Method and system for automated quality assurance and automated treatment planning in radiation therapy
Methods and systems for evaluating a proposed treatment plan for radiation therapy, for evaluating one or more delineated regions of interest for radiation therapy, and/or for generating a proposed treatment plan for radiation therapy. Machine learning based on historical data may be used.
1. A method for generating a proposed treatment plan for radiation therapy, the method comprising:
obtaining a set of patient data for a patient including a set of image data for at least one treatment site;
determining a treatment plan class, from a plurality of predefined treatment plan classes, each predefined treatment plan class defining one or more treatment plan features relevant to treatment of a respective treatment site;
calculating a proposed dose map by determining a dosage over a volume depicted in the set of image data according to a first set of rules defining expected relationships between an applied dosage, the determined treatment plan class, and at least one feature of the set of image data;
determining one or more treatment plan parameters for achieving the proposed dose map according to a second set of rules defining expected relationships between the applied dosage and treatment plans;
generating as output the proposed treatment plan including the one or more determined treatment plan parameters; and
displaying a visualization of the proposed dose map on a display device, wherein the visualization further comprises providing, with the visualization, a user interface for receiving user input,
wherein the first set of rules includes at least one of a rule generated by computer learning based on historical data, a mathematical function relating dose values of the proposed dose map to the at least one feature of the set of image data, or a general rule governing dose maps entered manually.
2. The method of claim 1 , wherein determining the treatment plan class comprises:
determining the treatment plan class based on the patient data; and
using an automated classification algorithm,
wherein the treatment plan class is determined based on one or more of:
a patient's electronic medical record;
similarity of features of the set of image data with image features of a given treatment plan class;
presence of one or more defined regions of interest (ROIs) corresponding to one or more ROIs of a given treatment plan class;
manually inputted data; and
metadata provided with the image data.
3. The method of claim 1 , further comprising:
obtaining a set of region of interest (ROI) data delineating at least one ROI in the set of image data;
wherein calculating the proposed dose map includes determining a dosage over the at least one ROI according to the first set of rules, the at least one feature of the set of image data comprising the at least one ROI, and further comprising:
automatically characterizing the at least one ROI according to one or more predefined features to determine at least one ROI characterization;
wherein the first set of rules comprises at least one rule defining expected relationships between the applied dosage and the at least one ROI characterization.
4. The method of claim 3 , wherein calculating the proposed dose map further includes determining a dosage over a volume of the image data according to the first set of rules, using a non-ROI feature of the set of image data.
5. The method of claim 4 , wherein the first set of rules includes at least one rule defining an expected relationship between the applied dosage and the at least one ROI, and at least one rule defining an expected relationship between the applied dosage and the non-ROI feature of the image data.
6. The method of claim 5 wherein the second set of rules includes one or more of:
historical treatment parameters for a given planned dosage;
a mathematical function; and
a general rule governing treatment parameters irrespective of the proposed dose map.
7. The method of claim 1 , wherein the proposed dose map defines voxel-by-voxel a range of possible doses, further comprising, prior to determining the one or more treatment plan parameters, adjusting the proposed dose map to define specific voxel-by-voxel doses within the range of possible doses according to a third set of rules defining expected relationships between the applied dosage and the determined treatment plan class.
8. The method of claim 1 , wherein the visualization comprises a voxel-by-voxel mapping of proposed dosages superimposed on the set of image data and a voxel-by-voxel indication of a confidence measure for each voxel of the proposed dose map,
wherein the user input received by the user interface modifies at least one of:
proposed dosage for at least one voxel of the proposed dose map, and one or more features of the image data; and
recalculating the proposed dose map in accordance with an inputted modification.
9. The method of claim 1 , wherein the first set of rules is the rule generated by computer learning based on historical data.
10. The method of claim 1 , wherein the first set of rules is the mathematical function relating dose values of the proposed dose map to the at least one feature of the set of image data.
11. The method of claim 1 , wherein the first set of rules is the general rule governing dose maps entered manually.