Image registration method, device and computer equipment

Abstract

The application relates to an image registration method, device and computer equipment. The method comprises the following steps: obtaining a to-be-registered image set containing at least one to-be-registered image; determining a reference image from the to-be-registered image set, and performing grouping processing on the to-be-registered image set based on the reference image to obtain at least one to-be-registered image subset of the to-be-registered image set; determining a representative image in each to-be-registered image subset, and determining a target image in each to-be-registered image subset except the representative image; and performing registration on the target image according to a first deformation field between the reference image and the representative image and a second deformation field between the representative image and the target image. The method can reduce registration error.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to an image registration method, apparatus and computer equipment. Background Technology

[0002] Dynamic magnetic resonance imaging (MRI) can continuously acquire three-dimensional volumes of the human body over a period of time, facilitating real-time assessment of lung volume at different stages of the respiratory cycle. It plays a crucial role in assessing lung motion and lung tumor mobility. To perform quantitative analysis of dynamic MRI images and obtain comprehensive evaluation results, registration of anatomical structures within the images is necessary.

[0003] Because dynamic magnetic resonance imaging (MRI) images exhibit similarity between adjacent time-series images, existing techniques typically involve selecting a first time-series image as a reference image to register a second time-series image, then using the second time-series image as a reference image to register a third time-series image, and so on, thereby achieving registration of the entire dynamic MRI image. However, this method suffers from the propagation and accumulation of registration errors, which can easily lead to significant registration errors in subsequent images.

[0004] Therefore, current dynamic magnetic resonance imaging technology suffers from a large registration error. Summary of the Invention

[0005] Therefore, it is necessary to provide an image registration method, apparatus, computer device, and computer-readable storage medium that can reduce registration errors in response to the above-mentioned technical problems.

[0006] Firstly, this application provides an image registration method. The method includes:

[0007] Obtain a set of images to be registered that contains at least one image to be registered;

[0008] A reference image is determined from the set of images to be registered, and based on the reference image, the set of images to be registered is grouped to obtain at least one subset of images to be registered.

[0009] Determine the representative image in each of the subsets of images to be registered, and in each subset of images to be registered, determine the images to be registered other than the representative images as the target images;

[0010] The target image is registered based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image.

[0011] In one embodiment, determining the reference image from the set of images to be registered includes:

[0012] Determine the first motion information corresponding to the set of images to be registered; the first motion information is the motion information of the target voxel on each of the images to be registered in the set of images to be registered;

[0013] First target information that meets the first preset conditions is determined from the first motion information;

[0014] The image to be registered corresponding to the first target information is determined as the reference image.

[0015] In one embodiment, the step of grouping the set of images to be registered based on the reference image to obtain at least one subset of images to be registered from the set of images to be registered includes:

[0016] Based on the reference image, the grouping interval is determined according to the maximum value in the first motion information and the preset number of groups;

[0017] Based on the grouping interval, the first motion information is grouped to obtain at least one set of motion information;

[0018] The image to be registered that corresponds to the first motion information in the motion information set is determined as the subset of images to be registered.

[0019] In one embodiment, determining representative images in each of the subsets of images to be registered includes:

[0020] Determine the second motion information corresponding to the subset of images to be registered; the second motion information is the motion information of the target voxel on each of the images to be registered in the subset of images to be registered.

[0021] From the second motion information, determine the second target information that meets the second preset conditions;

[0022] The image to be registered corresponding to the second target information is determined as the representative image.

[0023] In one embodiment, before registering the target image based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image, the method further includes:

[0024] Based on a preset mapping relationship, registration parameters associated with motion information are determined; the mapping relationship includes an inverse relationship between the motion information and the registration parameters; the motion information includes first motion information and second motion information, and the registration parameters include a first registration parameter associated with the first motion information and a second registration parameter associated with the second motion information;

[0025] Based on the first registration parameters, the reference image and the representative image are registered to obtain the first deformation field; and based on the second registration parameters, the representative image and the target image are registered to obtain the second deformation field.

[0026] In one embodiment, registering the target image based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image, includes:

[0027] The first deformation field and the second deformation field are superimposed to obtain a third deformation field between the reference image and the target image;

[0028] The target image is registered based on the third deformation field.

[0029] In one embodiment, after registering the target image based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image, the method further includes:

[0030] Obtain the segmentation template of the reference image;

[0031] The segmentation template of the reference image is deformed according to the third deformation field to obtain the segmentation template of the target image;

[0032] The target image is segmented according to the segmentation template of the target image.

[0033] In one embodiment, after obtaining the segmentation template of the reference image, the method further includes:

[0034] The segmentation template of the reference image is deformed according to the first deformation field to obtain the segmentation template of the representative image;

[0035] The representative image is segmented according to the segmentation template of the representative image.

[0036] Secondly, this application also provides an image registration apparatus. The apparatus includes:

[0037] The image acquisition module is used to acquire a set of images to be registered, which includes at least one image to be registered;

[0038] A reference image module is used to determine a reference image from the set of images to be registered, and based on the reference image, to perform grouping processing on the set of images to be registered to obtain at least one subset of images to be registered in the set of images to be registered;

[0039] The representative image module is used to determine the representative image in each subset of images to be registered, and to determine the images to be registered other than the representative images in each subset of images to be registered as the target images;

[0040] An image registration module is used to register the target image based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image.

[0041] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps:

[0042] Obtain a set of images to be registered that contains at least one image to be registered;

[0043] A reference image is determined from the set of images to be registered, and based on the reference image, the set of images to be registered is grouped to obtain at least one subset of images to be registered.

[0044] Determine the representative image in each of the subsets of images to be registered, and in each subset of images to be registered, determine the images to be registered other than the representative images as the target images;

[0045] The target image is registered based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image.

[0046] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps:

[0047] Obtain a set of images to be registered that contains at least one image to be registered;

[0048] A reference image is determined from the set of images to be registered, and based on the reference image, the set of images to be registered is grouped to obtain at least one subset of images to be registered.

[0049] Determine the representative image in each of the subsets of images to be registered, and in each subset of images to be registered, determine the images to be registered other than the representative images as the target images;

[0050] The target image is registered based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image.

[0051] The aforementioned image registration method, apparatus, computer equipment, and storage medium acquire a set of images to be registered, including at least one image to be registered; determine a reference image from the set of images to be registered; group the set of images to be registered based on the reference image to obtain at least one subset of images to be registered; determine a representative image in each subset of images to be registered; and determine the images to be registered other than the representative image in each subset of images to be registered as target images. The target images are registered based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image. This allows for the grouping of images to be registered based on a reasonably determined reference image, and the selection of a representative image from each group, resulting in smaller registration errors between the reference image and the representative image, and between the representative image in each group and the target image, effectively reducing the registration error of dynamic magnetic resonance images. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the result of selecting a reference image;

[0053] Figure 2 A schematic diagram of the selection result for another reference image;

[0054] Figure 3 This is a flowchart illustrating an image registration method in one embodiment;

[0055] Figure 4 This is a schematic diagram illustrating the extraction of spatiotemporal motion information of the diaphragm in one embodiment;

[0056] Figure 5 This is a schematic diagram illustrating the selection results of the reference image and the representative image in one embodiment;

[0057] Figure 6 This is a schematic diagram illustrating the relationship between registration parameters and motion information in one embodiment;

[0058] Figure 7 This is a flowchart illustrating an image segmentation method based on image registration in one embodiment;

[0059] Figure 8 This is a flowchart illustrating an image segmentation method based on image registration in another embodiment;

[0060] Figure 9 This is a flowchart illustrating an image segmentation method in one embodiment;

[0061] Figure 10 This is a structural block diagram of an image registration device in one embodiment;

[0062] Figure 11 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0064] Figure 1 A schematic diagram of the result of image selection in the prior art is provided. According to... Figure 1 For dynamic magnetic resonance imaging of the lungs within two respiratory cycles, image D can be randomly selected as the reference image (I). ref The image is then registered with other images AC and EN within the respiratory cycle. Generally, the greater the difference between the images, the more iterations are required for registration convergence, and the longer the registration process takes. For example, based on image D, the time required to register F is usually greater than the time required to register E. Therefore, the method of randomly selecting reference images can easily lead to a long computation time for the registration process when there are large potential differences between the images.

[0065] Figure 2 Another schematic diagram of the image selection result in the prior art is provided. According to... Figure 2 To reduce the computation time of the registration process, adjacent images can be registered sequentially using a reference image. For example, first select image A as the reference image, register image B, then use image B as the reference image, register image C, and so on. Since the differences between adjacent images are small, the number of iterations required for registration convergence is relatively small. However, for subsequent images, the propagation and accumulation of registration errors may be relatively large, for example... Figure 2 The deformation field of N may have a large registration error.

[0066] In one embodiment, such as Figure 3 As shown, an image registration method is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0067] Step S110: Obtain a set of images to be registered, which includes at least one image to be registered.

[0068] The image to be registered can be an image that needs to be registered, such as a two-dimensional magnetic resonance image or a three-dimensional magnetic resonance image.

[0069] The image set to be registered can be dynamic images that need to be registered, such as dynamic two-dimensional magnetic resonance images or dynamic three-dimensional magnetic resonance images.

[0070] In practice, a set of images to be registered can be input to the terminal, and the set of images to be registered contains at least one image to be registered.

[0071] In practical applications, dynamic magnetic resonance imaging (MRI) devices can be used to scan target objects, obtaining dynamic two-dimensional (2D) or dynamic three-dimensional (3D) MRI images. A dynamic 2D MRI image can be viewed as multiple 2D images in the time dimension, each corresponding to a specific acquisition time. The MRI device can input these images to a terminal, which uses the received images as a set of images to be registered, and selects multiple 2D images from the dynamic 2D MRI image set as the images to be registered. Similarly, a dynamic 3D MRI image can be viewed as multiple 3D images in the time dimension, each corresponding to a specific acquisition time. The MRI device can input these images to a terminal, which uses the received images as a set of images to be registered, and selects multiple 3D images from the dynamic 3D MRI image set as the images to be registered.

[0072] Step S120: Determine reference images from the set of images to be registered, and based on the reference images, group the set of images to be registered to obtain at least one subset of images to be registered.

[0073] The reference image can be an image selected from the set of images to be registered.

[0074] The subset of images to be registered can be a subset of the set of images to be registered.

[0075] In practice, reference images can be selected from the set of images to be registered based on preset selection rules. Alternatively, the set of images to be registered can be grouped according to preset grouping rules based on the selected reference images to obtain at least one subset of images to be registered.

[0076] In practical applications, taking dynamic two-dimensional magnetic resonance imaging (MRI) as an example, each pixel in the two-dimensional image corresponds to a magnetic resonance intensity value. A target point can be pre-selected on the target object, and multiple two-dimensional images in the dynamic MRI are sorted according to the order of acquisition time to determine the magnetic resonance intensity value of the target point in each two-dimensional image. By correlating the acquisition time of the two-dimensional image with the magnetic resonance intensity value of the target point, a motion curve is obtained, with the vertical axis representing the magnetic resonance intensity value and the horizontal axis representing the acquisition time. This motion curve reflects the movement of the target point at each acquisition time. A target intensity value is selected from all the magnetic resonance intensity values ​​corresponding to the vertical axis of the motion curve, and the two-dimensional image corresponding to the target intensity value is used as the reference image. The magnetic resonance intensity value closest to the median of all magnetic resonance intensity values ​​can be used as the target intensity value. Alternatively, all the magnetic resonance intensity values ​​corresponding to the vertical axis of the motion curve can be grouped to obtain at least one set of intensity values. The two-dimensional images corresponding to each magnetic resonance intensity value in a set of intensity values ​​can form a subset of images to be registered. Magnetic resonance intensity values ​​within a specified range can be grouped together.

[0077] Step S130: Determine the representative image in each subset of images to be registered, and in each subset of images to be registered, determine the images to be registered other than the representative image as the target image.

[0078] The representative image can be an image selected from a subset of images to be registered.

[0079] In practice, a representative image can be selected from the subset of images to be registered based on a preset selection rule, and the images other than the representative image in the subset of images to be registered can be used as the target image.

[0080] In practical applications, for the motion curve corresponding to the subset of images to be registered, a target intensity value can be selected from the corresponding magnetic resonance intensity values. The two-dimensional image corresponding to the target intensity value is used as the representative image, and the two-dimensional images corresponding to other magnetic resonance intensity values ​​are used as the target images. Among them, the magnetic resonance intensity value that is closest to the median value of the magnetic resonance intensity values ​​corresponding to the motion curve can be used as the target intensity value.

[0081] Step S140: Register the target image according to the first deformation field between the reference image and the representative image, and the second deformation field between the representative image and the target image.

[0082] The first deformation field can be the deformation field representing the image relative to the reference image.

[0083] The second deformation field can be the deformation field of the target image relative to the representative image.

[0084] In a specific implementation, the reference image can be registered with a representative image in the subset of images to be registered to obtain a first deformation field between the reference image and the representative image. Alternatively, for the subset of images to be registered, the representative image in the subset of images to be registered can be registered with a target image to obtain a second deformation field between the representative image and the target image. Then, based on the first and second deformation fields, the third deformation field of the target image relative to the reference image can be determined. Based on the reference image and the third deformation field, the target image can be registered.

[0085] In practical applications, the reference image and the representative image can be registered to obtain the deformation matrix W1 of the representative image relative to the reference image. Alternatively, the representative image and the target image can be registered in the subset of images to be registered to obtain the deformation matrix W2 of the target image relative to the representative image. By calculating W3 = W1 + W2, W3 can be determined as the deformation matrix of the target image relative to the reference image when registering the reference image and the target image.

[0086] Figure 4 A schematic diagram is provided for extracting spatiotemporal motion information of the diaphragm. Based on... Figure 4 Dynamic magnetic resonance images of the human lungs can be acquired under conditions of free breathing or guided breathing, resulting in a magnetic resonance image set {I(x,y,z,t); t=1,2,3,…}, where x and y represent two-dimensional images, z represents the magnetic resonance intensity value, and t represents the acquisition time of the two-dimensional image. Coronal slices containing the diaphragm apex p(x,y,z,t) are extracted from the magnetic resonance image set to obtain multiple coronal slices in the time dimension. A sagittal slice containing the diaphragm apex p(x,y,z,t) can also be determined. Within the orthogonal tangent plane between the sagittal slice and multiple coronal slices, the spatiotemporal motion information l(z,t) of the diaphragm apex p(x,y,z,t) in the craniocaudal direction is extracted.

[0087] Figure 5 A diagram showing the selection result of a reference image and a representative image is provided. For example... Figure 5As shown, the extracted spatiotemporal motion information l(z,t) of the diaphragm can be represented as a respiratory curve with the respiratory time t on the horizontal axis and the magnetic resonance intensity value z on the vertical axis. The respiratory curve contains 14 sampling points AN. Within the range of magnetic resonance intensity values ​​corresponding to the sampling point set {A,B,…,N}, the sample point A with the magnetic resonance intensity value closest to the median value can be selected, and the coronal slice corresponding to A can be determined as the reference image. Alternatively, based on the magnetic resonance intensity, the sampling points {B,…,N} other than A can be divided into 5 sampling point subsets, resulting in {F,L,M}, {E,G,N}, {K}, {D,H}, and {B,C,I,J}. The coronal slices corresponding to each sampling point in each sampling point subset can be combined into a magnetic resonance image subset. In each subset of sampling points, the sampling point whose magnetic resonance intensity value is closest to the median value of the subset is selected, and F, N, K, D, and C are obtained respectively. The coronal slices corresponding to F, N, K, D, and C can be the representative images of each subset. The coronal slices corresponding to {L,M}, {E,G}, {H}, and {B,I,J} are the target images of the 1st, 2nd, 4th, and 5th subsets, respectively. There is no target image in the 3rd subset. In the image registration method, the deformation field between the reference image and the representative image of each subset can be determined first. Then, for each subset, the deformation field between the representative image and the target image in the subset can be determined. The deformation fields between the reference image and the representative image, and between the representative image and the target image are superimposed to obtain the deformation field between the reference image and the target image. This deformation field is the deformation field obtained by registering the reference image and the target image.

[0088] The aforementioned image registration method involves acquiring a set of images to be registered, containing at least one image to be registered; determining a reference image from the set; grouping the set of images to be registered based on the reference image to obtain at least one subset of images to be registered; determining a representative image in each subset; and identifying the images to be registered other than the representative image in each subset as target images. The target images are then registered based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image. This method allows for the grouping of images to be registered based on a reasonably determined reference image, and the selection of a representative image from each group. This results in smaller registration errors between the reference image and the representative image, and between the representative image in each group and the target image, effectively reducing the registration error of dynamic magnetic resonance images.

[0089] In one embodiment, step S120 may specifically include: determining first motion information corresponding to the image set to be registered; the first motion information is the motion information of the target voxel on each image to be registered in the image set to be registered; determining first target information that meets the first preset condition from the first motion information; and determining the image to be registered corresponding to the first target information as the reference image.

[0090] The target voxel can be a voxel selected from the target object of the magnetic resonance scan, such as the apex of the diaphragm.

[0091] The first motion information can be the motion curve of the target voxel on each image to be registered in the image set to be registered, for example, the motion curve of the apex of the diaphragm.

[0092] The first preset condition can be that the median value of the first target information is closest to the median value of all first motion information.

[0093] In a specific implementation, a target voxel can be selected on the target object. Among the images to be registered in the image set, image data corresponding to the target voxel is determined. This image data is then used to form first motion information. Based on a first preset condition, first target information is selected from the first motion information, and the image to be registered corresponding to the first target information is determined as the reference image. Specifically, an intermediate value of the first motion information can be determined, and the first motion information closest to the intermediate value is selected as the first target information.

[0094] For example, according to Figure 4-5 The diaphragm apex can be selected as the target voxel. Fourteen coronal slices containing the diaphragm apex in the dynamic magnetic resonance imaging (MRI) image are used as images to be registered, and ordered according to the acquisition time. Alternatively, one sagittal slice containing the diaphragm apex can be determined. Within the orthogonal intersection plane of the coronal and sagittal slices, the MRI data of the diaphragm apex are used to form the first motion information l(z,t). The first motion information l(z,t) corresponds to 14 sampling points, with the horizontal axis representing time t and the vertical axis representing the magnetic resonance intensity value z, reflecting the motion of the diaphragm apex. The median value of the magnetic resonance intensity value is determined, and the sampling point A closest to the median value is selected as the first target information. The coronal slice corresponding to sampling point A is then used as the reference image. The specific formula for determining the reference image can be...

[0095] I(t)=I ref if l(t) = median(l{All})

[0096] Where All represents all first motion information, median represents the median value of the first motion information, and I ref This indicates the reference image that has been determined.

[0097] In this embodiment, by determining the first motion information corresponding to the set of images to be registered, the first target information that meets the first preset conditions is determined from the first motion information, and the image to be registered corresponding to the first target information is determined as the reference image. The reference image can be determined according to the motion of the target object, so that the determined reference image conforms to the motion of the target object, which helps to reduce the time of image registration and improve the efficiency of image registration.

[0098] In one embodiment, step S120 may further include: determining the grouping interval based on the reference image, according to the maximum value in the first motion information and the preset number of groups; grouping the first motion information according to the grouping interval to obtain at least one motion information set; and determining the image to be registered corresponding to the first motion information in the motion information set as a subset of images to be registered.

[0099] The grouping interval can be the difference between the maximum (or minimum) value of the first motion information in adjacent motion information sets.

[0100] In specific implementation, the maximum value in the first motion information can be determined, and the number of groups can be preset. According to the preset mapping relationship, the grouping interval is determined by the maximum value in the first motion information and the preset number of groups. All the first motion information is grouped according to the grouping interval to obtain at least one motion information set. For each motion information set, the image to be registered corresponding to each first motion information in the motion information set can be determined as a subset of images to be registered, thereby obtaining at least one subset of images to be registered.

[0101] For example, the maximum value l in the first motion information l(z,t) can be determined. max The number of groups N can also be preset, and the formula for calculating the group spacing Δz is:

[0102]

[0103] The first motion information l(z,t) is grouped according to Δz, with the height of each group being Δz, resulting in the following: Figure 5 The five motion information sets shown correspond to a subset of images to be registered. The formula for grouping processing can be...

[0104] I(t)∈S i if(i-1)(2Δz+1)≤l(t)≤i((2Δz+1)-1, i=1, 2, 3,...,N.

[0105] Among them, S i This represents all images to be registered in the i-th group.

[0106] It should be noted that the grouping interval can also be preset, and the number of groups for the first motion information can be determined based on the maximum value in the first motion information and the preset grouping interval.

[0107] Furthermore, the number of groups and the spacing between groups can be preset based on the reference image. For example, when the resolution of the reference image is low, fewer groups or larger spacing between groups can be set, and when the resolution of the reference image is high, more groups or smaller spacing between groups can be set.

[0108] In this embodiment, based on the reference image, the grouping interval is determined according to the maximum value in the first motion information and the preset number of groups. The first motion information is then grouped according to the grouping interval to obtain at least one set of motion information. The image to be registered corresponding to the first motion information in the set of motion information is determined as a subset of images to be registered. Grouping the images to be registered facilitates the subsequent selection of representative images from the subset of images to be registered, making the registration error between the reference image and the representative image, as well as the registration error between the representative image and the target image, smaller, thereby reducing the registration error of the entire dynamic image.

[0109] In one embodiment, step S130 may specifically include: determining second motion information corresponding to a subset of images to be registered; the second motion information being the motion information of a target voxel on each image to be registered in the subset of images to be registered; determining second target information that meets a second preset condition from the second motion information; and determining the image to be registered corresponding to the second target information as a representative image.

[0110] The second motion information can be the motion curve of the target voxel on each image to be registered in the subset of images to be registered, for example... Figure 5 The curves corresponding to {F,L,M}, {E,G,N}, {K}, {D,H}, and {B,C,I,J} are respectively.

[0111] The second preset condition can be that the median value of the second target information is closest to the median value of all the second motion information.

[0112] In specific implementation, for each subset of images to be registered, second motion information corresponding to each image to be registered in the subset can be selected from the first motion information. Based on a second preset condition, second target information is selected from the second motion information, and the image to be registered corresponding to the second target information is determined as the representative image. Specifically, an intermediate value of the second motion information can be determined, and the second motion information closest to the intermediate value is selected as the second target information.

[0113] For example, according to Figure 5For the subset {F,L,M}, the second motion information includes the curves corresponding to sampling points F, L, and M. The median value of the magnetic resonance intensity (MRI) values ​​corresponding to the subset {F,L,M} is determined. The sampling point F closest to this median value is selected from the sampling points F, L, and M as the second target information. Therefore, the image to be registered corresponding to sampling point F can be determined as the representative image of the subset {F,L,M}. The specific formula for determining the representative image can be...

[0114]

[0115] Among them, S i This represents all images to be registered in the i-th subset of images to be registered. It represents the representative image of the i-th subset of images to be registered.

[0116] In this embodiment, by determining the second motion information corresponding to the subset of images to be registered, the second target information that meets the second preset conditions is determined from the second motion information, and the image to be registered corresponding to the second target information is determined as the representative image. A representative image can be determined in each subset of images to be registered, so that the registration error between the reference image and the representative image, as well as the registration error between the representative image and the target image, are both small, thereby reducing the registration error of the entire dynamic image.

[0117] In one embodiment, prior to step S140, the method may further include: determining registration parameters associated with motion information based on a preset mapping relationship; the mapping relationship includes an inverse relationship between motion information and registration parameters; the motion information includes first motion information and second motion information; the registration parameters include a first registration parameter associated with the first motion information and a second registration parameter associated with the second motion information; registering a reference image and a representative image based on the first registration parameter to obtain a first deformation field; and registering a representative image and a target image based on the second registration parameter to obtain a second deformation field.

[0118] The registration parameter can be the iteration threshold of the loss function in the registration algorithm.

[0119] In practice, during the registration process, the mapping relationship between registration parameters and motion information can be preset. After the motion information is determined, the registration parameters associated with the motion information are determined according to the preset mapping relationship. Then, the registration algorithm can be executed using the registration parameters.

[0120] In practical applications, taking Demons (a registration algorithm) as an example, the deformation field between the reference image and the representative image is iteratively updated. The calculation formula can be:

[0121] r (n+1) =r (n) +dr(n+1) .

[0122] Where dr represents the displacement vector of the deformation field, r (n) Let r represent the nth deformation field during the iteration process. (n+1) Let represent the (n+1)th deformation field during the iteration process. Define the loss function.

[0123] loss (n) =∑|dr (n+1) | / ∑r (n) .

[0124] The registration algorithm stops when the difference between the loss functions of two consecutive iterations is less than a preset threshold e. The specific formula is as follows:

[0125] loss (n-1) -loss (n) ≤e.

[0126] Among these methods, the e-value can be pre-trained to be associated with the first or second motion information. For example, the e-value can be made to be inversely proportional to the motion information l(z,t), including linear or nonlinear inverse relationships, as well as negative logarithmic inverse relationships.

[0127] Figure 6 A schematic diagram illustrating the relationship between registration parameters and motion information is provided. Based on... Figure 6 As the relative similarity between two images increases, the threshold e required to complete the registration decreases. Taking the Demons registration algorithm as an example, using spatiotemporal motion information l(z,t) to represent image similarity, the result of fitting it with the threshold e is as follows: Figure 6 As shown, the fitting equation quantifies the relationship between the optimal threshold e and the image similarity based on the relative diaphragm displacement l, where e = 0.0343l. -2.705 .

[0128] In this embodiment, by determining the registration parameters associated with motion information according to a preset mapping relationship, registering the reference image and the representative image according to the first registration parameter to obtain a first deformation field, and registering the representative image and the target image according to the second registration parameter to obtain a second deformation field, the registration parameters can be associated with motion information. Furthermore, the registration parameters can be inversely proportional to the motion information, with smaller motion information corresponding to relatively larger registration parameters, reducing the number of iterations of the registration algorithm and thus reducing the processing time of the registration algorithm.

[0129] In one embodiment, step S140 may specifically include: superimposing the first deformation field and the second deformation field to obtain a third deformation field between the reference image and the target image; and registering the target image according to the third deformation field.

[0130] The third deformation field can be the deformation field of the target image relative to the reference image.

[0131] In practice, the first deformation field and the second deformation field can be superimposed to obtain the third deformation field of the target image relative to the reference image. Based on the reference image and the third deformation field, the target image can be registered.

[0132] In practical applications, we can first determine the reference image I. ref Representative images in the subset of images to be registered The first deformation field between Then, within the subset of images to be registered, determine the representative image. With target image I j The second deformation field between Adding the first deformation field to the second deformation field yields reference image I. ref With target image I j The third deformation field r(I) between ref →I j The specific formula for the above process can be:

[0133]

[0134]

[0135]

[0136] In this embodiment, a third deformation field is obtained by superimposing the first deformation field and the second deformation field. Based on the third deformation field, the target image is registered, which can quickly determine the deformation field between the reference image and the target image and improve the efficiency of registering the target image.

[0137] In one embodiment, after step S140, the process may further include: obtaining a segmentation template of the reference image; performing deformation processing on the segmentation template of the reference image according to a third deformation field to obtain a segmentation template of the target image; and performing segmentation processing on the target image according to the segmentation template of the target image.

[0138] The segmentation template can be a fixed graphic used for segmentation.

[0139] In practice, the reference image can be segmented to obtain a segmentation template for the reference image. Then, the segmentation template for the reference image can be deformed according to the third deformation field between the reference image and the target image to obtain a segmentation template for each target image. The segmentation template for the target image is then used to segment each target image to obtain the segmentation result of each target image.

[0140] For example, according to Figure 5 The coronal slice corresponding to image A is the reference image. In the five subsets {F,L,M}, {E,G,N}, {K}, {D,H}, and {B,C,I,J}, the coronal slices corresponding to F, N, K, D, and C are representative images, while the coronal slices corresponding to L, M, E, G, H, B, I, and J are target images. For the subset {F,L,M}, we can first determine the deformation matrix W1 between A and F, and then determine the deformation matrix W2 between F and L (or M). The deformation matrix between A and L (or M) is then W3 = W1 + W2. Segmenting the reference image A yields a segmentation template. The segmentation template of A is then deformed according to the deformation matrix W3 to obtain the segmentation template of the target image L (or M). Similar processing is applied to the target images in other subsets to obtain segmentation templates for each target image. These segmentation templates can then be used to segment each target image.

[0141] In this embodiment, by obtaining a segmentation template of the reference image, and performing deformation processing on the segmentation template of the reference image according to the third deformation field, a segmentation template of the target image is obtained. Based on the segmentation template of the target image, the target image is segmented. The segmentation template of the reference image can be transmitted to each target image with a relatively small registration error, so that each target image is segmented according to a more accurate segmentation template, thereby improving the accuracy of image segmentation.

[0142] Moreover, for the entire dynamic magnetic resonance image, only one segmentation of the reference image is required. The segmentation template of the target image can be obtained by passing the segmentation template, thereby realizing the segmentation of the entire dynamic magnetic resonance image and improving the efficiency of image segmentation.

[0143] In one embodiment, after obtaining the segmentation template of the reference image, the process may further include: deforming the segmentation template of the reference image according to a first deformation field to obtain a segmentation template representing the image; and segmenting the representative image according to the segmentation template representing the image.

[0144] In a specific implementation, the reference image can be segmented to obtain a segmentation template for the reference image. Then, the segmentation template for the reference image can be deformed according to the first deformation field between the reference image and the representative image to obtain the segmentation template for each representative image. The segmentation template for each representative image is then used to segment each representative image to obtain the segmentation result for each representative image.

[0145] For example, according to Figure 5Using the deformation matrix W1 between reference image A and representative image F, segment reference image A to obtain a segmentation template for A. Then, deform the segmentation template of A according to the deformation matrix W1 to obtain the segmentation template for representative image F. Perform similar processing on other representative images N, K, D, and C to obtain segmentation templates for representative images N, K, D, and C in sequence. Each representative image can be segmented according to the segmentation template.

[0146] In this embodiment, by deforming the segmentation template of the reference image according to the first deformation field, a segmentation template of the representative image is obtained. The representative image is then segmented according to the segmentation template of the representative image. This allows the segmentation template of the reference image to be transmitted to each representative image with a relatively small registration error, so that each representative image can be segmented according to a more accurate segmentation template, thereby improving the accuracy of image segmentation.

[0147] Moreover, for the entire dynamic magnetic resonance image, only one segmentation of the reference image is required. The representative image segmentation template can be obtained by passing the segmentation template, thereby achieving the segmentation of the entire dynamic magnetic resonance image and improving the efficiency of image segmentation.

[0148] To facilitate a deeper understanding of the embodiments of this application by those skilled in the art, a specific example will be used for illustration below.

[0149] Figure 7 A flowchart illustrating an image segmentation method based on image registration is provided in one embodiment. According to... Figure 7 For two-dimensional or three-dimensional images that change dynamically in the time dimension, such as dynamic two-dimensional magnetic resonance images or dynamic three-dimensional magnetic resonance images, after image acquisition, one two-dimensional or three-dimensional image can be selected as a reference image from the time dimension, and other two-dimensional or three-dimensional images in the time dimension can be used as target images. The reference image and each target image are then registered, for example, using three-dimensional deformable registration to obtain the deformation field of each target image relative to the reference image. The reference image can also be segmented to obtain a segmentation template. The segmentation template of the reference image is deformed using each deformation field, and then the segmentation template of the reference image is passed to each target image. Each target image can then be segmented using the deformed segmentation template. In this embodiment, since the dynamic scanning of dynamic magnetic resonance images typically has high temporal resolution, adjacent two-dimensional or three-dimensional images have a high degree of similarity in the time dimension. Based on image registration technology, the segmentation template of the reference image is passed to each target image, and the target image can be directly segmented according to the deformed segmentation template, without the need for separate segmentation of each target image, thus improving the efficiency of image segmentation.

[0150] Figure 8A flowchart illustrating an image segmentation method based on image registration is provided. According to... Figure 8 Taking dynamic magnetic resonance imaging of the lungs as an example, the specific steps may include:

[0151] Step S210: Continuously acquire dynamic magnetic resonance images of the lungs over a period of time. The breathing mode during the acquisition process can be free breathing or guided breathing.

[0152] Step S220: Extract the spatiotemporal motion information of the diaphragm. Extract the coronal slice containing the diaphragm apex p(x,y,z) from the dynamic magnetic resonance image to obtain a spatiotemporal three-dimensional (two-dimensional image + time dimension) image. Extract the motion image of the diaphragm apex p(x,y,z) in the craniotail direction through the orthogonal tangent plane between the sagittal plane of point p(x,y,z) and the spatiotemporal three-dimensional image. Extract the spatiotemporal motion information l(z,t) of the diaphragm apex p(x,y,z) in the craniotail direction from the motion image.

[0153] Step S230: Training the stopping condition for the registration algorithm. Taking the Demons registration algorithm as an example, this algorithm iteratively deforms the reference image by applying the displacement vector dr on a voxel-by-voxel basis. The overall deformation field r is iteratively updated, and the calculation formula can be...

[0154] r (n+1) =r (n) +dr (n+1) .

[0155] Define loss function

[0156] loss (n) =∑|dr (n+1) | / ∑r (n) .

[0157] Finally, the registration algorithm stops when the difference in the loss function between two consecutive iterations is less than a preset threshold e, which can be expressed as follows:

[0158] loss (n-1) -loss (n) ≤e.

[0159] Compared to traditional registration algorithms where the value of e remains constant within the same application, this application proposes to pre-train the relationship between the value of e and the spatiotemporal motion information l, so that when the value of l is small, e increases relatively, and the registration algorithm stops early to reduce unnecessary computation.

[0160] Step S240: Group the entire dynamic image set according to the spatiotemporal motion information. Each dynamic image within the dynamic image set is then classified according to the spatiotemporal motion information captured in step S220 above.

[0161] 1) Refer to the image:

[0162] I(t)=I rer if l(t) = median(l{All}).

[0163] 2) User-defined number of groups N, calculate the phase distance between each group of images.

[0164]

[0165] 3) Group the remaining images:

[0166] I(t)∈S i if(i-1)(2Δz+1)≤l(t)≤i((2Δz+1)-1, i=1, 2, 3,..., N

[0167] 4) Select representative images from each group of images:

[0168]

[0169] Step S250: Optimize the registration scheme. Perform a series of registrations and substitute the relationship between the e-value obtained from training in step S230 and the spatiotemporal motion information l:

[0170]

[0171]

[0172]

[0173] Step S260: Segment the reference image and use the deformation field r(I) obtained in step S250. ref →I) Pass the segmentation template to all target images for automatic matching.

[0174] It should be noted that the image segmentation method based on image registration described above can use a variety of different registration algorithms, requiring only retraining of the relationship between the registration algorithm's loss function threshold e and the spatiotemporal motion information l. Furthermore, the registration algorithm in the above method can be used independently without being combined with image segmentation; that is, steps S210-S250 can be executed independently, with step S260 being optional.

[0175] It should also be noted that the above image segmentation method based on image registration takes the four-dimensional magnetic resonance image of the lung as an example, but it can also be applied to two-dimensional dynamic magnetic resonance images with time, and can also be applied to other organs affected by respiratory motion, such as the magnetic resonance images of organs like the liver and spleen under free breathing.

[0176] The image segmentation method based on image registration described above utilizes the periodicity of images obtained along the respiratory cycle to directly extract spatiotemporal motion information from four-dimensional lung magnetic resonance images, selects reference images in the best way, and classifies other images into subsets based on spatiotemporal motion information, automatically generating an optimized registration scheme.

[0177] Furthermore, considering the characteristics of dynamic magnetic resonance imaging of the lungs, the registration tolerance parameter e is adaptively adjusted based on spatiotemporal motion information, which accelerates the registration convergence speed while ensuring registration accuracy.

[0178] Furthermore, the above method can be fully automatic after performing preliminary segmentation on the reference image to generate template volume, thus improving image segmentation efficiency.

[0179] In one embodiment, such as Figure 9 As shown, an image segmentation method is provided. Taking the application of this method to a terminal as an example, the method includes the following steps:

[0180] Step S310: Obtain a set of images to be registered, which includes at least one image to be registered;

[0181] Step S320: Determine reference images from the set of images to be registered, and based on the reference images, perform grouping processing on the set of images to be registered to obtain at least one subset of images to be registered.

[0182] Step S330: Determine the representative image in each subset of images to be registered, and in each subset of images to be registered, determine the images to be registered other than the representative image as the target image;

[0183] Step S340: The first deformation field between the reference image and the representative image and the second deformation field between the representative image and the target image are superimposed to obtain the third deformation field between the reference image and the target image.

[0184] Step S350: Obtain the segmentation template based on the reference image;

[0185] Step S360: Perform deformation processing on the segmentation template of the reference image according to the third deformation field to obtain the segmentation template of the target image, and perform segmentation processing on the target image according to the segmentation template of the target image;

[0186] Step S370: Perform deformation processing on the segmentation template of the reference image according to the first deformation field to obtain a segmentation template of the representative image; perform segmentation processing on the representative image according to the segmentation template of the representative image.

[0187] The aforementioned image segmentation method involves acquiring a set of images to be registered, containing at least one image to be registered; determining a reference image from the set; grouping the set of images to be registered based on the reference image to obtain at least one subset of images to be registered; determining a representative image for each subset; and identifying the images to be registered other than the representative image as target images within each subset. Finally, the method superimposes a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image, to obtain a third deformation field between the reference image and the target image. A segmentation template for the reference image is obtained. This template is then deformed using a third deformation field to obtain a segmentation template for the target image. The target image is then segmented using this template. A segmentation template for the representative image is then deformed using a first deformation field to obtain a segmentation template for the representative image. This template is then used to segment the representative image. The segmentation template for the reference image can be transferred to both the representative and target images with a relatively small registration error, resulting in more accurate segmentation templates for both images. This allows for accurate segmentation of both the representative and target images.

[0188] Moreover, for the entire dynamic magnetic resonance image, only one segmentation of the reference image is required. The segmentation templates representing the image and the target image can be obtained by passing the segmentation template, thereby achieving the segmentation of the entire dynamic magnetic resonance image and improving the efficiency of image segmentation.

[0189] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0190] Based on the same inventive concept, this application also provides an image registration apparatus for implementing the image registration method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more image registration apparatus embodiments provided below can be found in the limitations of the image registration method described above, and will not be repeated here.

[0191] In one embodiment, such as Figure 10 As shown, an image registration device is provided, including: an image acquisition module 410, a reference image module 420, a representative image module 430, and an image registration module 440, wherein:

[0192] Image acquisition module 410 is used to acquire a set of images to be registered, which includes at least one image to be registered;

[0193] The reference image module 420 is used to determine a reference image from the set of images to be registered, and based on the reference image, to perform grouping processing on the set of images to be registered to obtain at least one subset of images to be registered in the set of images to be registered.

[0194] The representative image module 430 is used to determine the representative image in each of the subsets of images to be registered, and to determine the images to be registered other than the representative images in each subset of images to be registered as the target images;

[0195] The image registration module 440 is used to register the target image based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image.

[0196] In one embodiment, the reference image module 420 is further configured to determine first motion information corresponding to the image set to be registered; the first motion information is the motion information of the target voxel on each of the images to be registered in the image set to be registered; determine first target information that meets the first preset condition from the first motion information; and determine the image to be registered corresponding to the first target information as the reference image.

[0197] In one embodiment, the reference image module 420 is further configured to determine the grouping interval based on the reference image, according to the maximum value in the first motion information and the preset number of groups; to group the first motion information according to the grouping interval to obtain at least one motion information set; and to determine the image to be registered corresponding to the first motion information in the motion information set as the subset of images to be registered.

[0198] In one embodiment, the aforementioned representative image module 430 is further configured to determine second motion information corresponding to the subset of images to be registered; the second motion information is the motion information of the target voxel on each of the images to be registered in the subset of images to be registered; determine second target information that meets the second preset conditions from the second motion information; and determine the image to be registered corresponding to the second target information as the representative image.

[0199] In one embodiment, the image registration device further includes:

[0200] The registration parameter determination module is used to determine registration parameters associated with motion information according to a preset mapping relationship; the mapping relationship includes an inverse proportional relationship between the motion information and the registration parameters; the motion information includes first motion information and second motion information, and the registration parameters include a first registration parameter associated with the first motion information and a second registration parameter associated with the second motion information;

[0201] The registration parameter registration module is used to register the reference image and the representative image according to the first registration parameter to obtain the first deformation field, and to register the representative image and the target image according to the second registration parameter to obtain the second deformation field.

[0202] In one embodiment, the image registration module 440 is further configured to superimpose the first deformation field and the second deformation field to obtain a third deformation field between the reference image and the target image; and to register the target image according to the third deformation field.

[0203] In one embodiment, the image registration device further includes:

[0204] A segmentation template acquisition module is used to acquire the segmentation template of the reference image;

[0205] The target image template delivery module is used to deform the segmentation template of the reference image according to the third deformation field to obtain the segmentation template of the target image;

[0206] The target image segmentation module is used to segment the target image according to the segmentation template of the target image.

[0207] In one embodiment, the image registration device further includes:

[0208] The representative image transmission module is used to deform the segmentation template of the reference image according to the first deformation field to obtain the segmentation template of the representative image;

[0209] The representative image segmentation module is used to segment the representative image according to the segmentation template of the representative image.

[0210] Each module in the aforementioned image registration device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0211] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 11 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements an image registration method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0212] Those skilled in the art will understand that Figure 11 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0213] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0214] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0215] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data shall comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0216] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0217] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0218] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. An image registration method, characterized in that, The method includes: Obtain a set of images to be registered that contains at least one image to be registered; The process further includes: determining a reference image from the set of images to be registered; grouping the set of images to be registered based on the reference image to obtain at least one subset of images to be registered; corresponding the acquisition time of the image to the magnetic resonance intensity value of the target point to obtain a motion curve with the magnetic resonance intensity value as the vertical axis and the acquisition time as the horizontal axis; selecting a target intensity value from all magnetic resonance intensity values ​​corresponding to the vertical axis of the motion curve; using the image corresponding to the target intensity value as the reference image; grouping all magnetic resonance intensity values ​​corresponding to the vertical axis of the motion curve to obtain at least one set of intensity values; and forming a subset of images to be registered by each magnetic resonance intensity value in the set of intensity values, wherein magnetic resonance intensity values ​​within a specified interval are grouped together. The method further includes: determining representative images in each subset of images to be registered, and determining the images to be registered other than the representative images in each subset of images to be registered as target images; for a motion curve corresponding to the subset of images to be registered, selecting a target intensity value from the corresponding magnetic resonance intensity values, using the image corresponding to the target intensity value as the representative image, and using the images corresponding to other magnetic resonance intensity values ​​as the target images. The target image is registered based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image.

2. The method according to claim 1, characterized in that, The step of determining the reference image from the set of images to be registered includes: Determine the first motion information corresponding to the set of images to be registered; the first motion information is the motion information of the target voxel on each of the images to be registered in the set of images to be registered; First target information that meets the first preset conditions is determined from the first motion information; The image to be registered corresponding to the first target information is determined as the reference image.

3. The method according to claim 2, characterized in that, The step of grouping the image set to be registered based on the reference image to obtain at least one subset of images to be registered, includes: Based on the reference image, the grouping interval is determined according to the maximum value in the first motion information and the preset number of groups; Based on the grouping interval, the first motion information is grouped to obtain at least one set of motion information; The image to be registered that corresponds to the first motion information in the motion information set is determined as the subset of images to be registered.

4. The method according to claim 3, characterized in that, The step of determining representative images in each subset of images to be registered includes: Determine the second motion information corresponding to the subset of images to be registered; the second motion information is the motion information of the target voxel on each of the images to be registered in the subset of images to be registered. From the second motion information, determine the second target information that meets the second preset conditions; The image to be registered corresponding to the second target information is determined as the representative image.

5. The method according to claim 4, characterized in that, Before registering the target image based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image, the method further includes: Based on a preset mapping relationship, registration parameters associated with motion information are determined; the mapping relationship includes an inverse relationship between the motion information and the registration parameters; the motion information includes first motion information and second motion information, and the registration parameters include a first registration parameter associated with the first motion information and a second registration parameter associated with the second motion information; Based on the first registration parameters, the reference image and the representative image are registered to obtain the first deformation field; and based on the second registration parameters, the representative image and the target image are registered to obtain the second deformation field.

6. The method according to claim 1, characterized in that, The registration of the target image based on the first deformation field between the reference image and the representative image, and the second deformation field between the representative image and the target image, includes: The first deformation field and the second deformation field are superimposed to obtain a third deformation field between the reference image and the target image; The target image is registered based on the third deformation field.

7. The method according to claim 6, characterized in that, After registering the target image based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image, the process further includes: Obtain the segmentation template of the reference image; The segmentation template of the reference image is deformed according to the third deformation field to obtain the segmentation template of the target image; The target image is segmented according to the segmentation template of the target image.

8. The method according to claim 7, characterized in that, After obtaining the segmentation template of the reference image, the process further includes: The segmentation template of the reference image is deformed according to the first deformation field to obtain the segmentation template of the representative image; The representative image is segmented according to the segmentation template of the representative image.

9. An image registration device, characterized in that, The device includes: The image acquisition module is used to acquire a set of images to be registered, which includes at least one image to be registered; A reference image module is used to determine a reference image from the set of images to be registered, and based on the reference image, to perform grouping processing on the set of images to be registered to obtain at least one subset of images to be registered in the set of images to be registered; The reference image module is further configured to correlate the image acquisition time with the magnetic resonance intensity value of the target point to obtain a motion curve with the magnetic resonance intensity value as the vertical axis and the acquisition time as the horizontal axis. A target intensity value is selected from all magnetic resonance intensity values ​​corresponding to the vertical axis of the motion curve. The image corresponding to the target intensity value is used as the reference image. All magnetic resonance intensity values ​​corresponding to the vertical axis of the motion curve are grouped to obtain at least one set of intensity values. The image corresponding to each magnetic resonance intensity value in a set of intensity values ​​forms a subset of the images to be registered. Magnetic resonance intensity values ​​within a specified interval range are grouped together. The representative image module is used to determine the representative image in each subset of images to be registered, and to determine the images to be registered other than the representative images in each subset of images to be registered as the target images; The representative image module is further configured to select a target intensity value from the corresponding magnetic resonance intensity values ​​for the motion curve corresponding to the subset of images to be registered, use the image corresponding to the target intensity value as the representative image, and use the images corresponding to other magnetic resonance intensity values ​​as the target images. An image registration module is used to register the target image based on a first deformation field between the reference image and the representative image, and a second deformation field between the representative image and the target image.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

Images