Image processing device, image processing system, image processing method, and program

Abstract

To reduce the remaining edge portions that correspond to the bone contour in soft tissue images.SOLUTION: An image processing device comprises the following: image acquisition means for obtaining two types of captured images, a high-energy image captured with high-energy radiation and a low-energy image captured with low-energy radiation; normalization means for normalizing the pixel values of each of these two captured images; and energy subtraction means for generating a first image of a first tissue region and a first image of a second tissue region based on the two normalized captured images, and generating a second image of the first tissue region using an edge image of the first tissue region obtained from the first image of the first tissue region and the weighting coefficients obtained from the first image of the second tissue region.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to an image processing device, an image processing system, an image processing method, and a program. [Background technology]

[0002] The energy subtraction method is an imaging method that makes it easier to see a region of interest in a radiographic image by utilizing the fact that the X-ray absorption characteristics of materials vary depending on the X-ray energy. For example, Patent Document 1 discloses a method for generating a radiographic image using the energy subtraction method. Specifically, a subject is irradiated with high-energy X-ray beams and low-energy X-ray beams, and X-ray image data relating to these beams after they pass through the subject is acquired by an image detection means. Energy subtraction image data is generated by subtracting the two. Furthermore, the weighting of X-ray image data captured with a high-energy X-ray beam (hereinafter referred to as a high-energy image) and X-ray image data captured with a low-energy X-ray beam (hereinafter referred to as a low-energy image) is changed. This makes it possible to change the region of interest in the energy subtraction image data. As a result, it is possible to obtain energy subtraction image data in which bone tissue is erased and soft tissue is extracted as the region of interest, or conversely, energy subtraction image data in which soft tissue is erased and bone regions are extracted as the region of interest. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 06-022218 Summary of the Invention [Problem to be solved by the invention]

[0004] In order to create soft tissue images and bone tissue images suitable for radiography from high-energy images and low-energy images, appropriate weighting is necessary. However, the appropriate weights vary depending on complex relationships such as the energy of the X-rays, the thickness of the subject, the ratio of fat and muscle in the subject, the density of the bone, and the detection sensitivity characteristics of the X-ray detector with respect to energy. For this reason, it is difficult to set appropriate weights, and depending on the set weights, there is a risk that edge portions corresponding to bone contour portions may remain in the soft tissue image. When more detailed radiography is desired, it is desirable to reduce the remaining of such edge portions.

[0005] In view of such a background, one object of the present disclosure is to provide an image processing apparatus that reduces the remaining of edge portions.

Means for Solving the Problems

[0006] In order to solve the above problems, an image processing apparatus according to one aspect of the present disclosure includes: image acquisition means for acquiring two types of captured images, a high-energy image captured with high-energy radiation and a low-energy image captured with low-energy radiation; normalization means for normalizing the pixel values of each of the two types of captured images; energy subtraction means for generating a first image of a first tissue part and a second image of a second tissue part based on the two types of captured images that have been normalized, and generating the second image of the first tissue part using a weight coefficient obtained based on the edge image of the first tissue part obtained from the first image of the first tissue part and the first image of the second tissue part; and includes.

Effects of the Invention

[0007] According to one aspect of the present disclosure, it is possible to reduce the remaining of edge portions.

Brief Description of the Drawings

[0008]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Mode for Carrying Out the Invention

[0009] Hereinafter, examples will be described in detail with reference to the accompanying drawings. Note that the following examples do not limit the invention according to the claims. Although a plurality of features are described in the examples, not all of these plurality of features are essential to the invention, and the plurality of features may be arbitrarily combined. Further, in the accompanying drawings, the same or similar configurations are denoted by the same reference numerals, and redundant descriptions are omitted.

[0010] In addition, the radiation in the present disclosure includes, in addition to α-rays, β-rays, γ-rays, etc., which are beams formed by particles (including photons) emitted by radioactive decay, beams having energy equal to or higher than the same level, for example, X-rays, particle beams, cosmic rays, etc. may also be included.

[0011] <Example> Hereinafter, with reference to FIGS. 1 to 7, an image processing apparatus, an image processing system, and an image processing method according to an embodiment of the present disclosure will be described. Note that in the embodiment described below, an example will be described in which the energy of X-rays is changed by the X-ray tube 100, and a high-energy image and a low-energy image are acquired by irradiating the subject with X-rays twice.

[0012] First, the configuration of an example of an X-ray imaging apparatus used to acquire high-energy images and low-energy images in this embodiment will be described below with reference to FIG. 1. The illustrated X-ray imaging apparatus 1 includes an X-ray tube 100, an FPD (Flat Panel Detector) 200, an image processing unit 300, a display unit 400, and an operation unit 500. The X-ray tube 100 irradiates a subject 000 with X-rays, and the FPD 200, located behind the subject 000, converts the irradiated X-rays into image data. The converted image data is sent to the image processing unit 300, which performs image processing, which will be described later, on the image data. The display unit 400 displays on a monitor a radiographic image and a subtraction image generated from the image data after image processing. The operation unit 500 can be used to operate the image processing unit 300, input imaging information, and set parameters for image processing, such as brightness and contrast adjustment.

[0013] A known FPD can be used for the FPD 200. There are two types of FPDs: a linear conversion type, which directly converts X-rays into electric charges to obtain image data, and an indirect conversion type, which converts X-rays into light and then converts it into electric charges to obtain image data. Either of these may be used. The operation unit 500 may be implemented using, for example, a keyboard or a mouse, but may also be implemented using a GUI (Graphical User Interface) of the display unit 400. The FPD 200 is communicably connected to the image processing unit 300. Communication between them may be performed by wire or wirelessly.

[0014] Next, the X-ray imaging of the subject 000 performed using the X-ray imaging apparatus 1 will be described. First, X-rays are irradiated from the X-ray tube 100 to the FPD 200 on the extension line of the subject 000. Next, the X-rays are converted into image data by the irradiated FPD 200 and sent to the I / O unit 301 of the image processing unit 300. At this time, information regarding imaging during image capture such as the dose and tube voltage (hereinafter referred to as imaging information) may be sent from the X-ray tube 100 to the image processing unit 300. The acquisition of image data by the X-ray tube 100 and the FPD 200 is performed under two imaging conditions in which the quality of the X-rays is changed using an additional filter for changing the tube voltage or the quality of the X-rays in the X-ray tube 100. The two pieces of image data obtained under these two imaging conditions are sent to the I / O unit 301, which is the input / output unit of the image processing unit 300.

[0015] Here, the configuration of the image processing unit 300 illustrated in this embodiment will be described. The image processing unit 300 includes an I / O unit 301, a storage medium 302, a memory 307, and a CPU 308. The I / O unit 301 is used for input / output of image data and the like, output of processed images and the like, and input / output of image processing parameters and the like. The storage medium 302 has functional areas (303 to 306) for storing programs related to each process shown in the flowchart described later. Here, although the form in which the image processing unit 300 is integrated with the X-ray imaging apparatus 1 has been illustrated, the image processing unit 300 may be provided as a single unit and connected to a server or the like in the hospital.

[0016] The image processing unit 300 acquires image data by the I / O unit 301 and stores the acquired image data in the image storage unit 303 of the storage medium 302. The stored image data is temporarily read into the memory 307 together with each image processing program such as the normalization unit 304 stored in the storage medium 302 and is calculated by the CPU 308. However, for the arithmetic processing, an arithmetic device such as a GPU or an image processing chip may be used instead of the CPU 308. The image processing is completed by displaying the image obtained as a result of the image processing on the display unit 400.

[0017] Next, the processing of the image processing program stored in the storage medium 302 of the image processing unit 300 will be described using the flowchart of the image processing shown in FIG. 2. The processing of each subsequent step is performed by the CPU 308 executing the programs stored in each part of the storage medium 302.

[0018] When the image processing starts, first, in step S101, an image acquisition process is performed to acquire two types of image data obtained from two different qualities of X-rays from the storage medium 302. In the following description, among the two images with different qualities, the image with the high-energy quality is referred to as the high-energy image, and the other is referred to as the low-energy image. When the two types of images are acquired, the flow proceeds to step S102. Here, the image data is acquired from the storage medium 302, but it is also possible to acquire the image data from an external storage medium such as a server in the hospital, for example.

[0019] Next, in step S102, alignment processing is performed. The alignment processing aims to correct the positional shift on the two images due to the body movement of the subject 000 that occurs when the shooting times are different. Specifically, the positional shift of the other image is corrected with an arbitrary image among the low-energy image and the high-energy image as a reference. For the correction of the positional shift, a known method can be used. For example, alignment by template matching, machine learning-based alignment using a Convolutional Neural Network (hereinafter referred to as CNN), etc. can be used. After the alignment is completed, the flow proceeds to step S103.

[0020] In step S103, normalization processing is performed. The normalization processing aims to normalize the differences due to the exposure dose and quality based on the soft tissues of the low-energy image and the high-energy image after alignment. Hereinafter, the details of an example of the normalization processing will be further described using the flowchart shown in FIG. 3.

[0021] When the normalization process starts, in step S201, the soft tissue area extraction process is performed. The purpose of the soft tissue area extraction process is to extract the soft tissue area serving as the correction reference and create a soft tissue area map. Note that the extraction of the soft tissue area may be performed, for example, by histogram analysis processing or by machine learning using a CNN.

[0022] In this embodiment, as an example, the case where histogram analysis processing is used for the extraction of the soft tissue area will be described below with reference to the flowchart of FIG. 4. The histogram analysis processing illustrated in FIG. 4 includes a histogram extraction process (step S301), a straight line area removal process (step S302), a high pixel value threshold process (step S303), and a soft tissue map creation process (step S304). Here, the processes up to the histogram extraction process, the straight line area removal process, and the high pixel value threshold process are performed individually for each of the high-energy image and the low-energy image. The processes executed in each step will be described below.

[0023] When the histogram analysis process (soft tissue area extraction process) starts, in step S301, the histogram extraction process is performed. In the histogram extraction process, a histogram showing the relationship between the number of pixels and the pixel values included in the image is extracted. When the histogram is extracted, the flow proceeds to step S302.

[0024] In step S302, the straight line area removal process is performed. In the straight line area removal process, using the extracted histogram, the area where the X-ray reaches the FPD200 without passing through the subject is removed. Specifically, among the pixel values in the histogram obtained in step S301 that are not 0, the pixel value at the position 10% from the top of the width from the highest pixel value to the lowest pixel value is set as the threshold, and the pixels with pixel values equal to or higher than the threshold are removed from the image. Thereby, with respect to the image data (straight line area) caused by the X-ray that reaches the FPD200 without passing through the subject, it is possible to remove it from the target for extracting the soft tissue area. After the removal of the straight line area, the flow proceeds to step S303.

[0025] In step S303, high pixel value threshold processing is performed. In the high pixel value threshold processing, threshold processing is performed using a histogram to leave a high pixel value region where bones are not expected to be included in the image data. Specifically, among the non-zero pixel counts in the histogram obtained in step S301, the pixel value at the 50% position from the highest pixel value to the lowest pixel value of the width is used as the threshold, and pixels with pixel values below the threshold are removed from the image data. As a result, the region assumed to be the bone tissue region in the image data is largely removed. When the pixels with pixel values below the threshold are removed from the image data, the flow proceeds to step S304.

[0026] In step S304, soft tissue map creation processing is performed. In the soft tissue map creation processing, a soft tissue map is created based on the image data related to the high-energy image and the image data related to the low-energy image for which the processing from step S301 to step S303 has been executed. As a specific example, for example, in the non-excluded regions of the high-energy image and the low-energy image that have been processed up to the high pixel value threshold processing (step S303), a soft tissue map is created with the non-excluded regions of the pixels in both image data as soft tissues. Specifically, for mapping, for example, the pixel value of the soft tissue pixels can be set to 1 and the others to 0 so that they can be distinguished. By executing these processes, the soft tissue region extraction process in step S201 is completed, and a soft tissue map with the soft tissue region extracted can be created. When the soft tissue region is extracted, in the flowchart shown in FIG. 3, the flow proceeds from step S201 to step S202.

[0027] In step S202, an approximate equation coefficient derivation process is performed. In the approximate equation coefficient derivation process, the soft tissue map created in the soft tissue region extraction process in step S201, the high-energy image, and the low-energy image are used to derive coefficients of a correction equation to be used in the normalization process executed in step S103. As a specific example, the relationship between the pixel values of the pixels in the high-energy image and the low-energy image corresponding to pixels set as soft tissue in the soft tissue map is expressed by a linear approximate equation such as Equation 1 below. The coefficients of this linear approximate equation can be derived by using the least squares method or the like. Io=a×Ii+b …Equation 1 Here, Io represents the pixel value of the reference image data, Ii represents the pixel value of the image data other than the reference image data, and a and b represent the coefficients of the approximation formula.

[0028] The image data used as the reference for deriving the approximate equation coefficients may be either high-energy images or low-energy images. For example, the image to be used as the reference may be determined in advance for each imaging region, and the approximate equation coefficient derivation process may be performed based on this. In this specific example, the high-energy images and low-energy images are image data obtained by logarithmically transforming images that are linear with respect to dose, or image data in which pixel values have logarithmic characteristics with respect to dose at the time of image acquisition. However, the approximate equation is not limited to the linear approximate equation described above. Any known approximate equation that determines the relationship between two corresponding data items can be used, and the image data is not limited to those having the characteristics illustrated here. Once the approximate equation coefficients are obtained, the flow proceeds to step S203.

[0029] In step S203, an approximation formula application process is performed. In the approximation formula application process, according to the coefficients obtained in step S202, these are applied to the image data. As a specific example, regarding the case of using the coefficients obtained by Equation 1, the approximation formula is applied using the following Equation 2. In this case, the approximation formula is applied to the image data that is not the reference image data, and thereby a corrected image is created. Ic = a × Ii + b … Equation 2 Here, Ic represents the pixel value of the image after correction, Ii represents the pixel value of the image data that is not the reference image data, and a and b represent the coefficients of the approximation formula.

[0030] By executing each process from step S201 to step S203 in the flowchart shown in FIG. 3 described above, the normalization process in step S103 ends. Thereby, it is possible to create image data in which either the high-energy image or the low-energy image is normalized to the other. When the normalized image data is obtained, in the flowchart shown in FIG. 2, the flow proceeds to step S104.

[0031] In step S104, an energy subtraction process is performed. The energy subtraction process is performed using the high-energy image and the low-energy image after the execution of the normalization process in step S103. Here, a specific example of the energy subtraction process will be described using the flowchart shown in FIG. 5.

[0032] When the energy subtraction process starts, first, in step S401, a calculation image creation process is performed. The calculation image creation process uses the high-energy image and the low-energy image after the normalization process in step S103. Specifically, an average image and a provisional bone tissue image are created from the high-energy image and the low-energy image. The average image is obtained by calculating the average value of the pixel values of corresponding pixels in the two types of image data. The provisional bone tissue image is obtained by subtracting the image data of the low-energy image from the image data of the high-energy image after the normalization process in step S103. By subtracting the image data, soft tissues are canceled out, and a provisional bone tissue image can be created in which bones with different X-ray absorption characteristics remain.

[0033] Although it is possible to cancel out soft tissues by subtracting image data of a high-energy image from image data of a low-energy image, this embodiment shows an example in which a low-energy image is subtracted from a high-energy image. The average image and provisional bone tissue image described here are used in subsequent processes: bone edge extraction process (step S402), weight optimization process (step S403), and weight coefficient application process (step S404). Once the average image and provisional bone tissue image, which are images for calculation, are obtained, the flow proceeds to step S402.

[0034] In step S402, a bone edge extraction process is performed. The bone edge extraction process is performed using a temporary bone tissue image. A specific example of the bone edge extraction process will be described with reference to the flowchart shown in FIG. 6. In the bone edge extraction process illustrated in FIG. 6, first, an extracted bone region image is created in step S501, then a bone edge image is created from the extracted bone region image in the next step S502, and finally, the bone edge image is binarized in step S503 to create a bone edge map. The processes performed in each step will be described in detail below.

[0035] When the bone edge extraction process in step S402 starts, first, the bone region derivation process in step S501 is performed. As a specific example, in the bone region derivation process in step S501, pixels with high absolute values are extracted from the provisional bone tissue image created in the arithmetic image creation process in step S401. Thereby, only the bone part in the image data can be extracted. Note that as the extraction method, for example, using the pixel number ratio determined in advance from the higher histogram as a threshold value, pixels with a number higher than the threshold value may be extracted. When the bone region is derived, the flow proceeds to step S502.

[0036] In step S502, edge filter processing is performed. The edge filter processing is performed, for example, by using a known edge filter such as a Prewitt filter, a Sobel filter, or a Gaussian filter on the bone region extraction image created in the bone region derivation process in step S501. When the edge filter processing is performed, the flow proceeds to step S503.

[0037] In step S503, bone edge map generation processing is performed. In the bone edge map creation process, for example, for the bone edge image obtained in the edge filter processing in step S502, pixels with a high number are extracted using the pixel number ratio determined in advance from the higher histogram as a threshold value. More specifically, a bone edge map is created by setting the pixel value of pixels exceeding the threshold value to 1 and the pixel value of other pixels to 0. When the bone edge map generation process in step S503 ends and the bone edge map is obtained, the bone edge extraction process in step S402 ends, and the flow proceeds to step S403.

[0038] In step S403, weight coefficient optimization processing is performed. In the weight coefficient optimization processing, based on the average image and the provisional bone tissue image created in step S401, and the bone edge map created in step S402, the weight coefficients used when creating the soft tissue image and the bone tissue image are derived. A specific example of the weight coefficient optimization processing will be described using the flowchart shown in FIG. 7.

[0039] When the weighting coefficient optimization process of step S403 is started, in the temporary weighting coefficient application process of step S601, a bone tissue image for evaluation is created by applying the temporary weighting coefficient to the temporary bone tissue image created in the calculation image creation process of step S401. Once the bone tissue image for evaluation is created, the flow proceeds to step S602.

[0040] In step S602, a temporary soft-tissue image is created by subtracting the bone tissue image for evaluation from the average image created in the calculation image creation process in step S401. After the temporary soft-tissue image is created, the flow proceeds to step S603.

[0041] In step S603, an evaluation value calculation process is performed. In the evaluation value calculation process, an evaluation value is calculated to evaluate the amount of bone edges remaining in the provisional soft tissue image from the provisional soft tissue image obtained in step S602 and the bone edge map obtained in step S402. The specific method for calculating the evaluation value used in this embodiment will be described later. After the evaluation value is calculated, the flow proceeds to step S604.

[0042] In step S604, a weighting factor calculation process is performed. In the weighting factor calculation process, a weighting factor is calculated based on the evaluation value calculated in step S603 so as to reduce the remaining edges of bones. Specifically, the temporary weighting factor applied in step S601 is changed so as to increase or decrease by a predetermined value. After the weighting factor is calculated, the flow proceeds to step S605.

[0043] In step S605, a weight coefficient determination process is performed. In the weight coefficient convergence determination process, it is determined whether the weight coefficient has converged using the weight coefficient calculated in the weight coefficient calculation process of step S604 or, in addition to the weight coefficient, the provisional weight coefficient used in step S601. If it is determined that convergence has occurred, the weight coefficient optimization process of step S403, which was described in detail in FIG. 7, ends, and the flow shown in FIG. 5 proceeds to step S404. If it is determined that the weight coefficient has not converged, the flow returns to step S601, and the provisional weight coefficient application process is performed again.

[0044] Here, a specific example of the weight coefficient optimization process executed in step S403 will be described next. First, in the provisional weight coefficient application process of step S601, as the initial value of the provisional weight coefficient, a value may be determined randomly, or if the range within which the weight coefficient can take values is known to some extent, the maximum value, minimum value, etc. of the range can be used. Then, the determined provisional weight coefficient is multiplied by the pixel value of each pixel in the provisional bone tissue image. When the flow is returned from the weight coefficient convergence determination process of step S605, the weight coefficient obtained in the weight coefficient calculation process of step S604, which is the previous step, is applied. In the next provisional soft tissue image creation process of step S602, the provisional soft tissue image is created by subtracting the bone tissue image to which the provisional weight coefficient has been applied from the average image created in the calculation image creation process of step S401.

[0045] In the next evaluation value calculation process of step S603, an edge filter is applied to the provisional soft tissue image to generate a soft tissue edge image. Then, in the soft tissue edge image, an evaluation value is calculated from the feature amount of the pixel corresponding to the bone edge in the bone edge map. At this time, as the edge filter, a known edge filter such as a Prewitt filter, a Sobel filter, or a Gaussian filter may be used. As the feature amount, the sum, average value, median value, mode value of the histogram, etc. of the pixel values may be used. In the next weight coefficient calculation process of step S604, the weight coefficient is calculated based on the evaluation value obtained in step S603. As the calculation method, a known technique such as Newton's method, Halley's method, or the bisection method may be used.

[0046] In the next step S605, the weighting coefficient convergence determination process determines whether convergence has occurred based on the weighting coefficient calculated in the weighting coefficient calculation process in step S604 or based on both the weighting coefficient and the provisional weighting coefficient used in step S601. For example, in a case where the weighting coefficient calculated in step S604 is used, step S605 compares the weighting coefficient with a preset threshold to determine whether a convergence determination condition is met. For example, if the average value of the edge is the evaluation value, the closer it is to 0, the greater the edge reduction. Therefore, convergence can be determined if the average value is lower than the threshold. In a case where the weighting coefficient calculated in step S604 and the provisional weighting coefficient used in step S601 are used, convergence can be determined if the difference between these two coefficients is equal to or less than the threshold. The weighting coefficient optimization process in step S403 can be performed through the above process. After the process shown in FIG. 7 is completed and the optimized weighting coefficient is obtained in step S403, the flow proceeds to step S404 in FIG. 5.

[0047] In step S404 of FIG. 5, a weighting coefficient application process is performed. In the weighting coefficient optimization process, a soft tissue image and a bone tissue image are created based on the weighting coefficient optimized in step S403. Specifically, the bone tissue image is created by multiplying each pixel value of the temporary bone tissue image obtained in step S401 by the weighting coefficient optimized in step S403. In addition, the soft tissue image can be created by dividing each pixel value of the bone tissue image by each pixel value of the average image obtained in the calculation image creation process in step S401. Once the bone tissue image and the soft tissue image are obtained, the energy subtraction process in step S104 of FIG. 2 is completed, and the flow proceeds to step S105.

[0048] Next, in step S105 of Fig. 2, image processing for display is performed. In the image processing for display, image processing is performed on the tissue image obtained in step S104 to obtain an image that is easy for the user to diagnose. Examples of well-known techniques used in step S105 include noise adjustment processing, gradation processing, and window adjustment processing. The image that has undergone these image processes is displayed on the display unit 400.

[0049] As described above, the image processing apparatus according to the present disclosure includes an image acquisition unit, a normalization unit, and an energy subtraction unit. In the present disclosure, the CPU 308 functions as an example of the image acquisition unit, and acquires two types of captured images, a high-energy image captured with high-energy radiation and a low-energy image captured with low-energy radiation. Further, the CPU 308 also functions as an example of the normalization unit by the program of the normalization unit 304, and can perform normalization of the pixel values of each of the two types of captured images. Furthermore, the CPU 308 also functions as an example of the energy subtraction unit by the program of the energy subtraction unit 305. Specifically, the CPU 308 can generate a first image (temporary bone tissue image) of the bone tissue and a first image (temporary soft tissue image) of the soft tissue based on the two types of captured images that have been normalized according to the program. Then, the CPU 308 can generate a second image (bone tissue image for evaluation) of the bone tissue using the weight coefficient obtained from the bone edge image obtained from the first image of the bone tissue and the first image of the soft tissue.

[0050] The CPU 308 that functions as an example of the energy subtraction unit can generate a second image of the bone tissue by multiplying each pixel value of the first image of the bone tissue by a weight coefficient (temporary weight coefficient and weight coefficient). Further, at that time, the CPU 308 can generate a bone edge map using the bone edge image. The CPU 308 obtains an evaluation value regarding the relationship between the edge pixels of the first image of the soft tissue and the corresponding pixels of the generated bone edge map, changes the weight coefficient based on the obtained evaluation value, and can optimize the weight coefficient.

[0051] Furthermore, the CPU 308, which functions as an example of an energy subtraction means, generates a difference image by subtracting the low-energy image from the high-energy image in the two normalized captured images. The generated difference image is a first image of bone tissue because the soft tissue areas in both energy images are offset. The CPU 308 can also extract a bone edge image by applying edge filtering to the obtained first image of bone tissue.

[0052] Furthermore, the CPU 308, which functions as an example of an energy subtraction means, generates an average image from the two normalized captured images. Then, a first image of the soft tissue can be generated by subtracting each pixel value obtained by multiplying each pixel value of the first image of the bone tissue by a temporary weighting coefficient from each pixel value of the average image. At this time, the CPU 308 can calculate an evaluation value for evaluating the weighting coefficient based on a soft tissue edge image obtained by applying edge filtering to the first image of the soft tissue and a bone edge map obtained using the bone edge image. Furthermore, the CPU 308 can calculate the sum of the pixel values of the pixels of the soft tissue edge image arranged in the bone edge map as the evaluation value. At this time, the CPU 308 can change the weighting coefficient so that the calculated evaluation value is an evaluation value in which the pixels of the bone edge image are reduced.

[0053] Furthermore, CPU 308, which functions as an example of an energy subtraction means, generates an average image from the two normalized captured images. Then, a second image of the soft tissue is generated by dividing each pixel value of the average image by each pixel value of the second image of the bone tissue. This second image of the soft tissue is displayed on display unit 400 and used for interpretation of the soft tissue.

[0054] Also, in the present disclosure, the CPU 308 can function as an example of normalization means. At this time, the CPU 308 extracts, for example, a region substantially corresponding to soft tissue from each of two types of captured images using a histogram. Then, the CPU 308 calculates a correction coefficient for normalization based on the pixel value of one pixel in the region extracted from one of the two types of captured images and the pixel value of the other pixel corresponding to the one pixel in the region extracted from the other captured image. At this time, assuming the obtained correction coefficients are a and b, Io is the pixel value of one of the captured images before correction, and Ic is the pixel value of the other captured image after correction, normalization of the two types of images can be performed based on the relational expression Ic = a × Io + b.

[0055] Note that the present disclosure can also constitute an image processing system. The image processing system includes a radiation imaging device (FPD 200) that acquires a radiation image of a subject 000 based on radiation, and the above-described image processing device (image processing unit 300). The radiation imaging device and the image processing device are communicably connected. Further, the present disclosure can also construct a control method for causing the above-described image processing device (image processing unit 300) to execute a blog program related to each step of the above-described image processing stored in the storage medium 302.

[0056] Furthermore, in the present disclosure, for the purpose of reducing bone edge remnants in radiographic images, the weighting coefficients are adjusted to reduce pixels associated with bone edges located in the soft tissue image obtained using the weighting coefficients. However, the soft tissue image and the bone tissue image used in image processing in the above-described embodiments may be interchanged, and the weighting coefficients may be adjusted to increase pixels associated with soft edges located in the bone tissue image obtained using the weighting coefficients. Obtaining a soft tissue image using the bone tissue image obtained in this manner can reduce bone edge remnants in radiographic images. Therefore, the bone and soft tissue in the above-described disclosure are preferably considered as first and second tissue portions that can be interchanged during image processing. Furthermore, the bone tissue image is preferably considered as an image of the first tissue portion, and the soft tissue image is preferably considered as an image of the second tissue portion. In this case, for example, the first image of the bone tissue described above can be considered as a first image of the first tissue portion. According to the above-described embodiments, it is possible to reduce bone edge remnants from high-energy images and low-energy images, thereby creating and displaying soft tissue images more suitable for diagnosis.

[0057] (Other Examples) The present disclosure can also be realized by supplying a program that achieves the above-mentioned functions to a system or device via a network or storage medium, and having one or more processors in the computer of that system or device read and execute the program.

[0058] Furthermore, various recording media can be used, such as flexible disks, optical disks (e.g., CD-ROMs, DVD-ROMs), magneto-optical disks, magnetic tapes, non-volatile memories (e.g., USB memories), ROMs, etc. Furthermore, the programs that implement the above-described functions may be downloaded via a network and executed by a computer.

[0059] Moreover, the present invention is not limited to the case where the functions of the above-described embodiments are realized only by executing the program code read by the computer. Based on the instructions of the program code, an OS (operating system) or the like running on the computer may perform part or all of the actual processing, and the functions of the above-described embodiments may be realized by such processing.

[0060] Furthermore, the program code read from the recording medium may be written into a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer. Based on the instructions of the program code, a CPU or the like provided in the function expansion board or function expansion unit may perform part or all of the actual processing, and the above-described functions may be realized by such processing.

[0061] The above disclosure includes the following configurations, methods, and programs. (Configuration 1) Image acquisition means for acquiring two types of captured images, a high-energy image captured with high-energy radiation and a low-energy image captured with low-energy radiation; Normalization means for normalizing the pixel values of each of the two types of captured images; Energy subtraction means for generating a first image of a first tissue part and a first image of a second tissue part based on the two types of captured images that have been normalized, and generating a second image of the first tissue part using a weight coefficient obtained based on an edge image of the first tissue part obtained from the first image of the first tissue part and the first image of the second tissue part; An image processing apparatus comprising the above. (Configuration 2) The energy subtraction means generates the second image of the first tissue part by multiplying the weight coefficient by each pixel value of the first image of the first tissue part. The image processing apparatus according to Configuration 1. (Configuration 3) 3. The image processing device according to configuration 1 or 2, wherein the energy subtraction means generates an edge map of the first tissue portion using the edge image of the first tissue portion, and changes the weighting coefficient based on an evaluation value regarding a relationship between an edge pixel of the first image of the second tissue portion and a corresponding pixel of the edge map of the first tissue portion. (Configuration 4) 4. The image processing device according to any one of configurations 1 to 3, wherein the energy subtraction means generates a difference image between the two normalized captured images, and generates a first image of the first tissue portion using the difference image. (Configuration 5) 5. The image processing device according to any one of configurations 1 to 4, wherein the energy subtraction means extracts an edge image of the first tissue portion by applying edge filtering to the first image of the first tissue portion. (Configuration 6) The image processing device according to any one of configurations 1 to 5, wherein the energy subtraction means generates an average image from the two types of normalized captured images, and generates the first image of the second tissue portion by subtracting each pixel value obtained by multiplying each pixel value of the first image of the first tissue portion by a temporary weighting coefficient from each pixel value of the average image. (Configuration 7) The image processing device according to configuration 6, wherein the energy subtraction means calculates an evaluation value for evaluating the weighting coefficient based on an edge image of the second tissue portion obtained by applying edge filter processing to a first image of the second tissue portion and an edge map of the first tissue portion obtained using the edge image of the first tissue portion. (Configuration 8) 8. The image processing device according to configuration 7, wherein the energy subtraction means calculates, as the evaluation value, a sum of pixel values of pixels of the edge image of the second tissue portion arranged in the edge map of the first tissue portion. (Configuration 9) 9. The image processing device according to configuration 8, wherein the energy subtraction means changes the weighting coefficient so that the calculated evaluation value is an evaluation value in which pixels of the edge image of the first texture portion are reduced. (Configuration 10) 10. The image processing device according to any one of configurations 1 to 9, wherein the energy subtraction means generates an average image from the two types of normalized captured images, and generates a second image of the second tissue portion by dividing each pixel value of the average image by each pixel value of the second image of the first tissue portion. (Configuration 11) The image processing device according to any one of configurations 1 to 10, wherein the normalization means extracts an area substantially corresponding to the second tissue portion from each of the two types of captured images, and calculates a correction coefficient for normalization based on a pixel value of one pixel in the area extracted from one of the two types of captured images and a pixel value of another pixel corresponding to the one pixel in the area extracted from the other captured image. (Configuration 12) The image processing device according to configuration 11, wherein the normalization means normalizes the two types of images based on a relational expression Ic=a×Io+b, where Io is a pixel value of one of the captured images before correction, a and b are correction coefficients, and Ic is a pixel value of the other captured image after correction. (Configuration 13) a radiographic imaging device for acquiring a radiographic image of a subject based on radiation; an image processing device according to any one of configurations 1 to 12, which is communicably connected to the radiation imaging device; An image processing system comprising: (method) Obtaining two types of images, a high-energy image taken with high-energy radiation and a low-energy image taken with low-energy radiation; normalizing pixel values of each of the two types of captured images; generating a first image of a first tissue portion and a first image of a second tissue portion based on the two types of normalized captured images, and generating a second image of the first tissue portion using a weighting coefficient obtained based on an edge image of the first tissue portion obtained from the first image of the first tissue portion and the first image of the second tissue portion; An image processing method comprising: (program) When executed by a computer, a program for causing the computer to execute each step of the image processing method described in the method on the computer.

[0062] As described above, the present disclosure has been described with reference to the embodiments, but the present disclosure is not limited to the above embodiments. Inventions modified within the scope not contrary to the gist of the present disclosure, and inventions equivalent to the present disclosure are also included in the present disclosure.

Explanation of Reference Numerals

[0063] 000 Subject 100 X-ray tube 200 FPD 300 Image processing apparatus 301 I / O unit 302 Storage medium 303 Image storage unit 304 Normalization unit 305 Energy subtraction unit 306 Display image unit 307 Memory 308 CPU 400 Display device 500 Operating device

Claims

1. Image acquisition means for acquiring two types of captured images, a high-energy image captured with high-energy radiation and a low-energy image captured with low-energy radiation; Normalization means for normalizing the pixel values of each of the two types of captured images; Energy subtraction means for generating a first image of a first tissue part and a first image of a second tissue part based on the two types of captured images that have been normalized, and generating a second image of the first tissue part using a weight coefficient obtained based on an edge image of the first tissue part obtained from the first image of the first tissue part and the first image of the second tissue part; An image processing apparatus comprising the above.

2. The energy subtraction means according to claim 1, wherein the second image of the first tissue part is generated by multiplying the weight coefficient by each pixel value of the first image of the first tissue part.

3. The energy subtraction means according to claim 1, wherein the energy subtraction means generates an edge map of the first tissue part using the edge image of the first tissue part, and changes the weight coefficient based on an evaluation value regarding the relationship between the edge pixels of the first image of the second tissue part and the corresponding pixels of the edge map of the first tissue part.

4. The energy subtraction means according to claim 1, wherein the energy subtraction means generates a difference image of the two types of captured images that have been normalized, and generates the first image of the first tissue part using the difference image.

5. The energy subtraction means according to claim 1, wherein the energy subtraction means extracts the edge image of the first tissue part by applying an edge filter process to the first image of the first tissue part.

6. The energy subtraction means according to claim 1, wherein the energy subtraction means generates an average image from the two types of captured images that have been normalized, and subtracts each pixel value obtained by multiplying each pixel value of the average image by a provisional weight coefficient from each pixel value of the first image of the first tissue part to generate the first image of the second tissue part.

7. The energy subtraction means according to claim 6, wherein the energy subtraction means calculates an evaluation value for evaluating the weight coefficient based on an edge image of the second tissue part obtained by applying an edge filter process to the first image of the second tissue part and an edge map of the first tissue part obtained using the edge image of the first tissue part.

8. The image processing apparatus according to claim 7, wherein the energy subtraction means calculates, as the evaluation value, the sum of the pixel values of the pixels of the edge image of the second tissue part arranged in the edge map of the first tissue part.

9. The image processing apparatus according to claim 8, wherein the energy subtraction means changes the weight coefficient so that the calculated evaluation value becomes an evaluation value in which the pixels of the edge image of the first tissue part are reduced.

10. The image processing apparatus according to claim 1, wherein the energy subtraction means generates an average image from the two types of normalized captured images, and divides each pixel value of the average image by each pixel value of the second image of the first tissue part, thereby generating a second image of the second tissue part.

11. The image processing apparatus according to claim 1, wherein the normalization means extracts regions substantially corresponding to the second tissue part from each of the two types of captured images, and based on the pixel value of one pixel in the region extracted from one of the two types of captured images and the pixel value of the other pixel corresponding to the one pixel in the region extracted from the other captured image, calculates a correction coefficient for normalization.

12. The image processing apparatus according to claim 11, wherein the normalization means performs normalization of the two types of images based on the relational expression Ic = a × Io + b, where Io is the pixel value of the one captured image before correction, a and b are correction coefficients, and Ic is the pixel value of the other captured image after correction.

13. A radiation imaging apparatus that acquires a radiation image of a subject based on radiation, The image processing apparatus according to any one of claims 1 to 12, communicably connected to the radiation imaging apparatus, An image processing system comprising:

14. Obtaining two types of captured images, a high-energy image captured with high-energy radiation and a low-energy image captured with low-energy radiation, Normalizing the pixel values of each of the two types of captured images, Generating a first image of a first tissue part and a first image of a second tissue part based on the two types of normalized captured images, and generating a second image of the first tissue part using a weight coefficient obtained based on the edge image of the first tissue part obtained from the first image of the first tissue part and the first image of the second tissue part, An image processing method including:

15. A program that, when executed by a computer, causes the computer to execute each step of the image processing method according to claim 14.

Images