Expansion of search space for multicriteria optimization in radiation treatment planning

1. A method for radiation treatment planning, the method comprising:

by control circuitry:

receiving user input defining a plurality of clinical goals for a radiation treatment;

defining a plurality of quality metrics corresponding to the clinical goals;

generating an initial set of base plans, each base plan having an associated set of machine parameters and an associated dose distribution;

generating at least one virtual plan by extrapolation from the initial set of base plans, the at least one virtual plan having an associated dose distribution, wherein generating the at least one virtual plan includes:

selecting a pair of base plans from the initial set of base plans;

assigning a weight to each of the base plans in the initial set of base plans such that a positive weight is assigned to each of the selected pair of base plans and a negative weight is assigned to each other base plan in the initial set of base plans; and

computing a dose distribution for the at least one virtual plan by combining the dose distributions of the initial set of base plans according to the assigned weights;

defining an expanded search space that includes the initial set of base plans and the at least one virtual plan;

providing a user interface for navigation of the expanded search space;

receiving, via the user interface, user input identifying a satisfactory outcome within the expanded search space;

generating a deliverable treatment plan based on the user input identifying the satisfactory outcome; and

operating a radiation treatment system in accordance with the deliverable treatment plan to perform a radiation treatment on a patient.

2. The method of claim 1 wherein generating the initial set of base plans includes:

generating a seed plan; and

determining, from the seed plan, a Pareto-optimal set of plans, wherein each plan in the Pareto-optimal set of plans is optimized to emphasize different ones of the quality metrics,

wherein the initial set of base plans includes the seed plan and the Pareto-optimal set of plans.

3. The method of claim 1 wherein generating the at least one virtual plan includes:

assigning a weight to each base plan in the initial set of base plans, wherein the weight assigned to at least one of the base plans is less than zero; and

computing a dose distribution for the at least one virtual plan by combining the dose distributions of the base plans according to the weights.

4. The method of claim 1 wherein the weights are selected such that a sum of the weights is equal to 1.

5. The method of claim 1 wherein the initial set of base plans includes a Pareto-optimal set of plans, wherein each plan in the Pareto-optimal set of plans is optimized to emphasize different ones of the quality metrics, and wherein generating the at least one virtual plan further includes:

rejecting the at least one virtual plan if the at least one virtual plan is non-Pareto-optimal or renders any of the initial base plans non-Pareto-optimal.

6. The method of claim 1 wherein the at least one virtual plan is rejected if the dose distribution computed for the at least one virtual plan includes more than a maximum number of voxels having negative doses.

7. The method of claim 1 further comprising:

presenting information about the deliverable treatment plan to a user.

8. The method of claim 1 wherein the deliverable treatment plan includes a control point sequence and a multileaf collimator leaf sequence and wherein operating the radiation treatment system in accordance with the deliverable treatment plan includes:

providing, by a treatment head of the radiation treatment system coupled with a radiation source, radiation at one or more angles specified by the control point sequence and using a sequence of movements of a multileaf collimator specified by the multileaf collimator leaf sequence, such that radiation in accordance with the treatment plan is delivered to the patient.

9. A system comprising:

a memory; and

a processor coupled to the memory and configured to:

receive user input defining a plurality of clinical goals for a radiation treatment;

define a plurality of quality metrics corresponding to the clinical goals;

generate an initial set of base plans, each base plan having an associated set of machine parameters and an associated dose distribution;

generate at least one virtual plan by extrapolation from the initial set of base plans and by:

selecting a pair of base plans from the initial set of base plans;

assigning a weight to each of the base plans in the initial set of base plans such that a positive weight is assigned to each of the selected pair of base plans and a negative weight is assigned to each other base plan in the initial set of base plans; and

computing a dose distribution for the at least one virtual plan by combining the dose distributions of the initial set of base plans according to the assigned weights, the at least one virtual plan having an associated dose distribution;

define an expanded search space that includes the initial set of base plans and the at least one virtual plan;

provide a user interface for navigation of the expanded search space;

receive user input identifying a satisfactory outcome within the expanded search space;

generate a deliverable treatment plan based on the user input identifying the satisfactory outcome; and

operate a radiation treatment system in accordance with the deliverable treatment plan to perform a radiation treatment on a patient.

10. The system of claim 9 wherein the processor is further configured such that generating the initial set of base plans includes:

generating a seed plan; and

determining, from the seed plan, a Pareto-optimal set of plans, wherein each plan in the Pareto-optimal set of plans is optimized to emphasize different ones of the quality metrics,

wherein the initial set of base plans includes the seed plan and the Pareto-optimal set of plans.

11. The system of claim 9 wherein the processor is further configured such that generating the at least one virtual plan includes:

assigning a weight to each base plan in the initial set of base plans, wherein the weight assigned to at least one of the base plans is less than zero; and

computing a dose distribution for the at least one virtual plan by combining the dose distributions of the base plans according to the weights.

12. The system of claim 9 wherein the weights are selected such that a sum of the weights is equal to 1.

13. The system of claim 9 wherein the initial set of base plans includes a Pareto-optimal set of plans, wherein each plan in the Pareto-optimal set of plans is optimized to emphasize different ones of the quality metrics, and wherein the processor is further configured such that generating the at least one virtual plan further includes:

rejecting the at least one virtual plan if the at least one virtual plan is non-Pareto-optimal or renders any of the initial base plans non-Pareto-optimal.

14. The system of claim 9 wherein the processor is further configured such that the at least one virtual plan is rejected if the dose distribution computed for the at least one virtual plan includes more than a maximum number of voxels having negative doses.

15. The system of claim 9 wherein the processor is further configured to:

present information about the deliverable treatment plan to a user.

16. The system of claim 9 further comprising:

a treatment head coupled with a radiation source; and

a multileaf collimator,

wherein the processor is further configured such that operating the radiation treatment system in accordance with the deliverable treatment plan includes providing, by the treatment head, radiation from the radiation source at one or more angles specified by the treatment plan and using a sequence of movements of the multileaf collimator specified by the treatment plan, such that radiation in accordance with the deliverable treatment plan is delivered to the patient.