Systems and methods for automated radiation treatment planning

Abstract

The radiation treatment planning method includes a step of optimizing a treatment plan based on at least one proposed dose map according to a set of clinical targets. The resulting dose distribution is compared with the at least one clinical target, and if the optimized dose distribution does not achieve the at least one clinical target, the method continues to a step d in which the dose map is adjusted before a new treatment plan is optimized. When the optimized dose distribution achieves the clinical target, the treatment plan is accepted. [Selected Figure] Figure 1

Description

[Technical Field]

[0001] The present invention relates to systems, methods and computer program products for radiation therapy planning, and in particular for contributing to the automation of such planning. [Background technology]

[0002] In radiation therapy, the goal is usually to deliver a sufficiently high radiation dose to a target (e.g., a tumor) within a patient while minimizing the radiation dose to surrounding normal tissue. In particular, it is important to minimize the dose to sensory organs close to the target. Treatment plans, which define the treatment parameters, such as treatment machine settings, to be used in a radiation therapy session, are usually determined with the aid of computer-based treatment planning systems.

[0003] In inverse treatment planning, an optimization algorithm is used to find a set of treatment parameters that results in an acceptable dose distribution within the subject that preferably satisfies all clinical goals defined by the clinician. Clinical goals can take many forms. Some common goals include: A requirement that a specified minimum or maximum dose be delivered to at least or at most a specified fraction or volume of a region of interest (ROI); A requirement that a specified minimum or maximum fraction or volume of the ROI should receive at least or at most a specified dose; Minimum, maximum, or average dose requirements to the ROI, ·Requirements for how well the dose matches the target ·Uniformity requirements at the target -Requirements on how the dose decays with distance to the target The requirement that a specific point on the patient should receive a specified dose, and Requirements for biological measures such as tumor control rate or probability of normal tissue damage in the ROI.

[0004] Traditionally, optimization of treatment plans requires a great deal of manual input. The results generally depend on the experience of the treatment planner and, for example, the selection of treatment goals used for optimization. Furthermore, the process by no means guarantees that the best possible treatment plan will be obtained. Even experienced treatment planners typically require a great deal of "trial and error" before an acceptable treatment plan is found. Furthermore, if the dose distribution of an optimized treatment plan is generally satisfactory but has some deficiencies, the treatment planner may not know how to adjust the optimization goals or constraints, or target weights, to address the deficiencies. Automating the optimization process has proven difficult because current methods rely heavily on human judgment, which is difficult to automate. Summary of the Invention [Problem to be solved by the invention]

[0005] SUMMARY OF THE INVENTION It is an object of the present invention to overcome or at least mitigate the aforementioned drawbacks, and in particular to facilitate automated treatment planning with optimal achievement of clinical goals. [Means for solving the problem]

[0006] The present invention provides a computer-based radiation therapy planning method, comprising: a. obtaining at least one proposed dose map to be used in treatment planning for a patient; the at least one dose map specifies at least one desired dose level for at least a first region of the patient according to a set of clinical goals including at least one clinical goal for the patient; b. optimizing the treatment plan based on the dose map to obtain an optimized dose distribution; c. comparing the optimized dose distribution of the treatment plan with at least one clinical goal, and continuing to step d if the optimized dose distribution does not achieve at least one clinical goal; d. adjusting at least one dose map in at least one region if the optimized dose distribution does not meet at least one clinical goal based on the results of the comparison; e. repeating steps b through d for at least one adjusted dose map; f. terminating the procedure when it is determined that the optimized dose distribution in step c achieves at least one clinical goal; This usually involves receiving a treatment plan that can be used to treat the patient.

[0007] The proposed method is therefore based on attempting to mimic a proposed dose map by optimizing the plan to result in a dose distribution corresponding to the dose map. The dose map is iteratively updated based on how well the optimized plan meets the clinical goals to mimic the dose map. According to the invention, this can be done in a fully automated manner. In particular, steps c and d are preferably performed computer-generated without the need for user input.

[0008] Optimization is performed at least in part with the aim of obtaining a specific dose distribution specified to achieve clinical goals. Such optimization is referred to herein as "dose simulation," indicating that the goal of optimization is to find a set of treatment parameters that results in a dose distribution that matches or "mimics" as closely as possible a specific desired dose distribution, called a dose map. Dose simulation can also be based on a spatial dose map, i.e., using a specific reference dose target that differs for each voxel. The optimization then aims, at least in part, to achieve a dose distribution in which each voxel is as similar as possible to or as little above / below the dose level specified in the spatial dose map. Alternatively or additionally, dose simulation can be based on a dose map corresponding to a specific dose-volume histogram (DVH), i.e., using a previously obtained DVH curve as a reference in the optimization. In this case, the optimization aims to achieve a dose distribution in a given ROI that does not necessarily have to be as spatially close as possible to the spatial dose map, but whose DVH is as similar as possible to or as small as possible above / below the DVH of the dose map.

[0009] The simulation may also involve more than one dose map. The dose map or maps preferably include: a. Clinical target dose and volume levels; b. A previously created manual plan for the current patient or another patient; c. Knowledge-based predictions, The method is based on one or more of the following:

[0010] The method preferably includes determining a direction and magnitude of at least one adjustment of the dose map, said direction and magnitude being determined with the aim of bringing the resulting dose distribution closer to achieving the clinical goal.

[0011] The dose map may be adjusted by setting at least one new dose value for at least one voxel. Alternatively, the dose map is adjusted by setting at least one new importance weight for at least one voxel. Another option is to adjust the dose map by adjusting a target DVH curve for at least one volume within the patient.

[0012] If, based on the results of the comparison, the optimized dose distribution does not meet at least one clinical target, the step of adjusting the dose map in at least one region preferably includes determining at least one voxel for which the calculated optimized dose deviates from at least one clinical target for the region at least partially containing the at least one voxel, determining the direction of the deviation, and adjusting the dose map in that voxel to bring the dose for the at least one voxel closer to the at least one clinical target for the region at least partially containing the at least one voxel.

[0013] The set of clinical targets preferably includes at least two clinical targets for the patient. The clinical targets may be prioritized into at least two priority levels, in which case the dose map adjustment may be determined based on the priority level of the clinical targets. The dose map adjustment may then be determined according to the clinical target with the highest priority. Alternatively, the dose map adjustment may be determined according to more than one target, for example as a weighted average of the targets. If there are targets associated with non-overlapping regions or targets with preferred adjustments in the same region in the same direction, the dose map adjustment may be determined according to multiple targets.

[0014] The invention also relates to a computer program product comprising computer readable code means which, when executed on a processor, cause the processor to perform the method of any one of the preceding claims. The computer program product may be stored on a memory unit, such as a non-transitory memory unit. Furthermore, the invention relates to a computer system comprising a processor and a program memory, such as a non-transitory program memory, for holding a computer program product according to the above.

[0015] The present invention also relates to a treatment planning system comprising a processor, at least one data memory comprising data for obtaining a treatment plan including a set of clinical targets and at least one dose map to be used for planning, and at least one program memory comprising a computer program product according to the above.

[0016] The invention will now be described in more detail, by way of example only, with reference to the accompanying drawings, in which: [Brief explanation of the drawings]

[0017] [Figure 1] FIG. 1 is a schematic diagram of a dose map. [Figure 2] 1 is a flowchart outlining a method according to one embodiment of the present invention. [Figure 3] 1 is an overview of a computer system in which methods according to some embodiments of the present invention may be implemented; DETAILED DESCRIPTION OF THE INVENTION

[0018] FIG. 1 is a schematic diagram of a dose map for a slice 11 through a patient. The first region 13 corresponds to the target, and a dark color indicates that this region should receive a high dose. Two regions 15, 17 that should receive a low dose, typically the organs at risk, are shown without color. The remainder of the slice is lightly colored, meaning that the dose should be limited but is not as critical as the organs at risk 15, 17. As will be appreciated, any suitable number of different dose levels can be established. Also, typically the slice will be divided into voxels, and there will be a table or the like specifying the dose level for each voxel as a numerical value.

[0019] FIG. 2 is a flowchart outlining a method according to one embodiment of the present invention. As will be appreciated, the method is implemented by one or more computer programs in a computer. In a first step S21, a proposed dose distribution for treating a particular patient is generated based on clinical goals and data about the patient. The proposed dose distribution defines dose values ​​for each region of the patient and is often referred to as a dose map. This initial dose map may be calculated using any suitable method, as described below. It is also possible to use more than one dose map, including two or more. For example, using one upper dose map and one lower dose map is well known in the art.

[0020] Dose planning in step S21 may be performed in any suitable manner, typically according to a set of rules defining a relationship between the applied dose, the class of treatment plan, and at least one feature of the image data, for example by interferometric techniques known per se from, for example, WO 2014 / 197994. The rules may include rules generated by machine learning, mathematical functions, and other rules well known to those skilled in the art. A proposed dose map may define a proposed dose across a volume of image data. For example, knowledge-based planning may be used to generate the initial dose map. Alternatively, the initial dose map may be based solely on clinical objectives.

[0021] Initially, one or more dose maps may be specified for each voxel. Each voxel or other subregion of the image may be characterized by one or more appearance features, such as the anatomical structure to which it belongs, density, or other features of the image. A set of rules may then be used to associate the voxel features with treatment plan classes and, optionally, other patient characteristics, such as age or body part. Clinical objectives may also be used to define isodose or DVH curves.

[0022] In step S22, dose simulation is performed. As used herein, the term dose simulation refers to optimizing a treatment plan based on a dose map (or multiple dose maps) with the goal of obtaining a dose distribution to be evaluated. The ultimate goal is to arrive at a dose distribution that achieves clinical goals within certain limits. Optimization can be performed in several ways, for example, by penalizing deviations of the dose at each voxel from the dose level of the corresponding voxel in the dose map and optimizing toward finding a plan with the lowest possible penalty. If an upper dose map limit is used, only deviations above the specified dose level are penalized. Similarly, if a lower dose map limit is used, only deviations below the specified dose level are penalized. The penalty for each voxel may have an individual importance weight. In the first iteration of loop S22-S24, optimization is based on the simulation of the initial dose map.

[0023] Step S23 is a decision step for determining whether to perform another loop in the optimization process. If the result of the dose simulation in step S22 is not satisfactory, the answer in step S23 is yes, and the process continues to step S24, where one or more dose maps are adjusted. The process then returns to step S22, where dose simulation is performed based on the adjusted dose map or maps. If the answer in step S23 is no, the method continues to step S25, where the treatment plan is accepted. This means that if the dose distribution obtained in step S22 is not deemed to achieve the clinical goal sufficiently well, one or more dose maps are adjusted to bring the obtained dose distribution closer to the clinical goal. If the dose distribution obtained in step S22 is deemed to achieve the clinical goal, no further adjustment of the dose maps is required, and the optimized treatment plan in step S22 is accepted. The treatment plan may then be used to treat the patient.

[0024] The adjustment of the dose map in step S24 is preferably done automatically by a computer program to adjust the resulting dose in one or more subregions of the dose map. For example, if in the treatment plan resulting from step S22, certain voxels or groups of voxels receive doses that are too high according to the clinical goal, the dose at those voxels in the dose map may be decreased. Similarly, if in the treatment plan resulting from step S22, certain voxels or groups of voxels receive doses that are too low according to the clinical goal, the dose at those voxels in the dose map may be increased. Thus, in steps S23 or S24, voxels or regions where the clinical goal is not being achieved are identified, and the direction and magnitude of the deviation are determined to allow appropriate adjustment of the dose values ​​in step S24. Thus, the adjusted dose map or maps output from step S24 may specify dose levels in one or more regions of the dose map or maps that do not correspond to the actual desired dose in one or more regions, but that are intended to influence the dose in a desired direction.

[0025] The dose map adjustment may be performed by individually setting new dose values ​​for the voxels or groups of voxels to be adjusted, or by specifying an increase or decrease for the voxels or groups of voxels. Alternatively, the adjustment may be performed by adjusting the target DVH curve.

[0026] The voxels to be adjusted, as well as the magnitude and direction of the adjustment, can be determined in several different ways. For example, for a clinical goal that specifies a minimum dose to a region, subregions with doses below this level can be selected to be adjusted. The magnitude of the adjustment can also be correlated with the deviation from the specified dose level. For a clinical goal that specifies a minimum or maximum dose level for a certain percentage x of a region of the patient, voxels can also be ordered according to their current dose level in the dose distribution, with x% of voxels with the highest or lowest dose, respectively, selected to be adjusted, or selected to be adjusted if they do not reach the minimum or maximum dose level, respectively. Similar methods for selecting voxels to be adjusted can also be specified for average dose goals, dose attenuation goals, etc.

[0027] In step S23, determining whether the resulting dose distribution achieves the clinical goals sufficiently well typically involves comparing the dose distribution for at least one region of the patient with at least one clinical goal for that region. If the dose distribution does not deviate from the at least one clinical goal by more than a specified amount or percentage, the dose distribution is considered to achieve that at least one goal. Similarly, if the dose distribution deviates by more than a specified amount or percentage, the dose distribution is not considered to achieve that at least one goal. In some cases, goals for one region may be met but goals for another region may not be met. In such cases, achievement of some goals may be determined to be sufficient to accept the treatment plan, and the treatment plan may be accepted in step S25, even though other goals are not achieved. Alternatively, it may be determined that the goals are not achieved overall, and the procedure should continue to step S24. Alternatively, the treatment plan may be determined to be acceptable in step S25 if computational resources, such as the number of acceptable steps to be taken, are exhausted.

[0028] When there are conflicting goals for the same sub-region, the direction of the adjustment can be determined based on the priority of the goals. For example, if one goal has a higher priority than another, the goal with the highest priority can be used to determine the adjustment. Alternatively, a compromise between the goals can be determined and used to determine the adjustment. For example, a weighted average of their desired updates can be applied. This is particularly true when the goals have the same priority level.

[0029] 3 is a schematic diagram of a system for radiation therapy planning and treatment in which the method of the present invention may be performed. A computer 31 comprises a processor 33, first and second data memories 34, 35, and first and second program memories 36, 37. One or more user input means 38, 39 are also present, preferably in the form of a keyboard, mouse, joystick, voice recognition means, or any other available user input means. The user input means may also be configured to receive data from an external memory unit.

[0030] The first data memory 34 comprises clinical data and / or other information used to derive a treatment plan. The second data memory 35 comprises one or more dose maps for one or more patients to be used for treatment planning according to embodiments of the present invention. The first program memory 36 holds a computer program known per se, configured for treatment plan optimization. The second program memory 37 holds a computer program configured to cause a computer to perform the method steps described in connection with FIG. 2.

[0031] It will be understood that the data memories 34, 35 and the program memory are shown and described schematically. There may be several data memory units each holding one or more different types of data, or one data memory holding all data in a suitably structured manner; the same applies to the program memory. One or more memories may also be mounted on other computers. For example, a computer may be configured only to perform one of the methods, and there may be another computer for performing the optimization.

Claims

1. A computer system (31) for generating a radiation therapy plan based on a computer program, the computer system (31) comprising a processor (33), readable code means, and a program memory (36, 37) for holding the computer program, wherein, when processing is executed in the processor (33), the processor (33) executes the following steps based on the computer program: a. obtaining (S21) at least one proposed dose map for a patient to be used in generating a treatment plan defining treatment parameters, said dose map being a dose distribution defining dose values ​​for each region of said patient; The at least one dose map specifies at least one desired dose level for at least a first region of a patient according to a set of clinical goals including at least one clinical goal for the patient; b) optimizing the treatment plan by adjusting the proposed dose map to minimize deviations from the dose levels specified in the dose map in order to obtain a dose distribution that is as close as possible to the dose map (S22); c) comparing the dose distribution obtained by the optimization of the treatment plan with the at least one clinical target and determining whether the dose distribution achieves the at least one clinical target (S23); d. If the obtained dose distribution does not achieve the at least one clinical target based on the result of the comparison, determining a penalty based on a direction and a magnitude of deviation of a dose level at each voxel of the obtained dose distribution from a dose level of a corresponding voxel based on the at least one clinical target, the penalty having an importance weight for the at least one voxel based on the deviation, and adjusting the dose map in at least one region that does not meet the at least one clinical target based on the determined penalty (S24); e. Repeating steps b through d with respect to the adjusted dose map; and f. accepting the treatment plan (S25) when it is determined in step c that the resulting dose distribution does not deviate from the at least one clinical goal by more than a specified amount or percentage and achieves the at least one clinical goal, the clinical goal being a requirement that a specified minimum or maximum dose be delivered to at least or at most a specified fraction or volume of a region of interest (ROI); a requirement that a specified minimum or maximum fraction or volume of the ROI should receive at least or at most a specified dose; Minimum, maximum, or average dose requirements to the ROI; the requirement for uniformity at the target, and The requirement that a specific point on the patient should receive a specified dose and satisfying at least one requirement selected from the group consisting of: A computer system configured to:

2. the dose map a. the clinical target dose and volume levels; b. A previously created manual plan for the current patient or another patient; c. educated prediction; The computer system of claim 1 , wherein the computer system is obtained based on one or more of:

3. 3. The computer system of claim 1, further comprising steps (S23, S24) of determining a direction and magnitude of at least one adjustment of the at least one dose map, wherein, with respect to a clinical goal defining a minimum dose to a region, the direction for subregions having a dose below this level can be selected as upward or increasing, and the magnitude of the adjustment is determined by correlating it with the deviation from the defined dose level.

4. 4. The computer system of claim 1, wherein the set of clinical goals includes at least two clinical goals related to a patient, the clinical goals are prioritized into at least two priority levels, and adjustments to the dose map are determined based on the priority levels of the clinical goals.

5. The computer system of claim 4 , wherein the dose map adjustments are determined according to the clinical goal having the highest priority.

6. The computer system of claim 5 , wherein the dose map adjustment is determined according to more than one goal as a weighted average of the goals.

Images