Medical image registration method and device, computer equipment and storage medium

Abstract

The invention relates to a medical image registration method and device, computer equipment, a storage medium and a computer program product, relates to the technical field of medical image processing, and can improve the internal registration accuracy of a target object. The method comprises the following steps: acquiring a reference medical image containing a target object in different periods and a medical image to be registered; according to the reference medical image and the to-be-registered medical image, obtaining a plurality of medical image block pairs containing local parts of the target object; each medical image block pair comprises a first image block representing the local part of the target object in the reference medical image and a second image block corresponding to the first image block in the to-be-registered medical image; for each medical image block pair, performing edge registration according to the edge of the target object in the first image block and the edge of the target object in the second image block, and determining a local deformation field corresponding to the local part of the target object; and performing image registration on the reference medical image and the to-be-registered medical image according to each local deformation field.

Description

Technical Field

[0001] The present application relates to the technical field of medical image processing, and in particular to a medical image registration method, apparatus, computer equipment, storage medium and computer program product. Background Art

[0002] With the development of computer technology, medical imaging equipment can be used to assist in the preoperative and intraoperative treatment of interventional surgery. Specifically, CT enhanced scanning can be performed before surgery, and the vascular conditions or other information inside the target object can be understood based on the enhanced images, and a surgical plan can be formulated; during the operation, CT plain scans are mainly performed, and the surgical method is further confirmed based on the acquired plain scan images. In order to better utilize the preoperative enhanced scan images, the enhanced scan images can be registered with the plain scan images acquired during the operation.

[0003] In the related art, the enhanced image and the plain scan image of the target object can be registered based on an image registration method or a finite element simulation method.

[0004] However, the former can only effectively register the external boundary of the target object, and the registration accuracy of the internal vessels cannot be guaranteed; although the latter can perform image registration on the inside of the target object, the credibility of the registration result is low. It can be seen that the relevant technology is difficult to achieve accurate image registration on the inside of the target object in the medical image. Summary of the invention

[0005] Based on this, it is necessary to provide a medical image registration method, apparatus, computer equipment, computer-readable storage medium and computer program product that can improve the accuracy of internal registration of a target object in order to address the above technical problems.

[0006] In a first aspect, the present application provides a medical image registration method, comprising:

[0007] Acquire a reference medical image and a to-be-registered medical image containing a target object at different periods;

[0008] Acquire a plurality of medical image block pairs containing a local portion of the target object according to the reference medical image and the medical image to be registered; each of the medical image block pairs comprises a first image block in the reference medical image representing the local portion of the target object, and a second image block in the medical image to be registered corresponding to the first image block;

[0009] For each of the medical image block pairs, performing edge registration according to an edge of the target object in the first image block and an edge of the target object in the second image block, and determining a local deformation field corresponding to a part of the target object;

[0010] Image registration is performed on the reference medical image and the medical image to be registered according to each of the local deformation fields.

[0011] In one of the embodiments, acquiring a plurality of medical image block pairs including a part of the target object according to the reference medical image and the medical image to be registered includes:

[0012] Performing a first rigid registration on the reference medical image and the target object in the medical image to be registered;

[0013] According to a predetermined number of image blocks, the target object in the first rigid registration result and the target object in the reference medical image are respectively subjected to block processing; the number of image blocks is determined from the plurality of candidate block numbers according to the registration accuracies corresponding to the respective numbers of candidate block numbers;

[0014] A plurality of medical image block pairs are obtained according to the plurality of image blocks obtained after the block processing.

[0015] In one embodiment, the number of image blocks is determined by the following steps:

[0016] Acquire sample reference medical images and sample to-be-registered medical images of the target object at different time periods;

[0017] According to the sample reference medical image and the sample medical image to be registered, a plurality of sample local deformation fields corresponding to the number of candidate blocks are obtained, and according to the plurality of sample local deformation fields, image registration is performed on the sample reference medical image and the sample medical image to be registered to obtain a sample medical image registration result;

[0018] Determining the image registration accuracy corresponding to the number of candidate blocks according to the reference point registration accuracy; the reference point registration accuracy is used to characterize the registration accuracy corresponding to the reference point in the sample medical image registration result;

[0019] The number of candidate blocks is updated, and among the plurality of candidate block numbers, the number of candidate blocks whose image registration accuracy satisfies a preset accuracy condition is determined as the number of image blocks.

[0020] In one of the embodiments, obtaining a plurality of sample local deformation fields corresponding to the number of candidate blocks according to the sample reference medical image and the sample to-be-registered medical image includes:

[0021] Determine a direction to be adjusted; the direction to be adjusted is one of a plurality of image block directions;

[0022] In the direction to be adjusted, the target object in the sample reference medical image and the sample medical image to be registered is divided into blocks according to the number of candidate blocks, and in the direction other than the direction to be adjusted, the target object in the sample reference medical image and the sample medical image to be registered is divided into blocks according to a preset number of blocks, and a plurality of sample local deformation fields are determined according to the obtained medical image blocks;

[0023] The step of determining, from among the plurality of candidate block numbers, the number of candidate block numbers whose image registration accuracy satisfies a preset accuracy condition as the number of image block numbers comprises:

[0024] Among the multiple candidate block numbers, the candidate block number whose image registration accuracy meets the preset registration accuracy condition is determined as the image block number in the direction to be adjusted, and the current direction to be adjusted is updated until the image block number in each image block direction is determined.

[0025] In one embodiment, the step of obtaining a plurality of medical image block pairs according to a plurality of image blocks obtained after block processing includes:

[0026] Determine a plurality of initial image blocks including the local part of the target object obtained after the block processing; the plurality of initial image blocks include initial image blocks of the reference medical image and the medical image to be registered;

[0027] If the initial image block includes the edge of the target object, the edge of the initial image block is expanded according to a preset edge expansion width to obtain a target image block after edge expansion;

[0028] If the initial image block does not include the edge of the target object, performing image block edge expansion on the initial image block until a portion of the edge of the target object is covered to obtain a target image block after edge expansion;

[0029] A plurality of medical image block pairs are obtained according to the target image blocks that are paired with the reference medical image and the medical image to be registered.

[0030] In one embodiment, performing image registration on the reference medical image and the medical image to be registered according to each of the local deformation fields includes:

[0031] Determining a rigid deformation field of the target object according to the first rigid registration result;

[0032] Acquire a fused deformation field according to each of the local deformation fields, and obtain a composite deformation field of the target object according to a composite result of the rigid deformation field of the target object and the fused deformation field;

[0033] Image registration is performed on the target object in the reference medical image and the target object in the medical image to be registered according to the composite deformation field.

[0034] In one embodiment, the method further comprises:

[0035] Performing a second rigid registration on the reference medical image and the medical image to be registered, determining an initial global deformation field according to the second rigid registration result, and fusing the initial global deformation field and the composite deformation field to obtain a target global deformation field;

[0036] According to the target global deformation field, the reference medical image and the region outside the target object in the medical image to be registered are registered to obtain a medical image after rigid registration.

[0037] In one of the embodiments, if it is determined to perform the first global elastic registration, after registering the reference medical image and the region outside the target object in the medical image to be registered according to the target global deformation field to obtain a medical image after rigid registration, the method further includes:

[0038] The deformation field of the target object in the rigidly registered medical image is kept unchanged, and a first global elastic registration is performed on the rigidly registered medical image and the reference image.

[0039] In one embodiment, if it is determined to perform a second global elastic registration, then determining the initial global deformation field according to the second rigid registration result includes:

[0040] A second global elastic registration is performed on the medical image obtained after the second rigid registration, and an initial global deformation field is determined according to the medical image obtained after the second global elastic registration.

[0041] In one embodiment, obtaining the composite deformation field of the target object according to the composite result of the rigid deformation field of the target object and the fused deformation field includes:

[0042] Performing elastic registration on the target object in the reference medical image and the target object in the medical image to be registered, and obtaining an elastic deformation field corresponding to the target object according to the elastic registration result;

[0043] The elastic deformation field, the rigid deformation field and the fused deformation field are compounded to obtain a compound deformation field of the target object.

[0044] In one of the embodiments, before acquiring a plurality of medical image block pairs according to the reference medical image and the medical image to be registered, the method further includes:

[0045] Performing edge recognition on the target object in the reference medical image and the medical image to be registered to obtain a first edge of the target object in the reference medical image and a second edge in the medical image to be registered;

[0046] The performing edge registration according to the edge of the target object in the first image block and the edge of the target object in the second image block comprises:

[0047] Edge registration is performed according to a first edge of the target object in the first image block and a second edge of the target object in the second image block.

[0048] In one embodiment, obtaining a reference medical image and a to-be-registered medical image of a target object at different time periods includes:

[0049] Acquire a first enhanced scan image and a second enhanced scan image of the target object at different periods, use the first enhanced scan image as the medical image to be registered, and perform plain scan image simulation processing on the second enhanced scan image to obtain a reference medical image;

[0050] After performing medical image registration on the reference medical image and the medical image to be registered according to each of the local deformation fields, the method further includes:

[0051] Acquire medical image registration parameters for registering the medical image to be registered to the reference medical image according to medical image registration results corresponding to the reference medical image and the medical image to be registered;

[0052] The reference points pre-marked in the first enhanced scan image are deformed according to the medical image registration parameters, and image registration evaluation information is determined according to the distance between the deformed blood vessel bifurcation point and the paired reference point in the second enhanced scan image.

[0053] In a second aspect, the present application also provides a medical image registration device, comprising:

[0054] A medical image acquisition module, used for acquiring reference medical images of target objects at different periods and medical images to be registered;

[0055] A medical image block acquisition module, configured to acquire a plurality of medical image block pairs containing a local portion of the target object according to the reference medical image and the medical image to be registered; each of the medical image block pairs comprises a first image block in the reference medical image representing a local portion of the target object, and a second image block in the medical image to be registered corresponding to the first image block;

[0056] A local deformation field acquisition module, configured to perform edge registration for each pair of medical image blocks according to an edge of the target object in the first image block and an edge of the target object in the second image block, and determine a local deformation field corresponding to a part of the target object;

[0057] A registration module is used to perform image registration on the reference medical image and the medical image to be registered according to each of the local deformation fields.

[0058] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0059] Acquire a reference medical image and a to-be-registered medical image containing a target object at different periods;

[0060] Acquire a plurality of medical image block pairs containing a local portion of the target object according to the reference medical image and the medical image to be registered; each of the medical image block pairs comprises a first image block in the reference medical image representing the local portion of the target object, and a second image block in the medical image to be registered corresponding to the first image block;

[0061] For each of the medical image block pairs, performing edge registration according to an edge of the target object in the first image block and an edge of the target object in the second image block, and determining a local deformation field corresponding to a part of the target object;

[0062] Image registration is performed on the reference medical image and the medical image to be registered according to each of the local deformation fields.

[0063] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0064] Acquire a reference medical image and a to-be-registered medical image containing a target object at different periods;

[0065] Acquire a plurality of medical image block pairs containing a local portion of the target object according to the reference medical image and the medical image to be registered; each of the medical image block pairs comprises a first image block in the reference medical image representing the local portion of the target object, and a second image block in the medical image to be registered corresponding to the first image block;

[0066] For each of the medical image block pairs, performing edge registration according to an edge of the target object in the first image block and an edge of the target object in the second image block, and determining a local deformation field corresponding to a part of the target object;

[0067] Image registration is performed on the reference medical image and the medical image to be registered according to each of the local deformation fields.

[0068] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0069] Acquire a reference medical image and a to-be-registered medical image containing a target object at different periods;

[0070] Acquire a plurality of medical image block pairs containing a local portion of the target object according to the reference medical image and the medical image to be registered; each of the medical image block pairs comprises a first image block in the reference medical image representing the local portion of the target object, and a second image block in the medical image to be registered corresponding to the first image block;

[0071] For each of the medical image block pairs, performing edge registration according to an edge of the target object in the first image block and an edge of the target object in the second image block, and determining a local deformation field corresponding to a part of the target object;

[0072] Image registration is performed on the reference medical image and the medical image to be registered according to each of the local deformation fields.

[0073] The above-mentioned medical image registration method, apparatus, computer equipment, storage medium and computer program product can obtain a reference medical image and a medical image to be registered containing a target object at different periods, and then can obtain multiple medical image block pairs containing a part of the target object based on the reference medical image and the medical image to be registered, wherein each medical image block pair includes a first image block in the reference medical image representing a part of the target object, and a second image block in the medical image to be registered corresponding to the first image block; for each medical image block pair, edge registration is performed based on the edge of the target object in the first image block and the edge of the target object in the second image block to determine the local deformation field corresponding to the part of the target object; and image registration is performed on the reference medical image and the medical image to be registered based on each local deformation field. In the present application, by dividing the target object in the reference medical image and the medical image to be registered into multiple parts and then performing edge registration, multiple local deformation fields that accurately reflect the edge deformation path of the target object can be obtained. Then, based on the correlation between the internal movement of the target object and the edge movement of the target object, with the help of multiple local deformation fields, the relevant image information inside the target object in the medical image to be registered is registered to the corresponding position on the reference medical image, thereby improving the registration accuracy of the target object during the image registration process. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0075] Figure 1 A diagram of an application environment of a medical image registration method in one embodiment;

[0076] Figure 2 A schematic diagram of a process of a medical image registration method in an embodiment;

[0077] Figure 3a is a schematic diagram of an edge registration in one embodiment;

[0078] Figure 3b is a schematic diagram of another edge registration in one embodiment;

[0079] Figure 4 A schematic flow chart of a step of acquiring multiple medical image block pairs in one embodiment;

[0080] Figure 5 is a schematic diagram of an image segmentation in one embodiment;

[0081] Figure 6a is a Gaussian weight map in one embodiment;

[0082] Figure 6b is a schematic diagram of a fused deformation field obtained based on a Gaussian weight map in one embodiment;

[0083] Figure 7 is a flow chart of another medical image registration method in one embodiment;

[0084] Figure 8 is a schematic diagram of a medical image in one embodiment;

[0085] Fig. 9 is a schematic diagram of annotating a medical image in one embodiment;

[0086] Fig.10 is a flow chart of another medical image registration method in one embodiment;

[0087] Fig.11 is a structural block diagram of a medical image registration device in one embodiment;

[0088] Fig.12 The figure is a diagram of the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0089] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0090] With the development of computer technology, medical imaging equipment can be used to assist in the preoperative and intraoperative treatment of interventional surgery. Specifically, CT enhanced scanning can be performed before surgery, and the vascular conditions or other information inside the target object can be understood based on the enhanced images, and a surgical plan can be formulated; during the operation, CT plain scans are mainly performed, and the surgical method is further confirmed based on the acquired plain scan images. In order to better utilize the preoperative enhanced images, the enhanced images can be registered with the plain scan images acquired during the operation.

[0091] In the related art, the enhanced image and the plain scan image of the target object can be registered based on an image registration method or a finite element simulation method.

[0092] Taking the image registration of the liver as the target object as an example, the image registration method usually registers the image according to the grayscale characteristics of the image. For the plain scan image of the liver, usually only the boundary of the liver can be identified. That is to say, the boundary of the liver can be registered by elastic methods, but the internal vascular registration cannot be guaranteed.

[0093] The registration based on finite element simulation is relatively in line with the actual situation. It actually considers the deformation of the liver and its impact on the intrahepatic blood vessels by segmenting the surface of the liver, modeling the liver tissue with appropriate materials, and considering the influence of external forces (such as gravity, breathing, and between organs). However, external forces are generally still driven by deformation fields based on traditional registration and cannot be truly obtained. Therefore, the credibility of the modeling of the entire process and the credibility of the final image registration results are low. In addition, the finite element method often faces the problem of overly complex solutions and long solution times, and even often fails to converge, making it difficult to meet the needs of medical image registration in clinical scenarios.

[0094] Based on this, the present application provides a medical image registration method to at least solve the problem of inaccurate internal registration of the target object existing in the related art.

[0095] The medical image registration method provided in the embodiment of the present application can be applied to Figure 1In the application environment shown, the application environment includes a terminal and a server, and the terminal communicates with the server through a network. The terminal may include a medical image acquisition device and / or a medical image storage device, such as a computer tomography (CT) device, a positron emission computed tomography (PET) device, or one or more types of medical image scanning devices. The server may be implemented as an independent server or a server cluster consisting of multiple servers.

[0096] In an exemplary embodiment, Figure 2 As shown, a medical image registration method is provided, which is applied to Figure 1 Taking the server in as an example, the method includes the following steps S201 to S204.

[0097] S201, obtaining reference medical images and medical images to be registered containing target objects at different stages.

[0098] The target object is an organ or tissue of a patient. In some embodiments, the target object may have the following regularity: the movement of its boundary is correlated with the movement of the internal vessels, such as the movement of the boundary of an organ tissue is positively correlated with the movement of the internal blood vessels; in other words, any type of tissue or organ having the above regularity may be used as a target object, for example, the patient's liver may be used as a target object.

[0099] In practical applications, images of the target object can be collected at different times to obtain reference medical images and medical images to be registered containing the target object at different periods. In some examples, the reference medical image can also be called a reference image, and the medical image to be registered can also be called a moving image.

[0100] In some embodiments, the reference medical image may be an image containing appearance details of the target object, and the medical image to be registered may be an image containing both appearance details and internal details of the target object. For example, the reference medical image may be a plain scan image, and the medical image to be registered may be an enhanced scan image. Taking the target object being the liver as an example, the reference medical image may be a plain scan image of the abdomen, in which the tissues that can be observed include the inferior vena cava and liver target organs, and the corresponding medical image to be registered may be an enhanced scan image of the abdomen, in which the blood vessels and liver target organs in the liver can be observed.

[0101] In some optional embodiments, an enhanced scan image of the patient's target object can be obtained before the operation, and a plain scan image of the target object can be obtained during the operation. On the one hand, since some of the enhanced scan images before the operation are collected from the outpatient process, there is a time difference of more than one week from the time of collecting the plain scan images during the operation, and the data are not obtained simultaneously in a short breath-holding process. Therefore, it is often encountered that the patient's respiratory movement, body position change, and peristalsis of the organ itself cause huge errors in the registration of the target organ and its internal blood vessels. On the other hand, since the intraoperative plain scan image cannot identify the vascular information (such as blood vessel information) inside the target object, it is difficult to make full use of the enhanced scan image collected in advance before the operation. In this regard, the plain scan image obtained during the operation can be used as a reference image, and the enhanced scan image obtained before the operation can be used as the medical image to be registered.

[0102] In some embodiments, the medical institution may collect the medical images to be registered for the target object of the patient in advance and store them in the data storage system of the server, wherein the data storage system may store the image data of the medical images to be processed by the server, and the data storage system may be integrated on the server, or placed on the cloud or other network servers. Then, the medical institution may upload the reference medical images collected in real time for the target object of the patient to the server, and the server may query and find the medical images to be registered corresponding to the same patient according to the patient identifier.

[0103] It is understandable that the reference medical image and the medical image to be registered may be uploaded to the data storage system of the server in advance, and after receiving the medical image registration instruction, the corresponding reference medical image and the medical image to be registered may be obtained for follow-up care.

[0104] S202, obtaining a plurality of medical image block pairs containing a local portion of a target object based on a reference medical image and a medical image to be registered; each medical image block pair includes a first image block representing a local portion of the target object in the reference medical image, and a second image block corresponding to the first image block in the medical image to be registered.

[0105] After obtaining a reference medical image and a medical image to be registered, the target objects in the reference medical image and the medical image to be registered can be processed in blocks to obtain a plurality of first image blocks representing local parts of the target object corresponding to the reference medical image, and a plurality of second image blocks representing local parts of the target object corresponding to the medical image to be registered.

[0106] Afterwards, by pairing the multiple image blocks corresponding to the reference medical image and the medical image to be registered, multiple medical image block pairs can be obtained.

[0107] S203 , for each medical image block pair, perform edge registration according to the edge of the target object in the first image block and the edge of the target object in the second image block to determine a local deformation field corresponding to a part of the target object.

[0108] Specifically, the first image block and the second image block in each medical image block pair include local edges of the target object (also called boundaries). For each obtained medical image block pair, edge alignment can be performed based on the local edges of the target object in the first image block and the local edges of the target object in the second image block.

[0109] After edge registration is performed on the edge of the target object in the first image block and the edge of the target object in the second image block, the local deformation field corresponding to the local part of the target object can be obtained according to the result of the edge registration. Thus, multiple local deformation fields of the target object can be obtained according to the local deformation fields corresponding to each of the multiple medical image blocks.

[0110] In some optional embodiments, for each medical image block pair, the first image block and the second image block in the medical image block pair i may be rigidly registered according to the edge to obtain a rigid registration matrix bound to the relative coordinates. Then, the corresponding local deformation field can be determined by matrix operation combined with the following formula:

[0111]

[0112] Among them, Disp index is the local deformation field corresponding to the medical image block pair i; index is the coordinate value of each pixel in the image block. If the image block is an image block of a three-dimensional image, index can be expressed as (x, y, z) representing the coordinate value of each pixel.

[0113] S204: performing image registration on the reference medical image and the medical image to be registered according to each local deformation field.

[0114] Specifically, each local deformation field is obtained by edge registration of the local edge of the target object in the medical image to be registered with the local edge of the target object in the reference medical image. It can be understood that the local deformation field can reflect the deformation path of the local edge of the target object moving from the corresponding position of the medical image to be registered to the corresponding position of the reference medical image, that is, it can characterize the movement or deformation mode of the edge of the target object, and the movement inside the target object is related to the movement of the edge of the target object. Accordingly, after determining the deformation or movement mode of the edge of the target object, the mode of moving the inside of the target object from the position of the medical image to be registered to the corresponding position on the reference medical image can be determined based on the correlation between the internal movement of the target object and the movement of the edge of the target object, so that the edge and internal information of the target object in the medical image to be registered can be registered to the reference medical image based on multiple local deformation fields, so as to achieve accurate registration between the reference medical image and the medical image to be registered.

[0115] In this embodiment, a reference medical image and a medical image to be registered containing a target object at different periods can be obtained, and then a plurality of medical image block pairs containing a local portion of the target object can be obtained based on the reference medical image and the medical image to be registered, wherein each medical image block pair includes a first image block representing a local portion of the target object in the reference medical image, and a second image block corresponding to the first image block in the medical image to be registered; for each medical image block pair, edge registration is performed based on an edge of the target object in the first image block and an edge of the target object in the second image block to determine a local deformation field corresponding to the local portion of the target object; and image registration is performed on the reference medical image and the medical image to be registered based on each local deformation field. In the present application, by dividing the target object in the reference medical image and the medical image to be registered into multiple parts and then performing edge registration, multiple local deformation fields that accurately reflect the edge deformation path of the target object can be obtained. Then, based on the correlation between the internal movement of the target object and the edge movement of the target object, with the help of multiple local deformation fields, the relevant image information inside the target object in the medical image to be registered is registered to the corresponding position on the reference medical image, thereby improving the registration accuracy of the target object during the image registration process.

[0116] In one embodiment, before acquiring a plurality of medical image block pairs according to the reference medical image and the medical image to be registered, the following steps may also be included:

[0117] Edge recognition is performed on the target object in the reference medical image and the medical image to be registered to obtain a first edge of the target object in the reference medical image and a second edge in the medical image to be registered.

[0118] In this embodiment, edge recognition can be first performed on the reference medical image and the medical image to be registered. For example, for the plain scan image and the enhanced scan image of the liver, edge recognition of the inferior vena cava and the liver can be performed on the plain scan image serving as the reference medical image, and edge recognition of the intrahepatic blood vessels and the liver can be performed on the enhanced scan image serving as the medical image to be registered.

[0119] Accordingly, edge registration is performed according to the edge of the target object in the first image block and the edge of the target object in the second image block, including:

[0120] Edge registration is performed according to a first edge of the target object in the first image block and a second edge of the target object in the second image block.

[0121] In some optional embodiments, performing edge registration on the edge of the target object may be understood as aligning the edge of the local target object in the first image block and the edge in the second image block.

[0122] Specifically, although the registration can be performed through Mask, this method is prone to cause errors in the registration. Figure 3a The two image blocks shown (i.e., the image block corresponding to the reference image and the image block corresponding to the moving image) will directly form a hard edge after segmentation based on the Mask registration method, such as Figure 3a The edge pointed by the arrow will be Figure 3a The arrow parts in the two image blocks are aligned instead of the edge of the target object (i.e., the curve formed by the junction of the black area and the white area in the image block). The alignment method is not reasonable and affects the accuracy of the registration result.

[0123] In this embodiment, if Figure 3b As shown, after edge recognition is performed to determine the edge of the local target object in the image block, if the edge of the local target object in the first image block and the edge in the second image block are aligned, a more reasonable and accurate registration effect for the local target object can be obtained, thereby reducing the registration error.

[0124] In one embodiment, Figure 4 As shown, in step S202, obtaining multiple medical image block pairs containing local parts of the target object according to the reference medical image and the medical image to be registered may include the following steps:

[0125] S401 : performing a first rigid registration on a target object in a reference medical image and a medical image to be registered.

[0126] In practical applications, after obtaining the reference medical image and the medical image to be registered, the target object in the image can be rigidly registered first. To distinguish it from other rigid registration processes in the following text, this rigid registration can also be called the first rigid registration. In this step, by rigidly registering the target object, that is, coarse registration, it can be ensured that the reference medical image and the target object in the medical image to be registered are roughly aligned.

[0127] In some optional embodiments, a mask image (i.e., Mask image) for the target object in the reference medical image and a mask image of the target object in the medical image to be registered can be obtained respectively, and then the target object can be subjected to mask-based local rigid registration based on the two mask images. The first rigid registration can be performed based on a deep learning algorithm or a traditional method.

[0128] S402, according to a predetermined number of image blocks, the target object in the first rigid registration result and the target object in the reference medical image are respectively divided into blocks; the number of image blocks is determined from multiple candidate block numbers according to the registration accuracies corresponding to each of the multiple candidate block numbers.

[0129] In the present application, the target object is divided into multiple blocks, and the local deformation field corresponding to each block is obtained respectively to characterize the movement path of the local edge of the target object between the medical image to be registered and the reference medical image. It can be understood that the appropriate number of blocks can better reflect the deformation mode of the local edge of the target object, so as to properly transfer the deformation of the edge of the target object to the inside of the target object, thereby improving the accuracy of subsequent registration.

[0130] For example, taking the example that the medical image to be registered and the reference image can represent a three-dimensional target object, if the number of blocks is 1×1×1, it means that there is no block, and the subsequent registration is only performed on the target object as a whole; if the number of blocks is large, for example, the number of blocks is 100×100×100, then the number of blocks is too small, which affects the force transmitted from the edge of the liver to the inside of the deformation field, resulting in a decrease in the accuracy of the deformation field inside the subsequent target object.

[0131] Based on this, the relevant samples can be divided into blocks according to different candidate block numbers in advance, and the corresponding multiple local deformation fields can be obtained and image registration can be performed to obtain the registration accuracy under each candidate block number, and then the image block number can be determined from multiple candidate block numbers according to the registration accuracy of each candidate block number. Then, the target object in the first rigid registration result and the target object in the reference medical image can be divided into blocks according to the image block number.

[0132] In some exemplary embodiments, the number of image blocks can be understood as a universal number determined based on a large number of data samples and statistical laws, that is, the number of image blocks can be used for reference medical images and mobile medical images of the same target object and different patients; and for different types of target objects, the corresponding number of image blocks can be determined respectively, and then when performing block processing, the number of image blocks related to the target object can be determined according to the target object contained in the image.

[0133] S403, obtaining a plurality of medical image block pairs according to the plurality of image blocks obtained after the block processing.

[0134] After the block processing, the obtained multiple image blocks can be paired to obtain multiple medical image block pairs.

[0135] In this embodiment, on the one hand, the target object can be locally rigidly registered first, and then divided and paired, so as to improve the correspondence between the two image blocks in the medical image block pair and reduce the difficulty of subsequent edge registration; on the other hand, the registration accuracy under different numbers of candidate blocks is determined by combining statistical methods, and the appropriate number of image blocks is screened to segment the medical image, so as to improve the accuracy of the deformation field obtained subsequently, so that it can more accurately characterize the influence of the edge deformation of the target object on the movement of the internal vessels of the target object.

[0136] In one embodiment, the number of image blocks can be determined by the following steps:

[0137] S301, obtaining sample reference medical images and sample to-be-registered medical images of target objects at different stages.

[0138] In some embodiments, the sample reference medical image may be an image containing appearance details of the target object, and the sample medical image to be registered may be an image containing both appearance details and internal details of the target object. For example, the sample reference medical image may be a plain scan image, and the sample medical image to be registered may be an enhanced scan image.

[0139] In some exemplary embodiments, sample reference medical images and sample medical images to be registered can be obtained from historical medical images acquired in the past. For example, with the authorization of the sample patient, the sample patient's preoperative enhanced scan image can be used as the sample reference medical image, and the sample patient's intraoperative plain scan image can be used to obtain the sample medical image to be registered.

[0140] In some other exemplary embodiments, a trained image generation model may also be used to obtain at least one of a sample reference medical image and a sample medical image to be registered. For example, image simulation may be performed on the two images at the same time to generate a sample reference medical image and a sample medical image to be registered. Alternatively, after obtaining a sample reference medical image or a sample medical image to be registered, another sample medical image to be registered or a sample reference medical image may be obtained through an image generation model.

[0141] S302, according to the sample reference medical image and the sample medical image to be registered, obtain multiple sample local deformation fields corresponding to the number of candidate blocks, and according to the multiple sample local deformation fields, perform image registration on the sample reference medical image and the sample medical image to be registered to obtain a sample medical image registration result.

[0142] In this step, after obtaining the sample reference medical image and the sample medical image to be registered, the current number of candidate blocks can be determined, and according to the number of candidate blocks, the target objects in the sample reference medical image and the sample medical image to be registered are respectively processed by blocks, and according to the obtained multiple sample medical image block pairs, multiple sample local deformation fields with the same number of candidate blocks are obtained. Then, according to the multiple sample local deformation fields, the sample reference medical image and the sample medical image to be registered are registered to obtain the sample medical image registration result.

[0143] Specifically, when acquiring multiple sample local deformation fields and performing image registration on the sample reference medical image and the sample medical image to be registered, reference may be made to the steps related to the registration of the reference medical image and the medical image to be registered in this application.

[0144] For the current number of candidate blocks, the target object in the sample reference medical image and the sample medical image to be registered can be processed into blocks according to the number of candidate blocks to obtain multiple sample medical image block pairs containing the local part of the target object. For example, if the current number of candidate blocks is M (M is a positive integer ≥ 2), then M first sample image blocks corresponding to the target object in the sample reference medical image and M second sample image blocks corresponding to the target object in the sample medical image to be registered can be obtained. Each sample medical image block pair includes a first sample image block representing the local part of the target object in the sample reference medical image and a second sample image block corresponding to the first sample image block in the medical image to be registered.

[0145] Then, for each pair of sample medical image blocks, edge alignment is performed according to the edge of the target object in the first sample image block and the edge of the target object in the second sample image block to determine the sample local deformation field corresponding to the local part of the target object. Then, according to each sample local deformation field, image alignment is performed on the sample reference medical image and the sample medical image to be aligned, thereby obtaining the sample medical image alignment result.

[0146] S303: Determine the image registration accuracy corresponding to the number of candidate blocks according to the reference point registration accuracy.

[0147] Among them, the reference point registration accuracy is used to characterize the registration accuracy corresponding to the reference points in the sample medical image registration results.

[0148] In this step, after obtaining the sample medical image registration result, the reference point in the result can be determined, and the reference point registration accuracy corresponding to the reference point can be obtained, and the reference point registration accuracy is used as the image registration accuracy corresponding to the current number of candidate blocks.

[0149] In some optional embodiments, if the sample reference medical image and the sample medical image to be registered contain vascular bifurcation points, the vascular bifurcation points can be used as reference points. If the sample reference medical image and the sample medical image to be registered do not contain vascular bifurcation points, the points marked by the user in the sample reference medical image and the sample medical image to be registered can be used as reference points.

[0150] S304, updating the number of candidate blocks, and determining the number of candidate blocks whose image registration accuracy meets a preset accuracy condition among the multiple candidate block numbers as the number of image blocks.

[0151] After obtaining the image registration accuracy corresponding to the number of candidate blocks, the current number of candidate blocks can be updated and step S302 can be re-executed until the image registration accuracy corresponding to each of the multiple candidate block numbers is obtained. Thereafter, the multiple candidate block numbers can be screened according to the image registration accuracy, and the candidate block number whose image registration accuracy meets the preset accuracy condition is determined as the image block number.

[0152] In this embodiment, for different numbers of candidate blocks, the image registration accuracy corresponding to the number of candidate blocks can be determined in combination with the reference point registration accuracy of the reference points in the sample medical image registration results, and the impact of different numbers of candidate blocks on image registration can be accurately quantified, which helps to screen out the appropriate number of image blocks.

[0153] In one embodiment, in step S302, obtaining a plurality of sample local deformation fields corresponding to the number of candidate blocks according to the sample reference medical image and the sample to-be-registered medical image may include the following steps:

[0154] Determine the direction to be adjusted; in the direction to be adjusted, divide the target objects in the sample reference medical image and the sample medical image to be registered into blocks according to the number of candidate blocks, and in the direction other than the direction to be adjusted, divide the target objects in the sample reference medical image and the sample medical image to be registered into blocks according to a preset number of blocks, and determine multiple sample local deformation fields based on the obtained medical image blocks.

[0155] The direction to be adjusted is one of multiple image block directions.

[0156] In a specific implementation, there may be multiple image block directions, and the multiple image block directions may be perpendicular to each other. For example, the multiple image block directions may be the X-axis direction and the Y-axis direction, or the X-axis direction, the Y-axis direction, and the Z-axis direction. Then, one of the multiple image block directions may be determined as the current direction to be adjusted. The direction to be adjusted may be understood as the direction in which the number of candidate blocks to be adjusted is determined and the appropriate number of image blocks in the direction is determined.

[0157] Then, the target objects in the sample reference medical image and the sample medical image to be registered can be blocked according to the number of candidate blocks in the direction to be adjusted, and the target objects in the sample reference medical image and the sample medical image to be registered can be blocked according to the preset number of blocks in other directions other than the direction to be adjusted. Thus, the sample reference medical image and the sample medical image to be registered can be blocked respectively in multiple image blocking directions to obtain multiple medical image blocks, namely, multiple first sample image blocks corresponding to the sample reference medical image and multiple second sample image blocks corresponding to the sample medical image to be registered. Afterwards, multiple sample local deformation fields can be determined based on the obtained medical image blocks, and the image registration accuracy in the direction to be adjusted under the current number of candidate blocks can be determined based on the multiple sample local deformation fields currently obtained.

[0158] Accordingly, step S304, determining the number of candidate blocks whose image registration accuracy satisfies a preset accuracy condition among the multiple candidate block numbers as the number of image blocks, may include:

[0159] Among multiple candidate block numbers, the candidate block number whose image registration accuracy meets the preset registration accuracy condition is determined as the image block number in the direction to be adjusted, and the current direction to be adjusted is updated until the image block number in each image block direction is determined.

[0160] Specifically, after obtaining the image registration accuracy in the direction to be adjusted under the current number of candidate blocks, the number of candidate blocks in the current direction to be adjusted can be adjusted, and the number of blocks in directions other than the direction to be adjusted can continue to remain unchanged, that is, still according to the preset number of blocks, block division is performed in directions other than the direction to be adjusted, and the above steps are repeated until the image registration accuracy under each number of candidate blocks in the direction to be adjusted is determined, and the number of candidate blocks whose image registration accuracy meets the preset registration accuracy condition is determined as the number of image blocks in the direction to be adjusted.

[0161] Afterwards, the current direction to be adjusted can be updated, and the aforementioned steps can be re-executed until the number of image blocks in each image block direction is determined. In some optional embodiments, after the number of image blocks in the direction to be adjusted is determined, if there is still a direction to be adjusted whose number of image blocks has not been determined, the determined number of image blocks can be used as the preset number of blocks in the corresponding direction. For example, when the image block direction includes three image block directions of X, Y, and Z, the number of image blocks in each of the three image block directions can be determined by the following steps:

[0162] (1) Control the Y and Z directions to distinguish only one block, traverse the X-axis direction with different numbers of blocks, and after registration, count the accuracy of the vascular bifurcation points. Determine the image registration accuracy based on the statistical results, and find the X-axis block number X_BlockSize (i.e., the number of image blocks in the X-axis direction) that makes the image registration accuracy optimal.

[0163] (2) Control X_BlockSize, divide the Z direction into 1 block, traverse the Y-axis direction with different numbers of blocks, and after registration, count the accuracy of the vascular bifurcation points. Determine the image registration accuracy based on the statistical results, and find the Y-axis block number that optimizes the image registration accuracy as Y_BlockSize (i.e., the number of image blocks in the Y-axis direction).

[0164] (3) Control X_BlockSize and Y_BlockSize, and traverse different numbers of blocks in the Z-axis direction. After registration, count the accuracy of the vascular bifurcation points, determine the image registration accuracy based on the statistical results, and find the Z-axis block number Z_BlockSize (i.e., the number of image blocks in the Z-axis direction) that achieves the best image registration accuracy.

[0165] For example, if the final number of image blocks in the three image block directions of X, Y, and Z are 4, 4, and 3 respectively, then Figure 5 As shown, the target object can be subsequently divided into 4×4×3=48 image blocks.

[0166] In this embodiment, the number of candidate blocks in each image block direction can be traversed in sequence, so that reasonable block division can be performed in each image block direction, thereby improving the adaptability of the number of image blocks to the target object.

[0167] In one embodiment, in step S403, obtaining multiple medical image block pairs according to multiple image blocks obtained after block processing may include the following steps:

[0168] Determine a plurality of initial image blocks including a part of the target object obtained after the block processing; if the initial image block includes an edge of the target object, expand the edge of the initial image block according to a preset edge expansion width to obtain an edge-expanded target image block; if the initial image block does not include an edge of the target object, expand the edge of the initial image block until the edge of part of the target object is covered to obtain an edge-expanded target image block; obtain a plurality of medical image block pairs based on the target image blocks paired with the reference medical image and the medical image to be registered.

[0169] The multiple initial image blocks include initial image blocks of the reference medical image and the medical image to be registered.

[0170] In a specific implementation, after the block processing, multiple initial image blocks containing local parts of the target object can be obtained from the multiple image blocks obtained by block processing of the reference medical image, and multiple initial image blocks containing local parts of the target object can be obtained from the multiple image blocks obtained by block processing of the medical image to be registered.

[0171] In some optional embodiments, the reference medical image and the medical image to be registered can be directly divided into blocks. In this case, in addition to the image area where the target object is located, the reference medical image and the medical image to be registered may also contain image areas other than the target object. Accordingly, there may be initial image blocks that do not contain the target object among the multiple initial image blocks obtained. In this regard, the boundaries of the target object in the reference medical image and the target object in the medical image to be registered can be pre-extracted and marked by an edge extraction algorithm. Then, after the block division, the image blocks outside the boundaries that do not contain the local part of the target object are removed, and the remaining image blocks are used as the initial image blocks.

[0172] Alternatively, the target object in the reference medical image and the target object in the medical image to be registered may be pre-identified using an edge extraction algorithm to determine the area where the target object is located in the reference medical image and the medical image to be registered, and then only the identified area containing the target object may be divided into blocks to obtain multiple initial image blocks.

[0173] After acquiring multiple initial image blocks, if the initial image block includes the edge of the target object, the edge of the initial image block can be expanded according to the preset edge expansion width to obtain the target image block after edge expansion, wherein the expanded width is also called padding, thereby ensuring a certain redundancy between image blocks, making the subsequent splicing and fusion of local deformation fields more coherent and smooth. For example, if the initial image block already contains the edge of the liver or blood vessels, it is expanded according to the pre-defined padding width, and the starting point and ending point after padding are recorded to facilitate the subsequent fusion of the deformation field: if the coordinates of the initial image block before padding are the starting point (x start ,y start , z start ), end point (x end ,y end , z end ), then the coordinate starting point of the target image block after padding is (x start -padding,y start -padding, z start -padding), the ending point coordinates are (x end -padding,y end -padding, z end -padding).

[0174] If the initial image block does not include the edge of the target object, such as the edge of the liver or blood vessels is not included in the current initial image block, the image block edge is expanded for the initial image block until the partial edge of the target object is covered to obtain the target image block after the edge expansion. For example, the padding range of the initial image block can be further expanded according to the preset ratio ratio*padding until it is expanded to the initial image block containing the edge of the target object. Correspondingly, the starting point and the ending point after padding can also be recorded to facilitate the subsequent fusion of the deformation field. By expanding the image block edge of the initial image block until the partial edge of the target object is covered, on the one hand, redundancy between image blocks is ensured, and on the other hand, the local edge of the target object is ensured in the image block, so that the edge alignment can be performed later to obtain a local deformation field reflecting the deformation path of the edge of the target object.

[0175] After obtaining a plurality of target image blocks, a plurality of medical image block pairs may be obtained based on the pairing relationship, the target image block in the reference medical image, and the target image block in the medical image to be registered that is paired with the target image block.

[0176] In one embodiment, in step S204, image registration is performed on the reference medical image and the medical image to be registered according to each local deformation field, which may include the following steps:

[0177] S2041: Determine a rigid deformation field of the target object according to the first rigid registration result.

[0178] Specifically, after the first rigid registration is performed on the target object in the reference medical image and the target object in the medical image to be registered, the rigid deformation field of the target object can be determined according to the first rigid registration result. In some optional embodiments, the rigid deformation field Disp can be obtained by combining the following formula: local :

[0179] Disp local =Trans2Disp(Mat local )

[0180] Among them, Mat local is a registration matrix (also called a local registration matrix) for the target object obtained according to the first rigid registration result.

[0181] S2042, obtaining a fused deformation field according to each local deformation field, and obtaining a composite deformation field of the target object according to a composite result of the rigid deformation field of the target object and the fused deformation field.

[0182] On the other hand, after obtaining multiple local deformation fields, the multiple local deformation fields can be fused to obtain a fused deformation field. In some examples, the fusion of local deformation fields can be understood as part of the multiple local deformation fields being replaced and part being retained.

[0183] During fusion, the deformation field weight of each local deformation field can be determined. In some examples, the weight of the local deformation field and the distance between the local deformation field and the center of the target object can be negatively correlated, that is, the shorter the distance between the local deformation field and the center of the target object (or the closer the local deformation field is to the center), the greater the weight of the local deformation field. In some examples, the weight of the local deformation field can be determined using any of the following weights: Gaussian weight, function weight (BSpline), inverse distance weighted weight.

[0184] Taking Gaussian weight as an example, for each local deformation field Disp index , can generate a Gaussian weight map Figure 6a A Gaussian weight map is shown. In one example, the weighted fusion deformation field Disp can be determined by combining the following formula: new :

[0185]

[0186] Among them, id is the id identifier corresponding to the local deformation field. Figure 6b A fused deformation field obtained according to the formula is shown.

[0187] Then, the rigid deformation field and the fused deformation field of the target object can be compounded, and the composite deformation field of the target object can be obtained according to the compound result, wherein the compound of the deformation field can be understood as the sum of the deformation field on the vector. In one embodiment, taking the target object as liver as an example, the rigid deformation field and the fused deformation field of the target object can be compounded according to the following formula to obtain the composite deformation field Disp liver :

[0188]

[0189] in, The symbol for the compound operation.

[0190] S2043: Perform image registration on the target object in the reference medical image and the target object in the medical image to be registered according to the composite deformation field.

[0191] In this embodiment, image registration is performed on the target object by combining the rigid deformation field of the target object and the local deformation field obtained by fusing multiple local deformation fields. This can combine the changes in the external edges of the target object to accurately register the internal information of the target object to the corresponding position, thereby improving the registration accuracy inside the target object.

[0192] In one embodiment, the method may further include the following steps:

[0193] A second rigid registration is performed on the reference medical image and the medical image to be registered, an initial global deformation field is determined according to the second rigid registration result, the initial global deformation field and the composite deformation field are fused to obtain a target global deformation field; based on the target global deformation field, the reference medical image and the medical image to be registered are registered with respect to the areas other than the target object to obtain a medical image after rigid registration.

[0194] Specifically, after obtaining the reference medical image and the medical image to be registered, the reference medical image and the medical image to be registered can be globally rigidly registered. For the sake of distinction, this registration can also be referred to as the second rigid registration. In some optional embodiments, the second rigid registration can be performed based on a deep learning algorithm or a traditional registration method, the purpose of which is to ensure that the reference medical image and the entire medical image to be registered can be roughly aligned as a whole.

[0195] After the second rigid registration, the registration matrix Mat can be obtained according to the corresponding second rigid registration result. global, which can also be called the global registration matrix. Afterwards, the registration matrix Mat global Obtain the initial global deformation field for the entire image. In one example, the registration matrix Mat can be combined with the following formula global Transformed into the initial global deformation field Disp global :

[0196] Disp global =Trans2Disp(Mat global )

[0197] Then, after obtaining the composite deformation field corresponding to the target object, the initial global deformation field and the composite deformation field can be fused to obtain the target global deformation field. In one example, the deformation field can be fused by combining the following formula:

[0198] Disp rigid =(1-weight liver )×Disp global +Disp liver ×weight liver

[0199] Among them, Disp rigid is the target global deformation field, weight liver The weight map with weights between 0 and 1 is obtained by smoothing the mask of the target object.

[0200] Afterwards, the regions outside the target object in the reference medical image and the medical image to be registered can be registered according to the target global deformation field to obtain a medical image after rigid registration.

[0201] In this embodiment, by aligning the areas outside the target object in the reference medical image and the medical image to be aligned according to the target global deformation field, accurate image alignment can be performed on the areas outside the target object by combining the correlation between the deformation of the target object and the deformation of other areas outside the target object.

[0202] In one embodiment, if it is determined to perform the first global elastic registration, after registering the reference medical image and the region outside the target object in the medical image to be registered according to the target global deformation field to obtain the medical image after rigid registration, the method may further include the following steps:

[0203] The deformation field of the target object in the rigidly registered medical image is kept unchanged, and the rigidly registered medical image and the reference image are first globally elastically registered.

[0204] In order to obtain a better registration effect, global elastic deformation can be performed. However, in the process of global elastic deformation, if the global elastic deformation is directly performed on the entire image, the deformation field outside the target object may affect the registration accuracy inside the target object. In this regard, the deformation field of the target object in the medical image after rigid registration can be kept unchanged, and then the medical image after rigid registration and the reference image are first globally elastically registered, thereby reducing the deformation inside the target object during the subsequent global elastic deformation. For example, if iterative registration is used, the Mask of the target object can be used during the iteration process to ensure that the deformation field inside the Mask is 0 before iteration.

[0205] In some other embodiments, if it is determined to perform the second global elastic registration, determining the initial global deformation field according to the second rigid registration result may include the following steps:

[0206] A second global elastic registration is performed on the medical image obtained after the second rigid registration, and an initial global deformation field is determined according to the medical image obtained after the second global elastic registration.

[0207] For example, after the second rigid registration, the medical image obtained after the second rigid registration can be subjected to the second global elastic registration, and then the registration matrix corresponding to the medical image obtained after the second global elastic registration can be obtained, and the registration matrix can be transformed to determine the initial global deformation field. Thus, the deformation inside the target object can be excluded in advance, and the deformation inside the target object can be reduced when the global rigid registration is performed later.

[0208] In one embodiment, Figure 7 As shown, in step S2042, a composite deformation field of the target object is obtained according to the composite result of the rigid deformation field and the fused deformation field of the target object, including:

[0209] S701 , elastically registering a target object in a reference medical image and a target object in a medical image to be registered, and obtaining an elastic deformation field corresponding to the target object according to the elastic registration result.

[0210] S702, compounding the elastic deformation field, the rigid deformation field and the fused deformation field to obtain a compound deformation field of the target object.

[0211] In practical applications, after performing image registration on the target object in the reference medical image and the target object in the medical image to be registered using the rigid deformation field of the target object and the composite deformation field corresponding to the fused deformation field, it is still possible that the liver contour cannot be aligned. In this case, a local elastic registration for the target object can be added.

[0212] Specifically, elastic registration may be performed on the target object in the reference medical image and the target object in the medical image to be registered, and the elastic deformation field corresponding to the target object may be obtained according to the elastic registration result.

[0213] Furthermore, after obtaining the elastic deformation field and rigid deformation field of the target object, the elastic deformation field, the rigid deformation field and the fused deformation field can be compounded to obtain the composite deformation field of the target object. According to the composite deformation field, image registration is performed on the target object in the reference medical image and the target object in the medical image to be registered. Moreover, according to the target global deformation field, the areas other than the target object in the reference medical image and the medical image to be registered can be registered to obtain a rigidly registered medical image. Finally, global elastic registration is performed on the rigidly registered medical image. Thus, the overall registration accuracy of the reference medical image and the medical image to be registered is improved through the combination of local elastic registration and global elastic registration.

[0214] In one embodiment, in step S201, obtaining a reference medical image and a medical image to be registered containing a target object at different stages may include:

[0215] A first enhanced scan image and a second enhanced scan image of a target object at different periods are obtained, the first enhanced scan image is used as a medical image to be registered, and a plain scan image simulation process is performed on the second enhanced scan image to obtain a reference medical image.

[0216] In a specific implementation, it is generally difficult to obtain the vascular information inside the target object from a plain scan image, and it is difficult to determine the registration effect of the medical image.

[0217] In this regard, in this embodiment, a first enhanced scan image and a second enhanced scan image of the target object at different periods can be obtained, and the first enhanced scan image can be used as the medical image to be registered, and the second enhanced scan image can be subjected to flat scan image simulation processing, wherein the flat scan image simulation processing can be understood as converting the second enhanced scan image into a flat scan image, thereby, the image obtained after the flat scan image simulation processing can be used as a reference medical image.

[0218] In an exemplary embodiment, with the authorization of the patient, dual enhanced images of the target object collected by the same patient at different time periods can be obtained, and the VP image (venous phase image) and DP image (delayed phase image) in the dual enhanced images are selected as the first enhanced scanning image and the second enhanced scanning image. These two images can obtain abundant reference points of the target object (such as vascular bifurcation points obtained based on the vascular information of the target object) and other relevant registration features. Then, one of the images can be used as a reference image and the other image can be used as a moving image. Figure 8As shown, the VP image and DP image of the liver are used as reference images and moving images respectively. Before the plain scan simulation processing, Figure 8 The reference image and the moving image can clearly observe the blood vessels inside the liver. Then, the moving image can be subjected to a plain scan simulation process, and the obtained simulated reference image can be used as a reference medical image for subsequent use.

[0219] In some optional embodiments, when performing plain scan simulation processing on the reference image, a histogram matching method can be used to perform plain scan simulation on the blood vessels in the reference image, and the blood vessel information can be smoothed to simulate a plain scan image. Specifically, the blood vessels and the target object in the reference image can be segmented. The histogram statistics of the area in the target object that does not contain the blood vessels are calculated, and according to the histogram, the grayscale of the blood vessels in the target object is re-assigned according to a preset probability to obtain a simulated reference image, for example, Figure 8 The simulated reference image shown in is an image obtained after the blood vessel information is erased in this way. Of course, other methods can also be selected for simulation, for example, blood vessel information can be erased based on a GAN network (Generative Adversarial Network), a diffusion model, etc.

[0220] Accordingly, after step S204, after performing medical image registration on the reference medical image and the medical image to be registered according to each local deformation field, the following steps may also be included:

[0221] According to the medical image registration results corresponding to the reference medical image and the medical image to be registered, medical image registration parameters for registering the medical image to be registered to the reference medical image are obtained; according to the medical image registration parameters, the reference points pre-marked in the first enhanced scanning image are deformed, and according to the distance between the deformed blood vessel bifurcation point and the paired reference point in the second enhanced scanning image, image registration evaluation information is determined.

[0222] Specifically, after obtaining the medical image to be registered and the reference medical image obtained after simulation, since both are actually obtained based on enhanced scanning images, the reference points in the medical image to be registered and the reference medical image can be marked based on the first enhanced scanning image and the second enhanced scanning image acquired previously.

[0223] In some optional embodiments, if the first enhanced scanning image and the second enhanced scanning image contain vascular bifurcation points, the vascular bifurcation points can be used as reference points; if the first enhanced scanning image and the second enhanced scanning image do not contain vascular bifurcation points, the points marked by the user in the first enhanced scanning image and the second enhanced scanning image can be used as reference points. Fig. 9An example of a reference image and a moving image is shown. Taking the reference point as a blood vessel bifurcation point as an example, one or more blood vessel bifurcation points may be annotated to obtain a series of paired blood vessel bifurcation points.

[0224] Then, after obtaining the medical image registration result, the corresponding medical image registration parameter can be determined, which is used to register the medical image to be registered to the reference medical image, for example, it can be a registration matrix or a deformation field. Then, the reference point pre-marked in the first enhanced scan image can be deformed according to the medical image registration parameter to obtain the deformed reference point, and the distance between the deformed reference point and the paired reference point in the second enhanced scan image is determined, and the image registration evaluation information is determined according to the distance, wherein the image registration evaluation information can be the image registration accuracy, and the image registration accuracy is negatively correlated with the distance, that is, the shorter the distance, the higher the image registration accuracy.

[0225] In this embodiment, a reference medical image is obtained by performing plain scan simulation processing on the second enhanced scan image, thereby pre-marking the vascular bifurcation points, and subsequently obtaining accurate and reliable image registration evaluation information based on the distances between the corresponding vascular bifurcation points.

[0226] In some optional embodiments, when determining the image registration accuracy corresponding to the number of candidate blocks, it can be obtained according to the steps of the above-mentioned embodiments, that is, based on the first sample enhanced scan image and the second sample enhanced scan image containing the target object at different periods, a sample reference medical image and a sample medical image to be registered can be obtained, and then the image registration accuracy can be determined according to the steps of the above-mentioned embodiments. The specific steps can be the same as those in the above-mentioned embodiments and will not be repeated here.

[0227] In order to enable those skilled in the art to better understand the above steps, the embodiment of the present application is illustrated below by using an example, but it should be understood that the embodiment of the present application is not limited to this.

[0228] like Fig.10 As shown in the figure, taking the target object as the liver, the plain scan image of the abdomen can be obtained as the reference image (i.e., the reference medical image), and the enhanced scan image of the abdomen can be used as the moving image (i.e., the medical image to be registered). Then, on the one hand, the reference image and the moving image can be globally rigidly registered in the abdomen to obtain the global rigid registration matrix Mat global , after transformation, the initial global deformation field Disp is obtained global .

[0229] On the other hand, the liver and blood vessels can be segmented for the reference image and the moving image. After that, the reference image and the moving image can be locally registered for the liver to obtain the liver local registration matrix Mat local, and then convert it into a local deformation field Disp local After that, for the pre-segmented liver and blood vessels, firstly use the edge extraction algorithm to extract their edges, then perform block processing in the X, Y, and Z directions of the liver, pair the image blocks corresponding to the reference image and the moving image, and obtain multiple medical image block pairs. Then, the local deformation field corresponding to each medical image block pair can be determined, and multiple local deformation fields can be fused to obtain the fused deformation field Disp new .

[0230] Then, you can use Disp new 、Disp local and Disp global Perform compound or fusion processing to obtain a compound deformation field corresponding to the target object and a target global deformation field for the global image, register the liver in the reference image and the liver in the moving image according to the compound deformation field, and register the area other than the liver in the reference image and the area other than the liver in the moving image according to the target global deformation field to obtain a medical image registration result. Furthermore, global elastic registration or global and local coordinated elastic registration can also be performed. Finally, the final registration result is output.

[0231] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0232] Based on the same inventive concept, the embodiment of the present application also provides a medical image registration device for implementing the medical image registration method involved above. The implementation solution provided by the device to solve the problem is similar to the implementation solution recorded in the above method, so the specific limitations in one or more medical image registration device embodiments provided below can refer to the limitations of the medical image registration method above, and will not be repeated here.

[0233] In an exemplary embodiment, Fig.11 As shown, a medical image registration device is provided, comprising:

[0234] A medical image acquisition module 1101 is used to acquire reference medical images of target objects at different periods and medical images to be registered;

[0235] The medical image block acquisition module 1102 is used to acquire a plurality of medical image block pairs containing a part of the target object according to the reference medical image and the medical image to be registered; each of the medical image block pairs includes a first image block representing a part of the target object in the reference medical image, and a second image block corresponding to the first image block in the medical image to be registered;

[0236] A local deformation field acquisition module 1103 is used to perform edge registration for each medical image block pair according to an edge of the target object in the first image block and an edge of the target object in the second image block, so as to determine a local deformation field corresponding to a part of the target object;

[0237] The registration module 1104 is used to perform image registration on the reference medical image and the medical image to be registered according to each of the local deformation fields.

[0238] In one embodiment, the medical image block acquisition module 1102 is used to:

[0239] Performing a first rigid registration on the reference medical image and the target object in the medical image to be registered;

[0240] According to a predetermined number of image blocks, the target object in the first rigid registration result and the target object in the reference medical image are respectively subjected to block processing; the number of image blocks is determined from the plurality of candidate block numbers according to the registration accuracies corresponding to the respective numbers of candidate block numbers;

[0241] A plurality of medical image block pairs are obtained according to the plurality of image blocks obtained after the block processing.

[0242] In one embodiment, the apparatus further comprises:

[0243] A sample image acquisition module, used for acquiring sample reference medical images and sample medical images to be registered containing the target object at different periods;

[0244] A sample image registration result acquisition module is used to acquire a plurality of sample local deformation fields corresponding to the number of candidate blocks according to the sample reference medical image and the sample medical image to be registered, and perform image registration on the sample reference medical image and the sample medical image to be registered according to the plurality of sample local deformation fields to obtain a sample medical image registration result;

[0245] An image registration accuracy acquisition module, used to determine the image registration accuracy corresponding to the number of candidate blocks according to the reference point registration accuracy; the reference point registration accuracy is used to characterize the registration accuracy corresponding to the reference point in the sample medical image registration result;

[0246] The updating module is used to update the number of candidate blocks, and determine the number of candidate blocks whose image registration accuracy meets the preset accuracy condition among the multiple candidate block numbers as the number of image blocks.

[0247] In one embodiment, the sample image registration result acquisition module is used to:

[0248] Determine a direction to be adjusted; the direction to be adjusted is one of a plurality of image block directions;

[0249] In the direction to be adjusted, the target object in the sample reference medical image and the sample medical image to be registered is divided into blocks according to the number of candidate blocks, and in the direction other than the direction to be adjusted, the target object in the sample reference medical image and the sample medical image to be registered is divided into blocks according to a preset number of blocks, and a plurality of sample local deformation fields are determined according to the obtained medical image blocks;

[0250] The step of determining, from among the plurality of candidate block numbers, the number of candidate block numbers whose image registration accuracy satisfies a preset accuracy condition as the number of image block numbers comprises:

[0251] Among the multiple candidate block numbers, the candidate block number whose image registration accuracy meets the preset registration accuracy condition is determined as the image block number in the direction to be adjusted, and the current direction to be adjusted is updated until the image block number in each image block direction is determined.

[0252] In one embodiment, the medical image block acquisition module 1102 is used to:

[0253] Determine a plurality of initial image blocks including the local part of the target object obtained after the block processing; the plurality of initial image blocks include initial image blocks of the reference medical image and the medical image to be registered;

[0254] If the initial image block includes the edge of the target object, the edge of the initial image block is expanded according to a preset edge expansion width to obtain a target image block after edge expansion;

[0255] If the initial image block does not include the edge of the target object, performing image block edge expansion on the initial image block until a portion of the edge of the target object is covered to obtain a target image block after edge expansion;

[0256] A plurality of medical image block pairs are obtained according to the target image blocks that are paired with the reference medical image and the medical image to be registered.

[0257] In one embodiment, the registration module 1104 is used to:

[0258] Determining a rigid deformation field of the target object according to the first rigid registration result;

[0259] Acquire a fused deformation field according to each of the local deformation fields, and obtain a composite deformation field of the target object according to a composite result of the rigid deformation field of the target object and the fused deformation field;

[0260] Image registration is performed on the target object in the reference medical image and the target object in the medical image to be registered according to the composite deformation field.

[0261] In one embodiment, the apparatus further comprises:

[0262] a target global deformation field acquisition module, used for performing a second rigid registration on the reference medical image and the medical image to be registered, determining an initial global deformation field according to the second rigid registration result, and fusing the initial global deformation field and the composite deformation field to obtain a target global deformation field;

[0263] The other region registration module is used to register the reference medical image and the region other than the target object in the medical image to be registered according to the target global deformation field to obtain a medical image after rigid registration.

[0264] In one embodiment, if it is determined to perform a first global elastic registration, the apparatus is further configured to:

[0265] The deformation field of the target object in the rigidly registered medical image is kept unchanged, and a first global elastic registration is performed on the rigidly registered medical image and the reference image.

[0266] In one embodiment, if it is determined to perform the second global elastic registration, the target global deformation field acquisition module is further used to:

[0267] A second global elastic registration is performed on the medical image obtained after the second rigid registration, and an initial global deformation field is determined according to the medical image obtained after the second global elastic registration.

[0268] In one embodiment, the registration module 1104 is used to:

[0269] Performing elastic registration on the target object in the reference medical image and the target object in the medical image to be registered, and obtaining an elastic deformation field corresponding to the target object according to the elastic registration result;

[0270] The elastic deformation field, the rigid deformation field and the fused deformation field are compounded to obtain a compound deformation field of the target object.

[0271] In one embodiment, the medical image block acquisition module 1102 is further used to:

[0272] Performing edge recognition on the target object in the reference medical image and the medical image to be registered to obtain a first edge of the target object in the reference medical image and a second edge in the medical image to be registered;

[0273] The local deformation field acquisition module 1103 is used to:

[0274] Edge registration is performed according to a first edge of the target object in the first image block and a second edge of the target object in the second image block.

[0275] In one embodiment, the medical image acquisition module is used to:

[0276] Acquire a first enhanced scan image and a second enhanced scan image of the target object at different periods, use the first enhanced scan image as the medical image to be registered, and perform plain scan image simulation processing on the second enhanced scan image to obtain a reference medical image;

[0277] The device also includes:

[0278] A registration parameter determination module, configured to obtain medical image registration parameters for registering the medical image to be registered to the reference medical image according to medical image registration results corresponding to the reference medical image and the medical image to be registered;

[0279] A registration evaluation module is used to deform the reference points pre-marked in the first enhanced scan image according to the medical image registration parameters, and determine the image registration evaluation information according to the distance between the deformed reference points and the paired blood vessel bifurcation points in the second enhanced scan image.

[0280] Each module in the above-mentioned medical image registration device can be implemented in whole or in part by software, hardware or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0281] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Fig.12As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store medical image data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a medical image registration method is implemented.

[0282] Those skilled in the art will understand that Fig.12 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0283] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0284] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0285] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0286] 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, stored data, displayed data, 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 relevant data must comply with relevant regulations.

[0287] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present 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), magnetoresistive 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. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0288] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, 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.

[0289] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A medical image registration method, characterized in that: The method comprises: Acquire a reference medical image and a to-be-registered medical image containing a target object at different periods; Acquire a plurality of medical image block pairs containing a local portion of the target object according to the reference medical image and the medical image to be registered; each of the medical image block pairs comprises a first image block in the reference medical image representing the local portion of the target object, and a second image block in the medical image to be registered corresponding to the first image block; For each of the medical image block pairs, performing edge registration according to an edge of the target object in the first image block and an edge of the target object in the second image block, and determining a local deformation field corresponding to a part of the target object; Image registration is performed on the reference medical image and the medical image to be registered according to each of the local deformation fields.

2. The method according to claim 1, characterized in that The step of acquiring a plurality of medical image block pairs including a local part of the target object according to the reference medical image and the medical image to be registered comprises: Performing a first rigid registration on the reference medical image and the target object in the medical image to be registered; According to a predetermined number of image blocks, the target object in the first rigid registration result and the target object in the reference medical image are respectively subjected to block processing; the number of image blocks is determined from the plurality of candidate block numbers according to the registration accuracies corresponding to the respective numbers of candidate block numbers; A plurality of medical image block pairs are obtained according to the plurality of image blocks obtained after the block processing.

3. The method according to claim 2, characterized in that The number of image blocks is determined by the following steps: Acquire sample reference medical images and sample to-be-registered medical images of the target object at different time periods; According to the sample reference medical image and the sample medical image to be registered, a plurality of sample local deformation fields corresponding to the number of candidate blocks are obtained, and according to the plurality of sample local deformation fields, image registration is performed on the sample reference medical image and the sample medical image to be registered to obtain a sample medical image registration result; Determining the image registration accuracy corresponding to the number of candidate blocks according to the reference point registration accuracy; the reference point registration accuracy is used to characterize the registration accuracy corresponding to the reference point in the sample medical image registration result; The number of candidate blocks is updated, and among the plurality of candidate block numbers, the number of candidate blocks whose image registration accuracy satisfies a preset accuracy condition is determined as the number of image blocks.

4. The method according to claim 3, characterized in that The step of acquiring a plurality of sample local deformation fields corresponding to the number of candidate blocks according to the sample reference medical image and the sample to-be-registered medical image comprises: Determine a direction to be adjusted; the direction to be adjusted is one of a plurality of image block directions; In the direction to be adjusted, the target object in the sample reference medical image and the sample medical image to be registered is divided into blocks according to the number of candidate blocks, and in the direction other than the direction to be adjusted, the target object in the sample reference medical image and the sample medical image to be registered is divided into blocks according to a preset number of blocks, and a plurality of sample local deformation fields are determined according to the obtained medical image blocks; The step of determining, from among the plurality of candidate block numbers, the number of candidate block numbers whose image registration accuracy satisfies a preset accuracy condition as the number of image block numbers comprises: Among the multiple candidate block numbers, the candidate block number whose image registration accuracy meets the preset registration accuracy condition is determined as the image block number in the direction to be adjusted, and the current direction to be adjusted is updated until the image block number in each image block direction is determined.

5. The method according to claim 2, characterized in that: The step of obtaining a plurality of medical image block pairs according to the plurality of image blocks obtained after the block processing comprises: Determine a plurality of initial image blocks including the local part of the target object obtained after the block processing; the plurality of initial image blocks include initial image blocks of the reference medical image and the medical image to be registered; If the initial image block includes the edge of the target object, the edge of the initial image block is expanded according to a preset edge expansion width to obtain a target image block after edge expansion; If the initial image block does not include the edge of the target object, performing image block edge expansion on the initial image block until a portion of the edge of the target object is covered to obtain a target image block after edge expansion; A plurality of medical image block pairs are obtained according to the target image blocks that are paired with the reference medical image and the medical image to be registered.

6. The method according to claim 2, characterized in that The performing image registration on the reference medical image and the medical image to be registered according to each of the local deformation fields comprises: Determining a rigid deformation field of the target object according to the first rigid registration result; Acquire a fused deformation field according to each of the local deformation fields, and obtain a composite deformation field of the target object according to a composite result of the rigid deformation field of the target object and the fused deformation field; Image registration is performed on the target object in the reference medical image and the target object in the medical image to be registered according to the composite deformation field.

7. The method according to claim 6, characterized in that The method further comprises: Performing a second rigid registration on the reference medical image and the medical image to be registered, determining an initial global deformation field according to the second rigid registration result, and fusing the initial global deformation field and the composite deformation field to obtain a target global deformation field; According to the target global deformation field, the reference medical image and the region outside the target object in the medical image to be registered are registered to obtain a medical image after rigid registration.

8. The method according to claim 7, characterized in that If it is determined to perform the first global elastic registration, after registering the reference medical image and the region outside the target object in the medical image to be registered according to the target global deformation field to obtain a medical image after rigid registration, the method further includes: The deformation field of the target object in the rigidly registered medical image is kept unchanged, and a first global elastic registration is performed on the rigidly registered medical image and the reference image.

9. The method according to claim 7, characterized in that: If it is determined to perform the second global elastic registration, then determining the initial global deformation field according to the second rigid registration result includes: A second global elastic registration is performed on the medical image obtained after the second rigid registration, and an initial global deformation field is determined according to the medical image obtained after the second global elastic registration.

10. The method according to claim 8 or 9, characterized in that: The step of obtaining a composite deformation field of the target object based on a composite result of the rigid deformation field of the target object and the fused deformation field comprises: Performing elastic registration on the target object in the reference medical image and the target object in the medical image to be registered, and obtaining an elastic deformation field corresponding to the target object according to the elastic registration result; The elastic deformation field, the rigid deformation field and the fused deformation field are compounded to obtain a compound deformation field of the target object.

11. The method according to any one of claims 1 to 9, characterized in that Before acquiring a plurality of medical image block pairs according to the reference medical image and the medical image to be registered, the method further includes: Performing edge recognition on the target object in the reference medical image and the medical image to be registered to obtain a first edge of the target object in the reference medical image and a second edge in the medical image to be registered; The performing edge registration according to the edge of the target object in the first image block and the edge of the target object in the second image block comprises: Edge registration is performed according to a first edge of the target object in the first image block and a second edge of the target object in the second image block.

12. The method according to any one of claims 1 to 9, characterized in that The obtaining of reference medical images of the target object at different periods and the medical images to be registered includes: Acquire a first enhanced scan image and a second enhanced scan image of the target object at different periods, use the first enhanced scan image as the medical image to be registered, and perform plain scan image simulation processing on the second enhanced scan image to obtain a reference medical image; After performing medical image registration on the reference medical image and the medical image to be registered according to each of the local deformation fields, the method further includes: Acquire medical image registration parameters for registering the medical image to be registered to the reference medical image according to medical image registration results corresponding to the reference medical image and the medical image to be registered; The reference points pre-marked in the first enhanced scan image are deformed according to the medical image registration parameters, and image registration evaluation information is determined according to the distance between the deformed reference points and the paired reference points in the second enhanced scan image.

13. A medical image registration device, characterized in that: The device comprises: A medical image acquisition module, used for acquiring reference medical images of target objects at different periods and medical images to be registered; A medical image block acquisition module, configured to acquire a plurality of medical image block pairs containing a local portion of the target object according to the reference medical image and the medical image to be registered; each of the medical image block pairs comprises a first image block in the reference medical image representing a local portion of the target object, and a second image block in the medical image to be registered corresponding to the first image block; A local deformation field acquisition module, configured to perform edge registration for each pair of medical image blocks according to an edge of the target object in the first image block and an edge of the target object in the second image block, and determine a local deformation field corresponding to a part of the target object; A registration module is used to perform image registration on the reference medical image and the medical image to be registered according to each of the local deformation fields.

14. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 12 are implemented.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

16. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

Images