Generation of realizable radiotherapy plans

1. A method for generating a radiotherapy treatment plan, the method comprising:

receiving, by processor circuitry, an input parameter related to a patient, the input parameter comprising at least one of dose information, one or more medical images, or one or more object segmentations;

processing, by the processor circuitry, the input parameter with a machine learning technique to estimate a realizable plan parameter of a radiotherapy treatment plan, wherein the machine learning technique is trained to establish a relationship between the at least one of the dose information, the one or more medical images, or the one or more object segmentations of input parameter and a set of realizable radiotherapy treatment plan parameters to achieve a target radiotherapy dose distribution; and

generating, by the processor circuitry, the radiotherapy treatment plan based on the estimated realizable plan parameter.

2. The method of claim 1 , wherein processing the input parameter comprises:

predicting, using the trained machine learning technique, an estimated treatment plan parameter of the radiotherapy treatment plan based on the input parameter; and

generating the realizable plan parameter that corresponds to the estimated plan parameter.

3. The method of claim 2 , wherein the realizable plan parameter is generated by solving an optimization problem.

4. The method of claim 3 , wherein the optimization problem uses a loss function based on the set of realizable radiotherapy treatment plan parameters and the estimated plan parameter.

5. The method of claim 3 , wherein the optimization problem is solved using at least one of a closed form solution, a gradient based scheme, or a stochastic optimization scheme.

6. The method of claim 2 , wherein generating the realizable plan parameter comprises using at least one of an explicit differentiation process, composition of a fixed number of differentiable iterations, or a differentiable machine learning technique trained to estimate the realizable plan parameter.

7. The method of claim 2 , wherein the estimated treatment plan parameter is incompatible with a radiotherapy device used to treat the patient.

8. The method of claim 7 , wherein the estimated treatment plan parameter corresponds to a first dose distribution that is incompatible with the radiotherapy device; and

wherein the realizable plan parameter is generated by identifying a subset of the set of the realizable radiotherapy treatment plan parameters that are compatible with the radiotherapy device and result in a second dose distribution that is within a specified threshold.

9. The method of claim 1 , wherein the input parameter is further processed to achieve at least one of selected points from a dose-volume histogram or one or more radiobiological properties.

10. The method of claim 1 , wherein the set of realizable radiotherapy treatment plan parameters includes at least one of machine parameters or a dose distribution.

11. The method of claim 1 , wherein the machine learning technique is trained by optimizing a loss function based on a given radiotherapy treatment dose distribution and a given set of realizable radiotherapy treatment plan parameters.

12. The method of claim 11 , where the optimization comprises at least one of a closed form solution, a gradient based scheme, or a stochastic optimization scheme.

13. The method of claim 1 further comprising training the machine learning technique by optimizing two loss functions using bilevel optimization.

14. The method of claim 13 , further comprising:

optimizing a first of the two loss functions based on a deviation between the at least one of the dose information, the one or more medical images, or the one or more object segmentations of input parameter and a set of estimated radiotherapy treatment plan parameters; and

optimizing a second of the two loss functions based on a deviation between a result of optimizing the first loss function and realizable radiotherapy treatment plan parameters corresponding to the set of estimated radiotherapy treatment plan parameters.

15. A non-transitory computer readable medium comprising non-transitory computer readable instructions for generating a radiotherapy treatment plan, the computer readable instructions comprising instructions for performing operations comprising:

receiving an input parameter related to a patient, the input parameter comprising at least one of dose information, one or more medical images, or one or more object segmentations;

processing the input parameter with a machine learning technique to estimate a realizable plan parameter of a radiotherapy treatment plan, wherein the machine learning technique is trained to establish a relationship between the at least one of the dose information, the one or more medical images, or the one or more object segmentations of input parameter and a set of realizable radiotherapy treatment plan parameters to achieve a target radiotherapy dose distribution; and

generating the radiotherapy treatment plan based on the estimated realizable plan parameter.

16. The non-transitory computer readable medium of claim 15 , wherein the operations further comprise:

predicting, using the trained machine learning technique, an estimated treatment plan parameter of the radiotherapy treatment plan based on the input parameter; and

generating the realizable plan parameter that corresponds to the estimated plan parameter.

17. The non-transitory computer readable medium of claim 15 , wherein the target radiotherapy dose distribution comprises at least one of a dose distribution, selected points from a dose-volume histogram, or one or more radiobiological properties.

18. A system for generating a radiotherapy treatment plan, the system comprising:

a memory for storing instructions; and

one or more processors for executing the instructions stored in the memory for performing operations comprising:

receiving an input parameter related to a patient, the input parameter comprising at least one of dose information, one or more medical images, or one or more object segmentations;

processing the input parameter with a machine learning technique to estimate a realizable plan parameter of a radiotherapy treatment plan, wherein the machine learning technique is trained to establish a relationship between the at least one of the dose information, the one or more medical images, or the one or more object segmentations of input parameter and a set of realizable radiotherapy treatment plan parameters to achieve a target radiotherapy dose distribution; and

generating the radiotherapy treatment plan based on the estimated realizable plan parameter.

19. The system of claim 18 , wherein the operations further comprise:

predicting, using the trained machine learning technique, an estimated treatment plan parameter of the radiotherapy treatment plan based on the input parameter; and

generating the realizable plan parameter that corresponds to the estimated plan parameter.