A CBCT multi-scan artifact image fusion method and system
The CBCT image stitching method, which combines scan position displacement information and multi-scale fusion algorithm, solves the problems of low automation, insufficient accuracy and efficiency in the existing technology, and achieves high-precision and fast image stitching, which is suitable for radiotherapy positioning.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SUPERACCURACY SCIENCE & TECHNOLOGY CO LTD
- Filing Date
- 2026-03-13
- Publication Date
- 2026-06-19
Smart Images
Figure CN122243762A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of radiotherapy image processing technology, specifically to a CBCT multi-scan image fusion method and system. Background Technology
[0002] In tumor radiotherapy, cone-beam computed tomography (CBCT) is one of the core technologies for image-guided radiotherapy. It can accurately locate the tumor by acquiring three-dimensional images of the patient before or during treatment, thereby improving the accuracy and safety of radiotherapy. However, due to the limited field of view of CBCT detectors, for large tumors or patients requiring full-body verification, a single scan cannot cover the entire target area and surrounding normal tissue. Therefore, multi-scan scanning is required to acquire CBCT images of multiple local fields of view, which are then stitched together to obtain a complete, large-scale three-dimensional image.
[0003] Currently, existing CBCT multi-scan image stitching technologies are mainly divided into two categories: one is the stitching method based on image feature matching, which extracts grayscale features, edge features, or texture features of images from different scan positions and uses feature point matching algorithms to achieve image alignment and stitching; the other is the stitching method based on mechanical positioning, which uses the positioning information of the scan position drive mechanism of the radiotherapy equipment to roughly determine the spatial position relationship of images from different scan positions, and then performs simple stitching and fusion.
[0004] Based on the above-mentioned existing technologies, the following shortcomings exist:
[0005] (1) The image feature matching-based stitching method is affected by slight changes in the patient's position, movement of tissues and organs and image noise. The accuracy of feature point extraction and matching is low, and stitching misalignment is easy to occur, resulting in image distortion after stitching and failing to meet the accuracy requirements of radiotherapy positioning. At the same time, this method requires a large number of feature calculations on the image, resulting in low stitching efficiency and difficulty in meeting the rapid positioning needs of clinical radiotherapy.
[0006] (2) The stitching method based on mechanical positioning relies solely on the mechanical positioning accuracy of the scanning position drive mechanism. However, the scanning position of the radiotherapy equipment is affected by factors such as mechanical gaps and load changes during the movement, resulting in positioning errors. High-precision image alignment cannot be achieved solely through mechanical positioning information, leading to obvious stitching gaps or overlapping area distortion in the stitched images, which also fails to meet the requirements of precise radiotherapy.
[0007] (3) Most existing splicing methods are semi-automated or manual splicing, which require medical staff to intervene manually, such as manually selecting matching feature points and adjusting splicing parameters, which increases the workload of medical staff. In addition, the subjectivity of manual operation is strong, which further affects the splicing accuracy and consistency.
[0008] Therefore, how to overcome the shortcomings of existing technologies and provide a CBCT multi-scan stitching solution that can achieve automation and improve accuracy and efficiency has become an urgent technical problem to be solved in the field of radiotherapy imaging technology. Summary of the Invention
[0009] The purpose of this invention is to provide a CBCT multi-scan image fusion method and system to solve the technical problems of low automation, low accuracy and low efficiency in existing CBCT multi-scan image stitching.
[0010] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:
[0011] In a first aspect, the present invention provides a CBCT multi-scan image fusion method, comprising:
[0012] The scanning object is controlled to move along a preset scanning path and stop at the reference scanning position and multiple target scanning positions respectively, and the original CBCT image information and corresponding displacement information of the reference scanning position and each of the target scanning positions are acquired simultaneously.
[0013] The original CBCT image information of the reference scan position and each of the target scan positions are reconstructed to obtain the initial three-dimensional CBCT images of the reference scan position and each of the target scan positions.
[0014] The initial three-dimensional CBCT images of each target scan position are aligned with the initial three-dimensional CBCT image of the reference scan position to obtain the aligned image.
[0015] A multi-scale fusion algorithm is used to fuse the aligned images to obtain a three-dimensional CBCT stitched image.
[0016] Furthermore, after obtaining the reference scan position and the initial three-dimensional CBCT images of each scan position, the obtained reference scan position and the initial three-dimensional CBCT images are preprocessed; the preprocessing includes one or more combinations of grayscale correction, noise removal and edge enhancement.
[0017] Furthermore, the initial three-dimensional CBCT images of each target scan position are aligned with the initial three-dimensional CBCT image of the reference scan position, including:
[0018] An initial spatial transformation matrix of each target scan position relative to the reference scan position is constructed based on the displacement information of each target scan position. The initial three-dimensional CBCT images of each target scan position are spatially transformed based on the initial spatial transformation matrix of each target scan position to achieve alignment between the initial three-dimensional CBCT images of each target scan position and the initial three-dimensional CBCT images of the reference scan position.
[0019] Furthermore, the aligned image is corrected to obtain a corrected image;
[0020] The corrected images are fused using a multi-scale fusion algorithm to obtain a three-dimensional CBCT mosaic image.
[0021] Furthermore, the correction of the aligned image includes:
[0022] Extract the overlapping areas of the reference scan position image and the images of each target scan position in the aligned image respectively;
[0023] Feature extraction and feature matching are performed on each of the overlapping regions, and the deviation value is calculated.
[0024] The initial spatial transformation matrix is corrected according to each corresponding deviation value to obtain the corresponding corrected spatial transformation matrix;
[0025] Based on each of the corrected spatial transformation matrices, the aligned image corresponding to each of the target scan positions is spatially transformed again to obtain the corrected image.
[0026] Furthermore, the obtained three-dimensional CBCT stitched image is post-processed, including grayscale equalization and / or artifact removal.
[0027] Secondly, the present invention also provides a CBCT multi-scan image fusion system, utilizing the CBCT multi-scan image fusion method described above, including,
[0028] The image acquisition module is used to acquire the raw CBCT image information of the reference scan position and each of the target scan positions;
[0029] The displacement information acquisition module is used to acquire displacement information of the reference scan position and each of the target scan positions;
[0030] The image reconstruction module is used to reconstruct the original CBCT image information of the reference scan position and each of the target scan positions respectively, so as to obtain the initial three-dimensional CBCT images of the reference scan position and each of the target scan positions.
[0031] The image alignment module is used to align the initial three-dimensional CBCT images of each target scan position with the initial three-dimensional CBCT image of the reference scan position to obtain the aligned image.
[0032] The image fusion module is used to perform image fusion on the aligned images using a multi-scale fusion algorithm to obtain a three-dimensional CBCT stitched image.
[0033] Furthermore, it also includes an image preprocessing module for preprocessing the obtained reference scan position and each of the initial three-dimensional CBCT images; the preprocessing module includes one or more combinations of a grayscale correction unit, a noise removal unit, and an edge enhancement unit.
[0034] Furthermore, it also includes an image correction module for correcting the alignment of the three-dimensional CBCT images of each of the target scan positions with the three-dimensional CBCT image of the reference scan position.
[0035] Furthermore, it also includes an image post-processing module to post-process the obtained three-dimensional CBCT stitched images; the post-processing module includes a grayscale equalization unit and / or an artifact removal module.
[0036] The present invention has the following beneficial effects: The CBCT multi-scan image fusion method of the present invention combines scan position displacement information to achieve alignment, and does not require a large amount of image feature calculation, thus greatly shortening the alignment time; the stitching fusion adopts a multi-scale fusion algorithm to process overlapping areas. This algorithm achieves pixel-level precise fusion through multi-scale decomposition and reconstruction, which can better preserve image details, eliminate stitching gaps, and further improve the quality of stitched images. The CBCT multi-scan image fusion method of the present invention can improve the automation level of CBCT multi-scan image stitching, and at the same time improve stitching accuracy and efficiency, solving the technical problems of weak automation, low accuracy and efficiency in existing CBCT multi-scan image stitching. Attached Figure Description
[0037] To make the objectives, technical solutions, and advantages of the invention clearer, the invention will now be described in further detail with reference to the accompanying drawings, wherein:
[0038] Figure 1 This is a flowchart illustrating the CBCT multi-scan image fusion method in an embodiment of the present invention.
[0039] Figure 2 This is a schematic diagram of the image fusion process in an embodiment of the present invention.
[0040] Figure 3 This is a schematic diagram of the image fusion module in an embodiment of the present invention. Detailed Implementation
[0041] The technical solutions of some embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments disclosed in the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments disclosed in the present invention are within the scope of protection of the present invention. It should be noted that in the drawings with reference numerals, the same reference numerals and letters represent similar parts. Once a part is defined in a drawing, it will not be defined and explained again in subsequent drawings.
[0042] This invention can be applied to the field of CBCT image stitching, and solves the technical problems of low automation, low accuracy and low efficiency in existing CBCT multi-scan image stitching.
[0043] The CBCT multi-scan image fusion method and system disclosed in this invention have the following technical effects:
[0044] Compared with the prior art, the present invention has the following advantages:
[0045] 1. This invention combines scan position displacement information to achieve initial alignment, and then performs fine correction through overlapping area feature matching, thus doubly ensuring alignment accuracy. The displacement information is acquired by a combination of grating ruler, encoder and IMU, which improves the accuracy of the displacement information and provides a reliable foundation for initial alignment. The fine correction process uses improved feature extraction and feature matching algorithms to further eliminate the deviation of the initial alignment and ensure that the stitched image is free of misalignment and distortion.
[0046] 2. The initial alignment based on displacement information in this invention does not require a large amount of image feature calculation, which greatly shortens the alignment time. At the same time, the entire stitching process is fully automated and does not require manual intervention, which reduces the workload of medical staff and avoids the inefficiency caused by manual operation, thus meeting the needs of rapid positioning in clinical radiotherapy.
[0047] 3. This invention eliminates the impact of image noise and grayscale deviation on stitching through preprocessing, removes erroneous matching points through the RANSAC algorithm, and ensures alignment through iterative correction, effectively improving the stability and anti-interference ability of the stitching process; the stitching fusion adopts a multi-scale fusion algorithm to process overlapping areas. This algorithm achieves pixel-level precise fusion through multi-scale decomposition and reconstruction, which can better preserve image details, eliminate stitching gaps, and further improve the quality of stitched images.
[0048] 4. This invention is applicable to multi-scan CBCT stitching of target areas of different locations and sizes, and is especially suitable for radiotherapy localization of large-scale tumors in the chest and abdomen. It can provide complete and clear large-scale imaging support for precise radiotherapy and has strong clinical applicability.
[0049] To further illustrate the CBCT multi-scan image fusion method and system provided by the present invention, the following embodiments are disclosed.
[0050] In some embodiments, please refer to Figure 1 This embodiment provides a CBCT multi-scan image fusion method, including:
[0051] The scanning object is controlled to move along a preset scanning path and stop at the reference scanning position and multiple target scanning positions respectively, and the original CBCT image information and corresponding displacement information of the reference scanning position and each of the target scanning positions are acquired simultaneously.
[0052] The original CBCT image information of the reference scan position and each of the target scan positions are reconstructed to obtain the initial three-dimensional CBCT images of the reference scan position and each of the target scan positions.
[0053] The initial three-dimensional CBCT images of each target scan position are aligned with the initial three-dimensional CBCT image of the reference scan position to obtain the aligned image.
[0054] A multi-scale fusion algorithm is used to fuse the aligned images to obtain a three-dimensional CBCT stitched image.
[0055] In some embodiments, the reference scan bit is the first scan bit, or a pre-defined reference scan bit.
[0056] In some embodiments, the scanning object is controlled to move along a preset scanning path and stop at a reference scanning position and multiple target scanning positions. In actual operation, the scanning object can be fixed on the surface of the treatment bed of the radiotherapy equipment, and the treatment bed of the radiotherapy equipment can be controlled to move along the preset scanning path to multiple scanning positions. When stopping at each scanning position, the corresponding area of the scanning object is scanned by the CBCT imaging system to obtain the raw CBCT projection data of each scanning position. At the same time, the displacement detection module of the treatment bed collects and records the displacement information of each scanning position relative to the reference scanning position in real time. The displacement information includes the translational displacement of the treatment bed in the X, Y, and Z directions to form a displacement information list. The displacement detection module adopts a combination of gantry sensors and encoders to achieve high-precision acquisition of displacement information. The acquisition frequency is synchronized with the CBCT scanning frequency to ensure that the CBCT data of each scanning position corresponds one-to-one with the displacement information.
[0057] Specifically, the original CBCT image information of the reference scan position and each of the target scan positions is reconstructed using the existing FDK algorithm. This is achieved by pre-weighting to compensate for geometric distortion caused by cone beam divergence, using a ramp filter to enhance edges and suppress blur, and then using a window function to reduce noise. Finally, the filtered projection of each angle is back-projected into 3D space and accumulated to obtain a three-dimensional image. In practice, commercially available software is used to reconstruct the original CBCT image information of the reference scan position and each of the target scan positions. The reconstruction process is not an improvement point for this technical solution and will not be described in detail here.
[0058] In some embodiments, after obtaining the reference scan position and the initial three-dimensional CBCT images of each scan position, the obtained reference scan position and the initial three-dimensional CBCT images are preprocessed; the preprocessing includes one or more combinations of grayscale correction, noise removal, and edge enhancement. Specifically, grayscale correction is used to eliminate image grayscale deviations caused by detector response inhomogeneity; noise removal employs a combination of median filtering and Gaussian filtering to remove quantum noise and electronic noise from the image; edge enhancement uses the Sobel operator to enhance the edge information of tissues and organs in the image, providing support for subsequent fine alignment.
[0059] Specifically, the Sobel operator used for edge enhancement is an edge detection / enhancement method based on first-order gradients: it uses two 3×3 convolutional kernels to calculate the gray-level change rate (approximate partial derivative) in the x and y directions, respectively. Because the gray-level change is fastest at the edges, locations with large gradient magnitudes correspond to edges; enhancing these locations highlights the contours of tissues and organs, facilitating subsequent registration / fine alignment. The Sobel convolutional kernel is shown in the following equation:
[0060] ;
[0061] in, The kernel is the convolution kernel in the x-direction. The kernel is the convolution kernel in the y-direction.
[0062] Corresponding gradient calculation ( (representing convolution) ;
[0063] in, Let be the rate of change of grayscale in the x-direction. y is the grayscale change rate in the y-direction.
[0064] In some embodiments, aligning the initial three-dimensional CBCT images of each target scan position with the initial three-dimensional CBCT image of the reference scan position includes:
[0065] An initial spatial transformation matrix of each target scan position relative to the reference scan position is constructed based on the displacement information of each target scan position. The initial three-dimensional CBCT images of each target scan position are spatially transformed based on the initial spatial transformation matrix of each target scan position to achieve alignment between the initial three-dimensional CBCT images of each target scan position and the initial three-dimensional CBCT images of the reference scan position.
[0066] Specifically, the initial spatial transformation matrix is a rigid transformation matrix, constructed based on a rigid body transformation model, and the formula is:
[0067] T(x,y,z)=R(x,y,z)+t; where R is the rotation matrix, corresponding to the rotational displacement about the X-axis, Y-axis, and Z-axis, and t is the translation vector, corresponding to the translational displacement in the X, Y, and Z directions.
[0068] In some embodiments, the aligned image is corrected to obtain a corrected image;
[0069] The corrected images are fused using a multi-scale fusion algorithm to obtain a three-dimensional CBCT mosaic image.
[0070] In some embodiments, correcting the aligned image includes:
[0071] Extract the overlapping areas of the reference scan position image and the images of each target scan position in the aligned image respectively;
[0072] Feature extraction and feature matching are performed on each of the overlapping regions, and the deviation value is calculated.
[0073] The initial spatial transformation matrix is corrected according to each corresponding deviation value to obtain the corresponding corrected spatial transformation matrix;
[0074] Based on each of the corrected spatial transformation matrices, the aligned image corresponding to each of the target scan positions is spatially transformed again to obtain the corrected image.
[0075] In some embodiments, the feature extraction employs an improved SIFT algorithm, which enhances the extraction capability of minute features by increasing the number of layers in the difference-of-Gaussian pyramid; the feature matching employs a FLANN matcher combined with the RANSAC algorithm to eliminate erroneous matching points and improve matching accuracy; the deviation value is calculated using the mean square error of grayscale (MSE) and the structural similarity index (SSIM) of the overlapping region. When the MSE is less than a preset threshold and the SSIM is greater than a preset threshold, the alignment is deemed qualified; otherwise, the initial spatial transformation matrix is iteratively corrected based on the deviation value until the qualification condition is met.
[0076] Specifically, SIFT descriptor Euclidean distance: the distance between two point descriptor vectors. , :
[0077] ;
[0078] in, , For the two selected points at Component values on the dimension.
[0079] Specifically, FLANN (Fast Library for Approximate Nearest Neighbors) is defined as an approximate nearest neighbor search, aiming to find the nearest neighbor in a database. Searching for:
[0080] ;
[0081] in, For the query results, The query descriptor (from a feature point to be registered). This is a feature database containing candidate descriptors.
[0082] For each Find the nearest neighbor distance d1 and the second nearest neighbor distance d2: ;in The threshold value is used.
[0083] In some embodiments, threshold The preferred value range is 0.7 to 0.8. Of course, in practical applications, it can be adjusted according to specific needs and actual data.
[0084] Specifically, RANSAC is used to eliminate mismatches and estimate the spatial transformation model, assuming the matching point pairs are... ,in From the reference scan f bits. From the target scan bit (target). Wherein, These are the coordinates of the i-th feature point in the reference bed image. These are the coordinates of the i-th feature point in the target bed image.
[0085] Perform 2D affine transformation:
[0086] ;
[0087] in, These are the coordinates of a point in the original image (target or moving); These are the coordinates of the points in the reference graph coordinate system after the affine transformation; It is a 2D affine transformation matrix (3×3, homogeneous coordinate form); These are linear transformation parameters, comprehensively representing rotation, scaling, and shearing. , Translation parameters.
[0088] RANSAC interior point determination is used to calculate reprojection error:
[0089] For a candidate transformation ( Prediction points are as follows: ;in, , This represents the process of restoring from homogeneous coordinates to two dimensions. These are the coordinates of the i-th feature point in the reference bed image.
[0090] The reprojection error is: ;in, The matching point for the target scan bit. This is the prediction point.
[0091] Set of interior points: ;in, This is the interior point threshold. Specifically, The preferred value range is 1 to 3 pixels.
[0092] The objective of the RANSAC optimal model is: ;in, Let this be the final set of interior points. And within the final set of interior points... Perform a least squares precise estimate: ;in, It is a 2D affine transformation matrix; The reference point is the point on the map (homogeneous coordinates are used in the calculation). The corresponding point (observation value) in the target image; π(⋅) represents the reconstruction from homogeneous coordinates to two dimensions; Let be the set of interior points.
[0093] Specifically, the calculation process for the deviation value is as follows:
[0094] First, calculate the mean squared difference between the two images in the overlapping region. ;
[0095] Let the set of pixels in the overlapping region be... Pixel count ; The grayscale value of the overlapping area of the reference scan bits; Define the grayscale of the overlapping region after the current transformation of the target scan bit; define the mean squared difference between the two images in the overlapping region:
[0096] ;
[0097] in, The grayscale distribution of the baseline bed image in the overlapping region; Ω represents the grayscale distribution of the target bed image in the overlapping region after correction by the current transformation matrix; Ω represents the set of voxels in the overlapping region after the two bed images are aligned. N represents the coordinates of a voxel (3D point) within the overlapping region; N = |Ω| represents the total number of voxels in the overlapping region.
[0098] Then, the structural similarity (SSIM) is calculated;
[0099] First, define the mean, variance, and covariance of the overlapping region:
[0100] ;
[0101] ;
[0102] ;
[0103] in, , The average brightness of the two images in the overlapping area; , Contrast / fluctuation intensity between the two images in the overlapping region; Covariance reflects whether the texture changes of the two images are synchronized; The grayscale distribution of the baseline bed image in the overlapping region; Ω represents the grayscale distribution of the target bed image in the overlapping region after correction by the current transformation matrix; Ω represents the set of voxels in the overlapping region after the two bed images are aligned. N represents the coordinates of a voxel (3D point) within the overlapping region; N = |Ω| represents the total number of voxels in the overlapping region.
[0104] The SSIM formula is defined as follows:
[0105] ;
[0106] in, , For is a constant, is Grayscale dynamic range, , This is a constant; in this embodiment, it is preferably set to 0.01 or 0.03. , The average brightness of the two images in the overlapping area; , Contrast / fluctuation intensity between the two images in the overlapping region; For covariance.
[0107] Then, perform iterative corrections: let the initial matrix be... , No. Wheel in RANSAC interior point set Upper estimation direct estimation : ;in, This is the new transformation matrix obtained in the (k+1)th iteration; To find the candidate T that minimizes the objective function among all candidates; Let be the set of interior points obtained from the k-th round of RANSAC filtering; This refers to the i-th point in the baseline graph; The observation point in the target graph that matches pi; In order to put The two-dimensional prediction points are obtained by T-transformation followed by dehomogenization; The squared reprojection error of the i-th interior point.
[0108] Then, a preset error threshold is set, and the iteration process is used to determine whether the error threshold is consistent. If it is consistent, the iteration is stopped.
[0109] In some embodiments, please refer to Figure 2 The aligned images are fused using a multi-scale fusion algorithm to obtain a three-dimensional CBCT stitched image. The multi-scale fusion algorithm includes:
[0110] First, the two images with overlapping areas are decomposed into layers of different scales using Gaussian pyramid decomposition.
[0111] Then, gradient-based weight allocation is performed on each scale layer to obtain fusion weights, and fusion is performed according to the fusion weights to achieve pixel-level fusion;
[0112] Then, the overlapping area image is obtained by reconstructing the Laplacian pyramid to ensure that the transition of the fused area is natural and the details are preserved, thus obtaining a 3D CBCT mosaic image.
[0113] Specifically, the formula derivation for the multi-scale fusion algorithm is as follows:
[0114] (1) Gaussian pyramid decomposition formula
[0115] Let the two images to be fused in the overlapping region be: the overlapping region of the reference scan image. Overlapping area with target scan image Gaussian pyramid decomposition was performed on the two images to obtain Gaussian layers at various scales. and ,in, ,in At the original image scale, This represents the total number of layers in the decomposition.
[0116] Gaussian pyramid decomposition is achieved through Gaussian filtering and downsampling. Gaussian layers are composed of the first The Gaussian layer is obtained by downsampling after Gaussian filtering, as shown in the formula:
[0117] ;
[0118] in, Represents a three-dimensional convolution operation. The three-dimensional Gaussian kernel function is expressed as follows: ; The standard deviation of the Gaussian kernel is preferred; take... , For the downsampling function, an intermittent sampling method is used to sample the th... The resolution of the layer image is reduced to half of its original value, meaning that the coordinates in the original image are retained. The pixels.
[0119] (2) Formula for calculating fusion weight
[0120] To preserve image edge and detail information, a gradient-based weighting strategy is employed, calculating the fusion weights for each Gaussian layer separately. First, the gradient values of each Gaussian layer are calculated; the gradient of the 3D image is calculated using the Sobel gradient operator. , , The gradient components in the three directions are respectively , , The formula for gradient magnitude is:
[0121] ;
[0122] An initial weight map is constructed based on the gradient magnitude. The larger the gradient magnitude, the richer the details in that region, and the greater the fusion weight should be assigned. The initial weight formula is:
[0123] ;
[0124] ;
[0125] in, , They are respectively , In the The initial weights of the Gaussian layer. To avoid local minima where the denominator is zero, and to make the weight transition smoother, the initial weight map is Gaussian smoothed. The final weight formula is:
[0126] ;
[0127] in, Use a Gaussian kernel for weight smoothing, and take... And satisfy .
[0128] (3) Gaussian layer blending formula
[0129] Pixel-level fusion is performed on the Gaussian layers at the corresponding scale based on the final weights to obtain the fused Gaussian layers. The formula is:
[0130] ;
[0131] (4) Laplace's Pyramid Reconstruction Formula
[0132] The Pyramid of Laplace is constructed by the differences between the levels of the Pyramid of Gauss. It was built first... , Corresponding Laplace Pyramid , Then, based on the decomposed Gaussian pyramid, a fused Laplace pyramid is constructed. The final image was obtained by reconstructing the Laplace pyramid.
[0133] The formula for calculating the Laplacian layer is:
[0134] ;
[0135] in, For the upsampling function, the first... The resolution of the layer image is increased to twice its original value, and blank pixels are filled using bilinear interpolation; when When it's the topmost Gaussian layer, the corresponding Laplacian layer... .
[0136] The formula for constructing a merged Laplacian layer is:
[0137] ;
[0138] The final fused image is obtained by progressively reconstructing it from the top to the bottom of the Laplacian pyramid. The reconstruction formula is as follows:
[0139] ;
[0140] in, This refers to the fused image of the overlapping areas.
[0141] In some embodiments, the obtained three-dimensional CBCT stitched images are post-processed, including grayscale equalization and / or artifact removal, to optimize image quality.
[0142] In some embodiments, this embodiment also provides a CBCT multi-scan image fusion system, which utilizes the CBCT multi-scan image fusion method described above, including,
[0143] The image acquisition module is used to acquire the raw CBCT image information of the reference scan position and each of the target scan positions;
[0144] The displacement information acquisition module is used to acquire displacement information of the reference scan position and each of the target scan positions;
[0145] The image reconstruction module is used to reconstruct the original CBCT image information of the reference scan position and each of the target scan positions respectively, so as to obtain the initial three-dimensional CBCT images of the reference scan position and each of the target scan positions.
[0146] The image alignment module is used to align the initial three-dimensional CBCT images of each target scan position with the initial three-dimensional CBCT image of the reference scan position to obtain the aligned image.
[0147] For the image fusion module, please refer to [link / reference]. Figure 3 It is used to perform image fusion on the aligned images using a multi-scale fusion algorithm to obtain a three-dimensional CBCT stitched image.
[0148] In some embodiments, the system further includes an image preprocessing module for preprocessing the obtained reference scan position and each of the initial three-dimensional CBCT images; the preprocessing module includes one or more combinations of a grayscale correction unit, a noise removal unit, and an edge enhancement unit.
[0149] In some embodiments, the system further includes an image correction module for correcting the alignment of the three-dimensional CBCT images of each of the target scan positions with the three-dimensional CBCT image of the reference scan position.
[0150] In some embodiments, the system further includes an image post-processing module for post-processing the obtained three-dimensional CBCT stitched images; the post-processing module includes a grayscale equalization unit and / or an artifact removal module.
[0151] In the description disclosed in this invention, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to has a specific orientation, or is constructed and operated in a specific orientation. Therefore, they should not be construed as limiting the disclosure of this invention.
[0152] Unless the context otherwise requires, throughout the specification and claims, the term "comprising" is interpreted as open-ended and encompassing, meaning "including, but not limited to." In the description, terms such as "one embodiment," "some embodiments," "exemplary embodiment," "exemplary," or "some examples" are intended to indicate that a particular feature, structure, material, or characteristic associated with that embodiment or example is included in at least one embodiment or example disclosed in the invention. The illustrative representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics mentioned may be included in any suitable manner in any one or more embodiments or examples.
[0153] The terms "first" and "second" are used merely to distinguish different descriptive objects and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated; that is, they do not limit the position, order, priority, quantity, or content of the described objects. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the embodiments disclosed in this invention, unless otherwise stated, "a plurality of" means two or more.
[0154] In describing some embodiments, the term "connection" and its derivative expressions may be used. For example, the term "connection" may be used in describing some embodiments to indicate that two or more components have direct physical or electrical contact with each other. The embodiments disclosed herein are not necessarily limited to the content of this document.
[0155] "At least one of A, B, and C" has the same meaning as "at least one of A, B, or C," both including the following combinations of A, B, and C: only A, only B, only C, a combination of A and B, a combination of A and C, a combination of B and C, and a combination of A, B, and C. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B; "A and / or B" includes the following three combinations: only A, only B, and a combination of A and B.
[0156] The use of “configured as” in this article implies an open and inclusive language that does not exclude the applicability to or configuration of devices to perform additional tasks or steps.
[0157] In addition, the use of “based on” implies openness and inclusivity, because processes, steps, calculations or other actions “based on” one or more of the stated conditions or values may in practice be based on additional conditions or values beyond those stated.
[0158] As used herein, “about” and “approximately” include the values stated and the average values within an acceptable range of deviation from a particular value, wherein the acceptable range of deviation is determined by a person skilled in the art taking into account the measurement under discussion and the error associated with the measurement of the particular quantity (i.e., the limitations of the measurement system).
[0159] This document describes exemplary embodiments with reference to cross-sectional views and / or plan views, which are idealized exemplary drawings. In the drawings, the thickness of layers and regions is enlarged for clarity. Therefore, variations in shape relative to the drawings are contemplated due to, for example, manufacturing techniques and / or tolerances. Thus, exemplary embodiments should not be construed as limited to the shapes of the regions shown herein, but rather include shape deviations due to, for example, manufacturing processes. For example, etched regions shown as rectangular would typically have curved features. Therefore, the regions shown in the drawings are schematic in nature, and their shapes are not intended to show the actual shapes of the regions of the device, nor are they intended to limit the scope of the exemplary embodiments.
[0160] The above description is merely a specific embodiment of the present invention, but the scope of protection disclosed in the present invention is not limited thereto. Any variations or substitutions conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection disclosed in the present invention. Therefore, the scope of protection disclosed in the present invention should be determined by the scope of the claims.
Claims
1. A CBCT multi-scan artifact image fusion method, characterized in that, include: The scanning object is controlled to move along a preset scanning path and stop at the reference scanning position and multiple target scanning positions respectively, and the original CBCT image information and corresponding displacement information of the reference scanning position and each of the target scanning positions are acquired simultaneously. The original CBCT image information of the reference scan position and each of the target scan positions are reconstructed to obtain the initial three-dimensional CBCT images of the reference scan position and each of the target scan positions. The initial three-dimensional CBCT images of each target scan position are aligned with the initial three-dimensional CBCT image of the reference scan position to obtain the aligned image. A multi-scale fusion algorithm is used to fuse the aligned images to obtain a three-dimensional CBCT stitched image.
2. The CBCT multi-scan artifact image fusion method of claim 1, wherein, After obtaining the reference scan position and the initial three-dimensional CBCT image of each scan position, the obtained reference scan position and the initial three-dimensional CBCT image are preprocessed; the preprocessing includes one or more combinations of grayscale correction, noise removal and edge enhancement.
3. The CBCT multi-scan artifact image fusion method of claim 1, wherein, Aligning the initial 3D CBCT images of each target scan position with the initial 3D CBCT image of the reference scan position, including: An initial spatial transformation matrix of each target scan position relative to the reference scan position is constructed based on the displacement information of each target scan position. The initial three-dimensional CBCT images of each target scan position are spatially transformed based on the initial spatial transformation matrix of each target scan position to achieve alignment between the initial three-dimensional CBCT images of each target scan position and the initial three-dimensional CBCT images of the reference scan position.
4. The CBCT multi-scan artifact image fusion method of claim 3, wherein, The aligned image is corrected to obtain the corrected image; The corrected images are fused using a multi-scale fusion algorithm to obtain a three-dimensional CBCT mosaic image.
5. The CBCT multi-scan artifact image fusion method of claim 4, wherein, Correcting the aligned image includes: Extract the overlapping areas of the reference scan position image and the images of each target scan position in the aligned image respectively; Feature extraction and feature matching are performed on each of the overlapping regions, and the deviation value is calculated. The initial spatial transformation matrix is corrected according to each corresponding deviation value to obtain the corresponding corrected spatial transformation matrix; Based on each of the corrected spatial transformation matrices, the aligned image corresponding to each of the target scan positions is spatially transformed again to obtain the corrected image.
6. The CBCT multi-scan artifact image fusion method of claim 1, wherein, The obtained three-dimensional CBCT stitched image is post-processed, including grayscale equalization and / or artifact removal.
7. A CBCT multi-scan image fusion system, utilizing the CBCT multi-scan image fusion method as described in claim 1, characterized in that, The image acquisition module is used to acquire the raw CBCT image information of the reference scan position and each of the target scan positions; The displacement information acquisition module is used to acquire displacement information of the reference scan position and each of the target scan positions; The image reconstruction module is used to reconstruct the original CBCT image information of the reference scan position and each of the target scan positions respectively, so as to obtain the initial three-dimensional CBCT images of the reference scan position and each of the target scan positions. The image alignment module is used to align the initial three-dimensional CBCT images of each target scan position with the initial three-dimensional CBCT image of the reference scan position to obtain the aligned image. The image fusion module is used to perform image fusion on the aligned images using a multi-scale fusion algorithm to obtain a three-dimensional CBCT stitched image.
8. The CBCT multi-scan image fusion system according to claim 7, characterized in that, It also includes an image preprocessing module for preprocessing the obtained reference scan position and each of the initial three-dimensional CBCT images; the preprocessing module includes one or more combinations of a grayscale correction unit, a noise removal unit and an edge enhancement unit.
9. The CBCT multi-scan image fusion system according to claim 7, characterized in that, It also includes an image correction module for correcting the alignment of the three-dimensional CBCT images of each of the target scan positions with the three-dimensional CBCT image of the reference scan position.
10. The CBCT multi-scan image fusion system according to claim 7, characterized in that, It also includes an image post-processing module to post-process the obtained three-dimensional CBCT stitched images; the post-processing module includes a grayscale equalization unit and / or an artifact removal module.