WO2018105493A1

WO2018105493A1 - Radiographic image capturing device, radiographic image capturing system, radiographic image capturing method, and program

Info

Publication number: WO2018105493A1
Application number: PCT/JP2017/043116
Authority: WO
Inventors: 明佃
Original assignee: キヤノン株式会社
Priority date: 2016-12-05
Filing date: 2017-11-30
Publication date: 2018-06-14

Abstract

A radiographic image capturing device capable of calculating with a high degree of accuracy an average energy or an average number of photons of radiation using a smaller number of pixel values than in the prior art captures a radiographic image by means of radiation, and is provided with: a determining means for determining a prior probability on the basis of predetermined criteria; an acquiring means for acquiring a pixel value measured from the radiographic image; and a calculating means for calculating at least one of the average energy and the average number of photons of the radiation on the basis of the prior probability and the pixel value.

Description

Radiation imaging apparatus, radiation imaging system, radiation imaging method, and program

The present invention relates to a radiation imaging apparatus, a radiation imaging system, a radiation imaging method, and a program.

In recent years, the practical application of photon counting type radiographic apparatuses that individually measure photons of radiation (for example, X-rays) in radiographic apparatuses has been progressing. In the photon counting type apparatus, it becomes possible to discriminate the energy per photon, and a new application is expected.

However, in order to realize a photon counting type radiation imaging apparatus, it is necessary to individually measure a large amount of incident radiation photons, so new hardware capable of high-speed driving is necessary, and development is high. There are barriers.

Therefore, in Patent Document 1, instead of individually measuring photons, the average energy and average of incident radiation are calculated from the average value and noise of a plurality of pixel values (values representing the amount of incident light output from the pixels of the radiation imaging apparatus). A method for estimating the number of photons has been proposed.

JP 2009-285356 A

However, the method described in Patent Document 1 has a problem that the number of pixels necessary for estimating the energy or the average number of photons with high accuracy becomes enormous.

The radiographic apparatus according to the present invention is a radiographic apparatus that captures a radiographic image with radiation, and includes a determination unit that determines a prior probability based on a predetermined condition, and obtains a pixel value measured from the radiographic image. And means for calculating at least one of the average energy and the average number of photons of the radiation based on the prior probability and the pixel value.

The present invention can calculate the average energy or the average number of photons with high accuracy with fewer pixels than the prior art.

It is a figure which shows an example of a structure of the radiography system which concerns on this invention. It is a flowchart which shows an example of operation | movement of the radiography system of a present Example. It is a figure which shows the effect of a present Example. It is a flowchart which calculates the solution of a 1st parameter and a 2nd parameter. It is a flowchart which shows an example of the prior probability determination method. It is a figure which shows distribution of the prior probability in case the number of photons is normal distribution. It is a flowchart which calculates the solution of a 1st parameter and a 2nd parameter based on the prior probability determined for every pixel. It is a figure which shows an example of the prior probability according to the imaging | photography site | part. It is a figure which shows an example of the prior probability according to to-be-photographed object information (body type). It is a flowchart which calculates the solution of a 1st parameter and a 2nd parameter based on the prior probability by contrast agent information. It is a figure which shows an example of the prior probability according to contrast agent information. The figure showing the relationship between an attention pixel and surrounding pixels. The figure showing the relationship between an attention pixel and surrounding pixels. The flowchart which showed an example of step 301 (determination of prior probability). The figure which shows the method of determining the spatial distribution of the coefficient of a regularization term. The figure which shows the method of determining the spatial distribution of the coefficient of a regularization term. The figure which shows the method of determining the spatial distribution of the coefficient of a regularization term. The flowchart which showed an example of step 301 (determination of prior probability). The figure which shows the method of determining the spatial distribution of the coefficient of a regularization term.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

FIG. 1 is a diagram showing an example of the configuration of a radiation imaging system according to the present invention. A radiation imaging system (or a radiation imaging apparatus) captures a radiation image with radiation. A radiation tube (radiation generator) 101 generates radiation. Radiation irradiated from the radiation tube 101 passes through the subject 102 and enters an FPD (Flat Panel Detector) 103 that is a radiation detection unit. The radiation tube 101 may be provided with a radiation filter for shielding low-energy radiation.

Although not shown, the FPD 103 includes a phosphor that converts radiation into visible light, and an image sensor that converts visible light into electric charge and voltage (a plurality of image sensors are provided, and one image sensor corresponds to one pixel). With. The FPD 103 also includes an image generation unit that converts the converted voltage into a digital radiographic image, and an image correction unit that performs offset correction, gain correction, and defect correction on the radiographic image.

For the FPD 103, a method of directly converting radiation using cadmium telluride (CdTe) or amorphous selenium (a-Se) into a voltage can be applied.

The radiation conditions (irradiation conditions) to be irradiated are set by the operation panel 104. The operation panel 104 includes a condition specifying unit 105, an imaging region specifying unit 106, and a condition display unit 107. The condition designation unit 105 can designate irradiation conditions such as a tube voltage, a tube current, and an irradiation time with a button or the like. When the imaging region and the radiation irradiation condition are associated in advance, and the imaging region designating unit 106 specifies the imaging region, it is possible to set the radiation irradiation condition corresponding to the imaging region.

The irradiation instruction unit 108 includes a foot pedal, and the radiation tube 101 can emit radiation when the operator depresses the foot pedal.

The radiation image output from the FPD 103 is transferred to the computer 109. The computer 109 includes an image acquisition unit 110, a prior information acquisition unit (determination unit) 111, an energy calculation unit 112, a prior information storage unit 113, an input device 114, and a display device 115. Note that the image correction unit included in the FPD 103 may be included in the computer 109. In some cases, the irradiation conditions of radiation are stored in advance in the computer 109 instead of the operation panel 104.

The image acquisition unit 110 has a function of capturing the radiation image generated by the FPD 103 into the computer 109. The image acquisition unit (acquisition unit) 110 acquires a radiographic image from the FPD 103 and acquires a pixel value measured from the radiographic image.

The prior information acquisition unit 111 acquires prior information from the radiation tube 101, the FPD 103, the operation panel 104, the prior information storage unit 113, the input device 114, the C arm 118, and the gantry 119, and determines a prior probability described later.

The prior information acquisition unit (determination unit) 111 includes radiation generation information related to a radiation tube (radiation generation unit) 101 that generates radiation, radiation detection information related to an FPD (radiation detection unit) 103 that detects radiation, and a subject 102 to be photographed. Subject information relating to the subject, imaging part information relating to the imaging part of the subject 102 to be photographed, contrast agent information relating to a contrast agent injected into the subject 102, support relating to the C-arm (support unit) 118 that supports the radiation imaging apparatus or the radiation tube 101 Based on the information and at least one of the gantry information about the gantry 119 on which the subject 102 is placed, a prior probability described later is determined.

The prior information acquisition unit (determination unit) 111 uses a radiographic image captured before a predetermined time among a plurality of captured radiographic images, to generate radiation generation information, radiation detection information, subject information, imaging region information, And at least one of the contrast agent information may be acquired.

Prior information acquired by the prior information acquisition unit 111 includes radiation generation information, radiation detection information, subject information, imaging region information, contrast agent information, support information, and gantry information, as well as a probability relating to the number of radiation (number of photons). There is probability distribution information of the distribution.

The probability distribution information includes functions of Poisson distribution and normal distribution, functions derived from these functions (functions represented by expressions (15), (16), (23), and (24) described later, and Poisson distribution and normal distribution parameters).

The radiation generation information includes at least one tube information of tube voltage of the radiation tube 101, tube current of the radiation tube 101, radiation irradiation time of the radiation tube 101, presence / absence of a radiation filter, and characteristics of the radiation filter. It is. The radiation generation information includes the radiation spectrum of the radiation tube 101.

The radiation detection information regarding the FPD 103 includes an accumulation time and the like. The imaging part information includes information for identifying the chest and head. The subject information includes the identification number, name, sex, age, height, weight, and the like of the subject 102 input by the input device 114. The contrast agent information includes the type and concentration of the contrast agent, the site where the contrast agent is injected, and the like. The support information regarding the C arm 118 includes the material, position, angle, and the like of the C arm 118. The gantry information regarding the gantry 119 includes the position of the gantry 119 and the like.

In the present embodiment, the prior information acquired by the prior information acquisition unit 111 is used for energy or photon number calculation, so that the average energy or average can be obtained with a smaller number of shots than the prior art (for example, the invention described in Patent Document 1). It becomes possible to calculate the number of photons with high accuracy.

The energy calculation unit 112 is based on the probability distribution regarding the number of radiations (number of photons) acquired by the prior information acquisition unit 111, the prior probability calculated by the prior information acquisition unit 111, and the radiation image acquired by the image acquisition unit 110. Calculate and output average energy or average number of photons.

The energy calculation unit (calculation unit) 112 calculates at least one of the average energy of radiation and the average number of photons by the Bayes' theorem described later based on the probability distribution, the prior probability, and the pixel value. The probability distribution of this embodiment is a probability distribution related to the number of photons of radiation.

The image acquisition unit 110, the advance information acquisition unit 111, the energy calculation unit 112, and the advance information storage unit 113 are a central processing unit of a computer, a main storage device, a storage device such as a hard disk, a graphic processing for high-speed calculation, It can be configured by general hardware such as a unit and a LAN (Local Area Network) adapter. The function of each component is implemented as software. Further, a circuit that executes the function of each component may be electrically mounted.

The input device 114 is, for example, a keyboard, a mouse, a touch panel, or the like. The operator can use the input device 114 to input imaging part information and subject information. The imaging part information and the subject information may be input through a LAN adapter mounted on the computer 109 in addition to the input by the input device 114.

The imaging part information and subject information are acquired as prior information by the prior information acquisition unit 111 and used to determine the prior probability.

The display device 115 is used to display a radiation image. In particular, when the display device 115 includes a color display, it is possible to express information on average energy by color (color or gradation), and improvement in diagnostic performance of radiographic images is expected.

Some of the functions of the operation panel 104 may be provided in the input device 114 and the display device (output device) 115. For example, the input of the radiation irradiation condition performed through the condition designating unit 105 or the imaging region designating unit 106 may be performed by the input device 114. In addition, radiation conditions such as tube voltage, tube current, and irradiation time may be displayed on the display device 115 instead of the condition display unit 107.

The radiation tube 101, the FPD 103, the operation panel 104, the irradiation instruction unit 108, and the computer 109 are connected to the synchronization device 116. The synchronization device 116 determines whether or not radiation exposure is possible from the state of the FPD 103, the pressing state of the irradiation instruction unit 108, and the processing state of the computer 109.

In addition, the radiographic system of the present embodiment includes a contrast medium injection device 117 for performing angiography. The contrast medium injection device 117 injects a blood vessel contrast medium such as iodine into the subject 102 based on an instruction from the operator. The contrast medium injection device 117 transmits contrast medium information as prior information to the prior information acquisition unit 111. The contrast agent information includes the type and concentration of the contrast agent.

FIG. 2 is a flowchart showing an example of the operation of the radiation imaging system of the present embodiment. In step 201, the operator inputs radiation irradiation conditions. The radiation irradiation condition is input through, for example, the condition specifying unit 105 or the imaging region specifying unit 106. Also, subject information such as the identification number of the subject 102 is input through the input device 114 together with the irradiation condition.

In step 202, the operator instructs radiation irradiation through the irradiation instruction unit. Thereafter, in step 203, the synchronization device 116 determines whether or not radiation exposure is possible from the state of the radiation tube 101, the FPD 103, the operation panel 104, the irradiation instruction unit 108, and the computer 109. If it is determined that exposure is not possible, a warning is output to the display device 115 in step 204.

If it is determined in step 203 that exposure is possible, the process proceeds to step 205. In step 205, the prior information acquisition unit 111 determines a probability distribution (for example, a probability distribution related to the number of radiations) and a prior probability necessary for calculating the average energy based on the prior information. The prior information is acquired from the radiation tube 101, the FPD 103, the operation panel 104, the prior information storage unit 113, the input device 114, the C arm 118, and the gantry 119. The prior information acquisition unit (determination unit) 111 determines a probability distribution and a priori probability of Bayes' theorem described later.

Also, the prior information acquisition unit 111 determines the number of pixel values necessary for estimating the energy or the average number of photons. The number of pixel values necessary to estimate the energy or the average number of photons from the imaging part information input from the operation panel 104 based on data (table or the like) in which the imaging part of the radiographic image is associated with the number of images to be taken. The number of radiographic images corresponding to is determined. Data in which the imaging region and the number of images are associated with each other is stored in advance in the pre-information storage unit 113.

Thereafter, the prior information acquisition unit 111 notifies the synchronization device 116 that radiation irradiation is possible, and the synchronization device 116 sends a radiation irradiation instruction signal to the radiation tube 101 and the FPD 103. Thus, in step 206, radiation (irradiation) acquisition is started by irradiating the FPD 103 with radiation.

When the pixel values necessary for estimating the average energy or the average number of photons are acquired, in step 207, the radiographic image capturing is terminated. In step 208, the average energy is calculated from the probability distribution and prior probabilities regarding the number of radiations obtained in step 205 and the pixel value obtained in step 206. The energy calculation unit 112 calculates average energy. A method for calculating the average energy will be described later.

In step 209, the information on the average energy calculated in step 208 is displayed and stored. For example, the information about the average energy can be displayed in color by the display device 115 by associating the obtained average energy with the color.

Next, the calculation method for calculating the average energy in step 208 will be described in detail. Here, the relationship between the average pixel value m of the pixel values acquired in step 206, the average number of photons n incident on the pixel, and the proportionality coefficient k (E) that is a function of the average energy E is expressed by equation (1). It is represented by

K (E) is a pixel value per photon and is a function of energy E. Since the pixel value (light emission amount) per photon can be obtained as a function of the energy E, if k (E) can be calculated from the pixel value, the average energy E is calculated.

For example, k (E) is expressed by a linear expression of average energy E. In the prior art (for example, the invention described in Patent Document 1), k (E) = E. That is, if the average pixel value m can be divided into the average photon number n and the proportional coefficient k (E), the average energy E can be calculated.

Therefore, hereinafter, a method of obtaining the average photon number n and the proportional coefficient k (E) from a plurality of pixel values (pixel values) will be described.

Here, it is assumed that in step 206, the FPD 103 takes L radiation images and L pixel values x = m ₁ , m ₂ ,..., _{M L} are acquired (measured) at a certain pixel. Taking one radiation image and obtaining one pixel value is said to be one measurement. When the measured value x is x = m ₁ , m ₂ ,..., _{M L} , the probability f (p, q | m) that a certain parameter p ₁ , p ₂ is p ₁ = p, p ₂ = q ₁ , m ₂ ,..., _{M L} ) is expressed by equation (2) based on Bayes' theorem.

K is a normalization constant. g (m ₁ , m ₂ ,..., m _L | p, q) indicates that the measured value x is x = m ₁ , when the parameter p ₁ is p ₁ = p and the parameter p ₂ is p ₂ = q. m ₂ ,... m _L is the probability of the event (including probability distribution). g (m _i | p, q) indicates that when the parameter p ₁ is p ₁ = p and the parameter p ₂ is p ₂ = q, the pixel value x _{i of the} i-th (or i-th) measurement is m _This is the probability (including probability distribution) of an event that is _i . h (p, q) is a prior probability.

As will be described later, in the parameters p ₁ and p ₂ , p ₁ is the average pixel value m, and p ₂ is the reciprocal 1 / k (E) of the proportional coefficient k (E). Alternatively, p ₁ is the average pixel value m, and p ₂ is the pixel value variance σ ² (or standard deviation).

Thus, the probability distribution g (m _i | p, q) and the prior probability h (p, q) are the first parameter p ₁ = p and the average energy of the radiation used for calculating the average pixel value of the radiation image. The second parameter p ₂ = q used in the calculation of is a variable. The first parameter p is at least one of a temporal average average pixel value at a predetermined pixel of the radiographic image and a spatial average average pixel value of a plurality of pixels of the radiographic image. The second parameter is at least one of a function having the average energy E as a variable, a variance σ ² of the pixel values of the radiographic image, and a standard deviation of the pixel values of the radiographic image.

According to the equation (2), if the probability distribution g (m _i | p, q) and the prior probability h (p, q) are calculated, the probability f (p, q | m ₁ , m ₂ ,... m _L ) can be calculated, and the parameters p and q can be calculated with high accuracy with a smaller number of pixel values than in the prior art.

FIG. 3 is a diagram showing the effect of this embodiment. In FIG. 3, the prior probability h (p, q) is determined by the equation (27). The horizontal axis represents the number of radiographic images (that is, the number of measurements), and the vertical axis represents the pixel value k (E) per photon corresponding to the average energy E. As shown in FIG. 3, when this embodiment is compared with the prior art, this embodiment calculates a value that approximates the true value of the average energy E with a smaller number of radiation images (number of measurements) than the prior art. can do.

As will be described later, as the probability distribution g (m _i | p, q), a probability distribution relating to the number of radiations (photon number) can be used in consideration of physical phenomena relating to generation and absorption of radiation. The prior probability h (p, q) can be calculated from information (that is, prior information) acquired separately from the measurement of the pixel value x.

Formula (2) is described as a product of functions, but in order to facilitate calculation, Formula (3) describing Formula (2) in logarithm is used hereinafter.

e is the base of the natural logarithm. If the parameter p ₁ = p and the parameter p ₂ = q are searched so that the value of the expression (3) becomes maximum, the average pixel value m which is p ₁ and 1 / k (E) or pixel which is p ₂ A variance σ ² (or standard deviation) of values is calculated. A known method can be applied to search for the parameter p ₁ = p and the parameter p ₂ = q so that the value of the expression (3) is maximized. For example, in Equation (3), searching for a local maximum value according to the local maximum condition of Equation (6) based on the stationary point (p, q) that becomes Equation (4) and the Hessian matrix of Equation (5), The parameter p ₁ = p and the parameter p ₂ = q that maximize the value of (3) can be searched.

As an easy method, the stationary point obtained from the equation (4) is substituted into the equation (3), and the parameters p ₁ = p and p ₂ = q that maximize the value of the equation (3) are set. You may search. In the numerical calculation method, a method such as a steepest descent method may be used.

The energy calculation unit (calculation unit) 112 calculates the probability f of the event that the first parameter p ₁ and the second parameter p ₂ become the first value and the second value, respectively, when the pixel value x _i is measured. The parameter solution is calculated so that (p, q | m ₁ , m ₂ ,..., _{M L} ) is maximized. The energy calculation unit (calculation unit) 112 has a probability distribution g (m _i | p, q) and an event probability f (p, q | m ₁ , m ₂ ,... _{M L} ) and the maximum. A first value p and a second value q, which are solutions of the first parameter p ₁ and the second parameter p ₂ of the prior probability h (p, q), are calculated.

FIG. 4 is a flowchart for calculating solutions of the first parameter and the second parameter. First, in step 301, the prior information acquisition unit 111 determines the prior probability log _e h (p, q) from the prior information. A method for determining the prior probability log _e h (p, q) will be described later. In step 302, the energy calculator (calculating unit) 112, a parameter p, and q as a variable, the probability distribution pixel value _{m i} is observed _{_{log e g (m i | p}} , q) is calculated. Probability _{_{log e g (m i | p}} , q) calculation will be described later.

In step 303, when the pixel values x = m ₁ , m ₂ ,..., _{M L} are observed using the equation (3) derived from Bayes' theorem, the energy calculation unit (calculation unit) 112 The probability log _e f (p, q | m ₁ , m ₂ ,..., _{M L} ) taking the parameters p and q is calculated. In step 304, as described above, the parameters p and q that maximize the probability log _ef (p, q | m ₁ , m ₂ ,..., _{M L} ) are determined.

In step 305, the average pixel value m and 1 / k (E) or the variance σ ^{2 of} pixel values (or standard deviation) are determined from the parameters p and q determined in step 304. Thereby, the energy calculation unit (calculation unit) 112 determines the average photon number n and the proportional coefficient k (E), and calculates the average energy E from the relational expression of the proportional coefficient k (E) and the average energy E.

Then, the probability _log e _g | showing _(m i p, q) and parameters _{_{_{p 1, p 2 (p 1}}} = p, p 2 = q) Specific examples of. In the following, a first case where the probability distribution of the number of photons is handled as a Poisson distribution and a second case where the probability distribution of the number of photons is handled as a normal distribution will be described. That is, in this embodiment, the probability distribution is at least one of a Poisson distribution and a normal distribution.

<First example: When the probability distribution of the number of photons is treated as a Poisson distribution>
Since the average number of photons is n, the probability that n _i number of photons incident on the i-th frame is known to follow a Poisson distribution of formula (7).

_ni ! Is the factorial of _{n i.} Here, the average photon number n is expressed by equation (8) using equation (1), and the photon number n _i of a certain frame is expressed by equation (9).

_mi is the pixel value of a certain pixel in the i-th frame.

Substituting Equation (8) and Equation (9) into Equation (7) yields Equation (10).

Equation (10) is a function of the pixel values _{m i,} and the average pixel value m, the reciprocal 1 / k of the proportional coefficient k (E) and (E). Therefore, the parameters p ₁ and p ₂ are the average pixel value m and the reciprocal 1 / k (E) of the proportionality coefficient, respectively, and the probability g (m _i | p, q) is expressed by Equation (11).

The first parameter is an average pixel value m that is a time average of predetermined pixels of the radiation image. The second parameter is a function 1 / k (E) having the average energy E as a variable. The calculation proceeds with the second parameter as the reciprocal 1 / k (E) of k (E), but the calculation is performed with the second parameter as the proportional coefficient k (E) instead of the reciprocal 1 / k (E). A similar result can be obtained by proceeding.

Substituting equation (11) into equation (3) yields equation (12).

The expression (12), factorial _{{m i / k (E)} }! Therefore, in order to facilitate handling, factorial {m _i / k (E)}! Is represented by a Γ function, Equation (12) becomes Equation (14).

Substituting equation (14) into equation (4) yields equations (15) and (16).

Ψ (x) = δlog _e Γ (x) / δx is a digamma function. In Expressions (15) and (16), for example, the Newton method is used, and the average pixel value m as a solution and the reciprocal 1 / k (E) of the proportional coefficient can be obtained. The average photon number n is calculated by substituting the average pixel value m obtained by the equations (15) and (16) and the inverse of the proportionality coefficient 1 / k (E) into the equation (1).

In this way, the energy calculation unit (calculation unit) 112 determines that the event probability f (p, q | m ₁ , m ₂ ,... _{M L} ) is a maximum value according to the equations (15) and (16). The solution of the first parameter m and the second parameter 1 / k (E) is calculated as follows.

<Second example: When the probability distribution of the number of photons is treated as a normal distribution>
In the first case where the probability distribution of the number of photons is treated as a Poisson distribution, the factorial {m _i / k (E)}! There are terms that are not easy to calculate, such as Γ and Γ functions. On the other hand, when the number of photons per unit time increases, the Poisson distribution is approximated by a normal distribution having an average n and a variance n as shown in Expression (17).

When Expression (8) and Expression (9) are substituted into Expression (16), Expression (17) becomes Expression (18). Further, considering the probability normalization condition, Equation (18) becomes Equation (19).

Therefore, when the distribution of the number of photons follows a normal distribution with mean n and variance n, the distribution of pixel values follows a normal distribution with average pixel value m and variance σ ² = k (E) · m. Therefore, the parameters p ₁ and p ₂ are the average pixel value m and the variance σ ² of the pixel values, respectively, and the probability g (m _i | p, q) is expressed by Equation (20). The first parameter is an average pixel value m that is a time average of predetermined pixels of the radiation image. The second parameter is the variance σ ² of the pixel values of the radiation image.

Comparing equation (19) and equation (20), the proportional coefficient k (E) becomes equation (21).

Therefore, by dividing the variance σ ² of pixel values by the average pixel value m, the proportional coefficient k (E) can be calculated, and the average photon number n can be calculated by Equation (7).

In the following, it advances the calculated one of the parameters as the dispersion sigma ² but, in place of the variance sigma ^2, the standard deviation √σ ^{2 = σ} as Get proceed even if similar results are obtained parameters.

From equation (3) and equation (20), the probability distribution of mean pixel value m and variance σ ² when _L pixel values x = m ₁ , m ₂ ,... (Logarithm) is expressed by equation (22).

Substituting equation (22) into equation (4) yields equations (23) and (24).

Equations (23) and (24) can determine the average pixel value m, which is the solution, and the variance σ ^{2 of} pixel values, for example, by using the Newton method.

As described above, the energy calculation unit (calculation unit) 112 determines that the event probability f (p, q | m ₁ , m ₂ ,... _{M L} ) is the maximum value according to the equations (23) and (24). The solution of the first parameter m and the second parameter σ ² is calculated as follows.

The prior probability h (p, q) is expressed in the form of a natural logarithm, for example, log _e h (p, q) in equation (3). Therefore, the prior probability h (p, q) is represented by a canonical distribution as shown in Equation (25).

Β is a coefficient. Z (β) is a normalization constant. e is a natural logarithm. E (p, q) is a function of parameters p and q. The function E (p, q) is a function having at least one extreme value.

By giving an extreme value to E (p, q), it is possible to make an assumption that the probability increases when the parameters p and q are predetermined values, and the number of pixel values is less than a predetermined threshold value. Even in this case, it is possible to converge to p or q with high accuracy. In order to have an extreme value, for example, E (p, q) may be expressed by a function of quadratic or higher (in the example described later, as shown in equation (29), E (p, q) is Expressed by a quartic function).

Examples of other probability distributions include beta distribution, gamma distribution, and chi-square distribution. These probability distributions can also have extreme values by appropriately setting parameters, and are suitable as prior probabilities. When the probability distribution is a normal distribution and Equation (4) becomes Equation (23) and Equation (24), and E (p, q) is expressed by a simple algebraic equation, Equation ( By substituting Equation (25) into Equation (23) and Equation (24) for transformation, an equation representing the solution may be obtained. In that case, the solution can be calculated without using the Newton method.

In this embodiment, an example in which L pixel values are acquired by one pixel by acquiring L radiation images has been described. However, the pixel values necessary for the calculation of the above parameters are set to a plurality of pixels. Can also be obtained. For example, as described above, when 4 pixels are used to obtain L pixel values, the number of acquired radiological images is L / 4. When a plurality of pixels are used, the number of radiographic images taken can be made smaller than L, and high-precision p or q can be calculated while shortening the radiographing time.

Next, a method for determining the prior probability from the prior information will be described. FIG. 5 is a diagram illustrating an example of a method for determining the prior probability h (p, q).

In step 401, tube information is acquired. As described above, the prior information acquisition unit 111 acquires the tube information from the operation panel 104 or the like.

In step 402, the prior information acquisition unit 111 calculates the minimum energy (or minimum average energy) of radiation and the minimum proportional coefficient _kmin . As the radiation passes through the object, the low energy photons of the radiation tend to be selectively absorbed by the object, so the average energy of the radiation after passing through the object is increased compared to before passing through the object.

Therefore, the corresponding radiation spectrum is obtained from the set tube voltage of the radiation tube 101 and the presence or absence of the radiation filter, and the average energy of the radiation when not passing through the object such as the subject 102 is calculated. Then, find the minimum energy of radiation. The energy calculation unit 112 calculates a corresponding proportionality coefficient _kmin based on the calculated minimum energy of radiation, and the prior information acquisition unit 111 acquires the calculated proportionality coefficient _kmin as prior information.

The radiation spectrum can be obtained by using a known model such as a TASMIP (Tungsten Anode Spectral Modeling Interpolating Polynomials) model.

In consideration of the fact that the radiation spectrum depends on the characteristics of the radiation tube 101 and the like, the result of evaluating the characteristics of the radiation tube 101 in advance using a spectrometer may be stored in the prior information storage unit 113. The minimum energy of radiation may be determined from the stored radiation spectrum. When a filter is inserted into the radiation tube 101, the prior information storage unit 113 stores information on the mass attenuation coefficient of the filter as prior information, and after passing through the filter based on the stored information on the mass attenuation coefficient. The radiation spectrum may be determined.

As another method for calculating the minimum average energy, the method described in Patent Document 1 can also be used. When calculating the average energy at the time of filter insertion by the method described in Patent Document 1, it is only necessary to obtain a radiographic image with the filter inserted and obtain the average energy.

In step 403, the maximum energy (or maximum average energy) of radiation and the maximum proportionality coefficient k _max are obtained. The maximum value of photon (radiation photon) energy is determined by the set tube voltage. For example, when the set tube voltage is 100 kilovolts, the maximum energy of radiation photons may be considered as 100 kilovolts. Therefore, the prior information acquisition unit 111 calculates the maximum energy of photons from the set tube voltage of the radiation tube 101, and calculates the corresponding proportional coefficient k _max from the calculated maximum energy.

In step 404, prior probabilities log (p, q) are obtained from the proportional coefficients k _min and k _max calculated in

steps

402 and 403.

The calculation of the prior probability h (p, q) is performed by the prior information acquisition unit (determination unit) 111. When the probability distribution of the number of photons is a Poisson distribution, the prior probability h (p, q) is expressed by Equation (26).

When the probability distribution of the number of photons is a normal distribution, the prior probability h (p, q) is expressed by Expression (27).

Here, A is a constant. Further, the region of 1 / k _max ≦ 1 / k (E) ≦ 1 / k _min is called an inner region of the probability distribution, and 1 / k (E) <1 / k _max and 1 / k _min <1 / k ( E) is called the outer region of the probability distribution. FIG. 6 is a diagram illustrating a distribution of prior probabilities h (p, q) when the number of photons is a normal distribution as shown in Expression (27). FIG. 6A is a diagram illustrating an inner region and an outer region where the horizontal axis is the average pixel value m and the vertical axis is the variance σ ² .

FIG. 6B is a diagram illustrating the relationship between the variance σ ² and the prior probability h (m, σ ² ) when the average pixel value m is a. As shown in FIG. 6B, the prior probability h (m, σ ² ) becomes maximum between the second parameters (inner regions) corresponding to the minimum energy and the maximum energy of radiation. The prior probability h (m, σ ² ) is 0 in a range (outer region) other than between the second parameters corresponding to the minimum energy and the maximum energy of radiation.

The prior probability h (m, σ ² ) may be maximized between the first parameters (inner regions) corresponding to the minimum energy and the maximum energy of radiation. The prior probability h (m, σ ² ) may be 0 in a range (outer region) other than between the first parameters corresponding to the minimum energy and the maximum energy of radiation.

Substituting the equation (27) of the prior probability h (p, q) when the number of photons is a normal distribution into the equation (22), the prior probability h (m, σ ² ) is 0 in the outer region. The logarithm log _e f (p, q | m ₁ , m ₂ ,..., _{M L} ) of the probability of (22) diverges to negative infinity. Similarly, when the equation (26) of the prior probability logh (p, q) when the number of photons is Poisson distribution is substituted into the equation (12) or (14), the prior probability h (m, 1 / Since k (E)) is 0, the logarithm of the probability of Expression (12) or Expression (14) diverges to negative infinity.

Therefore, in the outer region, the probability does not take the maximum value, and the obtained parameter (average pixel value m) is a parameter that always exists in the inner region.

When it is desired to avoid the logarithm of the probability diverging to negative infinity during the calculation, the prior probability h (p, q) in the outer region of the equation (26) or the equation (27) is set to A instead of 0. A finite number that is sufficiently smaller may be set.

FIG. 7 is a diagram showing another example of a method for determining the prior probability h (p, q). In this example, the prior probability h (p, q) is calculated using information input before radiographic image capturing.

In step 601, the prior information acquisition unit 111 acquires imaging part information. Next, in step 602, subject information is acquired. The subject information is information input through the input device 114 or the like and acquired by the prior information acquisition unit 111, and includes the identification number, name, sex, age, height, weight, and the like of the subject 102.

In step 603, a prior probability is determined for each pixel. In this example, since the imaging region is known, the anatomical feature of the subject 102 captured as a radiographic image is estimated for each pixel. For example, it is estimated whether each pixel images a bone part, a muscle part, or a lung part. Then, based on the prior probabilities determined for each pixel, the average energy in each pixel is estimated.

FIG. 8 shows a prior probability h (m = a, σ ² ) when the probability distribution of the number of photons is a normal distribution (when the parameters p ₁ and p ₂ are the average pixel value m and the variance σ ² of the pixel values). FIG. The prior probability h (m = a, σ ² ) is the maximum at the second parameter σ ² corresponding to the energy of the radiation that has passed through the imaging region of the subject 102 to be imaged. The prior probability h (m, σ ² ) may be the maximum in the first parameter m corresponding to the energy of the radiation that has passed through the imaging region of the subject 102 to be imaged.

When pixel A corresponds to a bone part, the prior probability h (m = a, σ ² ) at pixel A is such that the prior probability is maximized at the position of the average energy obtained after the radiation passes through the bone. Is set. In FIG. 8B, the prior probability h (m = a, σ ²⁾ so that the prior probability at the position where the variance σ ² is σ ² = k _bone · a and the average energy is indicated by a broken line is maximized. ) Is set. The proportionality coefficient k _bone is a proportionality coefficient corresponding to the average energy of the radiation transmitted through the bone part.

If pixel B corresponds to the heart or myocardium part, prior probability h (m = a, σ ²⁾ at pixel A so that the prior probability is maximized at the position of the average energy obtained after the radiation passes through the heart. ) Is set. In FIG. 8C, the prior probability h (m = a, σ ²⁾ so that the prior probability at the position where the variance σ ² is σ ² = k _muscle · a and the average energy is indicated by a broken line is maximized. ) Is set. The proportionality coefficient k _muscle is a proportionality coefficient corresponding to the average energy of the radiation transmitted through the heart or myocardial part.

As shown in FIG. 8B or FIG. 8C, the probability distribution having the maximum value at σ ² = k _bone · a or σ ² = k _muscle · a is expressed by, for example, the parameter β in Equation (28). And by adjusting the functions c ₁ (m), c ₂ (m), c ₃ (m), c ₄ (m), and c ₅ (m) shown in Equation (29).

Here, β is a parameter. Z (β) is a normalization constant. e is a natural logarithm. E (m, σ ² ) is represented by Expression (29).

The functions c ₁ (m), c ₂ (m), c ₃ (m), c ₄ (m), and c ₅ (m) are functions of the average pixel value m.

Also, in step 603, subject information such as the identification number of the subject 102 may be considered. Generally, when the body type of the subject 102 is obese, the amount of radiation absorbed is larger than that of the thin type, and thus the observed average energy is high. Therefore, when photographing the obese body type subject 102, the prior probability of the position of higher average energy is set to be larger than that of the thin type body subject 102.

FIG. 9 is a diagram illustrating changes in the prior probability distribution of obesity type and lean type. The prior probability h (m = a, σ ² ) is the maximum with the second parameter σ ² corresponding to the amount of radiation absorbed by the subject 102 to be imaged. The prior probability h (m, σ ² ) is a first parameter m corresponding to the radiation absorption amount of the subject 102 to be imaged, and may be maximized.

For example, BMI (Body Mass Index) may be calculated from the height and weight of the subject information, and the body type may be determined according to the BMI. Further, the body type may be determined using the body fat percentage data. The body shape may be determined according to the irradiation time of radiation from the radiation tube 101. The irradiation time of radiation can also be controlled using an AEC (Automatic Exposure Control) device or the like. Equation (30) represents a formula for calculating BMI.

In this embodiment, the imaging region, size, and body shape of the subject 102 are determined based on information (imaging region information, subject information, etc.) input by the operation panel 104 or the input device 114. In addition, the imaging region, size, and body shape of the subject 102 may be determined by a method such as pattern matching based on the captured radiographic image. In that case, in consideration of processing time such as pattern matching among a plurality of captured radiographic images, a determination is made using a radiographic image captured before a predetermined time, whereby the processing of the present embodiment Time can be shortened.

In this case, the prior information acquisition unit (determination unit) 111 acquires subject information or imaging region information using a radiographic image captured before a predetermined time among a plurality of captured radiographic images.

Further, a prior probability distribution may be generated by using past information associated with imaging part information and subject information as prior information. For example, the prior information storage unit 113 stores energy information obtained when a radiation image has been captured in the past as prior information, so that a prior probability distribution can be generated.

In this embodiment, a pixel value necessary for parameter determination is acquired using a certain pixel, but a pixel value required for parameter determination may be acquired using a plurality of pixels. In this case, instead of determining the prior probability for each pixel, the prior probability is determined for each set of a plurality of pixels. For example, as described above, when four pixels are used to obtain L pixel values, the prior probabilities corresponding to the four pixels are determined from imaging part information, subject information, images, and the like.

FIG. 10 is a diagram illustrating a method of determining the prior probability h (m, σ ² ) according to contrast agent information such as the type and concentration of contrast agent. The prior probability h (m = a, σ ² ) is the maximum at the second parameter σ ² corresponding to the energy of the radiation that has passed through the contrast agent injected into the object 102 to be imaged. For example, a case where angiography is performed using the contrast medium injection device 117 is assumed. The prior probability h (m, σ ² ) is the first parameter m corresponding to the energy of the radiation that has passed through the contrast agent injected into the object 102 to be imaged, and may be maximized.

In step 901, the prior information acquisition unit 111 acquires information on contrast medium to be injected (contrast medium type, concentration, and the like). When the contrast agent is iodine or the like, the average energy of the radiation increases after the radiation passes through the subject 102. On the other hand, when the contrast agent is carbon dioxide or the like, the average energy of the radiation decreases after the radiation passes through the subject 102.

Therefore, in step 902, the prior information acquisition unit 111 and the energy calculation unit 112 calculate an expected average energy and a proportional coefficient k (E) corresponding thereto from the contrast agent information.

When the pixel corresponds to the contrast portion of the contrast agent, the prior probability h (m, σ ² ) at the pixel is set so that the prior probability becomes maximum at the position of the average energy obtained after the radiation passes through the contrast agent. Is done. In FIG. 10B, the prior probability h (m, σ ² ) is set so that the prior probability at the position where the variance σ ² is σ ² = k _CM · a and the average energy is indicated by a broken line is maximized. . Proportionality factor k _CM is a proportionality coefficient corresponding to the average energy of the radiation transmitted through the imaging moiety of the contrast agent.

This prior probability can also be obtained by equation (28). Specifically, in Equation (28), the parameter β and the functions c ₁ (m) and c ₂ (m) shown in Equation (29) are set so that σ ² = k _CM · a and at least one maximum value. , C ₃ (m), c ₄ (m), and c ₅ (m) can be adjusted to obtain a probability distribution.

After the calculation of the prior probability and the like is completed, in step 903, a contrast medium is injected and a radiographic image is taken. Then, the average energy and the like are calculated from the radiation image (pixel value), the probability distribution information regarding the number of radiations, and the prior probability.

Note that the prior information acquisition unit (determination unit) 111 may acquire contrast agent information using a radiographic image captured before a predetermined time among a plurality of radiographic images captured. For example, the position of the blood vessel may be determined in advance from the radiographic image using pattern matching, and the prior probability in the pixel of the contrast portion may be calculated using the blood vessel through which the contrast agent passes as the contrast portion.

Thus, based on prior information such as subject information, imaging region information, and contrast agent information, the prior information acquisition unit (determination unit) 111 sets the first prior probability in the first pixel of the radiation image. Then, a second prior probability different from the first prior probability is set for the second pixel of the radiation image.

The energy calculation unit (calculation unit) 112 calculates at least one of the average energy of radiation and the average number of photons in the first pixel based on the probability distribution, the first prior probability, and the pixel value of the first pixel. calculate. The energy calculation unit (calculation unit) 112 calculates at least one of the average energy of radiation and the average number of photons in the second pixel based on the probability distribution, the second prior probability, and the pixel value of the second pixel. calculate.

The prior information acquisition unit 111 can also use pixel information around the target pixel for the prior probability log (p, q) of the target pixel. The prior information acquisition unit 111 acquires pixel information around the pixel of interest. The prior information acquisition unit 111 introduces an amount related to the difference between the target pixel and surrounding pixels into the prior probability, and the larger the difference is, the smaller the value of the above-described formula (3) is.

Thereby, the parameters p and q are calculated to values close to the surrounding pixels, and the noise of the image is reduced. The quantity related to the difference between the pixel of interest and the surrounding pixels is also called a regularization term. FIG. 11 shows the relationship between the target pixel and surrounding pixels.

As shown in FIG. 11A, examples of prior probabilities in which pixel information around a pixel of interest is introduced include, for example, the square of the difference between parameters p and q (L2 norm) as shown in Equation (31).

Here, (i, j) is the coordinate of the target pixel. λ _{(2) k, l} , μ _{(2) k, l} are coefficients (λ _{(2) k, l} , μ _{(2) k, l} > 0). p _i , _j is the value of the parameter p at the coordinates (i, j). q _i , _j is the value of the parameter q at the coordinates (i, j). In addition, _{(2) of} λ _{(2) k, l} means that the coefficient is related to the square of the difference.

Since the coefficients -λ _{(2) k, l} , -μ _{(2) k, l} are negative, the pixel of interest p _{i, j} , q _{i, j} and the surrounding pixels p _{i + k, j + k} , q _{i + k, j + k} The smaller the difference, the larger the value of log (p _i , q _j ). Therefore, the values p _{i, j} , q _{i, j} having a small difference from the surrounding pixels are determined to be more likely values.

Further, by differentiating the equation (31), a form that can be substituted into the equations (15), (16), (23), and (24) can be obtained.

An example of calculating the prior probability in which surrounding pixel information is introduced is, for example, a form using an absolute value of a difference (| · |) as shown in Expression (34).

As shown in Expression (35), the absolute value of the difference and the square thereof may be combined.

The coefficients λ _{k, l} , μ _{k, l} are changed according to the distance from the target pixel. For example, as shown in FIG. 11B, the upper left, upper right, lower left, and lower right coefficients for the pixel of interest (λ _{−1, −1} , λ _{1, −1} , λ ₋ if the coefficient λ is related to the parameter p). ₁ , ₁ , λ _1,1 ) are the upper, lower, right, and left coefficients (λ _{0, −1} , λ _0,1 , λ _0,1 if the coefficient λ is related to the parameter p). , Λ _{0, −1} ). That is, the relationship shown in Expression (36) is established.

What is necessary is just to obtain | require the distance between the gravity centers of two pixels about the distance from an attention pixel. Also, in order to improve the resolution in a specific direction, it is necessary to change the coefficients λ _{k, l} , μ _{k, l} .

The coefficients λ _{k, l} , μ _{k, l} and the number of surrounding pixels to be considered can be switched depending on the photographing conditions. In general, the larger the coefficients λ _{k, l} , μ _{k, l} (the smaller the coefficients -λ _{k, l} , -μ _{k, l} ) and the greater the number of surrounding pixels, the more noise However, the sharpness of the image tends to decrease. The coefficients λ _{k, l} , μ _{k, l} are defined as the intensity of the regularization term. Therefore, when the number of images in the time direction is small, when a high frame rate is required, when the dose is small, or when high resolution is not required, the coefficients λ _{k, l} , μ _{k, l} are increased and the surroundings are increased. Increase the number of pixels and give priority to noise reduction. For example, fluoroscopy is the case. In general shooting (still image shooting) that requires detailed description, the coefficients λ _{k, l} , μ _{k, l} are reduced, the number of surrounding pixels is reduced, and noise reduction is prioritized.

Further, the prior information acquisition unit 111 may extract edge information from the pixel value of the image and determine a regularization term coefficient for each pixel. Since the boundary portions having greatly different pixel values generally differ greatly in energy, the prior probability term (regularization term) in which peripheral pixel information is introduced is reduced in the boundary portion to reduce the influence of adjacent pixels.

FIG. 12 is a flow showing details of step 301 (determination of prior probabilities) in FIG. 4 in the present embodiment. The processing from step 501 to step 504 is processing of the prior information acquisition unit 111.

In step 501, a plurality of images are averaged to generate an average image. Since one image is noisy and unsuitable for edge extraction, a plurality of images are averaged to generate an average image.

In step 502, a boundary is extracted from the average image. Specifically, the prior information acquisition unit 111 includes a boundary extraction unit that extracts the boundary of the tissue in the image. For the boundary extraction, a generally used image filter such as a Laplacian filter may be used. The boundary is, for example, a boundary between tissues in the subject such as the lungs and abdomen, and a boundary between the subject and a portion where there is no subject (elementary omission).

In step 503, the spatial distribution of the regularization term coefficients λ _{k, l} , μ _{k, l} of each coordinate is determined. FIG. 13 is a diagram illustrating a method for determining the spatial distribution of the coefficient of the regularization term of each coordinate. FIG. 13A shows the spatial distribution of the intensity of the regularization term when a subject (represented by a white dotted line) is photographed. The spatial distribution is determined for each pixel (the cell in FIG. 13A). When an edge of an image taken through a subject is extracted, it becomes like a black portion of a grid. For the boundary pixels and the non-boundary pixels, the coefficients of the regularization term are determined by different methods. The boundary is where the structure of the subject changes rapidly. Since the boundary pixels show different values from the upper, lower, left, and right pixels, the regularization term becomes smaller. The coefficient of the regularization term of pixels other than the boundary is larger than the regularization term of the boundary. In order to determine the boundary, the prior information acquisition unit 111 holds a threshold value in advance, and the prior information acquisition unit 111 compares the pixel value of the edge with the threshold value, and determines a boundary that exceeds the threshold value.

Based on the spatial distribution of the coefficients determined in step 503, a prior probability formula is determined in step 504 for each pixel. The determination of the prior probability is performed using a prior probability formula that introduces surrounding pixel information such as Formula (31), Formula (34), and Formula (35). In FIG. 13, only the first parameter p and the coefficient λ are shown in order to avoid complexity. The same applies to the second parameter q and the coefficient μ. Although the form of the prior probability is the expression (31), the same applies to the expressions (34) and (35).

When determining the prior probability, the method for determining the coefficient of the regularization term is changed depending on whether or not the target pixel is a boundary. When the pixel of interest is other than the boundary, the regularization term is a coefficient of the regularization term of surrounding pixels to be compared (difference in this example).

As shown in FIG. 13B, the target pixel (i, j) of the pixel A in FIG. 13A is not a boundary. For example, the coefficient λ relating to the comparison term ( _{pi, j−} pi _{+ 1, j} ) ² with the pixel (i + 1, j) on the right side of the target pixel is the coefficient λ of the pixel (i + 1, j) on the right side. _{i + 1, j} . Since the pixel (i + 1, j) is not a boundary, a relatively large value is set for the coefficient λi _{+ 1, j} . The target pixel (i, j) is set so as to be easily influenced by the pixel (i + 1, j) on the right side. On the other hand, the coefficient λ applied to the comparison term (p _{i, j} −p _{i−1, j} ) ² with the pixel (i−1, _j ) on the left side of the target pixel is equal to the pixel (i−1, _j ) on the left side. j) is the coefficient λ _{i−1, j} . Unlike (i−1, j), which is the boundary on the right side, the coefficient λ _{i−1, j} is set to a relatively small value. The pixel of interest (i, j) is set so as not to be affected by the pixel (i-1, j) on the left side.

When the target pixel (i, j) is set as a boundary, coefficients λ _{k, l} , μ _{k, l} representing the size of the regularization term are used as the coefficients of the target pixel (i, j) in the boundary part. The coefficient of the regularization term of the pixel value of the boundary having a low correlation with the upper, lower, left and right pixels needs to be relatively small.

As shown in FIG. 13C, the target pixel (i, j) of the pixel B in FIG. 13A is a boundary. Therefore, a comparison term with the adjacent pixels (i + k, j + l) ((k, l) = (0, −1), (0, +1), (−1, 0), (1, 0)) on the upper, lower, left and right sides. _{_{(p i, j -p i +}} k, j + l) coefficient according to the ² lambda, all target pixel (i, j) be a factor lambda _{i, j} with the. That is, the target pixel (i, j) is set so as not to be affected by the upper, lower, left and right pixels.

Since the prior probability h (p, q) is determined by the steps shown in FIG. 4, the average number of photons based on the steps shown in FIG.

And a proportionality coefficient k (E) that is a function of average energy may be obtained.

FIG. 14 is a flow showing details of step 301 (determination of prior probabilities) in FIG. 4 in the present embodiment. The processing from step 701 to step 704 is the processing of the prior information acquisition unit 111.

In step 701, the prior information acquisition unit 111 acquires part information to be imaged from the operation panel 104. The acquired part information is used in step 702 and step 703. The part information is used to call a regularization term function associated with the imaging part. The regularization term function is a function that represents the pixel value of a pixel and the strength of the regularization term. For example, if the acquired imaging region is a chest front image, a regularization term function for the chest front is called. Details of the regularization term function will be described later. The regularization term function is stored in advance in the prior information storage unit 113.

Next, in step 501, an average image for edge extraction is generated, and in step 502, the boundary is extracted. This step is the same as the step described in FIG.

Step 702, region extraction and regularization term determination. Region extraction can be performed using, for example, a histogram. An example of an image histogram of the front of the chest is shown in FIG. In the histogram of the image in front of the chest, the highest pixel value is a portion where there is no subject (element missing). The next highest pixel value is in the lung field containing a lot of air, and the pixel value is the lowest in the mediastinum or the like. Therefore, by determining which peak (peak) of the histogram the pixel value of the pixel belongs to, it is possible to identify the site where the pixel is imaged. The relationship between the pixel value and the part is stored in the prior information storage unit 113 and acquired in step 701. Note that the region extraction method is not limited to a method using a histogram, and other known methods can be used.

Next, in step 703, the coefficient of the regularization term of each coordinate is determined. The coefficient is determined based on the area determined in step 702. For example, in the case of an image of the front of the chest, there are fine blood vessels. That is, the lung field has a small correlation with surrounding pixels. The coefficient of the regularization term in the lung field needs to be set lower than the coefficient of the regularization term in other parts.

The determination of the regularization term coefficient is performed using a regularization term function. The regularization term function is a function that represents the relationship between the pixel value and the regularization term, and is a function that differs for each imaging region. Therefore, it is called using the information on the imaging region acquired in step 701. The regularization term function is prepared as a curve smoothed by a Bezier curve, for example. FIG. 15B shows a regularization term function.

Also, the regularization term based on boundary extraction is determined in this step. The regularization term coefficient based on the boundary extraction is determined using the method described above. Note that the strength of the regularization term set at the boundary is set to be weaker than the strength of the regularization term determined based on the region. For example, in the lung field, the regularization term has a medium strength. For omissions and others, the strength of the regularization term is increased. The boundary makes the strength of the regularization term small.

In step 704, a prior probability formula is determined. The determination method is performed using the method described above. For example, the boundary uses a regularization term of the boundary, and uses a regularization term of surrounding pixels other than the boundary. Thereafter, the procedure after step 302 in FIG. 4 is executed.

In this embodiment, the description is made assuming a radiation angiography apparatus and a radiography apparatus, but the present embodiment can also be applied to other radiography apparatuses (for example, a mammography apparatus). In addition, the present embodiment can also be applied to a radiation imaging apparatus using radiation other than X-rays, for example, α rays, β rays, and γ rays.

The embodiments according to the present invention have been described above. However, the present invention is not limited to these embodiments, and can be changed or modified within the scope of the claims.

The present invention supplies software (program) that realizes the functions of the above-described embodiments to a system or apparatus via a network or various storage media, and a computer (CPU, MPU, etc.) of the system or apparatus reads the program. May be executed. The present invention can also be realized by a process in which one or more processors in a computer of a system or apparatus read and execute a program, and can also be realized by a circuit (for example, an ASIC) that realizes one or more functions.

This application claims priority from Japanese Patent Application No. 2016-235548 filed on December 5, 2016 and Japanese Patent Application No. 2017-163469 filed on August 28, 2017. The contents of which are incorporated herein by reference.

Claims

A radiographic apparatus that captures radiographic images by radiation,
Determining means for determining a prior probability based on a predetermined condition;
Obtaining means for obtaining a pixel value measured from the radiation image;
Calculating means for calculating at least one of an average energy and an average number of photons of the radiation based on the prior probability and the pixel value;
A radiation imaging apparatus comprising:
The determining means determines a probability distribution of Bayes' theorem and the prior probability,
2. The radiography according to claim 1, wherein the calculating unit calculates at least one of the average energy and the average number of photons of the radiation based on a probability distribution of the Bayes' theorem and the prior probability. apparatus.
3. The radiation imaging apparatus according to claim 2, wherein the probability distribution is a probability distribution related to the number of photons of the radiation.
2. The prior probability includes a first parameter used for calculating an average pixel value of the radiation image and a second parameter used for calculating an average energy of the radiation as variables. 4. The radiographic apparatus according to any one of 3 above.
The probability distribution uses a first parameter used for calculating an average pixel value of the radiation image and a second parameter used for calculating an average energy of the radiation as variables. The radiation imaging apparatus described in 1.
5. The first parameter is at least one of a temporal average average pixel value of predetermined pixels of the radiographic image and a spatial average average pixel value of a plurality of pixels of the radiographic image. Or the radiography apparatus of 5.
The second parameter is at least one of a function having the average energy as a variable, a variance of pixel values of the radiation image, and a standard deviation of pixel values of the radiation image. 5. The radiographic apparatus according to 5.
The radiographic apparatus according to any one of claims 2 to 7, wherein the probability distribution is at least one of a Poisson distribution and a normal distribution.
The calculation means is configured so that when the pixel value is measured, the probability distribution is such that a probability of an event in which the first parameter and the second parameter are the first value and the second value, respectively, is maximized. And calculating the first value and the second value which are solutions of the first parameter and the second parameter of the prior probability. The radiographic apparatus according to the item.
2. The radiation imaging apparatus according to claim 1, wherein the determining unit determines a prior probability based on pixel information around a pixel of interest in the radiation image.
The said determination means has a boundary extraction part which extracts the boundary of the structure | tissue in the said radiographic image, The prior probability is determined based on the boundary of the structure | tissue extracted by the said boundary extraction part. The radiation imaging apparatus described.
When the pixel of interest is a tissue boundary, the intensity of the pixel information is set based on the pixel of interest, and when the pixel of interest is not a tissue boundary, the intensity of the pixel information is pixel information around the pixel of interest. The apparatus according to claim 8, wherein the apparatus is set based on:
13. The prior probability is a maximum between at least one of the first parameters corresponding to the minimum energy and the maximum energy of the radiation and between the second parameters, respectively. The radiation imaging apparatus according to any one of the above.
14. The prior probability is 0 in a range other than at least one between the first parameters and the second parameters corresponding to the minimum energy of the radiation. Radiography equipment.
The prior probability is a maximum in at least one of the first parameter and the second parameter corresponding to the energy of the radiation that has passed through the imaging region of the subject to be imaged. The radiation imaging apparatus according to any one of 4 to 14.
15. The prior probability is a maximum in at least one of the first parameter and the second parameter corresponding to a radiation absorption amount of a subject to be photographed. The radiation imaging apparatus according to item 1.
The prior probability is maximized by at least one of the first parameter and the second parameter corresponding to the energy of the radiation transmitted through the contrast agent injected into the subject to be imaged. The radiation imaging apparatus according to any one of claims 4 to 14.
The determining means includes
Setting the first prior probability to the first pixel of the radiation image;
Setting a second prior probability different from the first prior probability to the second pixel of the radiation image;
The calculating means includes
Calculating at least one of the average energy and the average photon number of the radiation in the first pixel based on the first prior probability and the pixel value of the first pixel;
The at least one of the average energy and the average number of photons of the radiation in the second pixel is calculated based on the second prior probability and the pixel value of the second pixel. The radiation imaging apparatus according to any one of 1 to 17.
The determination means includes radiation generation information relating to the radiation generation means for generating the radiation, radiation detection information relating to the radiation detection means for detecting the radiation, subject information relating to the subject to be photographed, and a photographing part relating to the photographing part of the subject to be photographed. Based on at least one of information, contrast agent information about a contrast agent injected into the subject, support information about a support means for supporting the radiation imaging apparatus or the radiation generating means, and gantry information about a gantry on which the subject is placed The radiographic apparatus according to any one of claims 1 to 18, wherein the prior probability is determined.
The determining means uses the radiation image captured before a predetermined time among the plurality of captured radiation images, the radiation generation information, the radiation detection information, the subject information, the imaging region information, The radiographic apparatus according to claim 19, wherein at least one of the contrast medium information is acquired.
The radiation generation information includes a tube voltage of the radiation generation unit, a tube current of the radiation generation unit, a radiation irradiation time of the radiation generation unit, a radiation spectrum of the radiation generation unit, presence / absence of a radiation filter, and characteristics of the radiation filter 21. The radiographic apparatus according to claim 19 or 20, comprising at least one of the following.
Radiation generating means for generating radiation;
Radiation detection means for detecting radiation;
A radiation imaging system for capturing a radiation image with the radiation,
Determining means for determining a prior probability based on a predetermined condition;
Obtaining means for obtaining a pixel value measured from the radiation image;
Calculating means for calculating at least one of an average energy and an average number of photons of the radiation based on the prior probability and the pixel value;
A radiation imaging system comprising:
A radiography method for taking a radiographic image with radiation,
Determining a prior probability based on a predetermined condition;
Obtaining a pixel value measured from the radiation image;
Calculating at least one of an average energy and an average number of photons of the radiation based on the prior probability and the pixel value;
A radiation imaging method comprising:
A program for causing a computer to function as each unit of the radiation imaging apparatus according to any one of claims 1 to 21.