JPWO2018225234A1 - Positioning device and positioning method - Google Patents

Abstract

The function of the positioning unit 43 is to optimize each parameter relating to rotation and translation of the perspective projection by using a positioning algorithm for obtaining an optimal translation / rotation solution that minimizes the amount of deviation between the DRR image and the DR image. In running. The positioning unit 43 includes a three-axis optimizing unit 45, a six-axis optimizing unit 46, and a one-dimensional optimizing unit 47 as components for realizing the optimizing function. The functions of the three-axis optimizing unit 45, the six-axis optimizing unit 46, and the one-dimensional optimizing unit 47 are stored in a memory as a program and executed by the operation of the CPU.

Description

  The present invention relates to a positioning device and a positioning method for positioning a patient when performing radiation therapy on the patient.

  2. Description of the Related Art In radiotherapy for irradiating an affected part of a patient with radiation such as an X-ray, an electron beam, or a particle beam, it is necessary to accurately irradiate therapeutic radiation to the affected part. In such radiation treatment, first, an X-ray CT (Computed Tomography) scan is performed, and a treatment plan is formulated. When a treatment is performed by the radiation therapy apparatus, positioning is performed with the patient lying on the examination table in order to match the irradiation target of the treatment beam with the irradiation center.

  In an apparatus that uses an X-ray fluoroscopic image or CT data for positioning of a patient, a DR (Digital Radiography) image, which is an actual fluoroscopic image of an affected part of the patient and its surroundings, which is constrained to a treatment table by a fixture, is acquired, and a treatment plan is formulated. At this time, creation of a DRR (Digital Reconstructed Radiography) image, which is a virtual perspective projection, on the three-dimensional image data obtained by the X-ray CT scan at the time is executed. Then, by evaluating the similarity between the DR image and the DRR image (image registration), the shift amount between the actual position of the patient at the time of the radiation treatment and the position at the time of the treatment plan is calculated (Patent Document 1). And Patent Document 2).

  Before the positioning according to the positioning algorithm for matching the DR image and the DRR image, an initial setup called coarse positioning is performed. The initial setup may be performed, for example, by adjusting the position of the operator manually using a laser marking device or by superimposing the DR image and the DRR image displayed on the display by operating the input device. There is a deviation amount adjustment.

JP 2007-282877 A JP 2009-201556 A

  In a positioning algorithm that optimizes perspective projection parameters related to rotation and translation of CT data so that the degree of coincidence between the DR image and the DRR image is maximized, the amount of deviation between the DR image and the DRR image after the initial setup is optimized. Is the initial value. In the conventional parameter optimization, while changing the resolution of an image from a low resolution to a high resolution, an optimization operation is performed simultaneously in all six degrees of freedom of parallel movement of three orthogonal axes in three-dimensional space and rotational movement around each axis. Be executed.

  Since the initial setup is performed manually by the operator, there is an individual difference in the degree of deviation between the DR image and the DRR image. If the initial deviation amount between the DR image and the DRR image at the time of starting parameter optimization is large (for example, 1 cm or more), positioning may fail if optimization operations are performed simultaneously in all six degrees of freedom. was there. The cause of the failure may be a stagnation of the progress of the optimization calculation due to an attempt to forcibly adjust by rotation before the optimization of the parallel movement is completed. When the positioning fails, it is necessary to start over from the initial setup again, and the time from when the patient lays on the treatment table to when the patient can irradiate the treatment beam is long.

  The present invention has been made to solve the above-described problems, and has as its object to provide a positioning device and a positioning method that can easily perform positioning during treatment.

  According to a first aspect of the present invention, there is provided a positioning device for positioning a subject when performing radiation therapy for irradiating a treatment beam toward an affected part of the subject on a treatment table, wherein the radiation irradiating unit and a radiation detector are arranged. An image acquisition unit that acquires two-dimensional radiation images of the subject in two different directions from an imaging system having the imaging system, and reproduces the geometrical arrangement of the imaging system in space and virtually converts the CT data collected by computer tomography into virtual data. A DRR image creating unit that creates DRR images of the subject in two different directions by performing perspective projection, a positioning unit that performs alignment of the radiation image and the DRR image, and the positioning unit And a movement amount calculation unit that outputs a movement amount of the treatment table from a shift amount between the radiation image and the DRR image calculated by: Optimum calculation of the three-degree-of-freedom parameter of the parallel movement out of the six-degree-of-freedom parameter of the parallel movement and the rotation is performed to obtain the amount of parallel displacement between the radiation image and the DRR image. And a parameter having six degrees of freedom or a parameter having four or five degrees of freedom corresponding to the movement axis of the treatment table after reflecting the result of the optimization operation in the three-axis optimizing part. A multi-axis optimizing unit that calculates a shift amount in the parallel direction and the rotation direction between the radiation image and the DRR image by performing a calculation.

  In the second invention, the alignment unit executes the alignment while changing the resolution of the radiographic image and the DRR image stepwise from a low resolution to a high resolution, and the three-axis optimization unit performs The optimization operation of the three-degree-of-freedom parameter of the translation is performed at the resolution.

  In the third invention, when the multi-axis optimizing unit executes the optimization calculation of the parameter of six degrees of freedom, the movement amount calculating unit sets a parallel direction and a rotational direction between the radiation image and the DRR image. The amount of movement in six directions is calculated from the amount of displacement of the table, and among the amounts of movement in the six directions, the amount of movement corresponding to the movement axis of the treatment table is output.

  In a fourth aspect, the positioning unit includes a one-dimensional optimizing unit that executes a parameter optimizing operation for one-dimensional parallel movement of the imaging system along a shooting direction.

  According to a fifth aspect of the present invention, there is provided a positioning method for positioning a subject when performing radiation therapy for irradiating a treatment beam toward an affected part of the subject on a treatment table, comprising: An image acquisition step of acquiring two-dimensional radiation images of the subject in two different directions from an imaging system having the imaging system, and reproducing the geometrical arrangement of the imaging system in space, and virtualizing CT data collected in advance by computer tomography. A DRR image creating step of creating DRR images of the subject in two different directions by performing perspective projection, a positioning step of executing the alignment of the radiation image and the DRR image, and the positioning step A movement amount calculation step of outputting the movement amount of the treatment table from the deviation amount of the radiation image and the DRR image calculated by, the alignment step, A three-axis optimization step of executing an optimization operation for optimizing a three-degree-of-freedom parameter of the parallel movement among the six-degree-of-freedom parameters of the translation and rotation in the inter-coordinate system; A multi-axis optimization step of performing an optimization operation of a parameter of six degrees of freedom or a parameter of four or five degrees of freedom corresponding to the movement axis of the treatment table after reflecting the result of the conversion operation. It is characterized by the following.

  According to the first to fifth aspects, after optimizing the parameters for the three degrees of freedom of the translation, the parameters are optimized for the degrees of freedom including the rotation in the translation, so that the parameters are obtained from the imaging system. Even when the deviation amount of the initial position between the obtained radiation image and the DRR image created by performing virtual perspective projection on the CT data is large, it is possible to prevent a local solution from being obtained by the parameter optimization calculation. As described above, since the allowable range of the shift amount of the initial position between the radiographic image and the DRR image is increased, the case of re-starting from the initial setup as in the related art can be reduced without stalling the progress of the optimization calculation. Can be. Therefore, it is possible to shorten the workflow at the time of treatment with the radiation therapy apparatus.

  According to the second aspect, when positioning is performed by gradually increasing the resolution of the radiographic image and the DRR image from a low resolution to a high resolution, optimization of the three-degree-of-freedom parameter of the parallel movement is performed in the initial stage of the optimization. , The calculation cost of the optimization can be reduced.

  According to the third aspect, after calculating the amount of movement in six directions from the amount of deviation between the radiation image and the DRR image obtained by optimizing the parameters of six degrees of freedom, the necessary movement corresponding to the movement axis of the treatment table Since only the amount is output to the treatment table, there is no need to prepare a positioning algorithm for each degree of freedom of the movement axis of the treatment table. It is possible to reduce the positioning error as compared with the case where the parameters are optimized.

  According to the fourth aspect, since one-dimensional optimization is performed along the imaging direction of the imaging system, an SID (Source Image) is obtained as in an imaging system that performs imaging from two directions with a tilt angle to a patient. Even when the distance (Distance) becomes long, it is possible to prevent the progress of the optimization operation from stagnating. In addition, it is possible to minimize the optimization route and shorten the positioning time.

It is a schematic diagram of a radiotherapy device to which the positioning device according to the present invention is applied. FIG. 2 is a block diagram of a control system including the positioning device according to the present invention. 9 is a flowchart illustrating a procedure for positioning a subject. 9 is a flowchart illustrating a procedure for parameter optimization. It is explanatory drawing which shows typically the process of optimization of the valley structure of an evaluation function and the similarity of an image when the one-dimensional optimization operation of this invention is performed.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a schematic diagram of a radiotherapy device 1 to which a positioning device according to the present invention is applied.

  The positioning device of the present invention includes an X-ray imaging system, and is used together with the radiotherapy device 1. The radiotherapy apparatus 1 is for performing radiotherapy on a patient (subject) on a top plate 31 of a treatment table 30, and includes a head 55 for emitting a treatment beam and a base 55 installed on the floor of a treatment room. A gantry 53 rotatably supported by the table 52 is provided. The radiation treatment apparatus 1 can change the irradiation direction of the treatment beam by rotating the gantry 53.

  The X-ray imaging system is for performing X-ray fluoroscopy in order to specify the position of the affected part of the patient who is lying on the top plate 31 of the treatment table 30, and includes X-ray tubes 11a to 11d as radiation irradiation units. And flat panel detectors 21a to 21d, which are radiation detectors for detecting X-rays transmitted through the subject and the top plate 31. The X-ray tubes 11a to 11d and the flat panel detectors 21a to 21d are arranged at positions for performing X-ray fluoroscopy on the subject from an oblique direction. Although not shown in FIG. 1, the X-ray tubes 11a to 11d are arranged in concave portions formed on the floor surface, and the concave portions are covered by a lid member forming a part of the floor. Further, an image intensifier (II) may be used as the radiation detector.

  X-rays emitted from the X-ray tube 11a are detected by the flat panel detector 21a, and the X-ray tube 11a and the flat panel detector 21a constitute a first imaging system. X-rays emitted from the X-ray tube 11b are detected by the flat panel detector 21b, and the X-ray tube 11b and the flat panel detector 21b constitute a second imaging system. X-rays emitted from the X-ray tube 11c are detected by the flat panel detector 21c, and the X-ray tube 11c and the flat panel detector 21c constitute a third imaging system. X-rays emitted from the X-ray tube 11d are detected by the flat panel detector 21d, and the X-ray tube 11d and the flat panel detector 21d constitute a fourth imaging system. When positioning the subject, two of the first to fourth imaging systems are selected so that the gantry 53 does not overlap the field of view, and X-rays from two different directions with respect to the subject are obtained. A fluoroscopy is performed.

  FIG. 2 is a block diagram of a control system including the positioning device according to the present invention.

  This positioning device includes a CPU (Central Processing Unit) for executing a logical operation, a GPU (Graphics Processing Unit) for executing various image processing, a ROM in which a program necessary for controlling the device is stored, and data and the like during control are temporarily stored. The control unit 40 is implemented by a computer having a memory such as a RAM to be stored.

  The control unit 40 is connected to the X-ray tube control unit 10 that controls the irradiation of X-rays from the X-ray tubes 11a to 11d, and the flat panel detectors 21a to 21d. The X-ray tube control unit 10 is connected to the X-ray tubes 11a to 11d, and is necessary for irradiating X-rays to two imaging systems selected from the X-ray tubes 11a to 11d during X-ray fluoroscopy. Supply tube voltage and tube current. The control unit 40 is also connected to the network 17, the display unit 15, the input unit 16, the radiotherapy device 1, and the treatment table 30. The top board 31 of the treatment table 30 can be moved in six axes of horizontal movement and rotation by a top board moving mechanism 32.

  The control unit 40 has, as a functional configuration, an image acquisition unit 41 that acquires information on a two-dimensional radiation image (DR image) detected by two of the selected imaging systems among the flat panel detectors 21a to 21d, A DRR image creation unit 42 for creating DRR images in two different directions by virtually performing perspective projection on three-dimensional CT data collected by prior computer tomography (X-ray CT imaging) via the network 17. A positioning unit 43 that matches the DR image obtained by performing fluoroscopy with the two imaging systems with the DRR image, and a movement amount calculation unit 44 that calculates the amount of movement of the top board 31 from the amount of displacement between the images. Prepare. These functional configurations are stored in a memory as a program and executed by the action of the CPU.

  The positioning unit 43 executes an optimization calculation of each parameter related to rotation and parallel movement of the perspective projection using a positioning algorithm for obtaining an optimal translation / rotation solution that minimizes the amount of deviation between the DR image and the DRR image. It has a function to do. Then, the optimization calculation is executed centering on the target isocenter located in the irradiation field of the radiation therapy apparatus 1. The positioning unit 43 includes a three-axis optimizing unit 45, a six-axis optimizing unit 46, and a one-dimensional optimizing unit 47 as components for realizing parameter optimization. The functions of the three-axis optimizing unit 45, the six-axis optimizing unit 46, and the one-dimensional optimizing unit 47 are stored in a memory as a program and executed by the operation of the CPU.

  Next, a positioning method in the positioning device having the above-described configuration will be described. FIG. 3 is a flowchart showing a procedure for positioning the subject.

  An X-ray fluoroscopy is performed on the subject on the top board 31 of the treatment table 30 by two imaging systems selected from the first to fourth imaging systems, and images are obtained from two of the flat panel detectors 21a to 21d. Information is obtained, and DR images from two different directions are obtained (step S1: image obtaining step).

  The geometry (geometric arrangement) of the imaging system is reproduced in a virtual space on a computer, and virtual perspective projection is performed on the previously acquired three-dimensional CT data. The CT data is acquired by an X-ray CT apparatus when formulating a treatment plan, and stored in a patient DB (not shown). The control unit 40 acquires a treatment plan and CT data via the network 17. Thereafter, by the operation of the DRR image creation unit 42, two-dimensional DRR images of the subject in two different directions are created by virtual perspective projection onto the three-dimensional CT data (step S2: DRR image creation step).

  The geometry of the X-ray fluoroscopy includes the positions of any two of the X-ray tubes 11a to 11d, the positions of any two of the flat panel detectors 21a to 21d, and the position and orientation of the top plate 31 in the two selected imaging systems. Is included. Since the mechanical installation accuracy of these elements affects the final positioning accuracy, the installation positions are periodically configured, and the calibration results are reflected in the geometry of X-ray fluoroscopy.

  When the DRR image is created, the voxel values of the CT image data are integrated along the projection line from any one of the X-ray tubes 11a to 11d in the selected imaging system to any of the flat panel detectors 21a to 21d. Integration).

  The parameters related to the parallel movement and rotation of the perspective projection are optimized so that the DR image and the DRR image match, and alignment is performed (step S3: alignment step). Here, as an evaluation function for evaluating the degree of coincidence between two images, a normalized mutual information (NMI), a gradient difference (GD: Gradient Difference), and a zero-average normalized cross-correlation (ZNCC: Nero- An evaluation function such as mean normalized cross-correlation that has been conventionally used for multi-modality image registration can be employed. In addition, by using NMI, GD, and ZNCC in combination, it is also possible to improve the evaluation accuracy of the degree of coincidence between the DR image and the DRR image.

  It is preferable that the calculation of the evaluation function is performed only in the region where the subject is present in the image. In addition, it is desirable that a part having a deformation that is not reproducible with CT data, such as a moving organ or a joint in the subject, is excluded from the calculation of the evaluation function.

  The amount of misalignment between the DR image and the DRR image obtained as a result of the alignment by the operation of the alignment unit 43 is determined by executing the function of the movement amount calculation unit 44 by the operation of the CPU configuring the control unit 40. It is converted into a top plate movement amount (step S4: movement amount calculation step). Then, this movement amount is transmitted from the movement amount calculation unit 44 to the top moving mechanism 32 of the treatment table 30. Thereafter, the top 31 is moved by the operation of the top moving mechanism 32 (step S5: top moving step). As described above, by moving the top 31 so that the subject is displaced by the positional shift amount between the DR image and the DRR image created from the CT data, the subject can be treated by the treatment beam emitted from the radiation therapy apparatus 1. Is positioned at the position and angle as in the treatment plan. After transposition of the top plate 31 and positioning of the subject, X-ray fluoroscopy is performed again, and the DR image and the DRR image at that time are displayed on the display unit 15 so that the images match. It is determined by the user whether or not there is a visual check. Then, a treatment beam is emitted from the head 55 of the radiation therapy apparatus 1 toward the affected part of the subject.

  The functions of the three-axis optimizing unit 45, the six-axis optimizing unit 46, and the one-dimensional optimizing unit 47, which are components of the positioning unit 43, will be described in more detail. FIG. 4 is a flowchart showing a procedure of parameter optimization.

In this embodiment, the quasi-Newton method based on the BFGS formula is used as a method of the optimization calculation. The quasi-Newton method is a method that can obtain the minimum value and the maximum value of a k-dimensional function f (x). In the optimization operation, assuming that an evaluation function for evaluating the coincidence between the DRR image and the DR image is f (x), from the initial value x ₀ of the point in the treatment room space until x _{i + 1} at the ( _{i + 1)} -th point becomes sufficiently small, Iterative calculation is performed according to equation (1).

  Here, H is an approximation of the inverse matrix of the Hessian matrix. Although several approximate expressions for H have been proposed, the BFGS formula given by the following expressions (2) and (3) has the highest calculation efficiency.

  When performing image registration, parameters are optimized for a total of six degrees of freedom, ie, three degrees of freedom of translation and three degrees of rotation of each of the X, Y, and Z axes of the treatment room spatial coordinate system. I do. In the present invention, before simultaneously optimizing a total of six axes, that is, three axes in the parallel direction and three axes in the rotational direction, first, optimization is performed on three axes in the parallel direction (three degrees of freedom) (step S1). S31: Triaxial optimization step). In the three-axis optimization step, the three-dimensional position x is updated according to the above equation (1), using the evaluation function f (x) as a three-dimensional function. The iterative calculation is performed until the deviation amounts for the three parallel axes converge to the target value. Note that the three-axis optimization is realized by the CPU executing a program read from the three-axis optimization unit 45.

When the optimization for the three axes in the parallel direction is completed, the optimization for the six axes including the rotation direction is executed (step S32: a six-axis optimization process). In the six-axis optimization process, the six-dimensional position x is updated according to the above equation (1), assuming that the evaluation function f (x) is a sixth-order function that depends on six independent variables relating to six degrees of freedom of translation and rotation. I do. The initial value x ₀ of the six-dimensional position x will position after the three-axis parameters of translation in the preceding three axes optimization process is optimized. Then, if the shift amounts for a total of six axes, ie, three axes of translation and three axes of rotation, converge within a predetermined time or while repeating the calculation of a predetermined number of times (step S33), the optimization calculation is performed. finish. On the other hand, if the shift amounts of the six axes do not converge, one-dimensional optimization is performed (step S34: one-dimensional optimization step). The six-axis optimization is realized by the CPU executing a program read from the six-axis optimization unit 46. The six-axis optimizing unit 46 and the six-axis optimizing step correspond to the multi-axis optimizing unit and the multi-axis optimizing step of the present invention. In the present invention, the multi-axis means a movement axis having 4 to 6 degrees of freedom in which a rotation axis is added to three translation axes.

  Here, the X-ray fluoroscopy is performed from two different directions by the two selected imaging systems, and the values of the evaluation functions for evaluating the coincidence between the DRR image after the 6-axis optimization and the DR image are different. Obtained about the direction. The determination as to whether or not the shift amount between the two images has converged to the target value in step S33 is made based on whether or not the value of the evaluation function has reached the convergence determination value. When the values of the evaluation functions in two different directions are F1 and F2, the sum F1 + F2 of the evaluation functions in the two different directions is set as the value of the evaluation function to be compared with the convergence determination value. When one of the two different directions is more important in positioning, the sum may be obtained by weighting one of the two, instead of the simple sum of F1 and F2. After the three-axis optimization, it is determined whether or not to proceed to the six-axis optimization step. However, since the six-axis optimization is performed after the three axes of the parallel movement are adjusted, the three-axis optimization is performed. It is not necessary to set the convergence determination value more strictly as the convergence determination after the six-axis optimization (step S33), and the convergence determination of the optimization does not need to be performed. You may make it advance to the step of conversion.

  If the determination in step S33 after the calculation of the six-axis optimization is “No”, one-dimensional optimization is performed (step S34). In the one-dimensional optimization step, parameters relating to one-dimensional parallel movement along the imaging direction in the two selected imaging systems are optimized. The evaluation function in the one-dimensional optimization process is a linear function that depends on one variable related to one-dimensional translation along the imaging direction. In this optimization, the Brent method, the golden section method, or the like can be used.

  FIG. 5 is an explanatory diagram schematically showing the process of optimizing the valley structure and parameters of the evaluation function when the one-dimensional optimization operation of the present invention is executed.

  The one-dimensional optimization unit 47 optimizes parameters related to one-dimensional parallel movement along the imaging direction, so that the optimization path proceeds along the valley structure. Therefore, it is possible to perform optimization in consideration of the imaging direction not only at the beginning of optimization but also at the end of optimization, and X-rays can be obtained from two different directions having an inclination angle with respect to the subject as shown in FIG. This solves the problem of the stagnation of the progress of the optimization calculation due to the longer SID when fluoroscopy is performed. As a result, the optimization operation is efficiently performed on the shortest path, the total calculation time can be shortened as compared with the related art, and the positioning accuracy can be improved.

  When the evaluation function values in two different directions are F1 and F2, respectively, as in the case of the three-axis optimization and the six-axis optimization, May be used, but it is more preferable to use only the evaluation function having a large value (either F1 or F2). That is, when the optimization path is along the valley structure of the evaluation function, the evaluation function with a small value has a small contribution to the optimization, and therefore the objective function with a large value on the side where the displacement is not along the imaging direction. Performing the optimization only with the use of the above makes it possible to speed up the calculation.

  The one-dimensional optimization is repeatedly executed until the value of the evaluation function reaches the convergence determination value (Step S35: convergence determination step). In this embodiment, different evaluation functions are used for the optimization of three axes and six axes and one-dimensional optimization, and different values are used for the convergence determination value of the value of the evaluation function. As described above, by using the evaluation function and the convergence determination value suitable for each, the parameters can be more appropriately optimized. The one-dimensional optimization may be omitted when the difference between the evaluation function values F1 and F2 in two different directions is small.

  Further, when the convergence determination value is not reached even after reaching the predetermined calculation time or the predetermined number of calculations, the repetition calculation of the optimization is terminated (step S36: termination determination step). By providing such a time limit, the burden on the patient fixed on the top plate 31 in the same posture and a decrease in the throughput of the radiotherapy apparatus 1 are reduced.

  In addition, the above-described optimization operation is performed while gradually changing the resolution of the DR image and the DRR image from a low resolution to a high resolution using a downsampling method. For example, the calculation is repeated while sequentially increasing the resolution of the image from low resolution to high resolution in four steps. Since the amount of information decreases when the resolution is reduced, the calculation can be sped up by introducing such a calculation in which the calculation area is limited.

  Further, it is not necessary to perform the three-axis optimization at each resolution, and it is effective to perform the three-axis optimization only for the calculation with the lowest resolution. For example, the frequency at which the evaluation function falls into a local solution tends to increase as the resolution increases. Therefore, if the three-axis optimization is performed at the lowest resolution in the initial stage of the optimization that is most affected by the magnitude of the initial displacement between the DR image and the DRR image, and the large displacement is reduced, Even when only the six-axis optimization is performed in the high-resolution optimization calculation, it is possible to prevent the evaluation function from falling into a local solution. Since optimization of three axes of parallel movement is performed first on the low resolution side where the amount of information is small, the calculation load does not increase, and the stagnation of the progress of optimization in the subsequent six-axis optimization is reduced. , The calculation time required to reach the optimal solution can be reduced.

  Further, regarding the one-dimensional optimization, whether or not to perform the one-dimensional optimization may be switched for each resolution. That is, by omitting the one-dimensional optimization on the low resolution side where the direction of optimization in the solution space does not largely deviate from the preferred solution search direction without considering the shooting direction, the optimization operation can be further performed. It is possible to increase the speed.

  In the above-described positioning of the subject, the result of alignment between the DR image and the DRR image created from the CT data is used for moving the top board 31 before irradiating the treatment beam from the radiation therapy apparatus 1. It is not always necessary to move the top plate 31. For example, the result of the alignment may be used to confirm whether there is any displacement during the treatment.

  In this embodiment, since the treatment table 30 is configured to move the top board 31 in six axes, the movement amount of the top board 31 calculated by the movement amount calculation unit 44 corresponds to the six directions of the corresponding axis. Are output to the table moving mechanism 32. In this embodiment, irrespective of whether or not it corresponds to the movement axis of the treatment table, optimization of parameters with 6 degrees of freedom is executed, and the movement amount based on the shift amount between the obtained DR image and DRR image is The data is transmitted to the top moving mechanism 32 by the amount corresponding to the moving axis of the treatment table. For example, if the treatment table 30 moves the top 31 in four axes (three-axis parallel movement + vertical axis rotation), the movement amount calculation unit 44 outputs only the movement amounts of the corresponding movement axes in four directions. You. It should be noted that, after calculating a global optimal solution by performing a six-dimensional optimization rather than a solution obtained by performing a four-dimensional optimization using the number of moving axes of the treatment table as a variable, a constraint on an immovable axis is obtained. It has been experimentally confirmed by the inventor of the present invention that the solution obtained by projecting the condition has the shortest distance from the optimal solution and the positioning error is small.

  In this embodiment, as described above, the shift amount of six axes is calculated, and if the moving axis of the treatment table is four axes, the configuration of transmitting the four axes to the top moving mechanism 32 is adopted. The present invention is not limited to the embodiment. For example, instead of the 6-axis optimization calculation, a 4-axis or 5-axis optimization operation corresponding to the movement axis of the treatment table, for example, optimizing parameters of 4 degrees of freedom or 5 degrees of freedom is executed, and each movement axis is calculated. May be transmitted to the table moving mechanism 32.

DESCRIPTION OF SYMBOLS 1 Radiotherapy apparatus 10 X-ray tube control part 11 X-ray tube 15 display part 16 input part 17 network 21 flat panel detector 30 treatment table 31 top plate 32 top plate moving mechanism 40 control part 41 image acquisition part 42 DRR image creation part 43 Positioning unit 44 Moving amount calculation unit 45 3-axis optimization unit 46 6-axis optimization unit 47 One-dimensional optimization unit

  The present invention relates to a positioning device and a positioning method for positioning a patient when performing radiation therapy on the patient.

  2. Description of the Related Art In radiotherapy for irradiating an affected part of a patient with radiation such as an X-ray, an electron beam, or a particle beam, it is necessary to accurately irradiate therapeutic radiation to the affected part. In such radiation treatment, first, an X-ray CT (Computed Tomography) scan is performed, and a treatment plan is formulated. When a treatment is performed by the radiation therapy apparatus, positioning is performed with the patient lying on the examination table in order to match the irradiation target of the treatment beam with the irradiation center.

  In an apparatus that uses an X-ray fluoroscopic image or CT data for positioning of a patient, a DR (Digital Radiography) image, which is an actual fluoroscopic image of an affected part of the patient and its surroundings, which is constrained to a treatment table by a fixture, is acquired, and a treatment plan is formulated. At this time, creation of a DRR (Digital Reconstructed Radiography) image, which is a virtual perspective projection, on the three-dimensional image data obtained by the X-ray CT scan at the time is executed. Then, by evaluating the similarity between the DR image and the DRR image (image registration), the shift amount between the actual position of the patient at the time of the radiation treatment and the position at the time of the treatment plan is calculated (Patent Document 1). And Patent Document 2).

  Before the positioning according to the positioning algorithm for matching the DR image and the DRR image, an initial setup called coarse positioning is performed. The initial setup may be performed, for example, by adjusting the position of the operator manually using a laser marking device or by superimposing the DR image and the DRR image displayed on the display by operating the input device. There is a deviation amount adjustment.

JP 2007-282877 A JP 2009-201556 A

  In a positioning algorithm that optimizes perspective projection parameters related to rotation and translation of CT data so that the degree of coincidence between the DR image and the DRR image is maximized, the amount of deviation between the DR image and the DRR image after the initial setup is optimized. Is the initial value. In the conventional parameter optimization, while changing the resolution of an image from a low resolution to a high resolution, an optimization operation is performed simultaneously in all six degrees of freedom of parallel movement of three orthogonal axes in three-dimensional space and rotational movement around each axis. Be executed.

  Since the initial setup is performed manually by the operator, there is an individual difference in the degree of deviation between the DR image and the DRR image. If the initial deviation amount between the DR image and the DRR image at the time of starting parameter optimization is large (for example, 1 cm or more), positioning may fail if optimization operations are performed simultaneously in all six degrees of freedom. was there. The cause of the failure may be a stagnation of the progress of the optimization calculation due to an attempt to forcibly adjust by rotation before the optimization of the parallel movement is completed. When the positioning fails, it is necessary to start over from the initial setup again, and the time from when the patient lays on the treatment table to when the patient can irradiate the treatment beam is long.

  The present invention has been made to solve the above-described problems, and has as its object to provide a positioning device and a positioning method that can easily perform positioning during treatment.

According to a first aspect of the present invention, there is provided a positioning device for positioning a subject when performing radiation therapy for irradiating a treatment beam toward an affected part of the subject on a treatment table, wherein the radiation irradiating unit and a radiation detector are arranged. An image acquisition unit that acquires two-dimensional radiation images of the subject in two different directions from an imaging system having the imaging system, and reproduces the geometrical arrangement of the imaging system in space and virtually converts the CT data collected by computer tomography into virtual data. A DRR image creating unit that creates DRR images of the subject in two different directions by performing perspective projection, a positioning unit that performs alignment of the radiation image and the DRR image, and the positioning unit And a movement amount calculation unit that outputs a movement amount of the treatment table from a shift amount between the radiation image and the DRR image calculated by: Of six degrees of freedom of the parameters of translation and rotation kicking, by performing the optimization operation in three degrees of freedom parameters for parallel movement simultaneously, three axes to determine the direction parallel displacement amount of the radiation image and the DRR image After reflecting the result of the optimization operation in the optimization unit and the three-axis optimization unit, optimization of the parameter of six degrees of freedom or the parameter of four or five degrees of freedom corresponding to the movement axis of the treatment table is performed. And a multi-axis optimizing unit that calculates shift amounts in the parallel direction and the rotation direction between the radiation image and the DRR image by simultaneously executing a conversion operation.

In the second invention, the optimization calculation in the three-axis optimization unit is performed after the initial setup, which is coarse positioning.

In the third aspect, the three-axis optimizing unit sets a three-dimensional evaluation function for evaluating a degree of coincidence between the radiation image and the DRR image until the deviation amount for three axes converges to a target value. Perform an iterative calculation.

In a fourth aspect, the three-axis optimizing unit performs iterative calculation based on the following equation (1) until the deviation amount for three axes converges to a target value.

In a fifth aspect, the positioning unit executes the positioning while changing the resolution of the radiation image and the DRR image stepwise from a low resolution to a high resolution, and the three-axis optimization unit performs The optimization operation of the three-degree-of-freedom parameter of the translation is performed at the resolution.

In the sixth aspect, when the multi-axis optimizing unit executes an optimization calculation of a parameter having six degrees of freedom, the movement amount calculating unit sets a parallel direction and a rotational direction between the radiation image and the DRR image. The amount of movement in six directions is calculated from the amount of displacement of the table, and among the amounts of movement in the six directions, the amount of movement corresponding to the movement axis of the treatment table is output.

In the seventh invention, the positioning unit determines whether or not the deviation amount has converged to a target value as a result of the optimization calculation by the multi-axis optimizing unit. And a one-dimensional optimizing unit that executes a parameter optimizing operation for one-dimensional parallel movement of the image pickup system along a shooting direction.

In the eighth invention, as a result of the optimization operation by the multi-axis optimizing unit, it is determined whether or not the deviation amount has converged to a target value. It is performed by comparison.

In the ninth aspect, as a result of the optimization operation performed by the multi-axis optimization unit, it is determined whether or not the deviation amount has converged to a target value. This is performed by comparison with the convergence determination value.

In the tenth aspect, different evaluation functions are used for the evaluation functions of the three-axis optimization unit, the multi-axis optimization unit, and the one-dimensional optimization unit.

In the eleventh aspect, the positioning unit terminates the optimization calculation when the calculation by the one-dimensional optimization unit does not reach the convergence determination value even if the calculation has reached a predetermined time or a predetermined number of times.

In a twelfth aspect, the present invention provides a positioning method for positioning a subject when performing radiation therapy for irradiating a treatment beam toward an affected part of the subject on a treatment table, comprising: An image acquisition step of acquiring two-dimensional radiation images of the subject in two different directions from an imaging system having the imaging system, and reproducing the geometrical arrangement of the imaging system in space, and virtualizing CT data collected in advance by computer tomography. A DRR image creating step of creating DRR images of the subject in two different directions by performing perspective projection, a positioning step of executing the alignment of the radiation image and the DRR image, and the positioning step A moving amount calculating step of outputting a moving amount of the treatment table from a shift amount between the radiation image and the DRR image calculated by Of six degrees of freedom of the parameters of translation and rotation in the space coordinate system, and the three-axis optimization step of performing an optimization calculation to optimize the parameters of three degrees of freedom of translation at the same time, in the 3-axis optimization process A multi-axis optimization step of simultaneously executing the optimization calculation of the parameter of 6 degrees of freedom or the parameter of 4 or 5 degrees of freedom corresponding to the moving axis of the treatment table after reflecting the result of the optimization calculation; Is provided.

According to the first to twelfth aspects, after optimizing the parameters for the three degrees of freedom of the translation, the parameters are optimized for the degrees of freedom including the rotation in the translation, so that the parameters are obtained from the imaging system. Even when the deviation amount of the initial position between the obtained radiation image and the DRR image created by performing virtual perspective projection on the CT data is large, it is possible to prevent a local solution from being obtained by the parameter optimization calculation. As described above, since the allowable range of the shift amount of the initial position between the radiographic image and the DRR image is increased, the case of re-starting from the initial setup as in the related art can be reduced without stalling the progress of the optimization calculation. Can be. Therefore, it is possible to shorten the workflow at the time of treatment with the radiation therapy apparatus.

According to the fifth aspect , when positioning is performed by gradually increasing the resolution of the radiographic image and the DRR image from a low resolution to a high resolution, optimization of the three-degree-of-freedom parameter of the parallel movement is performed in the initial stage of the optimization. , The calculation cost of the optimization can be reduced.

According to the sixth invention, 6 after calculating the amount of movement in six directions from the shift amount of freedom of parameters optimized by radiographic images obtained and DRR image, necessary correspond to the couch movement axis moving Since only the amount is output to the treatment table, there is no need to prepare a positioning algorithm for each degree of freedom of the movement axis of the treatment table. It is possible to reduce the positioning error as compared with the case where the parameters are optimized.

According to the seventh to eleventh aspects, since the one-dimensional optimization is performed along the imaging direction of the imaging system, the SID can be changed like the imaging system that performs imaging from two directions with a tilt angle to the patient. Even when (Source Image Distance) is long, it is possible to prevent the progress of the optimization calculation from being stagnated. In addition, it is possible to minimize the optimization route and shorten the positioning time.

It is a schematic diagram of a radiotherapy device to which the positioning device according to the present invention is applied. FIG. 2 is a block diagram of a control system including the positioning device according to the present invention. 9 is a flowchart illustrating a procedure for positioning a subject. 9 is a flowchart illustrating a procedure for parameter optimization. It is explanatory drawing which shows typically the process of optimization of the valley structure of an evaluation function and the similarity of an image when the one-dimensional optimization operation of this invention is performed.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a schematic diagram of a radiotherapy device 1 to which a positioning device according to the present invention is applied.

  The positioning device of the present invention includes an X-ray imaging system, and is used together with the radiotherapy device 1. The radiotherapy apparatus 1 is for performing radiotherapy on a patient (subject) on a top plate 31 of a treatment table 30, and includes a head 55 for emitting a treatment beam and a base 55 installed on the floor of a treatment room. A gantry 53 rotatably supported by the table 52 is provided. The radiation treatment apparatus 1 can change the irradiation direction of the treatment beam by rotating the gantry 53.

  The X-ray imaging system is for performing X-ray fluoroscopy in order to specify the position of the affected part of the patient who is lying on the top plate 31 of the treatment table 30, and includes X-ray tubes 11a to 11d as radiation irradiation units. And flat panel detectors 21a to 21d, which are radiation detectors for detecting X-rays transmitted through the subject and the top plate 31. The X-ray tubes 11a to 11d and the flat panel detectors 21a to 21d are arranged at positions for performing X-ray fluoroscopy on the subject from an oblique direction. Although not shown in FIG. 1, the X-ray tubes 11a to 11d are arranged in concave portions formed on the floor surface, and the concave portions are covered by a lid member forming a part of the floor. Further, an image intensifier (II) may be used as the radiation detector.

  X-rays emitted from the X-ray tube 11a are detected by the flat panel detector 21a, and the X-ray tube 11a and the flat panel detector 21a constitute a first imaging system. X-rays emitted from the X-ray tube 11b are detected by the flat panel detector 21b, and the X-ray tube 11b and the flat panel detector 21b constitute a second imaging system. X-rays emitted from the X-ray tube 11c are detected by the flat panel detector 21c, and the X-ray tube 11c and the flat panel detector 21c constitute a third imaging system. X-rays emitted from the X-ray tube 11d are detected by the flat panel detector 21d, and the X-ray tube 11d and the flat panel detector 21d constitute a fourth imaging system. When positioning the subject, two of the first to fourth imaging systems are selected so that the gantry 53 does not overlap the field of view, and X-rays from two different directions with respect to the subject are obtained. A fluoroscopy is performed.

  FIG. 2 is a block diagram of a control system including the positioning device according to the present invention.

  This positioning device includes a CPU (Central Processing Unit) for executing a logical operation, a GPU (Graphics Processing Unit) for executing various image processing, a ROM in which a program necessary for controlling the device is stored, and data and the like during control are temporarily stored. The control unit 40 is implemented by a computer having a memory such as a RAM to be stored.

  The control unit 40 is connected to the X-ray tube control unit 10 that controls the irradiation of X-rays from the X-ray tubes 11a to 11d, and the flat panel detectors 21a to 21d. The X-ray tube control unit 10 is connected to the X-ray tubes 11a to 11d, and is necessary for irradiating X-rays to two imaging systems selected from the X-ray tubes 11a to 11d during X-ray fluoroscopy. Supply tube voltage and tube current. The control unit 40 is also connected to the network 17, the display unit 15, the input unit 16, the radiotherapy device 1, and the treatment table 30. The top board 31 of the treatment table 30 can be moved in six axes of horizontal movement and rotation by a top board moving mechanism 32.

  The control unit 40 has, as a functional configuration, an image acquisition unit 41 that acquires information on a two-dimensional radiation image (DR image) detected by two of the selected imaging systems among the flat panel detectors 21a to 21d, A DRR image creation unit 42 for creating DRR images in two different directions by virtually performing perspective projection on three-dimensional CT data collected by prior computer tomography (X-ray CT imaging) via the network 17. A positioning unit 43 that matches the DR image obtained by performing fluoroscopy with the two imaging systems with the DRR image, and a movement amount calculation unit 44 that calculates the amount of movement of the top board 31 from the amount of displacement between the images. Prepare. These functional configurations are stored in a memory as a program and executed by the action of the CPU.

  The positioning unit 43 executes an optimization calculation of each parameter related to rotation and parallel movement of the perspective projection using a positioning algorithm for obtaining an optimal translation / rotation solution that minimizes the amount of deviation between the DR image and the DRR image. It has a function to do. Then, the optimization calculation is executed centering on the target isocenter located in the irradiation field of the radiation therapy apparatus 1. The positioning unit 43 includes a three-axis optimizing unit 45, a six-axis optimizing unit 46, and a one-dimensional optimizing unit 47 as components for realizing parameter optimization. The functions of the three-axis optimizing unit 45, the six-axis optimizing unit 46, and the one-dimensional optimizing unit 47 are stored in a memory as a program and executed by the operation of the CPU.

  Next, a positioning method in the positioning device having the above-described configuration will be described. FIG. 3 is a flowchart showing a procedure for positioning the subject.

  An X-ray fluoroscopy is performed on the subject on the top board 31 of the treatment table 30 by two imaging systems selected from the first to fourth imaging systems, and images are obtained from two of the flat panel detectors 21a to 21d. Information is obtained, and DR images from two different directions are obtained (step S1: image obtaining step).

  The geometry (geometric arrangement) of the imaging system is reproduced in a virtual space on a computer, and virtual perspective projection is performed on the previously acquired three-dimensional CT data. The CT data is acquired by an X-ray CT apparatus when formulating a treatment plan, and stored in a patient DB (not shown). The control unit 40 acquires a treatment plan and CT data via the network 17. Thereafter, by the operation of the DRR image creation unit 42, two-dimensional DRR images of the subject in two different directions are created by virtual perspective projection onto the three-dimensional CT data (step S2: DRR image creation step).

The geometry of the X-ray fluoroscopy includes the positions of any two of the X-ray tubes 11a to 11d, the positions of any two of the flat panel detectors 21a to 21d, and the position and orientation of the top plate 31 in the two selected imaging systems. Is included. Since the mechanical installation accuracy of these elements affects the final positioning accuracy, the installation position is periodically calibrated , and the calibration result is reflected on the geometry of X-ray fluoroscopy.

  When the DRR image is created, the voxel values of the CT image data are integrated along the projection line from any one of the X-ray tubes 11a to 11d in the selected imaging system to any of the flat panel detectors 21a to 21d. Integration).

  The parameters related to the parallel movement and rotation of the perspective projection are optimized so that the DR image and the DRR image match, and alignment is performed (step S3: alignment step). Here, as an evaluation function for evaluating the degree of coincidence between two images, a normalized mutual information (NMI), a gradient difference (GD: Gradient Difference), and a zero-average normalized cross-correlation (ZNCC: Nero- An evaluation function such as mean normalized cross-correlation that has been conventionally used for multi-modality image registration can be employed. In addition, by using NMI, GD, and ZNCC in combination, it is also possible to improve the evaluation accuracy of the degree of coincidence between the DR image and the DRR image.

  It is preferable that the calculation of the evaluation function is performed only in the region where the subject is present in the image. In addition, it is desirable that a part having a deformation that is not reproducible with CT data, such as a moving organ or a joint in the subject, is excluded from the calculation of the evaluation function.

  The amount of misalignment between the DR image and the DRR image obtained as a result of the alignment by the operation of the alignment unit 43 is determined by executing the function of the movement amount calculation unit 44 by the operation of the CPU configuring the control unit 40. It is converted into a top plate movement amount (step S4: movement amount calculation step). Then, this movement amount is transmitted from the movement amount calculation unit 44 to the top moving mechanism 32 of the treatment table 30. Thereafter, the top 31 is moved by the operation of the top moving mechanism 32 (step S5: top moving step). As described above, by moving the top 31 so that the subject is displaced by the positional shift amount between the DR image and the DRR image created from the CT data, the subject can be treated by the treatment beam emitted from the radiation therapy apparatus 1. Is positioned at the position and angle as in the treatment plan. After transposition of the top plate 31 and positioning of the subject, X-ray fluoroscopy is performed again, and the DR image and the DRR image at that time are displayed on the display unit 15 so that the images match. It is determined by the user whether or not there is a visual check. Then, a treatment beam is emitted from the head 55 of the radiation therapy apparatus 1 toward the affected part of the subject.

  The functions of the three-axis optimizing unit 45, the six-axis optimizing unit 46, and the one-dimensional optimizing unit 47, which are components of the positioning unit 43, will be described in more detail. FIG. 4 is a flowchart showing a procedure of parameter optimization.

  In this embodiment, the quasi-Newton method based on the BFGS formula is used as a method of the optimization calculation. The quasi-Newton method is a method that can obtain the minimum value and the maximum value of a k-dimensional function f (x). In the optimization operation, assuming that an evaluation function for evaluating the coincidence between the DRR image and the DR image is f (x), the following equation (from the initial value x0 of the point in the treatment room space to xi + 1 at the (i + 1) th point) becomes sufficiently small. Calculate iteratively according to 1).

  Here, H is an approximation of the inverse matrix of the Hessian matrix. Although several approximate expressions for H have been proposed, the BFGS formula given by the following expressions (2) and (3) has the highest calculation efficiency.

  When performing image registration, parameters are optimized for a total of six degrees of freedom, ie, three degrees of freedom of translation and three degrees of rotation of each of the X, Y, and Z axes of the treatment room spatial coordinate system. I do. In the present invention, before simultaneously optimizing a total of six axes, ie, three axes in the parallel direction and three axes in the rotation direction, first, optimization is performed on three axes in the parallel direction (three degrees of freedom) (step S1). S31: Triaxial optimization step). In the three-axis optimization step, the three-dimensional position x is updated according to the above equation (1), using the evaluation function f (x) as a three-dimensional function. The iterative calculation is performed until the deviation amounts for the three parallel axes converge to the target value. Note that the three-axis optimization is realized by the CPU executing a program read from the three-axis optimization unit 45.

  When the optimization for the three axes in the parallel direction is completed, the optimization for the six axes including the rotation direction is executed (step S32: a six-axis optimization process). In the six-axis optimization process, the six-dimensional position x is updated according to the above equation (1), assuming that the evaluation function f (x) is a sixth-order function that depends on six independent variables relating to six degrees of freedom of translation and rotation. I do. The initial value x0 of the six-dimensional position x is a position after the three-axis parameters of the parallel movement are optimized in the previous three-axis optimization step. Then, if the deviation amounts for a total of six axes of three axes of translation and three axes of rotation converge within a predetermined time or while repeating the calculation of a predetermined number of times (step S33), the optimization calculation is performed. finish. On the other hand, if the shift amounts of the six axes do not converge, one-dimensional optimization is performed (step S34: one-dimensional optimization step). The six-axis optimization is realized by the CPU executing a program read from the six-axis optimization unit 46. The six-axis optimizing unit 46 and the six-axis optimizing step correspond to the multi-axis optimizing unit and the multi-axis optimizing step of the present invention. In the present invention, the multi-axis means a movement axis having 4 to 6 degrees of freedom in which a rotation axis is added to three translation axes.

  Here, the X-ray fluoroscopy is performed from two different directions by the two selected imaging systems, and the values of the evaluation functions for evaluating the coincidence between the DRR image after the 6-axis optimization and the DR image are different. Obtained about the direction. The determination as to whether or not the shift amount between the two images has converged to the target value in step S33 is made based on whether or not the value of the evaluation function has reached the convergence determination value. When the values of the evaluation functions in two different directions are F1 and F2, the sum F1 + F2 of the evaluation functions in the two different directions is set as the value of the evaluation function to be compared with the convergence determination value. When one of the two different directions is more important in positioning, the sum may be obtained by weighting one of the two, instead of the simple sum of F1 and F2. After the three-axis optimization, it is determined whether or not to proceed to the six-axis optimization step. However, since the six-axis optimization is performed after adjusting the three axes of the parallel movement, the three-axis optimization is performed. It is not necessary to set the convergence determination value more strictly as the convergence determination after the six-axis optimization (step S33), and to perform the convergence determination of the optimization. You may make it advance to the step of conversion.

  If the determination in step S33 after the calculation of the six-axis optimization is “No”, one-dimensional optimization is performed (step S34). In the one-dimensional optimization step, parameters relating to one-dimensional parallel movement along the imaging direction in the two selected imaging systems are optimized. The evaluation function in the one-dimensional optimization process is a linear function that depends on one variable related to one-dimensional translation along the imaging direction. In this optimization, the Brent method, the golden section method, or the like can be used.

  FIG. 5 is an explanatory diagram schematically showing the process of optimizing the valley structure and parameters of the evaluation function when the one-dimensional optimization operation of the present invention is executed.

  The one-dimensional optimization unit 47 optimizes parameters related to one-dimensional parallel movement along the imaging direction, so that the optimization path proceeds along the valley structure. Therefore, it is possible to perform optimization in consideration of the imaging direction not only at the beginning of optimization but also at the end of optimization, and X-rays can be obtained from two different directions having an inclination angle with respect to the subject as shown in FIG. This solves the problem of the stagnation of the progress of the optimization calculation due to the longer SID when fluoroscopy is performed. As a result, the optimization operation is efficiently performed on the shortest path, the total calculation time can be shortened as compared with the related art, and the positioning accuracy can be improved.

  When the evaluation function values in two different directions are F1 and F2, respectively, as in the case of the three-axis optimization and the six-axis optimization, May be used, but it is more preferable to use only the evaluation function having a large value (either F1 or F2). That is, when the optimization path is along the valley structure of the evaluation function, the evaluation function with a small value has a small contribution to the optimization, and therefore the objective function with a large value on the side where the displacement is not along the imaging direction. Performing the optimization only with the use of the above makes it possible to speed up the calculation.

  The one-dimensional optimization is repeatedly executed until the value of the evaluation function reaches the convergence determination value (Step S35: convergence determination step). In this embodiment, different evaluation functions are used for the optimization of three axes and six axes and one-dimensional optimization, and different values are used for the convergence determination value of the value of the evaluation function. As described above, by using the evaluation function and the convergence determination value suitable for each, the parameters can be more appropriately optimized. The one-dimensional optimization may be omitted when the difference between the evaluation function values F1 and F2 in two different directions is small.

  Further, when the convergence determination value is not reached even after reaching the predetermined calculation time or the predetermined number of calculations, the repetition calculation of the optimization is terminated (step S36: termination determination step). By providing such a time limit, the burden on the patient fixed on the top plate 31 in the same posture and a decrease in the throughput of the radiotherapy apparatus 1 are reduced.

  In addition, the above-described optimization operation is performed while gradually changing the resolution of the DR image and the DRR image from a low resolution to a high resolution using a downsampling method. For example, the calculation is repeated while sequentially increasing the resolution of the image from low resolution to high resolution in four steps. Since the amount of information decreases when the resolution is reduced, the calculation can be sped up by introducing such a calculation in which the calculation area is limited.

  Further, it is not necessary to perform the three-axis optimization at each resolution, and it is effective to perform the three-axis optimization only for the calculation with the lowest resolution. For example, the frequency at which the evaluation function falls into a local solution tends to increase as the resolution increases. Therefore, if the three-axis optimization is performed at the lowest resolution in the initial stage of the optimization that is most affected by the magnitude of the initial displacement between the DR image and the DRR image, and the large displacement is reduced, Even when only the six-axis optimization is performed in the high-resolution optimization calculation, it is possible to prevent the evaluation function from falling into a local solution. Since optimization of three axes of parallel movement is performed first on the low resolution side where the amount of information is small, the computational burden does not increase, and the stagnation of the progress of optimization in the subsequent six-axis optimization is reduced. , The calculation time required to reach the optimal solution can be reduced.

  Further, regarding the one-dimensional optimization, whether or not to perform the one-dimensional optimization may be switched for each resolution. That is, by omitting the one-dimensional optimization on the low resolution side where the direction of optimization in the solution space does not largely deviate from the preferred solution search direction without considering the shooting direction, the optimization operation can be further performed. It is possible to increase the speed.

  In the above-described positioning of the subject, the result of alignment between the DR image and the DRR image created from the CT data is used for moving the top board 31 before irradiating the treatment beam from the radiation therapy apparatus 1. It is not always necessary to move the top plate 31. For example, the result of the alignment may be used to confirm whether there is any displacement during the treatment.

  In this embodiment, since the treatment table 30 is configured to move the top board 31 in six axes, the movement amount of the top board 31 calculated by the movement amount calculation unit 44 corresponds to the six directions of the corresponding axis. Are output to the table moving mechanism 32. In this embodiment, irrespective of whether or not it corresponds to the movement axis of the treatment table, optimization of parameters with 6 degrees of freedom is executed, and the movement amount based on the shift amount between the obtained DR image and DRR image is The data is transmitted to the top moving mechanism 32 by the amount corresponding to the moving axis of the treatment table. For example, if the treatment table 30 moves the top 31 in four axes (three-axis parallel movement + vertical axis rotation), the movement amount calculation unit 44 outputs only the movement amounts of the corresponding movement axes in four directions. You. It should be noted that, after calculating a global optimal solution by performing a six-dimensional optimization rather than a solution obtained by performing a four-dimensional optimization using the number of moving axes of the treatment table as a variable, a constraint on an immovable axis is obtained. It has been experimentally confirmed by the inventor of the present invention that the solution obtained by projecting the condition has the shortest distance from the optimal solution and the positioning error is small.

  In this embodiment, as described above, the shift amount of six axes is calculated, and if the moving axis of the treatment table is four axes, the configuration of transmitting the four axes to the top moving mechanism 32 is adopted. The present invention is not limited to the embodiment. For example, instead of the 6-axis optimization calculation, a 4-axis or 5-axis optimization operation corresponding to the movement axis of the treatment table, for example, optimizing parameters of 4 degrees of freedom or 5 degrees of freedom is executed, and each movement axis is calculated. May be transmitted to the table moving mechanism 32.

DESCRIPTION OF SYMBOLS 1 Radiotherapy apparatus 10 X-ray tube control part 11 X-ray tube 15 display part 16 input part 17 network 21 flat panel detector 30 treatment table 31 top plate 32 top plate moving mechanism 40 control part 41 image acquisition part 42 DRR image creation part 43 Positioning unit 44 Moving amount calculation unit 45 3-axis optimization unit 46 6-axis optimization unit 47 One-dimensional optimization unit

Claims (5)

When performing radiotherapy to irradiate a treatment beam toward the affected part of the subject on the treatment table, a positioning device for positioning the subject,
An image acquisition unit that acquires a two-dimensional radiation image of the subject in two different directions from an imaging system having a radiation irradiation unit and a radiation detector;
A DRR image that reproduces the geometric arrangement of the imaging system in space and virtually performs perspective projection on CT data acquired in advance by computer tomography, thereby creating DRR images of the subject in two different directions. A creation department;
An alignment unit that performs alignment between the radiation image and the DRR image;
A movement amount calculation unit that outputs a movement amount of the treatment table from a shift amount between the radiation image and the DRR image calculated by the positioning unit;
With
The positioning unit,
By performing an optimization calculation of the three-degree-of-freedom parameters of the parallel movement among the six-degree-of-freedom parameters of the parallel movement and the rotation in the spatial coordinate system, the shift amount in the parallel direction between the radiation image and the DRR image is obtained. 3-axis optimization unit,
After reflecting the result of the optimization calculation in the three-axis optimization unit, the optimization calculation of the parameters of six degrees of freedom or the parameters of four or five degrees of freedom corresponding to the movement axis of the treatment table is executed. By this, a multi-axis optimization unit that determines the amount of shift in the parallel direction and the rotation direction between the radiation image and the DRR image,
And a positioning device. The positioning device according to claim 1,
The alignment unit performs alignment while gradually changing the resolution of the radiation image and the DRR image from a low resolution to a high resolution,
The positioning device, wherein the three-axis optimizing unit executes an optimization calculation of the three-degree-of-freedom parameter of the translation at the lowest resolution. The positioning device according to claim 1,
When the multi-axis optimizing unit performs the optimization calculation of the parameter having six degrees of freedom, the moving amount calculating unit calculates the six directions based on the amounts of displacement in the parallel direction and the rotation direction between the radiation image and the DRR image. A positioning device that calculates a moving amount of the treatment table and outputs a moving amount corresponding to a moving axis of the treatment table among the moving amounts in the six directions. The positioning device according to claim 1,
The positioning device, comprising: a one-dimensional optimizing unit configured to execute a parameter optimizing operation regarding one-dimensional parallel movement of the imaging system along a photographing direction. When performing radiotherapy to irradiate a treatment beam toward the affected part of the subject on the treatment table, a positioning method for positioning the subject,
An image acquisition step of acquiring a two-dimensional radiation image of the subject in two different directions from an imaging system having a radiation irradiation unit and a radiation detector,
A DRR image that reproduces the geometric arrangement of the imaging system in space and virtually performs perspective projection on CT data acquired in advance by computer tomography, thereby creating DRR images of the subject in two different directions. Creation process,
An alignment step of performing alignment between the radiation image and the DRR image;
A movement amount calculation step of outputting a movement amount of the treatment table from a shift amount between the radiation image and the DRR image calculated in the alignment step,
With
The positioning step includes:
A three-axis optimization step of executing an optimization operation for optimizing a parameter of three degrees of freedom of translation among parameters of six degrees of freedom of translation and rotation in a spatial coordinate system;
After reflecting the result of the optimization calculation in the three-axis optimization process, the optimization calculation of the parameter of six degrees of freedom or the parameter of four or five degrees of freedom corresponding to the movement axis of the treatment table is performed. Axis optimization process,
And a positioning method.

Images