase, and the converted myocardial image is consistent in morphology, contour and spatial coordinates; through spatial normalization transformation using affine transformation and nonlinear transformation, the converted MR image at the non-end-diastolic phase is precisely aligned with the MR template image at the end-diastolic phase in the standard space, and the corresponding conversion parameters are saved.

[0017] Furthermore, in step S12, according to the conversion parameters obtained in different phases, the myocardial PET image in the non-end-diastolic period of the cardiac cycle is spatially standardized with reference to the myocardial PET image in the end-diastolic period through spatial standardization conversion of affine transformation and nonlinear transformation, and the consistency of the converted myocardial image in morphology, contour and spatial coordinates is achieved; finally, the standardized non-end-diastolic myocardial PET image and the end-diastolic myocardial PET image are superimposed and recombined into a single myocardial PET image.

[0018] Compared with the prior art, the present invention has the following beneficial effects: First, based on the characteristics of the PET / MR all-in-one machine that synchronously acquires PET and MR signals, combined with the real-time monitoring of respiratory movement by non-magnetic sensors on the body surface, the present invention proposes a method for dynamic attenuation correction and motion correction. This method reconstructs the μ map of the respiratory and cardiac cycles by time phases and accurately aligns it with the PET data, thereby achieving high-quality attenuation correction of cardiac PET images while completing motion correction.

[0019] Secondly, the present invention significantly improves the accuracy of attenuation correction. Existing PET / MR myocardial imaging typically uses static MR attenuation correction (MRAC) methods that fail to account for cardiac and respiratory motion. This can easily lead to misalignment between the μ map and PET data or tissue misclassification due to motion, resulting in attenuation correction artifacts. To address these limitations, the present invention introduces a dynamic μ map generation technique with respiratory-cardiac dual gating. While preserving all PET data, this technique uses synchronized PET / MR acquisition and real-time gating to reconstruct the attenuation coefficient distribution for each respiratory and cardiac phase and accurately align it with the corresponding PET data. Compared to conventional MRAC, which only provides a single static μ map, the present invention's dynamic attenuation correction updates attenuation information in real time with cardiac and respiratory motion, avoiding correction errors caused by mismatches between PET images and μ maps. Furthermore, the present invention utilizes μ map optimization based on six tissue classifications, independently segmenting the myocardium into a sixth tissue type, in addition to the traditional five tissue types of soft tissue (fat, lung, bone, and air). This effectively reduces the interference of intracardiac blood and lung tissue on the accuracy of myocardial attenuation correction. Through the above improvements, the present invention can directly perform accurate frame-by-frame attenuation correction on the original myocardial PET data, greatly improving the accuracy and quantitative reliability of image attenuation correction.

[0020] Third, the present invention enhances motion compensation capabilities. The heart's cyclical contraction and relaxation, as well as respiratory motion, are the primary factors contributing to PET image blur and artifacts, and are considered one of the "most severe limitations" in traditional PET imaging. Existing technologies typically mitigate the effects of motion through either single respiratory gating or cardiac gating. However, single-gating methods cannot simultaneously correct for both cardiac and respiratory motion, and residual blurring still reduces image clarity. Furthermore, traditional motion correction based on gated reconstruction often results in increased image noise and decreased spatial resolution due to the splitting of data into multiple frames. To address this issue, the present invention utilizes MRI information acquired simultaneously by a PET / MR integrated device to obtain precise cardiac and respiratory motion fields, combines it with surface-based non-magnetic sensors to achieve respiratory-cardiac dual gating, and employs rigid body registration and nonlinear deformation to calibrate each phase of PET images to a unified reference space, thereby fully compensating for respiratory motion and cardiac displacement in myocardial PET. Compared to existing single-gating or no-correction schemes, the present invention's dual-gated motion correction virtually eliminates motion blur in myocardial images without increasing noise. After correction and alignment, the myocardial edge is clearer and motion artifacts are significantly reduced, which improves the reliability of lesion detection and quantitative analysis.

[0021] Fourthly, the data utilization efficiency of the present invention is higher: the gated imaging of traditional cardiac PET often divides the acquired data into multiple cardiac or respiratory phases and reconstructs them separately. Although this can reduce some motion blur, each frame of the image only contains a small part of the count of the original event, and the quality of a single frame is relatively general. Phase-by-phase gated reconstruction leads to a significant reduction in the effective count of a single image, and the overall data utilization is poor. To address this shortcoming, the present invention does not discard any acquired PET events, but subdivides the list mode data obtained in the free breathing state into a total of 64 clusters according to 8 respiratory phases and 8 cardiac phases, and after alignment and correction with the dynamic μ map of the corresponding phase, the 64 clusters of PET data of the same phase in all different cycles are superimposed and synthesized into a complete time-series image sequence. After spatial alignment and fusion of each gated sub-data, the present invention contributes all the counts of the original acquisition to image reconstruction, which greatly improves the utilization efficiency of PET data and the overall sensitivity of the image compared to the traditional gating method that only uses part of the data.

[0022] Fifth, the image signal-to-noise ratio of the present invention is significantly improved. Conventional double-gated reconstruction, while reducing motion blur, will cause the PET image noise level to increase significantly, about double the level of uncorrected full-data reconstruction; while MRI-based synchronous motion correction can eliminate blur while maintaining the noise at a level comparable to the full-data image. By fusing all gated frame data after motion compensation, the present invention recovers the counts lost due to gated splitting, significantly improving the signal-to-noise ratio of the reconstructed image. Therefore, the myocardial PET images obtained by the present invention have a higher signal-to-noise ratio and contrast, which is beneficial for the detection of subtle lesions and improving the accuracy of quantitative analysis.

[0023] Sixth, the present invention achieves higher myocardial segmentation accuracy. Existing MR attenuation correction methods often use limited tissue classification (e.g., only 4-5 categories, such as soft tissue, fat, lung, and bone), failing to independently identify the myocardium. This results in myocardial signal confounding with cardiac blood and lung tissue, leading to inaccurate attenuation coefficient assignment and reduced accuracy of myocardial attenuation correction. To address this, the present invention incorporates a six-tissue segmentation scheme into MR μ map generation: utilizing cardiac MRI cine sequences to rapidly and accurately delineate the ventricular endocardium, epicardium, and papillary muscle contours at each cardiac phase, clearly separating the myocardium from surrounding cardiac and lung tissue and treating it as a separate, sixth tissue category for attenuation correction. Leveraging this refined, frame-by-frame tissue classification, the present invention assigns more accurate attenuation coefficients to the myocardial region, avoiding confusion between the attenuation characteristics of the myocardium and surrounding tissues, thereby improving the accuracy and quantitative reliability of attenuation correction for myocardial PET images. Furthermore, the present invention combines high-resolution MRI structural imaging with partial volume effect correction on the corrected myocardial PET images, mitigating signal loss at the myocardial edge due to device resolution limitations, further enhancing image clarity and diagnostic accuracy.

[0024] Seventh, the present invention significantly improves the clinical applicability of existing PET / MR cardiac imaging procedures. For example, to obtain clear MR attenuation maps and cardiac function images, patients are typically required to hold their breath for approximately 10-20 seconds to reduce blurring and misalignment caused by respiratory motion. However, many heart disease patients find it difficult to hold their breath for extended periods, and inadequate cooperation can introduce attenuation correction errors and image artifacts. Furthermore, many methods rely on the injection of contrast agents to enhance myocardial signals or cumbersome manual procedures (such as manual myocardial delineation), increasing the complexity of the examination and the burden on the patient. To address these shortcomings, the present invention significantly improves the clinical practicality of this approach: the entire imaging process is compatible with free breathing, eliminating the need for breath-holding. It utilizes surface-mount non-magnetic sensors combined with MRI sequences in the absence of contrast agents to fully automatically acquire cardiac / respiratory gating signals, perform myocardial segmentation, and perform motion correction, all without manual intervention. This reduces the need for patient cooperation, simplifies the procedure, and reduces the risk of scan failure, making it more suitable for routine clinical use. Furthermore, the system is compatible with free-breathing scanning, eliminating the need for breath-holding and simplifying the imaging procedure. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 Schematic diagram of a flow chart of a method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on a PET / MR all-in-one machine according to an embodiment of the present invention; Figure 2 Generate a schematic diagram for the dynamic μ map of the respiratory cycle; Figure 3 Generate schematic diagrams for the μ-map of six tissue segmentations; Figure 4Schematic diagram of respiratory motion correction for PET data; Figure 5 Schematic diagram of cardiac cycle motion correction for PET data; In the figure, 101-respiratory cycle curve, 102-time phase, 103-dynamic μ map of respiratory cycle, 201-μ map of five-tissue segmentation, 202-myocardial tissue, 203-μ map of six-tissue segmentation, 301-myocardial PET images of different phases of the cardiac cycle in the non-end-inspiratory phase, 302-myocardial PET images of different phases of the cardiac cycle in the end-inspiratory phase, 303-myocardial PET images after registration and superposition reconstruction in the same phase of the cardiac cycle, 401-myocardial PET images after MR image registration, 402-myocardial MR images of different phases of the cardiac cycle after spatial normalization, 403-myocardial PET images of different phases of the cardiac cycle in spatially normalized. DETAILED DESCRIPTION

[0026] The following describes the implementation of the present invention in detail with reference to the examples, but they do not limit the present invention and are merely examples. At the same time, the advantages of the present invention will become clearer and easier to understand.

[0027] The present invention provides a method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on a PET / MR integrated machine, comprising the following steps: S1: Dynamic μ-map generation during the respiratory cycle PET data and 3DDIXON+UTE / ZTE sequence MR data in the free-breathing state are acquired simultaneously, and respiratory amplitude is monitored using surface non-magnetic sensors. The List-mode mode, combined with segmented K-space acquisition technology, reorganizes multiple respiratory cycle data into respiratory motion movie images that are arranged and played back in time sequence. The respiratory cycle curve 101 is divided into 8 phases 102 according to the respiratory motion amplitude, and a five-tissue segmentation μ map (water / soft tissue, fat, air, lung, bone) corresponding to each phase is generated, forming an 8-frame dynamic μ map 103 of the respiratory cycle, as shown in FIG. Figure 2 Shown is a schematic diagram of generating a respiratory-gated dynamic μ map.

[0028] S2: MR cardiac function movie sequence scan during the cardiac cycle High-resolution MR cardiac function cine sequence scanning was performed using compressed sensing three-dimensional steady-state free precession (3D SSFP). Each cardiac cycle was divided into eight phases based on the RR interval of ECG gating. The scanning could be performed in a single breath-hold or free breathing state.

[0029] S3: Precise Myocardial Segmentation Based on the significant differences in MR signals between the myocardium and the cardiac chambers' large blood vessels and lung tissue, myocardial segmentation is performed in each phase of the cardiac cycle to quickly and accurately define the inner and outer membranes of the ventricles and the contours of the papillary muscles, thereby accurately segmenting the signals of the myocardium, cardiac chambers' large blood vessels, and lung tissue in each phase.

[0030] S4: Dynamic μ-map of six tissue segmentation Based on the five-tissue segmentation μ-map 201, the myocardium is further segmented from the surrounding cardiac chambers, large blood vessels, and lung tissue, and the myocardial tissue 202 is separately segmented into the sixth tissue type. In each of the eight phases of the cardiac cycle in the respiratory cycle, a six-tissue segmentation μ-map 203 is generated separately, resulting in a total of 64 six-tissue segmentation dynamic μ-maps for the eight phases of the respiratory cycle and the eight phases of the cardiac cycle, as shown in FIG. Figure 3 Shown is a schematic diagram of the μ map generated for six tissue segmentations.

[0031] S5: PET data clustering In list-mode, 3D PET data is collected, with each respiratory cycle divided into eight phases based on the amplitude of the respiratory motion. Each respiratory phase is further divided into eight phases of PET data corresponding to the cardiac cycle. This results in a total of 64 clusters of raw PET data, spanning eight phases of the respiratory cycle and eight phases of the cardiac cycle.

[0032] S6: Spatiotemporal registration of PET data Using rigid body transformation technology, each cluster of PET data is registered with the μ-map of the corresponding phase to ensure spatial and temporal consistency. This achieves spatial standardization and consistency between PET data and MR images of the same phase, and also ensures spatial standardization and consistency between PET images of each phase across different respiratory and cardiac cycles.

[0033] S7: Merging PET data of the same phase After spatial normalization, all 64 clusters of PET data at the same phase between different respiratory and cardiac cycles are superimposed and reassembled to form 64 frames of PET images at the same phase between different respiratory and cardiac cycles, which are arranged in time sequence and are not attenuated and can be replayed in a loop. In this embodiment, myocardial PET images 301 at different phases of the cardiac cycle other than the end-inspiratory phase, myocardial PET images 302 at different phases of the cardiac cycle at the end-inspiratory phase, and myocardial PET images 303 after registration and superimposition of the same phase of the cardiac cycle are shown as follows: Figure 4 shown.

[0034] S8: Frame-by-frame dynamic attenuation correction The above 64 frames of PET images of different respiratory and cardiac cycles arranged in time sequence are subjected to scatter correction and attenuation correction based on the 64 μ maps of six tissue segmentation, thereby obtaining 64 frames of PET images of different respiratory and cardiac cycles, which are attenuation corrected, arranged in time sequence, and can be replayed in a loop.

[0035] S9: Respiratory motion correction Using rigid-body transformation technology, each PET image frame from the same cardiac cycle phase, after attenuation correction during the respiratory cycle, is then registered with the PET image at the end of inspiration during the respiratory cycle as a reference for myocardial image registration and data reconstructing. This step further improves the signal-to-noise ratio of the myocardial PET images, building on the effective respiratory motion correction and attenuation correction. The result is eight attenuation-corrected myocardial PET frames, arranged in a time sequence and capable of looped playback.

[0036] S10: Partial Volume Effect Correction Using rigid-body transformation technology, the PET cardiac images acquired over eight phases of the cardiac cycle were registered with the corresponding MR structural images, achieving spatial standardization and consistency between the myocardial PET and MR images. Based on the structural relationship between the myocardium and surrounding cardiac chambers, great vessels, and lung tissue in the 3D MR cardiac cine structural images, the eight attenuation-corrected myocardial PET images were corrected for partial volume effects, further improving image quality and diagnostic accuracy.

[0037] S11: Spatial normalization of myocardial MR images at different phases of the cardiac cycle The 3D MR cardiac cine images of the non-diastolic phase of the cardiac cycle are spatially normalized with reference to the MR cardiac structural images of the end-diastolic phase, ensuring consistency in morphology, contours, and spatial coordinates of the converted myocardial images. Using spatial normalization techniques such as affine transformation and nonlinear transformation, the deformed non-diastolic MR images are precisely aligned with the MR template images of the end-diastolic phase within the standard space, and the corresponding transformation parameters are saved.

[0038] S12: Spatial standardization of myocardial PET images at different phases of the cardiac cycle According to the conversion parameters of different time phases obtained in the previous step, the myocardial PET images at the non-end-diastolic phase of the cardiac cycle are spatially normalized with reference to the myocardial PET images at the end-diastolic phase through spatial normalization conversion techniques such as affine transformation and nonlinear transformation to ensure the consistency of the converted myocardial images in morphology, contour and spatial coordinates. Figure 5The figure shows a schematic diagram of cardiac cycle motion correction for PET data. Finally, the standardized non-end-diastolic myocardial PET image and the end-diastolic myocardial PET image are superimposed and reconstructed into a single myocardial PET image, which further improves the signal-to-noise ratio and facilitates subsequent image reading or semi-quantitative bull's eye image analysis. In this embodiment, based on the myocardial PET image 401 after MR image registration, the myocardial MR image 402 of different cardiac cycle phases after spatial standardization, and the myocardial PET image 403 of different cardiac cycle phases after spatial standardization, as shown in FIG. Figure 5 shown.

[0039] The above is only a specific embodiment of the present invention. It should be pointed out that any changes or substitutions that can be easily thought of by any technician familiar with the field within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention. The rest not described in detail are prior art.

Claims

1. A method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on a PET / MR integrated device, characterized by: The following steps are involved: S1: Synchronously acquire PET and MR data in the free-breathing state while monitoring the respiratory motion amplitude; divide the respiratory cycle into several phases according to the respiratory motion amplitude, generate the corresponding five-tissue segmentation μ map, and form a dynamic μ map of several frames of respiratory cycles; S2: 3D SSFP is used for MR cardiac function cine sequence scanning, and each cardiac cycle is divided into several phases based on the RR interval of ECG gating; S3: Based on the MR signal differences between the myocardium, cardiac vessels, and lung tissue, myocardial segmentation is performed at each phase of the cardiac cycle. S4: Based on the five-tissue segmentation μ-map, the myocardium is separately segmented as the sixth tissue type; in several phases of the cardiac cycle in each respiratory cycle, μ-maps of the six-tissue segmentation are separately generated, and dynamic μ-maps of the six-tissue segmentation of several phases of the respiratory cycle and their corresponding several phases of the cardiac cycle are obtained; S5: Acquire 3D PET data. Each respiratory cycle is divided into several phases according to the respiratory motion amplitude. Each phase of the respiratory cycle is further divided into PET data of several phases of the cardiac cycle to obtain several clusters of raw PET data. S6: Using rigid body transformation, each cluster of PET data is registered with the dynamic μ map of the corresponding phase; S7: superimposing and recombining several clusters of PET data of the same time phase to form several frames of PET images that are not attenuated and arranged in time sequence; S8: First perform scatter correction on several frames of PET images, and then perform attenuation correction based on the corresponding μ map of six tissue segmentations; S9: Using the end-inspiratory phase PET image as a reference, register and superimpose the attenuation-corrected images of the same phase of the cardiac cycle within the respiratory cycle to obtain several frames of attenuation-corrected myocardial PET images that are not affected by respiratory motion; S10: after registering the acquired attenuation-corrected myocardial PET images with the MR structural images of the corresponding time phase, the myocardial PET images are corrected for partial volume effects based on the 3D MR cardiac cine structural images; S11: Using the end-diastolic MR cardiac structural image as a template, perform affine transformation and nonlinear transformation on the non-end-diastolic image to complete spatial standardization and save the transformation parameters; S12: performing spatial normalization transformation on the non-end-diastolic myocardial PET image according to the acquired conversion parameters, and superimposing and recombining the image with the end-diastolic myocardial PET image into a single myocardial PET image.

2. The method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on the PET / MR integrated machine according to claim 1, characterized in that: In step S1, multiple respiratory cycles are reassembled into respiratory motion movie images that are arranged and played back in time sequence through List-mode combined with segmented K-space acquisition, and the respiratory cycle is divided into 8 phases according to the respiratory motion amplitude, and a five-tissue segmentation μ map corresponding to each phase is generated. The five-tissue segmentation μ map includes water / soft tissue, fat, air, lung, and bone.

3. The method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on the PET / MR integrated machine according to claim 2, characterized in that: In step S2, compressed sensing 3D SSFP is used to perform MR cardiac function movie sequence scanning, the scanning adopts a single breath-hold or free breathing state, and each cardiac cycle is equally divided into 8 phases based on the RR interval of electrocardiographic gating.

4. The method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on the PET / MR integrated machine according to claim 3, characterized in that: In step S4, a six-tissue segmentation μ map is generated separately in each of the eight phases of the cardiac cycle in each respiratory cycle, thereby obtaining 64 six-tissue segmentation dynamic μ maps for the eight phases of the respiratory cycle and the corresponding eight phases of the cardiac cycle.

5. The method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on the PET / MR integrated machine according to claim 4, characterized in that: In step S5, 3D PET data is collected in list-mode. Each respiratory cycle is divided into 8 phases according to the respiratory motion amplitude. The phases of each respiratory cycle are further divided into PET data of 8 phases of the cardiac cycle, thereby obtaining 64 clusters of PET raw data of 8 phases of the respiratory cycle and 8 phases of the corresponding cardiac cycle.

6. The method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on the PET / MR integrated machine according to claim 5, characterized in that: In step S7, the 64 clusters of PET data of the same phase between different respiratory cycles and cardiac cycles are superimposed and recombined to form 64 frames of PET images of the same phase between different respiratory cycles and cardiac cycles, which are arranged in time sequence and have not been attenuated and arranged in time sequence; In step S8, after scatter correction is performed on the PET images arranged in time sequence, attenuation correction is performed based on the 64 μ maps of six tissue segmentations to obtain 64 frames of PET images of different respiratory and cardiac cycles that have been attenuated and arranged in time sequence.

7. The method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on the PET / MR integrated machine according to claim 6, characterized in that: In step S9, a rigid body transformation is used to align the myocardial images of each frame of the PET image in the same phase of the cardiac cycle after attenuation correction within the respiratory cycle with the PET image of the end-inspiratory phase of the respiratory cycle as a reference, and the data is superimposed and reorganized to obtain 8 frames of myocardial PET images that are attenuated and corrected and arranged in time sequence and are not affected by respiratory motion factors.

8. The method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on the PET / MR integrated machine according to claim 7, characterized in that: In step S10, a rigid body transformation is used to register the acquired PET cardiac images from eight phases of the cardiac cycle with the MR structural images from the corresponding phases to achieve spatial standardization and consistency between the myocardial PET and MR images. Based on the structural relationship between the myocardium and the surrounding cardiac chambers, great vessels, and lung tissue in the 3D MR cardiac cine structural images, the eight frames of attenuation-corrected myocardial PET images are corrected for partial volume effects.

9. The method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on the PET / MR integrated machine according to claim 8, characterized in that: In step S11, the 3D MR cardiac movie structural image at the non-end-diastolic phase of the cardiac cycle is spatially normalized with reference to the MR cardiac structural image at the end-diastolic phase, and the converted myocardial image is consistent in morphology, contour, and spatial coordinates; through spatial normalization transformation using affine transformation and nonlinear transformation, the converted MR image at the non-end-diastolic phase is precisely aligned with the MR template image at the end-diastolic phase in the standard space, and the corresponding conversion parameters are saved.

10. The method for dynamic attenuation correction and respiratory and cardiac dual motion correction based on the PET / MR integrated machine according to claim 9, characterized in that: In step S12, according to the conversion parameters obtained at different time phases, the myocardial PET image at the non-end-diastolic phase in the cardiac cycle is spatially standardized with reference to the myocardial PET image at the end-diastolic phase through spatial standardization conversion using affine transformation and nonlinear transformation, and the converted myocardial image is consistent in morphology, contour and spatial coordinates; finally, the standardized myocardial PET image at the non-end-diastolic phase and the myocardial PET image at the end-diastolic phase are superimposed and recombined into a single myocardial PET image.

Images