JP2006138869A - Computed tomograph, and method and program for determining rotation center position - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a CT scanner capable of determining a rotation center from scan data itself of a subject without hardly any errors. <P>SOLUTION: When a rotary shaft 13 is moved in an x direction by an x-shift mechanism 8 relatively moving the rotary shaft 13 in a direction of an X-ray tube 1, the rotation center position moving on transmission data is predicted by a rotation center predicting part 21, and the rotation center position is determined by a rotation center determining part 22 by reflecting the predicted rotation center position on added transmission data of adding a multiplicity of pieces of transmission data of the subject 4 during rotation. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a computer tomography apparatus used for nondestructive inspection, and more particularly, to a high-resolution computer tomography apparatus for inspecting small electronic components and the like with high resolution, and a method and program for obtaining a rotation center position.

  In recent years, high-resolution industrial computer tomography apparatuses (hereinafter abbreviated as “CT scanners”) for inspecting small electronic components and the like with high resolution have been developed.

  For example, a high-resolution CT scanner known in Patent Document 1 uses X-rays as radiation, detects an X-ray beam generated from an X-ray tube and transmitted through a subject with a two-dimensional X-ray detector. A transmission image of the subject is obtained. When taking a cross-sectional image, a large number of transmission images are obtained while rotating the subject once (referred to as scanning). Data processing is performed on the many transmission images to obtain cross-sectional images (one or many) of the subject. For reconstruction of a cross-sectional image, a filtered back projection method (FBP: Filtered Back Projection method) is usually used.

  FIG. 12 is a conceptual diagram of a general high-resolution CT scanner. This high-resolution CT scanner can freely set the X-ray geometry, and has a feature capable of dealing with various objects. The rotation table 104 and the X-ray detector 103 for placing and rotating the subject can be moved closer to or away from the X-ray tube 101 (X-ray focal point F) (x direction), and an imaging distance FCD (Focus to rotation Center). Distance) and detection distance FDD (Focus to Detector Distance) can be changed continuously, and the imaging magnification (magnification rate) (= FDD / FCD) can be changed according to the subject. Further, the rotary table 104 can be moved up and down (z direction), and the imaging position of the subject can be changed.

  In FIG. 12, the scan area (cross-sectional image field of view) is a circle An centered on the rotation center Cn included in the imaging X-ray beam 102 on the rotation plane, and the smaller the imaging magnification, the smaller the circle. The rotation center Cn is usually slightly deviated from the center due to a mechanism error. However, if this deviation is large (at the same photographing magnification), the scan area becomes narrow, which is not preferable. For this reason, the rotary table is configured to be able to move in the direction (y direction) across the X-ray beam along the rotation plane and adjust the rotation center.

  In such a high-resolution CT scanner, in order to obtain a cross-sectional image with good resolution by processing a large number of transmission images, it is necessary to accurately know the rotation center position on the transmission image in units smaller than one pixel. .

  Prior to scanning the subject after finishing the geometric setting, the conventional high-resolution CT scanner is mounted on a pin-shaped phantom and photographed to calibrate (scale) the rotation center. Calibration is performed by obtaining the detected ch position corresponding to the rotation center and storing it in the data processing unit.

  As known in Patent Document 1, the center of rotation can also be obtained from the scan data itself of the subject. When this is adopted, calibration of the center of rotation can be omitted. This rotation center finding utilizes the fact that “transmission data added by 360 ° is symmetrical on the left and right of the rotation center”.

Further, this high resolution CT scanner can shift the center of rotation, scan the subject so that it protrudes on one side, and photograph a large subject (Patent Document 2). Since this scan is set (offset) by shifting the rotation center, it is called an offset scan. Referring to FIG. 12, the scan area (cross-sectional image field) in the offset scan is a circle Aof in contact with one side of the imaging X-ray beam 102 around the rotation center Cof on the rotation plane, and is larger than the normal scan in the same FCD. .
JP 2000-298105 A JP 2002-62268 A

  The high-resolution CT scanner disclosed in Patent Document 1 uses the transmission data obtained by adding (average) 360 ° to obtain the rotation center from the scan data itself of the subject, thereby rotating the rotation center on the pin-shaped phantom. Calibration is unnecessary.

  However, depending on the subject, finding the rotation center may cause an error. In particular, this error is frequent in the case of offset scanning, and this method cannot be used for offset scanning.

  FIG. 13 is a conceptual diagram showing an example of finding a conventional rotation center. FIG. 13A shows a situation where the capacitor 106 is set in the X-ray beam 102 as the subject shown in the sectional view. The center of rotation C is set close to the end of the X-ray beam 102 by offset scanning. FIG. 13B shows average transmission data (after logarithmic conversion). In this case, transmission data in the rotation plane is averaged by 360 ° and arranged at equal detection angle intervals.

  From this average transmission data, the center of rotation m that is symmetric is obtained by, for example, the correlation method. At this time, not only the correct center of rotation m1 but also the points m2 and m3, for example, can be seen. Since the left and right appear to be symmetrical, m2 and m3 may be erroneously determined as the rotation center.

  In addition, it is preferable to adopt offset scanning because the scanning area is widened. However, in the high resolution CT scanner, the xy movement of the table can be freely set, but the table is offset. The problem is that it is difficult to set the position. Appropriate X-ray geometry can be obtained by selecting the FCD so that the subject can fit in the scan area and setting the rotation center offset to the end of the field of view with a margin, but the rotation center is visually visible on the subject transmission image. In addition, if the FCD (or FDD) is changed, the rotation center is shifted on the screen due to a change in the enlargement ratio or due to a mechanism error, so that it is difficult to set an appropriate rotation center.

  An object of the present invention is to provide a CT scanner capable of obtaining the rotation center with less error from the scan data itself of the subject, and a method and program for obtaining the rotation center position.

In order to solve the above-mentioned problems, the present invention has a rotating means for rotating the subject relative to the radiation beam emitted from the radiation source, and the subject at a number of the rotational positions obtained by the radiation detector. In a method for obtaining the center position of rotation in a computed tomography apparatus that obtains a cross-sectional image of the subject from a large number of transmission data, transmission data at a number of points on a sinogram created by the large number of transmission data of the subject A step of correlating the multiple points determined by setting a virtual rotation center with transmission data at multiple points that respectively form a reverse radiation path, and changing the virtual rotation center to satisfy the correlation satisfying a predetermined condition. Obtaining the virtual rotation center to be provided as a rotation center position.

  Furthermore, in order to solve the above-mentioned problems, the present invention has a rotating means for rotating the subject relative to the radiation beam emitted from the radiation source, and the object at a number of the rotation positions obtained by the radiation detector. In a computer program for obtaining a center position of rotation in a computed tomography apparatus that obtains a cross-sectional image of the subject from a large number of transmission data of the specimen, the computer tomography apparatus includes a sinogram on a sinogram generated by the transmission data of the subject. The step of correlating the transmission data at a plurality of points with the transmission data at a plurality of points forming a reverse radiation path and the plurality of points determined by setting the virtual rotation center, and changing the virtual rotation center And obtaining the virtual rotation center that gives the correlation satisfying a predetermined condition as the rotation center position. To.

  In order to achieve the above object, the present invention further includes a rotating means for rotating the subject relative to the radiation beam emitted from the radiation source, and the object at a number of the rotation positions obtained by the radiation detector. In a computed tomography apparatus that obtains a cross-sectional image of a subject from a large number of transmission data of a specimen, transmission data at a plurality of points and a virtual rotation center are set on a sinogram created by the large number of transmission data of the subject. And correlate the transmission data at the multiple points forming the opposite radiation path with each of the multiple points determined in the above, and changing the virtual rotation center to give the correlation satisfying a predetermined condition as the rotation center position The center of rotation to be obtained is provided.

  According to the method, program or apparatus having such a configuration, since the rotation center is obtained by directly correlating the sinogram on the sinogram, the rotation direction is compared with the conventional method of obtaining the correlation on the average transmission data. Since there is no reduction in the amount of information due to the addition of, errors due to the subject are less likely to occur and the accuracy is good. This makes it possible to obtain the rotation center from the transmission data itself of the subject even for an offset scan that is likely to cause an error.

  As described above, according to the present invention, it is possible to provide a CT scanner capable of easily obtaining an offset scan from an object scan data itself and capable of easily performing an offset scan, and a method and program for obtaining a rotation center position. Can do.

(First embodiment)
A first embodiment of a CT scanner according to the present invention will be described with reference to FIGS.

  The CT scanner of this embodiment uses X-rays as radiation, and the X-ray tube 1 is generated as shown in FIG. 1A viewed from above and FIG. 1B viewed from the side. The X-ray focal point F is a microfocus X-ray tube with several to several tens of μm, and the X-ray detector 3 has an X-ray II (image intensifier: image intensifier tube) and a television camera.

  The X-ray tube 1 and the X-ray detector 3 are supported by an x shift mechanism 8 so as to face each other with the subject 4 interposed therebetween. The subject 4 is placed on the rotary table 5, and is rotated along the imaging plane 14 (cross-sectional image) within the X-ray beam 2 by the rotation / elevating mechanism 6 and is moved up and down at right angles to the imaging plane 14 (z direction). )

  The subject 4 is moved with the rotary table 5 by the y shift mechanism 7 across the X-ray beam 2 (in the y direction in the figure) and between the X-ray tube 1 and the X-ray detector 3 by the x shift mechanism 8. And the shooting distance FCD is changed. The X-ray detector 3 is moved by the x shift mechanism 8 and the detection distance FDD is changed. C is the rotation center, and D is the detection center.

  As other components, a data processing unit 19 that processes a transmission image from the X-ray detector 3, a display unit 20 that displays a processing result and the like, and a mechanism that controls the mechanism unit with a command from the data processing unit 19 The X-ray control unit 17 and the high-voltage generator 16 for controlling the tube voltage and tube current of the control unit 18 and the X-ray tube 1, and the portion including the X-ray tube 1, the subject 4 and the X-ray detector 3 are accommodated. There are no X-ray shielding boxes.

  An encoder (not shown) is attached to the x shift mechanism 8 and the y shift mechanism 7, and the FCD value, the FDD value, and the y value are read and sent to the data processing unit 19 through the mechanism control unit 18, respectively.

  The data processing unit 19 and the display unit 20 are configured by a normal computer, and the computer includes a CPU, a memory, a disk, a keyboard, an interface, and the like, and software for reconstructing a cross-sectional image from tomographic sequences and data. Etc. are remembered.

  The operator operates the data processing unit 19 and the display unit 20 to perform menu selection, condition setting, manual operation of the mechanism unit, start of tomography, apparatus status reading, cross-sectional image display, cross-sectional image analysis, etc. Do.

  The data processing unit 19 functions as a software functional block as a software function block, a rotation center predicting unit 21 that predicts a rotation center position that moves on the transmission image due to a magnification change or a mechanism error when x-shifted, and a rotation center on this image from the transmission image. A rotation center obtaining unit 22 for obtaining a position, a tomographic scan control unit 23, a reconstruction unit 24 for creating a cross-sectional image, and the like are included.

  Next, the operation of the CT scanner of this embodiment will be described with reference to FIG.

  First, the operator places the subject 4 on the rotary table 5 and inputs a command to the data processing unit 19 to radiate the X-ray beam 2 from the X-ray tube 1 and display a transmission image of the subject 4 on the display unit. Rotate the rotary table 5 manually or electrically while displaying it as a real-time moving image on 20, and change the FCD (and FDD) so that the transmitted image is just within the field of view of the image at all rotational positions to increase the shooting magnification. Adjust (normal scan).

  In the case of offset scanning, the center of rotation is moved near the edge of the field of view by shifting y. Next, the subject 4 is moved up and down so that the examination position is aligned with the imaging surface 14.

  At this time, the rotation center predicting unit 21 predicts a rotation center position that moves on the transmission image due to a change in magnification or a mechanism error, as will be described later, from the read FCD, FDD, and y values. Note that this expected rotation center is not accurate enough to be used for reconstruction.

  Next, when a scan command is input to the data processing unit 19, the scan control unit 23 of the data processing unit 19 captures the transmission image that is the output of the detector 3 while rotating the rotary table. From the transmission data of the imaging surface 14 position of the transmission image over 360 °, first, the rotation center obtaining unit 22 obtains the rotation center as will be described later, but the error is less likely to occur by reflecting the predicted rotation center position at this time. .

  Next, the reconstruction unit 24 reconstructs a cross-sectional image at the position of the imaging surface 14 by using a filter-corrected back projection method or the like using the rotation center as in the prior art. The reconstruction unit 24 reconstructs a number of cross-sectional images using transmission data other than the position of the imaging surface 14 by using the Feldkamp method in addition to the above-described reconstruction by the filter-corrected back projection method.

  Next, the operation of the rotation center prediction unit 21 will be described.

  FIG. 2 is a geometric view of the mechanism error δy. This shows the positional relationship between the X-ray focal point F, the rotation center C, and the detection center D on the imaging surface 14. The direction connecting F and D is the center line 30, that is, the x axis, and the x axis moves as D moves. The mechanism error δy is a deviation from the center line 30 of the rotation center C0 when the y shift is set to the origin 0. In FIG. 2, the y direction is extended in order to emphasize the mechanism error.

  In order to predict the center of rotation, it is necessary to perform mechanism error calibration in advance.

  FIG. 3 is an explanatory diagram of mechanism error calibration. This figure shows a calibration point 33 by a combination of FCD and FDD. In the FCD, the calibration point is increased at a position where the photographing magnification is large so as to increase the accuracy. A scan of a pin phantom or the like is performed with the y shift at the origin 0 at each calibration point 33, and the rotation center position δm0 (dc) is obtained from the transmission data, and the deviation δy of the rotation center C0 from the center line 30 from δm0 (dc). Ask for. This δy (FCD, FDD) is stored as a calibration table.

Referring to FIG. 2, in the rotation center prediction, first, a deviation δy is obtained by four-point interpolation using a calibration table δy (FCD, FDD) from FCD and FDD values. Next, using the y value,
Δθ · δm = atan ((δy + y) / FCD) (1)
Thus, the rotational center position δm from the detection center is predicted.

  Here, Δθ is an angular interval of transmission data. This δm is obtained by adding a change in magnification and a mechanism error.

  As described above, the rotation center predicting unit 21 performs mechanism error calibration in advance, creates a calibration table from transmission data, and predicts the rotation center position δm by, for example, four-point interpolation using the calibration table. it can.

  Next, the operation of the rotation center finding unit 22 will be described.

  This rotation center finding utilizes the fact that “transmission data added by 360 ° is symmetrical”. This is true at any stage of data preprocessing. This is the rotation center finding described in Patent Document 1, which will be briefly described here.

  FIG. 4 shows average transmission data P (m) (after logarithmic conversion). This is obtained by converting transmission data in the rotation plane into equal detection angle intervals (detection angle is an arrangement angle in a fan-shaped X-ray beam) and averaging 360 °. First, with m as a detection angle and n as a rotation angle, a sinogram P (m, n) created by transmission data is added (averaged) by 360 ° in the direction of n to obtain average transmission data, P (m). Here, the addition may be performed by thinning rather than all n. Next, the virtual rotation center m0 is set, P '(m) turned back here is obtained, the absolute value of the difference from P (m) is added for m, the correlation value is obtained, and the correlation value is obtained by changing m0. Let m0 that minimizes be the rotation center mc. As the correlation value, a value corresponding to the mean square error may be obtained instead of taking an absolute value. In addition, when the addition is performed, a correlation value is obtained by multiplying a weight which becomes smaller as the m value becomes farther from m0, so that the accuracy can be improved.

  Here, the center search area (m0 variable range) 35 is set around the position of δm from the detection center D (the center of the transmission data) using δm predicted by the “rotation center prediction”. The region size is determined in consideration of the stability of mechanism error and the accuracy of prediction. Here, instead of determining the center search region 35 by reflecting δm, the center of rotation may be obtained by multiplying the correlation value by a weight that increases as the distance from the δm position (m0) increases. Further, both determination of the center search area 35 and weighting may be performed. There is also a method in which a search is made to the left and right from the δm position and the first minimum value is set as the rotation center.

  As described above, according to the rotation center obtaining unit 22, the average transmission data obtained by adding a large number of transmission data of the subject 4 during the rotation, with respect to the transmission data at many points and the virtual rotation center. It is possible to obtain a correlation between the multiple points and transmission data at multiple points that are symmetrical points, and change the virtual rotation center to obtain the virtual rotation center that gives the best correlation as the rotation center position.

  According to the first embodiment of the present invention, the rotation center predicting unit 21 reflects the rotation center on the transmission data predicted from the magnification change and the mechanism error, that is, with an emphasis on the predicted position, the rotation center. Since the obtaining unit 22 obtains the rotation center from the transmission data itself of the subject, an error in obtaining the rotation center by the subject is less likely to occur. As a result, the center of rotation can be obtained from the transmission data itself of the subject even for an offset scan in which the correlation calculation range (m range where P (m) and P ′ (m) overlap) is small and error is likely to occur. Become.

  In addition, since the center search can be limited rather than the entire area, there is an effect that the calculation time can be shortened.

  In the first embodiment, it is possible to add a rotation center synthesizing unit that displays the predicted rotation center position so as to overlap the transmission image. The rotation center synthesis means is a software block in the data processing unit 19.

  FIG. 5 shows a transparent image screen 36. The operator places the subject 4 on the rotary table 5 and manually operates (electrically) FCD, FDD, and y shift (and rotation and elevation) to set the X-ray geometry.

  At this time, the operator performs adjustment while observing the transmission image screen 36 (real-time moving image) in FIG. The rotation center synthesizing means displays the rotation center position calculated in real time according to the FCD, FDD, and y values that change every moment on the transmission image as a rotation center 37. An imaging plane 38 is also displayed on the transparent image screen 36 separately.

  In the case of offset scanning, for example, the operator first adjusts the elevation, then stops at the position where the left portion L of the center of rotation is the longest while rotating, and then moves the FCD, FDD, y shift. Adjustment is performed so that the left end of the subject 4 is within the visual field and the center of rotation is at the right end of the visual field (including a margin).

  As described above, since the center of rotation is displayed over the transmission image, it is possible to easily set an optimum offset scan. In the case of normal scanning, adjustment can be easily performed so that the center of rotation is at the center of the field of view. The rotation center display may use the rotation center obtained instead of the expected rotation center, for example, when scanning continuously without changing the FCD and FDD.

  The combined display of the rotation center on the transmission image can also be used when the subject is placed at the rotation center position, and the placement position can be changed by observing the transmission image. This function can be used effectively for adjusting the placement position of the subject even in the case of an apparatus without a y-shift mechanism (no offset scan function). This function is particularly effective in the case of an apparatus provided with an electric mechanism for moving the subject xy on the rotary table.

  In the first embodiment, the rotation center prediction is predicted from the FCD, FDD, and y values. However, when the lift mechanism error is large, it may be predicted from the FCD, FDD, y value, and the lift value.

(Second Embodiment)
The configuration of the CT scanner of the second embodiment is the same as that of the first embodiment shown in FIGS. 1A and 1B, as shown in FIG. 6A viewed from above and FIG. 6B viewed from the side. The rotation center prediction unit 21 is deleted, and a rotation center calculation unit 22 ′ is provided in place of the rotation center calculation unit 22 utilizing the fact that “transmission data added by 360 ° is symmetrical”. It is. The rotation center obtaining unit 22 of the first embodiment utilizes the fact that “transmission data added by 360 ° is bilaterally symmetrical”, whereas the rotation center obtaining unit 22 ′ of the present embodiment is “reverse to each other”. The transmission data of the X-ray path in the direction is substantially the same ”.

  The operation of the second embodiment is that the “revolution center prediction” is excluded from the operation of the first embodiment, and the rotation center finding unit 22 ′ instead of the rotation center finding unit 22 realizes the rotation center finding unit 22 ′. is there.

  First, in the rotation center obtaining unit 22 ′, transmission data at the imaging plane 14 position of the subject transmission image over 360 ° is rotated on the horizontal axis for the detection angle m (arrangement angle in the fan-shaped X-ray beam) and on the vertical axis. Using the so-called sinogram P (m, n) arranged at the angle n, the rotation center position on the sinogram is obtained.

  The basic principle of finding the center of rotation is that “transmission data of X-ray paths in opposite directions are substantially the same”. This center finding uses reverse conversion. FIG. 7 is an X-ray path diagram for explaining reverse conversion. This indicates an X-ray path viewed from the direction of the rotation axis with coordinates fixed on the rotary table, and the X-ray focal point F rotates. The reverse conversion 40 obtains the reverse path (mr, nr) from one X-ray path (m, n). The reverse conversion 40 is expressed by the following equation.

mr−m0 = − (m−m0) That is, mr = 2 · m0−m (2)
nr · Δφ = n · Δφ + 180 ° −2 · (m−m0) · Δθ, ie, nr = n + 180 ° / Δφ−2 · (m−m0) · Δθ / Δφ (3)
Here, m0 is the rotation center (detection angle thereof), Δθ is the angular interval of detection angles, and Δφ is the angular interval of rotation.

  FIG. 8 is a sinogram P (m, n). If the virtual rotation center m0 is set in this figure, the reverse path (mr, nr) can be calculated from an arbitrary (m, n). In order to obtain the center of rotation, the sinogram value P is correlated with (m, n) and each reverse path (mr, nr) in a predetermined region. When m0 is correctly set as the rotation center, P (m, n) is approximately equal to P (mr, nr) and the correlation is improved by matching (reverse) each path. The center of rotation is obtained by changing m0 and taking the correlation. The sinogram P (m, n) is usually obtained by performing transparent data up to logarithmic conversion, but may be before logarithmic conversion or may be data at any stage of processing. This can be understood from the basic principle described above. As the predetermined area of (m, n), the left and right narrow sides (m0 to M, all n) of m0, and the area (2 · m0-M to m0, all n) where this area is folded around m0 are used. ) Is used as a correlation region (2 · m0-M to M, all n) 41, but may be a part thereof.

  In the sinogram P (m, n), m and n are integer values.

  FIG. 9 is a flowchart of the rotation center finding 2. The rotation center finding 2 will be specifically described with reference to FIGS. 9 and 8.

  S1: Initial setting of m0 (real number).

  S2: The correlation value and SOKAN are reset.

  S3: Determine the loop start value of m, ms (integer).

ms = INT (2 · m0−M) +1 (4)
S4: Enter a loop from m = ms to M.

  S5: Obtain mr by reverse conversion (formula (2)). Here, mr is generally a real number.

  S6: Enter a loop of n = 0 to N-1.

  S7: nr is obtained by reverse conversion (formula (3)). Here, nr is generally a real number.

  S8: Since mr and nr are real numbers, the point (mr, nr) is different from the data point. Therefore, a value Pr (mr, nr) at this point is obtained by interpolation calculation from surrounding data.

  S9: The absolute value of the difference between the data values of the reverse path is added to the correlation value.

SOKAN = SOKAN + ABS (Pr (mr, nr) −P (m, n)) (5)
Here, instead of taking the absolute value, it may be squared and a correlation value corresponding to the mean square error may be obtained, but there is no big difference in the final result (rotation center).

  S10: Repeat the loop of m and n.

  S11: It is determined whether the SOKAN value is minimum.

  S12: If it is determined that the SOKAN value is not minimum, m0 is changed and the process returns to S2.

  S13: When it is determined that the SOKAN value is the minimum, the process ends with m0 as the rotation center.

mc = m0 (6)
Here, the loop form for finding the minimum SOKAN value is simplified for the sake of clarity. Actually, for example, m0 is changed in a certain step to obtain respective SOKAN values, a region of m0 having a small SOKAN value is determined, a SOKAN value is calculated in small steps, and m0 having the smallest SOKAN value is set as the rotation center. Perform a calculation that

  The flowchart describes only basic calculations, and in practice, various processes for increasing the calculation accuracy can be added. For example, in the correlation value accumulation (formula (5)), the weight can be increased by multiplying the ABS (Pr (mr, nr) -P (m, n)) by a weight that becomes smaller as the m value moves away from m0. it can.

  Also, a weight depending on P (m, n) can be multiplied by the same calculation. In this case, accuracy is increased by using a small weight for a portion where absorption is too strong and the transmitted X-ray dose is small (P large) or a portion where the subject is hardly suspended and the transmitted X-ray dose is excessive (P small). be able to.

  In addition, various modifications are possible without regard to accuracy. For example, the SOKAN value may be obtained as an average value by dividing by the number of (m, n) points of calculation. Further, in order to increase the calculation speed, the roof range of m can be set to one side of m0, only the portion close to m0 on one side can be set, and the loop of n can also be calculated to be skipped.

  The rotation center finding method by the rotation center finding unit 22 ′ used in the second embodiment of the present invention has a direct correlation from the sinogram as compared with the rotation center finding method by the rotation center finding unit 22 in the first embodiment. Since the amount of information does not decrease due to the addition in the rotation direction, errors due to the subject are less likely to occur and the accuracy is good. In particular, even in the case of offset scanning where the center of rotation is close to the end of the sinogram, the center can be found satisfactorily (despite the correlation region 41 becoming smaller). Therefore, the center of rotation can be obtained from the transmission data itself of the subject even for the offset scan.

  In addition, even when the subject is long and thin and the absorption is very large in one direction, there is an advantage that the center can be found satisfactorily by dropping the weight of the region on the sinogram (P becomes large).

  In 2nd Embodiment, the effect | action by the rotation center estimation part 21 can be added similarly to 1st Embodiment. In this case, δm obtained by the rotation center predicting unit 21 described above is reflected when the rotation center is obtained by the rotation center obtaining unit 22 ′. Similarly, the center search region (m0 variable range) is set around the position of δm from the detection center D (the center of transmission data) using δm predicted by the “rotation center prediction”. The region size is determined in consideration of the stability of mechanism error and the accuracy of prediction. Here, instead of determining the central search region by reflecting δm, the center of rotation may be obtained by multiplying the correlation value by a weight that increases as the distance from the δm position (m0) increases. Further, both center search area determination and weighting may be performed. There is also a method in which a search is made to the left and right from the δm position and the first minimum value is set as the rotation center.

  According to this modified example, the rotation center predicting unit 21 reflects the rotation center on the transmission data predicted from the mechanism error, that is, with the emphasis on the predicted position, the rotation center obtaining unit 22 'applies the subject. Since the rotation center is obtained from the sinogram itself, an error in obtaining the rotation center due to the subject is less likely to occur. As a result, the center of rotation can be obtained from the transmission data itself of the subject even for an offset scan in which the correlation calculation range is small and errors are likely to occur.

  In addition, since the center search can be limited rather than the entire area, there is an effect that the calculation time can be shortened.

  In addition, it can deform | transform similarly to the modification of 1st Embodiment, and can acquire the same effect.

  In the embodiment of the present invention, the X-ray detector 3 uses an X-ray II (image intensifier tube) and a television camera, but is arranged along another two-dimensional X-ray detector or the imaging surface 14. It can be easily understood that a one-dimensional X-ray detector or the like may be used.

  In the embodiment of the present invention, the imaging position of the subject 4 is changed by raising and lowering the rotary table 5, but another raising and lowering mechanism may be installed on the rotary table 5. The X-ray detector 3 may be moved up and down as a single body. Further, the X-ray tube 1 and the X-ray detector 3 may be moved as a single body for the rotation of the subject 4 and the x-shift and y-shift of the rotary table 5 (rotation axis). Further, it is not always necessary to set the z direction in the up-and-down direction, and the entire apparatus may be inclined in any direction. That is, the present invention is also applied to other mechanism systems in which the movements among the X-ray tube 1, the X-ray detector 3, the subject 4, and the rotating shaft 13 that are radiation sources are relatively the same.

  Two examples of the mechanism system are illustrated. FIG. 10 shows another mechanism system 1. This rotates the X-ray tube 61 and the X-ray detector 63 arranged on the rotating frame 65 around the subject 64 as a single body. The X-ray tube 61 is moved to xy on the rotating frame. This method is used when examining a non-movable object such as a soft biological tissue.

  FIG. 11 shows another mechanism method 2. This is an apparatus for taking a cross-sectional image of a subject 64 during a load test. The subject rotates, but the xy movement is performed not by the subject but by the X-ray tube. The X-ray tube 61 is moved in the horizontal plane by the xy mechanism 66, and the X-ray detector 63 is moved by the x mechanism 80 in the direction of the X-ray focal point F. The subject 64 is rotated in a horizontal plane while a compression or tension load is applied. The subject 64 is fixed by the workpiece fixing portions 79a and 79b, and a load is applied when the workpiece fixing portion 79a supported by the lifting frame 72 moves up and down. The subject 64 is rotated by rotating the shaft 75 by the motor 74 and rotating the fixing portion 79 a via the pulley 76 a, the toothed belt 77 a, the pulley 78 a, and the spline shaft 76 fixed to the shaft 75. The fixing portion 79b is synchronously rotated through the pulley 76b, the toothed belt 77b, and the pulley 78b. While the connection between the pulley 78a and the spline shaft 76 is engaged with the rotation direction without any deviation, the rotation axis direction (the up-and-down direction) is connected to slide freely so that the rotation of the pulley 78a on the fixed side can be raised and lowered even during the up-and-down movement. Can be transmitted to the side spline shaft 76.

  The operation of the rotation center prediction unit 21 in each embodiment of the present invention can be variously changed without changing the gist of the present invention. For example, when the movement trajectory of points C and D can be approximated by a straight line, the coefficient of each straight line can be obtained by calibration and δy can be calculated by geometric calculation. Further, the relationship of δy-FCD is considered as a straight line or a curve (polynomial), and the coefficient of the straight line or curve is obtained for several points of FDD, and when calculating δy, the coefficient is interpolated with the value of FDD. Some methods are used. It is also possible to add the rotation center obtained by the rotation center obtaining unit 22 as calibration data and perform the next prediction with an emphasis on new calibration (learning type calibration).

  Further, the rotation center predicting means of the present invention predicts the position of the rotation center by taking into account only the change in magnification when the mechanism error is negligible.

Further, in the embodiment of the present invention, δy is calculated when FCD and FDD are changed. However, y can be controlled using δy so that the center of rotation is always at the same position in the image. For example, if the control is performed so that y = −δy, the center of rotation can be always aligned with the center. If you want to adjust to the position of dc from the center on the screen,
y = dc · FCD / FDD−δy (7)
Control may be performed so that Thereby, the enlargement ratio can be changed without changing the rotation center position on the transmission image by adjusting the offset scan.

  In each of the above-described embodiments, one or more of the rotation center prediction unit 21, the rotation center finding units 22, 22 ', the scan control unit 23, and the reconstruction unit 24 constituting the data processing unit 19 are general-purpose or dedicated computers. The computer can be configured as software that can be executed by the computer, and the computer can be configured by one or a plurality of computers or functions of a part of these computers.

1 is a configuration diagram showing a CT scanner according to a first embodiment of the present invention. FIG. FIG. 6 is a geometric view of a mechanism error δy in the CT scanner of the same embodiment. Mechanical error calibration explanatory diagram in the CT scanner of the embodiment. The figure which shows the average transmission data P (m) in CT scanner of the embodiment. The figure which shows the transparent image screen in CT scanner of the embodiment. The lineblock diagram showing the CT scanner concerning a 2nd embodiment of the present invention. FIG. 3 is an X-ray path diagram illustrating reverse conversion in the CT scanner of the embodiment. The figure which shows sinogram P (m, n) in CT scanner of the embodiment. The flowchart of the rotation center search 2 in the CT scanner of the embodiment. The figure which shows the other mechanism system 1 in CT scanner of the embodiment. The figure which shows the other mechanism system 2 in CT scanner of the embodiment. A conceptual diagram of a general high resolution CT scanner. The conceptual diagram which shows the example of a rotation center seek in the conventional CT scanner.

Explanation of symbols

Underline: All itself 1 ... X-ray tube, 2 ... X-ray beam, 3 ... X-ray detector, 4 ... subject, 5 ... rotary table,
6 ... Rotation / lifting mechanism, 7 ... y shift mechanism, 8 ... x shift mechanism,
9, 10 ... support frame, 13 ... rotating shaft, 14 ... photographing surface, 16 ... high voltage generator,
17 ... X-ray control unit, 18 ... Mechanism control unit, 19 ... Data processing unit, 20 ... Display unit,
21 ... Rotation center prediction unit, 22, 22 '... Rotation center finding unit, 23 ... Scan control unit,
24 ... Reconstruction unit, 30 ... Center line, 35 ... Center search area, 36 ... Transparent image screen,
37 ... Center of rotation (display), 38 ... Imaging plane (display), 41 ... Correlation region, 61 ... X-ray tube,
62 ... X-ray beam, 63 ... X-ray detector, 64 ... Subject, 65 ... Rotating frame,
66 ... xy mechanism, 70 ... load tester, 71 ... rotating part, 72 ... lifting frame,
73 ... fixed frame, 74 ... motor, 75 ... shaft, 76 ... pulley,
77 ... toothed belt, 78 ... pulley, 79 ... work fixing part, 80 ... x mechanism.

Claims (3)

Rotating means for rotating the subject relative to the radiation beam emitted from the radiation source, and a cross-sectional image of the subject from a plurality of transmission data of the subject at a number of the rotation positions obtained by a radiation detector In the method of obtaining the center position of the rotation in the computed tomography apparatus to obtain
On the sinogram created by a large number of transmission data of the subject, the transmission data at a large number of points, and the transmission data at a large number of points that form a reverse radiation path, respectively, are determined by setting a virtual rotation center. Correlating, and
And determining the rotation center position in the computed tomography apparatus, comprising: changing the virtual rotation center to determine the virtual rotation center that gives the correlation satisfying a predetermined condition as a rotation center position. Rotating means for rotating the subject relative to the radiation beam emitted from the radiation source, and a cross-sectional image of the subject from a plurality of transmission data of the subject at a number of the rotation positions obtained by a radiation detector In the program for obtaining the center position of the rotation in the computer tomography apparatus to obtain
In the computed tomography apparatus,
On the sinogram created by a large number of transmission data of the subject, the transmission data at a large number of points, and the transmission data at a large number of points that form a reverse radiation path, respectively, are determined by setting a virtual rotation center. Correlating, and
A program for determining the rotation center position in the computed tomography apparatus, wherein the virtual rotation center is changed and the virtual rotation center that gives the correlation satisfying a predetermined condition is determined as a rotation center position. Rotating means for rotating the subject relative to the radiation beam emitted from the radiation source, and a cross-sectional image of the subject from a plurality of transmission data of the subject at a number of the rotation positions obtained by a radiation detector In the computer tomography apparatus to obtain
On the sinogram created by a large number of transmission data of the subject, the transmission data at a large number of points, and the transmission data at a large number of points that form a reverse radiation path, respectively, are determined by setting a virtual rotation center. A computer tomography apparatus comprising: a rotation center obtaining unit that obtains a correlation and changes the virtual rotation center to obtain the virtual rotation center that gives the correlation satisfying a predetermined condition as a rotation center position.

Images