X-ray imaging device and X-ray imaging method using the same

Abstract

According to an embodiment of the present invention, an X-ray imaging apparatus includes an X-ray irradiation module, an X-ray detection module, and an image processor, the image processor being configured to perform the steps of: acquiring a high-energy image and a low-energy image obtained by X-rays of relatively high energy and relatively low energy, respectively; decomposing the high-energy image to generate a high-energy frequency component image for each of a plurality of frequency bands; decomposing the low-energy image to generate a low-energy frequency component image for each of a plurality of frequency bands; merging at least a portion of the high-energy frequency component images for each of the plurality of frequency bands and at least a portion of the low-energy frequency component images for each of the plurality of frequency bands to generate a merged frequency component image; and generating a standard image using the merged frequency component image.

Description

[Technical Field]

[0001] The present invention relates to an X-ray imaging apparatus for obtaining an image using X-rays and an X-ray imaging method using the same. [Background technology]

[0002] An X-ray imaging device is a device that irradiates X-rays onto an affected area of ​​a human or animal body, and captures an image of the affected area when the penetrating X-rays are incident on the affected area. Such X-ray imaging devices provide X-ray images of the affected area continuously or continuously, and are widely used for diagnosing and reading affected areas, as well as for various medical procedures.

[0003] Dual Energy Subtraction (DES) technology is used as an image processing technique to generate images that separate and extract soft tissue and bone from digital X-ray images of a body part, such as the chest. DES is particularly widely used to improve the accuracy of chest diagnoses. DES technology is based on the difference in attenuation between low-energy and high-energy X-rays in soft tissue and bone, and generates soft tissue and bone images by erasing areas corresponding to bone or soft tissue in chest X-ray images by calculating the ratio of the difference in X-ray attenuation between low-energy and high-energy conditions for each material. Removing specific areas from an X-ray image is generally called subtraction.

[0004] The soft tissue and bone images acquired by DES offer improved visibility compared to conventional chest X-ray images, enabling accurate diagnosis of lesions. The two-shot dual-energy X-ray images used in DES processing are acquired by irradiating the subject with high-energy and low-energy X-rays at intervals of several hundred milliseconds (msec). Therefore, if the subject moves during the two-shot acquisition process, the position of the images shifts, resulting in motion artifacts in the subtracted image. To reduce motion artifacts, the subject is typically held stationary during two-shot acquisition. However, significant changes in the heart and pulmonary blood vessels can still cause motion artifacts, negatively impacting image quality and hindering image diagnosis. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Korean Patent Registration No. 10-1501086 [Patent Document 2] Patent No. 6549561 [Patent Document 3] U.S. Patent No. 8,315,355 Summary of the Invention [Problem to be solved by the invention]

[0006] An object of the present invention is to provide a method for reducing motion artifacts in images generated by subtraction image processing in order to improve the accuracy of image diagnosis.

[0007] Another object of the present invention is to provide a method for improving the image quality of standard images, soft tissue images, and bone images obtained by irradiating dual energy X-rays. [Means for solving the problem]

[0008] An X-ray imaging apparatus according to an embodiment of the present invention includes an X-ray irradiation module configured to irradiate X-rays, an X-ray detection module configured to detect X-rays irradiated from the X-ray irradiation module and passing through an object to be inspected and output corresponding digital signals, and an image processor configured to generate an X-ray image using the output signal of the X-ray detection module, wherein the image processor is configured to perform the following steps: acquiring a high-energy image and a low-energy image obtained by X-rays of relatively high energy and low energy, respectively, decomposing the high-energy image to generate high-energy frequency component images for a plurality of frequency bands, decomposing the low-energy image to generate low-energy frequency component images for a plurality of frequency bands, merging at least a portion of the high-energy frequency component images for a plurality of frequency bands and at least a portion of the low-energy frequency component images for a plurality of frequency bands to generate a merged frequency component image, and generating a standard image using the merged frequency component images.

[0009] According to an embodiment of the present invention, the high-energy frequency component images for each frequency band may include a plurality of levels of high-energy Laplacian pyramid images and a plurality of levels of high-energy Gaussian pyramid images, and the low-energy frequency component images for each frequency band may include a plurality of levels of low-energy Laplacian pyramid images and a plurality of levels of low-energy Gaussian pyramid images. The merged frequency component image may be generated by merging the high-energy Laplacian pyramid images and the low-energy Laplacian pyramid images for each level.

[0010] The high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image may be merged with each other in a patch unit including a plurality of pixels for each level. When the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image are merged in a patch unit for each level, a patch having a larger contrast value may be selected from among corresponding patches of the high-energy Gaussian pyramid image and the low-energy Gaussian pyramid image at the corresponding level, and the merging may be performed. The high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image may be merged with each other in pixel units or in patch units including a plurality of pixels for each level. When the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image are merged for each level, a pixel or patch corresponding to a side having a statistical value indicating a larger contrast among corresponding patches of the high-energy Gaussian pyramid image and the low-energy Gaussian pyramid image at the corresponding level may be selected and merged, or a pixel or patch corresponding to a side having a statistical value indicating a larger contrast among corresponding patches of the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image at the corresponding level may be selected and merged.

[0011] When the merging is performed based on a comparison of statistics of corresponding patches of the high-energy Gaussian pyramid image and the low-energy Gaussian pyramid image, the statistical value indicating contrast may be a standard deviation or an average of absolute deviations of brightness values ​​of a plurality of pixels constituting the corresponding patches. On the other hand, when the merging is performed based on a comparison of statistics of corresponding patches of the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image, the statistical value indicating contrast may be an average or a median of brightness values ​​of a plurality of pixels constituting the corresponding patches.

[0012] Meanwhile, according to another embodiment of the present invention, the high-energy frequency component images for each of the plurality of frequency bands may include high-energy frequency component images for each of the plurality of frequency bands decomposed by Fourier transform or wavelet transform. The low-energy frequency component images for each of the plurality of frequency bands may include low-energy frequency component images for each of the plurality of frequency bands decomposed by Fourier transform or wavelet transform, and the merged frequency component image may be generated by merging the high-energy frequency component images and the low-energy frequency component images for each frequency band and then performing an inverse Fourier transform or an inverse wavelet transform.

[0013] The high-energy frequency component image and the low-energy frequency component image may be merged with each other in pixel units or in patch units including a plurality of pixels for each frequency band. When the high-energy frequency component image and the low-energy frequency component image are merged for each frequency band, a pixel or patch corresponding to a side having a statistical value indicating a larger contrast may be selected from among corresponding patches of the high-energy frequency component image and the low-energy frequency component image of the corresponding frequency band, and the merging may be performed.

[0014] When the merging is performed based on a comparison of the statistics of corresponding patches of the high-energy frequency component image and the low-energy frequency component image, the statistical value indicating the contrast may be the average or median of the absolute values ​​of the multiple pixel values ​​that make up the corresponding patch.

[0015] The standard image may be generated by merging a plurality of frequency component images for each frequency band that constitute the merged frequency component image.

[0016] The image processor may be configured to further perform a step of registering one or more of the high-energy image and the low-energy image, and the registration may include multiple image registrations performed using bone masking information and soft-tissue masking information, respectively, to reflect differences in bone and soft-tissue motion.

[0017] The registration step may include the steps of: generating a bone image and a soft-tissue image by subtracting the high-energy image and the low-energy image, respectively; generating a bone masking image and a soft-tissue masking image, respectively, including the bone masking information, from the generated bone image and the generated soft-tissue image; performing primary registration of one or more of the high-energy image and the low-energy image to be registered using the bone masking image; and additionally registering the primarily registered one or more of the high-energy image and the low-energy image using the soft-tissue masking image.

[0018] The bone masking image may include edge position information of bones included in the bone image, and the soft tissue masking image may include edge position information of soft tissues included in the soft tissue image.

[0019] The primary registration using the bone masking image may use a global optimization-based image registration algorithm.

[0020] The image registration algorithm may be an image registration algorithm that focuses on bone motion using a free-form deformation (FFD) method.

[0021] The additional registration using the soft tissue masked image can be performed by a technique of measuring the similarity between the high-energy image and the low-energy image in units of patches each including a predetermined number of pixels, and performing local registration.

[0022] The similarity may be measured through calculation of pixel value or information entropy between the high-energy image and the low-energy image.

[0023] The similarity may be measured using one or more of NCC (Normalized Cross Correlation) and MI (Mutual Information).

[0024] According to another embodiment of the present invention, an X-ray imaging apparatus includes an X-ray irradiation module configured to irradiate X-rays, an X-ray detection module configured to detect X-rays irradiated from the X-ray irradiation module and passing through an object to be examined and output corresponding digital signals, and an image processor configured to generate an X-ray image using the output signal from the X-ray detection module. The image processor is configured to perform the following steps: acquiring a high-energy image and a low-energy image obtained by using X-rays of relatively high energy and relatively low energy, respectively; registering one or more of the high-energy image and the low-energy image; and generating one or more of a standard image, a bone image, and a soft-tissue image through calculations using the high-energy image and the low-energy image after the registration step. Here, the registration includes multiple image registrations performed using bone masking information and soft-tissue masking information, respectively, to reflect differences in motion between bone and soft tissue.

[0025] An X-ray imaging method according to an embodiment of the present invention includes irradiating an object with X-rays, detecting the X-rays that have passed through the object and generating corresponding digital signals, and generating an X-ray image using the digital signals, wherein generating the X-ray image includes acquiring a high-energy image and a low-energy image obtained by X-rays of relatively high and low energy, respectively, decomposing the high-energy image to generate high-energy frequency component images for a plurality of frequency bands, decomposing the low-energy image to generate low-energy frequency component images for a plurality of frequency bands, merging at least some of the high-energy frequency component images for a plurality of frequency bands and at least some of the low-energy frequency component images for a plurality of frequency bands to generate a merged frequency component image, and generating a standard image using the merged frequency component images.

[0026] According to another embodiment of the present invention, an X-ray imaging method includes irradiating an object with X-rays, detecting the X-rays that have passed through the object and generating corresponding digital signals, and generating an X-ray image using the digital signals. The generating the X-ray image includes acquiring high-energy images and low-energy images obtained using X-rays with relatively high and low energies, respectively, registering one or more of the high-energy images and the low-energy images, and generating one or more of a standard image, a bone image, and a soft-tissue image through calculations using the high-energy images and the low-energy images that have undergone the registration. The registration includes multiple image registrations that are performed using bone masking information and soft-tissue masking information, respectively, to reflect differences in motion between bone and soft tissue. [Effects of the Invention]

[0027] The present invention can reduce motion artifacts in images generated by subtraction imaging, and can improve the image quality of standard images, soft tissue images, and bone images obtained by irradiating dual-energy X-rays. [Brief explanation of the drawings]

[0028] [Figure 1] 1 is a diagram schematically illustrating an X-ray imaging apparatus according to an embodiment of the present invention; [Figure 2] 1 is a schematic block diagram of an X-ray imaging apparatus according to an embodiment of the present invention; [Figure 3] 1 is a flowchart illustrating a dual energy subtraction image processing method according to an embodiment of the present invention; [Figure 4] 1 is a graph showing the change in mass attenuation coefficient depending on the energy of X-rays. [Figure 5] Examples of low-energy and high-energy images acquired with low-energy and high-energy X-rays, respectively, are shown. [Figure 6] 1 shows a bone image without and with artificial intelligence-based noise reduction technology applied, respectively. [Figure 7] 10 is a flowchart illustrating an image registration process according to an embodiment of the present invention; [Figure 8] 10 illustrates an example of a masked image in which bone and soft tissue positions and motion occurrence information generated for image registration according to an embodiment of the present invention are separately extracted. [Figure 9] 10 is a diagram illustrating image registration using a bone masking image in an image registration process according to an embodiment of the present invention; [Figure 10] 1 shows example images obtained by applying an image registration process according to an embodiment of the present invention. [Figure 11] 1 shows a schematic flow chart of a method for generating a standard video according to an embodiment of the present invention; [Figure 12]1 is a diagram illustrating generation of a Laplacian pyramid image and a Gaussian pyramid image for generating a standard image, and generation of a merged Laplacian pyramid image thereof, according to an embodiment of the present invention; [Figure 13] 1 is a diagram illustrating a process of generating a Gaussian pyramid image and a Laplacian pyramid image according to an embodiment of the present invention; [Figure 14] 10 is a diagram illustrating a process of merging a high-energy Laplacian pyramid image and a low-energy Laplacian pyramid image according to an embodiment of the present invention; [Figure 15] 10 is a diagram illustrating a pixel or patch unit merging method when merging a high-energy Laplacian pyramid image and a low-energy Laplacian pyramid image according to an embodiment of the present invention; [Figure 16] 1 is a diagram illustrating a method for generating a merged frequency component image according to an embodiment of the present invention; [Figure 17] 1 is a diagram illustrating a filtering process of a discrete wavelet transform according to an embodiment of the present invention; [Figure 18] 1 is a diagram illustrating a process of generating a subband component image through a discrete wavelet transform according to an embodiment of the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0029] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0023] The present invention will now be described in detail with reference to the accompanying drawings, so that those skilled in the art can easily understand the present invention. However, the present invention may be embodied in many different forms and is not limited to the described embodiments.

[0030] 1 shows an example of an X-ray imaging apparatus according to an embodiment of the present invention. Referring to FIG. 1, an X-ray imaging apparatus 10 according to the embodiment of the present invention includes an X-ray irradiation module 11 that generates and irradiates X-rays, and X-ray detection modules 12 and 13 that detect the X-rays irradiated from the X-ray irradiation module 11 and passing through a subject. The X-ray detection modules 12 and 13 may include a first X-ray detection module 12 for realizing a bed-type imaging structure and a second X-ray detection module 13 for realizing a stand-type imaging structure. The X-ray irradiation module 11 is configured to be capable of linear and rotational movement so that it can face either one of the first and second X-ray detection modules 12 and 13.

[0031] The X-ray irradiation module 11 may include an X-ray source that generates X-rays, and generates X-rays by receiving power from the power supply module 15. The connection frame 17 may have a guide groove 19 that extends upward on the power supply module 15 and guides the vertical movement of the X-ray irradiation module 11. The X-ray irradiation module 11 is fastened to the connection frame 17 in a manner that allows it to move up and down through the guide groove 19. At this time, a connection block 21 is fastened to the connection frame 17 so as to be vertically movable, and the X-ray irradiation module 11 is supported by the connection block 21 so as to move up and down together with the connection block 21. In addition, the X-ray irradiation module 11 may be supported by the connection block 21 so as to be rotatable about a horizontal axis.

[0032] The power supply module 15 may be configured to be horizontally movable on a horizontally extending moving frame 23. For example, the moving frame 23 may include a guide rail 25 extending horizontally, and the power supply module 15 is disposed on the moving frame 23 via the guide rail 25 so as to be horizontally movable.

[0033] The first X-ray detection module 12 may be installed on the connecting frame 17 so as to face the X-ray irradiation module 11 in the up-down direction. X-ray imaging may be performed using the X-ray irradiation module 11 and the first X-ray detection module 12. A support table configured to allow a subject, for example, a patient, to be placed thereon may be arranged, and X-ray imaging may be performed after the X-ray irradiation module 11 and the first X-ray detection module 12 are aligned so as to be positioned above and below the support table, respectively.

[0034] Meanwhile, the second X-ray detection module 13 may be supported on a separate stand 24 so as to be movable up and down. X-ray imaging may be performed using the X-ray irradiation module 11 and the second X-ray detection module 13. At this time, the X-ray irradiation module 11 is rotated with respect to the connecting block 21 and aligned to face the second X-ray detection module 13. With a patient standing closely in front of the second X-ray detection module 13, X-rays are irradiated through the X-ray irradiation module 11, and the second X-ray detection module 13 receives the X-rays that have passed through the patient, thereby performing X-ray imaging.

[0035] The first and second X-ray detection modules 12 and 13 may each include an X-ray detector for detecting X-rays. The X-ray detector may have a retractable or stationary structure. The X-ray detector may also be a cassette-type X-ray detector, which can be stored in a storage compartment such as a bucky as needed. A cassette-type digital X-ray detector converts incident X-rays into an electrical signal suitable for video signal processing. For example, the X-ray detector may include a pixel circuit board including thin film transistors and may have a number of switching cell elements and photoelectric conversion elements arranged in a matrix. The X-ray detector may be a direct-type detector that includes a photoconductor having photoconductivity together with the pixel circuit board, or an indirect-type detector that includes a light-emitting layer such as a scintillator together with the pixel circuit board.

[0036] 2 is a schematic block diagram of an X-ray imaging apparatus according to an embodiment of the present invention. The X-ray irradiation module 11 is configured to irradiate X-rays 27 toward a patient S lying on a table 28 that is transparent to ultrasound.

[0037] The system controller 31 controls the entire system, receives imaging commands input through a user interface 32 such as a keyboard, mouse, or touchscreen input device, and controls the corresponding processes. When an imaging command is input, the system controller 31 controls the exposure controller 33, which controls the operation of the power supply module 25. The exposure controller 33 then controls the power supply module 25 accordingly to perform X-ray exposure. When an image capture command is input, the system controller 31 outputs a signal to control the detector controller 34, which controls the X-ray detection module 12. When an imaging command is input, the system controller 31 outputs a corresponding signal to the image processor 35. At this time, the system controller 31 can send and receive necessary data from the exposure controller 33, detector controller 34, and image processor 35. The image processor 35 can receive necessary information from the detector controller 34 and transmit data necessary for imaging to the exposure controller 33.

[0038] The X-ray detection module 12 detects X-rays emitted from the X-ray irradiation module 11 and passing through the patient S, and generates a digital signal corresponding to the detected X-rays. The image processor 35 generates an image using the digital image signal received from the X-ray detection module 12. The image processing performed by the image processor 35 and the image obtained thereby will be described later.

[0039] The display 36 displays the image acquired by the image processor 35 and may be any display capable of displaying an image, such as a liquid crystal display, an OLED display, etc. The image storage device 37 may be a memory capable of storing the acquired image, and may be any type of memory, such as a memory, a database, a cloud storage device, etc.

[0040] 3 is a flowchart illustrating a dual-energy subtraction image processing method according to an embodiment of the present invention, in which the image processor 35 may be configured to perform dual-energy subtraction image processing. The image processor 35 may include a microprocessor, memory, and related hardware and software to perform the image processing steps described below.

[0041] Referring to Figure 3, dual energy X-rays, i.e., high-energy and low-energy X-rays, are used to obtain high-energy images (I H ) and low energy image (I L ) are acquired (101, 102). For example, a high-energy image can be acquired by irradiating relatively high-energy (120 kVp) X-rays, and a low-energy image can be acquired by irradiating relatively low-energy (60 kVp) X-rays. Different images are obtained under high-energy and low-energy conditions due to the difference in the mass attenuation coefficient of X-rays for soft tissue and bone.

[0042] 4 is a graph showing the change in mass attenuation coefficient depending on the energy of X-rays, and FIG. 5 shows examples of low-energy image (a) and high-energy image (b) acquired by low-energy X-rays and high-energy X-rays, respectively. High-energy and low-energy X-rays are irradiated through the control of the X-ray irradiation unit 11, and the images are acquired through the image acquisition unit 13, thereby enabling high-energy and low-energy images to be acquired.

[0043] Next, noise reduction is performed on one or more of the acquired high-energy and low-energy images (103, 104). While Figure 3 illustrates an example in which noise reduction is performed on both the high-energy and low-energy images, noise reduction can also be performed on only one of the high-energy and low-energy images. For example, noise reduction can be performed through learning using artificial intelligence (AI). Specifically, AI technology using the CycleGAN architecture can learn random noise and reduce noise while minimizing anatomical loss in the images. Applying AI-based noise reduction technology enables the acquisition of high-quality images even at low doses, thereby improving the accuracy of subsequent subtraction factor determination and image registration algorithms. Figure 6 shows a bone image without AI-based noise reduction (a) and a bone image with AI-based noise reduction (b). It can be seen that applying AI-based noise reduction technology improves the image quality of the final subtraction image.

[0044] Next, image registration is performed on one or more of the high-energy image and the low-energy image (105, 106). Although Fig. 3 illustrates an example in which image registration is performed on both the high-energy image and the low-energy image, image registration can also be performed on only one of the high-energy image and the low-energy image. Fig. 7 shows a schematic flowchart of an image registration process according to an embodiment of the present invention.

[0045] Image registration is an image processing method for correcting distortions caused by breathing, heartbeat, etc., that inevitably occur during imaging when comparing the same patient over time or at different points in time. When imaging is performed by irradiating X-rays with different energies twice, a time interval of approximately 100 to 200 ms inevitably occurs between the two images, resulting in motion artifacts caused by the movement of the heart, pulmonary blood vessels, etc., making the image registration process difficult. The motion artifacts have a negative impact on material separation performance and the accuracy of image diagnosis. To reduce the negative impact of the motion artifacts, an embodiment of the present invention applies an image registration algorithm that performs precise correction calculations while preventing artificial image distortion, thereby reducing the motion artifacts.

[0046] Generally, medical image registration uses a non-rigid registration method that takes into account the characteristics of medical images to prevent artificial image distortion. To account for differences in the magnitude and direction of motion among different human tissues, multi-resolution image processing techniques such as Gaussian or Laplacian pyramids are known. However, existing patch-based motion prediction and correction methods have limitations in accurately calculating motion for each human tissue. To overcome these limitations, embodiments of the present invention improve the accuracy and calculation speed of image registration algorithms by separating soft tissue and bone motion and applying separate algorithms. While the motion of the shoulders, spine, ribs, and other parts of a patient caused by minute movements or breathing appears globally, such as in the lung region, which is the primary target of interest, the motion of the heart and surrounding pulmonary blood vessels caused by cardiac pulsation appears locally and may be opposite in direction to the motion of bones. Therefore, embodiments of the present invention apply an image registration algorithm that separates and processes soft tissue and bone.

[0047] First, referring to FIG. 7, in order to apply the image registration algorithm separately to the motion of bones and soft tissues, a high-energy image (I H ) and low energy image (I L ) is subtracted from the bone image (I BONE ) and soft tissue imaging (I SOFT ) are generated (201). For the sake of easy understanding, the same symbols, i.e., I, are used for the high-energy image and the low-energy image generated by the digital signals acquired by the X-ray detection modules 12 and 13, the high-energy image and the low-energy image subjected to noise reduction processing, and the high-energy image and the low-energy image subjected to image registration. H and I L Use.

[0048] Then, the generated bone image (I BONE ) and soft tissue imaging (I SOFT ) from bone masking video (I BONE_MASK ) and soft tissue masking video (I SOFT_MASK ) are generated (202). BONE_MASK ) is bone image (I BONE ) and includes position information of bones included in the image, specifically edge position information, and the soft tissue masked image (I SOFT_MASK ) is a soft tissue image (I SOFT ) and may include position information of the soft tissue included in the image, specifically, position information of the edge of the soft tissue.

[0049] According to an embodiment of the present invention, a masking image is generated by extracting position information for each of bones and soft tissues in order to separately apply an image registration algorithm to the motion of bones and soft tissues. To extract the position information of each tissue, the principle of subtracting images obtained using high-energy X-rays and low-energy X-rays is utilized. The position information of each tissue is extracted from the soft-tissue image and bone image acquired temporarily by subtracting the high-energy image and the low-energy image in an unregistered state. The soft-tissue image and bone image contain only information about the soft tissue or bone, and because they are not registered, they also contain information about the occurrence of motion artifacts. Therefore, by extracting edge information from these images, masking images for the respective positions of bones and soft tissues and the motion occurrence positions of each tissue can be obtained. FIG. 8 shows masking images extracted by separately extracting the positions and motion occurrence information of bones and soft tissues, where (a) is the masking image for bones, and (b) is the masking image for soft tissues.

[0050] Referring again to FIG. 7, the generated bone masking image (I BONE_MASK ) to generate high energy (I H ) and / or low-energy imaging (I L ) are first aligned (203), and then the generated soft tissue masked image (I SOFT_MASK ) and the high-energy (I H ) and / or low-energy imaging (I L ) are secondarily aligned (204). BONE_MASK ) is a globally optimized image registration method, and it is suitable for soft tissue masking images (I SOFT_MASK ) is a local optimization method for image registration.

[0051] First, the bone masking video (I BONE_MASKThe primary image registration using the FFD (Free-Form Deformation) method prevents motion registration from being performed in mutually contradictory directions across the entire image, and employs a global optimization technique to derive optimized results with an emphasis on bone motion. In other words, the primary image registration can be performed with an emphasis on bone motion using the FFD (Free-Form Deformation) method, which is an image registration algorithm based on global optimization to naturally align the global movement of the human body.

[0052] An example of alignment using image deformation using FFD is shown in Figure 9. (a) in Figure 9 shows the image before deformation, and (b) shows the control points of the control grid selected from the image in (a). The control points shown in (b) in Figure 9 are deformed as shown in (c), and the image in (a) can be deformed as shown in (d). This FFD method is a representative image deformation technique, and by manipulating the control points of the control grid, objects can be deformed naturally so as not to violate the laws of physics.

[0053] Next, the soft tissue masking image (I SOFT_MASKThe secondary image registration using the CT scan is performed using a local technique to register the soft tissue motion to the image obtained by the primary registration performed around the bone, taking into account the fact that the motion of soft tissues, such as the heart and its surrounding pulmonary vessels, occurs relatively locally. In particular, unlike bone motion, the motion of pulmonary vessels may exhibit various directions and magnitudes independently of the surrounding area. Therefore, it is difficult to obtain accurate results if an FFD technique in which the surrounding area is deformed collectively by changing the control points is applied. Therefore, in an embodiment of the present invention, the secondary image registration is performed using a technique for local registration by calculating the similarity between high-energy and low-energy images in small-sized patches containing a predetermined number of pixels for the required soft tissue area. For example, methods such as normalized cross correlation (NCC) or mutual information (MI) can be used to measure the similarity. NCC and MI are methods for measuring the similarity by calculating the correlation between pixel values ​​or information entropy between two images, respectively.

[0054] FIG. 10 shows an image obtained by applying an image registration algorithm to an image of an extracted region of interest. (a) of FIG. 10 shows an image obtained by applying patch-based image registration according to a conventional method, and (b) of FIG. 10 shows an image obtained by separating the motions of bones and soft tissues, which have different characteristics, and applying an image registration algorithm to each of them according to an embodiment of the present invention. By separating and registering the motions of bones and soft tissues separately, the present invention can precisely register the motions of each tissue even if they conflict with each other. Applying a local registration technique to the entire image can result in unnatural results and the risk of registration noise. Therefore, even if a global optimization-based FFD technique, which has high computational complexity, is applied, a calculation method that focuses on bone motion can effectively reduce the amount of calculation required for registration. Furthermore, by focusing on the fact that the motion of soft tissues, such as the heart and its surrounding pulmonary blood vessels, exhibits localized behavior, a method of extracting only the relevant region and performing local registration minimizes the risk of registration noise and enables high-speed registration.

[0055] Next, referring to FIG. 3, the image-registered high-energy image (I H ) and low energy image (I L ) calculation (107) to generate target images, i.e., a standard image 111, a soft tissue image 112, and a bone image 113. The standard image 111 is a high energy image (I H ) and low energy image (I L ), and the soft tissue image 112 and the bone image 113 can be calculated by adding up the high energy image (I H ) and low energy image (I L) can be calculated by subtraction. Existing standard images generally use an image of one energy condition, for example, a high-energy image or a low-energy image, or simply blend a high-energy image and a low-energy image at a predetermined ratio, and the standard image obtained by such an existing method has a problem of not having good contrast. In contrast, the present invention provides a standard image with effectively improved contrast compared to existing standard images, and in this sense, the standard image according to the present invention can be called an enhanced standard image.

[0056] According to an embodiment of the present invention, a high-energy image and a low-energy image are decomposed to generate a plurality of frequency component images, i.e., a plurality of frequency band images, and the generated plurality of high-energy frequency component images and a plurality of low-energy frequency component images are merged for each corresponding frequency band to generate a merged frequency component image. A standard image is generated using the merged frequency component images for each frequency band. In one embodiment, the high-energy and low-energy frequency component images for each frequency band are generated by generating a multi-resolution pyramid image using the high-energy image and the low-energy image. Meanwhile, in another embodiment, the high-energy and low-energy frequency component images for each frequency band are generated using a Discrete Wavelet Transform (DWT), and in another embodiment, the high-energy and low-energy frequency component images for each frequency band are generated using a Discrete Fourier Transform (DFT). These embodiments will be described in order below.

[0057] First, a method for generating high-energy and low-energy frequency component images for each frequency band through generation of a multi-resolution pyramid image and generating a standard image based on the generated high-energy and low-energy frequency component images will be described with reference to FIGS. 11 to 16. FIG. 11 shows a schematic flowchart of a method for generating a standard image 111 according to an embodiment of the present invention, and FIG. 12 shows a schematic image processing procedure of the method for generating a standard image 111 according to an embodiment of the present invention. According to an embodiment of the present invention, a high-energy image (I) to which image registration is applied is generated. H ) and low energy image (I L ) are merged to generate a superior quality standard image 111 with improved global and local contrast.

[0058] First, referring to FIG. 11, the high energy image (I H ) and low energy image (I L ) to the logarithmic scale (302). This process is H ) and low energy image (I L This is done to generate a linear relationship between tissue strength and the

[0059]

number

[0060]

number

[0061] Next, multi-resolution pyramid images, i.e., high-energy and low-energy frequency component images for multiple frequency bands, are generated using the normalized low-energy image and high-energy image (304). The generated pyramid images with different resolutions represent frequency components for each band. By decomposing the image for each frequency band, noise components in the high-frequency band can be removed, thereby improving the accuracy of the algorithm.

[0062] When multi-resolution images are generated using normalized low-energy images and high-energy images, the generated pyramid images of different resolutions represent frequency component images for each frequency band. By generating multi-resolution pyramid images and decomposing the images into each frequency band, noise components in the high-frequency band can be removed, improving the accuracy of the algorithm. Furthermore, the highest level of Gaussian pyramid components, i.e., the image in the low-frequency band, represents global contrast, while the higher the frequency band, the more detailed the edge information, from thick edge information of bones and the heart to detailed edge information such as pulmonary blood vessels.

[0063] Referring to FIG. 12, the normalized high-energy image is converted into a high-energy Laplacian image pyramid (L H ) and high-energy Gaussian image pyramid (G H ) is generated, and the low-energy Laplacian image pyramid (L L ) and low-energy Gaussian image pyramid (G L) is generated. As the pyramid level increases, the resolution of the corresponding image decreases. For example, if a pyramid image has four levels, the pyramid image at the lowest level may have a resolution of 1000*1000, the pyramid image at the next level may have a resolution of 500*500, the pyramid image at the next level may have a resolution of 250*250, and the pyramid image at the highest level may have a resolution of 125*125. In this manner, multiple images of a pyramid structure having multi-resolution may be realized.

[0064] Specifically, Gaussian and Laplacian pyramid images can be generated as shown in Equation 3 below.

[0065]

number

[0066] Next, referring again to FIG. 11, at each level of the pyramid, a high-energy Laplacian pyramid image (L H ) and low-energy Laplacian pyramid images (L L ) are merged (305). In this merging step, components at each level are selected and merged to increase the global and local contrast. Referring to FIG. 12, the high-energy Laplacian pyramid image (L H ) and low-energy Laplacian pyramid images (L L ) are merged at each level to produce a merged Laplacian pyramid image (L R ) is generated.

[0067] Figures 14 and 15 show high-energy Laplacian pyramid images (L H ) and low-energy Laplacian pyramid images (L L ) is a schematic diagram of an image processing process for merging high-energy Laplacian pyramid images (L H ) ingredients at each level TIFF0007804772000024.tif9170 and low energy Laplacian pyramid image (L L ) with the same level of ingredients TIFF0007804772000025.tif10170 are merged to create a Laplacian pyramid image (L R ) each component Generates TIFF0007804772000026.tif10170.

[0068] In an embodiment of the present invention, when a low-energy Laplacian pyramid image and a high-energy Laplacian image at each level are merged, a high-energy Laplacian pyramid image (L H ) and low-energy Laplacian pyramid images (L L ) can be selected and merged. At this time, the high-energy Laplacian pyramid image (L H ) and low-energy Laplacian pyramid images (L L ) is selected based on the high-energy Gaussian pyramid image (G H ) and low-energy Gaussian pyramid images (G L ) at a level having the same resolution. Here, the selection of a corresponding pixel or patch in the Gaussian pyramid image may refer to a patch including a plurality of pixels surrounding a pixel in the Laplacian pyramid image of the corresponding patch in the Gaussian pyramid image, or a patch identical to the patch in the Laplacian pyramid image. For example, in relation to selecting a corresponding pixel or patch in the low-energy and high-energy Laplacian pyramid images, the selection of a high-energy Gaussian pyramid image (G H ) and low-energy Gaussian pyramid images (G L ), the patch of the Laplacian pyramid image on the side with the larger contrast statistic (standard deviation or average absolute deviation of the contrast of the pixels belonging to the patch) is selected.

[0069] Figure 15 shows the lowest level high-energy Gaussian pyramid image. TIFF0007804772000027.tif8170 and low-energy Gaussian pyramid image According to the comparison results of TIFF0007804772000028.tif9170, the lowest level high-energy Laplacian pyramid image TIFF0007804772000029.tif9170 and low-energy Laplacian pyramid image An example of merging TIFF0007804772000030.tif8170 is shown. Here, the Gaussian pyramid image TIFF0007804772000031.tif9170 and Laplacian pyramid image TIFF0007804772000032.tif9170 are pyramid images of the same level, i.e., images with the same resolution. First, referring to (a) of FIG. 15, a high-energy Laplacian pyramid image TIFF0007804772000033.tif8170 and low-energy Laplacian pyramid image High-energy Gaussian pyramid image related to selecting which of the first patches in TIFF0007804772000034.tif8170 TIFF0007804772000035.tif8170 and low-energy Gaussian pyramid image Statistics showing the contrast of the same patch of TIFF0007804772000036.tif8170, e.g., the standard deviation (STD H_11 , STDs L_11 ), i.e., the high-energy side, is selected, and the high-energy Laplacian pyramid image Brightness value of the patch in TIFF0007804772000037.tif8170 (I H_11 ) is merged Laplacian pyramid image (L R ) lowest level of video The brightness value of the patch in TIFF0007804772000038.tif8170 is selected as the brightness value of the patch. Next, in (b) of Figure 15, the merged Laplacian pyramid image (L R ) lowest level of video An example of selecting the brightness value of the second patch in TIFF0007804772000039.tif9170 is shown. Similar to (a), a high-energy Gaussian pyramid image TIFF0007804772000040.tif8170 and low-energy Gaussian pyramid image The standard deviation (STD H_12 , STDs L_12 ), i.e., the low-energy side, is selected, and the low-energy Laplacian image Brightness value of the patch in TIFF0007804772000042.tif8170 (I L_12 ) is merged Laplacian pyramid image (L R ) is selected as the brightness value of the corresponding patch.

[0070] Pixel contrast values ​​at higher levels of the pyramid represent global contrast, and pixel contrast values ​​at lower levels of the pyramid represent local contrast. In this manner, the Laplacian components at each pyramid level can be determined.

[0071] Meanwhile, in another embodiment of the present invention, a high-energy Laplacian pyramid image (L H ) and low-energy Laplacian pyramid images (L L ) is not based on a comparison of statistics indicating the contrast of the patch in the high-energy Gaussian pyramid image and the low-energy Gaussian pyramid image, but rather the one with the larger statistical value indicating the contrast of the patch in the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image, for example, the larger average or median brightness of the pixels that make up the patch, can be selected.

[0072] 11, a merged image is generated from the Laplacian components of the merged Laplacian pyramid obtained by merging the high-energy Laplacian image and the low-energy Laplacian image (306). At this time, the Laplacian components can be merged sequentially starting from the highest level to generate the merged image.

[0073] 16 shows an example of generating a merged image by reconstructing a merged Laplacian pyramid image. For example, referring to FIG. 16, the image at the highest level of the merged Laplacian pyramid is High-energy Gaussian image with the same resolution as TIFF0007804772000043.tif9170 TIFF0007804772000044.tif9170 is added together to form the first merged video component TIFF0007804772000045.tif8170 is generated. Here, adding the images can mean adding the brightness values ​​of the pixels to be merged in the two images. Then, the first merged image component Image obtained by upsampling TIFF0007804772000046.tif8170 TIFF0007804772000047.tif8170 and the next level merged Laplacian pyramid image TIFF0007804772000048.tif8170 is added to create the secondary merged image component Generate TIFF0007804772000049.tif9170. Then, the secondary merged image component Image obtained by upsampling TIFF0007804772000050.tif9170 TIFF0007804772000051.tif8170 and the next level merged Laplacian pyramid image TIFF0007804772000052.tif9170 is added to create a 3D merged image component. Generate TIFF0007804772000053.tif9170. Then, the 3rd order merged image component Image obtained by upsampling TIFF0007804772000054.tif9170 TIFF0007804772000055.tif9170 and the next level merged Laplacian pyramid image TIFF0007804772000056.tif8170 is added to create the fourth-order merged image component. Generate TIFF0007804772000057.tif8170. At this time, the generated fourth-order merged image component TIFF0007804772000058.tif8170 is the normalized merged image, i.e., the normalized standard image The file is TIFF0007804772000059.tif8170.

[0074] Merging Laplacian pyramid images is performed sequentially from the highest level of the pyramid to the lowest level (sequentially from the bottom up in FIG. 16). Since pyramid images at higher levels show global image components well and pyramid images at lower levels show local image components well, sequential merging from the highest level of the pyramid to the lowest level means first concentrating on global image components and then gradually concentrating on local image components.

[0075] Next, referring again to Figure 11, a scale transformation is performed (307) to return the log scale of the merged image to the original intensity scale, thereby generating a standard image.

[0076] Hereinafter, a method for generating a standard image using discrete wavelet transform will be described with reference to FIGS.

[0077] The normalized low-energy image and high-energy image can be decomposed using the discrete wavelet transform (DWT) to generate frequency component images for each frequency band. The discrete wavelet transform is a method that improves conversion efficiency by decomposing an image into different frequency components to match human visual characteristics using localized bases, and then irradiating and processing each component associated with the resolution corresponding to each frequency band.

[0078] The discrete wavelet transform can be expressed by the following Equation 4.

[0079]

number

[0080] Referring to Figure 17, the original discrete signal is decomposed into several sub-global signals with different frequencies through down-sampling of multi-resolution analysis, and then synthesized back into the original discrete signal through up-sampling. The original video signal is decomposed into a low-pass component (L) and a high-pass component (H) using a low-pass filter (LPF, h(n)) and a high-pass filter (HPF, g(n)). The scale function used at this time is TIFF0007804772000065.tif8169 is as shown in Equation 5 below, and the wavelet function TIFF0007804772000066.tif8169 can be expressed as Equation 6 below.

[0081]

number

[0082]

number

[0083] Referring to FIG. 18, a transformed image obtained by performing a discrete wavelet transform on an original image can include four subband component images (LL, LH, HL, and HH). The LL subband component image is composed of coefficients obtained by applying a horizontal low-pass filter (LPF_y) and a vertical low-pass filter (LPF_x) to the original image (x) to remove high-frequency components. The HH subband component image is obtained by applying a horizontal high-pass filter (HPF_y) and a vertical high-pass filter (HPF_x) to the original image (x) and, unlike the LL subband component image, only shows high-frequency components. The LH subband component image is obtained by applying a horizontal high-pass filter (HPF_y) to the original image (x) and includes horizontal frequency error components. The HL subband component image is obtained by applying a vertical high-frequency filter (HPF_x) to the original image (x) and includes vertical frequency error components. Here, first down sampling and second down sampling can be performed after applying a high- or low-pass filter, respectively.

[0084] Next, when merging the low energy component image and the high energy component image for each frequency band in each step, as in the above-described embodiment, one pixel or patch of either the high energy component or the low energy component may be selected for each pixel of the image or for each patch of pixels of the image, and merging may be performed. In this case, if merging is performed based on a comparison of statistics of corresponding patches of the high energy frequency component image and the low energy frequency component image, the statistics indicating contrast may be an average or median of absolute values ​​of multiple pixel values ​​constituting the corresponding patches.

[0085] According to another embodiment of the present invention, a high frequency component image and a low frequency component image for each frequency band may be generated by applying a discrete Fourier transform (DFT). The discrete Fourier transform may be expressed by the following Equation 7:

[0086]

number

[0087]

number

[0088] Although the embodiments of the present invention have been described above, the scope of the present invention is not limited to these, and various modifications and improvements made by those skilled in the art using the basic concept of the present invention defined in the claims also fall within the scope of the present invention. [Industrial Applicability]

[0089] INDUSTRIAL APPLICABILITY The present invention relates to an X-ray imaging apparatus and method, and is applicable to X-ray imaging equipment, and therefore has industrial applicability.

Claims

1. an X-ray irradiation module configured to be able to irradiate X-rays; an X-ray detection module configured to detect X-rays irradiated from the X-ray irradiation module and passing through an object to be inspected, and output a corresponding digital signal; an image processor configured to generate an X-ray image using an output signal of the X-ray detection module; Including, The image processor obtaining high-energy and low-energy images obtained by X-rays of relatively high and low energy, respectively; generating high energy frequency component images for each of a plurality of frequency bands by decomposing the high energy image; generating low-energy frequency component images for each of a plurality of frequency bands by decomposing the low-energy image; generating a merged frequency component image by merging at least a portion of the high-energy frequency component images for each frequency band and at least a portion of the low-energy frequency component images for each frequency band; generating a standard image using the merged frequency component image; configured to: the high-energy frequency component images for each of the plurality of frequency bands include high-energy Laplacian pyramid images at a plurality of levels and high-energy Gaussian pyramid images at a plurality of levels; the plurality of frequency band-specific low energy frequency component images include a plurality of levels of low energy Laplacian pyramid images and a plurality of levels of low energy Gaussian pyramid images; the merged frequency component image is generated by merging the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image for each level; The high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image are merged with each other in pixel units or in patch units including a plurality of pixels for each level, When merging the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image for each level, a pixel or patch corresponding to a side having a statistical value indicating a larger contrast among corresponding patches of the high-energy Gaussian pyramid image and the low-energy Gaussian pyramid image of the corresponding level is selected and merged, or a pixel or patch corresponding to a side having a statistical value indicating a larger contrast among corresponding patches of the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image of the corresponding level is selected and merged.

2. When merging by level, a pixel or patch corresponding to the side having a statistical value indicating a larger contrast is selected from the corresponding patches of the high-energy Gaussian pyramid image and the low-energy Gaussian pyramid image of the level, and the merging is performed, the statistical value is the average of the standard deviation or absolute deviation of the brightness values ​​of the multiple pixels constituting the corresponding patch, 2. The X-ray imaging device of claim 1, wherein when merging by level, a pixel or patch corresponding to a side having a statistical value indicating a larger contrast is selected from among corresponding patches of the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image of the corresponding level, and the statistical value is an average or median of brightness values ​​of multiple pixels constituting the corresponding patch.

3. An X-ray irradiation module configured to be able to irradiate X-rays; an X-ray detection module configured to detect X-rays irradiated from the X-ray irradiation module and passing through an object to be inspected, and output a corresponding digital signal; an image processor configured to generate an X-ray image using an output signal of the X-ray detection module; Including, The image processor obtaining high-energy and low-energy images obtained by X-rays of relatively high and low energy, respectively; generating high energy frequency component images for each of a plurality of frequency bands by decomposing the high energy image; generating low-energy frequency component images for each of a plurality of frequency bands by decomposing the low-energy image; generating a merged frequency component image by merging at least a portion of the high-energy frequency component images for each frequency band and at least a portion of the low-energy frequency component images for each frequency band; generating a standard image using the merged frequency component image; configured to: the high-energy frequency component images for each of the plurality of frequency bands include high-energy frequency component images for each of the plurality of frequency bands decomposed by Fourier transform or wavelet transform; the plurality of low energy frequency component images for each frequency band include a plurality of low energy frequency component images for each frequency band decomposed by Fourier transform or wavelet transform, The merged frequency component image is generated by merging the high-energy frequency component image and the low-energy frequency component image for each frequency band and performing an inverse Fourier transform or an inverse wavelet transform; The high-energy frequency component image and the low-energy frequency component image are merged with each other in units of pixels or in units of patches including a plurality of pixels for each frequency band, When the high-energy frequency component image and the low-energy frequency component image are merged for each frequency band, pixels or patches corresponding to a side having a statistical value indicating a larger contrast are selected from corresponding patches of the high-energy frequency component image and the low-energy frequency component image of the corresponding frequency band, and merging is performed.

4. An X-ray imaging device as described in Claim 3, wherein the statistical value is the average or median of the absolute values ​​of the multiple pixel values ​​that make up the corresponding patch.

5. The X-ray imaging apparatus according to claim 1 , wherein the standard image is generated by merging a plurality of frequency component images for each frequency band that constitute the merged frequency component image.

6. The image processor registering the acquired high-energy image and the acquired low-energy image; generating high-energy frequency component images for each of the plurality of frequency bands by decomposing the registered high-energy images; generating low-energy frequency component images for each of the plurality of frequency bands by decomposing the aligned low-energy images; configured to: The step of aligning includes: generating a bone image and a soft tissue image by subtracting the acquired high-energy image and the acquired low-energy image, respectively; generating a bone masking image from the generated bone image, the bone masking image including edge position information of bones included in the bone image, and generating a soft-tissue masking image from the generated soft-tissue image, the soft-tissue masking image including edge position information of soft tissues included in the soft-tissue image; performing primary registration between the high-energy image and the low-energy image, which are registration targets, using the bone masking image; additionally registering the first registered high-energy image and the first registered low-energy image using the soft tissue masking image; Including, The primary registration using the bone masking image is a global optimization-based image registration algorithm, and uses an image registration algorithm using a free-form deformation (FFD) method based on edge position information of bones included in the bone masking image; the additional registration using the soft-tissue masking image is performed by a technique of measuring a similarity between the primarily registered high-energy image and the primarily registered low-energy image in units of patches each including a predetermined number of pixels based on edge position information of soft tissue included in the soft-tissue masking image, and performing local registration; The X-ray imaging apparatus according to claim 1 , wherein the similarity is measured through calculation of pixel values ​​or information entropy between the primarily registered high-energy image and the primarily registered low-energy image.

7. The X-ray imaging apparatus according to claim 6 , wherein the similarity is measured using one or more of NCC (Normalized Cross Correlation) and MI (Mutual Information).

8. irradiating an X-ray beam toward an object to be inspected; detecting X-rays passing through the object to be inspected and generating a corresponding digital signal; generating an X-ray image using the digital signal; The step of generating an X-ray image comprises: obtaining high-energy and low-energy images obtained by X-rays of relatively high and low energy, respectively; generating high energy frequency component images for each of a plurality of frequency bands by decomposing the high energy image; generating low-energy frequency component images for each of a plurality of frequency bands by decomposing the low-energy image; generating a merged frequency component image by merging at least a portion of the high-energy frequency component images for each frequency band and at least a portion of the low-energy frequency component images for each frequency band; generating a standard image using the merged frequency component image; the high-energy frequency component images for each of the plurality of frequency bands include high-energy Laplacian pyramid images at a plurality of levels and high-energy Gaussian pyramid images at a plurality of levels; the plurality of frequency band-specific low energy frequency component images include a plurality of levels of low energy Laplacian pyramid images and a plurality of levels of low energy Gaussian pyramid images; the merged frequency component image is generated by merging the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image for each level; The high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image are merged with each other in pixel units or in patch units including a plurality of pixels for each level, and when merging the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image for each level, a pixel or patch corresponding to a side having a statistical value indicating a larger contrast among corresponding patches of the high-energy Gaussian pyramid image and the low-energy Gaussian pyramid image of the level is selected and merged, or a pixel or patch corresponding to a side having a statistical value indicating a larger contrast among corresponding patches of the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image of the level is selected and merged.

9. When merging by level, a pixel or patch corresponding to the side having a statistical value indicating a larger contrast is selected from the corresponding patches of the high-energy Gaussian pyramid image and the low-energy Gaussian pyramid image of the level, and the merging is performed, the statistical value is the average of the standard deviation or absolute deviation of the brightness values ​​of the multiple pixels constituting the corresponding patch, 9. The X-ray imaging method of claim 8, wherein when merging by level, a pixel or patch corresponding to a side having a statistical value indicating a larger contrast is selected from among corresponding patches of the high-energy Laplacian pyramid image and the low-energy Laplacian pyramid image of the corresponding level, and the statistical value is an average or median of brightness values ​​of a plurality of pixels constituting the corresponding patch.

10. A step of irradiating an X-ray toward an object to be inspected; detecting X-rays passing through the object to be inspected and generating a corresponding digital signal; generating an X-ray image using the digital signal; The step of generating an X-ray image comprises: obtaining high-energy and low-energy images obtained by X-rays of relatively high and low energy, respectively; generating high energy frequency component images for each of a plurality of frequency bands by decomposing the high energy image; generating low-energy frequency component images for each of a plurality of frequency bands by decomposing the low-energy image; generating a merged frequency component image by merging at least a portion of the high-energy frequency component images for each frequency band and at least a portion of the low-energy frequency component images for each frequency band; generating a standard image using the merged frequency component image; the high-energy frequency component images for each of the plurality of frequency bands include high-energy frequency component images for each of the plurality of frequency bands decomposed by Fourier transform or wavelet transform; the plurality of low energy frequency component images for each frequency band include a plurality of low energy frequency component images for each frequency band decomposed by Fourier transform or wavelet transform, The merged frequency component image is generated by merging the high-energy frequency component image and the low-energy frequency component image for each frequency band and performing an inverse Fourier transform or an inverse wavelet transform; The high-energy frequency component image and the low-energy frequency component image are merged with each other in units of pixels or in units of patches including a plurality of pixels for each frequency band, When the high-energy frequency component image and the low-energy frequency component image are merged for each frequency band, pixels or patches corresponding to a side having a statistical value indicating a larger contrast are selected from corresponding patches of the high-energy frequency component image and the low-energy frequency component image of the frequency band, and merging is performed.

11. An X-ray imaging method as described in Claim 10, wherein the statistical value is the average or median of the absolute values ​​of the multiple pixel values ​​that make up the corresponding patch.

12. The X-ray imaging method according to claim 8 , wherein the standard image is generated by merging a plurality of frequency component images for each frequency band that constitute the merged frequency component image.

13. generating the X-ray image further includes registering the acquired high-energy image and the acquired low-energy image; Decomposing the aligned high-energy images to generate high-energy frequency component images for each of the plurality of frequency bands; Decomposing the aligned low-energy images to generate low-energy frequency component images for each of the plurality of frequency bands; The step of aligning includes: generating a bone image and a soft tissue image by subtracting the acquired high-energy image and the acquired low-energy image, respectively; generating a bone masking image from the generated bone image, the bone masking image including edge position information of bones included in the bone image, and generating a soft-tissue masking image from the generated soft-tissue image, the soft-tissue masking image including edge position information of soft tissues included in the soft-tissue image; performing primary registration between the high-energy image and the low-energy image, which are registration targets, using the bone masking image; additionally registering the first registered high-energy image and the first registered low-energy image using the soft tissue masking image; Including, The primary registration using the bone masking image is a global optimization-based image registration algorithm, and uses an image registration algorithm using a free-form deformation (FFD) method based on edge position information of bones included in the bone masking image; the additional registration using the soft-tissue masking image is performed by a technique of measuring a similarity between the primarily registered high-energy image and the primarily registered low-energy image in units of patches each including a predetermined number of pixels based on edge position information of soft tissue included in the soft-tissue masking image, and performing local registration; the similarity is measured through calculation of pixel value or information entropy between the primary aligned high-energy image and the primary aligned low-energy image; 9. The X-ray imaging method according to claim 8.

Images