JP5286145B2 - Bed positioning method - Google Patents

Description

  The present invention relates to a bed positioning method in radiation therapy.

  In radiotherapy, in order to make the position of the treatment object coincide with the treatment plan, the bed on which the treatment object is placed is positioned. To determine the bed position, a treatment plan image and an X-ray fluoroscopic image taken at the time of treatment are used. A treatment target in the body or a target in the vicinity of the treatment target is set as a positioning target, a region including the positioning target is subjected to X-ray fluoroscopic imaging, and a displacement amount of the positioning target is obtained from a comparison between the captured image and a reference image of the treatment plan. The movement amount of the bed is determined so as to correct the deviation amount. In particle beam therapy, X-ray IMRT, and the like, a dose distribution concentrated on the target can be realized, and it is necessary to match the target position with the plan with high accuracy in positioning. In order to position the target position with high accuracy, it is required to calculate the deviation between the plan and the treatment for a total of 6 degrees of freedom, that is, 3 degrees of freedom of translation and 3 degrees of freedom of the positioning target. Further, in order to shorten the treatment time, it is required to shorten the bed positioning time.

  The treatment plan is performed on a three-dimensional image taken in advance before the treatment. An X-ray CT image composed of a plurality of slices is typically used for the three-dimensional image. In the treatment plan, the positional relationship with respect to the positioning object and the radiation irradiation apparatus shown in the three-dimensional image is planned.

  The positional relationship between the positioning target position and the radiation irradiation apparatus at the time of treatment is confirmed by an X-ray fluoroscopic image of the positioning target captured by an X-ray imaging apparatus whose positional relationship with the radiation irradiation apparatus is known. The reason why the X-ray fluoroscopic image is used is to perform positioning based on the internal structure of the positioning target. X-ray fluoroscopic images are typically taken by separate imaging devices from two orthogonal directions.

  In comparison between a treatment plan 3D image and a 2D X-ray fluoroscopic image, a DRR (Digitally Reconstracted Radiograph), which is an X-ray fluoroscopic image generated by numerical simulation from the treatment plan 3D image, is generated and Make a comparison. In the DRR generation, the positional relationship between the planned treatment plan image and the radiation treatment apparatus is used.

  However, it is known that it is difficult to automatically detect a 6-degree-of-freedom shift from two sets of corresponding fluoroscopic images and DRR. Although translation in a plane and rotation (in-plane conversion) with respect to an axis perpendicular to the plane can be detected with high accuracy, detection of rotation (out-of-plane rotation) with respect to an axis parallel to the plane is difficult.

  As a method for detecting a 6-degree-of-freedom shift, a plurality of DRRs in which the positional relationship between the plan image and the radiation irradiation apparatus is different from that of the treatment plan is generated, and the most consistent X-ray fluoroscopic image among the obtained DRRs A method (DRR generation optimization method) for aligning a three-dimensional image and an X-ray fluoroscopic image by selecting is known. However, when the DRR generation optimization method is used for all degrees of freedom, calculation time becomes long. Therefore, there are Patent Documents 1 and 2 as techniques for speeding up the planar image alignment in combination.

  In Patent Document 1, planar image alignment is performed for in-plane conversion in X-ray fluoroscopic images corresponding to each DRR, and DRR generation optimization is performed for two degrees of freedom of out-of-plane rotation. The amount of freedom deviation is determined.

  In Patent Document 2, an image is aligned by combining in-plane conversion and DRR regeneration on an image, and a patient shift amount is obtained from the shift amount of the image. When performing in-plane conversion, the images that have been converted in-plane are compared to maximize image similarity, and the amount of displacement of the planar image obtained in the process is converted into a three-dimensional positioning target displacement amount. A deviation amount of 6 degrees of freedom is determined.

US 7,204,640 US7,412,086

  In order to shorten the treatment time, it is desired to reduce the calculation time in bed positioning. In the method described in Patent Document 1, when two sets of X-ray fluoroscopic image capturing apparatuses are used, six degrees of freedom are determined for each image. Therefore, there is a degree of freedom determined redundantly from a plurality of images. . There is a possibility of shortening the calculation time by eliminating duplicate determination of the degree of freedom.

  Further, in the procedure for converting the in-plane conversion result described in Patent Document 2 into the patient displacement amount, there is a possibility that unnecessary calculation may be performed at the following two points. The first point relates to optimization convergence determination. In optimizing the amount of patient deviation, it is considered that the calculation is terminated when the search range becomes sufficiently narrow in the three-dimensional space, and unnecessary calculations can be omitted. However, in the method of Patent Document 2, since the patient shift amount is obtained after at least one in-plane conversion, there is a possibility that image conversion may be performed even though the search range of the patient shift amount is sufficiently narrow. The second point relates to a path of change of the position parameter in optimization. In optimizing the amount of patient deviation, an optimal solution is searched in a 6-degree-of-freedom space. However, if the optimization path is limited, many trial calculations may be required before convergence. For example, performing optimization by in-plane conversion of one image pair according to the above procedure is optimization that allows conversion of only 3 degrees of freedom among 6 degrees of freedom. In order to change other degrees of freedom, it is necessary to convert the amount of in-plane conversion deviation into patient deviation, regenerate the DRR, and optimize for other image pairs. There is a possibility of shortening the calculation time by reducing unnecessary calculations considered in the above two points.

  In order to shorten the calculation time in bed positioning, a positioning method with a small amount of calculation while maintaining accuracy is required. In order to shorten the calculation time compared with the conventional method, there was a problem of reducing the amount of calculation while maintaining accuracy.

  The bed positioning method of the present invention first acquires a set of fluoroscopic images taken from a plurality of directions, and according to the positional relationship of the treatment plan, (1) the DRR corresponding to the photographing direction of each image from the treatment plan image. And (2) provisional generation of patient deviation with 6 degrees of freedom by an optimization algorithm. After that, (3) the rotational component (out-of-plane rotational component) of the axis parallel to the entire DRR projection plane is calculated, and (4-A) when the out-of-plane rotational component is larger than the specified value, it is based on the deviation amount. If all DRRs are re-created based on the positional relationship, and the (4-B) out-of-plane rotation component is smaller than the specified value, the in-plane conversion component of each DRR of the deviation amount is calculated, and the in-plane conversion component is specified from the time of DRR generation. If the position is greater than or equal to the position, the DRR is plane-transformed. Next, (5) the total similarity that is the sum of the similarities is calculated for each pair of the DRR and the fluoroscopic image, and (6) the search range is updated by the optimization algorithm so as to increase the total similarity. (7) The above (2) to (6) are repeated, and the calculation ends when the optimization search range satisfies the specified convergence condition. (8) The bed positioning is performed by calculating the amount of bed movement for correcting the obtained positioning target deviation amount.

  The above method enables high-speed 6-DOF bed positioning and shortens the bed positioning time.

1 is a system diagram showing an embodiment of the present invention. The figure which shows the processing flow in the Example of this invention.

Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings.
[Example 1]
As shown in FIG. 1, a first X-ray fluoroscopic imaging apparatus, a second X-ray generator (including a bed positioning apparatus 101, a first X-ray generator (X-ray tube) 200, and a first X-ray detector 210 according to the present embodiment ( X-ray tube) 220 and second X-ray detector 230, second X-ray fluoroscopic image capturing device, X-ray fluoroscopic image capturing control device 240, positioning calculation device 400, treatment control device 500, unillustrated input means (keybird, mouse, etc.) ).

  The first X-ray generator (X-ray tube) 200 is attached to the treatment beam irradiation apparatus 100 and has a configuration that can move (rotate) together with the movement (rotation) of the treatment beam irradiation apparatus 100. The first X-ray detector 210 is located on the beam line of the treatment beam and is disposed on the opposite side of the treatment beam irradiation apparatus 100 with respect to the bed 120. In addition, the second X-ray fluoroscopic imaging apparatus has a movable structure. The 2nd X-ray detector 230 is arrange | positioned in the position which detects the X-ray radiated | emitted from the 2nd X-ray generator 220, and permeate | transmitted the irradiation object. That is, when the second X-ray generator 220 moves, the second X-ray detector 230 also moves and is installed at a position where X-rays are detected.

  The X-ray fluoroscopic image capturing control device 240 is connected to and controls the first X-ray fluoroscopic image capturing device and the second X-ray fluoroscopic image capturing device. Specifically, the X-ray fluoroscopic imaging control device 240 controls the start and stop of irradiation of X-rays from the first X-ray generator 200 and the second X-ray generator 220, the first X-ray detector 210 and the second X-ray detector 210. 2 Creates image data based on the X-ray detection signal from the X-ray detector 230 and controls the position where these are set.

  The positioning calculation device 400 includes a communication unit 410, a calculator 420, and a memory 430. The positioning calculation device 400 (specifically, the communication unit 410) is connected to the treatment plan image photographing device 102, the treatment control device 500, and the X-ray fluoroscopic image photographing control device 240. The communication unit 410 is connected to the memory 430 via the calculation unit 420.

  The treatment control device 500 is connected to the bed 120. The treatment control device 500 controls the positioning of the bed 120 based on the calculation result of the positioning calculation device 400.

  The treatment plan image capturing apparatus 102 includes a three-dimensional image capturing apparatus 300 and an image server 310.

  In the radiotherapy system, the treatment beam on the bed 120 is irradiated with the treatment beam from the treatment beam irradiation apparatus 100. In order to place the positioning target 130 at a planned position with respect to the treatment beam irradiation apparatus 100 during treatment, a fluoroscopic image is taken. Hereinafter, a method for positioning the bed 120 using the bed positioning device 101 will be described.

  When the positioning target (patient) 130 is placed on the bed 120 in order to start treatment with the treatment beam, an operator such as a doctor displays a cross mark (laser directly above the affected area) drawn on the body surface of the patient 130. The movement command of the bed is input to the treatment controller 500 using an input device (not shown) so that the mark or the like) is positioned on the beam line. The treatment control device 500 moves the bed 120 based on the movement command, and controls the cross mark on the body surface so as to coincide with the beam line.

  The operator inputs a movement start signal of the X-ray fluoroscopic imaging apparatus to the X-ray fluoroscopic imaging control apparatus 240 from an input device (not shown). The X-ray fluoroscopic image capturing control device 240 that has received the movement start signal moves the first X-ray fluoroscopic image capturing device and the second X-ray fluoroscopic image capturing device and arranges them at a predetermined position. The first X-ray fluoroscopic image capturing apparatus and the second X-ray fluoroscopic image capturing apparatus are arranged so that the X-ray imaging directions of the patient 130 are two directions orthogonal to each other. Thereafter, when the operator inputs an X-ray irradiation start signal to the X-ray fluoroscopic image capturing control device 240, the X-ray fluoroscopic image capturing control device 240 sends an X-ray from the first X-ray generating device 200 and the second X-ray generating device 220. The first X-ray detector 210 and the second X-ray detector 230 detect transmitted X-rays. The first X-ray detector 210 and the second X-ray detector 230 output an X-ray detection signal to the X-ray fluoroscopic imaging control device 240 when detecting the transmitted X-ray. The X-ray fluoroscopic imaging control device 240 creates image data based on these X-ray detection signals. As described above, the X-ray imaging directions are two orthogonal directions. The created image data (current image data) is output to the positioning calculation device 400.

  The positioning calculation device 400 receives image data from the X-ray fluoroscopic imaging control device 240 through the communication unit 410, and the computing unit 420 stores the image data in the memory 430.

  A three-dimensional image used at the time of treatment planning is captured by the three-dimensional image capturing apparatus 300. The captured image data is stored in the image server 310. During the treatment, the positioning calculation device 400 acquires a three-dimensional image from the image server via the communication unit and stores it in the memory 430.

  In the positioning calculation device 400, when the fluoroscopic image and the three-dimensional image data are stored, the computing unit 420 calculates the DRR from the three-dimensional image data. The computing unit compares the DRR and the fluoroscopic image, associates the fluoroscopic image with the positioning target position in the three-dimensional image by an optimization procedure, and obtains the deviation of the positioning target position from the plan. The computing unit 420 finally obtains a bed movement amount for correcting the deviation from the deviation of the positioning target position.

  In accordance with an instruction from the operator, the calculated bed movement amount is sent from the positioning calculation device 400 to the treatment control device 500. The treatment control device 500 moves and controls the bed 120 based on the received bed movement amount.

  The positioning calculation device 400 according to the present embodiment uses the current image data received from the X-ray fluoroscopic image capturing control device 240 and the image data stored in the image server 310 of the treatment plan image capturing device 102 to position the bed 120. Data is generated and the positioning data is output to the treatment control device 500. Hereinafter, processing means for the positioning device 400 to generate positioning data will be described with reference to the flowchart of FIG. Such processing means is stored in the memory 430 of the positioning calculation device 400.

  First, the positioning calculation device 400 acquires 3D image data to be positioned, which is previously captured by the 3D image capturing device 300 and stored in the image server 310, through the communication unit 410 and stores it in the memory 430 according to an instruction from the operator. (Step 100).

  When positioning the affected part of the patient 130 on the treatment beam line, first and second X-ray fluoroscopic images of the positioning target 130 placed on the bed 120 are taken (step S150). Each X-ray fluoroscopic image is taken by first and second X-ray fluoroscopic imaging devices. The captured current image data is sent from the X-ray image capturing control device 240 to the positioning calculation device 400. The positioning calculation device 400 receives the current image data by the communication unit 410 and stores it in the memory 430 via the calculator 420.

  The positioning calculation device 400 calculates the DRR according to the planned positional relationship between the three-dimensional image captured by the three-dimensional image capturing device 300 at the time of treatment planning and the first and second fluoroscopic image capturing devices (step S200). DRR is a substance converted from pixel values of a three-dimensional image on a straight line connecting an X-ray detector region corresponding to each pixel and an X-ray source (a point emitting X-rays of an X-ray generator). The amount is integrated and finally converted to X-ray transmittance.

  The positional relationship of the treatment plan is set as the reference position, and the DRR calculated at the reference position is set as the reference DRR. The positioning calculation device 400 obtains how much the positioning target position 130 at the time of X-ray fluoroscopic image photographing is shifted from the reference position by the following procedure. When the reference DRR matches the fluoroscopic image, the amount of deviation from the positioning target plan is zero for all degrees of freedom.

  The positioning calculation device 400 (specifically, the computing unit 420) generates a provisional deviation amount from the reference position of the positioning target position using an optimization algorithm (step S300). The amount of deviation is a set of 6 degrees of freedom values of 3 degrees of freedom for translation and 3 degrees of freedom for rotation. For the optimization algorithm, the Powell-Brent method or the like is used. If the DRR generation optimization is performed for all six degrees of freedom as described above, the calculation time becomes long. Therefore, it is desirable to approximate the deviation by planar conversion of the DRR as much as possible. However, rotation of the axis parallel to the projection plane of all DRRs (out-of-plane rotation) is handled separately because any DRR in-plane conversion cannot be approximated.

  Next, the positioning calculation device 400 (calculator 420) calculates an out-of-plane rotation component of the generated temporary deviation amount (step S400). Out-of-plane rotation is rotation of an axis parallel to all X-ray detector surfaces. As the out-of-plane rotation component, when there is an axis that coincides with the out-of-plane rotation axis among the coordinate axes representing the shift amount, the value of the matching axis is used. If there is no coincident axis, the inner product of the vector having the three degrees of freedom of rotation as a component and the unit vector in the direction of the rotation axis is obtained. Similarly, the out-of-plane rotation component of the deviation amount at the time of DRR generation is also obtained.

  The positioning calculation device 400 (calculator 420) compares the obtained out-of-plane rotation component of the deviation amount and the out-of-plane rotation component of the deviation amount at the time of DRR generation, and determines whether the change amount is equal to or greater than a specified value (step S500). . Here, the specified value is a reference value stored in the memory 430 in advance. If the amount of change in the out-of-plane rotation component is larger than the predetermined value, it is determined that approximation is impossible by the planar conversion of the DRR, and the DRR is regenerated in a positional relationship that is shifted from the positional relationship of the treatment plan by the temporary shift amount. . That is, when the value is equal to or greater than the specified value, the positioning calculation device 400 (calculator 420) regenerates all the DRRs in a positional relationship in which the shift amount is applied to the reference position (step S600).

  If the change amount of the out-of-plane rotation component is equal to or less than the specified value, the positioning calculation device 400 (calculator 420) converts the image into a plane in steps S610 and S620. When the out-of-plane rotation is negligibly small, the image shift can be approximated by in-plane conversion (plane conversion) in at least one DRR.

  As described above, when the out-of-plane rotation component of the shift amount is smaller than the predetermined value, the shift is approximated by planar conversion of each DRR. The two translational components in the plane and the in-plane rotation are combined and called in-plane conversion, and the shift amount in the three-dimensional space is projected onto the in-plane conversion component. In this projection, there may be a case where the in-plane conversion component has a smaller change compared to when DRR is generated, and in this case, the in-plane conversion is unnecessary. For this reason, in this embodiment, the DRR is subjected to in-plane conversion only when the in-plane conversion component changes by a predetermined value or more.

  For example, when the shift amount changes in a direction perpendicular to the DRR projection plane, the in-plane conversion component does not change, and thus DRR in-plane conversion is omitted. When the imaging directions of the X-ray fluoroscopic images are orthogonal, it is preferable because directions in which in-plane conversion can be omitted are increased. In addition, since the in-plane conversion of DRR is faster than the regeneration of DRR, the deviation evaluation is speeded up by approximating deviations other than out-of-plane rotation by in-plane conversion.

  A method in which the positioning calculation apparatus 400 (calculator 420) performs in-plane conversion will be described below. First, for each DRR, an in-plane conversion component of the deviation change amount is calculated by subtracting the deviation amount at the time of DRR generation from the deviation amount to be evaluated. The deviation amount at the time of DRR generation is 0 for all components when the DRR is the reference image generated at step S200, but has a finite value for the DRR recreated at step S600. The translational component 2 degrees of freedom of the in-plane conversion is obtained from the inner product of a vector having the translational component 3 degrees of freedom of the shift amount and a unit vector in the plane of the DRR. The one-degree-of-freedom rotational component of in-plane conversion is obtained by the inner product of a vector whose component is a three-degree-of-freedom rotational amount and the unit vertical vector of the DRR surface. Since the in-plane conversion component (two degrees of freedom of translation + one degree of freedom of rotation) is obtained for each DRR, the positioning calculation device 400 (calculator 420) converts the DRR into a plane (step S620). However, there are cases where identity conversion occurs when projected onto an in-plane conversion component, such as when the amount of deviation change is perpendicular to the DRR surface, and in this case, conversion is not necessary. Therefore, in step S620, the DRR plane conversion is performed only when the in-plane conversion component changes by a predetermined value or more.

  The similarity between the DRR converted in step S600 or in steps S610 and S620 and the fluoroscopic image is calculated (step S700). First, the similarity between each pair of DRR and fluoroscopic image is calculated individually. For the similarity, a sum of squares of pixel value differences or a standardized mutual information amount based on comparison of pixel values is used. All the similarities obtained individually are added to obtain the total similarity. This total similarity is used as an evaluation index for the positioning target deviation amount.

  The positioning calculation device 400 (calculator 420) updates the search range with an optimization algorithm so as to increase the obtained total similarity (step S800). In general, when the total amount of tentative deviation is larger than the end point of the search range, the search range is narrowed. For example, in the Powell-Brent method, the total similarity of the end points of the search range is compared with the total similarity at the trial points in the search range, and if the total similarity of the trial points is large, the search range is set as a new end point. To narrow. Here, a large total similarity indicates that the X-ray fluoroscopic image and the DRR image obtained from the three-dimensional image are similar, and these images can be regarded as matching when the total similarity is the largest. In other words, the total similarity is used as an evaluation index of the positioning target deviation amount, and it is determined that the larger the total similarity is, the closer the true deviation amount during the X-ray imaging of the positioning target is.

  In step S800, the positioning calculation device 400 (calculator 420) determines whether the updated search range satisfies the convergence condition (step S900). In this embodiment, a convergence condition is set for the search range. As the convergence condition, the search range is sufficiently smaller than the image resolution, or the positioning accuracy required from radiation irradiation is satisfied. If the search range is sufficiently smaller than the resolution of the treatment plan image, further search is meaningless. Therefore, a minimum search range in consideration of the image resolution is set in the memory 430 in advance, and the positioning calculation device 400 (calculator 420) ends the calculation when the search range becomes an area below the minimum search range. If the condition is not satisfied, the positioning calculation device 400 (calculator 420) returns to step S300 and repeats the calculation at a new trial point. As described above, when the search range satisfies the convergence condition, the calculation is terminated. When the search range is not satisfied, a temporary deviation amount is generated again by the optimization algorithm, and the calculation is repeated.

  Since the amount of deviation of the positioning target position from the treatment plan is determined in steps S100 to 900, the bed movement amount that finally corrects the deviation is calculated and the bed is positioned (step S1000).

  According to the present embodiment, all the DRRs are converted with respect to a set of 6-degree-of-freedom conversion parameters, and the conversion parameters are independently optimized for each DRR as in Patent Literature 1 in order to evaluate the total similarity. There is no duplication of optimization calculation that occurs in. Further, since optimization is not performed for each DRR, there is no limitation on the optimization path as in Patent Document 2, and optimization in a 6-degree-of-freedom space can be performed with the number of DRR generations being minimized. In addition, since convergence is determined based on the search range of the positioning target deviation amount, the optimization between the planar images does not optimize in a finer area than necessary.

100 therapeutic beam irradiation device 120 bed 130 positioning target 200 first X-ray generator (X-ray tube)
210 First X-ray detector 220 Second X-ray generator (X-ray tube)
230 Second X-ray detector 240 X-ray fluoroscopic image capturing control device 300 Three-dimensional image capturing device 310 Image server 400 Positioning calculation device 410 Communication unit 420 Calculator 430 Memory 500 Treatment control device

Claims (2)

A method of positioning the bed using an X-ray fluoroscopic image obtained by photographing the object to be positioned, which is placed on the three-dimensional image and the bed taken during treatment planning from a plurality of directions,
(1) a procedure for calculating a DRR corresponding to each X-ray fluoroscopic image from a three-dimensional image;
(2) a step of creating a hand first shift amount of the position-decided Me target optimization algorithm,
(3) calculating a rotation component (out-of-plane rotation component) of an axis parallel to the entire DRR projection plane of the first deviation amount;
(4-A) When the out-of-plane rotation component has changed by a predetermined value or more compared to when the DRR is generated, a procedure for recreating all the DRRs based on the positional relationship based on the first deviation amount;
(4-B) out-of-plane when the change amount of the rotational component is smaller than the specified value, the in-plane transform components for each DRR of the first shift amount is calculated and the in-plane transformation component DRR generated when either et al a step of planar convert DRR when changing over specified value,
(5) a step of calculating the total similarity is the sum of the similarities in each pair of DRR corresponding to a plurality of X-ray fluoroscopic image and the X-ray fluoroscopic image,
(6) updating the search range with an optimization algorithm to increase the total similarity;
(7) Repeat steps (2) to (6), and terminate the calculation when the optimization search range satisfies a specified convergence condition;
(8) A bed positioning method comprising a step of calculating a bed movement amount for correcting the second shift amount of the positioning object obtained as a result of the end of the calculation .   The bed positioning method according to claim 1, wherein the X-ray transmission image is taken from two orthogonal directions.

Images