WO2013051647A1

WO2013051647A1 - Radiography device and image processing method

Info

Publication number: WO2013051647A1
Application number: PCT/JP2012/075789
Authority: WO
Inventors: 拓司多田
Original assignee: 富士フイルム株式会社
Priority date: 2011-10-06
Filing date: 2012-10-04
Publication date: 2013-04-11
Also published as: JP2013090920A

Abstract

An objective of the present invention is to obtain a detailed phase differential image which has no noise. An x-ray radiography device (10) comprises an x-ray source (11), an x-ray detector (13), a lattice (12), a phase differential image generating unit (40), an unwrap processing unit (41), and a noise removal unit (43). The x-ray source (11) emits radiation toward a subject. The x-ray detector (13) detects the x-rays which are emitted from the x-ray source and generates image data. The lattice (12) is positioned between the x-ray source and the x-ray detector. The phase differential image generating unit (40) generates a phase differential image on the basis of image data (51, 52) which is obtained by the x-ray detector. The unwrap processing unit (41) carries out an unwrap process on the phase differential image. When subtracting an offset image which represents offset noise when a subject (H) is not present from a subject image which is obtained by photographing the subject (H), the noise removal unit (43) removes a trend which is noise which remains in the subject image from the subject image.

Description

Radiation imaging apparatus and image processing method

The present invention relates to a radiation imaging apparatus and an image processing method for detecting an image based on a phase change of radiation by a subject.

Radiation, such as X-rays, has a characteristic that it is absorbed and attenuated depending on the weight (atomic number) of the elements constituting the substance and the density and thickness of the substance. Focusing on this characteristic, X-rays are used as a probe for seeing through the inside of a subject in fields such as medical diagnosis and nondestructive inspection.

In a general X-ray imaging apparatus, an object is placed between an X-ray source that emits X-rays and an X-ray image detector that detects X-rays, and X-rays transmitted through the object are imaged. Do. In this case, the X-rays emitted from the X-ray source toward the X-ray image detector are absorbed and attenuated when passing through the subject, and then enter the X-ray image detector. An image based on the X-ray intensity change by the subject is detected by the X-ray image detector.

Since the X-ray absorption ability is lower as the element has a smaller atomic number, there is a problem that the change in the intensity of X-ray is small and sufficient contrast cannot be obtained in the soft tissue or soft material. For example, most of the components of the cartilage portion constituting the joint of the human body and the joint fluid in the vicinity thereof are water, and the difference in the X-ray absorption capacity between them is small, so that it is difficult to obtain contrast.

Against the background of such problems, research on X-ray phase imaging for obtaining an image based on the phase change of the X-ray by the subject instead of the change in the intensity of the X-ray by the subject has been actively conducted in recent years. X-ray phase imaging is a method of imaging the X-ray phase change based on the fact that the phase change of the X-ray incident on the subject is larger than the intensity change. A high-contrast image can be obtained. As one type of X-ray phase imaging, an X-ray imaging apparatus that detects an X-ray phase change by configuring an X-ray Talbot interferometer using two diffraction gratings and an X-ray image detector is known. (See, for example, WO 2004/058070).

In this X-ray imaging apparatus, the first diffraction grating is disposed behind the subject as viewed from the X-ray source, and the second diffraction grating is disposed at a position separated from the first diffraction grating by the Talbot distance. An X-ray image detector is arranged behind. The Talbot distance is the distance at which the X-rays that have passed through the first diffraction grating form a self-image (striped image) of the first diffraction grating due to the Talbot effect, and the grating pitch of the first diffraction grating and the X-ray wavelength Depends on and. This self-image is modulated by refraction caused by the phase change of X-rays in the subject. By detecting this modulation amount, the phase change of the X-ray is imaged.

The fringe scanning method is known as a method for detecting the modulation amount. In the fringe scanning method, the second diffraction grating is scanned with respect to the first diffraction grating in a direction parallel to the plane of the first diffraction grating and perpendicular to the grating line direction of the first diffraction grating. X-rays are radiated from the X-ray source at each scanning position while being translated (scanned) at a pitch, and the X-ray image passing through the subject and the first and second diffraction gratings is imaged by the X-ray image detector. Is the method. For the pixel value of each pixel obtained by this X-ray image detector, the phase shift amount (phase difference from the initial position when the subject does not exist) is calculated for the signal (intensity modulation signal) representing the change due to the scanning. By doing so, an image related to the modulation amount is obtained. This image is an image reflecting the refractive index of the subject, and corresponds to the differential amount of the X-ray phase change (phase shift), and is called a phase differential image.

As shown in WO2004 / 058070, the amount of phase shift is calculated using a function (arg [...]) for extracting the argument of a complex number or an arctangent function (tan ^-1 [...]). . For this reason, the phase differential image is expressed by a value convolved (wrapped) in the range of the function (−π to + π or −π / 2 to + π / 2). In the phase differential image wrapped in this way, discontinuous points may occur at locations where the upper limit of the range changes to the lower limit or at locations where the lower limit changes to the upper limit. It is known that unwrap processing is performed as described above (see, for example, JP 2011-045655 A).

The unwrap process is performed in order along a predetermined path from a predetermined position in the image as a starting point. When the discontinuous point is detected in the path, a value corresponding to the range of the function is uniformly added to or subtracted from data after the discontinuous point. Thereby, discontinuous points are eliminated and data is continuous.

Also, noise is superimposed on the phase differential image, resulting in image unevenness, which may hinder the observation of the subject. Noise occurs due to various causes depending on the shooting conditions, such as the placement error of each grid or radiation source, the temperature environment at the time of shooting, etc., but if the shooting conditions such as the placement error and temperature environment are almost the same Regardless of the presence or absence of the subject, it appears as noise in the same manner. For this reason, it can be obtained by imaging the subject by subtracting the phase differential image obtained by imaging in the same environment and without the subject from the phase differential image obtained by imaging the subject. It is known to remove noise components from the phase differential image (see WO 2004/058070, JP 2011-045655 A).

Previously, measures were taken only to remove noise components generated by the shooting environment. However, even after subtracting the phase differential image obtained by imaging in the absence of the subject from the phase differential image obtained by imaging the subject, noise components that are not related to the subject image (hereinafter referred to as trend) It was found through experiments that the This trend is different from the noise component removed by subtraction of the phase differential image captured in the absence of the subject described above in that the trend changes with each imaging.

It is an object of the present invention to provide a radiation imaging apparatus and an image processing method capable of removing a noise component that changes almost every time imaging is performed together with a noise component that is substantially unique to the radiation imaging apparatus.

The radiation imaging apparatus of the present invention includes a radiation source, a radiation detector, a phase differential image generation unit, and a noise removal unit. The radiation source emits radiation toward the subject. The radiation detector detects radiation emitted from a radiation source and generates image data. The phase differential image generation unit generates a phase differential image based on the grating unit arranged between the radiation source and the radiation detector and the image data obtained by the radiation detector. The noise removing unit remains when an unwrap processing unit that performs unwrap processing on the phase differential image and an offset image that represents offset noise when there is no subject are subtracted from the subject image that is the phase differential image after unwrap processing. To remove the trend.

The noise removal unit preferably includes an offset noise removal unit, a trend detection unit, and a trend removal unit. The offset noise removing unit removes the offset noise by subtracting the offset image from the subject image. The trend detection unit detects the trend based on the subject image from which the offset noise has been removed. The trend removal unit subtracts and removes the trend detected by the trend detection unit from the subject image.

The trend detection unit extracts a pixel value of the subject image from which the offset noise has been removed along a predetermined direction, and detects a predetermined direction component of the trend. Then, the trend removing unit generates a trend component image in which a predetermined direction component of the trend is arranged along a direction perpendicular to the predetermined direction, and subtracts the trend component image from the subject image, thereby detecting the trend from the subject image. A predetermined direction component is removed. In this way, it is preferable that the trend detection unit and the trend removal unit detect and remove trend components in two directions, ie, a first direction that is a predetermined direction and a second direction that is perpendicular to the predetermined direction. The subject image has a rectangular shape, and the first direction and the second direction are preferably a horizontal direction or a vertical direction along one side of the square.

The trend detection unit may detect a trend component based on at least the pixel value of the unexposed region where there is no subject among the extracted pixel values of the subject image.

The trend detection unit extracts the pixel value of the subject image along the first direction or the second direction, and then smoothes at least the subject region where the subject is present, whereby the first direction component of the trend or the second direction A direction component may be extracted.

Further, the trend detection unit extracts the pixel value of the subject image along the first direction or the second direction, and then interpolates and extracts the data of the subject region where the subject is present based on the data of the missing region. It is preferable to extract the first direction component or the second direction component of the trend by inserting or fitting.

The trend detection unit detects a first direction component or a second direction component of the trend for a plurality of rows or columns along the first direction or the second direction, and the first direction component or the first direction in each row or each column. Data obtained by averaging the two direction components may be detected as the first direction component or the second direction component.

The trend detection unit preferably sets at least one of the first direction and the second direction to a direction not passing through the subject. In addition, it is possible to provide an index that designates the placement of the subject and designates the placement of the subject so that a missing region without the subject appears in the phase differential image.

In the case of including an absorption differential image generation unit that generates an absorption image representing the absorption rate of radiation based on the image data obtained by the radiation detector and differentiates the absorption image to generate an absorption differential image, the trend detection unit is In the absorption differential image, a plurality of pixels whose pixel values are substantially equal to the surrounding pixels are selected, and the trend is detected based on the pixel values of the pixels of the phase differential image at positions corresponding to the selected pixels. May be.

The absorption differential image generation unit generates a first absorption image from the image data obtained when there is no subject, generates a second absorption image from the image data obtained when there is a subject, and generates the second absorption image as the first absorption image. It is preferable to generate an absorption differential image by differentiating the third absorption image obtained by dividing by one absorption image. In addition, it is preferable that the pixel selected by the trend detection unit is a pixel in a blank area where there is no subject.

The grating unit generates a second periodic pattern image by partially shielding the first periodic pattern image and a first grating that generates a first periodic pattern image by passing radiation from a radiation source. It is preferable that the radiation image detector includes the second grating and generates the image data by detecting the second periodic pattern image.

The grating unit includes a scanning mechanism that moves the first grating or the second grating at a predetermined scanning pitch and sequentially sets the plurality of scanning positions. The radiological image detector has a second periodic pattern at each scanning position. It is preferable that an image is detected to generate image data, and the phase differential image generation unit generates a phase differential image based on a plurality of image data generated by the radiation image detector.

The moving direction of the first lattice or the second lattice is a direction perpendicular to the lattice line. The moving direction of the first grating or the second grating may be a direction inclined with respect to the grating line.

The phase differential image generation unit may generate a phase differential image based on single image data obtained by the radiation detector.

The first grating is an absorption grating, and it is preferable to generate the first periodic pattern image by geometrically optically projecting the incident radiation. The first grating may be an absorption type grating or a phase type grating, and may generate a first periodic pattern image by causing a Talbot effect to incident radiation.

It is preferable to provide a multi-slit that partially blocks the radiation emitted from the radiation source and disperses the focal point.

The image processing method of the present invention includes an offset image generation step, a subject image generation step, and a noise removal step. In the offset image generation step, phase differentiation is performed regardless of the presence or absence of the subject, based on image data obtained by placing a grating portion between the radiation source and the radiation detector and performing imaging without the subject. An offset image that is a phase differential image in which offset noise superimposed on the image is projected is generated. In the object image generation step, an object image, which is a phase differential image, is generated based on image data obtained by imaging the object by arranging a grating portion between the radiation source and the radiation detector. . In the noise removal step, a trend that is noise remaining in the subject image when the offset image is subtracted from the subject image is removed from the subject image.

According to the present invention, it is possible to obtain a fine phase differential image by removing a noise component (trend) that changes every time imaging is performed together with a noise component (offset noise) that is substantially unique to the radiation imaging apparatus.

It is a block diagram which shows an X-ray imaging apparatus. It is a schematic diagram which shows the structure of a X-ray image detector. It is explanatory drawing explaining the structure of the 1st and 2nd grating | lattice. It is a graph which shows an intensity | strength modulation signal. It is a block diagram which shows the structure of an image process part. It is explanatory drawing which shows the aspect of an unwrap process. It is a flowchart which shows the process at the time of pre imaging | photography. It is a flowchart which shows the process at the time of this imaging | photography. It is explanatory drawing which shows a mode that a moire arises in pre imaging | photography image data. It is explanatory drawing which shows the aspect in which a moire appears as noise in a phase differential image. It is explanatory drawing which shows the aspect of offset noise. It is explanatory drawing which shows the aspect of the various images produced | generated at the time of this imaging | photography. It is explanatory drawing which shows the aspect in which a trend remains by the subtraction of an offset image. It is explanatory drawing which shows the aspect which detects the horizontal direction component of a trend. It is explanatory drawing which shows the example of a horizontal direction trend image. It is explanatory drawing which shows a mode that the horizontal direction component of a trend is removed. It is explanatory drawing which shows the aspect which detects the vertical direction component of a trend. It is explanatory drawing which shows the example of a vertical direction trend image. It is explanatory drawing which shows a mode that the vertical direction component of a trend is removed. It is a figure which shows the example of the phase differential image in which the subject was projected on the whole. It is explanatory drawing which shows the trend which has a periodic characteristic. It is a figure which shows the example of the parameter | index for restrict | limiting the arrangement | positioning of a subject. It is explanatory drawing which shows the example which provides a pseudo | simulation absorber in an element | child omission area | region. It is a block diagram which shows the image process part of 2nd Embodiment. It is explanatory drawing which shows the trend detection method of 2nd Embodiment. It is a block diagram which shows the aspect of the noise removal part of 3rd Embodiment. It is a flowchart which shows the image processing of 3rd Embodiment. It is a flowchart which shows the image processing of the main imaging | photography of 3rd Embodiment. It is explanatory drawing explaining the X-ray imaging apparatus which has a multi slit.

[First Embodiment]
In FIG. 1, an X-ray imaging apparatus 10 includes an X-ray source 11, a grating unit 12, an X-ray image detector 13, a memory 14, an image processing unit 15, an image recording unit 16, an imaging control unit 17, a console 18, and a system. A control unit 19 is provided. The X-ray source 11 includes, for example, a rotary anode type X-ray tube and a collimator that limits the X-ray irradiation field, and emits X-rays toward the subject H based on the control of the imaging control unit 17. To do.

The grating unit 12 includes a first grating 21, a second grating 22, and a scanning mechanism 23. The first and

second gratings

21 and 22 are disposed to face the X-ray source 11 in the Z direction, which is the X-ray irradiation direction. A subject H is disposed between the X-ray source 11 and the first grating 21. The X-ray image detector 13 is, for example, a flat panel detector using a semiconductor circuit, and is disposed behind the second grating 22 so that the detection surface 13a is orthogonal to the Z direction.

1st grating | lattice 21 is an absorption-type grating | lattice arrange | positioned orthogonally to Z direction, and is provided with the some X-ray absorption part 21a and X-ray transmissive part 21b which were extended | stretched in the Y direction on the grating | lattice surface. The X-ray absorbing portions 21a and the X-ray transmitting portions 21b are alternately arranged in the X direction and form a striped pattern. The second grating 22 is also an absorption-type grating, and includes a plurality of X-ray absorbing portions 22a and X-ray transmitting portions 22b that are extended in the Y direction and arranged alternately in the X direction, like the first grating 21. ing. The

X-ray absorbing portions

21a and 22a are formed of a material having X-ray absorption properties such as gold (Au) and platinum (Pt). The X-ray transmissive

portions

21b and 22b are formed of a material having X-ray permeability such as silicon (Si) or resin or a gap.

The first grating 21 partially passes the X-rays emitted from the X-ray source 11 to generate a first periodic pattern image (hereinafter referred to as a G1 image). The second grating 22 partially transmits the G1 image generated by the first grating 21 to generate a second periodic pattern image (hereinafter referred to as G2 image). When the subject H is not arranged, the G1 image substantially matches the lattice pattern of the second lattice 22.

The X-ray image detector 13 detects the G2 image and generates image data. The memory 14 temporarily stores the image data read from the X-ray image detector 13. The image processing unit 15 generates a phase differential image based on the image data stored in the memory 14, and generates a phase contrast image based on the phase differential image. The image recording unit 16 records a phase differential image and a phase contrast image.

The scanning mechanism 23 intermittently moves the second grating 22 in the X direction, and sequentially changes the relative position of the second grating 22 with respect to the first grating 21. The scanning mechanism 23 includes a piezoelectric actuator or an electrostatic actuator, and is driven based on the control of the imaging control unit 17 so as to detect an X-ray image at each scanning position (position after each intermittent movement). The X-ray image detector 13 detects an X-ray image while the intermittent movement of the second grating 22 is stopped, and image data of the X-ray image is stored in the memory 14. Instead of moving the second grating 22 intermittently, the second grating 22 may be moved continuously, and an X-ray image may be detected every time it moves a predetermined distance.

The console 18 includes an operation unit 18a and a monitor 18b. The operation unit 18a includes a keyboard, a mouse, and the like, and sets operation conditions such as setting of imaging conditions such as tube voltage, tube current, and irradiation time of the X-ray source 11, mode selection for main imaging or pre-imaging, and imaging execution instruction. Is possible. The main imaging is an imaging mode performed with the subject H placed between the X-ray source 11 and the first grating 21. Pre-imaging is an imaging mode performed without placing the subject H between the X-ray source 11 and the first grating 21. The pre-photographing is performed in order to acquire a background component caused by a manufacturing error or an arrangement error of the first and

second gratings

21 and 22 as an offset image.

The monitor 18b displays photographing information such as photographing conditions, and a phase differential image and a phase contrast image recorded in the image recording unit 16. The system control unit 19 comprehensively controls each unit according to a signal input from the operation unit 18a.

In FIG. 2, an X-ray image detector 13 includes a pixel electrode 31 that collects charges generated in a semiconductor film (not shown) by incident X-rays, and a TFT (Thin for reading charges collected by the pixel electrode 31). A plurality of pixel portions 30 having a film-transistor (32) 32 are arranged in a two-dimensional manner. The semiconductor film is made of amorphous selenium, for example.

The X-ray image detector 13 includes a gate scanning line 33, a scanning circuit 34, a signal line 35, and a readout circuit 36. The gate scanning line 33 is provided for each row of the pixel unit 30. The scanning circuit 34 applies a scanning signal for turning on / off the TFT 32 to each gate scanning line 33. The signal line 35 is provided for each column of the pixel unit 30. The readout circuit 36 reads out electric charges from the pixel unit 30 through the signal lines 35, converts them into image data, and outputs them. The detailed layer configuration of each pixel unit 30 is the same as the layer configuration described in Japanese Patent Application Laid-Open No. 2002-26300, for example.

The readout circuit 36 includes an integration amplifier, an A / D converter, a correction circuit (all not shown), and the like. The integrating amplifier integrates the charges output from each pixel unit 30 via the signal line 35 to generate an image signal. The A / D converter converts the image signal generated by the integrating amplifier into digital image data. The correction circuit performs dark current correction, gain correction, linearity correction, and the like on the image data. The corrected image data is stored in the memory 14.

The X-ray image detector 13 is not limited to the direct conversion type in which incident X-rays are directly converted into electric charges with a semiconductor film, and the incident X-rays are visible with a scintillator such as cesium iodide (CsI) or gadolinium oxysulfide (GOS). It may be an indirect conversion type that converts light into light and converts visible light into electric charge with a photodiode. Further, the X-ray image detector 13 may be configured by combining a scintillator and a CMOS sensor.

In FIG. 3, X-rays irradiated from the X-ray source 11 are cone beams having the X-ray focal point 11a as a light emitting point. The 1st grating | lattice 21 is comprised so that the Talbot effect may not arise and the X-rays which passed X-ray transmissive part 21b may be projected geometrically. Specifically, the width of the X-ray transmission part 21b in the X direction is set to a value sufficiently larger than the peak wavelength of X-rays irradiated from the X-ray source 11, and most of the X-rays are diffracted by the X-ray transmission part 21b. It is realized by not doing. When tungsten is used as the rotating anode of the X-ray source 11 and the tube voltage is 50 kV, the peak wavelength of the X-ray is about 0.4 mm. In this case, the width of the X-ray transmission part 21b may be about 1 to 10 μm.

Thus, the G1 image is always a self-image of the first grating 21 regardless of the distance from the first grating 21 downstream in the Z direction. The G1 image is enlarged in proportion to the distance from the X-ray focal point 11a to the downstream in the Z direction.

As described above, the grating pitch p ₂ of the second grating 22 is set so that the grating pattern of the second grating 22 matches the G1 image at the position of the second grating 22. Specifically, the grating pitch _{p 2} of the second grating 22, the distance _{L 1} between the grating pitch _p 1, X-ray focal point 11a and the first grating 21 of the first grating 21, the first grating 21 and the distance L ₂ between the second grating 22 is set to equation (1) so as to satisfy substantially. Hereinafter, the coordinates in the X, Y, and Z directions are x, y, and z.

The G1 image is modulated by refracting the subject H due to a phase change in the X-ray. This modulation amount reflects the X-ray refraction angle φ (x) of the subject H. The figure illustrates an X-ray path that is refracted in accordance with a phase shift distribution Φ (x) representing a phase change of the X-ray in the subject H. Reference numeral X1 indicates a path along which the X-ray goes straight when the subject H does not exist, and reference numeral X2 indicates an X-ray path refracted by the subject H.

The phase shift distribution Φ (x) is expressed by the equation (2) where λ is the wavelength of the X-ray and n (x, z) is the refractive index distribution of the subject H.

The refraction angle φ (x) is in the relationship of the phase shift distribution Φ (x) and the equation (3).

At the position of the second grating 22, the X-ray is displaced in the X direction by an amount corresponding to the refraction angle φ (x). This amount of displacement Δx is approximately expressed by equation (4) based on the small X-ray refraction angle φ (x).

Thus, the displacement amount Δx is proportional to the differential value of the phase shift distribution Φ (x). Therefore, by detecting the displacement amount Δx by fringe scanning, which will be described later, a differential value of the phase shift distribution Φ (x) is obtained, and a phase differential image is generated.

When a value obtained by dividing the grating pitch p ₂ into M pieces (p ₂ / M) is used as a scanning pitch, the scanning mechanism 23 causes the second grating 22 to be intermittently moved at this scanning pitch. Each time the second grating 22 is moved intermittently, X-rays are emitted from the X-ray source 11 and a G2 image is captured by the X-ray image detector 13. M is an integer greater than or equal to 3, for example, it is preferable that M = 5. Stripe scanning is performed by the intermittent movement of the second grating 22 and the imaging of the X-ray image detector 13.

When the expression (1) is not satisfied slightly, or when rotation around the Z direction or slight inclination with respect to the XY plane occurs between the first grating 21 and the second grating 22, the G2 image Moiré fringes occur. The moire fringes are moved along with the intermittent movement of the second grating 22, the movement distance in the X-direction coincides with the original moiré fringe reaches the grating pitch p _2. By confirming the movement of the moire fringes, the amount of intermittent movement of the second grating 22 can be verified.

M pixel values are obtained for each pixel unit 30 of the X-ray image detector 13 by the fringe scanning. As shown in FIG. 4, the M pixel values I _k periodically change with respect to the scanning position “k” of the second grating 22. The scanning position “k” is each stop position during the intermittent movement of the second grating 22. A signal indicating a change in the pixel value I _k with respect to the scanning position “k” is referred to as an intensity modulation signal.

The broken line in the figure indicates an intensity modulation signal obtained without the subject H being placed. On the other hand, a solid line indicates an intensity modulation signal in which the phase difference amount ψ (x) is generated by the subject H in a state where the subject H is arranged. This phase shift amount ψ (x) is in the relationship of the displacement amount Δx and the equation (5).

Therefore, for each pixel unit 30, a phase differential image is obtained by obtaining the phase shift amount ψ (x) of the intensity modulation signal based on the M pixel values I _k obtained by the fringe scanning.

Next, a method for calculating the phase shift amount ψ (x) will be described. The intensity modulation signal is generally expressed by Equation (6).

Here, A ₀ represents the average intensity of the incident X-ray, A _n represents the amplitude of the intensity-modulated signal. “N” is a positive integer and “i” is an imaginary unit. As shown in FIG. 4, when the intensity modulation signal draws a sine wave, n = 1.

In this embodiment, since the scanning pitch (p ₂ / M) is constant, Expression (7) is established.

When equation (7) is applied to equation (6), phase shift amount ψ (x) is represented by equation (8).

Here, arg [...] is a function that extracts the argument of a complex number. Further, the phase shift amount ψ (x) can also be expressed as an equation (9) using an arctangent function.

Since the deviation angle of the complex number ranges from −π to + π, when the phase shift amount ψ (x) is calculated based on the equation (8), the phase shift amount ψ (x) is −π Take a value that is convolved (wrapped) in the range from to + π. On the other hand, the arc tangent function usually has a range of −π / 2 to + π / 2. Therefore, when the phase shift amount ψ (x) is calculated based on Equation (9), the phase shift The deviation amount ψ (x) takes a value convolved in the range of −π / 2 to + π / 2. In Expression (9), the value range can be changed from −π to + π by discriminating the denominator and the numerator in the arctangent function, so that the phase shift amount ψ (x ) Can also be calculated.

In this embodiment, data having the pixel value as the phase shift amount ψ (x) calculated for each pixel unit 30 is referred to as a phase differential image. Note that an image represented by data obtained by multiplying or adding a phase shift amount ψ (x) by a constant may be a phase differential image. Hereinafter, it is assumed that the pixel values of the phase differential image are wrapped in a predetermined value range having a width α (for example, a range from 0 to α).

As shown in FIG. 5, the image processing unit 15 includes a phase differential image generation unit 40, an unwrap processing unit 41, an offset image storage unit 42, a noise removal unit 43, a phase contrast image generation unit 44, and the like.

The phase differential image generation unit 40 uses M pieces of image data (pre-photographed image data) 51 obtained by the X-ray image detector 13 by pre-imaging fringe scanning, and is based on the formula (8) or the formula (9). The phase differential image is generated by performing the calculation. Similarly, the phase differential image generation unit 40 generates a phase differential image based on the M pieces of image data (main captured image data) 52 obtained by the X-ray image detector 13 by the main scanning fringe scanning. The phase differential image generated by the phase differential image generation unit 40 is input to the unwrap processing unit 41.

The unwrap processing unit 41 performs unwrap processing on the phase differential image input from the phase differential image generation unit 40. As shown in FIG. 6, in the unwrap processing, a point where the pixel value of the phase differential image is largely changed (so-called phase jump) along a predetermined path is wrapped as a discontinuous point DP. This is a process for eliminating the discontinuous point DP by adding or subtracting the value range width α to the pixel values after the detected discontinuous point DP and making the change in the pixel value almost continuous. The detection of the discontinuous point DP is performed by obtaining a location where the change amount of the pixel value is a predetermined amount (for example, α / 2) or more.

The unwrap processing unit 41 sets a plurality of starting points for pixels located at the end of each row or each column of the phase differential image, and follows a predetermined path (for example, a straight path along the row or column) from a certain starting point. The unwrap processing is performed, and then the unwrap processing of the starting point adjacent to the starting point where the unwrapping processing is performed is performed, and the unwrap processing is performed along the route from the adjacent starting point in order while changing the starting point and the route. By repeating, the whole phase differential image is unwrapped.

The unwrap processing unit 41 performs an unwrap process on the phase differential image generated from the pre-captured image data 51 during pre-photographing, and stores it in the offset image storage unit 43 as an offset image. The offset image is generated due to distortion of the first grating 21 or the second grating 22, a slight positional deviation (including rotation or inclination), a slight arrangement error that occurs when the second grating 22 is scanned, and the like. A noise component (hereinafter referred to as offset noise) is projected. If the temperature environment of the X-ray source 11 at the time of actual imaging, the temperature environment of the X-ray detector 13 according to the number of times of imaging and the contact state with the subject are substantially the same as those at the time of pre-imaging, the phase obtained at the time of actual imaging Almost the same noise component as the offset noise projected on the offset image is also superimposed on the differential image. The offset image storage unit 43 stores the newly input offset image after erasing the already stored offset image when a new pre-photographing is performed and a new offset image is input. . In addition, at the time of actual photographing, the unwrap processing unit 41 performs unwrap processing on the phase differential image generated from the main captured image data 52 and inputs the processed image to the noise removing unit 43.

The noise removing unit 43 removes noise from the phase differential image obtained at the time of actual photographing. The noise removing unit 43 includes an offset removing unit 46, a trend detecting unit 47, and a trend removing unit 48.

The offset removing unit 46 removes offset noise from the phase differential image obtained at the time of actual photographing. More specifically, the offset removing unit 46 acquires an offset image from the offset image storage unit 42 when a phase differential image is input from the unwrap processing unit 41 during the main photographing. Then, a phase differential image from which offset noise has been removed is generated by subtracting the offset image from the input phase differential image. The phase differential image from which the offset noise has been removed is input to the trend detection unit 47.

The trend detection unit 47 and the trend removal unit 48 constitute a trend detection / removal unit that detects and removes the remaining trend from the phase differential image from which the offset noise has been removed. The trend detection unit 47 detects a noise component in a predetermined direction from the phase differential image from which the offset noise has been removed, and the trend removal unit 48 generates a trend image representing the trend detected by the trend detection unit. Then, the trend removing unit 48 generates a phase differential image from which the trend is removed by subtracting the trend image from the phase differential image from which the offset noise has been removed, and inputs the phase differential image to the phase contrast image generating unit 44.

More specifically, when the phase differential image is input, the trend detection unit 47 detects a noise component superimposed in the first direction of the phase differential image (for example, the horizontal direction (X direction) of the phase differential image). To do. The trend removing unit 48 generates a first trend image in which noise components in the first direction are uniformly arranged in a second direction (for example, the vertical direction (Y direction) of the phase differential image) orthogonal to the first direction. Subtract from the phase differential image. Thereby, the first direction component of the trend is removed from the phase differential image.

Next, the trend detection unit 47 detects the noise component superimposed in the second direction of the phase differential image from which the first direction component of the trend has been removed. Then, the trend removal unit 48 generates a second trend image in which the noise components in the second direction are uniformly arranged in the first direction, and further subtracts the phase trend image from which the first direction component of the trend has been removed. . Thereby, the trend is removed from the phase differential image.

The phase contrast image generation unit 44 integrates the phase differential image from which the offset noise and the trend have been removed by the noise removal unit 43 along the X direction to generate a phase contrast image representing the phase shift distribution. The phase differential image from which the trend is removed and the phase contrast image are recorded in the image recording unit 16.

Hereinafter, the operation of the X-ray imaging apparatus 10 will be described. When imaging the subject H using the X-ray imaging apparatus 10, pre-imaging is performed before imaging the subject H as shown in FIG. When the pre-shooting mode is selected as the shooting mode using the operation unit 18a (step S10), the camera enters a shooting instruction input standby state (step S11). Thereafter, when an imaging instruction is input using the operation unit 18a, the second grating 22 is translated by a predetermined scanning pitch by the scanning mechanism 23, and at each scanning position “k”, the X-ray from the X-ray source 11 is obtained. Radiation and detection of the G2 image by the X-ray image detector 13 are performed (step S12). As a result of the fringe scanning, M pieces of pre-captured image data 51 are generated and stored in the memory 14.

The pre-captured image data 51 is read by the image processing unit 15. In the image processing unit 15, a phase differential image is generated from the pre-captured image data 51 by the phase differential image generation unit 40 (step S13). This phase differential image is unwrapped by the unwrap processing unit 41 (step S14) and then stored in the offset image storage unit 42 as an offset image. The pre-photographing operation ends here. Note that this pre-imaging may be performed at least once in a state in which the subject H is not disposed when the X-ray imaging apparatus 10 is started up, and need not be performed every time before the main imaging. In addition, although an example in which pre-photographing is performed in order to generate an offset image in advance before main photographing has been described here, photographing for generating an offset image may be performed after main photographing.

Next, as shown in FIG. 8, the subject H is placed and the main imaging is performed. When performing the main shooting, the main shooting mode is selected using the operation unit 18a (step S20). When the main shooting mode is selected, a shooting instruction standby state is set (step S21). When a shooting instruction is given using the operation unit 18a, stripe scanning is performed (step S22), and M main captured image data 52 are stored in the memory 14.

Thereafter, the actual captured image data 52 is read out to the image processing unit 15. In the image processing unit 15, a phase differential image is generated from the actual captured image data 52 by the phase differential image generation unit 40 (step S23), and unwrap processing is performed by the unwrap processing unit 41 (step S24).

The phase differential image that has been subjected to the unwrapping process is input to the noise removing unit 43. In the noise removing unit 43, first, the offset noise is removed by subtracting the offset image from the phase differential image in the offset removing unit 46 (step S25).

Next, the trend is detected as a remaining noise component by the trend detection unit 47 (step S26), and the trend is removed by the trend removal unit 48 (step S27).

In FIG. 8, for the sake of simplicity, trend detection (step S28) and trend removal (step 27) are performed once, but trend detection and removal are performed as necessary until the trend is removed. It may be performed multiple times. For example, the X-ray imaging apparatus 10 performs trend detection and removal at least twice. Specifically, the X-ray imaging apparatus 10 detects and removes the trend X-direction component, and then detects and removes the trend Y-direction component.

Thus, the phase differential image from which the offset noise and the trend are removed is recorded in the image recording unit 16. At the same time, the phase contrast image generation unit 44 integrates the phase differential image from which the offset noise and the trend are removed to generate a phase contrast image representing the phase shift distribution (step S28), and is recorded in the image recording unit 16. The Thereafter, the phase differential image and the phase contrast image from which the offset noise and the trend have been removed are displayed on the monitor 18b (step S29).

As described above, the X-ray imaging apparatus 10 removes the offset noise by subtracting the offset image from the phase differential image generated in the main imaging, and further detects a trend that is a noise component remaining after the offset noise is removed. And remove. For this reason, according to the X-ray imaging apparatus 10, it is possible to obtain a fine phase differential image in which noise is further reduced as compared with the case where offset noise is simply removed. Further, since a fine phase differential image is obtained, the phase contrast image obtained by integrating the phase differential image is also low noise.

Hereinafter, the processing mode for removing the cause of the trend and the trend will be described more specifically. First, pre-photographed image data 51 is obtained when pre-photographing is performed, but since pre-photographing is performed in the absence of the subject H, almost nothing is projected in the pre-photographed image data 51 originally. . However, for example, as shown in FIG. 9, moire 56 may occur in the pre-captured image data 51. Moire 56, for example, Come to extract the pixel value I _k on line indicated by a broken line on the pre-photographed image data 51, a striped noise increases and decreases periodically in the X-direction. Further, when comparing the M pre-captured image data 51 by the stripe scanning, the moire 56 gradually moves in the X direction according to the scanning position “k”.

The moire 56 is generated by distortion of the first grating 21 or the second grating 22, a slight misalignment (including rotation or inclination), and a slight arrangement error that occurs when the second grating 22 is scanned. The distortion and slight misalignment of the first grating 21 and the second grating 22 may change depending on the temperature environment according to the number of imaging and the contact state with the subject. Similar moire 56 may occur due to expansion / contraction of the lattice 22 due to temperature changes.

As shown in FIG. 10, when the phase differential image 57 is generated based on the pre-captured image data 51 in which the moiré 56 is generated, the moiré 56 is, for example, ½ times the moire cycle on the original pre-captured image data 51. It is converted into noise 58 having a period. The noise 58 caused by the moire 56 is a so-called sawtooth wave having the range width α as the upper limit / lower limit. When the phase differential image 57 is unwrapped, the sawtooth noise 58 is superimposed on the offset image 59 as an offset noise 61 in which the pixel value ψ monotonously increases in the X direction, as shown in FIG. Note that the period of the noise 58 is ½ times the moire period because the phase differential image 57 is generated using an arctangent function whose value range represented by the equation (9) is −π / 2 to + π / 2. This is the case. On the other hand, when a declination extraction function having a range represented by equation (8) of −π to + π is used, or when the range of the arctangent function is expanded from −π to + π, the noise of the phase differential image 57 The period of 58 is the same as the moire period.

Similarly, as shown in FIG. 12A, moiré 62 may be superimposed on the captured image data 52 obtained by performing the actual capturing at the same time as the subject H is projected. When a phase differential image is generated based on the actual captured image data 52 in which the moire 62 is generated in this way, the moire 62 is a sawtooth wave having a period of 1/2 times as shown in the phase differential image 63 shown in FIG. When the unwrap processing is further performed, an offset noise 66 is obtained as in the phase differential image 65 (subject image) shown in FIG.

If the distortion and slight positional deviation of the first grating 21 and the second grating 22 are substantially the same during pre-photographing and actual photographing, the offset noise 61 on the offset image 59 and the actual photographing are The offset noise 66 on the obtained phase differential image 65 is roughly the same image. For this reason, if the offset image 59 is subtracted from the phase differential image 65, the offset noise 66 is removed from the phase differential image 65.

However, the distortion and slight misalignment of the first grating 21 and the second grating 22 are hardly the same in pre-shooting and main shooting, and the moire 56 and the book in pre-shooting are almost the same. When compared with the moire 62 at the time of shooting, it is normal that the period, the inclination in the image, etc. are slightly different. Therefore, there is a slight difference between the offset

noises

61 and 66 in which the

moires

56 and 62 are reflected. For this reason, as shown in FIG. 13, the phase differential image 71 obtained by subtracting the offset image 59 from the phase differential image 56 obtained at the time of the main photographing includes the offset noise 61 at the time of the pre-photographing and the offset at the time of the main photographing. A noise component due to the difference in the noise 66 remains as the trend 72. In FIG. 13, for the sake of convenience, the shading indicating the offset

noises

61 and 66 and the shading showing the trend 72 are drawn to the same extent, but the trend 72 has a size of about 1/10 of the offset

noises

61 and 66, for example. It is a noise component.

The trend detection unit 47 and the trend removal unit 48 detect and remove the trend 72 from the phase differential image 71 in which the trend 72 remains, as will be described below.

When the phase differential image 71 is input from the offset removal unit 46, the trend detection unit 47 detects a horizontal direction (X direction) component of the trend 72. Specifically, as illustrated in FIG. 14, the trend detection unit 47 extracts the pixel value ψ of the phase differential image 71 for each row. For example, the pixel value ψ in the row indicated by the broken line changes in the pixel value ψ according to the structure of the subject H in the region where the subject H is present (hereinafter referred to as subject region) E1, for example, but there is no subject H. In the areas (hereinafter referred to as “prime-missing areas”) E2a and E2b, the trend 72 is reflected to change almost uniformly. For this reason, the trend detection unit 47 smoothes the extracted pixel value ψ. As a result, the regions E2a and E2b that are unclear are not substantially changed as compared with those before the smoothing, but the subject region E1 is data that substantially matches the tendency of the trend 72 as shown by the broken line by the smoothing. As a horizontal component of the trend 72 that changes uniformly.

When the trend detection unit 47 generates the horizontal component of the trend 72 in each row by extracting and smoothing the pixel value ψ of each row as described above, the trend detection unit 47 averages the horizontal component of the trend in each row (hereinafter, referred to as “trend”). Horizontal direction trend data). The trend detection unit 47 detects the horizontal trend data as a horizontal component of the trend 72 and inputs it to the trend removal unit 48 together with the phase differential image 71.

As shown in FIG. 15, the trend removing unit 48 generates an image 76 (hereinafter referred to as a horizontal trend image) in which the horizontal trend data 72 detected by the trend detecting unit 47 is arranged in the vertical direction (Y direction). . Next, as shown in FIG. 16, the horizontal trend image 76 is subtracted from the original phase differential image 71 to generate a phase differential image 77 from which the horizontal component of the trend 72 is removed. The trend 78 is still superimposed on the phase differential image 77. The trend 78 is a vertical direction (Y direction) component of the trend 72 since the horizontal direction component is removed from the trend 72 of the original phase differential image 71. When the phase differential image 77 from which the horizontal component of the trend 72 is removed is generated in this manner, the trend removal unit 48 inputs the generated phase differential image 77 to the trend detection unit 47 again.

When the phase differential image 77 is input from the trend removal unit 48, the trend detection unit 47 detects a trend 78 (vertical component of the trend 72) from the phase differential image 77. Specifically, first, as shown in FIG. 17, the pixel value ψ of each column of the phase differential image 77 is extracted, and the trend 78 of each column is detected. For example, as illustrated in FIG. 17A, when the pixel value ψ of the column of the blank region is extracted as the “a” column, the extracted pixel value ψ is almost the trend 78 itself. On the other hand, as shown in FIG. 17B, when the pixel value ψ of the column on the subject H as shown in the “b” column is extracted, the trend 78 and the subject H are superimposed, so the extracted pixel value ψ Has a certain tendency reflecting the trend 78, but changes reflecting the subject H.

Therefore, the trend detection unit 47 smoothes the data regardless of whether or not the subject H is on the extracted column. When the pixel values ψ in each column are each smoothed in this way, in the case of a missing region as in the “a” column, the data mode does not change substantially before and after the smoothing process, but as in the “b” column. In a column where the subject H is present, the information of the subject H is almost lost (reduced), and the data has a tendency similar to the trend 78 as indicated by a one-dot chain line.

The trend detection unit 47 generates data (hereinafter, referred to as vertical trend data) obtained by averaging the trend 78 detected for each column as described above for all columns. The trend detection unit 47 detects this vertical trend data as the vertical component of the original trend 72 and inputs it to the trend removal unit 48.

18, the trend removal unit 48 generates an image (hereinafter referred to as a vertical trend image) in which the vertical trend data input from the trend detection unit 78 is arranged in the horizontal direction. Next, as shown in FIG. 19, the trend removing unit 48 generates a phase differential image 83 by subtracting the vertical direction trend image 81 from the phase differential image 77. The phase differential image 83 generated in this way is a phase differential image from which all the trends have been removed by removing both the horizontal direction component and the vertical direction component of the trend 72 from the phase differential image 71.

In the X-ray imaging apparatus 10, the phase differential image 83 from which all the trends have been removed is recorded in the image recording unit 16. The phase contrast image generation unit 44 generates a phase contrast image from the phase differential image 83.

In the first embodiment described above, the subject H is partially projected in the central portion of the phase differential image 71. However, as shown in FIG. Even when is projected, the X-ray imaging apparatus 10 can detect and remove the trend 72 suitably. This is because, when each direction component of the trend 72 is detected and removed, data obtained by averaging each direction component of the trend 72 detected for each row (column) is detected as each direction component of the trend 72. This is because it is difficult to be affected by the presence of H. In FIG. 20, the subject H is composed of

bone parts

20a and 20b and soft tissue 20c, and the upper left area of the phase differential image 71 is substantially occupied by the bone part 20a, and the lower right area is substantially occupied by the bone part 20b. A portion between the

bone portions

20a and 20b is the soft tissue 20c, and an example in which the phase differential image 71 does not have a blank region is shown.

In the first embodiment described above, after the offset noise 66 is removed, the horizontal component of the trend 72 is detected and removed, and thereafter the vertical component of the trend 72 is detected and removed. The removal mode is not limited to this. For example, the horizontal component of the trend 72 may be detected and removed after the vertical component of the trend 72 has been detected and removed first.

In the first embodiment described above, when detecting and removing the trend 72, the vertical component and the horizontal component are detected and removed separately. However, if the trend 72 can be detected and removed, It is not restricted to an aspect. For example, the first direction component of the trend 72 and the second direction are defined as an oblique direction inclined by a predetermined angle with respect to the vertical and horizontal sides of the phase differential image 71 as a first direction and a direction perpendicular to the first direction as a second direction. Components may be detected and removed. Also in this case, the phase differential image 83 from which the trend 72 is removed can be obtained as in the first embodiment described above.

Furthermore, when detecting and removing the trend 72, when the detection and removal are performed separately in two directions other than the vertical and horizontal directions of the phase differential image 71, a subject detection unit (not shown) is further provided in the noise removal unit 43, and the subject is detected. When the region where H is projected is detected and the region of the subject H has directionality, the direction along the direction of the region of the subject H is the first direction, and the direction perpendicular thereto is the second direction. It is also good.

For example, when the subject H is projected in an oblique 45 degree direction of the phase differential image, the oblique 135 degree direction perpendicular to the subject H is the first direction, and the oblique 45 degree direction along the subject H is the first direction. Two directions are assumed. This increases the number of rows that detect the component of the trend 72 so that it does not overlap the subject H in at least one direction (second direction), so that the detection accuracy of the trend 72 in this direction (second direction) is higher. improves. Further, in the above-described phase differential image 71, the subject H is projected along the vertical direction at approximately the center of the phase differential image 71. In this case, the horizontal direction is the first direction, and the vertical direction is The second direction. As described above, when a vertical component is detected (see FIG. 17), the pixel value ψ extracted along the vertical direction in the missing region is the trend 78 itself, so that every column crosses the subject H. The detection and removal of the vertical component is more accurate than the case of detecting and removing the horizontal component. In this way, the aspect of selecting the trend component detection direction is particularly suitable for detecting each direction component of the trend 72 except for data of a location where the subject H exists when detecting each direction component of the trend 72. It is.

In the first embodiment described above, each direction of the trend 72 is detected regardless of the presence or absence of the subject H by smoothing the extracted pixel value ψ when the vertical and horizontal direction components of the trend 72 are detected. However, except for data on the subject area (for example, data on the area E1 in FIG. 14), the trend 72 is detected based on the data on the missing areas (for example, areas E2a and E2b in FIG. 14). Each direction component may be detected. At this time, the method of calculating the data of the removed subject region portion is arbitrary. Data that simply connects the boundary point between the subject region and the missing region (for example, the boundary point between the region E1 and the regions E2a and E2b in FIG. 14) with a straight line may be calculated, or an arbitrary straight line, spline, other curve, etc. Thus, it may be calculated by interpolation, extrapolation, fitting, or the like. The subject region may be detected by providing a subject region detecting unit for detecting the subject region in the noise removing unit 43, or by detecting the subject region with the trend detecting unit 47. Also good. The subject area can be detected based on, for example, the amount of change in the pixel value ψ.

In the first embodiment described above, for the sake of simplicity, an example has been described in which the trend 72 has a tendency to increase (decrease) linearly in the phase differential image 71. However, like the trend 86 shown in FIG. , May have periodic features. Even in such a case, the X-ray imaging apparatus 10 can suitably detect and remove the trend. However, when the trend is detected by interpolation, extrapolation, fitting, or the like based on the data of the missing regions E2a and E2b except for the data of the subject region E1 as described above, a curve using a trigonometric function or the like is used. It is necessary to perform interpolation, extrapolation, fitting, etc.

Note that, as in the above-described modification, when the trend 72 is detected, if the data of only the blank region that does not cross the subject region can be used, the detection accuracy of the trend 72 is further improved. For this reason, it is preferable to make it easy to create a blank area in the phase differential image 71.

In order to make it easy to create a blank area in the phase differential image 71, the following may be performed. For example, as shown in FIG. 22, it is recommended to place the subject H on the front surface 91a (see FIG. 23) of the housing 91 in which the first and

second gratings

21 and 22 and the X-ray image detector 13 are housed. An index 92 indicating the range is provided. The recommended range indicated by the index 92 is set smaller than the size of the detection surface 13a of the X-ray detector 13. In this case, if the subject H is arranged in the index 92 according to the index 92, the difference area between the detection surface 13a and the index 92 becomes a blank area. In addition, even when the size of the subject H exceeds the range indicated by the index 92, it is easy to form a blank area in the difference area between the detection surface 13a and the index 92. In FIG. 22, the index 92 is provided at the center of the detection surface 13a. However, the index 92 may be provided so as to be biased in any direction, up, down, left, or right with respect to the detection surface 13a.

In addition, as described above, in order to make it easy to form a missing region in the phase differential image 71, when the arrangement of the subject H is limited by the index 92, the same amount of X-rays as the subject H is placed in the missing region. It is preferable to arrange a pseudo-absorber having an absorption rate. For example, as shown in FIG. 23, the pseudo-absorber 96 is disposed on the front surface of the detection surface 13a corresponding to the blank regions F2a and F2b. In this way, it is possible to prevent the pixel values ψ corresponding to the missing areas F2a and F2b from being saturated and the trend 72 from being detected. In FIG. 23, the pseudo absorber 96 is disposed in the X-ray detector 13, but the pseudo absorber 96 may be arbitrarily disposed as long as it is disposed at a position corresponding to the element missing regions F2a and F2b. is there. Further, the pseudo absorber 96 may be provided between the second grating 22 and the X-ray detection panel 13 or between the first and

second gratings

21 and 22, for example. Furthermore, the pseudo-absorber 96 may be arranged on the front surface 92a of the housing 92 (or inside the housing 92 of the front surface 92a) so as to surround the outer periphery of the index 92.

In the first embodiment described above, the pixel value ψ of the phase differential image is extracted for each row, the horizontal component of the trend in each row is detected, and the horizontal trend data obtained by averaging these is obtained as the horizontal trend data of the trend 72. It is detected as a direction component. However, for example, the horizontal component of the trend 72 detected in a certain line may be used as the horizontal component of the trend 72 of the entire phase differential image. Further, data obtained by averaging the horizontal components of the trend 72 detected in some rows, not the entire phase differential image, may be used as the horizontal component of the trend 72. For example, the horizontal component of the trend 72 detected in the line of the missing region is used as the horizontal component of the trend 72 as the entire phase differential image, or only the horizontal component of the trend 72 detected in the row of the missing region is used. When the averaged data is used as the horizontal component of the trend 72 as the entire phase differential image, the detection accuracy of the horizontal component of the trend 72 is improved as compared with the case where the data detected in the row across the subject region is used. The same applies to the detection of the vertical component. Further, as described above, the same applies to the case where the trend 72 components are detected in the first direction and the second direction that are not parallel to the sides of the phase differential image.

In the first embodiment described above, the trend detection unit 47 extracts the pixel value ψ of the phase differential image 71 for each row, and then smoothes the extracted pixel value ψ to detect the lateral component of the trend 72. Further, the horizontal component of the trend in each row is averaged to detect one horizontal trend data, but the smoothing of the extracted pixel value ψ may be omitted. That is, the trend detection unit 47 may average the extracted pixel values ψ themselves to obtain horizontal trend data without smoothing the pixel values ψ extracted for each row. This is effective when the signal of the subject H becomes dull and the data becomes almost trend only by averaging the pixel values between the rows.

Further, only the above-described smoothing may be performed, and averaging of the horizontal component of each row may be omitted. In this case, for example, the pixel value ψ of an arbitrary row is smoothed, and the smoothed pixel value ψ of this row is input to the trend removing unit 48 as lateral trend data. This is effective when the signal of the subject H becomes dull and the data becomes almost trend only by smoothing the pixel value ψ.

Furthermore, both the above smoothing and averaging may be omitted. For example, the trend detection unit 47 may extract a pixel value ψ in an arbitrary row of the phase differential image 71 and input the pixel value ψ in that row as it is to the trend removal unit 48 as horizontal trend data. This is effective, for example, when the trend 72 is larger than the signal of the subject H.

Of course, when detecting vertical trend data, smoothing or averaging may be omitted as described above, or both smoothing and averaging may be omitted. The same applies to each embodiment described later.

In the first embodiment described above, when the trend detection unit 47 detects horizontal trend data or vertical trend data, the trend detection unit 47 does not consider the size of the pixel value ψ. It is preferable to compare the pixel value ψ of the row (or column) with a predetermined threshold value, and based on the magnitude relationship, only the portion where the pixel value ψ is in the predetermined range is used for detecting trend data in each direction.

For example, since the pseudo-absorber 96 is not arranged, the pixel region where the pixel value ψ is saturated and the metal portion of the fixing jig of the subject H are reflected and the pixel value is extremely small (for example, approximately 0). The pixel value ψ (phase differential value) is not a normal value. For this reason, if trend data is detected in consideration of even pixel values ψ in a region where these pixel values ψ are extremely large or regions where pixel values ψ are extremely small, they are detected by being affected by these inappropriate values. Trend data may be incorrect. Accordingly, a first threshold value that determines the lower limit of the pixel value ψ to be used and a second threshold value that determines the upper limit of the pixel value ψ to be used are determined in advance, and the trend detection unit 47 extracts the pixel value ψ of each row or each column. It is preferable that the trend data in each direction is detected using only data in a range that falls within the range from the first threshold value to the second threshold value.

[Second Embodiment]
In the first embodiment described above, the horizontal and vertical components of the trend 72 are detected by extracting the pixel value ψ from the phase differential image 71 for each row and column and smoothing. The mode of detecting each direction component of the trend 72 is not limited to this. For example, each direction component of the trend 72 may be detected as described below as the second embodiment.

As shown in FIG. 24, the X-ray imaging apparatus of the second embodiment includes an absorption differential image generation unit 101 in the image processing unit 100. First, the absorption differential image generation unit 101 acquires M main captured image data 52 obtained by fringe scanning, and averages them to generate an absorption image. The absorption image is an image in which the X-ray absorption rate (transmittance) of the subject H is expressed as contrast.

In each of the actual captured image data 52, stripes by the first grating 21 and the second grating 22 are projected, and the stripes move according to the scanning position “k” of the second grating 22. If M pieces of actual captured image data 52 are compared in the order of photographing, they are moved by one cycle. For this reason, when the M actual captured image data 52 are averaged, the fringes by the first grating 21 and the second grating 22 are averaged to form a substantially uniform background, and thus there is no fringe in the absorption image. Only the image of the subject H is projected.

Next, the absorption differential image generation unit 101 generates an absorption differential image by differentiating the absorption image. Differentiation of the absorption image can be performed, for example, by taking a difference from an image shifted by one pixel in a predetermined direction (direction of differentiation). The absorption differential image generated by the absorption differential image generation unit 101 in this manner is input to the trend detection unit 102 and used when the trend 72 is detected. In addition, although the direction which differentiates an absorption image is arbitrary, for example, it is preferable to carry out to the X direction (refer FIG. 1) perpendicular | vertical to a grid line.

The trend detection unit 102 detects each direction component of the trend 72 from the phase differential image from which the offset noise has been removed, as with the trend detection unit 47 of the first embodiment, but the trend detection unit 102 refers to the absorption differential image. Meanwhile, each direction component of the trend 72 is detected.

As shown in FIG. 25A, it is assumed that the captured screen is 103 and the subject H is in the center of the captured screen 103. Here, for simplicity, the subject H is spherical. At this time, when a pixel value corresponding to a one-dot chain line crossing the center of the photographing screen 103 in FIG. 25A is extracted from the absorption image, as shown in FIG. 25B, a solid line 105a is displayed when there is no offset noise. As described above, the pixel value is flat in the portion of the unexposed region 104, and a curve or the like corresponding to the X-ray absorption rate is formed in the portion of the subject H. On the other hand, when there is offset noise, as indicated by the broken line 105b, the pixel value changes in the background region 104 reflecting the offset noise. Further, the curve shape also changes in the portion of the subject H due to the offset noise. However, the boundary position between the subject H and the blank area 104 is not changed by the offset noise.

Therefore, as shown by a solid line 106a (when there is no offset noise) and a broken line 106b (when there is offset noise) in FIG. 25C, in the absorption differential image, an area E2a and an area E2b corresponding to the missing region 104 In this case, the data is almost flat regardless of the presence or absence of offset noise.

The trend detection unit 102 detects the horizontal direction component of the trend 72 using the data of the area E2a and the area E2b corresponding to the blank area 104 as described above. Specifically, first, at least two pixels are selected from the region E2a or the region E2b where the pixel value is flat in the absorption differential image. Here, the pixel A1 with the region E2a and the pixel A2 with the region E2b are selected, but two points may be selected from either the region E2a or the region E2b. Further, three or more pixels may be selected, and if there are three or more regions where the change in the pixel value is flat corresponding to the blank region 104, one point may be selected from each region.

In this way, when the pixels A1 and A2 corresponding to the blank region 104 are selected, the trend detection unit 102 selects the pixel A1 selected as the absorption differential image from the phase differential image that has been subjected to the offset processing, as shown in FIG. The trend component 107 is detected by calculating a straight line connecting the pixel values of the pixels A1 ′ and A2 ′ corresponding to the pixels A1 and A2 (pixels at the same positions as the pixels A1 and A2). When three or more pixels are selected, the trend component 107 is detected by fitting or the like.

Here, although the horizontal direction component of the trend 72 is detected by the trend detection unit 102, the same applies to the case where the vertical direction component of the trend 72 is detected. Moreover, the aspect which removes the trend 72 from a phase differential image using the detected trend component 107 in the trend removal part 48 is the same as that of 1st Embodiment mentioned above.

When only the pixel values of the phase differential image are used, it may be difficult to clearly identify the background region 104 depending on the aspect of the subject H, and the extracted pixel values are smoothed as in the first embodiment. However, as described above, if the absorption differential image is used, the pixels A1 ′ and A2 ′ of the blank region 104 can be easily specified, and each direction component of the trend 72 can be detected easily and easily. Can be done quickly.

In the second embodiment described above, the absorption differential image generation unit 101 generates an absorption differential image from the actual captured image data 52, but the absorption differential image is preferably generated as follows. First, the absorption differential image generation unit 101 generates a first absorption image from the pre-photographed image data 51 at the time of pre-photographing, and stores it in the offset image storage unit 42, for example. Next, at the time of actual imaging, the absorption differential image 101 generates a second absorption image from the actual captured image data 52. Then, a third absorption image is generated by dividing the second absorption image by the first absorption image, and an absorption differential image to be input to the trend detecting unit 102 is generated by differentiating the third absorption image. In this way, noise components due to the uneven transmittance of the first grating 21 and the second grating 22 are reduced, so that an image more suitable as an absorption differential image used for specifying the missing region 104 in the trend detection unit 102 can be obtained. Become.

In the second embodiment described above, the pixels of the missing region 104 are selected from the absorption differential image. However, the trend detection unit 102 is not limited to the pixels of the missing region 104, but the entire absorption differential image (covered). It is only necessary to select a pixel whose pixel value is substantially equal to that of the surrounding pixels from the region of the specimen H). In the second embodiment described above, since the subject H is spherical, there are almost no places where the change in the pixel value is flat inside the subject H. However, in the case of a normal subject, This is because there is almost no change in the pixel value, and there may occur a region where the pixel value is substantially equal to the surrounding pixels. If the pixel value is substantially equal to the surrounding pixels, the trend can be removed even in the subject H as in the second embodiment. Further, as is clear from the fact that the pixels in the subject H can be used, even when the subject H is present in the entire imaging range 10, the trend is detected by the same method as in the second embodiment described above. Can do.

[Third Embodiment]
In the first and second embodiments described above, the trend 72 is removed after the offset noise 66 is removed, but the order in which the offset noise 66 and the trend 72 are removed is not limited to this. Next, with reference to FIGS. 26 and 27, a third embodiment of the present invention will be described. In the third embodiment, an image processing device 120 is used. In addition, the same code | symbol is attached | subjected to the member etc. which are common in the above-mentioned 1st Embodiment, and the description is abbreviate | omitted.

As shown in FIG. 26, when removing the trend 72 first, the noise removing unit 43 is configured by a trend removing unit 121 (first noise removing unit) and an offset removing unit 122 (second noise removing unit).

When the phase differential image that has been subjected to the unwrap processing is input from the unwrap processing unit 41, the trend removing unit 121 performs Fourier transform on the phase differential image to remove low-frequency components that are equal to or lower than a predetermined frequency. Thereafter, inverse Fourier transform is performed to generate a phase differential image from which low frequency components are removed. In the case of pre-shooting, the trend removing unit 121 stores the phase differential image from which the low frequency component has been removed in the offset image storage unit 42 as an offset image. In the case of actual photographing, the trend removing unit 121 inputs a phase differential image from which low frequency components are removed to the offset removing unit 122.

The offset removing unit 122 acquires an offset image from the offset image storage unit 42 when a phase differential image from which low frequency components are removed is input during the main photographing. Then, the offset noise is removed from the phase differential image by subtracting the offset image from the phase differential image obtained by the actual photographing.

The X-ray imaging apparatus configured as described above performs pre-imaging as shown in FIG. First, when the pre-shooting mode is selected as the shooting mode using the operation unit 18a (step S50), a shooting instruction input standby state is set (step S51). Thereafter, when an imaging instruction is input using the operation unit 18a, the X-ray source 11 performs X at each scanning position “k” while the second grating 22 is intermittently moved by a predetermined scanning pitch by the scanning mechanism 23. The G2 image is detected by the beam irradiation and the X-ray image detector 13 (step S52). As a result of the fringe scanning, M pieces of pre-captured image data 51 are generated and stored in the memory 14. Then, the pre-captured image data 51 is read out to the image processing unit 15, and a phase differential image is generated from the pre-captured image data 51 by the phase differential image generation unit 40 (step S 53) and unwrapped by the unwrap processing unit 41. Processing is performed (step S54).

Thereafter, the unwrapped phase differential image is input to the trend removing unit 121, subjected to Fourier transform (step S55), low frequency components equal to or lower than a predetermined frequency are removed (step S56), and inverse Fourier transform is performed. Is given. Thereby, the inputted phase differential image becomes a phase differential image from which a predetermined low frequency component is removed. The phase differential image from which the low frequency component has been removed is stored in the offset image storage unit 42 as an offset image, and pre-imaging is completed.

As shown in FIG. 28, during the main shooting, the main shooting mode is selected as the shooting mode using the operation unit 18a (step S70). When the main shooting mode is selected, a shooting instruction standby state is set (step S71). Thereafter, when an imaging instruction is given using the operation unit 18a, stripe scanning is performed (step S72), and M main captured image data 52 are stored in the memory 14. The captured image data 52 is read out by the image processing unit 15, a phase differential image is generated by the phase differential image generation unit 40 (step S73), and unwrap processing is performed by the unwrap processing unit 41 (step S74). ).

Thereafter, the unwrapped phase differential image is input to the trend removing unit 121, subjected to Fourier transform (step S75), low frequency components equal to or lower than a predetermined frequency are removed (step S76), and inverse Fourier transform is performed. Is given. Thereby, the inputted phase differential image becomes a phase differential image from which a predetermined low frequency component is removed. The phase differential image from which the low-frequency component has been removed is input to the offset removing unit 122, and the offset noise is further removed by subtracting the offset image (step S77). The phase differential image from which the offset noise has been removed is recorded in the image recording unit 16. Further, the phase contrast image generation unit 44 generates a phase contrast image from the phase differential image from which the offset noise has been removed, and records it in the image recording unit 16 (step S78). The phase differential image and phase contrast image generated in this way are displayed on the monitor 18b (step S79).

As described above, when the phase differential image generated from the pre-captured image data 51 and the main captured image data 52 is Fourier transformed to remove a predetermined low frequency component, the low frequency component of offset noise is removed from the phase differential image. . As described above, since the offset noise at the time of pre-photographing and that at the time of main photographing do not completely match, the trend remains when the offset image is subtracted. For this reason, as described above, no trend occurs when the difference between the phase differential image and the offset image from which the low-frequency component, which is mainly offset noise, is previously removed is taken.

Note that offset noise also has high frequency components. For example, a pixel defect in the X-ray detector 13 generates high-frequency offset noise. For this reason, in the above-described third embodiment, not only the low-frequency component is removed from the phase differential image generated at the time of actual imaging, but also an offset image is acquired and the phase differential image low-frequency component generated at the time of actual imaging. Is subtracted from the offset image. Thereby, high-frequency offset noise is also removed. When there is no pixel defect in the X-ray detector 13 or the pixel defect is corrected by another method (for example, by measuring the position of the pixel defect in advance and correcting the pixel value of the defective pixel at the time of the actual captured image data). In such a case, the offset noise may be removed by removing low frequency components from the phase differential image obtained by the actual photographing. Thereby, a phase differential image in which neither offset noise nor trend remains can be obtained.

In the first to third embodiments, the subject H is disposed between the X-ray source 11 and the first lattice 21. However, the subject H is disposed between the first lattice 21 and the second lattice. It may be arranged between the two.

In the first to third embodiments, the second grating 22 is moved in the direction perpendicular to the grating lines (X direction) during fringe scanning. However, the present applicant filed as Japanese Patent Application No. 2011-097090. As described above, the second grating 22 may be moved in a direction inclined with respect to the grid line (a direction not orthogonal to the X direction and the Y direction in the XY plane). This moving direction may be any direction as long as it is within the XY plane and other than the Y direction. In this case, the scanning position “k” may be set based on the X-direction component of the movement of the second grating 22. By moving the second grating 22 in a direction inclined with respect to the grating lines, the stroke (movement distance) required for scanning for one period of the fringe scanning becomes longer, so that there is an advantage that the movement accuracy is improved. .

In the first to third embodiments, the second grating 22 is moved at the time of fringe scanning. However, instead of the second grating 22, the first grating 21 is oriented in a direction perpendicular to the grating lines or inclined. It may be moved in the direction of

In the first to third embodiments, the X-ray source 11 that emits cone-beam X-rays emitted from the X-ray source 11 is used. However, X-rays that emit parallel-beam X-rays are used. It is also possible to use a source. In this case, instead of the expression (1), the first and

second gratings

21 and 22 may be configured so as to substantially satisfy p ₂ = p ₁ .

In the first to third embodiments, the X-rays emitted from the X-ray source 11 are incident on the first grating 21, and the X-ray source 11 has a single focal point. As described above, the X focus may be dispersed by providing a multi-slit (source grid) described in WO 2006/131235 or the like immediately after the exit side of the X-ray source 11. The grid line direction of the multi slit is the Y direction. As a result, it becomes possible to use a high-power X-ray source and the X-ray dose is improved, so that the image quality of the phase differential image is improved. The lattice lines (lattice grooves) of the multi-slit 150 extend in the Y direction and are parallel to at least one of the first and

second lattices

21 and 22. In this case, the lattice pitch p _{0 in} the X direction of the multi slit 150 needs to satisfy Expression (10). Here, the distance L ₀ is a distance in the Z direction from the multi slit 150 to the first grating 21.

When the multi-slit 150 is provided in this way, the position of the multi-slit 150 becomes the position of the X-ray focal point, and thus the distance L _{1 in the} above embodiment is replaced with the distance L ₀ .

In addition, when the multi-slit is provided, the first and second gratings are used in addition to performing the fringe scanning by moving the first grating 21 or the second grating 22 while the multi-slit 150 is fixed. It is possible to perform fringe scanning by moving the multi slit 150 while 21 and 22 are fixed. In this case, the multi slit 150 may be intermittently moved in the X direction using a value (p ₀ / M) obtained by dividing the pitch p ₀ of the multi slit 150 by M described above as a scanning pitch. As a result, the scanning position k of the multi slit 150 with respect to the first and

second gratings

21 and 22 is changed in order of k = 0, 1, 2,..., M−1.

In the first to third embodiments, the first grating 21 projects the incident X-ray geometrically, but the first grating 21 is known as disclosed in WO 2004/058070 and the like. May be configured to generate the Talbot effect. In order to generate the Talbot effect in the first grating 21, a small-focus X-ray light source or a multi-slit 150 may be used so as to enhance the spatial coherence of X-rays.

When the Talbot effect is generated in the first grating 21, the first grating 21 may be a phase grating instead of the absorption grating. The phase type grating is configured by replacing the X-ray absorption part of the absorption type grating with an X-ray phase forming part. The X-ray phase forming part is formed of a material (air, resin, etc.) having a predetermined refractive index difference with respect to the adjacent X-ray transmitting part. As the phase-type grating, there are known a phase grating that transmits incident X-rays with a phase modulation of π / 2 and a grating that transmits incident X-rays with a phase modulation of π / 2.

When the Talbot effect is generated in the first grating 21, the self-image (G1 image) of the first grating 21 is generated at a position away from the first grating 21 by the Talbot distance Z _m downstream in the Z direction. the distance _{L 2} from the first grid 21 to the second grid 22 is required to be Talbot distance _{Z m.}

Talbot distance Z _m is dependent on the beam shape of the structure and the X-ray of the first grating 21. The first grating 21 is absorption grating, if X-rays emitted from the X-ray source 11 is a cone beam shape, Talbot distance Z _m is represented by the formula (11). Here, “m” is a positive integer. In this case, the grating pitches p ₁ and p ₂ are set so as to substantially satisfy the expression (1) (however, when the multi slit 150 is used, the distance L ₁ is replaced with the distance L ₀ ).

Further, when the first grating 21 is a phase-type grating that applies phase modulation of π / 2 to the X-ray, and the X-ray emitted from the X-ray source 11 has a cone beam shape, the Talbot distance Z _m is And represented by equation (12). Here, “m” is “0” or a positive integer. In this case, the grating pitches p ₁ and p ₂ are set so as to substantially satisfy the expression (1) (however, when the multi slit 150 is used, the distance L ₁ is replaced with the distance L ₀ ).

In addition, when the first grating 21 is a phase-type grating that gives a phase modulation of π to the X-ray, and the X-ray emitted from the X-ray source 11 has a cone beam shape, the Talbot distance Z _m is expressed by the equation It is represented by (13). Here, “m” is “0” or a positive integer. Here, “m” is “0” or a positive integer. In this case, since the pattern period of the G1 image is ½ times the grating period of the first grating 21, the grating pitches p ₁ and p ₂ are set so as to substantially satisfy Expression (14) ( However, in the case of using the multi-slit 150, the distance _{L 1} is replaced by a distance _{L 0).}

The first grating 21 is absorption grating, if X-rays emitted from the X-ray source 11 is a parallel beam shape, Talbot distance Z _m is represented by the formula (15). Here, “m” is a positive integer. In this case, the lattice pitches p ₁ and p ₂ are set so as to substantially satisfy the relationship of p ₂ = p ₁ .

Further, when the first grating 21 is a phase-type grating that applies phase modulation of π / 2 to X-rays, and the X-rays emitted from the X-ray source 11 are parallel beams, the Talbot distance Z _m is And represented by equation (16). Here, “m” is “0” or a positive integer. In this case, the lattice pitches p ₁ and p ₂ are set so as to substantially satisfy the relationship of p ₂ = p ₁ .

When the first grating 21 is a phase-type grating that applies π phase modulation to X-rays, and the X-rays emitted from the X-ray source 11 are parallel beams, the Talbot distance Z _m It is represented by (17). Here, “m” is “0” or a positive integer. In this case, since the pattern period of the G1 image is ½ times the grating period of the first grating 21, the grating pitches p ₁ and p ₂ almost satisfy the relationship of p ₂ = p _1/2. Set to

In the first to third embodiments, the grating portion 12 is provided with the two gratings of the first and

second gratings

21 and 22. However, the second grating 22 is omitted and the first grating 21 is omitted. It is also possible to use only.

For example, by using an X-ray image detector described in Japanese Patent Laid-Open No. 2009-133823, the second grating 22 can be omitted and only the first grating 21 can be provided. This X-ray image detector is a direct conversion type X-ray image detector including a conversion layer that converts X-rays into electric charges and a charge collection electrode that collects electric charges converted in the conversion layer. The charge collection electrode includes a plurality of linear electrode groups. One linear electrode group is obtained by electrically connecting linear electrodes arranged at a constant period, and is arranged so that the phases thereof are different from those of other linear electrode groups. This linear electrode group functions as the second grating 22, and the presence of a plurality of linear electrode groups allows detection of a plurality of G2 images having different phases in one imaging. Therefore, in this configuration, the scanning mechanism 23 can be omitted.

Further, there is a method in which the scanning mechanism 23 is omitted and a phase differential image is generated based on single image data obtained by the X-ray image detector 13 via the first and

second gratings

21 and 22. As this method, there is a pixel division method filed by the present applicant as Japanese Patent Application No. 2010-256241. In this pixel division method, the first grating 21 and the second grating 22 are slightly rotated around the Z direction, and moire fringes having a period in the Y direction are generated in the G2 image. The single image data obtained by the X-ray image detector 13 is divided into groups of pixel rows (pixels arranged in the X direction) having different phases from each other with respect to the moire fringes, and a plurality of divided image data is obtained. A phase differential image is generated in the same procedure as the above-described fringe scanning method, assuming that the images are based on a plurality of different G2 images by fringe scanning. In this pixel division method, the intensity modulation signal described above is expressed as a change in intensity of pixel values for one cycle of moire fringes generated in single image data.

Further, similarly to the pixel division method, the scanning mechanism 23 is omitted, and the phase differential image is obtained based on the single image data obtained by the X-ray image detector 13 via the first and

second gratings

21 and 22. As a generation method, a Fourier transform method described in WO2010 / 050484 is known. This Fourier transform method obtains a Fourier spectrum by performing a Fourier transform on the single image data, separates a spectrum corresponding to a carrier frequency (a spectrum carrying phase information) from the Fourier spectrum, and then reverses the spectrum. This is a method of generating a phase differential image by performing Fourier transform. In this Fourier transform method, the intensity modulation signal described above is expressed as a change in intensity of pixel values for one cycle of moire fringes generated in a single image data, as in the case of the pixel division method.

The present invention can be applied to an industrial radiography apparatus and the like in addition to a radiography apparatus for medical diagnosis. In addition to X-rays, gamma rays or the like can be used as radiation.

Claims

A radiation source that emits radiation toward the subject;
A radiation detector that detects radiation emitted from the radiation source and generates image data;
A grating portion disposed between the radiation source and the radiation detector;
Based on the image data obtained by the radiation detector, a phase differential image generation unit that generates a phase differential image,
An unwrap processing unit that performs unwrap processing on the phase differential image;
A noise removing unit that removes a trend remaining when subtracting an offset image representing offset noise when there is no subject from the subject image that is the phase differential image after the unwrapping;
A radiation imaging apparatus comprising:
The noise removing unit
An offset noise removing unit that removes the offset noise by subtracting the offset image from the subject image;
A trend detector that detects the trend based on the subject image from which the offset noise has been removed;
A trend removing unit that subtracts and removes the trend detected by the trend detecting unit from the subject image;
The radiation imaging apparatus according to claim 1, further comprising:
The trend detection unit extracts a pixel value of the subject image from which the offset noise has been removed along a predetermined direction, detects the predetermined direction component of the trend,
The trend removing unit generates a trend component image in which the predetermined direction components of the trend are arranged along a direction perpendicular to the predetermined direction, and subtracts the trend component image from the subject image, thereby While removing the predetermined direction component of the trend from the specimen image,
The trend detection unit and the trend removal unit detect and remove the trend component in two directions, ie, a first direction that is the predetermined direction and a second direction that is perpendicular to the predetermined direction, from the subject image. The radiation imaging apparatus according to claim 2, wherein the trend is removed.
The said subject image is carrying out the quadrangle | tetragon and the said 1st direction and the said 2nd direction are the horizontal direction or the vertical direction along the one side of the said square, The range of Claim 3 characterized by the above-mentioned. Radiography equipment.
The trend detecting unit detects a component of the trend based on at least a pixel value of a missing region where the subject does not exist among pixel values of the extracted subject image. The radiographic apparatus according to the item.
The trend detection unit extracts the pixel value of the subject image along the first direction or the second direction, and then smoothes at least the subject region in which the subject is present, thereby The radiation imaging apparatus according to claim 3, wherein the first direction component or the second direction component is extracted.
The trend detection unit extracts pixel values of the subject image along the first direction or the second direction, and then, based on the data of the missing region, data of the subject region where the subject is located The radiation imaging apparatus according to claim 3, wherein the first direction component or the second direction component of the trend is extracted by interpolation, extrapolation, or fitting.
The trend detection unit detects the first direction component or the second direction component of the trend for a plurality of rows or columns along the first direction or the second direction, and detects the first direction component or the second direction component of the trend. The radiographic apparatus according to claim 3, wherein data obtained by averaging one direction component or the second direction component is detected as the first direction component or the second direction component.
4. The radiation imaging apparatus according to claim 3, wherein the trend detection unit sets at least one of the first direction and the second direction in a direction not passing through the subject.
2. The index according to claim 1, further comprising an index that designates the arrangement of the subject and designates the arrangement of the subject so that a missing region without the subject is generated in the phase differential image. Radiography equipment.
Based on the image data obtained by the radiation detector, an absorption image representing the absorption rate of the radiation is generated, and an absorption differential image generation unit that differentiates the absorption image to generate an absorption differential image,
The trend detection unit selects a plurality of pixels whose pixel values are substantially equal to neighboring pixels in the absorption differential image, and based on the pixel values of the pixels of the phase differential image at positions corresponding to the selected pixels. The radiation imaging apparatus according to claim 2, wherein the trend is detected.
The absorption differential image generation unit generates a first absorption image from the image data obtained when there is no subject, and generates a second absorption image from the image data obtained when the subject exists, The radiographic apparatus according to claim 11, wherein the absorption differential image is generated by differentiating a third absorption image obtained by dividing the second absorption image by the first absorption image.
12. The radiation imaging apparatus according to claim 11, wherein the pixel selected by the trend detection unit is a pixel in an unexposed area where the subject is not present.
The grating unit partially shields the first periodic pattern image by passing the radiation from the radiation source to generate the first periodic pattern image, and displays the second periodic pattern image. A second grid to generate,
The radiation imaging apparatus according to claim 1, wherein the radiation image detector generates the image data by detecting the second periodic pattern image.
The grating unit includes a scanning mechanism that moves the first grating or the second grating at a predetermined scanning pitch and sequentially sets a plurality of scanning positions.
The radiation image detector detects the second periodic pattern image at each scanning position to generate image data;
15. The radiation imaging apparatus according to claim 14, wherein the phase differential image generation unit generates a phase differential image based on a plurality of image data generated by the radiation image detector.
16. The radiation imaging apparatus according to claim 15, wherein the moving direction of the first grating or the second grating is a direction orthogonal to a grating line.
16. The radiation imaging apparatus according to claim 15, wherein the moving direction of the first grating or the second grating is a direction inclined with respect to a grating line.
15. The radiation imaging apparatus according to claim 14, wherein the phase differential image generation unit generates the phase differential image based on single image data obtained by the radiation detector.
15. The radiation according to claim 14, wherein the first grating is an absorption grating, and the first periodic pattern image is generated by geometrically projecting incident radiation. Shooting device.
15. The first periodic pattern image according to claim 14, wherein the first grating is an absorption grating or a phase grating, and generates the first periodic pattern image by causing a Talbot effect to incident radiation. Radiography equipment.
The radiation imaging apparatus according to claim 1, further comprising a multi-slit that partially blocks the radiation emitted from the radiation source and disperses the focal point.
An offset image representing offset noise when there is no subject is generated on the basis of image data obtained by arranging a grating portion between a radiation source and a radiation detector and performing imaging without a subject. An offset image generation step;
A subject image generation step for generating a subject image based on image data obtained by imaging a subject by arranging a grating portion between the radiation source and the radiation detector;
A noise removal step of removing, from the subject image, a trend that is noise remaining in the subject image when the offset image is subtracted from the subject image;
An image processing method comprising: