Intensity modulated radiotherapy plan optimization method based on guidance of individual prognosis prediction model

Abstract

The invention discloses an intensity modulated radiotherapy plan optimization method based on guidance of an individual prognosis prediction model. The method comprises the following steps: (1) establishing a prognosis model of radiotherapy of a patient; (2) taking the best patient prognosis as optimization main guidance, and combining the prognosis model established in the step (1) and a dose statistical index to construct a target functional expression; and (3) setting parameters for the objective function expression in the step (2) and carrying out numerical optimization solution to obtain an intensity modulated radiation therapy optimization plan. A reverse optimization objective function of radiotherapy plan design is formed by weighting a physical statistical item of dose distribution and an individual prognosis calculation item capable of causing a patient, and radiotherapy plan dose distribution approaching an optimal objective is automatically searched through a corresponding method for solving an optimization problem. And the radiation field intensity distribution of the therapeutic machine corresponding to the dose distribution is calculated, so that the radiotherapy plan design target is sublimated from the optimal physical dose statistical distribution to the optimal final treatment result.

Description

Technical Field

[0001] The present invention relates to the technical field of radiotherapy plan optimization methods, and in particular to an intensity modulated radiotherapy plan optimization method guided by an individual prognosis prediction model. Background Art

[0002] In the prior art, the optimization of intensity modulated radiotherapy plans mainly relies on pre-defined physical dose statistics and gradient targets to achieve the "shaping" of dose distribution, so as to ensure that the target area (tumor area) receives a high dose while the normal tissue receives the lowest possible irradiation dose. Radiotherapy plans for patients with the same type of tumor often have the same physical dose statistical target requirements, which only reflect the average level, while individual differences between patients are reflected due to differences in geometric anatomical structure and radiotherapy sensitivity. The same level of dose distribution has different prognoses for different patients, that is, the radiotherapy plan that meets the existing statistical physical dose index value may not be the most effective at the individual level. By combining the basic clinical characteristics of individual patients to construct a mathematical model of the response between the patient's radiotherapy dose and radiotherapy efficacy, and incorporating it into the inverse optimization objective function of the intensity modulated radiotherapy plan with the current physical dose statistical optimization, the radiotherapy plan is guided to automatically converge to the optimal direction of individual efficacy on the basis of meeting the dose statistical requirements, laying the foundation for the clinical development of personalized radiotherapy technology with higher efficacy and better quality of life for tumor patients.

[0003] Incorporating dose-individual efficacy mathematical models into the design of automatic guided plans for radiotherapy plan optimization has been a major research hotspot in the field of radiotherapy physics in recent years. In 2019, Mediri et al. proposed introducing an efficacy-guided radiotherapy plan irradiation field angle optimization method for three-dimensional conformal radiotherapy (3DCRT) for left-sided breast cancer. The efficacy is considered as the patient's all-cause mortality risk with the patient's radiotherapy plan dose distribution as the decision variable and individual clinical information as the covariate, and the objective term is to minimize the efficacy term. In 2021, Wilson et al. developed an efficacy-based objective function for 3DCRT, which guides the selection of the number, angle, and flux weight of beams in photon and proton radiotherapy by optimizing the predicted efficacy.

[0004] The above method has preliminarily achieved the introduction of individual efficacy items into the process of radiotherapy plan design optimization. However, the following problems still exist: 1) The optimization decision only considers the efficacy item, completely abandoning the statistical limit requirements of the existing physical dose indication items, which is very likely to result in radiotherapy plans that are contrary to the contemporary radiotherapy clinical evidence-based medicine standards; 2) Only 3DCRT is studied, without considering the mainstream clinical intensity-modulated radiation therapy (IMRT) technology, and its applicability needs to be discussed. Summary of the invention

[0005] In view of the above technical problems, the present invention provides an optimization method for intensity modulated radiotherapy plan guided by an individual prognosis prediction model. The inverse optimization objective function of the radiotherapy plan design proposed by the present invention is weighted by the physical statistical term of the dose distribution and the individual therapeutic effect calculation term of the patient. The dose distribution of the radiotherapy plan that is close to the above optimal target is automatically found through the corresponding method for solving the optimization problem, and the intensity distribution of the treatment machine field corresponding to the dose distribution is calculated, so as to achieve the sublimation of the radiotherapy plan design goal from the optimal physical dose statistical distribution to the optimal direction of the final treatment result.

[0006] To achieve the above object, the technical solution adopted by the present invention is:

[0007] The present invention provides an intensity modulated radiotherapy plan optimization method based on individual prognosis prediction model guidance, comprising the following steps:

[0008] (1) Establish a prognostic model for patients undergoing radiotherapy;

[0009] (2) Taking the best patient prognosis as the main optimization guide, the prognostic model established in step (1) and the dose statistical indicators are combined to construct the objective function;

[0010] (3) Setting parameters for the objective function in step (2) and performing numerical optimization to obtain an IMRT optimization plan.

[0011] In the technical solution of the present invention, the optimal patient prognosis is to maximize the patient's predicted efficacy and minimize the toxicity of radiotherapy, that is, minimize toxicity and maximize efficacy.

[0012] In the technical solution of the present invention, the dose statistical indicators include the reference dose, dose and dose-volume constraints of the region of interest.

[0013] Furthermore, in step (1), the method for establishing the prognosis model comprises the following steps:

[0014] a. Collect effective IMRT planning data, baseline clinical information, and clinical prognostic events to build a database;

[0015] b. constructing an association model with the baseline clinical information and the dose distribution information of the region of interest corresponding to the intensity modulated radiotherapy plan data as input and the probability of occurrence of clinical prognostic events corresponding to the region of interest as output; and obtaining a mathematical analytical expression of the prognostic model by substituting the actual collected data into the model parameters;

[0016] Wherein, the region of interest is a tumor and / or an organ at risk.

[0017] Furthermore, the effective intensity modulated radiotherapy plan data is effective intensity modulated radiotherapy plan data for lung cancer patients, and the prognostic events include local control failure of lung cancer after radiotherapy, radiation cardiac events ≥3, radiation pneumonitis events ≥3, and radiation esophagitis events ≥2;

[0018] The clinical baseline information of the local control failure event of lung cancer after radiotherapy includes the patient's age, gender, stage and whether combined chemotherapy is used;

[0019] The clinical baseline information of the grade ≥3 radiation-induced cardiac events included whether there was a history of previous heart disease;

[0020] The clinical baseline information of the grade ≥3 radiation pneumonitis events included age, smoking history, and chemotherapy;

[0021] The clinical baseline information of the grade ≥2 radiation esophagitis events included whether combination chemotherapy was used;

[0022] The association models are Logistic model and Lyman-Kutcher-Burman (LKB) model;

[0023] The method of substituting the actual collected data into the model parameters is the maximum likelihood estimation method (MLE).

[0024] Furthermore, in step (2), the mathematical expression of the objective function is as shown in Formula 1:

[0025]

[0026] St

[0027] (1)x≥0.

[0028] (2) d = Ax.

[0029] (3)C(d)≤0.

[0030] Formula 1

[0031] In formula 1,

[0032] F outcome (d) is the efficacy target item; f physical (d) is the physical target item; f smooth (x) is the smoothing target item;

[0033] x represents the flux intensity distribution of the irradiation field of the intensity modulated radiotherapy plan, which is the solution variable of the plan optimization model; A is the energy deposition matrix; d is the calculated dose distribution; β 1 ~β 3 is a parameter used to measure the weight between each objective item; C(d) is the dose and dose-volume constraint function.

[0034] Furthermore, in step (2), the efficacy target item f outcome The mathematical expression of (d) is shown in formula 2:

[0035]

[0036] In formula 2, d is the calculated dose distribution; f efficacy (d) is the efficacy model, including the tumor control rate model and the survival rate model; f toxicity (d) is the model of organ toxicity caused by radiotherapy dose, including the probability model of tumor control failure and the probability model of radiotherapy-induced normal tissue toxicity; the parameters γ and λ are used to balance the weights between various prognoses.

[0037] Furthermore, in step (2), the efficacy target item f outcome The acquisition process of (d) includes the following steps: taking the individualized optimal prognostic item caused by the patient's radiotherapy plan dose as the main guide, using the prognostic model established in step (1) to calculate the probability of occurrence of each prognostic event caused by the radiotherapy dose distribution, and forming the efficacy target item in the form of an objective function through weighting.

[0038] In the technical solution of the present invention, the efficacy target item f outcome (d) It is used to guide the plan to converge in the direction of maximizing the predicted patient efficacy and minimizing radiotherapy toxicity.

[0039] Furthermore, in step (2), the physical target item f physical (d) is constructed based on the physical dose statistics, and its mathematical expression is shown in Formula 3:

[0040]

[0041] In formula 3, d is the calculated dose distribution; d 0 is the reference dose; T is the transposition operator symbol; N is the total number of voxels corresponding to the region of interest; ω is the weight of different regions of interest.

[0042] In the technical solution of the present invention, the physical target item f physical (d) It is used to ensure that the optimized plan meets the current mainstream clinical evidence-based physical dose statistical objectives, that is, the target dose is as close to the prescribed dose as possible, and the OAR dose is as small as possible.

[0043] Furthermore, in step (2), the smoothing target item f smooth The smoothness of (x) is measured by the second-order difference of the flux; the expression of the smoothing objective term is shown in Equation 4:

[0044]

[0045] In formula 4, x is the flux intensity of the irradiation field of the intensity modulated radiotherapy plan; B is the field set; X is the beamlet set in the direction of the field blade; Y is the beamlet set perpendicular to the blade travel direction; h and k represent the resolution of the flux in the X direction and Y direction respectively.

[0046] In the technical solution of the present invention, the smoothing target item f smooth (x) Ensure that the plan generated by the optimization is machine-implementable.

[0047] In the technical solution of the present invention, the objective function in step (2) is obtained by weighting the above items. In some specific implementations, taking lung cancer patients undergoing intensity modulated radiation therapy as an example, the prognostic objective item f outcome (d) is the weighted sum of the estimated probability of local lung cancer control failure after radiotherapy, the probability of radiation-induced cardiac events ≥3, the probability of radiation-induced pneumonitis ≥3, and the probability of radiation-induced esophagitis ≥2, corresponding to the dose distribution of the radiotherapy plan under a certain alternative solution.

[0048] Furthermore, in step (3), the numerical optimization solution is solved by using the IPOPT interior point method (InteriorPoint OPTimizer) based on the concavity and convexity of the objective function and the clinical efficiency.

[0049] Compared with the existing planning design method, the present invention has the following advantages:

[0050] The intensity modulated radiotherapy plan optimization method proposed in the present invention is guided by an individual prognosis prediction model. The inverse optimization objective function of the radiotherapy plan design is weighted by the physical statistical terms of the dose distribution and the individual prognosis calculation terms that can cause the patient. The corresponding method for solving the optimization problem is used to automatically find the radiotherapy plan dose distribution that is close to the above-mentioned optimal target, and calculate the treatment machine field intensity distribution corresponding to the dose distribution, so as to achieve the sublimation of the radiotherapy plan design goal from the optimal physical dose statistical distribution to the optimal direction of the final treatment result.

[0051] Compared with the existing technology, it has the following advantages:

[0052] (1) The intensity modulated radiotherapy plan optimization method provided by the present invention can not only take into account the existing physical dose statistical level requirements, but also further guide the plan design to develop in the direction of optimal individual therapeutic effect, thereby further ensuring the individualized radiotherapy efficacy of tumor patients.

[0053] (2) The present invention extends the method of optimizing and generating individual patient efficacy-guided plans to the mainstream clinical intensity-modulated radiotherapy technology, thus fully broadening the wide applicability of the method. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present invention and, together with the description, serve to explain the principles of the present invention, in which:

[0055] Figure 1 This is a flow chart of the method for optimizing intensity modulated radiotherapy plan based on guidance of an individual prognosis prediction model in Example 1 of the present invention.

[0056] Figure 2 The comparison diagram is a dose volume histogram (DVH) and dose distribution of the clinical plan of 3 patients with non-small cell lung cancer and the efficacy plan constructed in Example 1 of the present invention. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solution, design method and advantages of the present invention clearer, the present invention is further described in detail by specific embodiments in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0058] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0059] Embodiment 1:

[0060] like Figure 1 As shown, this embodiment provides an intensity modulated radiotherapy plan optimization method guided by an individual prognosis prediction model, comprising the following steps:

[0061] Step S1: Establish a prognostic model for predicting the efficacy of patients after radiotherapy (hereinafter referred to as the prognostic model)

[0062] Specifically, this step includes the following sub-steps:

[0063] a. Collect the intensity-modulated radiotherapy plans, baseline clinical information, and prognostic follow-up data of multiple patients with the same disease who have been treated, and build the database required for patient efficacy modeling.

[0064] b. Using the patient's baseline clinical information and the dose distribution information of the organs corresponding to the radiotherapy plan as input, and the probability of occurrence of clinical prognostic events corresponding to tumors and organs at risk as output, an association model is constructed, and the model parameters are fitted by substituting the corresponding quantities of the actual collected data to obtain a mathematical analytical model for prognosis.

[0065] Taking the intensity modulated radiotherapy for lung cancer patients as an example, the related prognostic events include local control failure of lung cancer (LRF) after radiotherapy, ≥3 grade radiation cardiac events (CE3+), ≥3 grade radiation pneumonitis events (RILT3+), and ≥2 grade radiation esophagitis events (ARIE2+). The clinical baseline information of local control failure of lung cancer includes age, gender, stage, and whether combined chemotherapy is used; the clinical baseline information of ≥3 grade radiation cardiac events includes whether there is a previous history of heart disease; the clinical baseline information of ≥3 grade radiation pneumonitis includes age, smoking history, and whether chemotherapy is used; the clinical baseline information of ≥2 grade radiation esophagitis includes whether combined chemotherapy is used. The association model uses the Logistic model, the Lyman-Kutcher-Burman (LKB) model, and the COX proportional hazard model (hereinafter referred to as the COX model). Furthermore, the maximum likelihood estimate (MLE) method is used to fit the model parameters to obtain the efficacy prognostic model. The detailed information of the prognostic model is shown in Table 1. In the table, BED, EQD 2 , D eff They represent the biological equivalent dose of the corresponding anatomical structure, 2Gy fractionated equivalent dose and effective dose respectively.

[0066] Table 1. Efficacy models used in the objective functions for lung cancer patients

[0067] Model Model structure structure Dosage Indicators Covariates LRF Cox Target area BED Age, sex, stage, whether combined chemotherapy CE3+ Cox heart <![CDATA[EQD 2 ]]> Previous heart disease history RILT3+ Logistic lung <![CDATA[EQD 2 ]]> Age, smoking history, chemotherapy ARIE2+ LKB esophagus <![CDATA[D eff ]]> Whether combined with chemotherapy

[0068] Step S2: Incorporate the efficacy prediction model constructed above and mix it with the existing physical dose target term and smoothing term to form a weighted new optimization objective function:

[0069]

[0070] St

[0071] (1)x≥0.

[0072] (2) d = Ax.

[0073] (3)C(d)≤0.

[0074] Where x represents the intensity distribution of the field flux, i.e., the variable sought in the plan optimization; A is the energy deposition matrix; d = Ax is the calculated dose; parameter β i It is used to measure the weights between various objective items; C(d) is the dose constraint and dose-volume constraint, which is used to ensure the clinical acceptability of the plan.

[0075] f outcome (d) is the efficacy target item, which is a weighted combination of the prognostic model. This item is used to guide the optimized plan to have a better predicted efficacy; fphysical (d) A weighted combination of physical target items for tumors and organs at risk, which is used to ensure that the optimized plan meets the existing evidence-based indicators of clinical physical dose targets; smooth (x) is a smoothing term, which constrains the plan flux to be smoother and ensures the feasibility of the plan;

[0076] Specifically, the efficacy target item f outcome (d) It is expressed by the gain events caused by radiotherapy and the damage events caused by radiotherapy, as shown in formula (2). efficacy (d) is the efficacy model, which represents the prediction model of prognostic events that can benefit from radiotherapy, such as tumor control rate, survival rate, etc.; f toxicity (d) is the radiation dose-induced organ toxicity model, which represents the prediction model for negative prognostic events caused by radiation therapy, such as prognostic events related to tumor control failure or radiation-induced normal tissue toxicity; γ and λ are the relative weights of each prognostic event, which are used to express the degree of clinical consideration of the occurrence of each prognostic event.

[0077]

[0078] Specifically, the physical target item f physical (d) Based on the physical dose statistical indicators, the form is shown in formula (3):

[0079]

[0080] In formula (3), d is the calculated dose; d 0 is the reference dose, represents the prescription dose for the target area, and is set to 0 for the organ at risk; T is the transposition operator symbol; N is the total number of voxels corresponding to the region of interest; ω is the weight of different regions of interest.

[0081] Specifically, the smoothing target term f smooth In (x), the smoothness is measured by the second-order difference of the field flux map. The smaller the second-order difference, the smoother the flux representation. The second-order difference of the two-dimensional flux map with resolution h, k can be obtained by the flux element x i,j The standard difference formula is used to calculate, as shown in formula (4).

[0082]

[0083] In formula (4), x is the flux intensity of the irradiation field of the IMRT plan; B is the field set; X is the beamlet set in the direction of the field blade; Y is the beamlet set perpendicular to the blade direction; h and k represent the resolution of the flux in the X and Y directions, respectively.

[0084] It should be understood that in other embodiments, the objective / constraint function may only include one or more of the above-mentioned objective / constraint items, but not necessarily all of the objective / constraint items.

[0085] Taking the intensity modulated radiation therapy for lung cancer patients as an example, the efficacy target item f outcome (x) is weighted by the aforementioned LRF model, CE3+ model, RILT3+ model and ARIE2+ model. In addition, for lung cancer patients, the dose and dose-volume constraint function C(d) are shown in Table 2.

[0086] Table 2 Physical constraints used in the objective function for lung cancer patients

[0087]

[0088]

[0089] In Table 2, Gy is the unit of radiation dose.

[0090] Step S3: Setting the parameters of the efficacy objective function and analyzing the characteristics of the function, selecting an appropriate optimization solution method to solve the new objective function, obtaining a flux distribution that makes the objective function as small as possible, and obtaining an intensity-modulated radiotherapy optimization plan.

[0091] Taking the intensity modulated radiation therapy for lung cancer patients as an example, this embodiment simply sets the weight parameters λ and β to 1. The parameters can be adjusted according to the specific conditions of the patients.

[0092] Furthermore, due to the non-convexity of the objective function, this embodiment uses the IPOPT optimizer (Interior Point OPTimizer) to optimize the objective function. IPOPT is a nonlinear optimization solver, and the built-in solution algorithm used is the interior point linear search method. Before using IPOPT, the initial iteration point of the optimization variable needs to be given, which will affect whether the optimization problem converges or converges to a (local) optimal solution.

[0093] Furthermore, in this embodiment, the initial iteration point is set to satisfy that the average dose of the tumor is equal to the prescribed dose.

[0094] Step S4: Retrospectively collect multiple intensity modulated radiotherapy plans from the radiotherapy plan database of clinically treated patients, substitute the above-mentioned new plan optimization objective function, regenerate a new radiotherapy plan, and compare it with the original clinical plan to evaluate and verify the effectiveness of the method proposed in the present invention.

[0095] Specifically, the new radiotherapy plan is compared with the original clinical plan in terms of quantitative results and qualitative results. The quantitative results include the estimated efficacy value and the physical dose target value. The qualitative results include the DVH curve and dose distribution.

[0096] Taking the intensity modulated radiotherapy for lung cancer patients as an example, the intensity modulated radiotherapy plan optimization method with efficacy guidance in the present invention was used to regenerate new radiotherapy plans for three patients with non-small cell lung cancer. The optimized DVH and dose distribution are shown in Figure 2. Figure 2 As shown in the figure, the solid line and dotted line represent the efficacy optimization results and clinical plan, respectively, and the purple, blue, orange and yellow represent the heart, lung, esophagus and spinal cord, respectively. The comparison of efficacy prognosis evaluation of the efficacy plan and clinical plan of 3 patients with lung cancer is shown in Table 3. In addition, the 3-year and 5-year overall survival efficacy prognostic models (OS3+ and OS5+) were used to externally validate the method proposed in the present invention, as shown in Table 4. The experimental results show that the method for optimizing the intensity modulated plan with efficacy guidance proposed in the present invention can design a plan for patients to obtain better prognosis evaluation under the premise of meeting the basic clinical physical criteria.

[0097] Table 3 Comparison of efficacy and prognosis evaluation of 3 patients with lung cancer

[0098]

[0099] Table 4 External validation comparison of prognosis of 3 lung cancer patients

[0100]

[0101] Those skilled in the art will readily appreciate other embodiments of the invention after considering the specification and practicing the invention herein. The invention is intended to cover any variations, uses or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art of the invention. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the claims.

[0102] It should be understood that the present invention is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for optimizing intensity modulated radiotherapy plan based on individual prognosis prediction model, characterized in that: The following steps are involved: (1) Establish a prognostic model for patients undergoing radiotherapy; (2) Taking the best patient prognosis as the main optimization guide, the prognostic model established in step (1) and the dose statistical indicators are combined to construct the objective function; (3) Setting parameters for the objective function in step (2) and performing numerical optimization to obtain an IMRT optimization plan.

2. The method for optimizing intensity modulated radiotherapy plan according to claim 1, wherein: In step (1), the method for establishing the prognosis model comprises the following steps: a. Collect effective IMRT planning data, baseline clinical information, and clinical prognostic events to build a database; b. constructing an association model with the baseline clinical information and the dose distribution information of the region of interest corresponding to the intensity modulated radiotherapy plan data as input and the probability of occurrence of clinical prognostic events corresponding to the region of interest as output; and obtaining a mathematical analytical expression of the prognostic model by substituting the actual collected data into the model parameters; Wherein, the region of interest is a tumor and / or an organ at risk.

3. The method for optimizing intensity modulated radiotherapy plan according to claim 2, wherein: The effective intensity modulated radiotherapy plan data are effective intensity modulated radiotherapy plan data for lung cancer patients, and the prognostic events include local control failure of lung cancer after radiotherapy, radiation cardiac events ≥3, radiation pneumonitis events ≥3, and radiation esophagitis events ≥2; The clinical baseline information of the local control failure event of lung cancer after radiotherapy includes the patient's age, gender, stage and whether combined chemotherapy is used; The clinical baseline information of the grade ≥3 radiation-induced cardiac events included whether there was a history of previous heart disease; The clinical baseline information of the grade ≥3 radiation pneumonitis events included age, smoking history, and chemotherapy; The clinical baseline information of the grade ≥2 radiation esophagitis events included whether combination chemotherapy was used; The association models are Logistic model and Lyman-Kutcher-Burman model; The method of substituting the actual collected data into the model parameters is the maximum likelihood estimation method.

4. The method for optimizing intensity modulated radiotherapy plan according to claim 1, wherein: In step (2), the mathematical expression of the objective function is as shown in Formula 1: St (1)x≥0. (2) d = Ax. (3)C(d)≤0. Formula 1 In formula 1: F outcome (d) is the efficacy target item; f physical (d) is the physical target item; f smooth (x) is the smoothing target item; x represents the flux intensity distribution of the irradiation field of the intensity modulated radiotherapy plan and is the solution variable of the plan optimization model; A is the energy deposition matrix; d is the calculated dose distribution; β1~β3 are parameters used to measure the weights between the various objective items; C(d) is the dose and dose-volume constraint function.

5. The method for optimizing intensity modulated radiotherapy plan according to claim 4, wherein: In step (2), the efficacy target item f outcome The mathematical expression of (d) is shown in formula 2: In formula 2, d is the calculated dose distribution; f efficacy (d) is the efficacy model, including the tumor control rate model and the survival rate model; f toxicity (d) is the model of organ toxicity caused by radiotherapy dose, including the probability model of tumor control failure and the probability model of radiotherapy-induced normal tissue toxicity; the parameters γ and λ are used to balance the weights between various prognoses.

6. The method for optimizing intensity modulated radiotherapy plan according to claim 5, wherein: In step (2), the efficacy target item f outcome The acquisition process of (d) includes the following steps: taking the individualized optimal prognostic item caused by the patient's radiotherapy plan dose as the main guide, using the prognostic model established in step (1) to calculate the probability of occurrence of each prognostic event caused by the radiotherapy dose distribution, and forming the efficacy target item in the form of an objective function through weighting.

7. The method for optimizing intensity modulated radiotherapy plan according to claim 4, wherein: In step (2), the physical target item f physical (d) is constructed based on the physical dose statistics, and its mathematical expression is shown in Formula 3: In Formula 3, d is the calculated dose distribution; d0 is the reference dose; T is the transposition operator; N is the total number of voxels corresponding to the region of interest; and ω is the weight of different regions of interest.

8. The method for optimizing intensity modulated radiotherapy plan according to claim 4, wherein: In step (2), the smoothing target item f smooth In (x), smoothness is measured by the second-order difference of the flux; The expression of the smoothing target term is shown in Formula 4: In formula 4, x is the flux intensity of the irradiation field of the intensity modulated radiotherapy plan; B is the field set; X is the beamlet set in the direction of the field blade; Y is the beamlet set perpendicular to the blade travel direction; h and k represent the resolution of the flux in the X direction and Y direction respectively.

9. The method for optimizing intensity modulated radiotherapy plan according to claim 1, wherein: In step (3), the numerical optimization solution is solved by using the IPOPT interior point method based on the convexity and clinical efficiency of the objective function.

Images