A PET / MRI image fusion method and system based on multimodal registration

Abstract

The present invention discloses a PET / MRI image fusion method and system based on multimodal registration, which relates to the technical field of image fusion. It includes: obtaining a target image fusion region to obtain a target region item, where the target region item is used to represent the PET image acquisition region and the MRI image acquisition region to be fused of a target patient; based on the target region item, setting at least two marked regions as marked feature points, and assigning labels to the marked feature points to obtain a labeled feature set. After splitting the acquired PET and MRI image data into data segments, the present invention quickly screens out target data segments through a selected method, avoiding the processing of a large amount of irrelevant data. The target data segments are screened out according to the position information of the labeled feature set, and their actual position data is used as an alignment contour for alignment and fusion, reducing the influence caused by the difference in chest contours due to breathing during the fusion process and improving the accuracy of image fusion.

Description

Technical Field

[0001] The present invention relates to the technical field of image fusion, and specifically provides a PET / MRI image fusion method and system based on multimodal registration. Background Technique

[0002] PET imaging is positron emission computed tomography, which is an important medical imaging technology mainly used to observe metabolic activities in the body. MRI imaging is magnetic resonance imaging, which uses a strong magnetic field and radio waves to obtain detailed images of internal structures, including steps such as the generation of a strong magnetic field, the application of radiofrequency pulses, the release of signals, the acquisition and processing of signals, and image reconstruction. The PET and MRI image fusion method is a technology that combines positron emission tomography and magnetic resonance imaging images to improve the diagnostic ability of medical imaging. By integrating the advantages of these two imaging technologies, structural and functional information can be obtained simultaneously.

[0003] For the image fusion method with the patent publication number CN113362261A, based on the first fusion image obtained in one fusion, color transfer can be used to perform secondary fusion, so that the brightness and color of the first fusion image can be transferred to the visible light image, and thus a second fusion image with brightness and color closer to the visible light image can be obtained. Therefore, using the second fusion image as the output result of image fusion can reduce the color distortion caused by the brightness difference in the infrared image in the fusion image.

[0004] When performing PET and MRI image fusion on the chest area of a patient for the above and similar technical solutions, it is necessary to first obtain PET and MRI image data of the chest area. However, during this process, the patient is in a continuous breathing state, and the chest will expand and contract autonomously, which will lead to certain differences in the obtained PET and MRI image data, that is, the degree of chest expansion is inconsistent. At this time, when fusing PET and MRI image data, the accuracy of the fusion result will be affected. Summary of the Invention

[0005] The purpose of the present invention is to provide a PET / MRI image fusion method and system based on multimodal registration to solve the problems raised in the above background technique.

[0006] To achieve the above purpose, the present invention provides the following technical solution: A PET / MRI image fusion method based on multimodal registration, including:

[0007] Obtain the target image fusion area to obtain a target area item, and the target area item is used to represent the area for obtaining the PET image to be fused and the area for obtaining the MRI image of the target patient;

[0008] Based on the target region item, set at least two marked regions as marked feature points, assign labels to the marked feature points to obtain a labeled feature set;

[0009] Respectively obtain the PET image data and MRI image data of the target patient to obtain a first data set and a second data set. The first data set and the second data set respectively represent the target region item PET and MRI image data of the target patient obtained;

[0010] Split the first data set and the second data set, and split the first data set and the second data set into at least two data segments to obtain a first segment set and a second segment set;

[0011] Respectively obtain the position information of the labeled feature set in the first segment set and the second segment set to obtain a first position set and a second position set;

[0012] Obtain the target data segments in the first position set and the second position set through a selected method to obtain a first target segment set and a second target segment set;

[0013] Use the actual position data of the labeled feature set in the first target segment set and the second target segment set in the first position set and the second position set as the alignment contours to obtain a first contour set and a second contour set;

[0014] Based on the alignment of the first contour set and the second contour set respectively, and then align the first target segment set and the second target segment set respectively, so that the data segments in the first target segment set and the second target segment set are fused according to the corresponding positions, and then at least two fused segments are obtained.

[0015] Furthermore, the acquisition method of the target region item includes:

[0016] Obtain the fusion requirement. The fusion requirement includes a position requirement to obtain a target position item, and the target position item is used to represent the actual position to be fused of the target patient;

[0017] Based on the target position item, set a limiting region. The limiting region is a fixed range value. Based on the combination result of the limiting region and the target position item, obtain a range to be fused starting from the target position item, and then obtain the target region item.

[0018] Furthermore, the acquisition method of the labeled feature set includes:

[0019] Set a marking gap. The marking gap is a fixed percentage value. Based on the combination result of the marking gap and the target region item, obtain at least two gap feature points;

[0020] Using the gap feature points as the marked feature points, a marked feature set is obtained. The marked feature set is sorted and numbered according to the top-down sorting method, and then a numbered feature set is obtained.

[0021] Furthermore, the obtaining methods of the first data set and the second data set include:

[0022] Set the acquisition duration to obtain the target duration item. Using the boundary positions of the target area item as the acquisition start point and the acquisition end point, the start position item and the end position item are obtained;

[0023] Based on the target duration item, using the start position item and the end position item as the acquisition endpoints, PET image data and MRI image data are respectively obtained through the first target device and the second target device, and then the first data set and the second data set are obtained.

[0024] Furthermore, the obtaining methods of the first segment set and the second segment set include:

[0025] Obtain the breathing frequency and breathing duration of the target patient to obtain the breathing frequency item and the breathing duration item;

[0026] Set the splitting threshold, and the splitting threshold is a fixed percentage value. Based on the splitting threshold, the breathing duration item is split to obtain at least two breathing split items;

[0027] Based on the breathing frequency item, using the frequency trough point of the target patient as the splitting start point and the frequency peak point as the splitting end point, the splitting range is obtained. The splitting range is split by the breathing split items to obtain at least one first split set and at least one second split set;

[0028] Based on the quantity information of the breathing frequency items in the first data set and the second data set, the combination result of the first split set and the second split set is obtained, and then the first segment set and the second segment set are obtained.

[0029] Furthermore, the obtaining methods of the first position set and the second position set include:

[0030] Based on the numbered feature set, reflective patches are respectively pasted, and the real-time position information of the reflective patches is tracked through the target tracking device to obtain the first real-time position set and the second real-time position set;

[0031] Based on the first segment set and the second segment set, the position information of the reflective patches corresponding to the first segment set and the second segment set in the first real-time position set and the second real-time position set is respectively obtained to obtain the first position set and the second position set.

[0032] Furthermore, the selection method includes:

[0033] Obtain data segments when the reflective patch is at the highest position and the lowest position in the first position set and the second position set, and obtain extremely high segment items and extremely low segment items;

[0034] Obtain data segments when the height difference of the reflective patch is the smallest in the first position set and the second position set, and obtain extremely small difference items;

[0035] Based on the respective corresponding first position set and second position set in the combination results of the extremely high segment items, extremely low segment items and extremely small difference items, obtain the first target segment set and the second target segment set.

[0036] Furthermore, the method for obtaining the first contour set and the second contour set includes:

[0037] Obtain the real-time position data when the label feature sets in the first target segment set and the second target segment set are respectively at the highest position, the lowest position and the smallest difference, and obtain the first extremely high position set, the first extremely low position set, the first extremely small difference position set, the second extremely high position set, the second extremely low position set, and the second extremely small difference position set;

[0038] The first extremely high position set corresponds one by one to the second extremely high position set, the first extremely low position set corresponds one by one to the second extremely low position set, and the first extremely small difference position set corresponds one by one to the second extremely small difference position set, to obtain a combined correspondence set, and use the real-time position data of the label feature set in the combined correspondence set as the vertical section respectively to obtain the first contour set and the second contour set.

[0039] Furthermore, a PET / MRI image fusion system based on multimodal registration uses the above-mentioned method for PET / MRI image fusion based on multimodal registration, and includes:

[0040] Acquisition module: Acquire the target image fusion area to obtain the target area item, based on the target area item, set at least two marked areas as marked feature points, assign labels to the marked feature points to obtain a label feature set, and respectively acquire the PET image data and the MRI image data of the target patient to obtain the first data set and the second data set. The first data set and the second data set respectively represent the PET and MRI image data of the target area item of the target patient obtained;

[0041] Processing module: Split the first data set and the second data set, split the first data set and the second data set into at least two data segments to obtain the first segment set and the second segment set, respectively acquire the position information of the label feature set in the first segment set and the second segment set to obtain the first position set and the second position set, and obtain the first target segment set and the second target segment set by using a selected method to acquire the target data segments in the first position set and the second position set;

[0042] Fusion module: Using the actual position data of the labeled feature sets in the first target fragment set and the second target fragment set that are in the first position set and the second position set as the alignment profiles, the first profile set and the second profile set are obtained. Based on the first profile set and the second profile set, alignment is performed respectively, and then the first target fragment set and the second target fragment set are aligned respectively, so that the data fragments in the first target fragment set and the second target fragment set are fused according to the corresponding positions, and at least two fused fragments are obtained.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] This PET / MRI image fusion method and system based on multimodal registration obtains target region items by acquiring the target image fusion region, clarifies the acquisition regions of the PET image and the MRI image to be fused, sets the marked regions and assigns labels to form a labeled feature set, which is convenient for accurately positioning key features in the image data. After splitting the acquired PET and MRI image data into data fragments, the data splitting operation splits the first data set and the second data set into multiple data fragments. After obtaining the position information of the labeled feature set, the target data fragments can be quickly screened out by a selected method, avoiding the processing of a large amount of irrelevant data. The target data fragments are screened out according to the position information of the labeled feature set, and aligned and fused with their actual position data as the alignment profile, reducing the influence caused by the difference in chest contours due to breathing during the fusion process, and improving the accuracy of image fusion. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0046] Figure 2 It is a schematic diagram of the target position item and the target region item of the present invention;

[0047] Figure 3 It is a schematic diagram of the labeled feature set of the present invention;

[0048] Figure 4 It is a schematic diagram of the first split set of the present invention;

[0049] Figure 5 It is a schematic diagram of the selected method of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0050] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0051] As important clinical imaging techniques, positron emission tomography (PET) and magnetic resonance imaging (MRI) play an irreplaceable role in disease diagnosis and treatment evaluation. PET can reflect tissue metabolic activities, while MRI provides high-resolution anatomical structure information. Through PET / MRI image fusion, the advantages of both can be comprehensively utilized, thereby improving the accuracy of diagnosis and the personalization of treatment. However, during the PET / MRI image fusion process in the thoracic region, the challenges brought about by the patient's respiratory movement cannot be ignored. The expansion and contraction of the chest cavity caused by respiratory movement result in significant differences in the PET and MRI image data obtained separately, which seriously affects the accuracy of image fusion. During inspiration, the intercostal muscles contract, the diaphragm descends, the chest cavity volume increases, and the lung tissue expands. During expiration, these muscles relax, the chest cavity volume decreases, and the lung tissue contracts. This dynamic change in the thoracic cavity morphology directly affects the acquisition of PET and MRI images. The existence of such differences makes it difficult to directly fuse PET and MRI images. Uncorrected fusion will lead to image blurring, overlapping errors, and positional deviations, thus unable to accurately correspond metabolic information with anatomical structure information. And a PET / MRI image fusion method based on multimodal registration provided by this application obtains a target region item by acquiring the target image fusion region, clarifies the acquisition regions of the PET image and the MRI image to be fused, sets a marked region and assigns labels to form a labeled feature set, which is convenient for accurately positioning key features in the image data. After splitting the acquired PET and MRI image data into data segments, the target data segments are selected according to the position information of the labeled feature set, and their actual position data is used as the alignment contour for alignment and fusion, reducing the influence caused by the chest contour difference due to breathing during the fusion process and improving the accuracy of image fusion. The data splitting operation splits the first data set and the second data set into multiple data segments. After obtaining the position information of the labeled feature set, the target data segments can be quickly selected by a selected method, avoiding the processing of a large amount of irrelevant data, such as Figure 1 as shown, including steps S100 - S800.

[0052] Step S100: Acquire the target image fusion region to obtain the target region item.

[0053] It should be noted that the target region item is used to represent the acquisition regions of the PET image and the MRI image to be fused for the target patient. The acquisition method of the target region item includes: obtaining the fusion requirement, where the fusion requirement includes the position requirement, to obtain the target position item, and the target position item is used to represent the actual position to be fused of the target patient; based on the target position item, set a limiting region, and the limiting region is a fixed range value, and the size of the fixed range value of the limiting region is ±20 cm. Based on the combination result of the limiting region and the target position item, obtain the fusion range starting from the target position item, and then obtain the target region item.

[0054] Example 1

[0055] In the specific implementation process, as Figure 2 shown, after a certain patient seeks medical treatment, according to the treatment results of medical staff, it is determined that the patient needs to undergo positron emission tomography and magnetic resonance imaging scans on the lungs, that is, it is necessary to obtain PET images and MRI images of the lung region. At this time, the position requirement is the lung region, and the target position item is obtained. Then, according to the fixed range value of the set limit region, with the lung region as the starting point, the fusion range of ±20 cm up and down is obtained. Furthermore, the target region item is a range that extends 20 cm up and down with the lung region as the center point, and the target region item is obtained.

[0056] Step S200: Based on the target region item, set at least two marked regions as marked feature points, and assign labels to the marked feature points to obtain a labeled feature set.

[0057] It should be noted that the method for obtaining the labeled feature set includes: setting a marking gap, the marking gap is a fixed percentage value, the marking gap is 20%, and based on the combined result of the marking gap and the target region item, six gap feature points are obtained; using the gap feature points as marked feature points to obtain a marked feature set, and sorting and labeling the marked feature set according to the top-down sorting method, and then obtaining the labeled feature set.

[0058] Example 2

[0059] In the specific implementation process, as Figure 3 shown, after a certain patient seeks medical treatment, according to the treatment results of medical staff, the target region item is a range that extends 20 cm up and down with the lung region as the center point, and the target region item is obtained. At this time, according to the set marking gap, six marked feature points are set in the target region item, which are the bottom position of the target region item, and 8 cm, 16 cm, 24 cm, 32 cm, and 40 cm away from the bottom position. The position of 40 cm is exactly the top position of the target region item, and a marked feature set is obtained. At this time, the marked feature set is sorted and labeled according to the top-down sorting method. The marked feature point at the topmost end, that is, the top position of the target region item, is No. 1, and so on. The bottom position of the target region item, and the positions 8 cm, 16 cm, 24 cm, and 32 cm away from the bottom position are labeled 6, 5, 4, 3, and 2 respectively, and then the labeled feature set is obtained.

[0060] Step S300: Obtain the PET image data and MRI image data of the target patient respectively to obtain a first data set and a second data set.

[0061] It should be noted that the first data set and the second data set respectively represent the PET and MRI image data of the target region of the target patient. The acquisition methods of the first data set and the second data set include: setting an acquisition duration, where the acquisition duration is 10 minutes to obtain a target duration item, using the boundary position of the target region item as the acquisition starting point and the acquisition ending point to obtain a starting position item and an ending position item; based on the target duration item, using the starting position item and the ending position item as acquisition endpoints, respectively obtaining PET image data and MRI image data through a first target device and a second target device. The first target device and the second target device are a PET scanner and an MRI scanner respectively, thereby obtaining the first data set and the second data set.

[0062] Step S400: Split the first data set and the second data set, splitting the first data set and the second data set into at least two data segments to obtain a first segment set and a second segment set.

[0063] It should be noted that the acquisition methods of the first segment set and the second segment set include: obtaining the breathing frequency and breathing duration of the target patient to obtain a breathing frequency item and a breathing duration item; setting a splitting threshold, where the splitting threshold is a fixed percentage value, and the splitting threshold is 10%. Based on the splitting threshold, split the breathing duration item to obtain ten breathing split items; based on the breathing frequency item, use the frequency trough point of the target patient as the splitting starting point and the frequency peak point as the splitting ending point to obtain a splitting range, and split the splitting range with the breathing split items to obtain at least one first split set and at least one second split set; based on the quantity information of the breathing frequency items in the first data set and the second data set, obtain the combination result of the first split set and the second split set, thereby obtaining the first segment set and the second segment set.

[0064] Embodiment III

[0065] In the specific implementation process, such as Figure 4As shown, the single breath duration of the target patient is obtained as 10 s. At this time, according to the set splitting threshold, the breath duration is split to obtain ten breath split items, which are 1 s, 2 s, 3 s, up to 10 s, a total of 10. Among the obtained breath frequencies of this patient, the period from 1 - 5 s is inhalation, and the period from 5 - 10 s is exhalation. At this time, the position corresponding to the frequency trough point at 10 s is used as the splitting starting point, and the position corresponding to the frequency peak point at 5 s is used as the splitting ending point to obtain the splitting range. At this time, the splitting range is split by the breath split items to obtain 5 first splitting sets and 5 second splitting sets, which are 5 s - 6 s, 6 s - 7 s, 7 s - 8 s, 8 s - 9 s, 9 s - 10 s. Since within the target duration item of 10 min, the number of breath frequency items in the first data set and the second data set is very large. There are 5 first splitting sets and 5 second splitting sets within 10 s, 30 first splitting sets and second splitting sets within 1 min, and 300 first splitting sets and second splitting sets within 10 min. Therefore, a combined result of multiple first splitting sets and second splitting sets is obtained, and then a first segment set and a second segment set are obtained.

[0066] Step S500: Respectively obtain the position information of the label feature sets in the first segment set and the second segment set to obtain a first position set and a second position set.

[0067] It should be noted that the methods for obtaining the first position set and the second position set include: Based on the label feature sets, reflective patches are respectively pasted, and the real - time position information of the reflective patches is tracked by a target tracking device. The target tracking device is an infrared sensor, and the positions of each reflective patch are monitored in real time through the infrared sensor to obtain a first real - time position set and a second real - time position set; Based on the first segment set and the second segment set, the position information of the reflective patches corresponding to the first segment set and the second segment set in the first real - time position set and the second real - time position set is respectively obtained to obtain a first position set and a second position set.

[0068] Step S600: Obtain the target data segments in the first position set and the second position set through a selected method to obtain a first target segment set and a second target segment set.

[0069] It should be noted that as Figure 5 shown, the selected method includes: Obtain the data segments when the reflective patches are at the highest position and the lowest position in the first position set and the second position set to obtain extremely high segment items and extremely low segment items; Obtain the data segments when the height difference of the reflective patches is the smallest in the first position set and the second position set to obtain extremely small difference items; Based on the first position set and the second position set corresponding to each of the combined results of the extremely high segment items, the extremely low segment items, and the extremely small difference items, obtain a first target segment set and a second target segment set.

[0070] Embodiment 4

[0071] In a specific implementation process, after a certain patient's medical treatment, according to the treatment results of medical staff, the target area item is a range that extends 20 cm upwards and downwards with the lung area as the center point, obtaining the target area item. At this time, according to the set marking gap, six marking feature points are set in the target area item, namely the bottom position of the target area item, and 8 cm, 16 cm, 24 cm, 32 cm, and 40 cm away from the bottom position, and the position of 40 cm is exactly the top position of the target area item. Reflective patches are pasted at the six marking feature points respectively, and the serial numbers are 1-6 from top to bottom. The positions of each reflective patch are monitored in real time through an infrared sensor. Taking No. 1 as the reference point, the data segments when the reflective patch is at the highest position and the lowest position in the first position set are respectively the 5th first split set at the 3rd minute and the 2nd first split set at the 5th minute: the highest position of No. 1 is 0, and the lowest position is -0.8 cm; the highest position of No. 2 is 1.2 cm, and the lowest position is -1.5 cm; the highest position of No. 3 is 2.8 cm, and the lowest position is -3.2 cm; the highest position of No. 4 is 2.1 cm, and the lowest position is -2.9 cm; the highest position of No. 5 is 0.9 cm, and the lowest position is -1.8 cm; the highest position of No. 6 is 0.3 cm, and the lowest position is -2.1 cm; the data segments when the reflective patch is at the highest position and the lowest position in the second position set are respectively the 4th second split set at the 6th minute and the 3rd second split set at the 4th minute: the highest position of No. 1 is 0, and the lowest position is -0.7 cm; the highest position of No. 2 is 1.3 cm, and the lowest position is -1.4 cm; the highest position of No. 3 is 2.9 cm, and the lowest position is -3.3 cm; the highest position of No. 4 is 2.3 cm, and the lowest position is -2.2 cm; the highest position of No. 5 is 0.5 cm, and the lowest position is -1.5 cm; the highest position of No. 6 is 0.2 cm, and the lowest position is -2.3 cm, obtaining the extremely high fragment item and the extremely low fragment item; the data segment when the height difference of the reflective patch is the smallest in the first position set and the second position set is the 3rd first split set and the second split set at the 6th minute, where No. 1 is 0 cm, No. 2 is 0 cm, No. 3 is 1 cm, No. 4 is 0 cm, No. 5 is 0 cm, and No. 6 is 0 cm, obtaining the extremely small difference item; based on the respective corresponding first position sets and second position sets in the combination results of the extremely high fragment item, the extremely low fragment item, and the extremely small difference item, the first target fragment set and the second target fragment set are obtained.

[0072] Step S700: Using the actual position data of the labeled feature sets in the first target fragment set and the second target fragment set in the first position set and the second position set as the alignment contours, the first contour set and the second contour set are obtained.

[0073] It should be noted that the methods for obtaining the first contour set and the second contour set include: obtaining the real-time position data of the label feature sets in the first target segment set and the second target segment set that are respectively at the highest position, the lowest position, and the position with the smallest difference, to obtain the first extremely high position set, the first extremely low position set, the first extremely small difference position set, the second extremely high position set, the second extremely low position set, and the second extremely small difference position set; the first extremely high position set and the second extremely high position set, the first extremely low position set and the second extremely low position set, and the first extremely small difference position set and the second extremely small difference position set respectively correspond one by one to obtain a combined corresponding set, and the real-time position data of the label feature sets in the combined corresponding set are respectively used as vertical profiles to obtain the first contour set and the second contour set.

[0074] Step S800: Align the first target segment set and the second target segment set respectively based on the alignment of the first contour set and the second contour set.

[0075] It should be noted that the data segments in the first target segment set and the second target segment set are fused according to the corresponding positions to obtain at least two fused segments. Since the first extremely high position set and the second extremely high position set, the first extremely low position set and the second extremely low position set, and the first extremely small difference position set and the second extremely small difference position set in the combined corresponding set respectively correspond one by one, the first contour set and the second contour set obtained at this time are three groups of contour data segments with the largest contour, the smallest contour, and the closest contour respectively, and three fused segments are obtained after fusion.

[0076] A PET / MRI image fusion system based on multimodal registration uses a method for PET / MRI image fusion based on multimodal registration as described above, and includes: an acquisition module: acquiring a target image fusion region to obtain a target region item, setting at least two marked regions as marked feature points based on the target region item, assigning labels to the marked feature points to obtain a labeled feature set, respectively acquiring PET image data and MRI image data of a target patient to obtain a first data set and a second data set, where the first data set and the second data set respectively represent the PET and MRI image data of the target region item of the target patient; a processing module: splitting the first data set and the second data set, splitting the first data set and the second data set into at least two data segments to obtain a first segment set and a second segment set, respectively acquiring the position information of the labeled feature set in the first segment set and the second segment set to obtain a first position set and a second position set, obtaining a first target segment set and a second target segment set by using a selected method to acquire the target data segments in the first position set and the second position set; a fusion module: using the actual position data of the labeled feature set in the first target segment set and the second target segment set in the first position set and the second position set as alignment contours to obtain a first contour set and a second contour set, respectively aligning based on the first contour set and the second contour set, and then respectively aligning the first target segment set and the second target segment set, so that the data segments in the first target segment set and the second target segment set are fused according to the corresponding positions, and at least two fused segments are obtained.

[0077] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended embodiments and their equivalents.

Claims

1. A PET / MRI image fusion method based on multimodal registration, comprising: Obtaining a target image fusion region to obtain a target region item, where the target region item is used to represent the PET image acquisition region and the MRI image acquisition region to be fused of a target patient; It is characterized in that: Based on the target region item, setting at least two marker regions as marker feature points, and assigning labels to the marker feature points to obtain a labeled feature set; Respectively obtaining the PET image data and the MRI image data of the target patient to obtain a first data set and a second data set, where the first data set and the second data set respectively represent the PET and MRI image data of the target region item of the target patient obtained; Splitting the first data set and the second data set, and splitting the first data set and the second data set into at least two data segments to obtain a first segment set and a second segment set; Respectively obtaining the position information of the labeled feature set in the first segment set and the second segment set to obtain a first position set and a second position set; Obtaining target data segments in the first position set and the second position set through a selected method to obtain a first target segment set and a second target segment set; Using the actual position data of the labeled feature set in the first target segment set and the second target segment set in the first position set and the second position set as alignment contours to obtain a first contour set and a second contour set; Aligning based on the first contour set and the second contour set respectively, and then aligning the first target segment set and the second target segment set respectively, so that the data segments in the first target segment set and the second target segment set are fused according to the corresponding positions, and then obtaining at least two fused segments.

2. The method for PET / MRI image fusion based on multimodal registration according to claim 1, wherein: The obtaining method of the target region item includes: Obtaining a fusion requirement, where the fusion requirement includes a position requirement, to obtain a target position item, and the target position item is used to represent the actual position to be fused of the target patient; Based on the target position item, setting a limit region, where the limit region is a fixed range value, and based on the combined result of the limit region and the target position item, obtaining a range to be fused starting from the target position item, and then obtaining the target region item.

3. A method for PET / MRI image fusion based on multimodal registration according to claim 1, characterized in that: The obtaining method of the labeled feature set includes: Setting a marker gap, where the marker gap is a fixed percentage value, and based on the combined result of the marker gap and the target region item, obtaining at least two gap feature points; Using the gap feature points as marker feature points to obtain a marker feature set, sorting and assigning labels to the marker feature set according to the sorting method from top to bottom, and then obtaining the labeled feature set.

4. A PET / MRI image fusion method based on multimodal registration according to claim 1, characterized in that: The obtaining method of the first data set and the second data set includes: Setting an acquisition duration to obtain a target duration item, using the boundary positions of the target region item as the acquisition start point and the acquisition end point to obtain a start position item and an end position item; Based on the target duration item, using the start position item and the end position item as the acquisition end points, and respectively obtaining the PET image data and the MRI image data through a first target device and a second target device, and then obtaining the first data set and the second data set.

5. A PET / MRI image fusion method based on multimodal registration according to claim 1, characterized in that: The obtaining method of the first segment set and the second segment set includes: Obtaining the breathing frequency and breathing duration of the target patient to obtain a breathing frequency item and a breathing duration item; Set a splitting threshold, which is a fixed percentage value. Based on the splitting threshold, split the respiratory duration item to obtain at least two respiratory split items; Based on the respiratory frequency item, use the frequency trough point of the target patient as the splitting starting point and the frequency peak point as the splitting ending point to obtain a splitting range. Use the respiratory split items to split the splitting range to obtain at least one first splitting set and at least one second splitting set; Based on the quantity information of the respiratory frequency items in the first data set and the second data set, obtain the combined result of the first splitting set and the second splitting set, and then obtain the first segment set and the second segment set.

6. A PET / MRI image fusion method based on multimodal registration according to claim 1, characterized in that: The obtaining methods of the first position set and the second position set include: Based on the label feature set, paste reflective patches respectively, and use the target tracking device to track the real-time position information of the reflective patches to obtain the first real-time position set and the second real-time position set; Based on the first segment set and the second segment set, obtain the position information of the reflective patches corresponding to the first segment set and the second segment set in the first real-time position set and the second real-time position set respectively to obtain the first position set and the second position set.

7. A PET / MRI image fusion method based on multimodal registration according to claim 1, characterized in that: The selection method includes: Obtain the data segments when the reflective patches are at the highest position and the lowest position in the first position set and the second position set to obtain the extremely high segment item and the extremely low segment item; Obtain the data segment when the height difference of the reflective patches in the first position set and the second position set is the smallest to obtain the extremely small difference item; Based on the respective corresponding first position set and second position set in the combined result of the extremely high segment item, the extremely low segment item and the extremely small difference item, obtain the first target segment set and the second target segment set.

8. A method for PET / MRI image fusion based on multimodal registration according to claim 1, characterized in that: The obtaining methods of the first contour set and the second contour set include: Obtain the real-time position data when the label feature set in the first target segment set and the second target segment set is at the highest position, the lowest position and the smallest difference respectively to obtain the first extremely high position set, the first extremely low position set, the first extremely small difference position set, the second extremely high position set, the second extremely low position set, and the second extremely small difference position set; The first extremely high position set and the second extremely high position set, the first extremely low position set and the second extremely low position set, and the first extremely small difference position set and the second extremely small difference position set correspond one by one respectively to obtain a combined corresponding set. Use the real-time position data of the label feature set in the combined corresponding set as the vertical section to obtain the first contour set and the second contour set.

9. A PET / MRI image fusion system based on multimodal registration, characterized in that: Use a PET / MRI image fusion method based on multimodal registration according to any one of claims 1-8, including: Acquisition module: Acquire the target image fusion area to obtain the target area item. Based on the target area item, set at least two marked areas as marked feature points, assign labels to the marked feature points to obtain the label feature set, and respectively acquire the PET image data and the MRI image data of the target patient to obtain the first data set and the second data set. The first data set and the second data set respectively represent the PET and MRI image data of the target area item of the target patient obtained; Processing module: Split the first data set and the second data set, split the first data set and the second data set into at least two data segments to obtain a first segment set and a second segment set, respectively obtain the position information of the label feature sets in the first segment set and the second segment set to obtain a first position set and a second position set, and obtain the target data segments in the first position set and the second position set through a selected method to obtain a first target segment set and a second target segment set; Fusion module: Use the actual position data of the label feature sets in the first target segment set and the second target segment set in the first position set and the second position set as the alignment profiles to obtain a first profile set and a second profile set, align based on the first profile set and the second profile set respectively, and then align the first target segment set and the second target segment set respectively, so that the data segments in the first target segment set and the second target segment set are fused according to the corresponding positions, and then at least two fusion segments are obtained.

Images