Tomographic fusion imaging method and system based on multi-scale information fusion
Through multi-angle acquisition and adaptive enhancement methods, the problem of unstable fusion of three-dimensional tomography and two-dimensional projection images is solved, and high-quality S2D images are generated, which improves the diagnostic effect of breast tomography.
Patent Information
- Application Number
- CN202510368291.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-26
- Publication Date
- 2025-07-08
AI Technical Summary
The prior art cannot fusion three-dimensional tomographic images with two-dimensional projected images stably, resulting in low quality of the generated S2D images, especially in breast tomographic fusion imaging, problems such as uneven brightness, excessive skin lines, and metal artifacts, which affect the diagnostic effect.
By acquiring projected images from multiple angles, performing pre-processing and adaptive enhancement at different scales, three-dimensional tomographic images are generated, and information fusion is carried out at different scales to generate high-quality S2D images.
提高了S2D图像的质量,减少噪声,增强病灶显示,降低伪影,提升了诊断的准确性和图像质量。
Smart Images
Figure CN120279227A_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present application relate to the technical field of data processing, and in particular, to a tomographic fusion imaging method and system based on multi-scale information fusion. Background Art
[0002] Tomographic fusion imaging technology is a three-dimensional imaging technology based on X-rays, which can be used for medical diagnosis, industrial non-destructive testing, etc. Its main applications in medical diagnosis include: breast tomographic fusion imaging, chest tomographic fusion imaging, large-range parallel scan tomographic imaging, etc. The basic principle of tomographic fusion imaging technology is that an X-ray source exposes the object to be photographed at multiple different positions, and a flat panel detector generates a two-dimensional projection image by sensing the X-rays attenuated by the object and transmits it to the host computer. The host computer uses a specific algorithm to reconstruct these projection images according to the relative positions of the ray source focus, the detector, and the object to be photographed at the exposure positions of each projection image, obtains a three-dimensional tomographic image, stores the three-dimensional tomographic image on a disk or transmits it to other imaging systems, and displays it for doctors to diagnose the condition.
[0003] In a tomographic fusion imaging system, reconstructing a three-dimensional tomographic image from two-dimensional projection images is a key link. Conventional reconstruction algorithms include filtered back-projection reconstruction algorithms, iterative reconstruction algorithms, etc. Although the basic principles and the most basic applications of these algorithms are easily available in this technical field. However, how to optimize the parameters involved in these algorithms and how to improve the algorithms to adapt to different imaging systems or different imaging postures, parts, etc. is a crucial issue and also a difficult problem.
[0004] In practical applications, when the filtered back-projection reconstruction algorithm is used for breast tomographic fusion imaging reconstruction, there are the following two problems: the angular range of object sampling by the imaging system is generally less than 50 degrees, which belongs to limited-angle imaging. After reconstruction, problems such as uneven brightness in the breast area and excessive skin lines appear in the obtained three-dimensional tomographic image; and, due to the filtering of the projections of high-density objects such as calcified lesions and metal implants, metal artifacts, calcification artifacts, etc. will appear. Therefore, the quality of the tomographic image will be reduced, thus interfering with the doctor's diagnosis.
[0005] Further, the breast tomosynthesis fusion imaging system can generate a synthetic two-dimensional projection image (hereinafter referred to as the S2D image) based on the reconstructed three-dimensional tomographic image and the original projection image, so as to replace the separate FFDM (full-field digital mammography) image. The positive significance of the replacement of the S2D image for the FFDM image is as follows: shortening the examination time, reducing the pain of the patient, reducing the overall irradiation dose of the patient, and being able to provide high-quality two-dimensional images. The S2D image is obtained by fusing the information of the three-dimensional tomographic image and the two-dimensional projection image, so as to enhance the lesions and breast tissue structure, and reduce noise and artifacts.
[0006] In the prior art, it is impossible to stably fuse the information of the three-dimensional tomographic image and the two-dimensional projection image, resulting in a reduction in the quality of the generated S2D image. For example, in the process of generating the S2D image, it is necessary to detect lesions or objects of interest from the three-dimensional tomographic image, and further calculate the weights according to the detection results. This method is difficult to adapt to the changing clinical scenarios. Factors such as the type, shape, and gray-scale characteristics of the lesions or objects of interest will affect the detection results, resulting in uncertainty in the final image effect. Summary of the Invention
[0007] The embodiments of the present application provide a tomosynthesis fusion imaging method and system based on multi-scale information fusion, which can generate high-quality S2D images.
[0008] On the one hand, the embodiments of the present application provide a tomosynthesis fusion imaging method based on multi-scale information fusion, including:
[0009] Collecting projection images of the target object from multiple angles;
[0010] Performing preprocessing on the collected projection images;
[0011] Performing adaptive enhancement on the preprocessed projection images at different scales;
[0012] Performing backprojection using the enhanced projection images to generate a three-dimensional tomographic image;
[0013] Performing information fusion on the projection images at a preset angle in the three-dimensional tomographic image and the enhanced projection images at different scales to obtain a synthesized two-dimensional image.
[0014] On the one hand, the embodiments of the present application also provide a tomosynthesis fusion imaging system based on multi-scale information fusion, including: a projection acquisition device and a processor;
[0015] The projection acquisition device is configured to collect projection images of the target object from multiple angles;
[0016] The processor is configured to perform pre - processing on the acquired projection images, perform adaptive enhancement on the pre - processed projection images at different scales, and perform back - projection using the enhanced projection images to generate three - dimensional tomographic images, and perform information fusion on the three - dimensional tomographic images and the projection images at a preset angle in the enhanced projection images at different scales to obtain synthesized two - dimensional images.
[0017] In various embodiments of the present application, by acquiring projection images of a target object from multiple angles, performing pre - processing on the acquired projection images, performing adaptive enhancement on the pre - processed projection images at different scales, performing back - projection using the enhanced projection images to generate three - dimensional tomographic images, and performing information fusion on the three - dimensional tomographic images and the projection images at a preset angle in the enhanced projection images at different scales to obtain synthesized two - dimensional images, the quality of the synthesized two - dimensional images can be improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those skilled in the art, without creative efforts, other drawings can also be obtained based on these drawings.
[0019] Figure 1 It is a flowchart of a tomographic fusion imaging method based on multi - scale information fusion provided in this embodiment;
[0020] Figure 2 It is a structural schematic diagram of a projection acquisition device in the tomographic fusion imaging method based on multi - scale information fusion provided in this embodiment;
[0021] Figure 3 It is a flowchart of another tomographic fusion imaging method based on multi - scale information fusion provided in this embodiment;
[0022] Figure 4 It is an example diagram of a three - dimensional tomographic image after back - projection reconstruction in the tomographic fusion imaging method based on multi - scale information fusion provided in an embodiment of the present invention;
[0023] Figure 5 It is a flowchart of another tomographic fusion imaging method based on multi - scale information fusion provided in this embodiment;
[0024] Figure 6 It is an example diagram of an S2D image in the tomographic fusion imaging method based on multi - scale information fusion provided in an embodiment of the present invention;
[0025] Figure 7It is a schematic structural diagram of a tomosynthesis imaging system based on multi-scale information fusion provided by an embodiment of the present invention. Detailed implementation manners
[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Apparently, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.
[0027] In the following, the terms "including", "having", and their cognates that can be used in various embodiments of the present application are only intended to indicate specific features, numbers, steps, operations, elements, components, or combinations of the foregoing items, and should not be construed as first excluding the existence of one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing items or increasing the possibility of one or more features, numbers, steps, operations, elements, components, or combinations of the foregoing items.
[0028] In addition, the terms "first", "second", "third", etc. are only used for descriptive distinction and cannot be construed as indicating or implying relative importance.
[0029] Unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which various embodiments of the present application belong. The terms (such as those defined in a general-use dictionary) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having an idealized meaning or an overly formal meaning unless clearly defined in various embodiments of the present application.
[0030] Please refer to Figure 1 , Figure 1 which shows a tomosynthesis imaging method based on multi-scale information fusion. This method is applied to an X-ray tomosynthesis imaging system, which can specifically be a breast X-ray tomosynthesis imaging system, a chest X-ray tomosynthesis imaging system, or an extremity X-ray tomosynthesis imaging system. This method includes:
[0031] S101. Acquire projection images of a target object from multiple angles;
[0032] The X-ray tomosynthesis imaging system may include: a projection acquisition device, a memory, a processor, and a display;
[0033] Among them, the projection acquisition device is connected to the memory and the processor, and the processor is connected to the display.
[0034] Please refer to Figure 2 , the projection acquisition device may include: an X-ray source 21, an X-ray detector 22, a compression device 23, and a gantry 24.
[0035] Among them, the X-ray source 21 and the gantry 24 are connected by a fixing device ( Figure 2 not shown in the figure), and this fixing device makes a fan-shaped movement and can drive the X-ray source 21 to rotate relative to the gantry 24.
[0036] The projection acquisition device is used to obtain Figure 2 the original two-dimensional projected image of the target object 25 to be photographed in the figure. In practical applications, the target object 25 is various human tissue parts for X-ray examinations;
[0037] This two-dimensional projected image is transmitted to a memory and a processor. The processor is used to generate a three-dimensional tomographic image and an S2D image based on this two-dimensional projected image. The memory is used to store data such as images, algorithms, and various preset parameters. The display is used to display the received images.
[0038] During application, the target object 25 is compressed by the compression device 23 to separate the overlapping tissue structures of the target object 25, so as to more easily display the tissue abnormalities of the target object 25.
[0039] During the shooting process, the X-ray detector 22 is stationary, and the focus of the X-ray source 21 moves on a fan-shaped orbit.
[0040] By rotating the X-ray source 21, multiple frames of projected images of the target 25 at different angles can be collected. The angle interval is generally equally spaced. The angle range for collecting the projected images is symmetric with the center 0 degrees as the reference. This center 0 degrees refers to the angle when the X-ray source 21 is directly facing the target object 25. The angle range for the imaging system to sample the target object is generally less than 50 degrees, which belongs to limited-angle imaging and will cause problems such as uneven brightness in the target object area of the three-dimensional tomographic image and over-bright skin lines. The tomographic fusion imaging method and system provided in the embodiments of the present application are dedicated to solving these problems.
[0041] S102. Perform preprocessing on the collected projected images;
[0042] The preprocessing includes pre-treatment and filtering;
[0043] The pre-treatment refers to the basic processing of the image, such as correction, etc.;
[0044] The filtering is to filter the projected image after the pre-treatment.
[0045] In the filtering process, the selection of the filtering function is crucial, and the choice of the filtering function will affect the quality of the final tomographic image. Filtering functions such as Ram-Lak and Shepp-Logan can be selected. In the embodiment of the present application, for the tomographic fusion imaging method based on multi-scale information fusion, common filtering functions can be used, reducing the dependence on the filtering function and making the overall image quality better and more controllable.
[0046] S103. Perform adaptive enhancement on the preprocessed projection image at different scales;
[0047] The adaptive enhancement includes suppressing the parts that need to be weakened and enhancing the parts that need to be enhanced in the projection image, so as to make the projection image clearer.
[0048] Specifically, referring to Figure 3 , step S103 includes the following steps:
[0049] S1031. Construct a first Laplacian pyramid based on the filtered projection image;
[0050] According to the construction method of the Laplacian pyramid, decompose all the filtered projection images into a pyramid to generate a set of first Laplacian pyramids. The number of layers n of the first Laplacian pyramid can be appropriately adjusted according to the resolution of the projection image. The first Laplacian pyramid is composed of projection images at n scales. Let the j-th layer pyramid image of the i-th frame projection image be L i,j , where 0 ≤ i ≤ m - 1, 0 ≤ j ≤ n - 1, and m is the total number of projection images.
[0051] S1032. Calculate the adaptive stretching reference coefficients of each layer of projection images in the first Laplacian pyramid at each scale;
[0052] The reference coefficient includes a positive part reference coefficient G P-j and a negative part reference coefficient G N-j . In the filtered projection image, the gray-scale distributions of the positive part reference coefficient and the negative part reference coefficient are different.
[0053] The specific calculation method is as follows:
[0054] From all the filtered projection images, segment the target object from the projection images containing invalid regions such as air to obtain the first mask M i . Among them, the segmentation method can be direct gray-scale segmentation, clustering, Otus, etc.
[0055] Furthermore, according to the characteristics of the first Laplacian pyramid, for the first mask M iPerform multi-scale reduction to obtain the second mask M of the target object in each projection image at different scales i,j . Among them, the characteristics of the first Laplacian pyramid refer to the sizes at different scales.
[0056] According to the second mask M at different scales i,j , count the maximum and minimum values of all rows or columns within the tissue range of the target object in the i-th frame projection image corresponding to the second mask M i,j in the j-th layer pyramid image L i,j . Calculate the mean value G of the maximum values of all rows or columns and the mean value G max-i,j of the minimum values respectively min-j,j , and calculate the mean value G max-j of the maximum values and the mean value G min-j of the minimum values for all projection images at each scale.
[0057] . Among them, the tissue range of the target object refers to the image area represented on the second mask. The second mask is a binary image. In one example, 1 represents the tissue area of the target object and 0 represents the invalid background area.
[0058] The maximum value and the minimum value are collectively called the extreme values. Counting the mean value of the extreme values of all rows means first finding the extreme values of each row and then averaging the extreme values corresponding to all rows. In this way, the average value obtained represents the extreme values of all rows in the image, which is more reasonable than directly using the extreme values in the image as the extreme values of all rows. Counting the mean value of the extreme values of all columns is the same. That is, for a specific scale, each projection image at this scale has its own extreme values, and the extreme values obtained from each projection are averaged again.
[0059] It should be noted that
[0060] First, when using a conventional filtering function for filtering, it is certain that: G max-j >0, G min-j <0;
[0061] Second, averaging the extreme values of all projection images instead of allowing each projection image to use its own extreme values for subsequent processing can ensure the consistency of all projection images in terms of noise, balance, etc., so that the final generated image effect is stable.
[0062] Furthermore, determine the adaptive stretching reference coefficients G P-j , and G N-j :
[0063] G P-j = Coef P-j * G max-j , G N-j = Coef N-j * Gmin-j 。
[0064] Among them, Coef P-j and Coef N-j are proportionality coefficients, which can be adjusted according to the requirements of actual applications. The proportionality coefficients are different for each scale.
[0065] S1033. Stretch each layer of the projection image in the first Laplacian pyramid according to the calculated adaptive stretching reference coefficient to obtain a second Laplacian pyramid;
[0066] Specifically, let the pixel value of a certain pixel in the j-th layer pyramid image L i,j of the i-th frame projection image be g, and the calculation method of stretching g to g' is as follows:
[0067]
[0068] Among them, C N-j and C P-j are the negative part coefficient and the positive part coefficient respectively, and the value range is [0.1, 2].
[0069] Coef P-j 、Coef N-j 、C N-j and C P-j together determine the effect of multi-scale stretching of the filtered projection image. Different parameter combinations can produce different image effects. By adjusting these parameters, in actual applications, image noise suppression, metal artifacts, edge brightness adjustment, overall image equalization, etc. can achieve ideal effects.
[0070] S1034. Reconstruct the second Laplacian pyramid to obtain an enhanced projection image.
[0071] Reconstruct the second Laplacian pyramid obtained after stretching to obtain an enhanced projection image. This enhanced projection image is equivalent to reconstructing the filtered projection image of the first Laplacian pyramid constructed in step S1031.
[0072] The reconstruction algorithm in this embodiment is not specifically limited.
[0073] S104. Use the enhanced projection image for backprojection to generate a three-dimensional tomographic image;
[0074] Use the enhanced reconstructed filtered projection image obtained in step S103 for backprojection reconstruction to obtain a set of three-dimensional tomographic images. The backprojection algorithm is not specifically limited.
[0075] There are wavy artifacts on the three-dimensional tomographic image, which not only affect the appearance of the three-dimensional tomographic image itself but also affect the subsequent generation of S2D. Therefore, such artifacts must be removed. The specific removal method is as follows:
[0076] Reduce the size of the first mask M of the target object i to 1 / 4 or 1 / 8 of the original size to obtain a third mask with a reduced size;
[0077] Perform back-projection reconstruction on the third mask to obtain a mask of the three-dimensional tomographic image;
[0078] Determine whether the pixels in the mask of the three-dimensional tomographic image are in the tomographic effective region;
[0079] Specifically, it is determined whether a pixel is in the tomographic effective region by judging whether each pixel in the mask of the three-dimensional tomographic image is covered by projection images at all angles. If a pixel is covered by projection images at all angles, it is determined that the pixel is in the tomographic effective region;
[0080] In the embodiment of the present invention, the effective region refers to the effective tissue region obtained after excluding the above-mentioned ripple artifact region and air region.
[0081] Set the values of the pixels that are not in the tomographic effective region to 0 in the three-dimensional tomographic image.
[0082] By setting the pixel points that are not in the tomographic effective region in the three-dimensional tomographic image to 0, the artifacts in the three-dimensional tomographic image are removed, the quality of the three-dimensional tomographic image is improved, and problems such as metal artifacts and calcification artifacts caused by the filtering of projections of high-density objects such as calcified lesions and metal implants can be improved.
[0083] In another embodiment, the mask of the three-dimensional tomographic image can also be reduced to 1 / 4 or 1 / 8 on the scale of the real tomographic image, and when judging whether a pixel is in the tomographic effective region, it can be remapped to the original size, which can greatly reduce the occupation of the storage module by the program, reduce the load of the calculation module, and shorten the calculation time.
[0084] An example of the three-dimensional tomographic image after back-projection reconstruction is Figure 4 shown.
[0085] S105. Perform information fusion on the projection images at a preset angle in the three-dimensional tomographic image and the enhanced projection image at different scales to obtain a synthesized two-dimensional image.
[0086] Specifically, referring to Figure 5 , step S105 includes the following steps:
[0087] S1051. Generate the maximum intensity projection image of the three-dimensional tomographic image;
[0088] Select the projection image of the preset projection angle in the enhanced projection image, and generate the system projection matrix P of the preset projection angle according to the system parameters. Specifically, the system parameters can be obtained based on the geometric spatial positions of the detector, the X-ray tube, and the object being photographed corresponding to each frame of projection.
[0089] Preferably, the preset projection angle can be the 0 projection angle.
[0090] The 0 projection angle is the projection angle when the central ray of the X-ray tube is perpendicular to the detector;
[0091] The preset projection angle can also be the angle closest to the 0 projection angle, where "closest" means that the angular deviation between the projection position and the theoretical 0-angle position is the smallest.
[0092] Equivalent the three-dimensional tomographic image to the target object, and calculate the maximum intensity projection (MIP) image of the target object on the real detector according to the projection matrix P.
[0093] S1052. Generate two sets of Laplacian pyramids respectively according to the projection image of the preset projection angle in the enhanced projection image and the maximum intensity projection image.
[0094] Taking the 0 projection angle as an example of the preset projection angle, perform pyramid decomposition on the projection image Proj0 of the 0 projection angle in the enhanced projection image in step S103 and the MIP respectively to generate 2 sets of Laplacian pyramids. The number of layers of these 2 sets of Laplacian pyramids is also n, and each layer of pyramid image is L Proj0-j , and L MIP-j .
[0095] S1053. Fuse the two sets of Laplacian pyramids at different scales to generate a target pyramid, and reconstruct the target pyramid to obtain the S2D image.
[0096] Fuse the two sets of Laplacian pyramids at multiple scales to obtain a new Laplacian pyramid L S2D :
[0097]
[0098] where a j and c j are the weight coefficients at their respective scales, and b j and d j are the enhancement coefficients at their respective scales. The values of these coefficients will directly affect the overall quality of the final S2D image. At different scales, L Proj0-j and L MIP-jThe image information represented by the two has its own strengths. By adjusting these coefficients, the noise level of the S2D image can be significantly reduced, artifacts can be weakened, lesions can be enhanced, and the overall image performance can surpass the larger dose FFDM image.
[0099] For example, if the dose of a single-frame projection image is low, the projection image will show strong noise, and the same is true for the projection image Proj0 at 0 projection angle. The noise is generally a high-frequency component. In the Laplacian pyramid of the projection image at 0 projection angle, it is mainly manifested in the top layer L Proj0-0 , and L Proj0-0 The real image information contained in it is less. MIP-0 In the image, not only is the noise component very small, but it also contains a large number of subtle tissue structures, lesions, etc. Therefore, by controlling the top coefficients a0, b0, c0, and d0 in the two groups of Laplace pyramids, the weights of the two groups of Laplace pyramids can be controlled to achieve the purpose of enhancing lesions and reducing noise in the S2D image to be constructed.
[0100] Furthermore, the fused L S2D Reconstruction is performed to obtain a final S2D image, and the reconstruction algorithm is not specifically limited.
[0101] An example of an S2D image is Figure 6 shown.
[0102] It should be noted that the pyramid decomposition and reconstruction steps used in the tomographic fusion imaging method based on multi-scale information fusion can also be implemented by a multi-scale analysis method based on wavelet transform. The relevant images are decomposed layer by layer by using scale functions and wavelet functions, and then each layer of the image is enhanced or suppressed, and fused in a similar manner to the above steps, so as to achieve the same decomposition and reconstruction effect as the pyramid. The parts of the method involving the pyramid can all be implemented by the wavelet transform algorithm.
[0103] The tomographic fusion imaging method based on multi-scale information fusion in the above embodiment is also applicable to other medical X-ray tomographic fusion imaging systems other than breast X-ray tomographic fusion systems, such as chest X-ray tomographic fusion imaging systems, limbs X-ray tomographic fusion imaging systems, and large-scale linear scanning tomographic fusion systems, and can also be used for industrial X-ray tomographic imaging technology, etc.
[0104] In the embodiment of the present application, a tomographic fusion imaging method based on multi-scale information fusion is provided. By collecting projection images of a target object from multiple angles, preprocessing the collected projection images, adaptively enhancing the preprocessed projection images at different scales, using the enhanced projection images for backprojection to generate a three-dimensional tomographic image, and performing information fusion on the three-dimensional tomographic image and the projection images at a preset angle in the enhanced projection images at different scales to obtain a synthesized two-dimensional image, the quality of the synthesized two-dimensional image can be improved.
[0105] See Figure 7 , the embodiment of the present application also provides a tomographic fusion imaging system based on multi-scale information fusion, including: a projection acquisition device 701 and a processor 702, where the processor 702 is arranged in an electronic device, and the electronic device can be a computer with data processing capabilities, connected to the projection acquisition device 701, and can obtain data such as projection images from the projection acquisition device 701;
[0106] The projection acquisition device 701 is used to collect projection images of a target object from multiple angles;
[0107] The processor 702 is used to preprocess the collected projection images, perform adaptive enhancement on the preprocessed projection images at different scales, and use the enhanced projection images for backprojection to generate a three-dimensional tomographic image, and further, perform information fusion on the three-dimensional tomographic image and the projection images at a preset angle in the enhanced projection images at different scales to obtain a synthesized two-dimensional image.
[0108] The projection acquisition device 701 and the processor 702 are the projection acquisition device and the processor in the foregoing embodiment.
[0109] Further, the processor 702 is further used to perform preprocessing including preset items on the collected projection images, and perform filtering processing on the projection images after preprocessing to obtain filtered projection images.
[0110] Further, the processor 702 is further used to construct a first Laplacian pyramid according to the filtered projection images, and the first Laplacian pyramid includes projection images at multiple scales;
[0111] Calculate the adaptive stretching reference coefficients of the projection images at each layer in the first Laplacian pyramid at each scale;
[0112] Stretch the projection images at each layer in the first Laplacian pyramid according to the adaptive stretching reference coefficients to obtain a second Laplacian pyramid;
[0113] Reconstruct the Laplacian pyramid to obtain enhanced projection images.
[0114] Further, the adaptive stretching reference coefficient includes a positive part reference coefficient and a negative part reference coefficient;
[0115] The processor 702 is further configured to segment the projection image of the target object from each of the filtered projection images, and obtain a first mask of the target object according to the projection image of the target object;
[0116] According to the scale features of each layer of the first Laplacian pyramid, perform corresponding scale reduction on the mask to obtain a second mask of the target object at each scale of each projection image of the first Laplacian pyramid;
[0117] According to the second mask of the target object at each scale, count the maximum value and the minimum value of the projection images of each layer of the first Laplacian pyramid in all rows or all columns within the tissue range of the target object corresponding to the second mask, and calculate the mean value of the maximum values and the mean value of the minimum values in all rows or all columns respectively, and calculate the mean value of the maximum values and the mean value of the minimum values of all projection images at each scale;
[0118] Determine the positive part reference coefficient of the adaptive stretching reference coefficient according to the mean value of the maximum values of all projection images at each scale and a preset first proportionality coefficient, and determine the negative part reference coefficient of the adaptive stretching reference coefficient according to the mean value of the minimum values of all projection images at each scale and a preset second proportionality coefficient.
[0119] Further, the processor 702 is further configured to remove the artifacts of the three-dimensional tomographic image.
[0120] Specifically, the processor 702 is further configured to reduce the size of the first mask of the target object to a preset proportion of the original size to obtain a third mask with a reduced size;
[0121] Perform back-projection reconstruction on the third mask to obtain a mask of the three-dimensional tomographic image;
[0122] Determine whether each pixel in the mask of the three-dimensional tomographic image is in a tomographic effective area;
[0123] Set the pixel value of the pixel not in the tomographic effective area to 0 in the three-dimensional tomographic image.
[0124] Further, the processor 702 is further configured to determine whether the pixels in the mask of the three-dimensional tomographic image are covered by the projection images at all angles. If they are covered by the projection images at all angles, the pixels are in the tomographic effective area.
[0125] Further, the processor 702 is further configured to generate a maximum intensity projection image of the three-dimensional tomographic image;
[0126] According to the projection image of the preset projection angle and the maximum intensity projection image in the enhanced projection image, two sets of Laplacian pyramids are respectively generated;
[0127] The two sets of Laplacian pyramids are fused at different scales to generate a target pyramid, and the target pyramid is reconstructed to obtain an S2D image.
[0128] Further, the tomographic fusion imaging system based on multi-scale information fusion may further include Figure 7 The memory and the display not shown in .
[0129] For other technical details, refer to the description of the foregoing embodiments.
[0130] In the embodiments of the present application, by collecting projection images of a target object from multiple angles, preprocessing the collected projection images, adaptively enhancing the preprocessed projection images at different scales, performing backprojection using the enhanced projection images to generate a three-dimensional tomographic image, and performing information fusion at different scales on the three-dimensional tomographic image and the projection image of the preset angle in the enhanced projection image to obtain a synthesized two-dimensional image, the quality of the synthesized two-dimensional image can be improved.
[0131] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.
[0132] The above is the description of the tomographic fusion imaging method and system based on multi-scale information fusion provided by the present application. For those skilled in the art, according to the idea of the embodiments of the present application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present application.
Claims
1. A tomographic fusion imaging method based on multi-scale information fusion, characterized in that Including: Collecting projection images of the target object from multiple angles; Performing preprocessing on the collected projection images; Performing adaptive enhancement on the preprocessed projection images at different scales; Performing backprojection using the enhanced projection images to generate three-dimensional tomographic images; Performing information fusion on the three-dimensional tomographic images and the projection images at a preset angle in the enhanced projection images at different scales to obtain a synthesized two-dimensional image.
2. The method according to claim 1, wherein The performing preprocessing on the collected projection images includes: Performing preprocessing including preset items on the collected projection images; Performing filtering processing on the projection images after the preprocessing to obtain filtered projection images.
3. The method according to claim 2, wherein The performing adaptive enhancement on the preprocessed projection images at different scales includes: Constructing a first Laplacian pyramid based on the filtered projection images, the first Laplacian pyramid including projection images at multiple scales; Calculating the adaptive stretching reference coefficients of the projection images at each layer in the first Laplacian pyramid at each scale; Stretching the projection images at each layer in the first Laplacian pyramid according to the adaptive stretching reference coefficients to obtain a second Laplacian pyramid; Reconstructing the Laplacian pyramid to obtain enhanced projection images.
4. The method according to claim 3, wherein The adaptive stretching reference coefficients include a positive part reference coefficient and a negative part reference coefficient, and the calculating the adaptive stretching reference coefficients of the projection images at each layer in the Laplacian pyramid at each scale includes: Segmenting the projection images of the target object from each of the filtered projection images, and obtaining a first mask of the target object according to the projection images of the target object; According to the scale characteristics of each layer of the first Laplacian pyramid, reducing the mask corresponding to each scale to obtain a second mask of the target object corresponding to each projection image in the first Laplacian pyramid at each scale; According to the second masks of the target object at each scale, statistically calculating the maximum and minimum values of the projection images at each layer in the first Laplacian pyramid in all rows or all columns within the tissue range of the target object corresponding to the second masks, and respectively calculating the mean value of the maximum values and the mean value of the minimum values in all rows or all columns, as well as calculating the mean value of the maximum values and the mean value of the minimum values of all projection images at each scale; Determining the positive part reference coefficient of the adaptive stretching reference coefficient according to the mean value of the maximum values of all projection images at each scale and a preset first proportional coefficient, and determining the negative part reference coefficient of the adaptive stretching reference coefficient according to the mean value of the minimum values of all projection images at each scale and a preset second proportional coefficient.
5. The method according to claim 4, wherein After the performing backprojection using the enhanced projection images to generate three-dimensional tomographic images includes: Removing the artifacts of the three-dimensional tomographic images.
6. The method according to claim 5, wherein The removing the artifacts of the three-dimensional tomographic images includes: Reducing the size of the first mask of the target object to a preset proportion of the original size to obtain a third mask with a reduced size; Performing backprojection reconstruction on the third mask to obtain a mask of the three-dimensional tomographic image. Determine whether each pixel in the mask of the three-dimensional tomographic image is in the tomographic effective area; Set the pixel value of the pixels not in the tomographic effective area to 0 in the three-dimensional tomographic image.
7. The method according to claim 6, characterized in that The determination of whether each pixel in the mask of the three-dimensional tomographic image is in the tomographic effective area includes: Determine whether the pixels in the mask of the three-dimensional tomographic image are covered by the projection images at all angles; If covered by the projection images at all angles, the pixel is in the tomographic effective area.
8. The method according to claim 7, wherein The information fusion of the projection images at a preset angle in the three-dimensional tomographic image and the enhanced projection images at different scales to obtain a synthesized two-dimensional image includes: Generate the maximum intensity projection image of the three-dimensional tomographic image; Generate two sets of Laplacian pyramids respectively according to the projection image at the preset projection angle in the enhanced projection image and the maximum intensity projection image; Fuse the two sets of Laplacian pyramids at different scales to generate a target pyramid, and reconstruct the target pyramid to obtain the S2D image.
9. A tomographic fusion imaging system based on multi-scale information fusion, characterized in that, Include: A projection acquisition device and a processor; The projection acquisition device is used to acquire the projection images of the target object from multiple angles; The processor is used to perform preprocessing on the acquired projection images, perform adaptive enhancement on the preprocessed projection images at different scales, and perform backprojection using the enhanced projection images to generate a three-dimensional tomographic image, and perform information fusion on the projection images at a preset angle in the three-dimensional tomographic image and the enhanced projection images at different scales to obtain a synthesized two-dimensional image.
10. The determination device according to claim 9, characterized in that, The processor is further used to generate the maximum intensity projection image of the three-dimensional tomographic image; Generate two sets of Laplacian pyramids respectively according to the projection image at the preset projection angle in the enhanced projection image and the maximum intensity projection image; Fuse the two sets of Laplacian pyramids at different scales to generate a target pyramid, and reconstruct the target pyramid to obtain the S2D image.