CN1313969C - Method and system for optimizing radiotherapeutic radiation field orientation and strength distribution - Google Patents

Abstract

The present invention relates to a method and a system for simultaneously determining the radiation field directions and the intensity distribution of radiation therapy. The method comprises a modeling procedure, a model transformation procedure and an optimizing execution procedure, wherein the modeling procedure is used for denoting whether each radiation field direction is used or not by a binary variable and denoting the weight of each pencil beam by a floating-point variable to establish a model; the model transformation procedure is used for converting the binary variable to an equation for denoting weight variation from an equation for denoting a target function and a dose constraint condition; the optimizing execution procedure is used for optimizing the converted model to find out the optimal radiation field direction and an intensity distribution parameter. The system of the present invention comprises a data input unit, a modeling unit, a model conversion unit, an optimizing execution unit and a data output unit. The present invention not only accurately establishes the mathematical model of an optimizing problem, and meets the practical requirement of clinic; the present invention has the advantage of high solving speed.

Description

Optimize the system of radiotherapy launched field direction and intensity distributions

Technical field

The present invention relates to a kind of optimization method and system that determines radiocurable launched field direction and launched field intensity distributions or launched field weight simultaneously.

Background technology

Radiocurable purpose is to jeopardize organ in protection, promptly under the prerequisite of the normal structure that may sustain damage, gives the target area, and promptly tumor tissues is with lethal dose.In order to achieve this end, need in the treatment plan process, select suitable launched field parameter.To classical conformal therapy, the launched field parameter comprises launched field direction, launched field shape, launched field weight, clapboard; To more advanced strength regulated shape-adapted radiotherapy, the launched field parameter comprises the intensity distributions of launched field direction and launched field.For classical conformal therapy, artificial selection launched field parameter needs the time, also requires the planned personnel to have rich experience and knowwhy; For strength regulated shape-adapted radiotherapy, artificial selection launched field intensity distributions becomes an impossible thing.Therefore, determine that automatically the launched field parameter is the target that radiotherapy circle is pursued always.

At present existing multiple method based on Optimum Theory can be used for determining launched field weight, the clapboard angle and direction of classical conformal therapy, or is used for determining the launched field intensity distributions of strength regulated shape-adapted radiotherapy.These method computing velocitys are fast, have been used for radiocurable clinical practice, improved radiocurable efficient and quality.But because the complicacy of implications has only a few method can determine launched field direction and launched field intensity distributions simultaneously so far, these methods or computing time are oversize, or have done too much simplification to be similar to and do not have clinical value.Just because of the imperfection of method, corresponding optimization system can not be set up.

Summary of the invention

Therefore, in order to overcome the deficiencies in the prior art part, the object of the present invention is to provide a kind of new a kind of method or system that optimizes radiotherapy launched field direction and intensity distributions, so that this optimization system can be handled above-mentioned radiotherapy parameter optimization problem automatically, the result of its optimization can satisfy the requirement of clinical treatment fully, and make that the working time of computing machine itself will be those skilled in the art (the general requirement be no more than a few hours) in the clinical time span that allows, so method of the present invention has important industrial application value, and overcome still unsolved technical barrier.

In other words, the purpose of this invention is to provide a kind of launched field direction and the intensity distributions that can determine strength regulated shape-adapted radiotherapy simultaneously, or the radiotherapy parameter optimization method or the system of the launched field direction of classical conformal therapy, launched field weight and clapboard.It has not only set up the mathematical model of optimization problem exactly, and it is fast to find the solution speed, satisfies the needs of clinical practice.

For achieving the above object, the invention provides a kind of method of optimizing radiotherapy launched field direction and intensity distributions, may further comprise the steps: modeling procedure, the use that is used for representing each launched field direction with binary variable whether, represent the weight of each pencil beam with floating-point variable, meet the model of clinical prescription dosage requirement with foundation; Model conversion step is used for the model of being set up is carried out conversion, and described binary variable is transformed to from the equation of expression objective function and dose constraint condition in the equation of representing the weight variation range; Optimize execution in step, be used for finding out the launched field direction and the intensity distributions parameter of the optimum that meets the requirement of clinical prescription dosage adopting the model after the described conversion to be optimized.

The present invention also provides a kind of system that optimizes radiotherapy launched field direction and intensity distributions, comprises data input, modeling, model transferring, optimization execution and five unit of data output.Data input cell is used for from existing treatment planning systems input target area and jeopardizes the anatomical data of organ and the dose data of candidate's launched field.Whether the use that modeling unit is used for representing each launched field direction with binary variable represents the weight of each pencil beam with floating-point variable, meets the model of clinical prescription dosage requirement with foundation.The model transferring unit is used for the model of being set up is carried out conversion, described binary variable is transformed to from the equation of expression objective function and dose constraint condition in the equation of expression weight variation range.Optimize performance element and be used for the model after the conversion is optimized, find out the launched field direction and the intensity distributions parameter of the optimum that meets the requirement of clinical prescription dosage.The data output unit is used for exporting optimizes the launched field direction that obtains and intensity distribution data to existing treatment planning systems.

Description of drawings

Fig. 1 has illustrated the data exchange mechanism of the cellular construction of optimization system of the present invention and it and existing treatment planning systems.

Embodiment

1. set up mathematical model

After the explication of patient treatment target area, some treatment planning systems can judge automatically, search and obtain all and can not cause treating the launched field direction that frame and patient or therapeutic bed bump, and other planning system also can find all not cause the launched field direction of collision under planned personnel's operation.When the target area is positioned at the metastomium of health, such launched field direction has dozens of; When the target area is positioned at the incidence, particularly encephalic of health, such launched field direction has hundreds of.If adopt this dozens of even hundreds of launched field treatments, each treatment time will reach a few hours even longer.Therefore, be necessary that extracting a few optimum (generally being no more than 9) launched field from this out tens of even hundreds of launched fields is used for the treatment of.

In the method for the invention, suppose to represent the number that all can not cause the launched field of collision, use binary variable b with NB _iRepresent i launched field direction (i=1,2,3 ... NB), b _iValue is that 0 this launched field direction of expression is not used, and value is 1 this launched field direction of expression use; Use NP _iThe pencil beam number of representing i launched field direction; I launched field floating-point variable W _{I, j}Represent i launched field direction j pencil beam (j=1,2,3 ... NP _i) weight, target area or jeopardize the dosage D of a dose constraint point in the organ then _kCan be expressed as

$D_k=\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{b}_i*W_{i,j}*d_{i,j,k}---\left(1\right)$

Wherein k represents the numbering of dose constraint point; d _{I, j, k}The expression weight is the dosage contribution of j pencil beam of 1 o'clock i launched field to k dose constraint point.Need to prove that because classical conformal therapy can be considered a special case of strength regulated shape-adapted radiotherapy, promptly intensity distributions is uniformly, or along a direction linear change, will only do introduction at strength regulated shape-adapted radiotherapy below.

Suppose to require the launched field number to be no more than n, then optimization problem should satisfy following constraint condition

$\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}{b}_i\≤n---\left(2\right)$

The weight of single pencil beam should greater than, equal zero, and less than an enough big number, promptly optimization problem need satisfy following constraint condition

0.0≤w _{I, j}≤ WL arrives NB to i=1; J=1 is to NP _i

(3)

Wherein WL is the upper limit of weight variation range.

From the clinical dosimetry angle, the requirement that optimization problem also needs to satisfy the target dose uniformity coefficient and jeopardizes the organ tolerance dose.The requirement of target dose uniformity coefficient can be expressed as

L* D _T≤ D _{T, k}≤ U* D _TTo k=1 ..., N _T,

Promptly $L*{\stackrel{\‾}{D}}_T\≤\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{b}_i*W_{i,j}*d_{T,i,j,k}\≤U*{\stackrel{\‾}{D}}_{T,i,j,k}$ To k=1 ..., N _T, (4)

Wherein U and L represent that target dose allows the percentage of variation range bound.The requirement that jeopardizes the organ tolerance dose can be expressed as

D _{O, k}=≤T _OTo k=1 ..., N ₀,

Promptly $\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{b}_i*W_{i,j}*d_{{O,}_{i,j,k}}\≤T_O$ To k=1 ..., N ₀, (5)

D wherein _{O, k}It is the radioactive dose that jeopardizes an obligatory point in the organ; T _OIt is a tolerance dose that jeopardizes organ; N ₀It is a number that jeopardizes obligatory point in the organ.

As previously mentioned, the purpose of radiotherapy is to jeopardize under the tolerant prerequisite of organ, gives the target area with lethal dose.Therefore, the objective function of optimization may be defined as maximization target area mean dose,

Maximization ${\stackrel{\‾}{D}}_T=\underset{k=1}{\overset{N_T}{\mathrm{\Σ}}}{D}_{T,k}/N_T=\underset{k=1}{\overset{N_T}{\mathrm{\Σ}}}\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{b}_i*W_{i,j}*d_{T,i,j,k}/N_T---\left(6\right)$ N wherein _TIt is target dose obligatory point number; D _T, k is the dosage of k dose constraint point in the target area.

2. solving-optimizing problem

From the angle of mathematical programming, owing in the equation (4) of the equation (6) of expression objective function and expression dose constraint condition and (5), binary variable b is arranged all _iWith floating-point variable W _{I, j}Product term, the optimization problem of equation (2) to (6) definition is a mixed integer nonlinear programming problem (Mixed Integer Nonlinear Programming, abbreviation MINLP).During the method for solving of this class problem still is in and tries to explore, there is not proven technique to use.But the practical problems of above-mentioned radiotherapy parameter optimization is variable to be changed to:

Maximization ${\stackrel{\‾}{D}}_T=\underset{k=1}{\overset{N_T}{\mathrm{\Σ}}}{D}_{T,k}/N_T=\underset{k=1}{\overset{N_T}{\mathrm{\Σ}}}\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{W}_{i,j}*d_{T,i,j,k}/N_T---\left(7\right)$

Satisfy constraint

$L*{\stackrel{\‾}{D}}_T\≤\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{W}_{i,j}*d_{T,i,j,k}\≤U*{\stackrel{\‾}{D}}_{T,i,j,k}---\left(8\right)$

$\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{W}_{i,j}*d_{{O,}_{i,j,k}}\≤T_O---\left(9\right)$

0.0w _i，j≤b _i*WL (10)

$\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}{b}_i\≤n---\left(11\right)$

(7-11) finds out by formula, through binary variable b after the conversion _iDisappearance from the equation (8) of the equation (7) of expression objective function and expression dose constraint condition and (9), and only appear at the expression weights W _{I, j}In the linear equation of variation range (10), so optimization problem becomes a MILP (Mixed Integer Linear Programming) problem (Mixed Integer Linear Programming, abbreviation MILP).The mathematical programming technology and the high efficiency program (as freeware GLPK and business software CPLEX) of finding the solution existing maturation of this class problem can be used.For the radiotherapy planning optimization problem, the scale of problem generally is tens to two, 300 of binary variables, and floating-point variable is hundreds of to thousands of.With the operation of reality, the problem of this scale generally can be finished in clinical acceptable time in our experience.

As mentioned above, modeling procedure of the present invention (unit) is set up the model of above-mentioned equation (2)-(6); Model conversion step (unit) then goes out equation (7)-(11) according to above-mentioned equation (2)-(6) conversion; Optimize execution in step (unit) and adopt existing ripe mathematical programming technology and high efficiency program, find out the launched field direction and the intensity distributions parameter of the optimum that meets the requirement of clinical prescription dosage.

According to the difference that clinical prescription dosage requires, objective function and the constraint condition of optimizing launched field direction and intensity distributions can have multiple expression mode.Equation (6) and (7) are a kind of expression mode of objective function just, and equation (4), (5), (8) and (9) also are a kind of expression mode of constraint condition.The present invention is applicable to other expression mode of objective function and constraint condition fully.

The present invention has set up the mixed-integer programming model of optimizing launched field direction and intensity distributions.This model not only accurate description the treatment plan optimization problem, and can adopt ready-made freeware or business software rapid solving.Therefore having important use is worth.

The present invention has two kinds of embodiments.A kind of is to adopt the present invention to set up a plan optimization module in a radiotherapy planning system.Other module of this module and planning system (as dose calculation module) organically combines, the design of hitting the target.Another kind of embodiment is to adopt the present invention to set up an independently optimization system.As shown in Figure 1, before optimizing process began, this system obtained the target area and jeopardizes the anatomical data of organ and the dose data of candidate's launched field from an existing planning system by certain agreement (as DICOM RT); When optimizing process finished, this system sent launched field direction and intensity distributions (weight) that optimization obtains back to existing planning system.

The applicant adopts second kind of embodiment to set up optimization system involved in the present invention.There are data input, modeling, model transferring, optimization execution and five unit of data output in this system, and wherein data input cell obtains the target area and jeopardizes the anatomical data of organ and the dose data of candidate's launched field from PLUNC treatment planning systems (by a system of North Carolina, USA university exploitation) by the mode of read data file; The optimization performance element calls the GLPK optimizer and comes the solving-optimizing problem; The data output unit is optimized launched field direction and the intensity distribution data that obtains by the mode of write data file to the output of PLUNC treatment planning systems.

The applicant adopts a plurality of clinical cases that the actual usefulness of above-mentioned optimization system has been carried out strict test.The result shows, compares with the treatment plan that existing method is generated, and the treatment plan that native system generates can give the target area higher exposure dose guaranteeing to jeopardize under the prerequisite that the organ radioactive dose do not increase even reduce.Introduce the detailed results of two cases below:

Case one is applied to a patient who suffers from tumor of prostate with native system, compares with the treatment plan that method generated of existing clinical use, and the homogeneity of target dose is significantly increased, and the mean dose of target area is brought up to 68.5Gy by 67.5Gy; Simultaneously, the maximum dose that the fl head is accepted drops to 26.2Gy by 33.1Gy, and the maximum dose that right femoral head is accepted is constant substantially.

Case two is applied to a patient who suffers from brain tumor with native system, compares with the treatment plan that method generated of existing clinical use, and the homogeneity of target dose is constant substantially, and the mean dose of target area is brought up to 61.9Gy by 50.2Gy; Simultaneously, the mean dose that temporal lobe is accepted drops to 18.5Gy by 20.0Gy, and the maximum dose that spinal cord is accepted is constant.

In sum, the present invention uses a binary variable to represent a candidate's launched field direction, uses a floating-point variable to represent the weight of a pencil beam or launched field, thereby has set up a mixed integer nonlinear programming model of optimizing direction and weight.By being the MILP (Mixed Integer Linear Programming) model, make optimization problem can adopt ripe mathematical programming technology rapid solving with this model conversion.Adopt optimization method of the present invention and system, can realize actual clinical practice, improved efficient and quality that the radiotherapy parameter is determined greatly.

Claims (1)

1. system that optimizes radiotherapy launched field direction and intensity distributions comprises:

A data input block is used for importing the target area and jeopardizing the anatomical data of organ and the dose data of candidate's launched field from existing treatment planning systems;

Whether a modeling unit, the use that is used for representing each launched field direction with binary variable represent the weight of each pencil beam with floating-point variable, set up the following model that meets the requirement of clinical prescription dosage:

Represent the number that all can not cause the launched field of collision with NB, use binary variable b _iRepresent i launched field direction (i=1,2,3 ... NB), 0 this launched field direction of expression is not used, and this launched field direction is used in 1 expression; Use NP _iThe pencil beam number of representing i launched field direction; I launched field floating-point variable W _{I, j}Represent i launched field j pencil beam (j=1,2,3 ... NP _i) weight, target area or jeopardize the dosage D of a dose constraint point in the organ then _kBe expressed as

$D_k=\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{b}_i*W_{\mathrm{ij}}*d_{i,j,k}---\left(1\right)$

Wherein k represents the numbering of dose constraint point; d _{I, j, k}The expression weight is the dosage contribution of j pencil beam of 1 o'clock i launched field to k dose constraint point,

From implementing efficient, the launched field number is too much unsuitable, that is to say that optimization problem should satisfy following constraint condition

$\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}{b}_i\≤n---\left(2\right)$

Wherein n represents maximum launched field numbers of clinical permission,

The weight of single pencil beam should greater than, equal zero, and less than an enough big number, promptly optimization problem need satisfy following constraint condition

0.0≤w _{I, j}≤ WL arrives NB to i=1; J=1 is to NP _i(3)

Wherein WL is the upper limit of weight variation range,

The requirement of target dose uniformity coefficient:

L* D _T≤ D _{T, k}≤ U* D _TTo k=1 ..., N _T,

Promptly $L*{\stackrel{\‾}{D}}_T\≤\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{b}_i*W_{i,j}*d_{T,i,j,k}\≤U*{\stackrel{\‾}{D}}_{T,i,j,k}$ To k=1 ..., N _T, (4)

Wherein U and L represent that target dose allows the percentage of variation range bound,

Jeopardize organ tolerance dose's requirement:

D _{O, k}=≤T _OTo k=1 ..., N ₀,

Promptly $\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{b}_i*W_{i,j}*d_{O,i,j,k}\≤T_O$ To k=1 ..., N ₀, (5)

D wherein _{O, k}It is the radioactive dose that jeopardizes an obligatory point in the organ; T _OIt is a tolerance dose that jeopardizes organ; N ₀Be a number that jeopardizes obligatory point in the organ,

This system is maximization target area mean dose with the object definition of optimizing,

Maximization ${\stackrel{\‾}{D}}_T=\underset{k=1}{\overset{N_T}{\mathrm{\Σ}}}{D}_{T,k}/N_T=\underset{k=1}{\overset{N_T}{\mathrm{\Σ}}}\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{b}_i*W_{i,j}*d_{T,i,j,k}/N_T---\left(6\right)$

N wherein _TIt is target dose obligatory point number; D _{T, k}Be the dosage of k dose constraint point in the target area,

Formula (1) to (6) is complete has described the physical model that radiotherapy launched field direction and intensity distributions are determined in the optimization that modeling unit realized;

A model transferring unit, be used for the model of being set up is carried out conversion, described binary variable is transformed to from the equation of expression objective function and dose constraint condition in the equation of expression weight variation range, the physical model that is about to formula (1) to (6) description is transformed to the model that following formula (7) to (11) is described:

Maximization ${\stackrel{\‾}{D}}_T=\underset{k=1}{\overset{N_T}{\mathrm{\Σ}}}{D}_{T,k}/N_T=\underset{k=1}{\overset{N_T}{\mathrm{\Σ}}}\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{W}_{i,j}*d_{T,i,j,k}/N_T---\left(7\right)$

Satisfy constraint

$L*{\stackrel{\‾}{D}}_T\≤\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{W}_{i,j}*d_{T,i,j,k}\≤U*{\stackrel{\‾}{D}}_{T,i,j,k}---\left(8\right)$

$\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}\underset{j=1}{\overset{{\mathrm{NP}}_i}{\mathrm{\Σ}}}{W}_{i,j}*d_{O,i,j,k}\≤T_O---\left(9\right)$

0.0≤w _i，j≤b _i*WL (10)

$\underset{i=1}{\overset{\mathrm{NB}}{\mathrm{\Σ}}}{b}_i\≤n---\left(11\right)$

Optimize performance element for one, be used to adopt the model after the described conversion to be optimized, find out the launched field direction and the intensity distributions parameter of the optimum that meets the requirement of clinical prescription dosage;

A data output unit is used for exporting and optimizes the launched field direction that obtains and intensity distribution data to existing treatment planning systems.

Images