Method and device for removing artifacts of medical image and DSA system

Abstract

The invention relates to an artifact removal method and device for a medical image, computer equipment and a storage medium, and the method comprises the steps: obtaining an image set of each collection position in a target part, the image set comprising mask images and filling images which are in one-to-one correspondence; performing global registration on the mask image and the interference image of each acquisition position to obtain a first registration result; performing local registration on the image set based on the first registration result to obtain a corresponding second registration result; further, based on the second registration result, subtraction processing is performed on the image set, and a subtraction image of the target part is obtained. Through the method and the device, the problem that the artifacts in the DSA image cannot be effectively removed is solved, the image artifacts are removed, and the blood vessel image quality is improved.

Description

Technical Field

[0001] The present application relates to the field of image processing, and in particular to a method and device for removing artifacts from medical images and a DSA system. Background Art

[0002] Digital Subtraction Angiography (DSA) is a medical imaging technique used to observe the structure and function of the human vascular system, which can provide high-resolution and high-contrast vascular images. The basic principle of DSA technology is to subtract the images taken before and after the injection of contrast agent to eliminate the interference of structures such as bones and soft tissues, and obtain a subtraction image containing only blood vessels. Among them, the image taken before the injection of contrast agent is called the mask image, and the image taken after the injection of contrast agent is called the full image.

[0003] When photographing body parts such as the lower limbs, due to the constraints of the bed and the X-ray flat panel, it is necessary to use DSA technology to photograph and subtract different positions of the target part, and stitch the images at different positions into a complete DSA image. Due to the difference in the acquisition time of the mask and the fill film, the stitched DSA image often has motion artifacts, and the existing technology can only achieve simple stitching of images at different positions, and cannot effectively remove artifacts in the DSA image, resulting in low quality of vascular images.

[0004] With regard to the problem that the artifacts in the DSA images cannot be effectively removed in the related art, no effective solution has been proposed so far. Summary of the invention

[0005] In this embodiment, a method, apparatus, computer device and storage medium for removing artifacts from medical images are provided to solve the problem that artifacts in DSA images cannot be effectively removed in the related art.

[0006] In a first aspect, a method for removing artifacts from a medical image is provided in this embodiment, and the method includes:

[0007] Acquire an image set of each acquisition position in the target part; the image set includes a mask image and a full-image image corresponding to each other;

[0008] Performing global registration on the mask image and the clear image at each acquisition position to obtain a first registration result;

[0009] Performing local registration on the image set based on the first registration result to obtain a corresponding second registration result;

[0010] Based on the second registration result, subtraction processing is performed on the image set to obtain a subtraction image of the target part.

[0011] In some of the embodiments, before globally registering the mask image and the patch image at each acquisition position, the method further includes:

[0012] Acquire the mask image and the full-image image of each acquisition position in the target part;

[0013] The mask image and the patch image are preprocessed respectively; the preprocessing includes one or more combinations of operations including logarithmic transformation, regularization, normalization and denoising.

[0014] In some of the embodiments, globally registering the mask image and the full-length image at each acquisition position to obtain a first registration result includes:

[0015] Determining a region of interest in the film image;

[0016] Extracting a block image having the same size as the region of interest from the mask image, and matching the block image with the region of interest;

[0017] Determine the optimal matching position corresponding to the region of interest according to the matching result;

[0018] Determining an offset corresponding to the optimal matching position, and performing global registration on the mask image based on the offset to obtain the first registration result;

[0019] Or, the globally registering the mask image and the full-length image at each acquisition position to obtain a first registration result includes:

[0020] Determining a region of interest in the mask image;

[0021] Extracting a block image having the same size as the region of interest from the film image, and matching the block image with the region of interest;

[0022] Determine the optimal matching position corresponding to the region of interest according to the matching result;

[0023] An offset corresponding to the optimal matching position is determined, and the film image is globally registered based on the offset to obtain the first registration result.

[0024] In some embodiments, determining the region of interest in the film image includes:

[0025] Determine a target area in the radiographic image according to the target part, and use the target area as the region of interest;

[0026] or, determining the region of interest in the patch image based on the information degree of each local region in the patch image;

[0027] Or, the determining the region of interest in the mask image includes:

[0028] Determine a target region in the mask image according to the target part, and use the target region as the region of interest;

[0029] Or, based on the information degree of each local area in the mask image, the region of interest in the mask image is determined.

[0030] In some embodiments, extracting a block image having the same size as the region of interest from the mask image and matching the block image with the region of interest includes:

[0031] Processing the mask image by an image search algorithm to obtain a plurality of block images having the same size as the region of interest;

[0032] Matching each of the block images with the region of interest;

[0033] Or, extracting a block image having the same size as the region of interest from the film image and matching the block image with the region of interest includes:

[0034] Processing the patch image by an image search algorithm to obtain a plurality of block images having the same size as the region of interest;

[0035] Each of the block images is matched with the region of interest.

[0036] In some embodiments, determining the offset corresponding to the optimal matching position, and performing global registration on the mask image based on the offset to obtain the first registration result includes:

[0037] Determining the offset according to the optimal matching position and the center of the film image;

[0038] Expanding the boundary of the mask image based on a preset maximum offset;

[0039] Based on the offset, performing registration processing on the mask image after boundary expansion to obtain the first registration result;

[0040] Or, the determining the offset corresponding to the optimal matching position, and performing global registration on the film image based on the offset to obtain the first registration result, includes:

[0041] Determining the offset according to the optimal matching position and the center of the mask image;

[0042] Expanding the boundary of the patch image based on a preset maximum offset;

[0043] Based on the offset, the patch image after boundary expansion is registered to obtain the first registration result.

[0044] In some embodiments, locally registering the image set based on the first registration result to obtain a corresponding second registration result includes:

[0045] Based on the first registration result, performing block matching on the image set according to preset image control points;

[0046] Based on the block matching result, the image set is processed by a registration algorithm to obtain the corresponding second registration result.

[0047] In some embodiments, performing subtraction processing on the image set based on the second registration result to obtain the subtraction image of the target part includes:

[0048] splicing the respective patch images into an overall image of the target part, and performing subtraction processing on the overall image based on the second registration result to obtain a subtracted image of the target part;

[0049] or, performing subtraction processing on the solid film images based on the second registration result, and splicing the solid film images after the subtraction processing to obtain the subtraction image of the target part;

[0050] Or, performing subtraction processing on the image set based on the second registration result to obtain a subtraction image of the target part includes:

[0051] splicing the mask images into an overall image of the target part, and performing subtraction processing on the overall image based on the second registration result to obtain a subtracted image of the target part;

[0052] Or, based on the second registration result, the mask image is subjected to subtraction processing, and the mask images after the subtraction processing are spliced ​​to obtain the subtraction image of the target part.

[0053] In a second aspect, a device for removing artifacts from medical images is provided in this embodiment, the device comprising: an acquisition module, a matching module, a registration module, and a subtraction module;

[0054] The acquisition module is used to acquire an image set at each acquisition position in the target part; the image set includes a mask image and a full-image image in one-to-one correspondence;

[0055] The matching module is used to globally register the mask image and the full image at each acquisition position to obtain a first registration result;

[0056] The registration module is used to perform local registration on the image set based on the first registration result to obtain a corresponding second registration result;

[0057] The subtraction module is used to perform subtraction processing on the image set based on the second registration result to obtain a subtraction image of the target part.

[0058] In a third aspect, a DSA system is provided in this embodiment, including a DSA main body device and a processing device, wherein the processing device executes the medical image artifact removal method as described in the first aspect above.

[0059] Compared with the related art, the medical image artifact removal method, apparatus, computer equipment and storage medium provided in the present embodiment obtain an image set at each acquisition position in the target part, the image set including a one-to-one corresponding mask image and full image; globally align the mask image and full image of each acquisition position to obtain a first alignment result; locally align the image set based on the first alignment result to obtain a corresponding second alignment result; further, based on the second alignment result, subtraction processing is performed on the image set to obtain a subtraction image of the target part, thereby solving the problem of being unable to effectively remove artifacts in DSA images, achieving the removal of image artifacts and improving the quality of vascular images.

[0060] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0062] Figure 1 It is a structural diagram of a DSA system provided by an embodiment of the present application;

[0063] Figure 2 is a flow chart of a method for removing artifacts from medical images provided by an embodiment of the present application;

[0064] Figure 3It is a flow chart of a method for removing artifacts from medical images provided by a preferred embodiment of the present application;

[0065] Figure 4 It is a structural block diagram of a medical image artifact removal device provided in one embodiment of the present application.

[0066] In the figure: 110, DSA main device; 111, examination bed; 112, fixed frame; 113, robotic arm; 114, X-ray transmitter; 115, X-ray detector; 120, processing equipment; 10, acquisition module; 20, matching module; 30, registration module; 40, subtraction module. DETAILED DESCRIPTION

[0067] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0068] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the general meaning understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the", "these" and the like in this application do not represent quantitative restrictions, and they can be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusions; for example, a process, method and system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether directly or indirectly. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there can be three relationships. For example, "A and / or B" can mean: A exists alone, A and B exist at the same time, and B exists alone. Usually, the character " / " indicates that the objects associated with each other are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0069] Figure 1 is a schematic diagram of the structure of a DSA system provided in an embodiment of the present application, such as Figure 1 As shown, the DSA system includes a DSA main device 110 and a processing device 120. The DSA main device may include: an examination bed 111, a fixed frame 112, and a rotatable mechanical arm 113 (such as Figure 1 The figure shows a C-shaped arm), an X-ray emitter 114 and an X-ray detector 115 arranged at both ends of the mechanical arm 113; wherein the processing device 120 is connected to the X-ray detector 115, and is used to obtain the contrast image and mask image collected by the X-ray detector.

[0070] During the angiography process, the subject can lie flat on the examination bed 111. When the mechanical arm 113 drives the fixed frame 112 to move to the target position, the positions of the X-ray emitter 114 and the X-ray detector 115 are also relatively fixed. At this time, the X-ray emitter 114 is used to emit X-rays, and the corresponding X-ray detector 115 is used to detect the X-rays and convert them into digital images, and output the digital images to the processing device 120. Before the contrast agent is injected into the subject, the X-ray detector 115 collects the mask image of the subject, and after a certain dose of contrast agent is injected into the subject through the contrast agent injector, the X-ray detector 115 collects the contrast image of the subject.

[0071] In this embodiment, a method for removing artifacts from medical images is provided. Figure 2 is a flowchart of the method for removing artifacts from medical images of this embodiment. Figure 2 As shown, the process includes the following steps:

[0072] Step S210, obtaining an image set of each acquisition position in the target part; the image set includes a mask image and a full-image image in one-to-one correspondence.

[0073] Specifically, for each acquisition position in the target part, a first mask image and a first full-image image corresponding to the acquisition position are acquired. It should be noted that there are N acquisition positions preset on the target part, and the present embodiment acquires images of each position step by step, and sequentially acquires at least one mask image frame and at least one full-image frame of the i-th (1≤i≤N) acquisition position, wherein the mask image is an image taken before the contrast agent is injected, and the full-image image is an image taken after the contrast agent is injected.

[0074] Furthermore, the mask image and the full image of each frame are globally registered, and the mask image and the full image with the highest similarity measure are determined according to the registration result, and they are respectively used as the mask image and the full image corresponding to the acquisition position.

[0075] Step S220: globally register the mask image and the full image at each acquisition position to obtain a first registration result.

[0076] In this embodiment, the mask image can be globally registered based on the patch image of each acquisition position, or the mask image can be globally registered based on the patch image of each acquisition position. Specifically, when the mask image is used as the transformation object, the region of interest in the patch image is determined, and a block image of the same size as the region of interest is extracted from the mask image, the block image is matched with the region of interest, and the block image with the highest similarity to the region of interest is screened out from the mask image, and the position of the block image is the optimal matching position. Afterwards, the corresponding offset is calculated according to the optimal matching position, and the mask image is registered based on the offset to obtain a first registration result, that is, the mask image after global registration.

[0077] When the patch image is used as the transformation object, the region of interest in the mask image is determined; a block image of the same size as the region of interest is extracted from the patch image, and the block image is matched with the region of interest. After that, the optimal matching position corresponding to the region of interest is determined according to the matching result, the offset corresponding to the optimal matching position is determined, and the patch image is globally registered based on the offset to obtain a first registration result, that is, the patch image after global registration.

[0078] Step S230: performing local registration on the image set based on the first registration result to obtain a corresponding second registration result.

[0079] In this embodiment, the mask image in the image set can be locally registered based on the first registration result, or the patch image in the image set can be locally registered based on the first registration result. Specifically, when the mask image is used as the transformation object, the mask image after global registration and the patch image are block matched according to the preset image control points to obtain the corresponding relationship between each local area in the mask image after global registration and the patch image, and then based on the corresponding relationship, the mask image after global registration and the patch image are registered by the registration algorithm to obtain the corresponding second registration result, that is, the mask image after global registration is transformed into the target mask image.

[0080] When the patch image is taken as the transformation object, the patch image and the mask image after global registration are block matched according to the preset image control points to obtain the correspondence between each local area in the patch image after global registration and the mask image, and then based on the correspondence, the patch image and the mask image after global registration are registered through the registration algorithm to obtain the corresponding second registration result, that is, the patch image after global registration is transformed into the target patch image.

[0081] It should be noted that in the global registration and local registration process, the method of selecting the mask image as the transformation object has priority over the full film image as the transformation object. This is because the essence of the DSA registration process is the vascular subtraction process, in which the mask image is an image taken before the injection of contrast agent, and the full film image is an image taken after the injection of contrast agent. The difference between the two is that the blood vessels cannot be observed from the mask image, and the full film image contains obvious blood vessels. Therefore, based on the difference between the mask image and the full film image, the two need to be registered. In actual application scenarios, the vascular position reflected by the full film image is the real vascular position. If the full film image is used as the transformation object for registration, it may cause the current real vascular position to change, reducing the accuracy of image transformation.

[0082] Step S240: Based on the second registration result, subtraction processing is performed on the image set to obtain a subtraction image of the target part.

[0083] Specifically, a complete subtraction image can be obtained by stitching first and then subtracting. Each of the patch images is stitched into an overall image of the target part, and based on the target mask image corresponding to each acquisition position, the overall image is subtracted to obtain a subtraction image of the target part; or, each of the mask images is stitched into an overall image of the target part, and based on the target patch image corresponding to each acquisition position, the overall image is subtracted to obtain a subtraction image of the target part.

[0084] In addition, a method of first subtracting and then stitching is adopted, in which the first film image is subtracted based on the target mask image, or the mask image is subtracted based on the target film image to eliminate interference from structures such as bones and soft tissues, and the subtracted film images or mask images are stitched together to obtain a subtracted image of the target part.

[0085] At present, when photographing body parts such as lower limbs, due to the constraints of the bed and the X-ray flat panel, it is necessary to use DSA technology to photograph and subtract different positions of the target part, and stitch the images of different positions into a complete DSA image. Due to the difference in the acquisition time of the mask and the full film, the stitched DSA image often has motion artifacts, and the existing technology can only achieve simple stitching of images at different positions, and cannot effectively remove artifacts in the DSA image, resulting in low quality of vascular images.

[0086] Compared with the prior art, the present application obtains an image set of each acquisition position in the target part, and the image set includes a one-to-one corresponding mask image and full image; the mask image and full image of each acquisition position are globally registered to obtain a first registration result; based on the first registration result, the image set is locally registered to obtain a corresponding second registration result; further, based on the second registration result, the image set is subtracted to obtain a subtracted image of the target part. Based on this, by globally registering and locally registering the image of each acquisition position, the image is corrected according to the registration result, so that accurate subtraction processing can be achieved based on the corrected image, which solves the problem of being unable to effectively remove artifacts in DSA images, achieves the removal of image artifacts, and improves the quality of vascular images.

[0087] In some of the embodiments, before globally registering the mask image and the full image at each acquisition position, the following steps are also included:

[0088] Obtaining a mask image and a full-length image of each acquisition position in the target area;

[0089] The mask image and the patch image are preprocessed respectively; the preprocessing includes one or more combinations of operations including logarithmic transformation, regularization, normalization and denoising.

[0090] Specifically, a mask image and a patch image of the i-th acquisition position are obtained, and the mask image and the patch image are preprocessed respectively. In this embodiment, the preprocessing method includes but is not limited to logarithmic transformation, regularization, normalization and denoising, and one or more of these methods can be selected to process the image.

[0091] It should be noted that the logarithmic transformation is used to convert the acquired image from the exponential domain to the linear domain. At this time, the image is analyzed from the perspective of grayscale distribution. The preprocessed mask image and the full film image will be at the same level, which means that when the patient is not moving, the subtraction of the preprocessed mask image and the full film image will result in the vascular structure and does not include other structures.

[0092] In addition, operations such as regularization, normalization, and denoising in this embodiment are compatible with the image registration method actually selected, so that the mask image and the full image can be preprocessed in a targeted manner based on the current image registration method, thereby improving the accuracy and quality of image registration and significantly improving the registration effect.

[0093] Through this embodiment, the mask image and the full film image of each acquisition position in the target part are obtained, and the mask image and the full film image are preprocessed respectively, so as to convert the image from the exponential domain to the linear domain, so that the detail level of the transformed image is clearer, so as to optimize the subsequent image registration results.

[0094] In some of the embodiments, globally registering the mask image and the full-length image at each acquisition position to obtain a first registration result includes the following steps:

[0095] Determine the region of interest in the film image;

[0096] Extracting a block image with the same size as the region of interest from the mask image, and matching the block image with the region of interest;

[0097] Determine the optimal matching position corresponding to the region of interest according to the matching results;

[0098] An offset corresponding to the optimal matching position is determined, and the mask image is globally registered based on the offset to obtain a first registration result.

[0099] Specifically, a region of interest (ROI) in the full-length image is selected, and the size of the region of interest is t (0 < t ≤ 1) times the full-length image; wherein the specific value of t is selected based on historical experience information. The mask image and the region of interest are processed by block matching, and a block image of the same size as the region of interest is extracted from the mask image, and the block image is matched with the region of interest.

[0100] Furthermore, according to the matching results between the mask image and the region of interest, the best matching position corresponding to the region of interest is determined, that is, the center position of the region most similar to the region of interest is screened out from the mask image. After that, the distance between the best matching position and the center of the mask image is calculated as an offset, and the mask image is registered based on the offset to obtain a first registration result, that is, the mask image after global registration.

[0101] Alternatively, globally registering the mask image and the full-length image at each acquisition position to obtain a first registration result comprises the following steps:

[0102] determining a region of interest in the mask image;

[0103] Extracting a block image with the same size as the region of interest from the film image, and matching the block image with the region of interest;

[0104] Determine the optimal matching position corresponding to the region of interest according to the matching results;

[0105] An offset corresponding to the optimal matching position is determined, and global registration is performed on the film images based on the offset to obtain a first registration result.

[0106] Specifically, a region of interest in the mask image is selected, and the size of the region of interest is t (0 < t ≤ 1) times the size of the mask image; wherein the specific value of t is selected based on historical experience information. The patch image and the region of interest are processed by block matching, a block image of the same size as the region of interest is extracted from the patch image, and the block image is matched with the region of interest.

[0107] Furthermore, according to the matching results between the patch image and the region of interest, the best matching position corresponding to the region of interest is determined, that is, the center position of the region most similar to the region of interest is screened out from the patch image. After that, the distance between the best matching position and the center of the patch image is calculated as an offset, and the patch image is registered based on the offset to obtain a first registration result, that is, the patch image after global registration.

[0108] It should be noted that in order to solve the problem of motion artifacts caused by the gantry shaking during the DSA step scanning process, the global registration algorithm used in this embodiment is based on the registration calculation of the region of interest. The common registration method is based on the global calculation of feature points, which uses the image feature structure. For the overall offset caused by the bed movement and gantry shaking during the DSA step scanning process, the global registration algorithm based on the region of interest can achieve better registration effect than the global registration algorithm based on feature points.

[0109] Through this embodiment, the mask image and the full image of each acquisition position are globally registered to obtain a first registration result, and both the mask image and the full image can be used as transformation objects to correct image deviation caused by factors such as rack shaking and subject movement.

[0110] In some embodiments, determining a region of interest in a film image comprises the following steps:

[0111] Determine the target area in the film image according to the target part, and take the target area as the region of interest;

[0112] Or, based on the information degree of each local area in the patch image, the region of interest in the patch image is determined.

[0113] Specifically, in the patch image, an image region corresponding to the target part is selected as the target region, and the target region is the region of interest of the patch image. For example, when the target part is the liver, the liver region in the image is selected as the region of interest. At the same time, an area with a size of t (0 < t ≤ 1) times the patch image can be selected at the center of the patch image as the region of interest.

[0114] In addition, when determining the region of interest in the patch image, the information degree of each local area in the image can be combined for selection. First, the information entropy of different areas in the patch image is calculated, or the patch image is processed by a feature extraction algorithm to obtain the corresponding information template map, and then the information entropy or information template map is used to preliminarily screen out the area with large information in the patch image, and the area is framed multiple times to extract the desired region of interest.

[0115] Or, determining a region of interest in a mask image comprises the following steps:

[0116] Determine the target area in the mask image according to the target part, and use the target area as the region of interest;

[0117] Or, based on the information degree of each local area in the mask image, the region of interest in the mask image is determined.

[0118] Specifically, in the mask image, an image region corresponding to the target part is selected as the target region, and the target region is the region of interest of the mask image. For example, when the target part is the liver, the liver region in the image is selected as the region of interest. At the same time, an area with a size of t (0 < t ≤ 1) times the size of the mask image can be selected in the center of the mask image as the region of interest.

[0119] In addition, when determining the region of interest in the mask image, the information degree of each local area in the image can be combined for selection. First, the information entropy of different regions in the mask image is calculated, or the mask image is processed by a feature extraction algorithm to obtain the corresponding information template map, and then the information entropy or information template map is used to preliminarily screen out the region with large information in the mask image, and the region is framed multiple times to extract the desired region of interest.

[0120] Through this embodiment, the target area is determined according to the target part, and the target area is used as the region of interest, or the region of interest is determined based on the information degree of each local area in the image, so as to accurately select the region of interest.

[0121] In some of the embodiments, extracting a block image having the same size as the region of interest from the mask image and matching the block image with the region of interest comprises the following steps:

[0122] The mask image is processed by an image search algorithm to obtain a plurality of block images of the same size as the region of interest;

[0123] Match each patch image to the region of interest.

[0124] Specifically, when matching the mask image with the region of interest, the mask image is processed by an image search algorithm to obtain multiple block images of the same size as the region of interest. The image search algorithm in this embodiment includes but is not limited to a full traversal search, a three-step search method, a four-step search method, and a diamond search method.

[0125] Furthermore, each block image is matched with the region of interest, and the region most similar to the region of interest is screened out from the mask image. The center position of the region is the optimal matching position, so that the image offset can be calculated based on the optimal matching position and the center of the mask image.

[0126] Alternatively, a block image having the same size as the region of interest is extracted from the film image, and the block image is matched with the region of interest, including the following steps:

[0127] The patch image is processed by an image search algorithm to obtain a plurality of block images of the same size as the region of interest;

[0128] Match each patch image to the region of interest.

[0129] Specifically, when matching the patch image with the region of interest, the patch image is processed by an image search algorithm to obtain multiple block images of the same size as the region of interest. The image search algorithm in this embodiment includes but is not limited to full traversal search, three-step search method, four-step search method and diamond search method.

[0130] Furthermore, each block image is matched with the region of interest, and the region most similar to the region of interest is screened out from the patch image. The center position of the region is the optimal matching position, so that the image offset can be calculated based on the optimal matching position and the center of the patch image.

[0131] Through this embodiment, the mask image or the patch image is processed by an image search algorithm to obtain multiple block images of the same size as the region of interest, and each block image is matched with the region of interest, so as to facilitate the acquisition of the degree of image offset and improve the accuracy of image correction.

[0132] In some of the embodiments, determining an offset corresponding to the optimal matching position, and performing global registration on the mask image based on the offset to obtain a first registration result includes the following steps:

[0133] Determine the offset according to the optimal matching position and the center of the film image;

[0134] Expanding the boundary of the mask image based on a preset maximum offset;

[0135] Based on the offset, the mask image after boundary expansion is registered to obtain a first registration result.

[0136] It should be noted that the best matching position is the center of the area in the mask image that is most similar to the region of interest. This embodiment calculates the distance between the best matching position and the center of the mask image to obtain the corresponding offset.

[0137] Furthermore, based on a preset maximum offset, a symmetric mapping method is used to extend the boundaries of the mask image. That is, mirror-symmetric pixel values ​​are used to fill the expanded area on the boundary of the image, so that each side of the image is expanded by the maximum offset, thereby avoiding blank areas when the mask image is selected as the global registration after translation.

[0138] After that, according to the calculated image offset, the image area at the corresponding position is selected in the mask image after boundary expansion, which is the mask image after global registration. For example, when the distance between the best matching position in the mask image and the center of the patch image is (1, 3), indicating that the offset of the current image is (1, 3), then in the expanded mask image, the image area at a distance of (1, 3) from the image center is selected as the mask image after global registration, and the size of the mask image is the same as the size of the region of interest.

[0139] Or, determining an offset corresponding to the optimal matching position, and performing global registration on the film image based on the offset to obtain a first registration result, comprising the following steps:

[0140] Determine the offset according to the optimal matching position and the center of the mask image;

[0141] Expanding the boundary of the patch image based on a preset maximum offset;

[0142] Based on the offset, the patch image after boundary expansion is registered to obtain a first registration result.

[0143] It should be noted that the best matching position is the center of the region in the mask image that is most similar to the region of interest. This embodiment calculates the distance between the best matching position and the center of the mask image to obtain the corresponding offset.

[0144] Furthermore, based on the preset maximum offset, the patch image is extended at the boundary by a symmetric mapping method, that is, the expanded area is filled with mirror-symmetric pixel values ​​at the boundary of the image, so that each side of the image is expanded by the maximum offset, thereby avoiding the appearance of blank areas when the patch image is selected as the global registration after translation.

[0145] After that, according to the calculated image offset, the image area at the corresponding position in the patch image after boundary expansion is selected, which is the patch image after global registration. For example, when the distance between the best matching position in the patch image and the center of the mask image is (1, 3), indicating that the offset of the current image is (1, 3), then in the expanded patch image, the image area at a distance of (1, 3) from the image center is selected as the patch image after global registration, and the size of the patch image is the same as the size of the region of interest.

[0146] Through this embodiment, the offset corresponding to the optimal matching position is determined, and the mask image is globally registered based on the offset to obtain a first registration result, thereby realizing global registration of the mask image or the mask image and achieving the effect of accurately correcting the image.

[0147] In some of the embodiments, locally registering the image set based on the first registration result to obtain a corresponding second registration result includes the following steps:

[0148] Perform block matching on the image set according to the preset image control points;

[0149] Based on the block matching result, the image set is processed by a registration algorithm to obtain a corresponding second registration result.

[0150] Specifically, when the mask image is used as the transformation object, the present embodiment uses the pixel displacement method to perform local registration on the mask image after global registration and the full image. First, multiple image control points are selected to determine the key positions in the image, and block matching is performed on the mask image after global registration and the full image according to each image control point, thereby establishing a corresponding relationship between each local area in the mask image after global registration and the full image.

[0151] Furthermore, based on the correspondence between the mask image and the full image after global registration, image registration is performed using a rigid registration algorithm or an elastic registration algorithm, and each local area in the mask image after global registration is moved and adjusted to align with the full image to obtain a corresponding second registration result, i.e., the target mask image after global and local registration is completed.

[0152] When the patch image is used as the transformation object, the present embodiment uses the pixel displacement method to perform local registration of the patch image and the mask image after global registration. First, multiple image control points are selected to determine the key positions in the image, and block matching is performed on the patch image and the mask image after global registration according to each image control point, thereby establishing a corresponding relationship between each local area in the patch image after global registration and the mask image.

[0153] Afterwards, based on the correspondence between the global registered full film image and the mask image, image registration is performed using a rigid registration algorithm or an elastic registration algorithm, and each local area in the globally registered full film image is moved and adjusted to align with the mask image to obtain the corresponding second registration result, that is, the target full film image with global and local registration is completed.

[0154] Through this embodiment, the image set is locally registered based on the first registration result to obtain a corresponding second registration result, thereby achieving local registration of the mask image or the full image, thereby effectively eliminating image artifacts caused by patient movement.

[0155] In some of the embodiments, based on the second registration result, performing subtraction processing on the image set to obtain a subtraction image of the target part includes the following steps:

[0156] splicing the individual film images into an overall image of the target part, and performing subtraction processing on the overall image based on the second registration result to obtain a subtracted image of the target part;

[0157] Or, subtraction processing is performed on the solid image based on the second registration result, and each solid image after the subtraction processing is spliced ​​to obtain a subtraction image of the target part.

[0158] Specifically, the present embodiment can use different processing methods to complete the subtraction and splicing of the patch images to obtain a complete subtracted image of the target part. Exemplarily, according to the acquisition position corresponding to the patch images, each patch image is spliced ​​into an overall image of the target part, and a local area associated with each target mask image is determined in the overall image, and each time the target mask image is subtracted from the associated local area, and finally a complete subtracted image is obtained.

[0159] In addition, the target mask image and the patch image corresponding to each acquisition position can be determined in advance, and the target mask image is subtracted from the patch image at the same acquisition position each time to obtain multiple subtracted patch images, and then the subtracted patch images are spliced ​​to obtain a complete subtracted image of the target part.

[0160] Or, based on the second registration result, performing subtraction processing on the image set to obtain a subtraction image of the target part includes the following steps:

[0161] splicing the mask images into an overall image of the target part, and performing subtraction processing on the overall image based on the second registration result to obtain a subtracted image of the target part;

[0162] Or, the mask image is subtracted based on the second registration result, and the mask images after the subtraction are spliced ​​to obtain a subtraction image of the target part.

[0163] Specifically, the present embodiment may adopt different processing methods to obtain a complete subtraction image of the target part. For example, according to the acquisition position corresponding to the mask image, each mask image is spliced ​​into an overall image of the target part, a local area associated with each target patch image is determined in the overall image, and each time the target patch image is subtracted from the associated local area, and finally a complete subtraction image is obtained.

[0164] In addition, the target full-image and mask image corresponding to each acquisition position can be predetermined, and the target full-image and the mask image at the same acquisition position are subtracted each time, and then the subtracted images are spliced ​​to obtain a complete subtracted image of the target part.

[0165] It should be noted that the target mask image or target patch image obtained through global registration and local registration has corrected the offset generated during the image acquisition process, avoiding large differences between the mask image and the patch image, and effectively improving the probability of successful image stitching.

[0166] Through this embodiment, the image set is subtracted based on the second registration result to obtain a subtracted image of the target part, thereby eliminating motion artifacts generated during the image acquisition process and significantly improving the quality of the subtracted image.

[0167] The present embodiment is described and illustrated below through preferred embodiments.

[0168] Figure 3 is a flowchart of the method for removing artifacts from medical images of this preferred embodiment, such as Figure 3 As shown, the medical image artifact removal method includes the following steps:

[0169] Step S310, logarithmically transforming the first mask image and the first patch image of each acquisition position in the target part to obtain a corresponding second mask image and a second patch image;

[0170] Step S320, determining the region of interest in the second mask image, extracting a block image of the same size as the region of interest from the second mask image, and matching the block image with the region of interest;

[0171] Step S330, determining the best matching position corresponding to the region of interest according to the matching result, and calculating the offset corresponding to the best matching position;

[0172] Step S340, performing boundary extension on the second mask image based on a preset maximum offset, and performing registration processing on the boundary-extended second mask image based on the offset, to obtain a third mask image;

[0173] Step S350, performing block matching on the third mask image and the second patch image according to preset image control points to obtain a correspondence between each local area in the third mask image and the second patch image;

[0174] Step S360, based on the corresponding relationship, the third mask image and the second mask image are processed by a registration algorithm to obtain a target mask image corresponding to the acquisition position;

[0175] Step S370: stitching the first patch images into an overall image of the target part, and performing subtraction processing on the overall image based on the target mask image corresponding to each acquisition position to obtain a subtracted image of the target part.

[0176] Through this embodiment, the first mask image and the first patch image of each acquisition position in the target part are logarithmically transformed to obtain the corresponding second mask image and the second patch image, so as to convert the image from the exponential domain to the linear domain. On this basis, the region of interest in the second patch image is determined, and the second mask image is matched with the region of interest, and the offset of the image is obtained according to the matching result, so that the second mask image after the boundary expansion can be registered based on the offset, and the image offset caused by factors such as rack shaking can be corrected.

[0177] Furthermore, the third mask image is locally registered with the second fill image to obtain a target mask image corresponding to the acquisition position, and the first fill images are spliced ​​into an overall image of the target part. The overall image is subtracted based on the target mask image corresponding to each acquisition position to obtain a final subtracted image, which solves the problem of being unable to effectively remove artifacts in DSA images, realizes the removal of image artifacts, and improves the quality of vascular images.

[0178] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0179] In this embodiment, a device for removing artifacts from medical images is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. The terms "module", "unit", "subunit", etc. used below can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0180] Figure 4 is a structural block diagram of the medical image artifact removal device of this embodiment, such as Figure 4 As shown, the device includes: an acquisition module 10, a matching module 20, a registration module 30 and a subtraction module 40;

[0181] An acquisition module 10 is used to acquire an image set at each acquisition position in the target part; the image set includes a mask image and a full-image image in one-to-one correspondence;

[0182] A matching module 20 is used to globally register the mask image and the full image at each acquisition position to obtain a first registration result;

[0183] A registration module 30, configured to perform local registration on the image set based on the first registration result to obtain a corresponding second registration result;

[0184] The subtraction module 40 is used to perform subtraction processing on the image set based on the second registration result to obtain a subtraction image of the target part.

[0185] In some of these embodiments, Figure 4 On the basis of this, the device also includes a preprocessing module, which is used to obtain a mask image and a full-image image of each acquisition position in the target part; the mask image and the full-image image are preprocessed respectively; the preprocessing includes one or more combinations of operations such as logarithmic transformation, regularization, normalization and denoising.

[0186] In some of these embodiments, Figure 4 On the basis of, the device also includes a global registration module, which is used to determine the region of interest in the film image; extract a block image with the same size as the region of interest from the mask image, and match the block image with the region of interest; determine the optimal matching position corresponding to the region of interest according to the matching result; determine the offset corresponding to the optimal matching position, and perform global registration on the mask image based on the offset to obtain a first registration result; or, it is used to determine the region of interest in the mask image; extract a block image with the same size as the region of interest from the film image, and match the block image with the region of interest; determine the optimal matching position corresponding to the region of interest according to the matching result; determine the offset corresponding to the optimal matching position, and perform global registration on the film image based on the offset to obtain a first registration result.

[0187] In some of these embodiments, Figure 4 On the basis of, the device also includes a screening module, which is used to determine the target area in the full-film image according to the target part, and use the target area as the region of interest; or, based on the information degree of each local area in the full-film image, determine the region of interest in the full-film image; or, used to determine the target area in the mask image according to the target part, and use the target area as the region of interest; or, based on the information degree of each local area in the mask image, determine the region of interest in the mask image.

[0188] In some of these embodiments, Figure 4 On the basis of, the device also includes a search module, which is used to process the mask image through an image search algorithm to obtain multiple block images with the same size as the region of interest; match each block image with the region of interest; or, is used to process the mask image through an image search algorithm to obtain multiple block images with the same size as the region of interest; match each block image with the region of interest.

[0189] In some of these embodiments, Figure 4 On the basis of, the device also includes an expansion module, which is used to determine the offset according to the optimal matching position and the center of the mask image; perform boundary expansion on the mask image based on a preset maximum offset; based on the offset, perform registration processing on the mask image after the boundary expansion to obtain a first registration result; or, it is used to determine the offset according to the optimal matching position and the center of the mask image; perform boundary expansion on the mask image based on a preset maximum offset; based on the offset, perform registration processing on the mask image after the boundary expansion to obtain a first registration result.

[0190] In some of these embodiments, Figure 4 On the basis of, the device also includes an establishment module for performing block matching on the image set according to preset image control points based on the first registration result; based on the block matching result, the image set is processed by a registration algorithm to obtain a corresponding second registration result.

[0191] In some of these embodiments, Figure 4 On the basis of, the device also includes a processing module, which is used to stitch the various patch images into an overall image of the target part, and perform subtraction processing on the overall image based on the second registration result to obtain a subtraction image of the target part; or, perform subtraction processing on the patch images based on the second registration result, and stitch the various patch images after the subtraction processing to obtain a subtraction image of the target part; or, splice the various mask images into an overall image of the target part, and perform subtraction processing on the overall image based on the second registration result to obtain a subtraction image of the target part; or, perform subtraction processing on the mask images based on the second registration result, and stitch the various mask images after the subtraction processing to obtain a subtraction image of the target part.

[0192] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0193] In this embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0194] Optionally, the computer device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0195] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and will not be repeated in this embodiment.

[0196] In addition, in combination with the medical image artifact removal method provided in the above embodiments, a storage medium may also be provided in this embodiment to implement the method. The storage medium stores a computer program; when the computer program is executed by a processor, any medical image artifact removal method in the above embodiments is implemented.

[0197] It should be understood that the specific embodiments described herein are only used to explain the application, rather than to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of this application.

[0198] Obviously, the drawings are only some examples or embodiments of the present application. For ordinary technicians in the field, the present application can also be applied to other similar situations based on these drawings without creative work. In addition, it is understandable that although the work done in this development process may be complicated and lengthy, for ordinary technicians in the field, certain changes in design, manufacturing or production based on the technical content disclosed in this application are only conventional technical means and should not be regarded as insufficient content disclosed in this application.

[0199] The term "embodiment" in this application refers to a specific feature, structure or characteristic described in conjunction with the embodiment that can be included in at least one embodiment of the present application. The appearance of this phrase in various locations in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. It is clearly or implicitly understood by those of ordinary skill in the art that the embodiments described in this application can be combined with other embodiments without conflict.

[0200] The above-mentioned 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 patent protection. 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 scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the attached claims.

Claims

1. A method for removing artifacts from medical images. It is characterized in that The method comprises: Acquire an image set of each acquisition position in the target part; the image set includes a mask image and a full-image image corresponding to each other; Performing global registration on the mask image and the clear image at each acquisition position to obtain a first registration result; Performing local registration on the image set based on the first registration result to obtain a corresponding second registration result; Based on the second registration result, subtraction processing is performed on the image set to obtain a subtraction image of the target part.

2. The method for removing artifacts from medical images according to claim 1, It is characterized in that Before globally registering the mask image and the patch image at each acquisition position, the method further includes: Acquire the mask image and the full-image image of each acquisition position in the target part; The mask image and the patch image are preprocessed respectively; the preprocessing includes one or more combinations of operations including logarithmic transformation, regularization, normalization and denoising.

3. The method for removing artifacts from medical images according to claim 1, It is characterized in that The globally registering the mask image and the full-length image at each acquisition position to obtain a first registration result includes: Determining a region of interest in the film image; Extracting a block image having the same size as the region of interest from the mask image, and matching the block image with the region of interest; Determine the optimal matching position corresponding to the region of interest according to the matching result; Determining an offset corresponding to the optimal matching position, and performing global registration on the mask image based on the offset to obtain the first registration result; Or, the globally registering the mask image and the full-length image at each acquisition position to obtain a first registration result includes: Determining a region of interest in the mask image; Extracting a block image having the same size as the region of interest from the film image, and matching the block image with the region of interest; Determine the optimal matching position corresponding to the region of interest according to the matching result; An offset corresponding to the optimal matching position is determined, and the film image is globally registered based on the offset to obtain the first registration result.

4. The method for removing artifacts from medical images according to claim 3, It is characterized in that The determining of the region of interest in the film image comprises: Determine a target area in the radiographic image according to the target part, and use the target area as the region of interest; or, determining the region of interest in the patch image based on the information degree of each local region in the patch image; Or, the determining the region of interest in the mask image includes: Determine a target region in the mask image according to the target part, and use the target region as the region of interest; Or, based on the information degree of each local area in the mask image, the region of interest in the mask image is determined.

5. The method for removing artifacts from medical images according to claim 3, It is characterized in that The step of extracting a block image having the same size as the region of interest from the mask image and matching the block image with the region of interest comprises: Processing the mask image by an image search algorithm to obtain a plurality of block images having the same size as the region of interest; Matching each of the block images with the region of interest; Or, extracting a block image having the same size as the region of interest from the film image and matching the block image with the region of interest includes: Processing the patch image by an image search algorithm to obtain a plurality of block images having the same size as the region of interest; Each of the block images is matched with the region of interest.

6. The method for removing artifacts from medical images according to claim 3, It is characterized in that The determining of the offset corresponding to the optimal matching position, and performing global registration on the mask image based on the offset to obtain the first registration result, includes: Determining the offset according to the optimal matching position and the center of the film image; Expanding the boundary of the mask image based on a preset maximum offset; Based on the offset, performing registration processing on the mask image after boundary expansion to obtain the first registration result; Or, determining the offset corresponding to the optimal matching position, and performing global registration on the film image based on the offset to obtain the first registration result, includes: Determining the offset according to the optimal matching position and the center of the mask image; Expanding the boundary of the patch image based on a preset maximum offset; Based on the offset, the patch image after boundary expansion is registered to obtain the first registration result.

7. The method for removing artifacts from medical images according to claim 1, It is characterized in that The locally registering the image set based on the first registration result to obtain a corresponding second registration result includes: Based on the first registration result, performing block matching on the image set according to preset image control points; Based on the block matching result, the image set is processed by a registration algorithm to obtain the corresponding second registration result.

8. The method for removing artifacts from medical images according to claim 1, It is characterized in that The step of performing subtraction processing on the image set based on the second registration result to obtain a subtraction image of the target part includes: splicing the respective patch images into an overall image of the target part, and performing subtraction processing on the overall image based on the second registration result to obtain a subtracted image of the target part; or, performing subtraction processing on the solid film images based on the second registration result, and splicing the solid film images after the subtraction processing to obtain the subtraction image of the target part; Or, performing subtraction processing on the image set based on the second registration result to obtain a subtraction image of the target part includes: splicing the mask images into an overall image of the target part, and performing subtraction processing on the overall image based on the second registration result to obtain a subtracted image of the target part; Or, based on the second registration result, the mask image is subjected to subtraction processing, and the mask images after the subtraction processing are spliced ​​to obtain the subtraction image of the target part.

9. A device for removing artifacts from medical images, It is characterized in that The device comprises: an acquisition module, a matching module, a registration module and a subtraction module; The acquisition module is used to acquire an image set at each acquisition position in the target part; the image set includes a mask image and a full-image image in one-to-one correspondence; The matching module is used to globally register the mask image and the full image at each acquisition position to obtain a first registration result; The registration module is used to perform local registration on the image set based on the first registration result to obtain a corresponding second registration result; The subtraction module is used to perform subtraction processing on the image set based on the second registration result to obtain a subtraction image of the target part.

10. A DSA system, comprising a DSA main body device and a processing device, It is characterized in that The processing device is configured to perform the steps of the medical image artifact removal method according to any one of claims 1 to 8.

Images