WO2020125332A1 - 用于正交双层光栅装置的动态调强子野分割方法 - Google Patents

用于正交双层光栅装置的动态调强子野分割方法 Download PDF

Info

Publication number
WO2020125332A1
WO2020125332A1 PCT/CN2019/120600 CN2019120600W WO2020125332A1 WO 2020125332 A1 WO2020125332 A1 WO 2020125332A1 CN 2019120600 W CN2019120600 W CN 2019120600W WO 2020125332 A1 WO2020125332 A1 WO 2020125332A1
Authority
WO
WIPO (PCT)
Prior art keywords
blade
subfield
intensity
grating
blades
Prior art date
Application number
PCT/CN2019/120600
Other languages
English (en)
French (fr)
Inventor
文虎儿
姚毅
Original Assignee
苏州雷泰医疗科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 苏州雷泰医疗科技有限公司 filed Critical 苏州雷泰医疗科技有限公司
Priority to US17/417,085 priority Critical patent/US20220047893A1/en
Publication of WO2020125332A1 publication Critical patent/WO2020125332A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61N—ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY
    • A61N5/00—Radiation therapy
    • A61N5/10—X-ray therapy; Gamma-ray therapy; Particle-irradiation therapy
    • A61N5/103—Treatment planning systems
    • A61N5/1031—Treatment planning systems using a specific method of dose optimization
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61N—ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY
    • A61N5/00—Radiation therapy
    • A61N5/10—X-ray therapy; Gamma-ray therapy; Particle-irradiation therapy
    • A61N5/1042—X-ray therapy; Gamma-ray therapy; Particle-irradiation therapy with spatial modulation of the radiation beam within the treatment head
    • A61N5/1043—Scanning the radiation beam, e.g. spot scanning or raster scanning
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61N—ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY
    • A61N5/00—Radiation therapy
    • A61N5/10—X-ray therapy; Gamma-ray therapy; Particle-irradiation therapy
    • A61N5/1042—X-ray therapy; Gamma-ray therapy; Particle-irradiation therapy with spatial modulation of the radiation beam within the treatment head
    • A61N5/1045—X-ray therapy; Gamma-ray therapy; Particle-irradiation therapy with spatial modulation of the radiation beam within the treatment head using a multi-leaf collimator, e.g. for intensity modulated radiation therapy or IMRT
    • G—PHYSICS
    • G21—NUCLEAR PHYSICS; NUCLEAR ENGINEERING
    • G21K—TECHNIQUES FOR HANDLING PARTICLES OR IONISING RADIATION NOT OTHERWISE PROVIDED FOR; IRRADIATION DEVICES; GAMMA RAY OR X-RAY MICROSCOPES
    • G21K1/00—Arrangements for handling particles or ionising radiation, e.g. focusing or moderating
    • G21K1/02—Arrangements for handling particles or ionising radiation, e.g. focusing or moderating using diaphragms, collimators
    • G21K1/04—Arrangements for handling particles or ionising radiation, e.g. focusing or moderating using diaphragms, collimators using variable diaphragms, shutters, choppers
    • G21K1/046—Arrangements for handling particles or ionising radiation, e.g. focusing or moderating using diaphragms, collimators using variable diaphragms, shutters, choppers varying the contour of the field, e.g. multileaf collimators
    • G—PHYSICS
    • G21—NUCLEAR PHYSICS; NUCLEAR ENGINEERING
    • G21K—TECHNIQUES FOR HANDLING PARTICLES OR IONISING RADIATION NOT OTHERWISE PROVIDED FOR; IRRADIATION DEVICES; GAMMA RAY OR X-RAY MICROSCOPES
    • G21K1/00—Arrangements for handling particles or ionising radiation, e.g. focusing or moderating
    • G21K1/06—Arrangements for handling particles or ionising radiation, e.g. focusing or moderating using diffraction, refraction or reflection, e.g. monochromators
    • G21K1/067—Arrangements for handling particles or ionising radiation, e.g. focusing or moderating using diffraction, refraction or reflection, e.g. monochromators using surface reflection, e.g. grazing incidence mirrors, gratings
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61N—ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY
    • A61N5/00—Radiation therapy
    • A61N5/10—X-ray therapy; Gamma-ray therapy; Particle-irradiation therapy
    • A61N5/103—Treatment planning systems
    • A61N5/1031—Treatment planning systems using a specific method of dose optimization
    • A61N2005/1032—Genetic optimization methods
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61N—ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY
    • A61N5/00—Radiation therapy
    • A61N5/10—X-ray therapy; Gamma-ray therapy; Particle-irradiation therapy
    • A61N2005/1092—Details
    • A61N2005/1094—Shielding, protecting against radiation

Definitions

  • the invention relates to the field of medical technology, and mainly relates to a dynamic sliding-window subfield segmentation method for an orthogonal double-layer grating device for reverse intensity modulated radiotherapy.
  • the more common technique in radiotherapy is the intensity modulation using grating.
  • the movement of the grating blade can have a very good conformation effect on the target area, and at the same time can reduce the radiation damage of normal tissues.
  • the thinner the blades of the multi-leaf collimator the greater the number, the better the conformity of the multi-leaf collimator, but for the conventional single-layer grating, because the blade can only move in one direction, its blade
  • the conformability in the thickness direction is limited.
  • the blade thickness direction is improved compared to the single-layer grating, it is still limited by the influence of the blade thickness and cannot be moved or formed arbitrarily. Location of the irradiation unit.
  • orthogonal double-layer gratings (shown in Figures 1 and 2) have several advantages:
  • Dynamic intensity-modulated subfield segmentation achieves segmentation of intensity maps by simultaneously controlling the movement speed and irradiation dose rate of each pair of blades.
  • the segmentation algorithms of each pair of blades are calculated independently without disturbing each other.
  • dynamic segmentation has high irradiation efficiency and steep target area dose curve, which has obvious advantages.
  • dynamic segmentation requires more MU to complete, and the control system is more complicated.
  • the present invention proposes a dynamic sliding-window subfield segmentation method for orthogonal double-layer grating devices.
  • S1 Construct a virtual single-layer grating.
  • the speed of the virtual single-layer grating is the speed composite of the orthogonal double-layer grating.
  • step S2 Obtain the optimized intensity map, and rotate the optimized intensity map according to the installation angle of the virtual single-layer grating obtained in step S1;
  • step S3 Perform dynamic intensity calculation on the rotated optimized intensity map obtained in step S2.
  • the calculation result includes the time point of the speed inflection point and the shape of each subfield.
  • a preliminary calculated intensity map can be obtained and optimized
  • the intensity map comparison can check whether the subfield formed by the virtual single-layer grating meets expectations;
  • step S1 the speed of the virtual single-layer grating is the speed composite of the orthogonal grating A and the grating B, and the speed magnitude and direction are v, ⁇ ;
  • v 1 is the maximum speed of the blades in the horizontal direction
  • v 2 is the maximum speed of the blades in the vertical direction
  • the thickness of the virtual single-layer grating takes a smaller value
  • the number of blade pairs takes a larger value
  • the remaining properties are the same as those of grating A or grating B is consistent.
  • step S2 further includes the following content.
  • the sampling interval of the optimized intensity map in the blade thickness direction is directly divided according to the blade thickness.
  • the direction of movement is divided at a custom interval, with a value of 0.25.
  • step S3 specifically includes the following steps:
  • the minimum distance (Gap) between the blade pairs is known, so the minimum increment per unit length needs to be calculated, as shown in equation (3), when the blade runs at a physical maximum speed per unit length , The flux per unit length is the smallest, and the product of min slope and Gap is the minimum intensity value that the grating can achieve;
  • S3.4 Determine the position of the left and right blades in the subfield, and use the speed inflection point as the dividing point to calculate the position of the left and right blades at each dividing point;
  • it also includes an optimized orthogonal double-layer grating dynamic intensity modulation subfield segmentation method, which is specifically a subfield weight optimization method;
  • the subfield weight optimization method is to optimize the time point of each subfield under the condition of fixed blade motion trajectory.
  • the optimization goal is the second norm of the difference between the segmented intensity map and the optimized intensity map, as shown in equation (4);
  • the segmentation intensity graph can be regarded as a linear superposition of each subfield, as shown in equation (5);
  • I seg is the intensity map formed by a single subfield
  • u i is the subfield weight
  • the objective function is J obj .
  • an optimized orthogonal double-layer grating dynamic intensity-modulated subfield segmentation method which is specifically a blade motion trajectory optimization method
  • the blade motion trajectory optimization method is to optimize the objective function under certain conditions under the condition that the subfield weight is fixed and the motion trajectory of each blade is used as a variable.
  • the blade motion trajectory optimization method specifically includes the following steps:
  • the constraints on the blade movement are: tungsten gate constraints, blade physical constraints and speed constraints.
  • the constraints are as shown in equation (8):
  • i, j is the blade number of the horizontal grating and the vertical grating corresponding to the point (x, y);
  • i, j is the blade number of the horizontal grating and the vertical grating corresponding to the point (x, y);
  • the segmentation intensity graph can be regarded as a linear superposition of each subfield, as shown in equation (5);
  • I seg is the intensity map formed by a single subfield
  • I seg I(x, y), u i is the subfield weight
  • the objective function is J obj
  • the optimization objective function is the second norm of the difference between the segmentation intensity graph and the optimization intensity graph, as shown in equation (4);
  • S8 Start an external loop, find the row and column with the largest difference between the segmentation intensity map and the optimized intensity map, and the evaluation standard of the difference value is the second norm;
  • S9 Find the raster serial number corresponding to the row and column with the largest difference.
  • the raster has a fixed correspondence with the intensity map.
  • the raster serial number can be calculated according to the number of the intensity map;
  • step S11 Cyclically change the positions of the upper, lower, left and right blades. According to step S10, all four blades will get an active range. The maximum and minimum values of the four blade ranges are cyclically taken to disturb the positions of the four blades;
  • step S12 If the objective function drops or the internal loop exceeds the limit, step S13 is entered, otherwise steps S10-S11 are repeated;
  • Fig. 1 is an example diagram 1 of orthogonal double-layer grating solving sub-field segmentation of multiple connected regions
  • FIG. 2 is an example diagram 2 of orthogonal double-layer grating solving sub-field segmentation of multiple connected regions
  • FIG. 3 is an overall flowchart of a dynamic intensity-modulated subfield segmentation method for an orthogonal double-layer grating device provided by an embodiment of the present invention
  • FIG. 4 is a flowchart of a blade motion trajectory optimization method provided by an embodiment of the present invention.
  • Figure 5 is one of the comparison charts between the optimized intensity map and the rotated intensity map
  • FIG. 5(a) is one of the optimized intensity graphs provided by the embodiment of the present invention.
  • Figure 5(b) is one of the optimized intensity maps corresponding to Figure 5(a) after rotation;
  • Figure 6 is the second comparison between the optimized intensity map and the rotated intensity map
  • FIG. 6(a) is the second optimized intensity graph provided by an embodiment of the present invention.
  • Figure 6(b) is the second optimized intensity graph corresponding to Figure 6(a) after rotation
  • Figure 7(b) is the subfield overlay corresponding to Figure 7(a);
  • FIG. 9 is a second conformal diagram of orthogonal double-layer grating subfields provided by an embodiment of the present invention.
  • Figure 10 is an illustration of the overspeed condition of orthogonal double-layer grating blades
  • Figure 11 is the iterative curve of blade position optimization
  • FIG. 12(a) is the third optimized intensity graph provided by an embodiment of the present invention.
  • FIG. 12(b) is the segmentation intensity diagram corresponding to FIG. 12(a);
  • Figure 12(c) is the optimized intensity graph corresponding to Figure 12(a);
  • FIG. 13 is a schematic diagram of the upper and lower two layers of grating cooperate to divide multiple shooting fields
  • FIG. 14 is a schematic diagram of the upper and lower two-layer grating cooperate to optimize the fitting of the envelope of the shooting field edge.
  • the upper grating 1 and the lower grating 2 can cooperate with each other to complete the simultaneous construction of multiple shooting fields 3 (four shooting fields 3 in the figure) as illustrated in FIG. 13, It can also cooperate with each other.
  • the upper and lower gratings cooperate to optimize the envelope of the shooting edge. 13 and 14 are mainly for ease of understanding, and the number and size of the actual grating blades are different.
  • the present invention proposes a dynamic intensity-modulated subfield segmentation method for an orthogonal double-layer grating device, which is based on an orthogonal double-layer grating device for radiotherapy equipment,
  • the device is installed under the accelerator head of the radiotherapy equipment and includes:
  • the upper grating blades and the lower grating blades, the planes of the upper grating blades and the lower grating blades are parallel to each other and perpendicular to the direction of the rays emitted by the accelerator, and the movement directions of the upper grating blades and the lower grating blades are orthogonal;
  • the upper grating blades include the left blade and the right blade, which are used to search and move to the left and right sides of the target area;
  • the lower grating blades include upper blades and lower blades, which are used to search and move to the upper and lower sides of the target area;
  • the controller is used to drive each sub-blade in the left and right blades, the upper and lower blades to move independently, so as to achieve the purpose of conforming to the target area.
  • the orthogonal double-layer grating device Compared with the traditional single-layer grating and double-layer parallel grating, the orthogonal double-layer grating device has higher conformity, and can achieve the positioning accuracy of less than 1 mm in both directions.
  • a dynamic intensity modulated subfield segmentation method for an orthogonal double-layer grating device it specifically includes the following steps, as shown in FIG. 3:
  • S1 Construct a virtual single-layer grating.
  • the speed of the virtual single-layer grating is the speed composite of the orthogonal double-layer grating.
  • step S2 Obtain the optimized intensity map. To facilitate subfield segmentation, rotate the optimized intensity map according to the installation angle of the virtual single-layer grating obtained in step S1;
  • step S3 Perform dynamic intensity calculation on the rotated optimized intensity map obtained in step S2.
  • the calculation result includes the time point of the speed inflection point and the shape of each subfield.
  • a preliminary calculated intensity map can be obtained and optimized
  • the intensity map comparison can check whether the subfield formed by the virtual single-layer grating meets expectations;
  • step S3 is a dynamic intensity-modulated subfield segmentation step of the virtual single-layer grating.
  • step S4 obtains the subfield shapes at each speed inflection point, and step S4 reversely rotates the subfield shapes by ⁇ .
  • step S5 the orthogonal double-layer grating conforms to the derotated subfield obtained in step S4.
  • the subfield formed by the virtual single-layer grating is implemented with an orthogonal double-layer grating. The motion curve of the double-layer grating.
  • step S1 the speed of the virtual single-layer grating is orthogonal grating A (MLC1) and grating B ( MLC2) speed synthesis, speed magnitude and direction are v, ⁇ ;
  • v 1 is the maximum speed of the blades in the horizontal direction
  • v 2 is the maximum speed of the blades in the vertical direction
  • the thickness of the virtual single-layer grating takes a smaller value (here 0.25 or other)
  • the logarithm of the blade takes a larger value (It can be 256 or other here)
  • the rest of the properties are consistent with raster A or raster B.
  • step S2 the following content is included.
  • the sampling interval of the optimized intensity map in the blade thickness direction is directly divided by the blade thickness in the blade movement direction It is divided according to a custom interval, and the value is 0.25.
  • step S3 specifically includes the following steps:
  • the minimum distance (Gap) between the blade pairs is known, so the minimum increment per unit length needs to be calculated, as shown in equation (3), when the blade runs at a physical maximum speed per unit length , The flux per unit length is the smallest, and the product of min slope and Gap is the minimum intensity value that the grating can achieve;
  • S3.4 Determine the position of the left and right blades in the subfield, and use the speed inflection point as the dividing point to calculate the position of the left and right blades at each dividing point;
  • the remaining characteristic technologies are the same, the difference is that it also includes an optimized orthogonal double-layer grating dynamic intensity modulation subfield segmentation method, which is specifically subfield weight optimization law;
  • the subfield weight optimization method is to optimize the time point of each subfield under the condition of fixed blade motion trajectory.
  • the optimization goal is the second norm of the difference between the segmented intensity map and the optimized intensity map, as shown in equation (4);
  • the segmentation intensity graph can be regarded as a linear superposition of each subfield, as shown in equation (5);
  • I seg is the intensity map formed by a single subfield
  • u i is the subfield weight
  • the objective function is J obj .
  • the subfield weight optimization method is a constrained quadratic programming problem, which can be solved according to the general solution.
  • the remaining feature technologies are the same, the difference is that it also includes an optimized orthogonal double-layer grating dynamic intensity-modulated subfield segmentation method, which specifically is blade motion trajectory optimization law;
  • the blade motion trajectory optimization method is to optimize the objective function under certain conditions under the condition that the subfield weight is fixed and the motion trajectory of each blade is used as a variable.
  • the MU step time point of speed inflection point calculated by the virtual single-layer grating is the time interval.
  • the blade trajectory can be represented by a series of points. According to preliminary statistics, the number of trajectory points of all blades is about 20,000, that is to say, the optimization variables are about 20,000. Common optimization methods, such as nonlinear optimization, genetic algorithm, and particle swarm algorithm are not significant. To this end, an optimization algorithm is provided here. The whole optimization process is divided into outer loop and inner loop. The outer loop searches for the row and column with the largest difference between the segmentation intensity map and the optimization intensity map, and calculates the corresponding blade serial number.
  • the purpose of this step is to lock the blade serial number to be optimized and reduce the number of variables to be optimized; if the number of outer loop iterations exceeds The limit or the objective function meets the requirements, then exit the entire loop.
  • the inner loop randomly selects the blade moment, calculates the blade's active range according to the constraints, changes between the maximum and minimum values of the blade's active range, and observes whether the objective function becomes smaller; if the objective function drops or the number of inner loop iterations exceeds, Jump out of the inner loop.
  • blade motion trajectory optimization method specifically includes the following steps, as shown in Figure 4:
  • the constraints on the blade movement are: tungsten gate constraints, blade physical constraints (the position of the right blade is greater than the left blade, the position of the upper blade is greater than the lower blade), and the speed constraint.
  • the constraints are as shown in equation (8):
  • i, j is the blade number of the horizontal grating and the vertical grating corresponding to the point (x, y);
  • i, j is the blade number of the horizontal grating and the vertical grating corresponding to the point (x, y);
  • the segmentation intensity graph can be regarded as a linear superposition of each subfield, as shown in equation (5);
  • I seg is the intensity map formed by a single subfield
  • I seg I(x, y), u i is the subfield weight
  • the objective function is J obj
  • the optimization objective function is the second norm of the difference between the segmentation intensity graph and the optimization intensity graph, as shown in equation (4);
  • S8 Start the external loop, find the row and column with the largest difference between the segmentation intensity map and the optimized intensity map, and the evaluation standard of the difference value is the second norm;
  • S9 Find the raster serial number corresponding to the row and column with the largest difference.
  • the raster has a fixed correspondence with the intensity map.
  • the raster serial number can be calculated according to the number of the intensity map;
  • step S11 Cyclically change the positions of the upper, lower, left and right blades. According to step S10, all four blades will get an active range. The maximum and minimum values of the four blade ranges are cyclically taken to disturb the positions of the four blades;
  • step S12 If the objective function drops or the internal loop exceeds the limit, step S13 is entered, otherwise steps S10-S11 are repeated;
  • the present invention provides a dynamic intensity-modulated subfield division method for orthogonal double-layer grating devices, and provides two methods for optimizing this division: blade motion trajectory optimization method and subfield weight optimization method.
  • the subfield weight optimization method is to optimize the time point of each subfield under the fixed blade motion trajectory.
  • the blade motion trajectory optimization method is to use the blade motion trajectory as a variable under the condition that the subfield weight is fixed, under certain constraints, Optimize the objective function.
  • the invention starts by constructing a virtual single-layer grating, and the installation direction of the grating is the speed synthesis direction of the orthogonal double-layer grating.
  • the grating installation angle is rotated on the obtained optimized intensity map.
  • the rotated graph is still rectangular, and the dimension is expanded, and the entire intensity map becomes larger.
  • the invention defines the sampling interval of the graphics after rotation, which is not completely consistent with the sampling interval of the optimized intensity map.
  • the optimized intensity maps before and after rotation are shown in FIGS. 5 and 6.
  • the dynamic intensity adjustment calculation is performed on the optimized intensity map after rotation.
  • the calculation result includes the time point of the speed inflection point and the shape of each subfield.
  • a preliminary calculated intensity map can be obtained. Comparison with the optimized intensity map can be checked Whether the subfield formed by the virtual grating is as expected. It can be seen from Fig. 7 that the subfield formed by the virtual single-layer grating is reasonable.
  • the orthogonal double-layer grating After the subfield formed by the virtual single-layer grating is reversely rotated, the real subfield is obtained, and the existing orthogonal double-layer grating is used to conform the reversely rotated subfield, so that the orthogonal double-layer grating is obtained.
  • An example of the conformal shape of the orthogonal double-layer grating subfields is shown in Figures 8 and 9, where the blank area is the subfield after reverse rotation, and the network shape is the orthogonal double-layer grating. It can be seen that the orthogonal double-layer grating can be properly shaped Ziye.
  • the present invention proposes two optimization methods:
  • the blade motion trajectory optimization method takes the blade position as a variable, the initial value is the blade position formed by the speed synthesis algorithm, adds constraints, and optimizes with the second norm of the difference of the superposition intensity map of the subfield as the goal;
  • the constraint function includes tungsten gate constraints, physical constraints, and velocity constraints.
  • the external data required are: tungsten gate parameters, grating speed, and grating minimum interval.
  • the constraint function is shown in equation (8). At the same time, according to this constraint function, the movement range of a certain blade at a certain moment can be calculated.
  • the blade position is optimized according to the steps shown in Figure 4.
  • the curve of the number of iterations and the objective function is shown in Figure 11. It can be seen that during the iteration process, the objective function is slowly decreasing. After optimization, the segmentation intensity graph is closer to the optimization intensity graph. .
  • the subfield weight optimization method is completed at one time, without repeated iteration. After optimization, the objective function drops from 2.8254 to 1.9457.
  • the comparison between the optimized intensity graph, the segmented intensity graph and the optimized intensity graph is shown in Figure 12, which shows that the optimization effect is obvious.
  • the orthogonal double-layer grating has obvious advantages in the conformity of the target area and the treatment efficiency.
  • the invention provides a dynamic intensity-modulated subfield division method for the orthogonal double-layer grating device.
  • the core of the segmentation algorithm is to construct the virtual single-layer grating after synthesizing the speed of the two-layer grating, and perform the sliding-window segmentation of the single-layer grating, and finally use the two-layer grating to shape each subfield. This method provides a feasible solution for the dynamic sliding-window segmentation of orthogonal double-layer gratings.
  • the segmentation algorithm provides two optimization methods: blade motion trajectory optimization method and subfield weight optimization method.
  • the blade motion trajectory optimization method is to optimize the objective function under certain constraints under the condition that the subfield weights are fixed and the motion trajectory of each blade is used as a variable.
  • the subfield weight optimization method is to optimize the time point of each subfield under the condition of fixed blade movement trajectory. Both optimization methods can reduce the error value of the segmentation intensity map and improve the optimization effect.

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Pathology (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Physics & Mathematics (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Radiation-Therapy Devices (AREA)
  • Mechanical Light Control Or Optical Switches (AREA)
  • Exposure And Positioning Against Photoresist Photosensitive Materials (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种用于正交双层光栅装置的动态调强子野分割方法，分割算法的核心是将两层光栅的速度合成后构造虚拟单层光栅，进行单层光栅的动态调强分割，最后用两层光栅适形各子野。为了进一步减少分割误差，动态调强子野分割方法提供了两种优化方法：叶片运动轨迹优化法和子野权重优化法。叶片运动轨迹优化法是在子野权重固定的条件下，以各叶片的运动轨迹为变量，在一定约束下，优化目标函数。子野权重优化法是叶片运动轨迹固定条件下，优化各子野的时间点。两种优化方法都可减少分割强度图的误差值，改善优化效果。

Description

用于正交双层光栅装置的动态调强子野分割方法 技术领域
本发明涉及医疗技术领域，主要涉及一种逆向调强放疗的用于正交双层光栅装置的动态调强(sliding-window)子野分割方法。
背景技术
目前，放射治疗中较普遍的技术是应用光栅进行强度调制。通过光栅叶片的运动，可以对靶区有着非常好的适形效果，同时可以减少正常组织的放射损伤。一般情况下，多叶准直器的叶片越薄，数量越多，多叶准直器的适形度越好，但是对于常规的单层光栅，由于叶片只能在一个方向上运动，其叶片厚度方向的适形能力有限，对于平行的双层光栅，虽然其叶片厚度方向相比于单层光栅的适形能力有所提高，但还是受限于叶片厚度的影响，并不能运动或者形成任意位置的照射单元。
相比之下，正交双层光栅(如图1和图2所示)有着几点优势：
1)叶片厚度方向的适形度好；
2)对复杂射野的照射效率高；
3)可以有效减少光栅漏射，可以更好地保护危及器官。
动态调强子野分割通过同时控制各对叶片的运动速度和照射剂量率，实现强度图的分割。在动态分割时，若叶片可交叠，各对叶片的分割算法独立计算，互不干扰。相较静态子野分割，动态分割照射效率高，靶区剂量曲线陡峭，有着明显优势。然而，动态分割需要更多的MU才能完成，而且控制系统比较复杂。
正交双层光栅的动态分割理论上会有很高的照射效率，同时具备正交双层光栅和动态分割的优势，但其缺点是做动态分割时，各对叶 片的分割算法不再独立，两个方向的叶片运动相互耦合，同时影响到分割强度图，大大增加了算法难度。它的实现至今未见诸于文献和各大放疗公司的发布中，换言之，这部分的研究至今空白。
因此，一种适用于正交双层光栅装置的动态调强(sliding-window)子野分割方法亟待提出，其可填补此领域的空白，为后续研究提供借鉴。
发明内容
为了解决上述技术问题，本发明提出了一种用于正交双层光栅装置的动态调强(sliding-window)子野分割方法。
为了达到上述目的，本发明的技术方案如下：
用于正交双层光栅装置的动态调强子野分割方法，具体包括以下步骤：
S1：构造虚拟单层光栅，虚拟单层光栅的速度是正交双层光栅的速度合成，设置虚拟单层光栅的叶片厚度和叶片对数；
S2：得到优化强度图，将优化强度图按照步骤S1得到的虚拟单层光栅的安装角度旋转；
S3：对步骤S2得到的旋转后的优化强度图进行动态调强计算，计算结果包括速度拐点的时间点和各个子野形状，通过叠加各个子野形状即可得到初步的计算强度图，与优化强度图对比可以检查虚拟单层光栅形成的子野是否符合预期；
S4：对虚拟单层光栅形成的子野进行反旋转，得到真实的子野；
S5：采用正交双层光栅对步骤S4得到的反旋转后的子野进行适形；
S6：检查正交双层光栅的叶片是否超速，若超速，则将超速的叶片加以限制，其允许叶片回退。
在上述技术方案的基础上，还可做如下改进：
作为优选的方案，在步骤S1中，虚拟单层光栅的速度是正交光栅A和光栅B的速度合成，速度大小和方向为v，θ；
Figure PCTCN2019120600-appb-000001
Figure PCTCN2019120600-appb-000002
其中，v 1为水平方向叶片的最大速度，v 2为竖直方向叶片的最大速度；虚拟单层光栅的叶片厚度取较小值，叶片对数取较大值，其余属性与光栅A或光栅B保持一致。
作为优选的方案，在步骤S2还包括以下内容，当优化强度图按照步骤S1得到的虚拟单层光栅的安装角度旋转后，优化强度图在叶片厚度方向的采样间隔直接按叶片厚度划分，在叶片运动方向则按照自定义的间隔划分，取值0.25。
作为优选的方案，步骤S3具体包括以下步骤：
S3.1：计算通量的最小增量；
为确保叶片在运动过程中不会关闭，已知叶片对之间的最小距离(Gap)，因此需要计算单位长度的最小增量，如式(3)知，叶片以物理最大速度运行单位长度时，单位长度的通量最小，min slope与Gap的乘积即为光栅能达到的最小强度值；
Figure PCTCN2019120600-appb-000003
S3.2：计算通量与距离关系，根据步骤S2旋转后的优化强度图及通量的最小增量确定左右叶片运动过程中强度通量与位置的对应曲线；
S3.3：设定速度拐点，根据通量与距离关系曲线计算出速度的拐点；
S3.4：子野中左右叶片位置的确定，以速度拐点为分割点，计算各分割点处左右叶片的位置；
S3.5：叶片加速度检查，若叶片最大加速度超出限制，则降低光 栅的最大速度，重复步骤S3.1-S3.4；
S3.6：强度图对比，计算当前分割下的强度图，与优化强度图对比，当误差大于阈值，则调整当前分割下的强度图，重复步骤S3.1-S3.5。
作为优选的方案，还包括优化正交双层光栅动态调强子野分割方法，其具体为子野权重优化法；
子野权重优化法是叶片运动轨迹固定条件下，优化各子野的时间点，优化目标为分割强度图与优化强度图差值的二范数，如式(4)所示；
J obj＝||J opt-J cal|| 2       (4)
分割强度图可视为各子野的线性叠加，如式(5)所示；
Figure PCTCN2019120600-appb-000004
其中，I seg为单个子野形成的强度图，u i为子野权重，目标函数为J obj。
作为优选的方案，还包括优化正交双层光栅动态调强子野分割方法，其具体为叶片运动轨迹优化法；
叶片运动轨迹优化法是在子野权重固定的条件下，以各叶片的运动轨迹为变量，在一定条件下，优化目标函数。
作为优选的方案，叶片运动轨迹优化法具体包括以下步骤：
S7：计算当前叶片位置叠加形成的强度图；
水平光栅每对叶片的运动轨迹记作：
Figure PCTCN2019120600-appb-000005
竖直光栅每对叶片的运动轨迹记作：
Figure PCTCN2019120600-appb-000006
其中，
Figure PCTCN2019120600-appb-000007
表示：第i对叶片(对应的Y轴坐标为y i)的左叶片运动轨迹；
Figure PCTCN2019120600-appb-000008
表示：第i对叶片(对应的Y轴坐标为y i)的右叶片运动轨迹；
Figure PCTCN2019120600-appb-000009
表示：第j对叶片(对应的X轴坐标为x j)的下叶 片运动轨迹；
Figure PCTCN2019120600-appb-000010
表示：第j对叶片(对应的X轴坐标为x j)的上叶片运动轨迹；水平叶片共n对，竖直叶片共m对；
叶片运动所受约束有：钨门约束、叶片物理约束以及速度约束，约束条件如式(8)所示：
Figure PCTCN2019120600-appb-000011
则用上述叶片轨迹得到的计算强度图为：
Figure PCTCN2019120600-appb-000012
Figure PCTCN2019120600-appb-000013
其中，i，j为点(x，y)对应的水平光栅和竖直光栅的叶片序号；
其中，i，j为点(x，y)对应的水平光栅和竖直光栅的叶片序号；
分割强度图可视为各子野的线性叠加，如式(5)所示；
Figure PCTCN2019120600-appb-000014
其中，I seg为单个子野形成的强度图，而I seg＝I(x，y)，u i为子野权重；
目标函数为J obj，优化目标函数为分割强度图与优化强度图差值的二范数，如式(4)所示；
J obj＝||J opt-J cal|| 2       (4)
S8：开始外部循环，找到分割强度图与优化强度图差异最大的行和列，差异值的评价标准是二范数；
S9：找到差异最大的行和列对应的光栅序号，光栅与强度图有着固定的对应关系，可依照强度图的编号计算出光栅序号；
S10：开始内部循环，随机选取一个时刻，根据约束条件计算上下左右四个叶片的活动范围，计算方法见式(8)；
S11：循环改变上下左右叶片的位置，根据步骤S10，四个叶片都会得到一个活动范围，循环取四个叶片范围的最大值和最小值，对四个叶片的位置分别做扰动；
S12：若目标函数有下降或者内部循环超限，则进入步骤S13，否则重复步骤S10-S11；
S13：若目标函数小于阈值或者外部循环超限，则停止优化，否则重复步骤S8-S12。
附图说明
图1为正交双层光栅解决多连通区域子野分割的实例图一；
图2为正交双层光栅解决多连通区域子野分割的实例图二；
图3为本发明实施例提供的用于正交双层光栅装置的动态调强子野分割方法的整体流程图；
图4为本发明实施例提供的叶片运动轨迹优化法的流程图；
图5为优化强度图与旋转后强度图对比图之一；
图5(a)为本发明实施例提供的优化强度图之一；
图5(b)为图5(a)对应的旋转后的优化强度图之一；
图6为优化强度图与旋转后强度图对比图之二；
图6(a)为本发明实施例提供的优化强度图之二；
图6(b)为图6(a)对应的旋转后的优化强度图之二；
图7为本发明实施例提供的旋转后的优化强度图与子野叠加图的对比；
图7(a)为本发明实施例提供的旋转后的优化强度图之一；
图7(b)为图7(a)对应的子野叠加图；
图8为本发明实施例提供的正交双层光栅子野适形图之一；
图9为本发明实施例提供的正交双层光栅子野适形图之二；
图10为正交双层光栅叶片超速情况说明；
图11为叶片位置优化迭代曲线；
图12为本发明实施例提供的子野权重优化法的优化强度图、分割强度图和优化后强度图的对比；
图12(a)为本发明实施例提供的优化强度图之三；
图12(b)为图12(a)对应的分割强度图；
图12(c)为图12(a)对应的优化后强度图；
图13是上下两层光栅配合分割出多个射野的示意图；
图14是上下两层光栅配合对射野边沿包络线进行拟合优化的示意图。
具体实施方式
下面结合附图详细说明本发明的优选实施方式。
为了便于理解，如图13和图14所示，上层光栅1和下层光栅2可以相互配合，完成如图13所示例的多个射野3(图中为4个射野3)的同时构建，也可以相互配合，如图14所示的，上下两层光栅配合对射野边沿包络线进行拟合优化。图13和图14主要是为了便于理解进行的图示，其和实际的光栅叶片的数量和尺寸是不同的。
为了达到本发明的目的，本发明提出一种用于正交双层光栅装置的动态调强子野分割方法，该方法是基于一种用于放疗设备的正交双层光栅装置来实现的，该装置安装于放疗设备的加速器机头下，包括：
上层光栅叶片和下层光栅叶片，上层光栅叶片和下层光栅叶片所在平面互相平行，且垂直于加速器机头发出的射线方向，上层光栅叶片和下层光栅叶片的运动方向为正交；
上层光栅叶片包括左边叶片和右边叶片，用于向靶区左右两侧搜索移动；
下层光栅叶片包括上边叶片和下边叶片，用于向靶区上下两侧搜索移动；
控制器，用于驱动左边叶片和右边叶片、上边叶片和下边叶片中的每片子叶片单独运动，以达到和靶区适形的目的。
该正交双层光栅装置相对于传统单层光栅，及双层平行光栅来说，适形度更高，两个方向均能够达到小于1mm的走位精度。
下面就对本发明进行详细的描述，在一种用于正交双层光栅装置的动态调强子野分割方法的其中一些实施例中，其具体包括以下步骤，如图3所示：
S1：构造虚拟单层光栅，虚拟单层光栅的速度是正交双层光栅的速度合成，设置虚拟单层光栅的叶片厚度和叶片对数；
S2：得到优化强度图，为了便于子野分割，将优化强度图按照步骤S1得到的虚拟单层光栅的安装角度旋转；
S3：对步骤S2得到的旋转后的优化强度图进行动态调强计算，计算结果包括速度拐点的时间点和各个子野形状，通过叠加各个子野形状即可得到初步的计算强度图，与优化强度图对比可以检查虚拟单层光栅形成的子野是否符合预期；
S4：对虚拟单层光栅形成的子野进行反旋转，得到真实的子野；
S5：采用正交双层光栅对步骤S4得到的反旋转后的子野进行适形；
S6：检查正交双层光栅的叶片是否超速，若超速，则将超速的叶片加以限制，其允许叶片回退。
其中，步骤S3为虚拟单层光栅的动态调强子野分割步骤，分割完成后步骤S4得到了在各个速度拐点下的子野形状，步骤S4将各子野形状反旋转θ。步骤S5中，正交双层光栅对步骤S4得到的反旋转后的子野适形，这里把虚拟单层光栅形成的子野用正交双层光栅实现，得到了各个速度拐点下，正交双层光栅的运动曲线。
为了进一步地优化本发明的实施效果，在另外一些实施方式中，其余特征技术相同，不同之处在于，在步骤S1中，虚拟单层光栅的 速度是正交光栅A(MLC1)和光栅B(MLC2)的速度合成，速度大小和方向为v，θ；
Figure PCTCN2019120600-appb-000015
Figure PCTCN2019120600-appb-000016
其中，v 1为水平方向叶片的最大速度，v 2为竖直方向叶片的最大速度；虚拟单层光栅的叶片厚度取较小值(这里可以取0.25或其他)，叶片对数取较大值(这里可以取256或其他)，其余属性与光栅A或光栅B保持一致。
进一步，在步骤S2还包括以下内容，当优化强度图按照步骤S1得到的虚拟单层光栅的安装角度θ旋转后，优化强度图在叶片厚度方向的采样间隔直接按叶片厚度划分，在叶片运动方向则按照自定义的间隔划分，取值0.25。
进一步，步骤S3具体包括以下步骤：
S3.1：计算通量的最小增量；
为确保叶片在运动过程中不会关闭，已知叶片对之间的最小距离(Gap)，因此需要计算单位长度的最小增量，如式(3)知，叶片以物理最大速度运行单位长度时，单位长度的通量最小，min slope与Gap的乘积即为光栅能达到的最小强度值；
Figure PCTCN2019120600-appb-000017
S3.2：计算通量与距离关系，根据步骤S2旋转后的优化强度图及通量的最小增量确定左右叶片运动过程中强度通量与位置的对应曲线；
S3.3：设定速度拐点，根据通量与距离关系曲线计算出速度的拐点；
S3.4：子野中左右叶片位置的确定，以速度拐点为分割点，计算各分割点处左右叶片的位置；
S3.5：叶片加速度检查，若叶片最大加速度超出限制，则降低光栅的最大速度，重复步骤S3.1-S3.4；
S3.6：强度图对比，计算当前分割下的强度图，与优化强度图对比，当误差大于阈值，则调整当前分割下的强度图，重复步骤S3.1-S3.5。
为了进一步地优化本发明的实施效果，在另外一些实施方式中，其余特征技术相同，不同之处在于，还包括优化正交双层光栅动态调强子野分割方法，其具体为子野权重优化法；
子野权重优化法是叶片运动轨迹固定条件下，优化各子野的时间点，优化目标为分割强度图与优化强度图差值的二范数，如式(4)所示；
J obj＝||J opt-J cal|| 2     (4)
分割强度图可视为各子野的线性叠加，如式(5)所示；
Figure PCTCN2019120600-appb-000018
其中，I seg为单个子野形成的强度图，u i为子野权重，目标函数为J obj。
因此，子野权重优化法是一个带约束的二次规划问题，按照一般的求解思路求解即可。
为了进一步地优化本发明的实施效果，在另外一些实施方式中，其余特征技术相同，不同之处在于，还包括优化正交双层光栅动态调强子野分割方法，其具体为叶片运动轨迹优化法；
叶片运动轨迹优化法是在子野权重固定的条件下，以各叶片的运动轨迹为变量，在一定条件下，优化目标函数。
将叶片运动离散化，用虚拟单层光栅计算出的MU step(速度拐点的时间点)为时间间隔。叶片离散化后，叶片轨迹可用一系列点表示。初步统计，所有叶片运动轨迹点数约20000个，也就是说，优化变量约为20000个，普通优化方法，如非线性优化、遗传算法、粒子 群算法效果均不显著。为此，这里提供了一种寻优算法。整个寻优过程分为外循环和内循环。外循环寻找分割强度图和优化强度图差异最大的行和列，并计算出对应的叶片序号，这一步目的是锁定待优化的叶片序号，减少待优化变量的个数；若外循环迭代次数超限或目标函数达到要求，则退出整个循环。内循环随机选择叶片时刻，根据约束条件计算出叶片的活动范围，在叶片活动范围的最大值和最小值之间变动，观察目标函数是否变小；若目标函数下降或内循环迭代次数超限，跳出内循环。
进一步，叶片运动轨迹优化法具体包括以下步骤，如图4所示：
S7：计算当前叶片位置叠加形成的强度图；
水平光栅每对叶片的运动轨迹记作：
Figure PCTCN2019120600-appb-000019
竖直光栅每对叶片的运动轨迹记作：
Figure PCTCN2019120600-appb-000020
其中，
Figure PCTCN2019120600-appb-000021
表示：第i对叶片(对应的Y轴坐标为y i)的左叶片运动轨迹；
Figure PCTCN2019120600-appb-000022
表示：第i对叶片(对应的Y轴坐标为y i)的右叶片运动轨迹；
Figure PCTCN2019120600-appb-000023
表示：第j对叶片(对应的X轴坐标为x j)的下叶片运动轨迹；
Figure PCTCN2019120600-appb-000024
表示：第j对叶片(对应的X轴坐标为x j)的上叶片运动轨迹；水平叶片共n对，竖直叶片共m对；
叶片运动所受约束有：钨门约束、叶片物理约束(右叶片位置大于左叶片，上叶片位置大于下叶片)以及速度约束，约束条件如式(8)所示：
Figure PCTCN2019120600-appb-000025
则用上述叶片轨迹得到的计算强度图为：
Figure PCTCN2019120600-appb-000026
Figure PCTCN2019120600-appb-000027
其中，i，j为点(x，y)对应的水平光栅和竖直光栅的叶片序号；
其中，i，j为点(x，y)对应的水平光栅和竖直光栅的叶片序号；
分割强度图可视为各子野的线性叠加，如式(5)所示；
Figure PCTCN2019120600-appb-000028
其中，I seg为单个子野形成的强度图，而I seg＝I(x，y)，u i为子野权重；
目标函数为J obj，优化目标函数为分割强度图与优化强度图差值的二范数，如式(4)所示；
J obj＝||J opt-J cal|| 2     (4)
S8：开始外部循环，找到分割强度图与优化强度图差异最大的行和列，差异值的评价标准是二范数；
S9：找到差异最大的行和列对应的光栅序号，光栅与强度图有着固定的对应关系，可依照强度图的编号计算出光栅序号；
S10：开始内部循环，随机选取一个时刻，根据约束条件计算上下左右四个叶片的活动范围，根据叶片约束(钨门约束、物理约束、速度约束)得到某一时刻叶片允许的活动范围，计算方法见式(8)；
S11：循环改变上下左右叶片的位置，根据步骤S10，四个叶片都会得到一个活动范围，循环取四个叶片范围的最大值和最小值，对四个叶片的位置分别做扰动；
S12：若目标函数有下降或者内部循环超限，则进入步骤S13，否则重复步骤S10-S11；
S13：若目标函数小于阈值或者外部循环超限，则停止优化，否 则重复步骤S8-S12。
综上，本发明提供了一种用于正交双层光栅装置的动态调强子野分割方法，并提供了两种优化此分割的方法：叶片运动轨迹优化法和子野权重优化法。子野权重优化法是叶片运动轨迹固定条件下，优化各子野的时间点，叶片运动轨迹优化法是在子野权重固定的条件下，以各叶片的运动轨迹为变量，在一定约束下，优化目标函数。
在现有的子野分割技术中，无论是单层光栅，还是平行双层光栅，都会面临两个问题：叶片厚度方向的适形度不够；一个复杂的射野，需要多个子野才能够形成，照射效率低。正交双层光栅在适形度和照射效率上均占有优势，此外，双层光栅可有效减少光栅漏射，可以更好地保护危及器官。相较静态子野分割，动态分割照射效率高，靶区剂量曲线陡峭，有着明显优势。种用于正交双层光栅装置的动态调强子野分割方法。
本发明开始先构造虚拟单层光栅，光栅安装方向是正交双层光栅的速度合成方向。为了便于进行虚拟单层光栅的动态调强(sliding-window)分割，对得到的优化强度图旋转光栅安装角。旋转后的图形仍为矩形，同时维数扩张，整个强度图范围变大。本发明自行定义旋转后图形的采样间隔，与优化强度图采样间隔并不完全一致，旋转前后的优化强度图见图5和图6。
接着对旋转后的优化强度图进行动态调强计算，计算结果包括速度拐点的时间点和各个子野形状，通过叠加各个子野形状即可得到初步的计算强度图，与优化强度图对比可以检查虚拟光栅形成的子野是否符合预期。由图7可见，虚拟单层光栅形成的子野是合理的。
虚拟单层光栅形成的子野经过反旋转后，得到的是真实的子野，用现有的正交双层光栅对反旋转后的子野进行适形，这样就得到了正交双层光栅的动态调强(sliding-window)计算结果。正交双层光栅 子野适形的范例见图8和图9，其中的空白区域是反旋转后的子野，网络状即为正交双层光栅，可见正交双层光栅可以正确适形子野。
最后，检查叶片是否超速。上一步已经形成了一系列时间点下的子野形状，这里检查叶片是否超速。值得注意的是，虚拟单层光栅形成动态调强(sliding-window)子野时，已经检查了叶片超速问题。然而，在速度合成的虚拟单层光栅不超速的情况下，正交双层光栅不一定不超速。如图10所示，虚拟叶片运动距离较小，但造成的形状偏差较大，水平光栅在适形时运动距离陡增，造成叶片超速，有时也会造成叶片回退。因此，这里的速度检查有别于单层光栅的动态调强(sliding-window)，允许叶片回退。
为了改善现有的正交双层光栅动态调强(sliding-window)分割，本发明提出了两种优化方法：
a)叶片运动轨迹优化法，以叶片位置为变量，初值即为速度合成算法形成的叶片位置，添加约束条件，以子野叠加强度图之差的二范数为目标进行优化；
b)子野权重优化法，各时刻叶片位置固定，优化各时刻的时间(子野权重)。
对于叶片运动轨迹优化法而言，叶片位置优化开始前需将叶片位置初始化，这里的变量排列规则的伪代码为：
for i＝1：所有时刻
顺序排列水平光栅的所有左叶片
顺序排列水平光栅的所有右叶片
顺序排列竖直光栅的所有下叶片
顺序排列竖直光栅的所有上叶片
end
这样可将所有叶片的所有时刻排成一列，便于维护；
接着，定义约束函数，约束函数包含钨门约束、物理约束和速度约束，其中需要的外部数据有：钨门参数、光栅速度，光栅最小间隔，约束函数按照式(8)所示。同时，根据此约束函数可计算出某时刻某叶片的活动范围。
定义目标函数。这里需要根据已有的叶片位置和时间信息计算出强度图。由离散的叶片位置形成离散的强度图会经由实际位置转化，因此这里需要的外部数据有：叶片厚度、强度图采样间距，MU(时间信息)。
最终根据图4所示的步骤优化叶片位置，得到的迭代次数与目标函数的曲线如图11所示，可见在迭代过程中，目标函数在缓慢下降，优化后分割强度图与优化强度图更加接近。
子野权重优化法是一次完成的，不需要反复迭代。优化后目标函数从2.8254降至1.9457。优化强度图、分割强度图和优化后的强度图的对比见图12，可见优化效果明显。
正交双层光栅在靶区适形度和治疗效率上都有着明显优势，本发明提供了一种用于正交双层光栅装置的动态调强子野分割方法。分割算法的核心是将两层光栅的速度合成后构造虚拟单层光栅，进行单层光栅的动态调强(sliding-window)分割，最后用两层光栅适形各子野。此方法为正交双层光栅的动态调强(sliding-window)分割提供了一个可行方案。
为了进一步减少分割误差，分割算法提供了两种优化方法：叶片运动轨迹优化法和子野权重优化法。叶片运动轨迹优化法是在子野权重固定的条件下，以各叶片的运动轨迹为变量，在一定约束下，优化目标函数。子野权重优化法是叶片运动轨迹固定条件下，优化各子野的时间点。这两种优化方法都可减少分割强度图的误差值，改善优化 效果。
对于本发明的优选实施方式，应当指出，对于本领域的普通技术人员来说，在不脱离本发明创造构思的前提下，还可以做出若干变形和改进，这些都属于本发明的保护范围。

Claims (7)

  1. 用于正交双层光栅装置的动态调强子野分割方法，其特征在于，具体包括以下步骤：
    S1：构造虚拟单层光栅，虚拟单层光栅的速度是正交双层光栅的速度合成，设置虚拟单层光栅的叶片厚度和叶片对数；
    S2：得到优化强度图，将优化强度图按照步骤S1得到的虚拟单层光栅的安装角度旋转；
    S3：对步骤S2得到的旋转后的优化强度图进行动态调强计算，计算结果包括速度拐点的时间点和各个子野形状，通过叠加各个子野形状即可得到初步的计算强度图，与优化强度图对比可以检查虚拟单层光栅形成的子野是否符合预期；
    S4：对虚拟单层光栅形成的子野进行反旋转，得到真实的子野；
    S5：采用正交双层光栅对步骤S4得到的反旋转后的子野进行适形；
    S6：检查正交双层光栅的叶片是否超速，若超速，则将超速的叶片加以限制，其允许叶片回退。
  2. 根据权利要求1所述的用于正交双层光栅装置的动态调强子野分割方法，其特征在于，在所述步骤S1中，虚拟单层光栅的速度是正交光栅A和光栅B的速度合成，速度大小和方向为v，θ；
    Figure PCTCN2019120600-appb-100001
    Figure PCTCN2019120600-appb-100002
    其中，v 1为水平方向叶片的最大速度，v 2为竖直方向叶片的最大速度；虚拟单层光栅的叶片厚度取较小值，叶片对数取较大值，其余属性与光栅A或光栅B保持一致。
  3. 根据权利要求2所述的用于正交双层光栅装置的动态调强子野分割方法，其特征在于，在所述步骤S2还包括以下内容，当优化强度图按照步骤S1得到的虚拟单层光栅的安装角度旋转后，优化强度图在叶片厚度方向的采样间隔直接按叶片厚度划分，在叶片运动方向则按照自定义 的间隔划分，取值0.25。
  4. 根据权利要求3所述的用于正交双层光栅装置的动态调强子野分割方法，其特征在于，所述步骤S3具体包括以下步骤：
    S3.1：计算通量的最小增量；
    为确保叶片在运动过程中不会关闭，已知叶片对之间的最小距离(Gap)，因此需要计算单位长度的最小增量，如式(3)知，叶片以物理最大速度运行单位长度时，单位长度的通量最小，min s/ope与Gap的乘积即为光栅能达到的最小强度值；
    Figure PCTCN2019120600-appb-100003
    S3.2：计算通量与距离关系，根据所述步骤S2旋转后的优化强度图及通量的最小增量确定左右叶片运动过程中强度通量与位置的对应曲线；
    S3.3：设定速度拐点，根据通量与距离关系曲线计算出速度的拐点；
    S3.4：子野中左右叶片位置的确定，以速度拐点为分割点，计算各分割点处左右叶片的位置；
    S3.5：叶片加速度检查，若叶片最大加速度超出限制，则降低光栅的最大速度，重复步骤S3.1-S3.4；
    S3.6：强度图对比，计算当前分割下的强度图，与优化强度图对比，当误差大于阈值，则调整当前分割下的强度图，重复步骤S3.1-S3.5。
  5. 根据权利要求1-4任一项所述的用于正交双层光栅装置的动态调强子野分割方法，其特征在于，还包括优化正交双层光栅动态调强子野分割方法，其具体为子野权重优化法；
    子野权重优化法是叶片运动轨迹固定条件下，优化各子野的时间点，优化目标为分割强度图与优化强度图差值的二范数，如式(4)所示；
    J obj＝||J opt-J cal|| 2  (4)
    分割强度图可视为各子野的线性叠加，如式(5)所示；
    Figure PCTCN2019120600-appb-100004
    其中，I seg为单个子野形成的强度图，u i为子野权重，目标函数为J obj。
  6. 根据权利要求1-4任一项所述的用于正交双层光栅装置的动态调强子野分割方法，其特征在于，还包括优化正交双层光栅动态调强子野分割方法，其具体为叶片运动轨迹优化法；
    所述叶片运动轨迹优化法是在子野权重固定的条件下，以各叶片的运动轨迹为变量，在一定条件下，优化目标函数。
  7. 根据权利要求6所述的用于正交双层光栅装置的动态调强子野分割方法，其特征在于，所述叶片运动轨迹优化法具体包括以下步骤：
    S7：计算当前叶片位置叠加形成的强度图；
    水平光栅每对叶片的运动轨迹记作：
    Figure PCTCN2019120600-appb-100005
    竖直光栅每对叶片的运动轨迹记作：
    Figure PCTCN2019120600-appb-100006
    其中，
    Figure PCTCN2019120600-appb-100007
    表示：第i对叶片(对应的Y轴坐标为y i)的左叶片运动轨迹；
    Figure PCTCN2019120600-appb-100008
    表示：第i对叶片(对应的Y轴坐标为y i)的右叶片运动轨迹；
    Figure PCTCN2019120600-appb-100009
    表示：第j对叶片(对应的X轴坐标为x j)的下叶片运动轨迹；
    Figure PCTCN2019120600-appb-100010
    表示：第j对叶片(对应的X轴坐标为x j)的上叶片运动轨迹；水平叶片共n对，竖直叶片共m对；
    叶片运动所受约束有：钨门约束、叶片物理约束以及速度约束，约束条件如式(8)所示：
    Figure PCTCN2019120600-appb-100011
    则用上述叶片轨迹得到的计算强度图为：
    Figure PCTCN2019120600-appb-100012
    Figure PCTCN2019120600-appb-100013
    其中，i，j为点(x，y)对应的水平光栅和竖直光栅的叶片序号；
    分割强度图可视为各子野的线性叠加，如式(5)所示；
    Figure PCTCN2019120600-appb-100014
    其中，I seg为单个子野形成的强度图，而I seg＝I(x，y)，u i为子野权重；
    目标函数为J obj，优化目标函数为分割强度图与优化强度图差值的二范数，如式(4)所示；
    J obj＝||J opt-J cal| 2  (4)
    S8：开始外部循环，找到分割强度图与优化强度图差异最大的行和列，差异值的评价标准是二范数；
    S9：找到差异最大的行和列对应的光栅序号，光栅与强度图有着固定的对应关系，可依照强度图的编号计算出光栅序号；
    S10：开始内部循环，随机选取一个时刻，根据约束条件计算上下左右四个叶片的活动范围，计算方法见式(8)；
    S11：循环改变上下左右叶片的位置，根据所述步骤S10，四个叶片都会得到一个活动范围，循环取四个叶片范围的最大值和最小值，对四个叶片的位置分别做扰动；
    S12：若目标函数有下降或者内部循环超限，则进入步骤S13，否则 重复所述步骤S10-S11；
    S13：若目标函数小于阈值或者外部循环超限，则停止优化，否则重复所述步骤S8-S12。
PCT/CN2019/120600 2018-12-21 2019-11-25 用于正交双层光栅装置的动态调强子野分割方法 WO2020125332A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/417,085 US20220047893A1 (en) 2018-12-21 2019-11-25 Dynamic intensity-modulated segmentation method for orthogonal double-layer grating device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811579085.8 2018-12-21
CN201811579085.8A CN109499011B (zh) 2018-12-21 2018-12-21 用于正交双层光栅装置的动态调强子野分割方法

Publications (1)

Publication Number Publication Date
WO2020125332A1 true WO2020125332A1 (zh) 2020-06-25

Family

ID=65755125

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/120600 WO2020125332A1 (zh) 2018-12-21 2019-11-25 用于正交双层光栅装置的动态调强子野分割方法

Country Status (3)

Country Link
US (1) US20220047893A1 (zh)
CN (1) CN109499011B (zh)
WO (1) WO2020125332A1 (zh)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109499011B (zh) * 2018-12-21 2020-11-10 苏州雷泰医疗科技有限公司 用于正交双层光栅装置的动态调强子野分割方法
CN110404185A (zh) * 2019-07-09 2019-11-05 苏州雷泰智能科技有限公司 一种正交双层光栅的旋转调强分割方法及装置
CN111681736B (zh) * 2020-06-08 2023-05-12 苏州雷泰医疗科技有限公司 一种基于分象限的正交双层光栅动态调强分割方法
CN113521561A (zh) * 2021-06-22 2021-10-22 苏州雷泰医疗科技有限公司 正交双层光栅逆向旋转调强的实现方法、存储介质及装置
CN115317811A (zh) * 2022-08-17 2022-11-11 中科超精(南京)科技有限公司 一种基于动态半影模型的动态调强装置及优化方法

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101246757A (zh) * 2007-08-23 2008-08-20 中国医学科学院肿瘤医院 动态独立准直器准直块运动轨迹的生成方法
CN102915784A (zh) * 2012-08-24 2013-02-06 山东新华医疗器械股份有限公司 双层多叶准直器
CN103079643A (zh) * 2010-08-23 2013-05-01 瓦里安医疗系统公司 多级多叶准直器
KR20150008706A (ko) * 2013-07-15 2015-01-23 가톨릭대학교 산학협력단 방사선 치료효과의 극대화를 위한 다층형 픽셀제한 방사선 치료용 콜리메이터
CN107349531A (zh) * 2017-07-04 2017-11-17 上海联影医疗科技有限公司 一种放疗设备、存储介质、子野形状生成方法及装置
CN108066902A (zh) * 2016-11-14 2018-05-25 上海东软医疗科技有限公司 一种子野信息确定方法和装置
CN108211136A (zh) * 2017-12-29 2018-06-29 苏州雷泰医疗科技有限公司 基于动态光栅的旋转调强优化方法及放射治疗设备
CN109499011A (zh) * 2018-12-21 2019-03-22 苏州雷泰医疗科技有限公司 用于正交双层光栅装置的动态调强子野分割方法

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2009072443A (ja) * 2007-09-21 2009-04-09 Toshiba Corp マルチリーフコリメータおよび放射線治療装置
WO2012089535A1 (en) * 2010-12-29 2012-07-05 Deutsches Krebsforschungszentrum Multi-leaf collimator with leaf drive
JP5630666B2 (ja) * 2012-03-30 2014-11-26 住友重機械工業株式会社 中性子捕捉療法用コリメータ及び中性子捕捉療法装置
CN106573152B (zh) * 2014-08-15 2021-01-26 皇家飞利浦有限公司 用于自适应放射治疗规划的受监管的4d剂量图变形
CN104759032B (zh) * 2015-04-08 2017-08-25 苏州雷泰医疗科技有限公司 一种用于放射治疗设备的光栅装置及其控制方法及放射治疗设备
US10987523B2 (en) * 2015-10-23 2021-04-27 The Regents Of The University Of California Platform for intensity modulated radiation therapy
CN108175953B (zh) * 2017-12-29 2020-07-21 苏州雷泰医疗科技有限公司 一种基于动态光栅的旋转调强优化方法及放射治疗设备

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101246757A (zh) * 2007-08-23 2008-08-20 中国医学科学院肿瘤医院 动态独立准直器准直块运动轨迹的生成方法
CN103079643A (zh) * 2010-08-23 2013-05-01 瓦里安医疗系统公司 多级多叶准直器
CN102915784A (zh) * 2012-08-24 2013-02-06 山东新华医疗器械股份有限公司 双层多叶准直器
KR20150008706A (ko) * 2013-07-15 2015-01-23 가톨릭대학교 산학협력단 방사선 치료효과의 극대화를 위한 다층형 픽셀제한 방사선 치료용 콜리메이터
CN108066902A (zh) * 2016-11-14 2018-05-25 上海东软医疗科技有限公司 一种子野信息确定方法和装置
CN107349531A (zh) * 2017-07-04 2017-11-17 上海联影医疗科技有限公司 一种放疗设备、存储介质、子野形状生成方法及装置
CN108211136A (zh) * 2017-12-29 2018-06-29 苏州雷泰医疗科技有限公司 基于动态光栅的旋转调强优化方法及放射治疗设备
CN109499011A (zh) * 2018-12-21 2019-03-22 苏州雷泰医疗科技有限公司 用于正交双层光栅装置的动态调强子野分割方法

Also Published As

Publication number Publication date
CN109499011A (zh) 2019-03-22
CN109499011B (zh) 2020-11-10
US20220047893A1 (en) 2022-02-17

Similar Documents

Publication Publication Date Title
WO2020125332A1 (zh) 用于正交双层光栅装置的动态调强子野分割方法
WO2021003957A1 (zh) 一种正交双层光栅的旋转调强分割方法及装置
CN109310880B (zh) 辐射疗法中的使用切片的轨迹优化
CN109621231B (zh) 多组件同步运动的轨迹优化方法、系统、装置及存储介质
JP6421194B2 (ja) 放射線治療計画装置、放射線治療計画方法および放射線治療システム
CN105664376B (zh) 一种剂量分布确定和放射治疗计划优化的装置及设备
CN104318122A (zh) 一种基于扰动的旋转调强优化方法
US11833370B2 (en) Methods and systems for irradiation based on a fluence map
WO2022052324A1 (zh) 一种基于正交双层光栅旋转扫掠的动态调强方法及装置
CN108771795B (zh) 确定放射治疗的照射分布的装置和放射治疗系统
US11147985B2 (en) Warm start initialization for external beam radiotherapy plan optimization
JP6823299B2 (ja) 放射線治療計画装置および放射線治療システム
CN110170109B (zh) 用于放疗设备的正交双层光栅装置及其子野分割控制方法
US20110293071A1 (en) Method for calculating head scatter phase space for radiation treatment using a multi-leaf collimator with dynamic jaws
CN108109678B (zh) 确定弧形放射治疗的照射分布的方法、装置和计算机可读存储介质
Saw et al. Leaf sequencing techniques for MLC-based IMRT
WO2021248837A1 (zh) 一种基于分象限的正交双层光栅动态调强分割方法
CN103198229A (zh) 一种铁路既有曲线的拨距计算方法
CN109432611B (zh) 扩充型动态楔形板控制点的生成方法、设备和存储介质
CN113521562B (zh) 一种基于正交双层光栅区域缩小的动态调强方法及装置
CN113521561A (zh) 正交双层光栅逆向旋转调强的实现方法、存储介质及装置
US9014338B2 (en) Method for determining a 4D plan for carrying out intensity-modulated radiation therapy
Chen et al. Optimal terrain construction problems and applications in intensity-modulated radiation therapy
US8368042B2 (en) Physical wedge positioning
CN113749685B (zh) 准直器的叶片优化方法、放射治疗计划转换方法和系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19898895

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19898895

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 19898895

Country of ref document: EP

Kind code of ref document: A1