Methods and systems relating to radiation therapy treatment planning - Patents.com

Abstract

The present disclosure relates to the use of machine learning to determine initial machine setting parameters for radiation therapy treatment planning. A machine learning system is trained on a data set that includes a dose distribution and a set of machine parameter settings resulting from the dose distribution. The trained system can be used to determine machine parameter settings based on a desired dose distribution, and the machine parameter settings can be used as initial machine parameter settings for radiation treatment optimization.

Description

[Technical field]

[0001] Technical Field The present invention relates to radiation therapy treatment plan optimization, and in particular to parameter initialization for such optimization procedures. [Background technology]

[0002] background

[0002] In radiotherapy treatment plan optimization, an optimization problem is set up and a radiotherapy treatment plan is optimized to achieve a desired dose distribution in a patient given a set of variables, typically including machine parameters of the radiotherapy treatment device used. Initial values ​​of these parameters can be set in various ways.

[0003]

[0003] For photon therapy, the machine parameters typically include MLC leaf sequencing. These initial machine parameter values ​​can be determined by solving a fluence map optimization problem and then converting the optimized fluence map dose to feasible machine parameters. This includes target projection, fluence map optimization and conversion to machine parameters. This is a time-consuming process, and the conversion step is generally a source of inaccuracies in the planning procedure, leading to long optimization times because many iterations are required.

[0004]

[0004] For ion therapy such as proton therapy, the machine parameters include spot placement, spot weight and beam energy. In pencil beam scanning, as an example, the initial values ​​of these parameters can be set in several different ways, and the implementations vary across different clinics. One possible implementation includes calculating the target projection and then using some formula to determine the initial value of the spot weight. Summary of the Invention [Problem to be solved by the invention]

[0005]

[0005] The present disclosure aims to make the treatment plan optimization procedure faster and enable better treatment plans resulting from the treatment plan optimization procedure. [Means for solving the problem]

[0006] Summary of the Invention

[0006] The present disclosure relates to the use of machine learning to determine initial machine setting parameters for radiation therapy treatment planning. Accordingly, the present disclosure relates to a computer-based method of training a machine learning system, comprising inputting a plurality of data sets, each data set including one or more dose distributions and a set of machine parameter settings including at least one machine parameter setting resulting from the one or more dose distributions in a planning procedure, to the machine learning system, and training the machine learning system to output at least one set of machine parameter settings based on a reference dose distribution.

[0007] The present disclosure also relates to a machine learning system trained according to the above, said machine learning system being configured to receive input data in the form of one or more dose distributions, and to output at least one machine parameter setting suitable for generating the dose distributions for a particular radiation therapy delivery device.

[0008]

[0008] The present disclosure also relates to a computer-based method of using such a machine learning system to determine machine parameter settings. The method includes inputting one or more reference dose distributions to the machine learning system, performing parameter initialization by the machine learning system, and outputting a set of machine parameter settings from the machine learning system, the set including at least one machine parameter setting of a radiation therapy delivery device. A better set of machine parameter settings can be obtained because the machine parameter settings are based on knowledge of suitable machine parameter settings for similar dose distributions.

[0009]

[0009] Therefore, according to the present invention, more accurate input data for the machine parameters of the radiotherapy delivery device can be obtained efficiently by machine learning. This means that the planning optimization time can be reduced because the initial data is more accurate. The method is particularly useful for machine parameter initialization used in machine learning-based optimization, but is also useful for any other type of optimization procedure.

[0010]

[0010] For photon therapy, the set of machine parameter settings may include one or more of MLC leaf settings, MU settings, start and stop angles, table angles, and pitch. For intensity modulated radiation therapy (IMRT) applications, the set of machine parameter settings may include one or more of segmented MLC (SMLC) or dynamic MLC (DMLC). For ion therapy, such as proton therapy, the set of machine parameter settings may include, for example, one or more of spot placement, spot weight, and beam energy.

[0011]

[0011] The present disclosure also relates to a computer-based method for optimizing a radiation therapy treatment plan, which includes performing a method for determining a set of machine parameter settings according to any of the embodiments outlined above prior to performing plan optimization, and using the resulting set of machine parameter settings as initial settings for corresponding machine parameters in the radiation therapy treatment plan optimization.

[0012]

[0012] One or more machine parameter settings determined according to the above may be used in any optimization procedure capable of generating a baseline dose to derive predictions of initial machine parameters to be used as optimization variables. Some examples of such optimization procedures are: Dose mimicking optimization as implemented in machine learning planning. A machine learning model predicts a reference dose, which is then converted to a deliverable dose by dose mimicking optimization. · Multi-criteria optimization (MCO): A reference dose is calculated from the fluence map and then optimized using dose mimicry by simultaneously optimizing two or more objective functions. If a plan has already been optimized for delivery with one delivery machine, recreate the same dose distribution for delivery with a different delivery machine by using the original plan as the reference dose distribution for dose mimicking optimization. · Conventional inverse treatment planning for Intensity Modulated Arc Therapy (VMAT), where the optimized fluence map is used to calculate the reference dose (in this case the invention replaces the leaf transform algorithm).

[0013]

[0013] The methods disclosed herein may be used with any treatment modality, including both photon-based and ion-based modalities, such as proton therapy, carbon ion therapy or helium ion therapy.

[0014]

[0014] The present disclosure also relates to a computer program product comprising computer readable code means which, when executed on a computer, causes the computer to perform any of the methods outlined above. The computer program product may comprise non-transitory storage means having stored thereon the computer readable code means. The present disclosure also relates to a computer comprising a processor and a program memory, said program memory storing the computer program product.

[0015] BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The invention will now be described in more detail, by way of example and with reference to the accompanying drawings, in which: [Brief description of the drawings]

[0016] [Figure 1] 1 generally illustrates steps for obtaining machine parameter settings according to the present disclosure. [Diagram 2] 1 illustrates steps for obtaining machine parameter settings for photon-based therapy according to one embodiment. [Diagram 3] 1 is a flow chart of a treatment plan optimization method using the parameter optimization method of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] Detailed Description of the Preferred Embodiments

[0016] The present disclosure relates to a method for training a machine learning system to return suitable machine parameter settings for a radiotherapy treatment plan based on one or more reference doses. Such machine parameter settings reflect the capabilities of the radiation delivery device used to deliver the plan. For photon-based plans, the parameter settings may relate to MLC leaf positions and / or other parameters, such as monitor units (MUs), start and stop angles, table angles and pitch, the latter being the possible tilt of the patient support. For proton or other ion-based plans, other parameter settings may be relevant, such as spot positions, spot weights and beam energies. The present disclosure also relates to a method for obtaining machine parameter settings for a radiotherapy treatment plan using the machine learning system thus trained. The machine learning system may be any type of machine learning system, including neural networks, such as deep neural networks.

[0018]

[0017] Exactly which machine parameter settings are required depends on the delivery system used to deliver the plan. The machine learning system can be adapted to determine all relevant machine parameter settings or only a subset of them. Those that are not determined by the machine learning system need to be provided by some other method.

[0019]

[0018] The parameter settings returned from the trained neural network can be used as initial values ​​to the treatment planning procedure. The treatment planning procedure can be any suitable planning method, such as a conventional optimization procedure using dose mimicking or optimization problems. Such procedures are known in the art and typically use an optimization problem designed to ensure a suitable dose distribution across the treatment site that corresponds to one or more reference dose distributions, possibly together with other objectives such as treatment time. The machine parameter settings determined based on the one or more reference dose distributions are used as initial values ​​for the optimization variables.

[0020]

[0019] In the application of dose mimicking procedures, the input data to the optimization process includes one or more reference dose distributions, each of which is a 3D volume with dose values ​​corresponding to the desired dose distribution of a patient. The dose distributions can be generated in any suitable manner. For example, the reference doses can be ML predicted doses, fluence map optimized doses, or doses obtained from another planning or treatment technique to be mimicked.

[0021]

[0020] An input data set to a training process for training a machine learning system includes a set of machine parameter settings and a dose distribution generated from each set of machine parameter settings, which may be, for example, a clinical plan. Thus, the machine parameter settings constitute a solution that is considered correct in light of the dose distributions of the same input data set.

[0022]

[0021] Figure 1 shows an example of a procedure for obtaining machine parameter settings by a machine learning system 11. As will be appreciated, the nature of the machine parameter settings returned from the process will vary depending on the treatment modality and the type of delivery device used in the treatment. On the left is a reference dose distribution 13 that is used as input data to the machine learning system 11. This is the dose distribution that the patient should ideally receive. The reference dose distribution in this example is given as slices of dose. The machine learning system is a Unet model. The functioning of such models is well known in the art.

[0023]

[0022] Figure 2 shows an example of photon-based therapy. As in Figure 1, a machine learning system 21 receives input data in the form of a reference dose distribution 23, which is the desired dose distribution to be used in treating the patient, and returns machine parameter settings. In Figure 2, a subset of the parameter settings is returned in the form of MLC leaf positions 25. The output from the machine learning system in this example is a leaf opening matrix, indicated diagrammatically with reference numeral 15, that defines the leaf positions of the MLC to be used in treatment plan optimization.

[0024]

[0023] In this example, the MU setting is required but is not determined by the machine learning system. Therefore, the MU optimization 27 is performed based on the machine parameter settings and returns an initial value of the MU. Alternatively, the initial MU setting can be predicted or determined in any other suitable manner and used in the subsequent optimization.

[0025]

[0024] Figure 3 shows an overall flow of treatment plan optimization incorporating steps according to an embodiment of the present invention. Initially, steps corresponding to the present disclosure in Figure 1 are performed. A reference dose S31 corresponding to the patient's desired dose distribution is input to a machine learning system trained as outlined above for parameterization S32. The output S33 from the parameterization procedure includes machine parameter settings of the delivery device used in the treatment. These output machine parameter settings and an optimization problem S34 are used as input to an optimization procedure S35. The optimization problem S34 and optimization procedure S35 may be as conventionally used in the art. Thus, the initial steps S31-S33 replace steps conventionally performed to provide machine parameter settings, including target projection, fluence map optimization, and leaf sequencing. As will be appreciated, the use of a machine learning system capable of providing these parameter settings S33 directly from the reference dose saves time and effort.

[0026] The optimization procedure in step S35 may rely on one of the following methods. Dose mimicking optimization as implemented in machine learning planning. A machine learning model predicts a reference dose, which is then converted to a deliverable dose by dose mimicking optimization. · Multi-criteria optimization (MCO): A reference dose is calculated from the fluence map and then optimized using dose mimicry by simultaneously optimizing two or more objective functions. If a plan has already been optimized for delivery with one delivery machine, recreate the same dose distribution for delivery with a different delivery machine by using the original plan as the reference dose distribution for dose mimicking optimization. · Conventional inverse treatment planning for Intensity Modulated Arc Therapy (VMAT), where the optimized fluence map is used to calculate the reference dose (in this case the invention replaces the leaf transform algorithm).

[0027]

[0026] Figure 4 is a schematic diagram of a computer configured to perform one or more of the methods according to the present disclosure. The computer 41 includes a processor 43, a data memory 44 and a program memory 45. Preferably, one or more user input means 47, 48 are also present in the form of a keyboard, a mouse, a joystick, a voice recognition means and / or any other available user input means. The user input means may be configured to receive data from an external memory unit.

[0028]

[0027] The program memory 45 holds a computer program configured to control the processor to perform the procedures. Like the data memory 44, the program memory may be implemented as one or several units as deemed appropriate. The data memory 44 holds input data that may be used in each procedure and output data generated from the plans. Input data to the training procedures includes a training data set. Input data to the method for determining machine parameter settings includes one or more reference dose distributions. Input data to the method for optimizing the plan includes, for example, a reference dose distribution and an objective function derived from the clinical objectives.

Claims

1. 1. A computer-based method for training a machine learning system, comprising: inputting a plurality of datasets to the machine learning system, each dataset including one or more dose distributions and a set of machine parameter settings including at least one machine parameter setting that results from the one or more dose distributions in a planning procedure, and training the machine learning system to output at least one set of machine parameter settings based on a reference dose distribution.

2. 13. A machine learning system trained according to claim 1 , configured to receive input data in the form of one or more reference dose distributions, and to output a set of machine parameter settings including at least one machine parameter setting suitable for generating the one or more reference dose distributions for a particular radiation therapy delivery device.

3. 3. A computer-based method for determining machine parameter settings using the machine learning system of claim 2, comprising the steps of inputting one or more reference dose distributions into the machine learning system, parameter initialization by the machine learning system, and outputting a set of machine parameter settings for a radiation therapy delivery device from the machine learning system.

4. The method of claim 3 , wherein the set of machine parameter settings includes an MLC leaf setting.

5. The method of claim 3 or 4, wherein the set of machine parameter settings includes MU settings.

6. The method of claim 3 , wherein the set of machine parameter settings includes one or more of spot geometry, spot weight, and beam energy.

7. 7. A computer-based method for optimizing a radiation therapy treatment plan, comprising: performing a method for determining machine parameter settings according to any one of claims 3 to 6 before performing a plan optimization; and using the obtained at least one machine parameter setting as an initial setting for said machine parameters in the radiation therapy treatment plan optimization.

8. The method of claim 7 , wherein the plan optimization is performed by optimizing an optimization problem.

9. The method of claim 8 , wherein the plan optimization is performed by dose mimicking.

10. A computer program product comprising computer readable code means which, when executed on a computer, causes said computer to carry out the method according to claim 1 or the method according to any one of claims 3 to 9.

11. A computer program product comprising non-transitory storage means storing computer readable code means which, when executed on a computer, causes said computer to perform the method of claim 1 or the method of any one of claims 3 to 9.

12. 11. A computer system including a processor and a program memory, the program memory storing the computer program product of claim 10.

Images