JP7198457B2 - Interior CT phase imaging X-ray microscope equipment - Google Patents

Description

The present invention relates to an X-ray microscope (XRM), and more particularly to a phase-contrast X-ray microscope that measures the structure of a sample by converting the phase of X-rays into contrast.

XRM is widely used to observe submicron-level structures of samples. This is a method of irradiating a sample with X-rays and projecting the X-rays that have passed through the sample in an enlarged manner onto a detector to see through the structure inside the sample, reflecting the absorption of X-rays inside the sample object. It is already known, for example, from Non-Patent Document 1 below.

Furthermore, non-patent document 2 already discloses a technique for acquiring a phase image, which is an image of the phase of X-rays that change as they pass through a substance, rather than the absorbance of X-rays. A Talbot interferometer-XRM that obtains a magnified image of a sample by using a phase image acquisition method using a phase grating and a magnifying optical system with a Fresnel zone plate (FZP) is already known from Non-Patent Document 3.

Next, the method of constructing a three-dimensional image is to irradiate X-rays that completely cover the cross section of the sample, measure projection data on all straight lines that pass through the cross section of the sample, and use that data to reconstruct an image. It has become something to do. Such a technique is known from Non-Patent Document 4. However, even if you want a tomographic image of only a small region of interest (ROI) in the object (specimen), in general, projection data on all straight lines that pass through the cross section of the sample, not just the ROI was necessary.

In general, when imaging with high magnification and high spatial resolution, the projected image is often measured at the ROI part of the sample at high magnification, and with conventional CT methods, it is not always possible to reconstruct a 3D image with high accuracy due to the lack of surrounding data. rice field. This is because in the Filtered Back Projection (FBP) method, which is a calculation procedure used for CT image reconstruction, projection data on a straight line that does not pass through the ROI is also required to generate the ROI image. . However, projection data on straight lines that do not pass through the ROI do not contain any ROI information. Therefore, in the field of X-ray CT, a CT imaging method in which only the ROI is irradiated with X-rays and only the projection data on (all) straight lines passing through the ROI is measured to generate only the image of the ROI has been adopted. Developed as CT.

Compared to conventional X-ray CT, which also measures unnecessary projection data, this interior CT has (1) a significant reduction in radiation exposure (specimen damage) outside the ROI, (2) detector size and X-ray (3) It is possible to photograph large objects that do not fit in the field of view. is mentioned.

In interior CT, projection data on a straight line that does not pass through the ROI is not measured, so a method of reconstructing an image from incomplete projection data with partial loss is required. In projection data acquisition with a parallel beam, projection data p(r, θ) (where r is the radius, θ angle) can be measured. In this case, since p(r, θ) where the straight line does not pass through the ROI is not measured, the left and right projection data of each angle θ are truncated and lost. It is necessary to correctly reconstruct the image f(x,y) at the ROI from such truncated projection data.

Much research has been done on this interior CT reconstruction for many years, and Non-Patent Document 5 states that the solution for interior CT image reconstruction is determined solely from projection data and is not determined as mathematically correct image reconstruction. is mathematically proven. Known for this non-uniqueness, many approximate image reconstruction methods have been investigated.

Typical methods include (1) image reconstruction after extrapolating the left and right missing portions of projection data in each direction using a smooth function, and (2) image reconstruction by iterative approximation using incomplete projection data. However, it was not put to practical use due to artifacts caused by approximation errors (Non-Patent Documents 6, 7, 8). Examples of typical artifacts that occur in interior CT include shading artifacts that distort low-frequency image components (see Fig. 2(a)) and cupping effects that increase values around the ROI (Fig. 2). (b)) occurs, and the image values are known to be unstable.

In contrast to these previous studies, a rigorous image reconstruction method for interior CT was devised (Non-Patent Documents 8 and 9), and an arbitrary small region B inside the ROI (that is, the ROI in Fig. 3(a) It was proved that if there is a priori knowledge that the values of the image f(x, y) are known in advance in the area circled inside), the solution for interior CT image reconstruction can be uniquely determined. If there are a plurality of regions B for guaranteeing the uniqueness of the solution, they may be located anywhere within the ROIS even if they are small. This result is known as US Pat.

On the other hand, another method has been devised that enables precise image reconstruction using a priori knowledge (Non-Patent Document 7), which performs highly accurate signal restoration from insufficient measurement data called compressed sensing. Based on the method, it was shown that if the image f(x,y) is piecewise uniform throughout the ROI, the interior CT image reconstruction solution is uniquely determined. Here, the piecewise uniformity means that the image is composed of a finite number of regions having completely constant values, like a numerical phantom (see FIG. 3(b)). This work is known as US Pat.

Y. Yoneda: New Emission X-ray Microscope, Review of Scientific Instruments, Vol.33, (1962), 529-532 U. Bonse and M. Hart: An X-ray Interferometer, Applyied Physics Letter, 6, (1965) 155-156 Y. Takeda, W. Yashiro, T. Hattori, A. Takeuchi, Y. Suzuki and A. Momose: Differential Phase X-ray Imaging Microscopy with X-ray Talbot Interferometer, Applied Physics Express, Vol.1, 11,(2008 ) 117002 Natterer F: The Mathematics of Computerized Tomography. Wiley, 1986 Ye Y, Yu H, Wei Y, Wang G: A general local reconstruction approach based on a truncated Hilbert transform. International Journal of Biomedical Imaging 2007: Article ID 63634, 2007 Kudo H, Courdurier M, Noo F, Defrise M: Tiny a priori knowledge solves the interior problem in computed tomography. Physics in Medicine and Biology 53: 2207-2231, 2008 Yu H, Wang G: Compressed sensing based interior tomography. Physics in Medicine and Biology 54: 2791-2805, 2009 Yang J, Yu H, Jiang M, Wang G: High order total variation minimization for interior tomography. Inverse Problems, 26: Article ID 35013, 2010 Courdurier M, Noo F, Defrise M, Kudo H: Solving the interior problem of computed tomography using a priori knowledge. Inverse Problems 24: Paper No. 065001, 2008 Yasuhiro Wataru, Yoshihiro Takeda, and Atsushi Momose: Efficiency of capturing a phase image using cone-beam x-ray Talbot interferometry, Journal of the Optical Society of America A, 8 (2008) 2025-2039 Atsushi Momose: X-ray phase imaging using the Talbot effect, Synchrotron, 23 (2010) 382-392

U.S. Patent No. 7,697,658 U.S. Patent No. 8,811,700

In XRM, when the target is an unstained biological soft tissue that absorbs little X-rays and is difficult to obtain contrast, or a structure in which resin or thin films with a similar composition are laminated, the phase of the electron beam that passes through the sample object must be adjusted to So-called phase-difference XRM is used, which converts to contrast for observation. Among the above-described phase-difference XRMs, the present invention can construct a high-resolution image even when X-rays are irradiated around the ROI of the sample. It relates to imaging XRM.

In the phase-difference XRM related to the present invention, it is based on the principle of absorption contrast that the higher the density of heavier elements in the sample object, the greater the scattering of X-rays, and the clearer the contrast can be obtained. Image contrast occurs. However, when the target is an unstained biological tissue whose main component is the same kind of light element, or a structure laminated with resin or thin films with a similar composition, it is difficult to obtain contrast in the image, so this is a method to solve this problem. For this purpose, phase imaging XRM is used in which an X-ray grating is used to convert the phase of X-rays that pass through a sample object into contrast for observation.

However, when attempting to obtain an image with a high spatial resolution using conventional phase-difference XRM, the image recording range becomes narrow due to the limitations of the size of the optical system and image detector. Even if phase-contrast X-ray image data for using the CT method is acquired in such a state that phase-contrast X-ray CT data for the entire sample cannot be acquired, image reconstruction with desired accuracy cannot be performed. There is a drawback that high spatial resolution CT data cannot be acquired.

In order to overcome the above drawbacks, an interior CT image analysis method is used. Knowledge (image values in a small region B inside the ROI in FIG. 3a) must be known, but the situation where the image values are known before imaging is very rare. In addition, since the exact solution method of Non-Patent Document 7 assumes that part of the ROI of the image is piecewise uniform, this method loses smooth density changes.

Therefore, the present invention improves the image contrast and improves operability even when targeting unstained biological tissue containing the same kind of light element as a main component, or a structure in which resin or thin films having a similar composition are laminated. An object of the present invention is to provide an interior CT phase imaging XRM capable of measuring a high spatial resolution three-dimensional tomographic image of a desired region of a part of a sample by an excellent and more practical interior CT image reconstruction method.

In order to achieve the above-mentioned object, a phase imaging XRM equipped with an optical system for phase difference image data measurement that can easily exchange two or more types of image recording ranges was devised. According to the present invention, first, a phase-contrast XRM image of the entire sample is captured at low magnification with a large image recording range, and based on the low-magnification phase-contrast XRM image, the sample is captured at high magnification and high spatial resolution. A region ROI was determined, and a high-magnification X-ray optical system was used to obtain high-magnification and high-spatial-resolution data. By doing so, interior CT image reconstruction can be performed by using the reconstruction data of the measurement points inside the desired high-magnification measurement ROI among the phase-difference XRM reconstruction images during low-magnification measurement. It is possible to construct a capable interior CT phase imaging XRM.

Further, in the present invention, in the interior CT phase imaging XRM described above, optical elements such as X-ray gratings and Fresnel zone plates (FZP) of X-ray optical systems with different magnifications are optimal for each optical system with different magnifications. In terms of usability of the interior CT phase imaging XRM, it is preferable that the interior CT phase imaging XRM is installed on a jig that can be adjusted in position and tilt so that the interior CT phase imaging XRM can be adjusted, and has a mechanism that allows these optical systems to be easily switched by operations such as translation and rotation. To get the maximum field of view, some optics do not have the FZP attached, including 1:1 optics that do not magnify.

Furthermore, in the interior CT phase imaging XRM described above, the position of the optical element is slightly changed to acquire a plurality of images, thereby detecting the phase change, amplitude change, or scattering by the sample as an image. is preferred.

For the imaged low-magnification image data, projection data is acquired by X-rays passing through the sample under conditions where all projection images of a cross section perpendicular to at least one axis on which the sample is rotated for cross-sectional imaging can be captured. Using the obtained projection data, the first stage of approximate reconstruction is performed by the CT image reconstruction method. Next, the projection data is measured at the maximum magnification at which all projection images of the ROI are measurable, and the image numerical value representing the physical quantity is at least piecewise uniform or uniform within the ROI based on the reconstructed CT image. By specifying a region that is piecewise represented by a polynomial, and representing the physical quantity piecewise uniformly or piecewise by a polynomial in the position and inside of the specified region, rather than the reconstruction in the first stage The image reconstruction method for interior CT performs a highly accurate second-stage reconstruction.

In the present invention, in the interior CT image reconstruction method described above, the numerical value of the projection data representing the physical quantity may include the absorption of the X-rays by the imaging target, or the X-ray absorption by the imaging target. It may comprise a phase shift of x-rays, or it may comprise diffraction of said x-rays by said imaging object, or it may comprise scattering of said electron beam by said imaging object.

In the interior CT image reconstruction method, the at least piecewise uniform or piecewise polynomial region identified within the ROI is obtained using the CT image obtained by the first stage reconstruction. It may be set manually by a person, or may be set by image processing using the CT image obtained by the reconstruction of the first stage. Alternatively, the ROI may be set in advance by at least one of specifying from a previously acquired CT image of the object to be imaged, and specifying from a model representing the structure of the object to be imaged and a priori information. wherein the at least piecewise uniform or piecewise polynomial region identified within the ROI using the CT image obtained by the approximate reconstruction of the first stage is a boundary of the ROI; It may be formed including a part.

In the interior CT image reconstruction method, the approximate reconstruction of the first stage is performed by a conventional CT image reconstruction method including filtered back projection (FBP) method, iterative approximation method, and statistical reconstruction method. Alternatively, the second stage reconstruction may be performed by CT including differential backprojection Hilbert transform, constrained successive approximation, and constrained statistical reconstruction. It may be performed by at least one of image reconstruction methods.

According to the present invention described above, the phase difference of X-rays passing through a sample object is imaged, the distribution of the phase difference of X-rays is quantitatively measured, and a precise analysis is performed with far less a priori knowledge than in the prior art. It is possible to provide an interior CT phase imaging XRM equipped with a more practical interior CT image reconstruction method capable of image reconstruction.

FIG. 3 is a diagram for explaining interior CT to which the present invention relates by comparison with normal CT. FIG. 10 is a diagram showing a case where conventional exact reconstruction is possible and an example where conventional exact reconstruction is possible; FIG. 4 is a diagram showing an example of typical artifacts that occur in interior CT; It is a figure explaining the X-ray optical system used by this invention. It is a figure explaining arrangement|positioning of each optical element in the X-ray optical system used by this invention. It is a figure explaining adjustment and switching of each optical element in two types of X-ray optical systems used in the present invention. It is a figure explaining the principal part composition of the interior CT phase imaging XRM apparatus of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a figure explaining the whole structure of the interior CT phase imaging XRM apparatus of this invention. It is a figure explaining the measuring method of the interior CT phase imaging XRM apparatus of this invention. FIG. 4 is a flow chart diagram showing details of the two-step image reconstruction method of the present invention; FIG. 3 shows an example of an imperfect image and an exact image in the two-stage image reconstruction method of the present invention; FIG. 10 is a diagram showing a case where conventional exact reconstruction is possible and an example where conventional exact reconstruction is possible; It is a figure explaining the piecewise uniformity and the piecewise polynomial referred to in the present invention. FIG. 10 is a diagram showing a specific example of setting an a priori information region B in a part of the ROI S in the a priori knowledge non-identifying image reconstruction method (Example 3); FIG. 10 is a diagram showing a specific simulation experiment example of the a priori knowledge identifying image reconstruction method and the a priori knowledge non-identifying image reconstruction method; FIG. 4 is a diagram for explaining a method of realizing mathematically strict image reconstruction of interior CT without using a priori information, which is the image reconstruction method of the present invention. FIG. 10 is a diagram for explaining a way of thinking when generalizing the image reconstruction method of the present invention; FIG. 10 is an explanatory diagram when the above image reconstruction method is applied to 180-degree parallel beam, fan beam short scan, and polygonal trajectory fan beam scan.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the accompanying drawings.
<Example 1>

First, as a basic idea of the present invention, an apparatus was constructed by adding an optical system switching mechanism to a commercially available XRM apparatus. Of the equipment, the X-ray source, sample mounting mechanism, and X-ray detector were fixed from the viewpoint of size and weight. It is also possible for these devices to be repositionable optical systems. In this embodiment, the distance between the X-ray source and the center of the sample is set to 500 mm, and the distance between the center of the sample and the X-ray detector/element surface is set to 760 mm, and optical systems with two different magnifications are incorporated.

FIG. 4(a) shows the arrangement of the optical elements of a Talbot-Lau interferometer having a magnifying optical system to which the FZP employed in the present invention is added. Here, the condensing mirror 2 has two focal points, one of which is the X-ray source 1 and the other of which is the sample 4. The inner surface of the condensing mirror 2 is formed into a spheroidal surface so that the X-rays can be totally reflected on the inner surface. there is The inside of the hollow glass tube may be formed with a single-layer metal film or a multi-layer film of two or more types of films with different densities. The X-rays transmitted through the G0 grating 3 and the sample 4 pass through the Fresnel zone plate (FZP) 5 and the G1 grating 6, enter the X-ray detector 7, and are recorded as an image. In the present invention, the FZP5 is provided with magnifications of 10x and 40x, so that the entire optical system can be switched.

In FIG. 4(b), the G0 grating 3 has a virtual X-ray source at the part of the grating that transmits X-rays, and the Talbot-Lau interferometer has an optical system in which many fine X-ray sources are superimposed. As a result, even if the X-ray source 1 has a certain area, it functions as a high spatial resolution microscope instead of a microfocus X-ray source. Since an X-ray source with high X-ray intensity has a finite area, the G0 grid 3 is necessary to obtain a high spatial resolution. be. In this example, an X-ray source with a diameter of 70 μm was used.

FIG. 5 shows the distances from the sample position of G0 gratings 3-1 and 3-2, FZPs 5-1 and 5-2, and G1 gratings 6-1 and 6-2 used in this example. The grating pitch of the G0 grating 3 was set to 3.0 μm for both low and high magnifications. The low-magnification FZP5-1 is x10, the high-magnification FZP5-2 is x40, the pitch of the low-magnification G1 grating 6-1 is 2.9 μm, and the pitch of the high-magnification G1 grating 6-2 is 2.4 μm. Table 1 summarizes the installation positions of the optical elements of this example. A method for determining the positions of these optical elements is described in detail in Non-Patent Document 10.

In this embodiment, a two-dimensional detector with a pixel size of 0.65 μm×0.65 μm and 1024 ch×1024 ch is used as the X-ray detector. The imaging area size and spatial resolution of this embodiment are determined by the detection area and pixel size of this detector. .65 μm/10=65 nm detection pixel size. Similarly, at high magnification (×40), the detection area is 16.4 μm (square) and the detection pixel size is 16.3 nm. As for the actual spatial resolution, it was affected by the accuracy of the finished optical elements, etc., and was about 150 nm at low magnification and about 50 nm at high magnification from the measurement of the standard sample.

FIG. 6 shows the configuration of the magnification exchange mechanism of the optical system in this embodiment. In FIG. 6, the X-rays passing through the G0 grating 3 pass through the sample 4 and then pass through the FZP5-1 and G1 grating 6-1, or pass through the FZP5-2 and G1 grating 6-2. As shown in Table 1 above, the FZP is used to determine the magnification of the microscope, and the G1 grating is used in combination to match it. Therefore, in the present invention, the FZP 5-1 and the G1 grating 6-1, and the FZP 5-2 and the G1 grating 6-2 are used in their respective combinations. there is

In FIG. 6, the G0 grating 3 is fixed to a mounting jig 31, and has a swivel 32 that can be rotated within the X vertical plane around the X-ray optical path X near the center of the G0 grating, a rotatable rotating mechanism 33, and the position of the G0 grating. can be adjusted in its position and mounting state by the X-ray optical path X and a moving mechanism 34 capable of moving in the vertical direction. It should be noted that the movement in the X direction of the X-ray optical path is provided with an externally controllable moving mechanism. The sample 4 is fixed to a sample mounting jig 41 and fixed to a sample XY moving table 42 . The sample XY moving table 42 is fixed to the sample rotating table 43 . The sample turntable 43 is rotatable about an axis perpendicular to the X-ray optical path X at an externally controllable rotation angle θ. In addition, the sample turntable 43 has a structure that allows Z-translation in the direction of the rotation axis.

After capturing the entire image of the sample at a low magnification, the position of the sample is moved by the Z translation function of the sample XY moving table 42 and the sample rotating table, and the desired specific region of the sample (Region of Interest, ROI) is viewed. It can be centered.

The FZPs 5-1 and 5-2 are each fixed to a mounting jig 51 capable of height adjustment Z, fixed to an FZP XY moving table 52, and mounted on an optical system switching table 8. FIG. As a result, the FZP 5-1 and FZP 5-2 are independently adjustable on the optical system switching table with respect to the X-ray optical path X direction and two axes YZ perpendicular thereto.

The G1 gratings 6-1 and 6-2 are each fixed to a mounting jig 61 capable of height adjustment Z, and a swivel 62 capable of rotation adjustment in the X vertical plane around the X-ray optical path X near the center of the G1 grating, It is mounted on the X-ray optical system switching table 8 after being fixed to the rotatable rotation mechanism 63 and the G1 grating XY moving table 64 . As a result, the G1 gratings 6-1 and G1 gratings 6-2 are independently placed on the X-ray optical system switching table 8, and the X-ray optical path X direction and two axes YZ perpendicular thereto and the plane perpendicular to the X-ray grating of the G1 gratings are arranged. It is adjustable for rotation in and about an axis perpendicular to the x-ray path X.

Adjustment items of FZPs 5-1 and 5-2 and G1 gratings 6-1 and 6-2 are respectively adjusted for X-ray optical paths of two different microscope magnifications. Some of these adjustments can be made manually, but it is desirable to allow them to be adjusted by external control. It should be noted that the vertical Y movement of the G1 gratings 6-1 and 6-2 with respect to the X-ray optical path must be externally controlled for the measurement of acquiring the phase image information.

As described above, the FZPs 5-1 and 5-2 and the G1 gratings 6-1 and 6-2 configured on the X-ray optical system switching table 8 constitute X-ray optical paths with two microscope magnifications. By moving the entire optical system switching table 8 with respect to the X-ray optical path X, it is possible to easily switch the XRM magnification. In this embodiment, the structure is such that two types of XRM magnification are switched, but it is also possible to switch between two or more types of optical systems. When the magnification is 1X, the X-ray optical path X does not have the FZP.

FIG. 7 shows a block diagram including the apparatus control system of this embodiment. The X-ray source 1 is controlled via the X-ray source controller 93, and the G0 grating adjustment mechanism, the sample position adjustment mechanism, the FZP position adjustment mechanism, the G1 position adjustment mechanism and the X-ray optical system switching table 8 are controlled for position adjustment as shown. Via device 94 , X-ray detector 9 is also connected to XRM controller 91 via X-ray detector controller 95 . These adjustment parameters and X-ray detector measurement data are stored in data storage device 92 .

FIG. 8 shows a block diagram of the entire microscope apparatus of this embodiment. The equipment shown in FIG. 7 constitutes an XRM installed on an optical bench 103 placed on a task 104 waiting for a vibration isolating function to prevent vibrations from the floor. The XRM is mounted in an X-ray shield and device exterior panel 101 with a temperature control device 102 to tightly control the temperature. The internal temperature was controlled to 0.1°C.
(Method of Acquiring Phase CT Image Data)

A transmission X-ray image of the sample 4 is obtained on the X-ray detector 8 by the phase-difference XRM apparatus of this embodiment described above. In order to capture a phase image, the G1 grating 6-1 or 6-2 used for imaging is moved at a pitch of 1/3 or less of the grating pitch in a direction perpendicular to the X-ray optical path X and perpendicular to the grating stripes. A transmitted image is captured each time the microscopic amount ΔY is moved (in the Y direction in this case). X-rays slightly refracted by the structure of the sample pass through the G1 grating, distorting the image of the grating, measuring it as image data with the X-ray detector 8 and recording it in the data storage device 92 . When taking a low-magnification (X10) X-ray microscopic image of this example, 0, 0.725 μm, 1.45 μm, 2, where 0.725 μm, which is 1/4 of the pitch (2.9 μm) of the G1 grating, is set as ΔY. Four transmission images are captured with a movement amount of 0.175 μm. Phase images are obtained from these four images using the method described in Non-Patent Document 11. At this time, not only a phase image but also an absorption image and a scattering image can be obtained.

In order to obtain a CT image, the transmission X-ray image pickup method described above is used, and G1 is moved by ΔY every time the rotation axis of the sample rotating table 43 shown in FIG. is imaged, and the CT data of the phase image for each sample rotation angle is recorded in the data storage device 92 . At this time, the absorption image and the scattering image can be simultaneously recorded as CT data for each rotation angle of the sample. In this measurement method, ΔY is kept constant, and the transmitted image is recorded by moving the angle by Δθ, and the transmitted image data is measured for each Δθ each time ΔY is moved, recorded in a data storage device, and all the data are measured. After completing the data processing, it is also possible to obtain CT data of a phase image, an absorption image, and a scattering image.
(Method of Acquiring Phase Interior CT Image Data)

Some samples 4 may include the need to perform particularly high spatial resolution data measurements. In the case of the apparatus of this embodiment, a detection area of 66 μm (square) and a detection pixel size of 65 nm at low magnification (×10) and a detection area of 16.4 μm (square) and 16.3 nm at high magnification (×40) detection pixel size. For example, this example corresponds to a case where the sample size is 50 μm in diameter and a 10 μm portion within it is desired to be measured with the highest spatial resolution CT. Assuming that the sample size is 50 μm in diameter, the low magnification XRM will be measured at a low magnification setting and a CT image will be acquired with a detection pixel size of 65 nm. Strict CT image reconstruction cannot be performed at high magnification settings.

The interior CT image reconstruction method of the present invention enables highly accurate image reconstruction even in such a case, and a method of acquiring data for phase interior CT image reconstruction will be described.

When measuring at a low magnification, the entire sample 4 is irradiated with X-rays as shown in FIG. . It is possible to acquire CT data under this condition and perform strict CT image reconstruction. On the other hand, when the same sample is imaged by XRM with a high magnification setting, the state shown in FIG. Although the conditions are such that omnidirectional X-ray transmission necessary for CT image reconstruction is performed, measurement data for other portions is insufficient, and accurate CT image reconstruction including the sample 41 portion cannot be performed. . Therefore, in the present invention, a method of performing precise CT image reconstruction of the sample 41 portion of FIG. developed. Measurement data obtained by low-magnification measurement or reconstructed CT images can be used as a priori information of the sample 4 whose structure is unknown.

Furthermore, according to this method, after acquiring the entire CT image data in FIG. 9(a), the measurement is performed with the settings shown in FIG. 9(b). 9A and 9B, the X-ray shutter 11 attached to the X-ray source 1 moves the X-ray shutter 11 from the X-ray optical path X only when it is necessary to acquire data, The sample 4 is made to be irradiated with X-rays. Furthermore, when acquiring data from only a part of the sample with a high magnification setting, the X-rays are irradiated only to the necessary part using the four-quadrant X-ray slit 12 or the X-ray pinhole. The advantage is that the overall X-ray exposure can be significantly reduced.
(Image reconstruction method for interior CT)

In the present invention, a technique for identifying a priori knowledge from the measured projection data (a priori knowledge identification type image reconstruction method) and a technique for identifying the a priori knowledge (a priori knowledge-based image reconstruction method), and for the peripheral part of the image (approximately applicable to any image) without identification. A fixed method (a priori knowledge non-identifying image reconstruction method) was developed.

FIG. 10 shows the processing flow of the two-stage image reconstruction method of the present invention. In the first step (S61), an imperfect image containing artifacts is generated using the conventional FBP method, iterative image reconstruction method, statistical image reconstruction method, etc. without a priori knowledge. Although this imperfect image contains artifacts, since artifacts generated in interior CT are low-frequency components, information on boundaries of tissues, structures, etc. is generated accurately in most cases. From the incomplete image shown in FIG. 11(a), the user manually or by image analysis (processing) using software identifies an arbitrary small region B within the ROI that can be used as a priori knowledge. (S62).

In the second step, using the region B obtained in the first step as a priori knowledge, a rigorous image reconstruction method (i.e., the first-stage reconstruction image reconstruction with higher precision than the original (S63). It should be noted that it is determined by what kind of a priori information area B could be extracted in the first step.

Here, the results obtained from the above incomplete images are summarized in the following four [results 1] to [results 4]. For example, if B is a constant value, [Result 1] is selected. ], and if B is close to the density variation of the piecewise polynomial, select [Result 4].

[Result 1 (constant a priori knowledge)]
As shown in FIG. 12(a), there is an arbitrary small a priori information region B inside the ROI, and it is known that the image f(x,y) has a constant value C (constant) in B. If there is, the solution for interior CT image reconstruction is uniquely determined. This result 1 is obtained by reducing a priori knowledge of the exact solution methods of Non-Patent Documents 5 and 6.

[Result 2 (polynomial prior knowledge)]
As shown in FIG. 12(a), there is an arbitrary small a priori information region B inside the ROI, and in B it is known that the image f(x,y) is an M-th order polynomial. If there is, the solution for interior CT image reconstruction is uniquely determined. Here, the polynomial means that the density change of the image f(x,y) has the following form.

However, the degree M of the polynomial must be known, and the coefficient a _mn of the polynomial may be unknown. [Result 1] is a stronger restriction of the form of the function in [Result 2] with the degree of the polynomial set to M=0, that is, [Result 2] uses the a priori knowledge of [Result 1]. It has become less.

[Result 3 (piecewise uniform a priori knowledge)]
As shown in FIG. 12(a), if there is an arbitrary small area B inside the ROI and it is known that the image f(x,y) is piecewise constant in B , the solution of the interior CT image reconstruction is uniquely determined. Here, the piecewise uniformity means that, as shown in _FIG . 13, B is composed of _a finite number ( _L ) of regions D ₁ , D ₂ , . , …, C _L . However, the number of regions _L and the constant values C ₁ , C ₂ , .

[Result 4 (piecewise polynomial a priori knowledge)]
As shown in FIG. 12(a), there is an arbitrary small region B inside the ROI, and in B the image f(x,y) is known to be an M-th order piecewise polynomial. If there is, the solution for interior CT image reconstruction is uniquely determined. Here, the _piecewise polynomial _{is an image f l} ( _{x ,} y ) has the following form.

However, the degree M of the polynomial must be known, the number of regions L and the polynomial coefficient a _mn ^(l) may be unknown, and [Result 4] is a priori knowledge of [Result 2] and [Result 3]. is reduced.

In this way, by theoretically considering the a priori knowledge about the object necessary for exact image reconstruction, the above-described conventional techniques (Non-Patent Documents 5 and 6, Patent Document 1) can be used. It enabled exact image reconstruction with much less a priori knowledge than empirical knowledge.

Non-Patent Documents 5 and 6 above use a priori knowledge of an arbitrary small region B within the same ROI as [Result 1] to [Result 4], but the value of the image f(x, y) itself is required. In contrast, in the present invention, [Result 1] to [Result 4] have much less a priori values such as (piecewise) uniform or (piecewise) polynomial Knowledge alone makes a big difference in a good way. Also, in Non-Patent Document 5, piecewise uniform type a priori knowledge is used, but it is necessary to make an unreasonable assumption that the entire ROI is piecewise uniform, not an arbitrary small region B in the ROI. different.

14A to 14C show examples of actual reconstruction using the a priori knowledge of [Result 1] to [Result 3]. In all cases, it can be seen that a small amount of a priori knowledge can achieve large artifact reductions.
<Example 2>

The invention described in the above example allowed far less a priori knowledge to be required for rigorous interior CT image reconstruction. However, in actual CT imaging, there are relatively few cases where a priori knowledge of the object of interest is known before imaging. Therefore, in the second embodiment, in the process flow shown in FIG. 13, the region B, which is an arbitrary small region within the ROI that can be used as a priori knowledge, is automatically identified at step S62. This automatic identification of the a priori information region can be realized, for example, by using image analysis (processing) techniques.
<Example 3>

The success or failure of the a priori knowledge automatic estimation type image reconstruction method of the first embodiment depends on whether or not the a priori knowledge (area B) can be successfully identified in the first step. Therefore, in the third embodiment, an "a priori knowledge non-identification type image reconstruction method" for reconstructing an image without identifying the region B is adopted. Since the assumption that the edge portion around the ROI is piecewise uniform or piecewise polynomial is not strictly true, the method remains an approximate image reconstruction method, but in many CT imaging situations Since [Knowledge 3] holds, by fixing and arranging the ROI around the edge, other approximations that do not use a priori knowledge can be obtained as shown in FIGS. Compared with the conventional image reconstruction method, the image can be reconstructed with much higher accuracy.

15A to 15F show specific simulation experiments of the a priori knowledge identifying image reconstruction method (Example 1) and the a priori knowledge non-identifying image reconstruction method (Example 3). An example is given by comparison with the prior art. A CT image of the chest was used in the experiment, and image reconstruction was performed with the heart portion located in the center as the ROI (see FIG. 15(a)).

In Example 1, the user viewed the image reconstructed by the FBP method in the first step, specified the a priori information region B within the ROIS, and performed strict image reconstruction in the second step. The a priori knowledge obtained is piecewise uniform in B corresponding to [Result 3]). The results are shown in FIG. 15(e). Further, in Example 3, image reconstruction was performed by fixing the a priori information region B to the periphery of the ROIs (the frame shape in FIG. 14(c)) (the a priori knowledge used was [result 3] is piecewise uniform in B). The results are shown in FIG. 15(f).

As a comparative example, FIG. 15B shows the results of the local FBP method in which the missing part of the projection data is extrapolated with a smooth function and the FBP method is applied. ) as a priori knowledge (Non - Patent Documents 5 and 6). FIG. 15(d) shows the results of the compression sensing method (Non-Patent Document 5) applying total variation (TV), which is a constraint of .

As is clear from these results, the local FBP method causes a strong cupping effect and significantly degrades the image, while the compression sensing method loses considerable detail and smooth density changes due to the effects of TV. On the other hand, it can be seen that the methods of Examples 1 and 3 of the present invention are able to reduce artifacts and reconstruct images fairly well.

[Image reconstruction method]
An image reconstruction method for generating an image from projection data based on the uniqueness of the solutions of [Result 1] to [Result 4] described above will be described. First, vectors obtained by discretizing an image f(x, y) and projection data p(r, θ) are represented by f and b, respectively, and a projection operation matrix that associates projection data with an image is represented by A. However, the image f includes not only the pixels in the ROIs but also all the pixels belonging to the object existence region in the cross section (care must be taken), and the projection data vector b is created by arranging all the measured values in a row. An evaluation function for evaluating whether or not the a priori knowledge in the a priori information area B is satisfied is represented by F(x). Image reconstruction can then be formulated as one of the following three optimization problems.

[Formulation 1]

[Formulation 2]

[Formulation 3]

However, x∈C represents a previously known constraint condition for the image, and the following is often used.
(a) (Support Constraint) The image f is nulled outside the a priori known support region Ω _OBJ .
(b) (Non-negative condition) Components of the image f do not take negative values.
(c) (Projection data value on the Hilbert straight line) In the image reconstruction method using the Hilbert transform, which will be described later, in order to compensate for the loss of information when rewriting Af=b to Hf=c, the Hilbert straight line L(u ) are used.

If F(x) is a function without a local minimum, called a convex function, there are many iterative or non-iterative methods for solving the above problem in the fields of mathematical optimization and image reconstruction. known and all of these techniques are available. For example, a constrained statistical image reconstruction method or a constrained successive approximation method can be used. As another class of image reconstruction method, it is possible to construct an image reconstruction method based on a framework called Differentiated Back Projection (DBP), which will be described later. Although the details of the DBP method will be described later, in the DBP method, the relational expression between the image f and the projection data b expressed by Af=b is not used as it is. ) of an imperfect image c, and is formulated as follows for image reconstruction. Methods of this class are called by names such as the differential backprojection + truncation Hilbert transform method (Non-Patent Documents 5 and 6, Patent Document 1).

[Formulation 4]

[Formulation 5]

[Formulation 6]

Of course, the problem formulated above is solved using iterative or non-iterative solutions.

Next, the evaluation function F(x) for evaluating a priori knowledge in the a priori information area B, which is the key to exact or accurate image reconstruction in the present invention, will be described.

First, in designing F(x), unlike the methods of Non-Patent Document 5 and Non-Patent Document 6 that impose a constraint condition based on a priori knowledge on the entire ROI S, the image reconstruction method of the present invention uses S A constraint condition is imposed only on the a priori information region B, which is an arbitrary small region in . In the case of [Result 1], the first derivative of f(x,y) is a priori knowledge that is zero in B , so the norm of the first derivative of f(x,y) in B is the minimum or minimize the variation (dispersion) of the concentration change in B. In the case of [Result 2], since it is a priori knowledge that the M+1 derivative of f(x,y) is zero in B, we either minimize the norm of the M+1 derivative, or Minimize the error between f(x, y) and f(x, y) ((x, y)∈B) with a function that fits an M-order polynomial. In the case of [Result 3], since f( _x ,y) is _a priori knowledge that becomes piecewise uniform in B , the total variation (TV: Total Variation) norm should be minimized. In the case of [Result 4], TV Minimize the norm.

An example of typical F(x) is shown in Table 2. However, in Table 2, the parameter p is the order of the norm, and its value can be 0 ≤ p ≤ 2 in [Result 1] and [Result 2], but in [Result 3] and [Result 4] it is uniform or M 0≤p≤1 should be used to avoid overestimating the influence of the boundary of a finite number (L) of subregions with density variations of the following polynomial. However, when 0≦p<1, F(x) does not become a convex function, so it is necessary to devise an iterative solution method or a non-iterative solution method, and it can be said that p=1 should be used. Numerical experiments were performed on the F(x) shown in Table 2 that had multiple candidates.

Furthermore, as described above, a method for realizing mathematically rigorous image reconstruction for interior CT using low-magnification overall image data without using a priori information about the object in the prior art will be described in detail below. describe.
<Example 4>

In order to ensure the uniqueness of the solution in the interior CT (that the solution of the image reconstruction problem is determined as one) and to enable strict image reconstruction, some additional information is required in addition to the interior CT projection data. is required. As an alternative to this, in the fourth embodiment, projection data that are not normally measured in interior CT, that is, extra projection data that does not pass through the region ROI Ω for which an image is to be obtained, are measured and this supplementary data is used. That is, an important feature of the exact solution method of the fourth embodiment of the present invention is to measure the minimum necessary extra projection data that does not pass through the ROIΩ.

In particular, by having an imaging optical system capable of changing the magnification of the specimen imaging, it is an effective method to image the entire specimen image and use supplementary data as extra projection data, part of which does not pass through the ROI Ω. Become.

The above method has generality that can be applied to various optical systems of microscopic CT such as electron beam, X-ray, and neutron beam. The principle of fan-beam CT that moves 360 degrees will be explained. As shown in FIG. 16A, consider the situation of interior CT in which projection data is measured by irradiating only electron beams passing through ROIΩ in 360-degree circular orbital fan-beam CT. Of course, a mathematically exact image reconstruction of the ROI Ω is impossible only with the interior CT projection data, and as shown in FIG. irradiate the entire object with an electron beam and measure the projection data of the entire object. By adding this partial overall projection data (hereinafter also simply referred to as "overall projection data") to the interior CT projection data, it is possible to reconstruct a strict ROI Ω, and an arbitrary, It has been mathematically proven that exact ROI Ω image reconstruction is possible if there is global projection data with a small arc segment rather than a single point.

In addition, for the above problem, the Differentiated Back Projection (DBP) method, which was used as a tool to show the uniqueness of the solution in the study of the interior CT rigorous solution method using a priori information on past objects, and the truncation Hilbert transform (Non-Patent Documents [5]-[9]) cannot prove it, but it can be proved by using a new image reconstruction method that introduces the truncation Hilbert transform into the filtering process in the FBP method. did. Here, the uniqueness of the solution to the interior CT image reconstruction problem in the 360-degree circular orbit fan-beam CT clarified by the present invention can be summarized as follows.

[Uniqueness of solution]
Given the interior CT projection data plus the full scan (without left and right truncation) projection data of any arc segment E (no matter how small), the interior CT image reconstruction solution is unique.

<Uniqueness of Solution in Obtained Interior CT>
Here, we will summarize the result of the uniqueness of the solution in interior CT image reconstruction, which was successfully proved based on a new image reconstruction method that introduces the truncation Hilbert transform into the filtering process of the FBP method, which will be described later. As definitions of symbols, the target image (object) is represented by f(x, y), and the ROI is represented by Ω.

First, based on FIG. 17, a concept for generalization will be described. When an arbitrary geometric system is used, the common point is that a set of line integral values on a straight line of an object is measured, and a relation of coordinate transformation is established between the projection data of both. Therefore, as shown in FIG. 17(a), projection data measured in an arbitrary geometric system are once coordinate-transformed into projection data of a virtual 360-degree circular orbit fan-beam CT, and the data are transformed into 360-degree projection data. Investigating whether the uniqueness condition in the case of circular orbital fan-beam CT is satisfied and deriving the uniqueness of the solution applicable to projection data measured in arbitrary geometric systems, we finally reach the following conclusions: be done.

[Uniqueness of solution (arbitrary geometric system)]
If the projection data are measured so that both of the following two conditions are satisfied, the image f(x, y) is uniquely defined in the ROI Ω, and exact reconstruction of Ω is possible.

(Condition 1) All projection data passing through the ROI Ω must be measured (interior CT measurement conditions).

(Condition 2) Projection data (partial total projection data) on all straight lines passing through an arbitrary small arc segment E outside the object (corresponding to a circle containing the object inside) is measured ( extra measurement of projection data to uniquely define the solution).

However, the length of the small arc segment E is a segment corresponding to a certain circle that contains the object inside, and it can be as short as possible as long as it is not at one point, and it can be at any location. The meaning of the above two geometric conditions is shown in FIG. 17(b).

In addition, three cases in which the above generalized solution uniqueness is useful are described below.

(a) Case of 180-degree parallel beam scanning Among CT data acquisition methods, consider the case where interior CT is performed by 180-degree parallel beam scanning, which is the most basic method for electron beam imaging. Projection data is represented by p(r, θ) (projection angle range −π/2≦θ<π/2) where r is the radius and θ is the angle. Now assume that in the angle range -ε≤θ≤ε (where ε is a small angle) the whole projection data (without truncation) is measured, and only the interior CT projection data is measured otherwise. At this time, if the small arc segment E is taken as shown in FIG. 18(a), it can be seen that the above-mentioned uniqueness condition of the solution is satisfied, and the ROI reconstruction solution is unique.

[Uniqueness of solution (180 degree parallel beam)]
In addition to the interior CT projection data in the angle range of -π/2 ≤ θ < π/2, if the overall projection data is measured in an arbitrary small angle range E (no matter how small), the image f(x, y ) is uniquely determined and exact reconstruction of Ω is possible.

This uniqueness can be interpreted as the limit of the uniqueness of the solution in the case of 360-degree circular orbit fan-beam CT, where the radius of the circular orbit is infinite. Similar solution uniqueness holds for beam CT. Note that this result can only be proved by using uniqueness generalized to arbitrary geometric systems.

(b) Case of fan beam short scan Consider the case of fan beam short scan shown in FIG. 18(b). The position of the X-ray source on the circular orbit is β ∈ [-π/2-α _max ,π/2+α _max ) (α _max is the overscan angle determined from the short scan conditions, Non-Patent Document [7]), Fan-beam projection data is represented by g(u, β), where u is the coordinate on the linear detector. Now suppose that the whole projection data (without truncation) is measured in the angular range -ε≤β≤ε (where ε is a small angle), and only the interior CT projection data is measured otherwise. At this time, if the small arc segment E is taken as shown in FIG. 21(b), it can be seen that the above-mentioned uniqueness condition of the solution is satisfied, and the solution of ROI reconstruction is unique.

[Uniqueness of solution (fan beam short scan)]
In addition to the interior CT projection data in the angle range of -π/2-α _max ≤ β < π/2 + α _max , if the entire projection data is measured in an arbitrary small angle range E (no matter how small), the ROI The image f(x,y) is uniquely defined in Ω, and exact reconstruction of Ω is possible.

(c) Fan-beam scan with polygonal trajectory:
Consider a fan beam scan using a regular pentagonal trajectory shown in FIG. 18(c). The position of the X-ray source on the regular pentagonal orbit is β∈[-π,π) (β is the azimuth of the point on the orbit from the center of the regular pentagon), and the coordinate on the linear detector is u. Data are represented by g(u,β). Now, in the angle range -ε≦β≦ε (ε is a small angle and determined so that one side of the regular pentagonal orbit is E), the whole projection data (without truncation) is measured, and the interior CT projection data is measured otherwise. only be measured.

At this time, if the small circular arc segment E is taken as shown in FIG. 18(c), it can be seen that the above-mentioned uniqueness condition of the solution is satisfied, and the ROI reconstruction solution is unique.

[Uniqueness of solution (fan beam scan using polygonal trajectory)]
In addition to the interior CT projection data in the angular range of -π≦β<π, if we measure the global projection data in an arbitrary small angular range E (no matter how small), the image f(x,y) at the ROI Ω is is uniquely determined and exact reconstruction of Ω is possible.

(b) Use Scout-View scan projection data for positioning as overall projection data A Scout-View scan is performed. Using this Scout-View scan function, low-magnification projection data can be used as the entire projection data of the partial trajectory E.

Next, the ROI Ω specified by the Scout-View scan is measured a second time, and by rotating the Ry axis, only the electron beam passing through the ROI is irradiated, and all the interior CT projection data is measured, and the image information data It is saved in the storage device 61 . This enables data acquisition for reconstructing an interior CT image.

Using the described image reconstruction method using all the measurement data of the partial data of the whole specimen projection of the first measurement and the second interior CT projection data performed on ROIΩ stored in the image information data storage device 61, Reconstruct interior CT images.

1... X-ray source, 2... X-ray focusing mirror, 3... G0 grating, 4... Sample, 5-1, 5-2... FZP, 6-1, 6-2... G1 grating, 7... X-ray detector 8... Optical element mounting table 11... X-ray shutter 12... Four-quadrant X-ray slit 31... G0 grating mounting jig 32... G0 grating swivel 33... G0 grating rotating mechanism 34... G0 grating moving mechanism 41 ... sample mounting jig, 42 ... sample XY moving table, 43 ... sample rotating table, 51 ... FZP mounting jig, 52 ... FZP XY moving table, 61 ... G1 grating mounting jig, 62 ... G1 grating swivel, 63 ... G1 grating rotating mechanism , 64... G1 grating XY moving table, 91... XRM controller, 92... Data storage device, 93... X-ray source controller, 94... Position adjustment controller, 95... X-ray detector controller, 101... Anti-X Wire shield and apparatus external panel, 102... temperature controller, 103... optical bench, 104... pedestal.

Claims (10)

a radiation source that emits radiation ;
a first grid interposed between the radiation source and the sample for shaping the radiation source into a grid;
a sample stage for rotating and holding the sample;
a radiation imaging optical system arranged behind the sample;
a second grating positioned behind the sample;
A radiation microscope apparatus comprising means for detecting a radiation image of the sample as an intensity distribution,
An interior CT phase imaging radiation microscope apparatus characterized by comprising a mechanism capable of simultaneously exchanging two or more sets of the radiation imaging optical system and the second grating in a plurality of combinations. 2. The interior CT phase imaging radiation microscope apparatus according to claim 1, further comprising a condenser lens for condensing radiation emitted from said radiation source between said radiation source and said first grating, The first grating installed between the optical lens and the sample has a mechanism capable of changing the position together with the operation of the mechanism capable of simultaneously exchanging the radiation imaging optical system and the second grating in a plurality of combinations. An interior CT phase imaging radiation microscope device characterized by: Acquire projection data by radiation that passes through a region inside the sample where measurement is desired, and use a plurality of projection data captured at each rotational step of a single axis that can be controlled to rotate, using a tomography (CT) image reconstruction method. A first stage of approximate reconstruction is performed, and based on the CT image obtained by this reconstruction, the image numerical value of the projection data representing the physical quantity is at least piecewise uniform within the region inside the sample where measurement is desired. or identifying a region A piecewise represented by a polynomial, and using the position of the region A and the property that the physical quantity is piecewise uniformly or piecewise represented by a polynomial in the region A, the first step 2. An interior CT image reconstruction method for performing three-dimensional measurement of a structure in a region desired to be measured inside the sample by performing a second-stage reconstruction that is more accurate than the reconstruction of . Fan-beam CT with a 360-degree circular orbit as the orbit of the radiation source, fan-beam short scan CT with the orbit of the radiation source as part of the circular orbit, or polygonal fan with a polygonal orbit as the orbit of the radiation source. At a low magnification that can acquire CT data of the entire sample as the first stage data for acquiring projection data by radiation passing through a region desired to be measured inside the sample in interior CT, such as beam CT , Irradiate radiation covering the entire sample from a segment of the trajectory that is not a single point, measure the entire projection data , and perform the first stage data and highly accurate second stage reconstruction. Three-dimensional measurement of the structure in the region desired to be measured inside the sample is performed by using together with projection data obtained by irradiating only the radiation passing through the region desired to be measured. Image reconstruction method for interior CT. In parallel beam scanning CT with a circular orbit as the orbit of the radiation source, CT data for the entire sample can be obtained as the first stage data for obtaining projection data using radiation that passes through a region desired to be measured inside the sample. At a low magnification, the radiation covering the entire sample is irradiated from a partial arc segment of the circular trajectory, not one point, and the entire projection data is measured, and the data of the first stage and the second stage with high accuracy Structure in the region desired to be measured inside the sample by using together with projection data obtained by irradiating only the radiation passing through the region desired to be measured for reconstructing the steps. An interior CT image reconstruction method for performing three-dimensional measurement of 6. The interior CT image reconstruction method according to any one of claims 3 to 5 , wherein the image value of the projection data includes a phase shift of the radiation due to the sample. The interior CT image reconstruction method according to any one of claims 3 to 5 , wherein the image values of the projection data include diffraction of the radiation by the sample. Inside the sample, the region A specified in the region desired to be measured is piecewise uniform or piecewise polynomial specified from the CT image obtained in the approximate reconstruction of the first step 6. A method for interior CT image reconstruction according to any one of claims 3 to 5, wherein a selection is made among a plurality of regions A represented. The region A specified in the region desired to be measured inside the sample can be manually set by the user using the CT image obtained by the approximate reconstruction of the first stage. The image reconstruction method for interior CT according to any one of claims 3 to 5 . The region A specified in the region where measurement inside the sample is desired using the CT image obtained by the approximate reconstruction in the first stage is within the region where measurement inside the sample is desired , identification from a CT image obtained in the past for the same or similar sample, a model representing the structure of the sample, and identification from a priori information. An interior CT image reconstruction method according to any one of claims 3 to 5 .

Images