Knowledge based multi-criteria optimization for radiotherapy treatment planning
A method of generating a treatment plan for treating a patient with radiotherapy, the method includes obtaining a plurality of sample plans, which are generated by use of a knowledge base comprising historical treatment plans and patient data. The method also includes performing a multi-criteria optimization based on the plurality of sample plans to construct a Pareto frontier, where the plurality of sample plans are evaluated with at least two objectives measuring qualities of the plurality of sample plans such that treatment plans on the constructed Pareto frontier are Pareto optimal with respect to the objectives. The method further includes identifying a treatment plan by use of the constructed Pareto frontier.
1. A method of generating a treatment plan for treating a patient with radiotherapy, the method comprising:
identifying a plurality of organs at risk for a planning target volume;
obtaining a plurality of sample plans, wherein the plurality of sample plans are generated by use of a knowledge base comprising historical treatment plans and patient data;
performing a multi-criteria optimization, based on the plurality of sample plans, to construct a Pareto frontier, wherein the plurality of sample plans are evaluated with at least two objectives measuring qualities of the plurality of sample plans, wherein the performing the multi-criteria optimization to construct the Pareto frontier comprises using a confidence level associated with a predicted sample plan, wherein treatment plans on the constructed Pareto frontier are Pareto optimal with respect to the objectives, wherein the performing the multi-criteria optimization further comprises generating a Pareto frontier with one organ at risk at a time of the plurality of organs at risk;
receiving a selected tradeoff point on the constructed Pareto frontier;
utilizing a minimum Euclidean distance criterion such that a point on the constructed Pareto frontier closest to the selected trade off point is an identified treatment plan; and
generating the identified treatment plan by use of the constructed Pareto frontier, wherein said generating the identified treatment plan by use of the constructed Pareto frontier comprises generating (i) a combinational treatment plan comprising a combination of sample plans and (ii) a machine deliverable control point sequence corresponding to the combinational treatment plan, wherein the control point sequence is a sequence that can be delivered by a treatment machine.
2. The method of claim 1 , wherein a sample plan of the plurality of sample plans is selected from training plans of the knowledge base.
3. The method of claim 1 , wherein a sample plan of the plurality of sample plans is generated utilizing an estimation model of the knowledge base, wherein the estimation model is constructed by use of training plans of the knowledge base.
4. The method of claim 1 , wherein the performing the multi-criteria optimization further comprises each of the plurality of sample plans is evaluated with a value related to a Tumor Control Probability (TCP).
5. The method of claim 1 , wherein said generating the identified treatment plan by use of the constructed Pareto frontier is performed by adjusting plan metrics of the plurality of sample plans.
6. The method of claim 1 , wherein the objectives are selected from plan quality metrics of the plurality of sample plans.
7. The method of claim 1 , further comprising updating the knowledge base with the generated identified treatment plan.
8. The method of claim 1 , wherein generating the machine deliverable control point sequence corresponding to the combinational treatment plan comprises:
combining a plurality of fluences associated with the plurality of sample plans to form combined fluences; and
transforming the combined fluences into a machine deliverable control point sequence using a leaf sequencing algorithm.
9. A system for generating a treatment plan for treating a patient with radiotherapy, the system comprising:
a memory that stores machine-readable instructions; and
a processor communicatively coupled to the memory, the processor operable to execute the instructions to
identify a plurality of organs at risk for a planning target volume;
obtain a plurality of sample plans, wherein the plurality of sample plans are generated by use of a knowledge base, the knowledge base comprising historical treatment plans and patient data;
perform a multi-criteria optimization, based on the plurality of sample plans, to construct a Pareto frontier, wherein the plurality of sample plans are evaluated with at least two objectives measuring qualities of the plurality of sample plans, wherein the perform the multi-criteria optimization to construct the Pareto frontier comprises use of a confidence level associated with a predicted sample plan, wherein treatment plans on the constructed Pareto frontier are Pareto optimal with respect to the objectives, wherein the perform the multi-criteria optimization further comprises generate a Pareto frontier with one organ at risk at a time of the plurality of organs at risk;
receive a selected tradeoff point on the constructed Pareto frontier;
utilize a minimum Euclidean distance criterion such that a point on the constructed Pareto frontier closest to the selected trade off point is an identified treatment plan; and
generate the identified treatment plan by use of the constructed Pareto frontier, wherein said generate the identified treatment plan by use of the constructed Pareto frontier comprises generate (i) a combinational treatment plan comprising a combination of sample plans and (ii) a machine deliverable control point sequence corresponding to the combinational treatment plan, wherein the control point sequence is a sequence that can be delivered by a treatment machine.
10. The system of claim 9 , further comprising a displaying device, wherein at least a portion of the constructed Pareto frontier is presented on the displaying device.
11. The system of claim 9 , wherein a sample plan of the plurality of sample plans is selected from training plans of the knowledge base.
12. The system of claim 9 , wherein a sample plan of the plurality of sample plans is generated utilizing an estimation model of the knowledge base, wherein the estimation model is constructed by use of training plans of the knowledge base.
13. The system of claim 9 , wherein the perform the multi-criteria optimization further comprises each of the plurality of sample plans is evaluated with a value related to a Normal Tissue Complication Probability (NTCP).
14. The system of claim 9 , wherein said generate the identified treatment plan by use of the constructed Pareto frontier is performed by adjusting plan metrics of the plurality of sample plans.
15. The system of claim 9 , wherein the objectives are selected from plan quality metrics of the plurality of sample plans.
16. The system of claim 9 , wherein the processor is further operable to update the knowledge base with the generated identified treatment plan.
17. The system of claim 9 , wherein the generate the machine deliverable control point sequence corresponding to the combinational treatment plan comprises:
combine a plurality of fluences associated with the plurality of sample plans to form combined fluences; and
transform the combined fluences into a machine deliverable control point sequence using a leaf sequencing algorithm.
18. A non-transitory computer readable storage medium having embedded therein program instructions, when executed by one or more processors of a device, causes the device to execute a process for generating a treatment plan for treating a patient with radiotherapy, the process comprising:
identifying a plurality of organs at risk for a planning target volume;
obtaining a plurality of sample plans, wherein the plurality of sample plans are generated by use of a knowledge base, the knowledge base comprising historical treatment plans and patient data;
performing a multi-criteria optimization, based on the plurality of sample plans, to construct a Pareto frontier, wherein the plurality of sample plans are evaluated with at least two objectives measuring qualities of the plurality of sample plans, wherein the performing the multi-criteria optimization to construct the Pareto frontier comprises using a confidence level associated with a predicted sample plan, wherein treatment plans on the constructed Pareto frontier are Pareto optimal with respect to the objectives, wherein the performing the multi-criteria optimization further comprises generating a Pareto frontier with one organ at risk at a time of the plurality of organs at risk;
receiving a selected tradeoff point on the constructed Pareto frontier;
utilizing a minimum Euclidean distance criterion such that a point on the constructed Pareto frontier closest to the selected trade off point is an identified treatment plan; and
generating the identified treatment plan by use of the constructed Pareto frontier, wherein said generating the identified treatment plan by use of the constructed Pareto frontier comprises generating (i) a combinational treatment plan comprising a combination of sample plans and (ii) a machine deliverable control point sequence corresponding to the combinational treatment plan, wherein the control point sequence is a sequence that can be delivered by a treatment machine.
19. The non-transitory computer readable storage medium of claim 18 , wherein the performing the multi-criteria optimization further comprises each of the plurality of sample plans is evaluated with at least one of an Equivalent Uniform Dose (EUD) for a planning target volume (PTV), an EUD for an Organ at Risk (OAR), Dose Volume Indices (DVI), a Heterogeneity Index (HI), or a value related to a Normal Tissue Complication Probability (NTCP).
20. The non-transitory computer readable storage medium of claim 18 , wherein generating the machine deliverable control point sequence corresponding to the combinational treatment plan comprises:
combining a plurality of fluences associated with the plurality of sample plans to form combined fluences; and
transforming the combined fluences into a machine deliverable control point sequence using a leaf sequencing algorithm.