JP2005095329A - Superresolving apparatus and medical diagnostic imaging apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To superresolve image data obtained by a medical diagnostic imaging apparatus. <P>SOLUTION: This superresolving apparatus is provided with a storage part 31 storing data of a point image intensity spread function related to an X-ray computed tomograph which are obtained by using a phantom; and a superresolving part 32 superresolving the image data related to a subject using the stored point image intensity spread function. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a super-resolution processing apparatus and a medical image diagnostic apparatus that perform super-resolution processing on volume data or multi-slice data collected by an X-ray computed tomography apparatus.

  In recent years, multi-slice X-ray computed tomography apparatus has been developed in response to the strong demand from the medical field to obtain a high-definition (high-resolution) and wide-range image taking place. It has become widespread. The multi-slice X-ray computed tomography apparatus includes an X-ray source that emits a fan beam X-ray having a spread in the slice direction (longitudinal direction of the bed), and a plurality of rows (4 rows, 8 rows, 16 rows, etc.). And a two-dimensional detector having a structure in which the detector element arrays are arranged in the slice direction, and this is operated by multi-scan or helical scan. Thereby, compared with a single slice X-ray CT apparatus, volume data over a wide range of the subject can be obtained with high accuracy and in a short time.

  The volume data obtained in this way is not only displayed and observed, but has various uses in recent years. For example, in the case of medical use, there is a measurement of the stenosis rate of a blood vessel and wrinkles. Specifically, volume data obtained by imaging the distribution state of the contrast agent flowing through the blood vessel can be obtained by administering the X-ray contrast agent to the subject and performing imaging using the X-ray CT apparatus. For this reason, the stenosis rate of the blood vessels and the size of the wrinkles are measured from the distribution of the CT value of the contrast agent reflected in the volume data. For example, in the case of measuring the stenosis rate, the thickness of the blood vessel inner wall (the range occupied by the contrast medium) is measured from the volume data, and the thickness of the blood vessel in the portion considered to be normal and the thickness of the blood vessel in the narrowed portion are measured. It is calculated by comparing. In the measurement of the thickness of the blood vessel, a threshold value for the CT value is usually set.

On the other hand, Patent Documents 1 and 2 show other processing examples of volume data obtained from various medical image apparatuses including not only an X-ray computed tomography apparatus but also an ultrasonic diagnostic apparatus and a magnetic resonance imaging apparatus. Yes. An object of Patent Document 1 is to perform reliable blood vessel measurement based on a display image, and sets a region of interest that crosses a blood vessel wall vertically on a tomographic image of a blood vessel, and a profile of pixel values in this region The dimension value related to the blood vessel is measured based on the above. On the other hand, the object described in Patent Document 2 is intended to accurately measure the length of an object of interest (blood vessel, intestine, etc.) having a curvature in a direction not parallel to the projection plane using an MIP image.
JP-A-11-342132 Here, not only the X-ray computed tomography apparatus but all image apparatuses have a resolution limit depending on the detection element pitch of the detector. In particular, in the X-ray computed tomography apparatus, as shown in FIG. 11, a CT image is generated as a back projection sum for a number of views, so blurring depending on the resolution limit is data of one view of a minute point. Appears on the image as a result of multiple synthesis of profiles in multiple directions. In other words, the CT image can be said to be an aggregate of intensity distribution functions of minute point images representing blur, that is, PSF (point spread function). However, no consideration has been given to this blur.

  An object of the present invention is to put a super-resolution process to image data obtained by a medical image diagnostic apparatus into practical use.

  A super-resolution processing apparatus according to an aspect of the present invention includes a unit that stores data of a point image intensity distribution function according to an X-ray computed tomography apparatus acquired using a phantom, and the image data related to a subject. Means for performing super-resolution processing using the stored point image intensity distribution function.

  According to the present invention, super-resolution processing for image data obtained by a medical image diagnostic apparatus can be put into practical use.

  Embodiments of a medical image diagnostic apparatus equipped with a super-resolution processing apparatus according to the present invention will be described below with reference to the drawings. Here, an X-ray computed tomography apparatus will be described as an example of a medical image diagnostic apparatus. However, other modalities such as a magnetic resonance imaging apparatus (MRI apparatus), an ultrasonic diagnostic apparatus, PET, SPECT, a gamma camera, and an X-ray diagnostic apparatus are used. You may apply to.

  The X-ray computed tomography apparatus includes a rotation / rotation type in which an X-ray tube and a radiation detector are rotated as one body, and a large number of detection elements are arrayed in a ring shape. There are various types such as a fixed / rotation type in which only the subject rotates around the subject, and the present invention can be applied to any type. Here, the rotation / rotation type that currently occupies the mainstream will be described. In addition, to reconstruct one slice of tomographic image data, projection data for about 360 ° around the object and projection data for about 360 ° is obtained, and projection data for 180 ° + α (α: fan angle) is also obtained by the half scan method. Needed. The present invention can be applied to any reconstruction method. Here, the half scan method will be described as an example. In addition, the mechanism for converting incident X-rays into electric charges is based on an indirect conversion type in which X-rays are converted into light by a phosphor such as a scintillator and the light is further converted into electric charges by a photoelectric conversion element such as a photodiode. The generation of electron-hole pairs in semiconductors and their transfer to the electrode, that is, the direct conversion type utilizing a photoconductive phenomenon, is the mainstream. Any of these methods may be employed as the X-ray detection element, but here, the former indirect conversion type will be described. In recent years, the so-called multi-tube X-ray computed tomography apparatus in which a plurality of pairs of an X-ray tube and an X-ray detector are mounted on a rotating ring has been commercialized, and the development of peripheral technologies has progressed. Yes. The present invention can be applied to both a conventional single-tube X-ray computed tomography apparatus and a multi-tube X-ray computed tomography apparatus. Here, a single tube type will be described.

  As shown in FIG. 1, the X-ray computed tomography apparatus has a gantry 1 configured to collect projection data relating to a subject. The gantry 1 includes an X-ray tube 10 and an X-ray detector 23. The X-ray tube 10 and the X-ray detector 23 are mounted on a ring-shaped rotating frame 12 that is rotationally driven by a gantry driving device 25. The central portion of the rotating frame 12 is opened, and the subject P placed on the top 2a of the bed 2 is inserted into the opening. The rotation center axis of the rotary frame 12 is defined as a Z axis (slice direction axis), and a plane orthogonal to the Z axis is defined as two orthogonal axes of XY.

  A tube voltage is applied between the cathode and anode of the X-ray tube 10 from the high voltage generator 21, and a filament current is supplied to the filament of the X-ray tube 10 from the high voltage generator 21. X-rays are generated by applying a tube voltage and supplying a filament current. As the X-ray detector 23, either a one-dimensional array detector or a two-dimensional array detector (also referred to as a multi-slice detector) may be employed. The X-ray detection element has a square light receiving surface of 0.5 mm × 0.5 mm, for example. For example, 916 X-ray detection elements are arranged in the channel direction. A two-dimensional array type detector has 40 rows arranged in the slice direction, for example. What consists of a single column is a one-dimensional array type detector.

  A data acquisition device 26 generally called a DAS (data acquisition system) converts a signal output from the detector 23 for each channel into a voltage signal, amplifies it, and converts it into a digital signal. This data (raw data) is supplied to the computer unit 3 outside the gantry. The pre-processing unit 34 of the computer unit 3 performs correction processing such as sensitivity correction on the data (raw data) output from the data collection device 26 and outputs projection data. This projection data is sent to and stored in the data storage device 37 of the computer system 3.

  The computer system 3 includes a system controller 29, a scan controller 30, a reconstruction processing unit 36, a display 38, an input device 39, a PSF storage unit 31, a super-resolution processing unit 32, an image, together with the preprocessing unit 34 and the storage device 37. The processing unit 33 is configured. The reconstruction processing unit 36 reconstructs image data based on projection data collected by, for example, helical scanning, volume scanning using cone beam X-rays, or a combination thereof.

  The PSF storage unit 31 stores in advance data of a point spread function (PSF: point spread function) regarding each XYZ direction unique to the X-ray computed tomography apparatus. The “intensity” of the point spread function is a CT value here. The PSF data is a wire phantom having a diameter (0.05 mm) smaller than 0.5 mm, for example, which is the detector pitch (resolution limit) of the detector 23, or a microsphere if it is three-dimensional. 2D or 3D image (blurred image) relating to the wire phantom or microsphere obtained by reconstruction from the projection data, here, enlarged reconstruction of the partial area centered on the wire phantom or microsphere. Obtained as data. That is, the PSF is obtained as image data related to a minute object having a size (diameter) smaller than the resolution limit that has been enlarged and reconstructed.

  The PSF data acquired from the wire phantom is measured in advance and stored in the PSF storage unit 31. The PSF data is commonly used for all subjects. Since it is not necessary to acquire PSF data for each subject, a scan for acquiring PSF data for each subject is not necessary, so that exposure to the subject can be reduced and commercialization can be promoted. .

  The super-resolution processing unit 32 performs super-resolution processing on the super-resolution processing target image using the PSF data. Here, the image that has been enlarged and reconstructed by the reconstruction processing unit 36 is the super-resolution processing target. Note that an image extracted by volume conversion (MPR) or the like from the volume data by the image processing unit 33 may be enlarged, that is, the pixel matrix size may be increased to make the image a super-resolution processing target. In the super-resolution processing, the super-resolution processing unit 32, the image processing unit 33, and the reconstruction processing unit 36 are linked under the control of the system controller 29 in accordance with a processing sequence predetermined for the processing. Works. Note that the super-resolution processing according to the present embodiment includes first to third super-resolution processing as will be described in sequence, and a desired one of these processing is performed by the operator via the input device 39. Is arbitrarily selected.

  FIG. 2 shows a first super-resolution processing procedure according to this embodiment. First, under the control of the system controller 29, in step S1, a reference image (three-dimensional image) is generated from the volume data or multi-slice data generated by volume reconstruction or the like related to the three-dimensional region of the subject. And displayed on the display 38 (see FIG. 3A). When an axial plane orthogonal to the Z axis including the super-resolution processing range is designated on the displayed reference image by the operator via the input device 39 (S2), an image related to the designated axial plane is obtained. The volume data is generated by the image processing unit 33 and displayed (S3). Next, in S4, a narrower local range (super-resolution processing range) including the super-resolution target portion is designated by the operator via the input device 39 on the displayed image regarding the axial plane (FIG. 3). (See (b)). By limiting the super-resolution processing range as much as possible, an image in that range can be reconstructed (enlarged reconstructed) with a very high resolution, here, a resolution that can gently represent the PSF curve. Further, as is well known, super-resolution processing includes processing for individually deconvolving PSF for all pixels, so that the number of processing steps increases depending on the number of target pixels (matrix size). By limiting the processing range as much as possible, the processing time can be effectively shortened and its practicality can be improved. Needless to say, when the processing area is limited due to the speedup of the processing apparatus, the super-resolution processing may be performed for the entire range of the initial reconstruction field of view (reconstruction FOV). In this case, super-resolution processing is performed after the image is reconstructed with the number of matrices multiplied by the enlargement ratio (for example, when the display matrix is 512 × 512 and the enlargement ratio is 8 times, a 4096 × 4096 matrix). It will be.

  Next, a reconstruction parameter for enlarging and reconstructing (zooming reconstruction) an image of a circular reconstruction FOV (reconstruction field of view) including a designated super-resolution processing range by the reconstruction processing unit 36 is obtained from the system controller. 29 or the super-resolution processing unit 32 (S5). Basically, the reconstruction parameters are determined based on clinical requirements such as the size of the target site. In addition to the center position and size (diameter or radius) of the reconstruction FOV, the reconstruction parameters include an enlargement ratio, a reconstruction function, a reconstruction matrix, a slice pitch, and the like. The enlargement ratio is determined so that the PSF curve can be expressed smoothly. For example, if the reconstruction field of view is a 512 × 512 matrix, it is set to 3 times or more, preferably 8 times the reconstruction FOV.

  When the reconfiguration parameter is determined, the reconfiguration parameter is automatically supplied to the reconfiguration processing unit 36 under the control of the system controller 29 (S6). Then, the reconfiguration processing unit 36 starts the reconfiguration batch process under the control of the system controller 29 (S7). As a result, an image relating to the reconstruction FOV (reconstruction field of view) including the super-resolution processing range is generated with the highest spatial resolution of the X-ray computed tomography apparatus (see FIG. 3C). By enlarging and reconstructing only the processing range, super-resolution processing can be applied to images with the highest spatial resolution of X-ray computed tomography equipment, and the effectiveness of super-resolution processing can be improved. It can improve and increase its practicality.

  Under the control of the system controller 29, the enlarged and reconstructed image data is supplied from the reconstruction processing unit 36 to the super-resolution processing unit 32 (S8) and used for the super-resolution processing (S9). First, PSF data is supplied from the PSF storage unit 31 to the super-resolution processing unit 32 and stored in the internal storage unit of the super-resolution processing unit 32 together with the enlarged and reconstructed image data. If the resolution of the stored PSF data (image data of the reconstructed wire phantom) is different from the resolution of the image reconstructed in S7, the resolution of the PSF is expanded in S7. In order to match the resolution of the reconstructed image, the PSF is resampled by the super-resolution processing unit 32 or the image processing unit 33. FIG. 3D shows the resampled PSF.

In the super-resolution processing unit 32, an image in the super-resolution processing range is subjected to the super-resolution processing using the PSF in the XY2 direction resampled as necessary (S9). In this super-resolution processing, first, PSF is deconvolved (*) into the image M in the super-resolution processing range.
M '= M (*) PSF
By this deconvolution processing, the resolution of the image in the super-resolution processing range is improved. In this embodiment, in order to improve the accuracy with respect to the true value, an approximate solution is obtained by applying an iterative solution method, typically a jacobi method. In the iterative solution method, an initial solution is required. Here, the image M ′ subjected to the deconvolution process is adopted as the initial solution. However, as an initial solution, a null image may be employed instead of the image M ′ that has undergone the deconvolution process. The error is denoted as E, the convolution process is denoted as *, and the solution is denoted as O.
E = (M-PSF * O) ²
A correction vector dE / dO is obtained from the gradient of E so as to minimize the error amount E. a is a constant.
_{O N + 1 = O N -a} * dE / dO
Thus, an image having super-resolution exceeding the resolution (detection element pitch) of the detector 23 is displayed on the display 38 as shown in FIG. 3E (S10). FIG. 10 shows the result of the simulation. A simulated blood vessel (true value image; FIG. 10 (a)) with a lime portion attached to the inner wall of the blood vessel is defined, and a simulated blurred measurement image (FIG. 10 (b)) is generated by convolving the PSF. did. It can be seen that by performing the super-resolution processing on the measurement image, as shown in FIG. 10C, the resolution is improved and the blood vessel and the lime portion can be separated and recognized.

  Next, the second super-resolution processing procedure will be described with reference to FIG. In the first super-resolution processing, the processing range is defined on the reconstruction plane, that is, the axial plane orthogonal to the Z-axis, whereas in the second super-resolution processing, the processing range is Z-axis. Is provided as a process in the case of being set on a so-called oblique surface that is inclined with respect to. Under the control of the system controller 29, in step S11, a reference image is generated from the volume data by the image processing unit 33 and displayed on the display 38 (see FIG. 5A). When the oblique surface OP inclined to the Z axis and the super-resolution processing range on the surface are designated by the operator via the input device 39 on the displayed reference image (S12), the image processing unit 33 A plurality of axial planes (XY planes) are set at minute intervals so as to intersect the processing range on the designated oblique plane (S13). The minute interval between the axial planes is set to such a degree that the PSF curve can be expressed smoothly in the Z-axis direction. For example, if the slice thickness of the axial image is 0.5 mm, the interval between the axial surfaces is set to 1/4 or less.

  Similar to S5 and S6 of the first super-resolution processing, a reconstruction parameter for enlarging and reconstructing (zooming reconstruction) a plurality of images related to a plurality of axial planes by using the system controller 29 or Determined by the super-resolution processing unit 32, the reconfiguration parameter is supplied to the reconfiguration processing unit 36 under the control of the system controller 29, and the reconfiguration batch processing is started in the reconfiguration processing unit 36 under the control of the system controller 29. (S14). As a result, a plurality of images (axial image sets) relating to a plurality of axial surfaces are generated. Data of a plurality of images is supplied from the reconstruction processing unit 36 to the image processing unit 33 under the control of the system controller 29, and as shown in FIG. 5B, the image to be processed on the oblique surface from the plurality of images. Is generated by the cross-section conversion process (MPR) (S15).

  Next, the three-dimensional PSF data illustrated in FIG. 5C is supplied from the PSF storage unit 31 to the super-resolution processing unit 32, and the internal storage unit of the super-resolution processing unit 32 together with the data of the oblique plane image. Is remembered. As shown in FIG. 5D, the super-resolution processing unit 32 converts the two-dimensional PSF on the oblique plane OP from the stored three-dimensional PSF data (image data of the enlarged and reconstructed microsphere). Generate (S16). Using the generated oblique surface PSF, the image of the oblique surface is subjected to super-resolution processing (S17) and displayed (S18) in the same manner as S9 of the first super-resolution processing.

  Next, a third super-resolution processing procedure will be described with reference to FIG. In the first and second super-resolution processes, the processing range is defined in two dimensions on the axial plane or oblique plane, and the super-resolution process is performed on the two-dimensional image. In the super-resolution processing of 3, the processing range is defined in three dimensions, and the super-resolution processing is performed on the three-dimensional image. Under the control of the system controller 29, in step S21, a reference image is generated from the volume data by the image processing unit 33 and displayed on the display 38 (see FIG. 7A). When the three-dimensional processing range 3D-ROI is designated by the operator via the input device 39 on the displayed reference image (S22), the designated three-dimensional processing range is designated by the image processing unit 33. A plurality of axial planes (XY planes) included in the 3D-ROI are set at minute intervals. The minute interval between the axial planes is set to such a degree that the PSF curve can be expressed smoothly in the Z-axis direction. A reconstruction parameter for enlarging and reconstructing (zooming reconstruction) a plurality of images related to the set plurality of axial planes is determined by the system controller 29 or the super-resolution processing unit 32, The configuration parameters are supplied to the reconfiguration processing unit 36 under the control of the system controller 29, and the reconfiguration batch processing is started in the reconfiguration processing unit 36 under the control of the system controller 29 (S23). As a result, a plurality of images (axial image sets) relating to a plurality of axial surfaces are generated. The data of the plurality of images is supplied to the super-resolution processing unit 32 together with the data of the three-dimensional PSF (FIG. 7B) in the PSF storage unit 31 under the control of the system controller 29. The super-resolution processing unit 32 performs super-resolution processing on a plurality of images related to a plurality of axial planes using three-dimensional PSF data (image data of enlarged and reconstructed microspheres) (S24). ) And display (S25).

  In the above-described super-resolution processing of the oblique surface, in the above description, enlargement reconstruction of a plurality of axial surfaces, generation of an oblique image of the oblique surface by cross-section conversion processing from a plurality of enlarged and reconstructed axial images, and the oblique image Although super-resolution processing is performed, other processing order may be used. For example, enlargement reconstruction of multiple axial planes, super-resolution processing is performed on each of the multiple axial images that have been enlarged and reconstructed, and the oblique image of the oblique plane is cross-sectional converted from the multiple axial images that have undergone super-resolution processing It may be generated by processing.

Next, an application example of the super-resolution processing in the super-resolution processing unit 32 will be described. Here, a device for improving the convergence and image accuracy of the iterative solution of super-resolution processing is provided. First, as is well known, for example, in the case of a coronary artery CT examination, examples of CT values of each part are as follows.
Blood vessel: 60HU
Blood (contrast agent): 200-300HU
Fat: -80 ~ -50HU
Stent: 500HU or more
In particular, since the CT value of a stent is high, the influence of blur is large, and the number of iterations is required for convergence due to the size. On the other hand, other tissues have a low CT value and are easily affected by image noise. Therefore, when the repetition is repeated many times as in the case of a stent, the tissue does not converge and tends to diverge. Therefore, it is effective to repeat a high contrast portion and a low contrast portion such as a stent under different convergence conditions or by dividing an image. As shown in FIG. 8, when an object having a high CT value such as a stent and an object having a low CT value such as a blood vessel, fat, or myocardium coexist in the measurement image (A), the stent image and its periphery are first extracted from the image. That is, a low contrast portion is removed by threshold processing on the measurement image, and a high contrast portion such as a stent is extracted together with its peripheral portion. For example, a portion within the distance of the radius of the PSF function is extracted as a peripheral portion from each high contrast pixel extracted by threshold processing. For example, the minimum pixel value in the measurement image or the threshold value used in the threshold processing is given to the pixel that has not been extracted. Using PSF under certain constraints on the extracted image of the high-contrast portion and its peripheral portion, for example, under the constraint that the presence of a value below the minimum pixel value or threshold is not allowed A deconvolution process is performed, and then an iterative method is applied, for example, 147 times using it as an initial solution. Thereby, a super-resolution image (B) of a high contrast portion (here, a stent) alone is generated.

  Next, PSF is convoluted with respect to the super-resolution image (B) of the high-contrast portion alone, thereby generating a blurred image (C) of the high-contrast portion alone. By subtracting the blurred image (C) of the high contrast portion alone from the measurement image (A), an image (D) of the low contrast portion alone other than the high contrast portion is generated, and the image (D) of the low contrast portion alone is generated. Is combined with the super-resolution image (B) of the high contrast portion alone and displayed. Accordingly, a super-resolution image of the high contrast portion can be obtained with high accuracy without being affected by the low contrast portion, and at the same time, the image of the low contrast portion can be separated from the high contrast portion.

  In addition, as shown in FIG. 9, the super-resolution processing is performed on the blur image (D) of the low contrast portion generated by subtracting the blur image (C) of the high contrast portion alone from the measurement image (A). The super-resolution image (E) in the low contrast portion is generated, and the super-resolution image (E) in the low contrast portion is combined with the super-resolution image (B) in the high contrast portion and displayed. Also good.

  As described above, by separating the high-contrast portion and the low-contrast portion and separately performing the super-resolution processing, it is possible to improve the convergence of the iterative solution and the image accuracy of the super-resolution processing.

  Similarly to the above, when the spatial frequency of the super-resolution processing target image becomes higher than the required resolution, the convergence becomes worse in the iterative solution. Therefore, it is necessary to limit the recovery band so as not to diverge. Therefore, band limitation (low pass in the frequency space) is performed so that the initial solution does not include unnecessary high frequency components.

O ₁ = M (*) PSF * F F: Band Limit Filter Also, each correction vector is band-limited so as not to include a higher frequency component than necessary in the iterative solution.

_{O N + 1 = O N -a} * dE / dx * F
In the two-step iterative solution of A), first, a high-resolution stent image is obtained by leaving a high frequency component with a loose band limitation for a stent. Next, a high-resolution stent image is used as an initial solution, and an image of a blood vessel or the like is restored by strong band limitation in the second iterative solution.

First
O '= M (*) PSF * F ₁ F ₁ : Filter that passes high-frequency components relatively well
O _{N + 1} = O _N -a * dE / dx * F ₁
When it converges, the band limiting filter is changed, and iterative processing is performed again with O _{N + 1} as the initial solution.

O _{N + 1,2} = O _{N, 2} -a * dE / dx * F ₂ F ₂ : Filter that does not pass high-frequency components so that the correction vector is not included in the correction vector in the second iteration, and the first iteration In order to preserve the high frequency component of the final solution ON _{+ 1} , the stent image is preserved. On the other hand, high-frequency components other than the stent image are not included in the second iterative solution ON _{+ 1,2} , and convergence is maintained.

  Also, if the iterative solution is repeated many times, the influence of image noise increases and the solution is likely to diverge. Therefore, it is also important to improve the convergence condition by giving a constraint condition according to the pixel value of the image.

  A pixel value exceeding a certain range is converted into a certain value. The correction amount is limited based on the CT value of each pixel of the measurement image. A certain variation allowable range is determined from each pixel CT value of the measurement image, and each repetitive solution value is clipped. The allowable range may be different for each pixel of the measurement image.

Iterative solution _{values: O N + 1 = CLIP (} O N - a * dE / dx)
CLIP (): Clipping function Example: If a pixel of the measurement image O is 50HU, the value of each iterative solution pixel is clipped to 0HU if the value is 0HU or less, and to 100HU if the value is 100HU or more.

  For pixels with a small size and high CT value such as a stent, pixel values fluctuate greatly due to recovery, and it is difficult to set an allowable range. For low CT values such as blood vessels and blood images, even if only pixels are clipped good.

Alternatively, a fixed variation allowable range for each iteration is determined from each pixel CT value of the measurement image, and the correction value is clipped. The allowable range may be different for each pixel of the measurement image.
Modify _{value: O N + 1 = O N} - CLIP (a * dE / dx)
Example: If a pixel in the measurement image O is 50HU, the allowable correction amount per iteration is 5HU.

Alternatively, the inclination vector is obtained from the pixels on the measurement image, and the correction tolerance is increased at a portion where the inclination is large.
Reduce the correction tolerance at a small slope.

  These various constraint conditions and iterative algorithms other than the Jacobi method may be combined.

  Note that the present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. In addition, various inventions can be formed by appropriately combining a plurality of components disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

1 is a diagram showing a schematic configuration of an X-ray computed tomography apparatus equipped with a super-resolution processing apparatus according to an embodiment of the present invention. 5 is a flowchart showing a first super-resolution processing procedure according to the present embodiment. FIG. 3 is a supplementary diagram of the first super-resolution processing procedure of FIG. 2. 9 is a flowchart showing a second super-resolution processing procedure according to the present embodiment. FIG. 5 is a supplementary diagram of the second super-resolution processing procedure of FIG. 4. 9 is a flowchart showing a third super-resolution processing procedure according to the present embodiment. FIG. 7 is a supplementary diagram of the third super-resolution processing procedure of FIG. 6. The figure which shows the application example of the super-resolution process by this embodiment. The figure which shows the other application example of the super-resolution process by this embodiment. The figure which shows the example of an image which received the super-resolution process by this embodiment. The figure which shows the generation | occurrence | production principle of the blur of CT image.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Base, 2 ... Bed, 3 ... Computer unit, 10 ... X-ray tube, 12 ... Rotating frame, 21 ... High voltage generator, 23 ... X-ray detector, 25 ... Base drive device, 26 ... Data collection device, DESCRIPTION OF SYMBOLS 29 ... System controller, 30 ... Scan controller, 31 ... PSF storage part, 32 ... Super-resolution processing part, 34 ... Pre-processing part, 36 ... Reconstruction processing part, 37 ... Projection data storage part, 38 ... Display, 39 ... Input device.

Claims (6)

Means for storing point image intensity distribution function data relating to an X-ray computed tomography apparatus acquired using a phantom;
Means for performing super-resolution processing on the image data relating to the subject using the stored point image intensity distribution function.   2. The super-resolution processing apparatus according to claim 1, wherein the means for performing the super-resolution processing includes a process of deconvolving the stored point spread function with respect to the image data.   3. The super solution according to claim 2, wherein the means for performing the super resolving process includes an iterative process for minimizing an error between the image data and the deconvoluted image data as an initial solution. Image processing device.   2. The super-resolution processing apparatus according to claim 1, wherein two-dimensional or three-dimensional image data relating to the phantom is stored as data of the point image intensity distribution function.   The means for performing the super-resolution processing includes means for resampling image data as data of the stored point image intensity distribution function in accordance with resolution of image data relating to the subject. 4. The super-resolution processing device according to 4. Image acquisition means for acquiring medical image data from a subject;
Means for storing data of a point spread function related to the image acquisition means acquired using a phantom;
A medical image diagnostic apparatus comprising: means for performing a super-resolution process on the medical image data using the stored point spread function.

Images