JP5866958B2 - Radiation tomography apparatus, radiation tomographic image generation processing apparatus, and radiation tomographic image generation program - Google Patents

Description

  The present invention relates to a radiation tomography apparatus, a radiation tomographic image generation processing apparatus, and a radiation tomographic image generation program that collect projection data (projection images) and reconstruct the acquired projection data to generate a radiation tomographic image.

  As a conventional radiation tomography apparatus, for example, there is an X-ray tomography apparatus (for example, see Non-Patent Document 1). The X-ray tomography apparatus includes an X-ray tube that irradiates X-rays toward a subject placed on a top board, and an X-ray detector that detects X-rays transmitted through the subject. As a technique for obtaining an X-ray tomographic image, for example, there is a technique called tomosynthesis, and a plurality of projection data collected in one tomographic image is reconstructed to obtain an X-ray tomographic image of an arbitrary cutting height. (FIG. 9). That is, the X-ray tomography apparatus collects projection data for each preset predetermined angle by translating the X-ray tube and the X-ray detector in opposite directions. The acquired projection data is subjected to image reconstruction by shift addition method, filtered back-projection (FBP) method, ML-EM (maximum likelihood-expectation maximization) method, etc. Is generated.

Takeshi Shiomi, “Principles and Applications of Tomosynthesis: New Technology Created by FPD”, Journal of Medical Image Information Society, Vol.24 No.2, P22-27, 2007

  However, the conventional example having such a configuration has the following problems. Data loss may occur in the imaging stage of conventional radiation tomographic image generation. For example, when a high-absorption body made of metal such as a metal artificial joint, external fixator, or tooth filling is present in the subject, radiation is shielded by the high-absorption body. As a result, the radiation should not reach the detection point of the radiation detector on the extension of the line connecting the radiation source and the high-absorbent body, or only a very small amount of radiation should have arrived. (FIG. 10). The loss of data due to such a phenomenon is called data loss. When a radiation tomographic image is generated using a pixel value of a detection point where data loss has occurred in the same manner as a normal pixel value of another detection point, an artifact occurs.

  As a radiation tomographic image generation method for reducing such artifacts, the above-described shift addition method and, for example, an FBP method in which a low-frequency component is added to a Shepp-Logan filter are known, but a radiation tomography with poor spatial resolution is known. It may be an image, or artifacts may remain, such as a black radiation tomogram around the superabsorber.

  Further, the data loss includes what is called truncation caused by an uncollected portion in the projection data collection trajectory. For example, in the X-ray CT apparatus, the uncollected portion nc in the collection trajectory in FIG. 11 is truncation. In addition, since projection data is acquired every predetermined angle, projection data at an angle between them is also called truncation. When such truncation occurs, in the reconstructed image, the number of projections contributing to the image reconstruction differs depending on the pixel. In this case, the reliability of a pixel with a small number of projections contributing to image reconstruction is lower than that of a pixel with a large number of projections contributing to image reconstruction.

  The present invention has been made in view of such circumstances, and provides a radiation tomography apparatus, a radiation tomographic image generation processing apparatus, and a radiation tomographic image generation program capable of reducing artifacts caused by data loss. The purpose is to do.

In order to achieve such an object, the present invention has the following configuration. That is, the radiation tomography apparatus according to the present invention is a radiation tomography apparatus that generates a radiation tomographic image, an actual projection data collection unit that collects a plurality of actual projection data from different directions with respect to a subject, A data loss extraction unit that extracts data loss pixels from the actual projection data; and when calculating calculated projection data by forward projection from the reconstructed image, the calculated projection data corresponding to the data loss pixels are sequentially And an image reconstruction execution unit that executes image reconstruction by a successive approximation method without performing projection .

According to the radiation tomography apparatus according to the present invention, the actual projection data collection unit collects a plurality of actual projection data from different directions with respect to the subject, and the data defect extraction unit extracts the data defect pixel from the actual projection data. To extract. Then, when calculating the calculated projection data by forward projection from the reconstructed image, the image reconstruction execution unit does not perform forward projection on the pixels of the calculated projection data corresponding to the data missing pixels, and the image by the successive approximation method Perform a reconfiguration. Thereby, it is possible to calculate the calculated projection data excluding the pixels corresponding to the data missing pixels. That is, data missing pixels are extracted from the actual projection data, and the extracted data missing pixels are prevented from contributing to image reconstruction (pixel value update) by the successive approximation method. For this reason, since image reconstruction is performed using normal pixels other than data loss pixels, artifacts due to data loss can be suppressed.

Further, the radiation tomography apparatus according to the present invention is a radiation tomography apparatus for generating a radiation tomography image, an actual projection data collection unit for collecting a plurality of actual projection data from different directions with respect to a subject, A data loss extraction unit that extracts data loss pixels from the actual projection data, and an image reconstruction execution unit that executes image reconstruction by a successive approximation method without performing back projection from pixels corresponding to the data loss pixels. It is characterized by having.
According to the radiation tomography apparatus according to the present invention, the actual projection data collection unit collects a plurality of actual projection data from different directions with respect to the subject, and the data defect extraction unit extracts the data defect pixel from the actual projection data. To extract. Then, the image reconstruction execution unit executes image reconstruction by the successive approximation method without performing back projection from the pixel corresponding to the data missing pixel. Thereby, it is possible to perform back projection using normal pixels excluding the pixels corresponding to the data missing pixels. That is, data missing pixels are extracted from the actual projection data, and the extracted data missing pixels are prevented from contributing to image reconstruction (pixel value update) by the successive approximation method. For this reason, since image reconstruction is performed using normal pixels other than data loss pixels, artifacts due to data loss can be suppressed.

In the radiation tomography apparatus according to the present invention, it is preferable that the image reconstruction execution unit further does not perform back projection from a pixel corresponding to the data missing pixel. Thereby, it is possible to perform back projection using normal pixels excluding the pixels corresponding to the data missing pixels.

  In the radiation tomography apparatus according to the present invention, a projection number setting unit that sets the number of projections that contributes to reconstruct the pixels of the reconstructed image, and a pixel value of the reconstructed image as the number of projections decreases It is preferable that a pixel value update determination unit that determines whether or not to update the pixel value so as not to be updated. That is, the projection number setting unit sets the projection number of the pixel that contributes to reconstruct the pixel of the reconstructed image, and the pixel value update determination unit determines the pixel value of the reconstructed image as the pixel has a smaller projection number. It is determined whether or not to update the pixel value so as not to update. For this reason, pixel values that do not update (execution of image reconstruction) are not updated for pixels that have a smaller number of projections that contribute to image reconstruction. Therefore, pixel values that have a higher number of projections that contribute to reconstruction are prioritized. Will be updated. Thereby, the artifact resulting from the update of the pixel value of a pixel with a small number of projections can be suppressed.

  Further, in the radiation tomography apparatus according to the present invention, when the projection number setting unit collects the actual projection data in each direction, when the projection line passes through the pixel of the reconstructed image, the projection line is Projection is performed at each pixel of the reconstructed image through which the projection line that reaches the pixel of the actual projection data passes when the pixel of the actual projection data is counted as a data deficient pixel. It is preferable not to count the passage of lines. As a result, even if the pixel of the actual projection data is a data defect pixel, even if the pixel counts the passage of the projection line, the projection line passes through each pixel of the reconstructed image through which the projection line that reaches the data defect pixel passes. Is not counted. For this reason, since the number of projections is reduced, it is possible to suppress artifacts due to the update of the pixel values of the pixels having a small number of projections.

  Further, in the radiation tomography apparatus according to the present invention, the projection number setting unit sets the projection number and the maximum value thereof, and the pixel value update determination unit is determined by a ratio based on the projection number and the maximum value. It is preferable to do. Thereby, whether or not to update the pixel value can be determined by the ratio based on the number of projections and the maximum value.

  In the radiation tomography apparatus according to the present invention, it is preferable that the pixel value update determination unit determines the ratio based on the number of projections and a random number. As a result, the timing at which the pixel value is not updated can be made variable without being fixed.

A radiation tomographic image generation processing device according to the present invention is a radiation tomographic image generation processing device that generates a radiation tomographic image based on a plurality of actual projection data collected from different directions with respect to a subject, A data loss extraction unit that extracts data loss pixels from actual projection data , and forward projection on the pixels of the calculated projection data corresponding to the data loss pixels when calculating the calculation projection data by forward projection from the reconstructed image And an image reconstruction execution unit that performs image reconstruction by the successive approximation method without performing the above.
According to the radiation tomographic image generation processing device of the present invention, the data loss extraction unit extracts data loss pixels from a plurality of actual projection data collected from different directions with respect to the subject. Then, when calculating the calculated projection data by forward projection from the reconstructed image, the image reconstruction execution unit does not perform forward projection on the pixels of the calculated projection data corresponding to the data missing pixels, and the image by the successive approximation method Perform a reconfiguration. Thereby, it is possible to calculate the calculated projection data excluding the pixels corresponding to the data missing pixels. That is, data missing pixels are extracted from the actual projection data, and the extracted data missing pixels are prevented from contributing to image reconstruction (pixel value update) by the successive approximation method. For this reason, since image reconstruction is performed using normal pixels other than data loss pixels, artifacts due to data loss can be suppressed.

A radiation tomographic image generation processing device according to the present invention is a radiation tomographic image generation processing device that generates a radiation tomographic image based on a plurality of actual projection data collected from different directions with respect to a subject, A data loss extraction unit that extracts data loss pixels from actual projection data; and an image reconstruction execution unit that executes image reconstruction by a successive approximation method without performing back projection from pixels corresponding to the data loss pixels. It is characterized by that.
According to the radiation tomographic image generation processing device of the present invention, the data loss extraction unit extracts data loss pixels from a plurality of actual projection data collected from different directions with respect to the subject. Then, the image reconstruction execution unit executes image reconstruction by the successive approximation method without performing back projection from the pixel corresponding to the data missing pixel . Thereby, it is possible to perform back projection using normal pixels excluding the pixels corresponding to the data missing pixels. That is, data missing pixels are extracted from the actual projection data, and the extracted data missing pixels are prevented from contributing to image reconstruction (pixel value update) by the successive approximation method. For this reason, since image reconstruction is performed using normal pixels other than data loss pixels, artifacts due to data loss can be suppressed.

Further, the radiation tomographic image generation program according to the present invention is a radiation tomographic image generation for causing a computer to execute a process of generating a radiation tomographic image based on a plurality of actual projection data collected from different directions with respect to a subject. In the program, when calculating the calculated projection data by forward projection from the reconstructed image and extracting the data missing pixels from the actual projection data, the calculated projection data corresponding to the data missing pixels And a step of performing image reconstruction by a successive approximation method without performing forward projection .
According to the radiation tomographic image generation program according to the present invention, data missing pixels are extracted from a plurality of actual projection data collected from different directions with respect to the subject. Then, when calculating the calculated projection data from the reconstructed image by forward projection, image reconstruction by the successive approximation method is executed without performing forward projection on the pixels of the calculated projection data corresponding to the data-missing pixels. Thereby, it is possible to calculate the calculated projection data excluding the pixels corresponding to the data missing pixels. That is, data missing pixels are extracted from the actual projection data, and the extracted data missing pixels are prevented from contributing to image reconstruction (pixel value update) by the successive approximation method. For this reason, since image reconstruction is performed using normal pixels other than data loss pixels, artifacts due to data loss can be suppressed.

Further, the radiation tomographic image generation program according to the present invention is a radiation tomographic image generation for causing a computer to execute a process of generating a radiation tomographic image based on a plurality of actual projection data collected from different directions with respect to a subject. A program, comprising: extracting data-deficient pixels from the actual projection data; and executing image reconstruction by a successive approximation method without performing back projection from pixels corresponding to the data-deficient pixels. It is characterized by being.
According to the radiation tomographic image generation program according to the present invention, data missing pixels are extracted from a plurality of actual projection data collected from different directions with respect to the subject. Then, image reconstruction by the successive approximation method is executed without performing back projection from the pixel corresponding to the data-missing pixel . Thereby, it is possible to perform back projection using normal pixels excluding the pixels corresponding to the data missing pixels. That is, data missing pixels are extracted from the actual projection data, and the extracted data missing pixels are prevented from contributing to image reconstruction (pixel value update) by the successive approximation method. For this reason, since image reconstruction is performed using normal pixels other than data loss pixels, artifacts due to data loss can be suppressed.

  According to the radiation tomography apparatus, the radiation tomographic image generation processing apparatus, and the radiation tomographic image generation program according to the present invention, data missing pixels are extracted from a plurality of actual projection data collected from different directions with respect to the subject. Then, image reconstruction by the successive approximation method is executed without using the data missing pixel. That is, data missing pixels are extracted from the actual projection data, and the extracted data missing pixels are prevented from contributing to image reconstruction (pixel value update) by the successive approximation method. For this reason, since image reconstruction is performed using normal pixels other than data loss pixels, artifacts due to data loss can be suppressed.

It is a block diagram which shows the structure of the X-ray tomography apparatus which concerns on an Example. It is a figure where it uses for description of a detection probability setting part. (A) is a diagram for explaining the detection probability in the case of a normal pixel, and (b) is a diagram for explaining the detection probability in the case of a data loss pixel. (A)-(d) is a figure where it uses for description of a projection number setting part. (A) is a figure used for description of forward projection with respect to the pixel corresponding to a data defect pixel, (b) is a figure for description of back projection from the pixel corresponding to a data defect pixel. It is a flowchart which shows operation | movement of a X-ray tomography apparatus. It is a figure which shows an example of the radiation tomography apparatus which concerns on a modification. It is a figure which shows an example of the radiation tomography apparatus which concerns on a modification. It is a figure where it uses for description of the conventional X-ray tomography. It is a figure with which it uses for description when a high absorber exists between the conventional radiation source and a radiation detector. It is a figure which shows an example of the radiation tomography apparatus which concerns on a modification while showing an example of truncation.

  Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration of an X-ray tomography apparatus according to an embodiment. FIG. 2 is a diagram for explaining the detection probability setting unit. FIG. 3A is a diagram for explaining the detection probability in the case of a normal pixel, and FIG. 3B is a diagram for explaining the detection probability in the case of a data loss pixel. 4A to 4D are diagrams for explaining the projection number setting unit. FIG. 5A is a diagram for explaining forward projection with respect to a pixel corresponding to a data defect pixel, and FIG. 5B is a diagram for explaining back projection from a pixel corresponding to the data defect pixel. In this embodiment, an X-ray tomography apparatus will be described as an example of a radiation tomography apparatus.

  Please refer to FIG. The X-ray tomography apparatus 1 is disposed so as to face the top plate 2 on which the subject M is placed, the X-ray tube 3 that irradiates the subject M with X-rays, and the X-ray tube 3. And a flat panel X-ray detector (hereinafter abbreviated as “FPD” as appropriate) 4 for detecting X-rays transmitted through M.

  The X-ray tube 3 is subjected to control necessary for X-ray irradiation by the X-ray tube controller 5. The X-ray tube control unit 5 includes a high voltage generation unit 6 that generates a tube voltage and a tube current of the X-ray tube 3. The X-ray tube control unit 5 emits X-rays from the X-ray tube 3 in accordance with X-ray irradiation (exposure) conditions such as tube voltage, tube current, and irradiation time.

  The FPD 4 has a large number of X-ray detection elements (DUs in FIGS. 2 to 5) that convert X-rays into electrical signals and detect them on an X-ray detection surface on which a transmission X-ray image to be detected is projected. Are arranged in a two-dimensional matrix. Examples of the array matrix of the X-ray detection elements include horizontal: several thousand × vertical: several thousand. The X-ray detection element may be a direct conversion type in which X-rays are directly converted into electric signals, or an indirect conversion type in which X-rays are once converted into light and then converted into electric signals.

  The X-ray tube 3 and the FPD 4 move in parallel along the body axis ax of the subject M in FIG. The X-ray tube 3 and the FPD 4 are translated in a synchronous manner and in opposite directions. The X-ray tube 3 and the FPD 4 are moved by a rack, pinion, motor, or the like (not shown). The FPD 4 collects a plurality of X-ray detection signals from different directions (angles) with respect to the subject M. That is, the FPD 4 moves in the reverse direction in synchronization with the X-ray tube 3 and collects a plurality of X-ray detection signals by X-rays from the X-ray tube 3 irradiated from different directions with respect to the subject M. . The X-ray detection signal is actual projection data. The FPD 4 corresponds to the actual projection data collection unit of the present invention.

  An A / D converter 7, an image processing unit 8, and a main control unit 9 are provided in the subsequent stage of the FPD 4. The A / D converter 7 converts each analog X-ray detection signal output from the FPD 4 into a digital signal. The image processing unit 8 performs various necessary processes on the X-ray image based on the digitally converted X-ray detection signal. The main control unit 9 comprehensively controls each component of the X-ray tomography apparatus 1 and includes a central processing unit (CPU) and the like. The main control unit 9 performs control to move the top plate 2, the X-ray tube 3 or the FPD 4, for example. Similarly, the top plate 2 is moved by a rack, pinion, motor, or the like (not shown).

  The X-ray tomography apparatus 1 includes a display unit 11, an input unit 12, and a storage unit 13. The display unit 11 includes a monitor or the like. The input unit 12 includes a keyboard, a mouse, and the like. The storage unit 13 includes a storage medium such as a ROM (Read-only Memory), a RAM (Random-Access Memory), or a hard disk. The storage unit 13 stores a plurality of X-ray images for different directions.

  The X-ray tomography apparatus 1 includes an X-ray tomographic image generation unit 20 that generates X-ray tomographic images based on a plurality of X-ray images from different directions with respect to the subject M, that is, a plurality of actual projection data. Yes. The X-ray tomographic image generation unit 20 includes a data loss extraction unit 21, a projection number setting unit 23, and an image reconstruction unit 25. Note that the image reconstruction unit 25 generates an X-ray tomographic image (reconstructed image) by image reconstruction by the successive approximation method (sequential approximation image reconstruction method) based on the actual projection data. The ML-EM method is used as the successive approximation image reconstruction method. The X-ray tomographic image generation unit 20 corresponds to the radiation tomographic image generation processing device of the present invention.

  The data loss extraction unit 21 extracts data loss pixels from each of the plurality of actual projection data. In other words, the data loss extraction unit 21 extracts pixels having abnormal pixel values as data loss pixels from the collected projection data compared to other pixels. For example, a detection point on an extension line of a line connecting the X-ray focal point (X-ray source) of the X-ray tube 3 and a high absorber such as metal, that is, a pixel detected by the X-ray detection element DU of the FPD 4 is mentioned. . Since this pixel is shielded by the high absorber and X-rays do not reach or a very small amount of X-rays reach, the pixel value is close to 0 (zero). Therefore, a threshold value is provided, and the pixel direction of the actual projection data is extracted with the pixel having such a value as a data defect pixel. The data-missing pixel is made to know which projection direction and which pixel of the actual projection data. Similarly, when the pixel value indicates an abnormally high value or an abnormally low value due to the X-ray detection element DU or the like of the FPD 4, similarly, a threshold value is set in advance and a data-missing pixel is extracted.

The projection number setting unit 23 determines the number of projections that contribute to image reconstruction of a certain pixel in the X-ray tomographic image (reconstructed image) generated based on the actual projection data and the maximum value of the projection number. Set. The number of projections is represented by the sum Σ _i a _ij (including the detection probability a _ij ) of probability detection. The projection number setting unit 23 includes a detection probability setting unit 27, an adding unit 29, and a maximum value setting unit 31.

The detection probability setting unit 27 sets a detection probability a _ij that contributes to reconstruct a pixel (voxel or pixel) j of a reconstructed image (X-ray tomographic image) λ _j from a predetermined direction. As shown in FIG. 2, the detection probability setting unit 27 connects the X-ray focal point of the X-ray tube 3 and the X-ray detection element DU of the FPD 4 when collecting actual projection data in each direction with respect to the subject M. Suppose that there is a projection line. When this projection line passes through the pixel j of the reconstructed image λ _j , the passage of the projection line at each pixel j is counted as a _ij = 1, and when it does not pass, a _ij = 0 is set so as not to be counted. To do. In the present embodiment, a method that does not consider out-of-focus or scattered rays is shown, but detection probability may be set in consideration of these. Further, the detection probability may be set by other known methods.

In addition, as shown in FIGS. 3A and 3B, the detection probability setting unit 27 detects that the pixel i of the actual projection data y _i detected by the X-ray detection elements DU1 to DU3 of the FPD 4 is a data-missing pixel. In some cases, the projection line passing through the projection line reaching the pixel i of the actual projection data y _i is set not to count the passage of the projection line in each pixel j of the reconstructed image λ _j . That is, as shown in FIG. 3A, when the pixel i of the actual projection data y _{i is} a normal pixel without data loss (symbol ◯), the projection line that reaches the pixel i of the actual projection data y _i passes. The passage of the projection line at each pixel j of the reconstructed image λ _j is counted as a _ij = 1. On the other hand, as shown in FIG. 3B, when the pixel i of the actual projection data y _i is a data-missing pixel (symbol ×), the reconstruction that passes the projection line that reaches the pixel i of the actual projection data y _i passes. The passage of the projection line at each pixel j of the image λ _j is not counted as a _ij = 0. Note that the detection probability a _ij in each direction that does not consider the data loss pixel may be set in advance and stored in the storage unit 13. From the detection probability a _ij in each direction that does not consider the data loss, the data loss pixel The passage of the projection line may not be counted as a _ij = 0 at each pixel j of the reconstructed pixel λ _j through which the projection line arriving at is passed.

The adding unit 29 obtains a sum Σ _i a _ij of detection probabilities collected in each direction. For example, in the case of collecting actual projection data whose projection directions on the subject M are 0 °, 1 °, 2 °,..., D, the detection probabilities a _ij in all these directions are added to obtain the sum Σ _i a _ij . . As shown in FIGS. 4A to 4C, for example, the projection directions are set to −45 °, 0 °, and 45 °. In each projection direction, for the convenience of explanation, the detection probability a _ij is set assuming that X-rays are incident only on the X-ray detection element DU2 of the PPD4. The total detection probability Σ _i a _ij is obtained by adding the detection probabilities a _ij in all projection directions (−45 °, 0 °, 45 °) as shown in FIG.

The maximum value setting unit 31 extracts the maximum value max _j Σ _i a _ij of the total detection probability Σ _i a _{ij from} the reconstructed image λ _j . For example, in the case of FIG. 4D, max _j Σ _i a _ij = 3. Note that the maximum value max _j Σ _i a _ij is a value that does not include the case where the number of projections is not counted based on the data-missing pixel in FIG.

Returning to FIG. The image reconstruction unit 25 generates an X-ray tomographic image (reconstructed image) by the ML-EM method as a successive approximation image reconstruction method based on the actual projection data y _i . The image reconstruction unit 25 includes a random number generation unit 33, a pixel value update determination unit 35, and an image reconstruction execution unit 37. First, the image reconstruction execution unit 37 will be described.

The image reconstruction execution unit 37 performs image reconstruction by the ML-EM method. That is, the image reconstruction execution unit 37 updates the pixel value of the reconstruction image λ _j (X-ray tomographic image). The calculation formula of the ML-EM method is represented by the following formula (1).

λ_ｊ ^{（ｋ＋１）}：ｋ＋１回目の再構成画像
λ_ｊ ^（ｋ）：ｋ回目の再構成画像
Σ_ｊ′＝１ ^ｍａ_ｉｊ′λ_ｊ′ ^（ｋ）：順投影
Σ_ｉ＝１ ^ｎ〔（ｙ_ｉａ_ｉｊ）／（Σ_ｊ′＝１ ^ｍａ_ｉｊ′λ_ｊ′ ^（ｋ））〕：逆投影
Σ_ｉ＝１ ^ｎａ_ｉｊ：検出確率ａ_ｉｊの総和
ｙ_ｉ：実投影データ
ｉ：実投影データの画素（１〜ｎまでの通し番号が付され、ｎはＦＰＤの一列のＸ線検出素子数×投影方向数で算出される）
ｊ：再構成画像の画素（１〜ｍまでの通し番号が付される）

^{_{λ j (k + 1):}} k + 1 -th reconstructed image λ _j ^{(k): k-th} reconstructed image _{^{Σ j '= 1 m a ij}} ' λ j '(k): the forward projection Σ _{i =} ^{1 n} [(y _i a _ij ) / (Σ _{j ′ = 1} ^m a _{ij ′} λ _{j ′} ^(k) )]: Back projection Σ _{i = 1} ⁿ a _ij : Sum of detection probabilities a _ij y _i : Actual projection data i: Actual projection Data pixels (serial numbers from 1 to n are assigned, where n is calculated by the number of X-ray detection elements in one column of the FPD × the number of projection directions)
j: Reconstructed image pixels (serial numbers from 1 to m are given)

Based on equation (1), updating of each pixel j in the reconstructed image lambda j is performed in the following procedure. First, an initial value of each pixel j of the reconstructed image λ j (k) is assumed. For example, the initial value of all the pixels j is set to λ j (k) = 1. Next, projection data is calculated by forward projection from the reconstructed image λ j (k) (λ j ′ (k) ). The calculated projection data is referred to as calculated projection data. The ratio between the actual projection data y i and the calculated projection data is calculated. The ratio between the actual projection data y i and the calculated projection data is back projected. The backprojected value calculated by normalizing the sum Σ i = 1 n a ij of detection probability, multiplied by the reconstructed image lambda j (k) to calculate the reconstructed image λ j (k + 1) and. When updating each pixel j of the reconstructed image λ j repeatedly for a preset number of times, the calculated projection data is calculated by forward projection with the calculated λ j (k + 1) as λ j (k). Return to the process.

Note that the image reconstruction execution unit 37 repeats the pixel value update (image reconstruction) for all the pixels _j of the reconstructed image λ _j , but the pixel j for which determination is established by the pixel value update determination unit 35 described later. Only the pixel value is updated. In addition, the image reconstruction execution unit 37 repeats the pixel value update of the reconstructed image λ _j by a preset number of times.

Further, in this series of procedures, the image reconstruction execution unit 37 updates the pixel value j of the reconstructed image λ _j without using the pixel corresponding to the data defect pixel (the pixel at the data defect position) (image). Perform reconfiguration). That is, the image reconstruction execution unit 37 includes at least one of the following two restrictions. First, when calculating the calculated projection data from the reconstructed image λ _j by forward projection, the image reconstruction execution unit 37 does not perform forward projection on the pixels of the calculated projection data corresponding to the data-missing pixels. . That is, as shown in FIG. 5A, first, it is assumed that data loss occurs in, for example, the pixel i = 5 of the actual projection data y _i detected by the X-ray detection element DU2. In this case, the value calculated by the pixel value of λ ₂ + λ ₅ + λ ₈ corresponding to the data-missing pixel is set as the pixel value of the pixel i = 5 having the same number in the calculated projection data cp, and forward projection is not performed. For example, the pixel value of the data missing pixel in the actual projection data y _i is substituted.

On the other hand, as the second, the image reconstruction execution unit 37 does not perform back projection from the pixel corresponding to the data missing pixel. Specifically, for example, when the ratio between the actual projection data y _i and the calculated projection data is back-projected, the back projection is not performed from the pixel i corresponding to the data missing pixel of the actual projection data y _i . That is, as shown in FIG. 5B, first, similarly, assume that data loss occurs in, for example, pixel i = 5 in the actual projection data y _i detected by the X-ray detection element DU2. In this case, the back projection is not performed from the pixel i = 5 having the same number corresponding to the data missing pixel i = 5. For example, as shown in FIG. 3B, do not count the passage of the projection line as a _ij = 0 in each pixel j of the reconstructed image λ _j through which the projection line that reaches the data-missing pixel (symbol x) passes. A detection probability a _ij is set. Thus, back projection is not performed from pixel i = 5 in the back projection calculation of equation (1).

Such processing of the image reconstruction execution unit 37 is executed by the determination of the pixel value update determination unit 35. A random number is used to determine whether or not the pixel value update determination unit 35 updates the pixel value. The random number generator 33 generates a random number p in the range of 0 ≦ p <1. Pixel value updating determination unit 35 updates the pixel value by the ratio based on the sum sigma _i a _ij probability of detection not to update the pixel values of the higher sum sigma _i a _ij is small pixel detection probability reconstructed image lambda _j It is determined whether or not. Specifically, the ratio based on the total detection probability Σ _i a _ij is a ratio based on the total detection probability Σ _i a _ij (the number of projections) and its maximum value max _j Σ _i a _ij . That is, the pixel value update determination unit 35 compares the sum of detection probabilities Σ _i a _ij (the number of projections) and the ratio based on the maximum value max _j Σ _i a _ij with the random number p generated by the random number generation unit 33. Judge by.

More specifically, a ratio (a value corresponding to the number of projections) based on the sum Σ _i a _ij of detection probabilities contributing to constructing a predetermined pixel j of the reconstructed image λ _j and its maximum value max _j Σ _i a _ij ) _Is used as 0.5 × (1 + Σ _i a _ij / max _j Σ _i a _ij ). That is, the pixel value update determination unit 35 determines whether or not to update the pixel value depending on whether or not the following formula (2) is satisfied.
p <0.5 × (1 + Σ i a ij / max j Σ i a ij) ... (2)

For example, when p = 0.85, Σ _ia a _ij = 80, max _j Σ _ia a _ij = 100, 0.85 <0.9 = [0.5 × (1 + 80/100) according to Equation (2) ] And the formula is established. When this expression is established, the pixel value update determination unit 35 determines to update the pixel value. On the other hand, if the expression does not hold, the pixel j is not updated and the next pixel j is determined. When there is no missing data, the pixel value is always updated by the ML-EM method. The possibility that the pixel value is updated decreases according to the amount of missing data.

  Next, the operation of the X-ray tomography apparatus 1 will be described with reference to the flowchart of FIG.

[Step S01] Collection of Actual Projection Data y _i The X-ray tube 3 and the FPD 4 are translated in parallel and in opposite directions along the body axis ax of the subject M in FIG. At that time, a plurality of actual projection data y _i from different directions with respect to the subject M is collected. The collected actual projection data y _i is subjected to necessary image processing and temporarily stored in the storage unit 13.

[Step S02] Extraction of Data Loss Pixel The data loss extraction unit 21 extracts data loss pixels from each of the plurality of actual projection data. For example, pixels of actual projection data detected by the X-ray detection element DU of the FPD 4 on the extension line of the line segment connecting the X-ray focal point (X-ray source) of the X-ray tube 3 and a high absorber such as metal, or the FPD 4 Pixels indicating abnormal values due to the X-ray detection elements DU and the like are extracted. Whether or not there is data loss is determined using, for example, a threshold value.

[Step S03] Setting Detection Probability _aij The detection probability setting unit 27 sets a detection probability _aij that contributes to reconstruct the pixels of the reconstructed image from a predetermined direction. As shown in FIG. 2, the detection probability setting unit 27 uses the projection line when the projection line when collecting the actual projection data y _i in each direction with respect to the subject passes through the pixel j of the reconstructed image λ _j. For each pixel j that passes, the passage of the projection line is counted as a _ij = 1, and when it does not pass, a _ij = 0 is set so as not to be counted.

In addition, when the pixel i of the actual projection data y _i detected by the X-ray detection element DU of the FPD 4 is a data deficient pixel, the detection probability setting unit 27 detects the projection line that reaches the pixel i of the actual projection data y _i. Setting is made so that the passage of the projection line is not counted in each pixel j of the reconstructed image λ _j that passes. That is, as shown in FIG. 3B, when the pixel i of the actual projection data y _i detected by the X-ray detection element DU2 is a data deficient pixel (symbol x), the reconstructed image λ through which the projection line passes. the passage of the projection beam at each pixel j in the _j so as not counted as a _{ij = 0.}

[Step S04] Setting of Detection Probabilities Sum Σ _i a _{ij As} shown in FIG. 4D, the adder 29 calculates the sum of detection probabilities Σ _i a _ij collected in each direction. That is, the total detection probability Σ _i a _ij is the number of projections collected in different directions with respect to the subject M.

[Step S05] Setting of Maximum Value max _j Σ _i a _ij The maximum value setting unit 31 extracts the maximum pixel value max _j Σ _i a _ij from the total detection probability Σ _i a _ij of the reconstructed image λ _j. To set. For example, in the case of FIG. 4D, max _j Σ _i a _ij = 3 is set. Note that the maximum value max _j Σ _i a _ij is a value that does not include the case where the number of projections is not counted based on the data-missing pixel in FIG.

Steps S06 to S10 are steps for generating an X-ray tomographic image by the ML-EM method based on the collected actual projection data y _i .

[Step S06] Generation of Random Number p The random number generator 33 generates a random number p in the range of 0 ≦ p <1.

[Step S07] Pixel Value Update Determination The pixel value update determination unit 35 sets the detection probability sum Σ _i a _ij so as not to update the pixel value of the reconstructed image λ _j for pixels with a smaller detection probability sum Σ _i a _ij . It is determined whether or not to update the pixel value based on the ratio based. That is, the pixel value update determination unit 35 determines the comparison by comparing the sum of detection probabilities Σ _i a _ij and the ratio based on the maximum value max _j Σ _i a _ij with the random number p. Specifically, the determination is made based on whether or not Expression (2) is satisfied. If it is established, the process proceeds to step S08. If not, the pixel value is not updated for the pixel i, and the process proceeds to step S09.

[Step S08] Update Pixel Value The image reconstruction execution unit 37 updates the pixel value i of the reconstructed image λ _j by the ML-EM method (equation (1)) without using the data-missing pixel (image reconstruction). Execution). That is, the image reconstruction execution unit 37 includes at least one of the following two restrictions. First, as shown in FIG. 5A, the image reconstruction execution unit 37 calculates calculated projection data corresponding to a data-missing pixel when calculating calculated projection data from the reconstructed image λ _j by forward projection. No forward projection is performed on the pixel i. On the other hand, as shown in FIG. 5B, the image reconstruction execution unit 37 does not perform back projection from the pixel corresponding to the data loss pixel.

[Step S09] all pixel processing determination image reconstruction unit 25 determines whether or not the processing of the existence of pixel value updating for all pixels j of the reconstructed image lambda _j. When the process for determining whether or not the pixel value is updated is performed for all the pixels j, the process proceeds to step S10. If the process for determining whether or not the pixel value has been updated is not performed for all the pixels j, the process returns to step S06, and the process for the next pixel j is performed.

[Step S10] Update Count Determination The image reconstruction unit 25 determines whether or not a preset number of pixel value updates of the reconstructed image λ _j has been performed. When the pixel value is updated a preset number of times for the reconstructed image λ _j , the process proceeds to END. If the preset number of times is not satisfied, the process returns to step S06.

  With the above operation, the X-ray tomographic image generation unit 20 generates an X-ray tomographic image. The generated X-ray tomographic image is stored in the storage unit 13 or an X-ray tomographic image having an arbitrary cutting height is displayed on the display unit 11.

According to the X-ray tomography apparatus 1 according to the present embodiment, the FPD 4 collects a plurality of actual projection data y _i from different directions with respect to the subject M, and the data loss extraction unit 21 uses the actual projection data y. Data deficient pixels are extracted from each _i . Then, the image reconstruction execution unit 37 updates the pixel value of the reconstructed image λ _j by the ML-EM method (sequential approximate image reconstruction method) without using the data loss pixel (execution of image reconstruction). That is, the data missing pixel of the actual projection data y _i is extracted, and the extracted data missing pixel is not contributed to the pixel value update (image reconstruction) by the ML-EM method. Therefore, since the pixel value is updated with the normal pixel i other than the pixel corresponding to the data missing pixel, artifacts due to the data missing can be suppressed.

Further, when calculating the calculated projection data by forward projection from the reconstructed image λ _j , the image reconstruction execution unit 37 does not perform forward projection on the pixel i of the calculated projection data corresponding to the data missing pixel. Thereby, the calculated projection data excluding the pixel i corresponding to the data-missing pixel can be calculated. Further, the image reconstruction execution unit 37 does not perform back projection from the pixel i corresponding to the data missing pixel. Thereby, it is possible to perform back projection using a normal pixel i excluding the pixel i corresponding to the data missing pixel.

The projection number setting unit 23 sets the sum Σ _i a _ij of the detection probabilities of the pixel j that contributes to reconstruct the pixel j of the reconstructed image λ _j , and the pixel value update determination unit 35 It is determined whether or not to update the pixel value so that the pixel value of the reconstructed image λ _j is not updated as the pixel j has a smaller total probability Σ _i a _ij . Therefore, since the pixel value is not updated as the pixel j has a smaller total detection probability Σ _i a _ij that contributes to image reconstruction, the reliability is higher because the total detection probability Σ _i a _ij that contributes to reconstruction is larger. The pixel value is preferentially updated from the highly efficient pixel j. As a result, artifacts resulting from the update of the pixel value of the pixel j having a small detection probability sum Σ _i a _ij can be suppressed.

Further, when the projection number setting unit 23 collects the actual projection data y _i in each direction, if the projection line passes through the pixel j of the reconstructed image λ _j , the projection number setting unit 23 projects at each pixel j through which the projection line passes. counting the passage of the line, when the pixel i of the actual projection data y _i is data defective pixel at each pixel j in the reconstructed image lambda _j projection rays reaching the pixel i of the actual projection data y _i passes Does not count the passage of projection lines. Thus, even pixel j for counting the passage of projection line, when the pixel i of the actual projection data y _i is data defective pixel, each of the reconstructed image lambda _j the projection line to reach the data defective pixel passes Do not count the passage of projection lines at pixel j. Therefore, since the sum Σ _i a _ij of the detection probabilities is reduced, an artifact caused by the update of the pixel value of the pixel j having a small sum Σ _i a _ij of the detection probabilities can be suppressed.

Further, the pixel value update determination unit 35 compares the random number p with a ratio based on the sum Σ _i a _ij of the detection probabilities and the maximum value max _j Σ _i a _ij of the sum of the detection probabilities as shown in Expression (2). Judging by. Thereby, it is possible to change (random) without fixing the timing at which the pixel value is not updated.

  The present invention is not limited to the above embodiment, and can be modified as follows.

  (1) In the embodiment described above, in the X-ray tomography apparatus 1, the X-ray tube 3 and the FPD 4 are translated along the body axis ax of the subject M synchronously and in opposite directions to obtain actual projection data. Collected. However, as shown in FIG. 7, the data may be collected by rotating at a preset angle synchronously and in opposite directions. The X-ray tomography apparatus 1 may be an X-ray CT apparatus shown in FIG. In this case, the X-ray CT apparatus includes an X-ray tube 103 and an X-ray detector 104 arranged to face each other with the subject M interposed therebetween, and collects actual projection data while rotating around the subject M.

  (2) In the above-described embodiments, actual projection data is collected by detecting X-rays as radiation. However, actual projection data may be collected by detecting γ rays emitted from the subject as radiation. In this case, for example, as shown in FIG. 8, the apparatus includes two (plurality) γ-ray detectors 53 and 54 arranged with the subject M interposed therebetween, and is rotated around the subject M. You may make it acquire projection data.

  (3) In the embodiment described above, the successive approximation image reconstruction method uses the ML-EM method, but is not limited to this. As the successive approximation image reconstruction method, for example, a simulated annealing method, an OS-EM (ordered subsets-expectation maximization) method, a RAMLA (row-action maximum likelihood algorithm) method, or a DRAMA (dynamic RAMLA) method is used. May be. In the simulated annealing method, the above-described embodiments can be applied in forward projection.

(4) In the above-described embodiment, the pixel value update determination unit 35 calculates a value between 0.5 and 1 calculated by 0.5 × (1 + Σ _i a _ij / max _j Σ _i a _ij ) in Expression (2). Compared with random numbers, the timing at which pixel values are not updated is variable without being fixed. However, the timing at which pixel values are not updated may be fixed without being compared with random numbers. Further, the pixel value update determination unit 35 may determine whether or not to update the pixel value from the second and subsequent pixel value updates.

(5) In the above-described embodiment, the pixel value update of the reconstructed image λ _j is performed a preset number of times. However, it may be determined to converge so that the pixel value update is terminated when the change in the pixel values is equal to or less than a predetermined value by comparing before and after the reconstructed image λ _j .

  (6) In the above-described embodiment, the X-ray tomographic image generation unit 20 is configured separately from the image processing unit 8. However, the X-ray tomographic image generation unit 20 may be provided in the image processing unit 8.

  (7) In the embodiment described above, the X-ray tomographic image generation unit 20 includes a control unit configured by a CPU and the like, an input unit, a display unit, and a storage configured by a storage medium such as a ROM, a RAM, and a hard disk. And a workstation or personal computer equipped with a unit and the like. The operations in steps S02 to S10 described above are stored as a program in those storage units, and an X-ray tomographic image is generated by executing the program in the control unit. 1 may be stored in the storage unit 13 of the X-ray tomography apparatus 1 of FIG. 1 as a program, and the main control unit 9 may execute the program. In addition, the operation program of steps S02 to S10 may be executed on a workstation or personal computer connected to the X-ray tomography apparatus 1 by a network system such as a LAN.

DESCRIPTION OF SYMBOLS 1 ... X-ray tomography apparatus 3 ... X-ray tube 4 ... Flat panel type X-ray detector (FPD)
DESCRIPTION OF SYMBOLS 9 ... Main control part 20 ... X-ray tomographic image generation part 21 ... Data defect extraction part 23 ... Projection number setting part 25 ... Image reconstruction part 27 ... Detection probability setting part 29 ... Addition part 31 ... Maximum value setting part 33 ... Random number Generation unit 35 ... pixel value update determination unit 37 ... image reconstruction execution unit

Claims (11)

A radiation tomography apparatus for generating a radiation tomographic image,
An actual projection data collection unit for collecting a plurality of actual projection data from different directions with respect to the subject;
A data loss extraction unit for extracting data loss pixels from the actual projection data;
When calculating the calculated projection data by forward projection from the reconstructed image, image reconstruction is performed by performing image reconstruction by the successive approximation method without performing forward projection on the pixels of the calculated projection data corresponding to the data missing pixels. A configuration execution unit;
A radiation tomography apparatus comprising: A radiation tomography apparatus for generating a radiation tomographic image,
An actual projection data collection unit for collecting a plurality of actual projection data from different directions with respect to the subject;
A data loss extraction unit for extracting data loss pixels from the actual projection data;
An image reconstruction execution unit that performs image reconstruction by a successive approximation method without performing back projection from a pixel corresponding to the data-missing pixel;
A radiation tomography apparatus comprising: The radiation tomography apparatus according to claim 1 ,
The radiation tomography apparatus according to claim 1, wherein the image reconstruction execution unit further does not perform back projection from a pixel corresponding to the data missing pixel. In the radiation tomography apparatus in any one of Claim 1 to 3,
A projection number setting unit that sets the number of projections that contribute to reconstruct the pixels of the reconstructed image;
A pixel value update determination unit that determines whether or not to update the pixel value so as not to update the pixel value of the reconstructed image as the number of projections decreases.
A radiation tomography apparatus comprising: The radiation tomography apparatus according to claim 4,
When the projection line passes through the pixels of the reconstructed image when collecting the actual projection data in each direction, the projection number setting unit counts the passage of the projection line at each pixel through which the projection line passes. When the pixels of the actual projection data are data deficient pixels, the radiation that does not count the passage of the projection line at each pixel of the reconstructed image through which the projection line that reaches the pixel of the actual projection data passes Tomography equipment. The radiation tomography apparatus according to claim 4 or 5,
The projection number setting unit sets the projection number and its maximum value,
The said pixel value update determination part determines with the ratio based on the said projection number and its maximum value, The radiation tomography apparatus characterized by the above-mentioned. The radiation tomography apparatus according to any one of claims 4 to 6,
The radiation value tomography apparatus according to claim 1, wherein the pixel value update determination unit determines the ratio by comparing a ratio based on the number of projections with a random number. A radiation tomographic image generation processing device that generates a radiation tomographic image based on a plurality of actual projection data collected from different directions with respect to a subject,
A data loss extraction unit for extracting data loss pixels from the actual projection data;
When calculating the calculated projection data by forward projection from the reconstructed image, image reconstruction is performed by performing image reconstruction by the successive approximation method without performing forward projection on the pixels of the calculated projection data corresponding to the data missing pixels. A configuration execution unit;
A radiation tomographic image generation processing apparatus comprising: A radiation tomographic image generation processing device that generates a radiation tomographic image based on a plurality of actual projection data collected from different directions with respect to a subject,
A data loss extraction unit for extracting data loss pixels from the actual projection data;
An image reconstruction execution unit that performs image reconstruction by a successive approximation method without performing back projection from a pixel corresponding to the data-missing pixel;
A radiation tomographic image generation processing apparatus comprising: A radiation tomographic image generation program for causing a computer to execute a process of generating a radiation tomographic image based on a plurality of actual projection data collected from different directions with respect to a subject,
Extracting data missing pixels from the actual projection data;
A step of performing image reconstruction by a successive approximation method without performing forward projection on the pixels of the calculated projection data corresponding to the data missing pixels when calculating the calculated projection data from the reconstructed image by forward projection ; ,
A radiation tomographic image generation program comprising: A radiation tomographic image generation program for causing a computer to execute a process of generating a radiation tomographic image based on a plurality of actual projection data collected from different directions with respect to a subject,
Extracting data missing pixels from the actual projection data;
Performing image reconstruction by a successive approximation method without performing back projection from pixels corresponding to the data missing pixels;
A radiation tomographic image generation program comprising:

Images