JP2013119021A - X-ray ct device and image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To recover the contrast of a blurred structure without generating image quality difference or unevenness in a contour of the structure while holding noise reduction effects in an image (SS reconfigured image) reconfigured using corrected projection data.SOLUTION: An arithmetic unit 5 generates an SS edge map representing an edge region in an SS reconfigured image (S403), generates, from the SS edge map, an SS combination map indicating a boundary to combine an NSS reconfigured image to the SS reconfigured image (S404), and generates an NSS edge map representing an edge region in the NSS reconfigured image (S405). Then, an SM-NSS reconfigured image is generated by applying smoothing processing to the NSS reconfigured image (S406), the SS combination map is corrected using the NSS edge map and an NSS combination map is generated (S407). On the basis of the NSS combination map, a synthetic image is generated by combining the SS reconfigured image, the NSS reconfigured image and the SM-NSS reconfigured image (S408).

Description

  The present invention irradiates a subject with X-rays, measures X-rays transmitted through the subject with an X-ray detector, and reconstructs measurement data (projection data) from multiple directions, thereby tomographic images of the subject. The present invention relates to an X-ray CT apparatus for obtaining In particular, the present invention relates to an X-ray CT apparatus or the like that corrects projection data and performs contrast recovery processing on an image reconstructed using the corrected projection data.

  In the X-ray CT apparatus, X-rays are irradiated from around the subject, and real projection data acquired at a plurality of projection angles are back-projected, whereby a tomographic image of the subject is imaged.

  In the X-ray CT apparatus as described above, a technique for correcting projection data and generating a reconstructed image with reduced noise using the corrected projection data has been proposed. One of the techniques is a technique using a successive approximation projection data filter. This is a technique in which projection data evaluation indexes are set in advance, and projection data is sequentially updated so that an evaluation value obtained by quantifying the evaluation index takes a maximum value or a minimum value. As the evaluation index, a contradiction between the temporary projection data to be updated and the actual projection data, probabilistic likelihood, or the like is used. Non-Patent Document 1 proposes a successive approximation projection data filter that uses a penalized weighted square error function as an evaluation function. By applying the successive approximation reconstruction and the successive approximation projection data filter, it is possible to effectively reduce streak-like artifacts and quantum noise generated on the image when the detector output is small.

Thereafter, an image reconstructed using the projection data corrected in this way is referred to as SS (Sinogram).
Smoothing) Reconstructed image, NSS (Non Sinogram)
Smoothing) This is called a reconstructed image.

  By the way, the SS reconstructed image has a problem that the contrast is lowered due to the outline of the structure being blurred. Such blurring of the structure is not desirable because it may reduce the visibility of the diagnostic object, and in particular, may reduce the accuracy of diagnosis for a diagnostic object having a fine and low contrast structure. In particular, in the case of performing noise reduction by correcting projection data such as a successive approximation projection data filter, the degree of smoothing of data corresponding to an element having a small detector output is large. Therefore, the blur of the outline of a structure having a high CT value such as a bone becomes large.

  On the other hand, in Patent Document 1, a sharpened image (an image subjected to edge enhancement processing) is mixed for each matrix of an arbitrary size surrounding each pixel according to the amount of analysis related to a plurality of CT values in the matrix. Thus, there has been proposed a method for suppressing a reduction in image resolution while removing noise.

WO2005 / 110232 publication

T. Li et. Al., “NonlinearSinogram Smoothing for Low-Dose X-Ray CT,” IEEE. Trans. Nucl. Sci., Vol. 51, No. 5, pp. 2505-2513, 2004 K. Saueret. Al., “A Local Update Strategy for Iterative Reconstru CTion form ProjeCTions,” IEEE. Trans. Signal.Proc., Vol.41, No.2, pp.534-548, 1993

  However, the method described in Patent Document 1 has a problem that a strong streak artifact or strong noise is treated equally with the structure and noise, and may not be reduced.

  The present invention has been made in view of the above-described problems, and an object thereof is to maintain the noise reduction effect of an image reconstructed using the corrected projection data (SS reconstructed image). On the other hand, it is an object of the present invention to provide an X-ray CT apparatus or the like that can restore the contrast of a blurred structure without causing a gap in image quality or unevenness in the outline of the structure.

  In order to achieve the above-described object, the first invention includes an X-ray source for irradiating X-rays, an X-ray detector for detecting the X-rays, a bed for placing a subject, and the X-ray source. A rotating disk mounted with the X-ray detector and rotating around the subject; and a storage device for storing actual projection data obtained by converting the X-ray detected by the X-ray detector. A line CT apparatus, wherein a first image is reconstructed using data obtained by performing a first noise reduction process on the actual projection data, and further, the actual projection data is compared with the first noise reduction process. Also included in both the first image and the second image, the data subjected to the second noise reduction processing with a low noise reduction degree, or the image reconstruction unit for reconstructing the second image using the actual projection data A predetermined structure to be obtained from the first image of the structure. A boundary extraction unit that extracts a first boundary that is a boundary, and extracts a second boundary that is a boundary of the structure from the second image, and a region between the first boundary and the second boundary, An X-ray CT apparatus comprising: an image composition unit that performs composition processing using at least the first image and the second image and generates a composite image.

  A second invention is equipped with an X-ray source for irradiating X-rays, an X-ray detector for detecting the X-rays, a bed for placing a subject, the X-ray source and the X-ray detector. Image processing executed by an X-ray CT apparatus comprising: a rotating disk that rotates around the subject; and a storage device that stores actual projection data obtained by converting the X-rays detected by the X-ray detector. A first image is reconstructed using data obtained by subjecting the actual projection data to a first noise reduction process; and further, the actual projection data is subjected to noise more than the first noise reduction process. A step of reconstructing the second image using the data subjected to the second noise reduction processing having a low degree of reduction or the actual projection data, and a predetermined structure included in both the first image and the second image Is the boundary of the structure from the first image Extracting one boundary, and further extracting a second boundary, which is a boundary of the structure, from the second image; and at least the first image and the region between the first boundary and the second boundary; Performing a synthesis process using the second image to generate a synthesized image.

  According to the present invention, while maintaining the noise reduction effect of an image reconstructed using the corrected projection data (SS reconstructed image), the outline of the structure is blurred without causing image quality gaps or irregularities. It is possible to restore the contrast of the structure that has been closed. In addition, the degree of contrast recovery can be adjusted.

Overall appearance of X-ray CT apparatus 1 Configuration diagram of X-ray CT apparatus 1 The figure which shows the outline | summary of this invention The figure which shows the flow of the image processing of this invention The figure which shows the flow of air area judgment processing Diagram showing the flow of edge map generation processing Diagram showing edge extraction by Sobel filter The figure which shows the flow of an edge area | region binarization process The figure which shows the flow of SS synthetic | combination map production | generation process by SS reconstruction image Diagram showing the flow of edge-preserving smoothing processing The figure which shows the flow of SS reconstruction image direction utilization direction dependence filter processing The figure which shows the flow of SS synthetic map correction processing using the edge of a NSS reconstruction image Diagram showing the flow of boundary area weighted addition processing Figure showing an example of a composite image Diagram showing an example of a large difference map The figure which shows the production | generation process of the distance image from the boundary with SM-NSS reconstruction image area | region The figure which shows the production | generation process of the distance image from the boundary with SS reconstruction image area | region The figure which shows an example in case the boundary with SS reconstruction image area is lost The figure which shows the production | generation process process of the distance image from the boundary with SS reconstruction image area (continuation of FIG. 17)

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. First, the configuration of the X-ray CT apparatus 1 will be described with reference to FIGS. 1 and 2.
As shown in FIG. 1, the X-ray CT apparatus 1 processes data obtained from a scanner 2 on which an X-ray tube 11 and a detector 12 are mounted, a bed 4 on which a subject 10 is placed, and a detector 12. An arithmetic device 5, an input device 6 such as a mouse, a trackball, a keyboard, and a touch panel, a display device 7 for displaying a reconstructed image, and the like are included.
The operator inputs imaging conditions, reconstruction parameters, and the like via the input device 6. The imaging conditions are, for example, a bed feeding speed, a tube current, a tube voltage, and a slice position. The reconstruction parameter is, for example, a region of interest, a reconstructed image size, a backprojection phase width, a reconstruction filter function, or the like.

As shown in FIG. 2, the X-ray CT apparatus 1 is roughly composed of a scanner 2, an operation unit 3, and a bed 4.
The scanner 2 includes an X-ray tube 11 (X-ray source), a detector 12 (X-ray detector), a collimator 13, a driving device 14, a central control device 15, an X-ray control device 16, a high voltage generation device 17, and a scanner control. The apparatus 18 includes a bed control device 19, a bed movement measurement device 20, a collimator control device 21, a preamplifier 22, an A / D converter 23, and the like.

The central control device 15 inputs imaging conditions and reconstruction parameters from the input device 6 in the operation unit 3, and sends control signals necessary for imaging to the collimator control device 21, the X-ray control device 16, the scanner control device 18, and the bed control. Transmit to device 19.
The collimator control device 21 controls the position of the collimator 13 based on the control signal.
When imaging is started in response to the imaging start signal, the X-ray control device 16 controls the high voltage generator 17 based on the control signal. The high voltage generator 17 applies a tube voltage and a tube current to the X-ray tube 11. In the X-ray tube 11, electrons having energy corresponding to the applied tube voltage are emitted from the cathode, and the emitted electrons collide with the target (anode), whereby X-rays having energy corresponding to the electron energy are converted into the subject 10. Is irradiated.
The scanner control device 18 controls the drive device 14 based on the control signal. The driving device 14 circulates around the subject 10 around the gantry portion on which the X-ray tube 11, the detector 12, the preamplifier 22, and the like are mounted.
The bed control device 19 controls the bed 4 based on the control signal.

  The X-ray irradiated from the X-ray tube 11 is limited in the irradiation region by the collimator 13, absorbed (attenuated) in accordance with the X-ray attenuation coefficient in each tissue in the subject 10, passes through the subject 10, It is detected by a detector 12 arranged at a position facing the line tube 11. The detector 12 is configured by a plurality of detection elements arranged in a two-dimensional direction (a channel direction and a column direction perpendicular thereto). X-rays received by each detection element are converted into actual projection data. That is, the X-rays detected by the detector 12 are converted into current, amplified by the preamplifier 22, converted into digital data by the A / D converter 23, LOG converted, calibrated, and used as actual projection data. Input to the arithmetic unit 5.

  At this time, since the X-ray tube 11 and the detector 12 facing each other rotate around the subject 10, the actual projection data is obtained at discrete X-ray tube positions (and corresponding detector positions) in the rotation direction. Collected. A unit of acquisition of actual projection data at each X-ray tube position is called a “view”.

The arithmetic device 5 includes a reconstruction arithmetic device 31, an image processing device 32, and the like. The input / output device 9 includes an input device 6, a display device 7, a storage device 8 (storage unit), and the like.
The reconstruction calculation device 31 performs an image reconstruction process using the actual projection data, and generates a reconstructed image. The reconstruction calculation device 31 generates filter correction projection data by superimposing a reconstruction filter on the actual projection data of each view, and weights the filter correction projection data in the view direction (hereinafter referred to as “view direction weight”). .) Is weighted and the back projection process is performed to form a non-destructive tomographic image as a distribution map of the X-ray attenuation coefficient inside the subject 10.
The reconstruction calculation device 31 stores the generated reconstruction image in the storage device 8. In addition, the reconstruction calculation device 31 displays the CT image on the display device 7. Alternatively, the image processing device 32 performs image processing on the reconstructed image stored in the storage device 8 and displays it on the display device 7 as a CT image.

  The X-ray CT apparatus 1 is a multi-slice CT that uses a detector 12 in which detection elements are arranged in a two-dimensional direction, and a single that uses a detector 12 in which detection elements are arranged in one row, that is, in a one-dimensional direction (channel direction only). It is roughly divided into slice CT. In multi-slice CT, an X-ray beam that spreads in a cone shape or a pyramid shape is emitted from an X-ray tube 11 that is an X-ray source in accordance with the detector 12. In the single slice CT, an X-ray beam spreading in a fan shape is emitted from the X-ray tube 11. Usually, in imaging by the X-ray CT apparatus 1, X-ray irradiation is performed while the gantry unit circulates around the subject 10 placed on the bed 4.

  First, an outline of the present embodiment will be described with reference to FIG. In the present embodiment, the arithmetic device 5 reconstructs the SS reconstructed image 50 using data obtained by performing noise reduction processing on the actual projection data, and further, the SS reconstructed image 50 with respect to the actual projection data. The NSS reconstructed image 60 is reconstructed using data that has been subjected to noise reduction processing that has a lower noise reduction degree than the time of reconstruction, or actual projection data. For example, the arithmetic unit 5 corrects the actual projection data using the successive approximation projection data filter, sets the image reconstructed using the modified data as the SS reconstructed image 50, and uses the successive approximation projection data filter. An image reconstructed using uncorrected data (= actual projection data) is referred to as an NSS reconstructed image 60. As the actual projection data correction process using the successive approximation projection data filter, for example, the actual projection data is corrected by the successive approximation method including the photon weight according to the detector output of the X-ray detector and the penalty term in the evaluation function. The process to do is mentioned. For example, the following expression described in Non-Patent Document 1 is used as described below.

1,..., I,..., I are channel and column, view serial numbers. p _i and y _i are provisional projection data and actual projection data in the i-th detection element. p = (p ₁ ,..., p _i ,..., p _I ) is a vector of provisional projection data. κ _i is a set of proximity elements of the i-th detection element. ν _ik is a constant that determines the correlation between the i th detection element and the k th detection element. In Non-Patent Document 1, ν _ik is determined empirically in the channel, column, and view direction. Ψ (p _i −p _k ) is a potential function having the contrast between the i-th provisional projection data and the k-th provisional projection data as a variable. In Non-Patent Document 1, a quadratic function is used as the potential function. d _i is a weight coefficient (hereinafter referred to as “photon weight”) weighted to the difference value between the i-th actual projection data and the forward projection data. In the image reconstruction of the X-ray CT apparatus, the value reflects the detector output of the i-th detection element.

  In the update formula derived from the formula (1) by various numerical analysis methods, the provisional projection data in the update process is limited by the second term while being constrained by the actual projection data and the photon weight by the first term. Smoothing is performed according to the contrast between the projection data. Therefore, the arbitrary constant β in the equation (1) that plays the role of adjusting the balance between the first term and the second term is a parameter for adjusting the degree of smoothing of the provisional projection data.

  Photon weights play an important role in successive approximation reconstruction and successive approximation projection data filters. In order to make the explanation easy to understand, it is assumed here that the view direction and the channel direction are two-dimensional. In a horizontal view (a view in which the X-ray tube position is to the left or right of the subject), X-rays are transmitted in the body width direction of the human body, so a vertical view (the X-ray tube position is above or below the subject). There are many channels with a long transmission length in the human body and a small detector output. As a result, streak-like artifacts are generated mainly in the horizontal direction in the tomographic image. Since the photon weight in equation (1) is a value corresponding to the detector output, it is smaller in the horizontal view channel than in the vertical view channel. Accordingly, in the view in the horizontal direction rather than the view in the vertical direction, the restriction by the first term of the expression (1) is relatively weak and the smoothing effect by the second term is strongly received.

  The SS reconstructed image 50 and the NSS reconstructed image 60 shown in FIG. 3 are schematic diagrams in which a portion of the same structure 40 (for example, a bone) is enlarged. A hatched area indicates the original position of the structure 40.

  Since the edge (contour) of the structure 40 is blurred in the SS reconstructed image 50, an edge that is wider than the original structure 40 is extracted when the edge extraction process (boundary extraction process) is performed. In the SS reconstructed image 50 shown in FIG. 3, the SS edge 51 illustrated by the solid line indicates the edge extracted by the edge extraction process.

  On the other hand, the NSS reconstructed image 60 includes the noise represented by the streak artifact 62 although the edge (contour) of the structure 40 is clear. When edge extraction processing (boundary extraction processing) is performed on the NSS reconstructed image 60, an edge close to the original structure 40 is extracted. In the NSS reconstructed image 60 shown in FIG. 3, an NSS edge 61 illustrated by a solid line indicates an edge extracted by the edge extraction process. However, the NSS reconstructed image 60 includes the streak artifact 62, and there is a problem in the image quality around the edge of the structure 40.

  Therefore, in the present embodiment, the arithmetic device 5 extracts the edge (boundary) of the structure 40 from the SS reconstruction image 50 for the structure 40 included in both the SS reconstruction image 50 and the NSS reconstruction image 60. Further, the edge (boundary) of the structure 40 is extracted from the NSS reconstructed image 60. Then, at least the SS reconstructed image 50 and the NSS reconstructed image 60 for the region between the SS edge 51 extracted from the SS reconstructed image 50 and the NSS edge 61 extracted from the NSS reconstructed image 60 is calculated by the arithmetic unit 5. Then, a composite image 70 is generated. The composite image 70 generated in this way retains the noise reduction effect of the SS reconstructed image 50, and does not cause a gap in image quality or unevenness in the outline of the structure 40. The contrast has been restored.

  As will be described later, for the outer side of the SS edge 51 and the inner side of the NSS edge 61, the arithmetic device 5 is an image obtained by smoothing the SS reconstructed image 50, the NSS reconstructed image 60, or the NSS reconstructed image 60. The composite image 70 is generated using the CT value of an appropriate image. Further, the arithmetic device 5 also generates a composite image 70 using a CT value of an image obtained by smoothing the NSS reconstructed image 60 for a part of the region between the SS edge 51 and the NSS edge 61.

  Next, the features and effects of image processing in this embodiment will be described with reference to FIG. The image processing in this embodiment is divided into 10 steps as shown in FIG.

In S401, the arithmetic unit 5 generates a reconstruction area map indicating a reconstruction FOV range such as a circle.
In S402, the arithmetic unit 5 generates a map (= air map) indicating the air region.
In S403, the arithmetic unit 5 generates edge data indicating the edge strength of the SS reconstructed image and an edge map (= SS edge map) representing an area obtained by binarizing the edge data.
In S404, the arithmetic unit 5 generates a composite map (= SS composite map) indicating a boundary for compositing the NSS reconstructed image with the SS reconstructed image from the SS edge map.
In S405, the arithmetic unit 5 generates an edge map (= NSS edge map) representing the edge region of the NSS reconstructed image.
In S406, the arithmetic unit 5 generates an image (= SM (SMoothed) -NSS reconstructed image) obtained by performing edge preserving smoothing processing on the NSS reconstructed image.
In S407, the arithmetic unit 5 modifies the SS composite map using the NSS edge map, and generates an NSS composite map.
In S408, the arithmetic unit 5 generates an image (= composite image) obtained by combining the SS reconstructed image, the NSS reconstructed image, and the SM-NSS reconstructed image based on the NSS composite map.
In S409, the arithmetic unit 5 uses the values of the composite image for the pixels around the boundary between the region using the SS reconstructed image and the region using the SM-NSS reconstructed image in the composite image. The value and the value of the SM-NSS reconstructed image are converted into a value obtained by weighted addition.
In S410, the arithmetic unit 5 generates an image weighted and added by the weight value designated by the operator on the entire surface of the SS reconstructed image and the synthesized image.

  The features and effects of image processing in this embodiment include at least the following five.

  First, in S404, from the edge map of the SS reconstructed image, an SS composite map indicating a boundary for compositing the NSS reconstructed image with the SS reconstructed image is generated, and in S407, SS synthesis is performed using the NSS edge map. Modifying the map to generate an NSS composite map. This is equivalent to extracting the boundary with less influence of streak artifact by first using the SS reconstruction image and extracting the boundary with less blur using the NSS reconstruction image inside the boundary. . For this reason, the influence of streak-like artifacts is small, and a boundary with little influence of blur can be captured.

  Second, in S404 and S407, not only the edge strength but also the reference of the CT value and whether or not it is adjacent to the air region are determined. As a result, only the high-contrast structure such as a blurred bone can be restored, and the noise reduction effect by the successive approximation projection data filter can be maintained.

  The third is to apply a direction-dependent filter to the contour of the structure in the NSS reconstructed image using the direction of the SS reconstructed image in S408. Thereby, there is little blur of the outline of the structure of the NSS reconstructed image, and the unevenness in the outline is reduced.

  Fourth, in S409, the values around the boundary between the region using the SS reconstructed image and the region using the SM-NSS reconstructed image in the composite image are changed to the values of the SS reconstructed image and the SM-NSS reconstructed image. This is to convert the value of the component image into a value obtained by weighted addition. This is equivalent to performing weighted addition of the SS reconstructed image and the SM-NSS reconstructed image around the boundary position of the structure. For this reason, the separation and unevenness of the outline of the structure of the composite image are reduced.

  The fifth is to generate an image weighted and added by the weight value designated by the operator over the entire surface of the SS reconstructed image and the synthesized image in S410. Accordingly, the operator can adjust whether to generate an image close to the composite image or to generate an image close to the SS reconstructed image. That is, the degree of contrast recovery can be adjusted.

  Below, the detail of S401-S410 is demonstrated, referring FIGS.

<S401>
In the reconstruction range extraction process of S401, the arithmetic unit 5 generates a reconstruction area map indicating the FOV range in which the reconstruction image is output in the two-dimensional image plane. Usually it is a circular range.

<S402>
In the air region determination process of S402, the arithmetic device 5 inputs the SS reconstructed image, and calculates an air map including pixels that are air regions and pixels that have intermediate CT values that are boundaries between the air region and the structure. Generate. As shown in FIG. 5, the flow of the air region determination process is divided into four steps.

In S501, the arithmetic unit 5 initializes all pixels of the air map to “0”.
In S502, the arithmetic unit 5 scans all the pixels of the air map one by one, and “1” is set to the pixel of the air map at the same position as the pixel whose CT value in the SS reconstructed image is equal to or smaller than the predetermined parameter D_AIR_CTVAL_1. Set.
In S <b> 503, as a result of S <b> 502, the arithmetic unit 5 still positions adjacent to the pixel having the air map value “1” and the same position in the SS reconstructed image with respect to the pixel having the air map value “0”. It is determined whether or not the CT value of this pixel satisfies the condition of a predetermined parameter D_AIR_CTVAL_2 or less, and “2” is set to the air map pixel that satisfies the condition.
In S <b> 504, as a result of S <b> 503, the arithmetic unit 5 still positions adjacent to the pixel having the air map value “2” and the same position in the SS reconstructed image with respect to the pixel having the air map value “0”. It is determined whether or not the CT value of this pixel satisfies the condition of the predetermined parameter D_AIR_CTVAL_2 or less, and “3” is set to the air map pixel that satisfies the condition.

  With the processing shown in FIG. 5, it is possible to reliably extract pixels that become air regions and pixels adjacent to the air regions, and generate an air map. Then, by using the air map generated in this way, it is possible to distinguish the edge of the body surface or lung blood vessel in contact with the air region from the edge of the desired structure.

<S403>
In the SS edge map generation process of S403, the arithmetic unit 5 receives the SS reconstructed image, the reconstructed area map, and the air map, and generates an SS edge map indicating a region having a high edge. As shown in FIG. 6, the SS edge map generation process flow is divided into two steps.

  In the edge extraction process of S601, the arithmetic unit 5 calculates the value of the Sobel filter and generates edge data using these as edge strengths. The value “G” of the Sobel filter is obtained by using the value “Gh” obtained by multiplying each pixel by the horizontal coefficient matrix shown in FIG. 7 and the value “Gv” obtained by multiplying the vertical coefficient matrix shown in FIG. It becomes the following formula. Note that the process of S601 may use another method for calculating the edge strength.

  In the edge region binarization processing in S602, the arithmetic unit 5 generates an SS edge map in which pixels with high edge strength in the edge data generated in S601 are binarized. This edge region binarization process is divided into three steps as shown in FIG.

In step S <b> 801, the arithmetic device 5 creates a histogram of edge data values at the positions of pixels (= non-air pixels) whose reconstruction area map value is “1” and air map value is “0”.
In S802, the arithmetic unit 5 integrates the frequencies from the higher edge strength in the histogram created in S801, and calculates a threshold thHist that is the total frequency × D_EDGE_RATE. D_EDGE_RATE is a predetermined parameter.
In S803, the arithmetic unit 5 sets “1” in the SS edge map pixel that is equal to or greater than the threshold thHist calculated in S802 in the edge data generated in S601.
For example, if D_EDGE_RATE is “0.4”, “1” indicating an edge region is set in the SS edge map in pixels that fall in 40% of the entire edge strength value.

<S404>
In the SS composite map generation process using the SS reconstructed image in S404, the arithmetic unit 5 inputs the SS reconstructed image, the SS edge data, the SS edge map, and the air map, and selects the SS reconstructed image and the NSS reconstructed image at which position. An SS composite map is generated that indicates whether or not to be combined. In the process of S404, the arithmetic unit 5 performs the condition determination shown in FIG. 9 for each pixel.

  First, the arithmetic unit 5 determines whether the CT value of the SS reconstructed image is equal to or greater than a predetermined parameter CTV_TH2 (S901). If it is equal to or greater than CTV_TH2 (Yes in S901), “1” is set in the SS composite map. (S902). The pixel thus determined is an image having a high CT value. For such pixels, a composite image is basically generated using the CT value of the NSS reconstructed image. For example, when the structure 40 in the example of FIG. 3 is like a bone having a high CT value, the calculation device 5 basically has the NSS reconstructed image of the hatched area of the composite image 70 shown in FIG. A composite image is generated using the CT value. However, redetermination is performed by the process of S407 described later, and the CT value of the NSS reconstructed image may not be used.

  When the CT value of the SS reconstructed image is smaller than CTV_TH2 (No in S901), the arithmetic unit 5 determines whether the SS edge map value is “1” (S903), and the SS edge map value is “1”. If it is not “1” (No in S903), “5” is set in the SS composite map (S904). The pixel determined in this way is a pixel that is not an edge.

  When the value of the SS edge map is “1” (Yes in S903), the arithmetic unit 5 determines whether the CT value of the SS reconstructed image is equal to or greater than a predetermined parameter CTV_TH (S905). If (Yes in S905), “1” is set in the SS composite map (S906). The pixel thus determined is a pixel corresponding to a desired structure or an edge of the desired structure.

  When the CT value of the SS reconstructed image is smaller than CTV_TH (No in S905), the arithmetic unit 5 determines whether the air map value is larger than “0” (S907).

  When the value of the air map is “0”, that is, when it is not the air region (No in S907), the arithmetic unit 5 determines whether the value of the SS edge data is equal to or greater than a predetermined parameter EDV_TH (S908), and EDV_TH If it is above (Yes in S907), “2” is set in the SS composite map (S909). The pixel thus determined is a pixel having a high CT value and a high edge strength. For such pixels, a composite image is basically generated using the CT value of the NSS reconstructed image. However, redetermination is performed by the process of S407 described later, and the CT value of the NSS reconstructed image may not be used.

  When the value of the SS edge data is smaller than EDV_TH (No in S908), the arithmetic unit 5 sets “3” in the SS composite map (S910).

  On the other hand, when the value of the air map is “1” or more, that is, in the case of the pixel determined as the air region (Yes in S907), the arithmetic unit 5 determines whether the value of the SS edge data is equal to or greater than EDV_TH ( S911), if it is equal to or greater than EDV_TH (Yes in S911), “4” is set in the SS composite map (S912). The pixels determined in this way are basically pixels corresponding to the body surface and the edges of blood vessels in the lung field. And about such a pixel, in order to preserve | save an edge, a synthesized image will be produced | generated using CT value of a NSS reconstruction image.

  When the value of the SS edge data is smaller than EDV_TH (No in S911), the arithmetic unit 5 sets “2” in the SS composite map (S913). The pixel determined in this way is a pixel having a low edge strength at the boundary of the air region. And about such a pixel, a redetermination is performed in S407 mentioned later.

<S405>
In the NSS edge map generation process of S405, the arithmetic unit 5 performs the same process as the SS edge map generation process of S403, and generates NSS edge data and an NSS edge map.

<S406>
In the edge preserving smoothing process of the NSS reconstructed image in S406, the arithmetic unit 5 applies a direction-dependent filter whose direction is determined based on the SS reconstructed image in a specific region by the condition determination process shown in FIG. Generated image (= SM-NSS reconstructed image) is generated. First, the condition determination process shown in FIG. 10 will be described. Next, the direction-dependent filter process shown in FIG. 11 will be described.

  In the condition determination process shown in FIG. 10, the arithmetic device 5 determines whether or not it is at a position in contact with the air region (S1001). When the position is in contact with the air region (Yes in S1001), the arithmetic unit 5 performs no processing (S1002). This is because the pixels that are in contact with the air region are pixels that are in contact with the body surface and blood vessels in the lung field, and are not smoothed.

  When the position is not in contact with the air region (No in S1001), the arithmetic device 5 has the CT value of the NSS reconstructed image larger than the predetermined parameter D_SM_CTVAL, or the edge strength of the NSS reconstructed image is larger than the predetermined parameter D_SM_EVAL. Is determined (S1003).

  When the CT value of the NSS reconstructed image is greater than D_SM_CTVAL, or when the edge strength of the NSS reconstructed image is greater than the predetermined parameter D_SM_EVAL (Yes in S1003), the arithmetic unit 5 determines that the CT value of the SS reconstructed image is D_SM_CTVAL. And the edge strength of the NSS reconstructed image is smaller than D_SM_EVAL (S1004).

  When the CT value of the SS reconstructed image is larger than D_SM_CTVAL and the edge strength of the NSS reconstructed image is smaller than D_SM_EVAL (Yes in S1004), the arithmetic unit 5 performs no processing (S1005). Since the pixel determined in this way is a pixel in the inside of a bone or the contour of an organ, smoothing is not performed.

  When the value of the SS reconstructed image is equal to or less than D_SM_CTVAL, or the edge strength of the NSS reconstructed image is equal to or greater than D_SM_EVAL (No in S1004), the arithmetic unit 5 uses the direction-dependent filter whose direction is determined based on the SS reconstructed image Processing is performed (S1006). Since the pixels determined in this way are the pixels of the bone contour, etc., if the CT value of the NSS reconstructed image is used as it is, there is a possibility that the contour of the bone or the like may be uneven due to the streak artifact. There is. Therefore, a direction dependent filter process is performed.

  On the other hand, when the CT value of the NSS reconstructed image is D_SM_CTVAL or less and the edge strength of the NSS reconstructed image is D_SM_EVAL or less (No in S1003), the arithmetic unit 5 determines whether the CT value of the NSS reconstructed image is smaller than D_SM_CTVAL_MTH. Determination is made (S1007).

  When the CT value of the NSS reconstructed image is smaller than D_SM_CTVAL_MTH (Yes in S1007), the arithmetic unit 5 performs no processing (S1008). Since the pixels determined in this way are mainly pixels in the air region, smoothing is not performed.

  When the CT value of the NSS reconstructed image is greater than or equal to D_SM_CTVAL_MTH (No in S1007), the arithmetic unit 5 performs a direction-dependent filter process in which the direction is determined based on the SS reconstructed image (S1009). Since the pixel determined in this way becomes a weighted addition region to be described later, it is necessary to reduce unevenness due to noise. Therefore, a direction dependent filter process is performed.

  In the direction-dependent filter processing illustrated in FIG. 11, the arithmetic device 5 performs smoothing of the NSS reconstructed image using the edge direction of the SS reconstructed image in which the streak artifact is reduced. As shown in FIG. 11, the flow of the direction-dependent filtering process includes two steps.

  In the smoothing direction determination process using the SS reconstructed image in S1101, the arithmetic unit 5 uses, for example, four adjacent pixels in the four directions of the left-right direction, the up-down direction, the right-up direction, and the left-up direction. The absolute value of the difference from the pixel is calculated. In S1102, the arithmetic unit 5 performs smoothing using the pixel in the direction of the smallest value among the four values. A specific calculation formula is as follows. First, the arithmetic unit 5 calculates the absolute value of the difference between the pixel of interest and the adjacent pixel in each direction using Equations (3) to (6).

However, the negative direction of the X axis is the left direction, the positive direction of the X axis is the right direction, the negative direction of the Y axis is the upward direction, the positive direction of the Y axis is the downward direction, and P _if (x, y): SS image CT value, F1 _if : absolute value of the difference with the adjacent pixel in the horizontal direction, F2 _if : absolute value of the difference with the adjacent pixel in the vertical direction, F3 _if : absolute value of the difference with the adjacent pixel in the upward direction, F4 _if : absolute value of the difference from the adjacent pixel in the upward left direction.

Next, Min (F1 _if , F2 _if , F3 _if ,
If F4 _if ) = F1 _if , the arithmetic unit 5 performs an NSS image smoothing process using a direction-dependent filter using adjacent pixels in the left-right direction shown in Expression (7).

However, P _f (x, y): CT value of NSS image, P _sm (x, y): CT value after smoothing (= CT value of SM-NSS reconstructed image).

Also, Min (F1 _if , F2 _if , F3 _if ,
If F4 _if ) = F2 _if , the arithmetic unit 5 performs an NSS image smoothing process using a direction-dependent filter using adjacent pixels in the vertical direction shown in Expression (8).

Also, Min (F1 _if , F2 _if , F3 _if ,
If F4 _if ) = F3 _if , the arithmetic unit 5 performs the smoothing process of the NSS image by the direction-dependent filter using the adjacent pixels in the upward direction shown in Expression (9).

Also, Min (F1 _if , F2 _if , F3 _if ,
If F4 _if ) = F4 _if , the arithmetic unit 5 performs the smoothing process of the NSS image by the direction-dependent filter using the adjacent pixels in the upward left direction shown in Expression (10).

  In the above formulas (7) to (10), a simple average filter is used, but a weighted average filter may be used. As described above, by smoothing the NSS reconstructed image using the direction-dependent filter using the direction determined based on the information of the SS reconstructed image, the outline of the structure of the NSS reconstructed image is less blurred. The SM-NSS reconstructed image in which the unevenness (for example, streak artifact etc.) in the contour is reduced can be generated.

  However, the smoothing process in the present embodiment is not limited to this, and the NSS reconstructed image may be smoothed using an existing smoothing filter.

<S407>
In the SS composite map correction process using the NSS edge map in S407, the arithmetic unit 5 corrects the SS composite map generated in S404 based on the condition determination in FIG. 12, and generates an NSS composite map. In the processing of S407, the arithmetic device 5 performs the condition determination shown in FIG. 12 for each pixel.

  First, the arithmetic unit 5 determines whether the SS composite map is “4” (S1201). When the SS composite map is “4” (Yes in S1201), the arithmetic unit 5 sets “4” as it is in the NSS composite map.

  When the SS composite map is not “4” (No in S1201), the arithmetic device 5 determines whether the SS composite map is “2” or less (S1203). When the SS composite map is larger than “2” (No in S1203), the arithmetic unit 5 sets “5” in the NSS composite map.

  When the SS composite map is “2” or less (No in S1203), the arithmetic unit 5 determines whether the CT value of the SS reconstructed image is equal to or greater than CTV_TH2_NSS (S1205), and if it is equal to or greater than CTV_TH2_NSS (Yes in S1205). ), “1” is set in the NSS composite map (S1206). The pixel determined in this way is a pixel having a high CT value.

  When the CT value of the SS reconstructed image is smaller than CTV_TH2_NSS (No in S1205), the arithmetic unit 5 determines whether the value of the NSS edge map is “1” (S1207), and when it is not “1” (S1207). No), “5” is set in the NSS composite map (S1208). The pixel determined in this way is a pixel that is not an edge.

  When the value of the NSS edge map is “1” (Yes in S1207), the arithmetic unit 5 determines whether the CT value of the SM-NSS reconstructed image is greater than or equal to CTV_TH_NSS (S1209). (Yes in S1209), “1” is set in the NSS composite map (S1210).

  When the CT value of the SM-NSS reconstructed image is smaller than CTV_TH_NSS (No in S1209), the arithmetic unit 5 determines whether the value of NSS edge data is equal to or greater than EDV_TH_NSS (S1211), and smaller than EDV_TH_NSS (S1211). No), “3” is set in the NSS composite map (S1212).

  When the value of the NSS edge data is equal to or greater than EDV_TH_NSS (Yes in S1211), the arithmetic unit 5 determines whether the value of the air map is greater than “0” (S1213). When the value of the air map is “0” or less, that is, when the air map does not contact the air region (No in S1213), the arithmetic unit 5 sets “2” in the NSS composite map (S1214).

  When the value of the air map is “1” or more (Yes in S1213), the arithmetic unit 5 sets “4” in the NSS composite map (S1215). The pixels thus determined are pixels corresponding to the body surface or the edge of a blood vessel in the lung field.

<S408>
In the combining process of S408, the arithmetic device 5 generates a combined image using an appropriate image from among the SS reconstructed image, the NSS reconstructed image, and the SM-NSS reconstructed image according to the NSS composite map.

The combinations of the determination results in FIG. 12 and the images used for the synthesis process are summarized as follows.
(1) The value of the NSS composite map is “1” (pixel having a high CT value = pixel corresponding to a desired structure)... SM-NSS reconstructed image (2) The value of the NSS composite map is “2” ( Pixel corresponding to the edge of the desired structure) ... SM-NSS reconstructed image (3) The value of the NSS composite map is "3" (pixel that is neither the desired structure nor the edge) ... SS re Composition image (4) The value of the NSS composite map is “4” (pixel corresponding to the edge of the body surface or blood vessel in the lung field)... NSS reconstructed image (5) The value of the NSS composite map is “5” (desired Pixel that is neither a structure nor an edge) ... SS reconstruction image

<S409>
In the boundary area weighted addition process of S409, the arithmetic unit 5 provides an area (weighted addition area) calculated by weighted addition of the values of the SS reconstructed image and the NSS reconstructed image outside the structure. Reduce boundary separation and blurring outside the structure. The arithmetic unit 5 starts from the boundary of the structure, and extracts a boundary (= CT value difference boundary) in which the CT value difference between the NSS reconstructed image and the SS reconstructed image is smaller than the threshold value. Then, the arithmetic unit 5 uses the CT value from the boundary of the structure to the CT value difference boundary in the composite image as the CT value of the NSS reconstructed image using the distance from the boundary of the structure and the distance from the CT value difference boundary. A weighted addition value of the value and the CT value of the SS reconstructed image. The process of S409 is divided into four steps as shown in FIG.

  In S1301, the arithmetic unit 5 calculates the absolute value of the difference between the SS reconstructed image and the NSS reconstructed image, sets “1” to pixels whose value is DIFMAP_TH or more, and sets “0” to other pixels. By setting, a large difference map is generated.

  In S1302, the arithmetic device 5 generates a distance image representing the distance from the boundary with the SM-NSS reconstructed image region. FIG. 14 shows an example of a composite image, and FIG. 15 shows an example of a large difference map. FIG. 16 shows the process of generating a distance image from the boundary with the SM-NSS reconstructed image area in S1302 when these are input. Show.

  In the process 1, the arithmetic unit 5 sets “−1” to the pixel that is the outline of the SM-NSS reconstructed image use area, “−2” to the pixel inside the outline of the SM-NSS reconstructed image use area, NSS “−10” is set to the pixel in the reconstructed image utilization area.

  In the process 2, the arithmetic unit 5 sets “1” in the pixel adjacent to the SM-NSS reconstructed image area in the SS reconstructed image area. This is the boundary between the weighted addition area and the SM-NSS reconstructed image area, and “1” is the distance from the boundary. In the example of FIG. 16, “adjacent pixels” are eight pixels of an upper, lower, right, left, upper right, lower right, upper left, and lower left pixels of interest.

  In process 3, the arithmetic unit 5 generates a distance image that represents a pixel within a predetermined distance D_MAX_DIST within a range of a flag set on the large difference map from a pixel for which “1” is set. In the example of FIG. 16, D_MAX_DIST is set to “6”.

  There are various methods for generating a distance image. For example, the following method is used. First, the arithmetic unit 5 sets “2” to a pixel adjacent to the pixel set to “1” among the pixels set to “0” at the present time. Next, the arithmetic unit 5 sets “3” to the pixel adjacent to the pixel set to “2” among the pixels set to “0” at the present time. The arithmetic unit 5 repeats such processing until the set value becomes D_MAX_DIST. The image of the process 3 obtained in this way is a distance image from the boundary with the SM-NSS reconstructed image area.

  In S1303, the arithmetic unit 5 generates a distance image representing the distance from the boundary with the SS reconstructed image area in the weighted addition area. This process is shown in FIGS.

  In the process 1, the arithmetic unit 5 sets “1” to the boundary between the pixel to which the distance value is assigned and the pixel to which the distance value is not assigned in the distance image from the boundary with the SM-NSS reconstructed image region. That is, the pixel group in which “1” is set becomes a boundary between the weighted addition region and the SS reconstructed image region.

  In the process 2, the arithmetic unit 5 generates a distance image representing the distance from the boundary with the SS reconstructed image area to the boundary of the SM-NSS reconstructed image area. The arithmetic unit 5 basically performs a distance image generation process from the boundary with the SM-NSS reconstructed image area for pixels of “1” or more in the distance image from the boundary with the SM-NSS reconstructed image area. Perform the same process as Thereby, a distance image from the boundary with the SS reconstructed image area is generated.

  However, in the distance image generation process from the boundary with the SS reconstructed image area in the process 2, the distance may not be calculated by the above-described calculation process. For example, when the distance image from the boundary with the composite image, the large difference map, and the SM-NSS reconstructed image area is the example shown in FIG. 18, the distance image from the boundary with the SM-NSS reconstructed image area There is no boundary between a pixel to which a value is assigned and a pixel to which no value is assigned, that is, a boundary between a weighted addition area and an SS reconstructed image area. In such a case, the above calculation process cannot calculate the distance from the boundary with the SS reconstructed image area.

  On the other hand, in the process 3 shown in FIG. 19, the arithmetic unit 5 has a pixel value equal to or larger than the values of all the adjacent eight neighboring pixels in the distance image from the boundary with the SM-NSS reconstructed image region. Pixels (shaded pixels in FIG. 19) are extracted, and “1” is set to the corresponding pixel in the distance image from the boundary with the SS reconstructed image.

  In the process 4, the arithmetic unit 5 performs the same processing as the generation of the distance image from the boundary with the SM-NSS reconstructed image region on the pixels of “1” or more in the distance image from the SM-NSS reconstructed image boundary. Thus, a distance image from the SS reconstructed image area is generated. The image of the process 4 obtained in this way becomes a distance image from the boundary with the SS reconstructed image area.

  In S1304, the arithmetic unit 5 uses the value of the distance image from the boundary with the SM-NSS reconstructed image area and the value of the distance image from the boundary with the SS reconstructed image area as a weighting coefficient, and performs SM-NSS reconstruction. A weighted sum of the value of the constituent image and the value of the SS reconstructed image is calculated. Examples of calculation formulas are shown below.

  Where ds (x, y): distance from the boundary with the SM-NSS image area, di (x, y): distance from the boundary with the SS image area, Ps (x, y): SM-NSS image CT value, Pi (x, y): CT value of SS image, Pm (x, y): Pixel value obtained by weighted addition.

<S410>
In the entire weighted addition process of S410, the arithmetic unit 5 generates an image weighted and added by the weight value designated by the operator on the entire surface of the SS reconstructed image and the synthesized image. Examples of calculation formulas are shown below.

  Where w: weight specified by the operator, Pa (x, y): pixel value of the composite image, Pb (x, y): pixel value of the SS image, Pc (x, y): pixel value obtained by weighted addition . The weight w is a value from 0.0 to 1.0. An image closer to the composite image is generated as w is increased, and an image closer to the SS reconstructed image is generated as the value is decreased. As a result, the degree of contrast recovery can be adjusted.

  In this embodiment, an image to which the successive approximation projection data filter is applied and an image to which the successive approximation projection data filter is applied are used. However, an image generated by successive approximation image reconstruction as proposed in Non-Patent Document 2 is used for SS reconstruction. It is possible to apply the same as the NSS reconstructed image as the constituent image and the image before the successive approximation reconstruction or the image generated by the conventional reconstruction method.

  The preferred embodiments of the X-ray CT apparatus and the like according to the present invention have been described above with reference to the accompanying drawings, but the present invention is not limited to such examples. It will be apparent to those skilled in the art that various changes or modifications can be conceived within the scope of the technical idea disclosed in the present application, and these naturally belong to the technical scope of the present invention. Understood.

DESCRIPTION OF SYMBOLS 1 ......... X-ray CT apparatus 2 ......... Scanner 3 ......... Operation unit 4 ......... Bed 5 ......... Calculation device 6 ......... Input device 7 ......... Display device 8 ......... Storage device 10 ... …… Subject 11 ……… X-ray tube 40 ………… Structure 50 ……… SS reconstruction image 51 ……… SS edge 60 ……… NSS reconstruction image 61 ……… NSS edge 62 ……… Streaks Artifact 70 ……… Composite image

Claims (9)

An X-ray source that irradiates X-rays, an X-ray detector that detects the X-rays, a bed on which the subject is placed, the X-ray source and the X-ray detector, and the surroundings of the subject An X-ray CT apparatus comprising: a rotating disk that rotates a rotating disk; and a storage device that stores actual projection data obtained by converting the X-ray detected by the X-ray detector,
A first image is reconstructed using data obtained by subjecting the actual projection data to a first noise reduction process, and the noise reduction degree of the actual projection data is lower than that of the first noise reduction process. 2 an image reconstruction unit that reconstructs the second image using the data subjected to noise reduction processing or the actual projection data;
For a predetermined structure included in both the first image and the second image, a first boundary that is a boundary of the structure is extracted from the first image, and further, the structure of the structure is extracted from the second image. A boundary extraction unit for extracting a second boundary which is a boundary;
An image composition unit that performs composition processing using at least the first image and the second image for a region between the first boundary and the second boundary, and generates a composite image;
An X-ray CT apparatus comprising: An image smoothing unit that determines a direction of a direction-dependent filter based on the first image, applies the direction-dependent filter to the second image, performs a smoothing process, and generates a third image;
Further comprising
The X-ray CT apparatus according to claim 1, wherein the image synthesis unit further performs a synthesis process using the third image. The image smoothing unit calculates an absolute value of a difference between the target pixel and the adjacent pixel with respect to adjacent pixels in a plurality of directions adjacent to the target pixel of the first image, and the calculated value is the largest. The X-ray CT apparatus according to claim 2, wherein a direction of the small adjacent pixel is determined as a direction of the direction-dependent filter. The image reconstruction unit, as the first noise reduction processing and the second noise reduction processing, by a successive approximation method including a photon weight corresponding to a detector output of the X-ray detector and a penalty term in an evaluation function. The X-ray CT apparatus according to claim 1, wherein the actual projection data is corrected. A first composite map generation unit that generates a first composite map indicating how to combine the first image and the second image based on the first image;
A second composition map that corrects the first composition map based on at least the second image and generates a second composition map that indicates how to compose the first image, the second image, and the third image; A composite map generator;
Further comprising
The X-ray CT apparatus according to claim 2, wherein the image composition unit performs composition processing according to the second composition map. For the pixels around the boundary between the region using the first image and the region using the third image indicated by the second composite map, the value of the composite image, the value of the first image, and the value of the third image A boundary area weighted adder for converting the value into a weighted value;
The X-ray CT apparatus according to claim 5, further comprising: An image weighted addition unit that generates an image weighted and added according to a weight value designated by an operator over the entire surface of the first image and the composite image;
The X-ray CT apparatus according to claim 1, further comprising: An air map generating unit that generates an air map indicating an air region in the first image and a region in contact with the air region;
Further comprising
The first composite map generation unit further generates the first composite map based on the air map,
The X-ray CT apparatus according to claim 5, wherein the second composite map generation unit further corrects the first composite map based on the air map to generate the second composite map. An X-ray source that irradiates X-rays, an X-ray detector that detects the X-rays, a bed on which the subject is placed, the X-ray source and the X-ray detector, and the surroundings of the subject An image processing method executed by an X-ray CT apparatus, comprising: a rotating disk that rotates a rotating disk; and a storage device that stores actual projection data obtained by converting the X-rays detected by the X-ray detector,
A first image is reconstructed using data obtained by subjecting the actual projection data to a first noise reduction process, and the noise reduction degree of the actual projection data is lower than that of the first noise reduction process. Reconstructing the second image using the data subjected to two noise reduction processing or the actual projection data;
For a predetermined structure included in both the first image and the second image, a first boundary that is a boundary of the structure is extracted from the first image, and further, the structure of the structure is extracted from the second image. Extracting a second boundary which is a boundary;
Performing a combining process on the region between the first boundary and the second boundary using at least the first image and the second image to generate a combined image;
An image processing method including:

Images