Method and system for robust radiotherapy treatment planning for dose mapping uncertainties
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-03-17
- Publication Date
- 2026-03-16
AI Technical Summary
Current methods of non-rigid image registration (DIR) in radiation therapy are ambiguous and do not accurately account for anatomical changes, leading to uncertain dose mapping and potential inaccuracies in radiation treatment planning.
A robust optimization approach is employed to handle dose mapping uncertainties by considering multiple plausible deformation vector fields (DVF) and using optimization functions to generate a radiation treatment plan that accounts for uncertainties in dose distribution, incorporating error estimates and multiple image registrations.
This approach ensures adequate total dose delivery even in regions with steep dose gradients and uncertain image registration, providing a more reliable and robust radiation treatment plan.
Smart Images
Figure 00000000_0000_ABST
Abstract
Description
Technical Field
[0001] The present disclosure generally relates to radiation therapy, and specifically to the fields of generating, optimizing, and evaluating radiation treatment plans.
Background Art
[0002] Medical imaging methods are widely used in radiation treatment planning to identify and contour targets (tumors) and risk organs (Organ At Risk: OAR) in patients. Contoured targets and OARs from the acquired images, along with other radiation treatment planning parameters such as the minimum dose or target dose irradiated to the tumor and the maximum dose to the OAR, serve as inputs for optimizing the planned dose distribution so that radiation is concentrated on the target while limiting the exposure of the OAR.
[0003] In many treatment situations, multiple images of a patient are acquired at different times and / or spaces. Examples include adaptive planning where an initial plan is adapted over the course of radiation therapy, retreatment where the patient receives additional radiation therapy, and 4D robust optimization where multiple images are acquired over, for example, one respiratory cycle, but are not limited thereto.
[0004] Medical images of the same patient acquired at different times and / or from different viewpoints are compared, and the images undergo image registration (Image Registration: IR) to convert different sets of data into one coordinate system, for example, one first image, in order to obtain information on the irradiated dose. However, during radiation therapy, changes may occur that affect the accuracy of dose irradiation and the spatial distribution of the effective dose. Such changes may include movement, shrinkage or expansion of the tumor, or shape changes of the tumor and surrounding anatomical structures in the patient due to, for example, weight loss / gain.
[0005] To mitigate the effects of anatomical changes, deformable image registration (DIR) is used, where individual volume elements are mapped from one image to another using non-rigid translation and / or rotation, resulting in a deformation map or deformable vector field (DVF). However, since no exemplary deformation completely deforms all anatomical regions, the current method of DIR is ambiguous, and therefore any dose mapping based on DIR is inherently uncertain.
[0006] Conventional techniques do not accurately account for this uncertainty. Rather, the choice lies between relying on the deformed vector field (DVF) in a given region, or completely rejecting it and applying a potentially overly conservative estimate as the mapped dose.
[0007] International Publication No. 2012 / 069965 A1 discloses a system and method for manually or automatically correcting deformation maps arising from the registration of two sets of planned images. However, the correction process is time-consuming, requires user expertise, and does not adequately eliminate the uncertainties of dose mapping. [Overview of the project]
[0008] The object of this disclosure is to provide an improved solution that provides a robust optimization technique for dealing with dose mapping uncertainty in image registration. This object is achieved in a first embodiment of this disclosure which provides a method for generating a robust radiotherapy plan for a subject's treatment volume, where the treatment volume is defined using a plurality of voxels, and the method The steps include receiving a first image of the treatment volume, The steps include receiving at least one second image of the treatment volume, The steps include generating a mapped dose distribution in a first image by mapping doses defined in at least one second image to a first image using image registration, and A step of defining an optimization problem using at least one optimization function for the total dose related to radiotherapy, wherein the total dose is a function of the dose defined in a first image and the dose defined in at least one second image. A step of calculating at least one optimization function value based on at least two mapped doses in the mapped dose distribution in the first image, The process includes the step of generating a radiotherapy plan by optimizing at least one optimization function value evaluated by considering at least two mapped doses in a mapped dose distribution.
[0009] By considering DVF to be uncertain within reasonable limits, or by considering multiple reasonable DVFs, this disclosure enables obtaining a total dose that is robust to modeled dose mapping uncertainties. The main advantage is that by considering not only the nominal mapped dose but also the distribution of mapped doses, it is possible to guide the optimizer to a solution that ensures an appropriate total dose, even in regions with steep dose gradients and regions where IR is uncertain.
[0010] While conventional methods only consider what will be irradiated in the future, this disclosure provides robustness to uncertainties regarding what has already been irradiated. Furthermore, the generation of the mapped dose distribution in the first image can be carried out in several ways and is not limited to any particular dependency structure between voxels in the treatment volume. In this regard, the distribution can be defined as a set of mapped doses and, optionally, as the corresponding probability for each member in the set. Alternatively, the distribution can be defined as nominal values and error estimates. These may be, for example, the mean value per voxel and the standard deviation per voxel, or alternatively, the maximum deviation per voxel.
[0011] In one embodiment, mapping doses in at least one second image to a first image is performed using non-rigid image registration (DIR). DIR can account for non-rigid movement, rotation, and / or other changes in anatomy.
[0012] In one embodiment, mapping doses in at least one second image to a first image is performed using at least two different image registrations, and the mapped dose distribution is a set containing at least two mapped doses. For example, increasing the number of image registrations using several different algorithms or selections of algorithmic parameters can lead to a more reliable estimate of dose mapping uncertainty.
[0013] In one embodiment, mapping doses in at least one second image to a first image is performed by making an error estimate of each vector in a deformed vector field resulting from image registration, and the mapped dose distribution is based on at least two mapped doses resulting from a deformed vector field perturbed according to the error estimate using at least two different perturbations, one of which may be a zero perturbation.
[0014] In one embodiment, the dose in at least one second image is a dose from a previous treatment, part of a treatment, or part of a partial treatment, and the step of generating a radiotherapy plan includes generating a radioretreatment plan or adapting an existing radiotherapy plan.
[0015] In one embodiment, the dose in at least one second image is a partial beam dose in a four-dimensional radiotherapy plan.
[0016] In one embodiment, the optimization problem includes constraints that define parameters to be maintained during optimization. These constraints may, for example, be in the form of a predetermined dose in a predefined sub-volume (e.g., a target) that does not change during optimization. Thus, the target dose is maintained to ensure a specific dose distribution, while the remaining portion of the radiotherapy plan is robustly optimized.
[0017] In one embodiment, the optimization problem includes biological or physical objectives. Preferably, the physical objectives include dose limits to target and risk organs (OARs) in therapeutic volume, dose-volume histogram (DVH) limits, LET limits, particle stopping locations, and / or homogeneity and consistency indices. Thus, biological uncertainties can also be combined with other (physical) objectives. The planning optimization and evaluation must be able to use different biological models in combination and in combination with physical optimization functions and objectives. The optimization problem may be a combination of physical objectives, e.g., minimum and maximum doses to target and risk organs, DVH limits, and biological objectives, e.g., BED, EQD2, EUD, TCP, and NTCP.
[0018] In one embodiment, the physical objectives include dose limits to target and risk organs (OARs) in therapeutic volume, dose-volume histogram (DVH) limits, linear energy transfer (LET) limits, particle stopping locations, and / or homogeneity and consistency indices.
[0019] In one embodiment, the optimization includes a probabilistic programming method in which the expected value of the optimization function is minimized for at least two mapped doses in the mapped dose distribution; a minimax method in which the maximum value of the optimization function is minimized for at least two mapped doses in the mapped dose distribution; or any combination of two methods generally referred to as minimax probabilistic programming; or a voxel worst-case method in which the worst-case dose to each separately considered voxel is optimized.
[0020] In one embodiment, at least two mapped doses in the mapped dose distribution are further combined into an additional set of error scenarios, the additional set of error scenarios representing a specific realization of uncertainty in one or more parameters related to the treatment plan, the parameters including particle range, spatial location of treatment volume, radiotherapy device settings, density of irradiated tissue, interaction effects, organ motion, and / or biological model parameter values.
[0021] According to a second aspect, a computer program product is provided which comprises computer-readable code means that, when running on a computer, causes the computer to implement a method according to the first aspect.
[0022] According to a third aspect, a computer system is provided comprising a processor coupled to a memory that stores computer-readable instructions causing the processor to perform a method according to the first aspect when executed by the processor.
[0023] According to the fourth aspect, a treatment planning system including a computer system as described above is provided.
Brief Description of Drawings
[0024] These features, aspects, and advantages of the present disclosure will be further described in the following description with reference to the accompanying drawings. [Figure 1] Shows three mappings of the dose irradiated on the old image onto the new image. [Figure 2] Shows a flowchart representing steps of a computer-based method for generating a robust radiation treatment plan according to an embodiment of the present disclosure. [Figure 3a] Shows the generation of a mapped dose distribution according to an embodiment of the present disclosure. [Figure 3b] Shows the generation of a mapped dose distribution according to an embodiment of the present disclosure. [Figure 4a] Shows the generation of a mapped dose distribution according to another embodiment of the present disclosure. [Figure 4b] Shows the generation of a mapped dose distribution according to another embodiment of the present disclosure. [Figure 5] Schematically shows a computer-based system for evaluation, visualization, generation, and improvement of a radiation treatment plan according to an embodiment of the present disclosure.
[0025] In this specification, when indicating the same elements common to the drawings, the same reference numbers are used if possible. Further, the images in the drawings are simplified for illustrative purposes and are not necessarily shown to scale.
Modes for Carrying Out the Invention
[0026] Referring to Figure 1, the problem underlying dose mapping uncertainty is illustrated. The left side of Figure 1 shows an image of the patient, where the irradiated dose is indicated. Different shades of color represent the amount of dose irradiated to a specific region of the treatment volume. The right side of Figure 1 shows three different images of the same treatment volume at different time points. For example, to map the dose irradiated on the older image to the newer image, the data obtained from these different images is compared or integrated, or the older image is aligned to the newer image through image registration. Each of the three images represents a different image registration using a different algorithm, resulting in three different mapped doses on the newer image. Therefore, inherent uncertainty exists in dose mapping, which affects radiotherapy planning.
[0027] Figure 2 is a flowchart of one embodiment of the method according to the present disclosure, which can be used in connection with the generation of a radiotherapy plan. In one embodiment, the starting point is an initial treatment plan and the number of scenarios to consider, and the method is intended to obtain an improved treatment plan based on the initial treatment plan, to modify the initial plan with certain constraints, or to obtain a treatment plan that can be provided when the initial treatment plan does not satisfy all mechanical limitations. Depending on the type of data included in the plan, other input data, such as patient-related data, may be required for dose calculation. The initial treatment plan can be obtained in any form known in the art, including scenario-based and non-scenario-based methods.
[0028] A treatment plan is generated for the purpose of providing a therapeutic volume of radiotherapy to a subject (patient), and the therapeutic volume includes a target which may be an organ, a tumor, or a cluster of tumor cells. The therapeutic volume is defined using multiple voxels, as is known in the art.
[0029] In step S100, a first image of the treatment volume is received. The first image may be a reference image representing the patient's current condition, which will be used in the radiotherapy planning. In step S102, at least one second image of the treatment volume is received. The second image may be a target image representing the patient's condition at a previous point in time, for example, during a previous treatment or part of a treatment. Alternatively, the second image may be a phase image representing the patient's condition at a different phase, for example, at a different phase of the patient's respiratory cycle.
[0030] Both the first and second images can be obtained using any suitable medical imaging technique, including but not limited to radiographic computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), ultrasound, and single-photon emission tomography (SPECT).
[0031] In step S104, doses defined in at least one second image are mapped to the first image using image registration, so that a distribution of the mapped doses is generated in the first image. In one embodiment, non-rigid image registration is used to account for non-rigid movement, rotation, and / or other changes in anatomy. Image registration generates a deformation vector field in which vectors representing deformations between the first and second images are assigned to each voxel of the treatment volume.
[0032] In step S106, the optimization problem is defined using at least one optimization function for the total dose related to radiotherapy. The total dose is a function of the dose defined in the first image and the dose defined in at least one second image.
[0033] In step S108, an optimization function value is calculated based on at least two mapped doses in the mapped dose distribution in the first image. The optimization function value can be calculated using the mapped doses and the doses defined in the first image as input variables to obtain one optimization function value. Alternatively, each of the mapped doses can be used as an input variable, with the first function value based on the first mapped dose and the second function value based on the second mapped dose.
[0034] In step S110, the optimization function value is optimized by evaluating the distribution of at least two mapped doses in the mapped doses, and a radiotherapy plan is generated.
[0035] The mapped dose distribution can be generated in several ways and is not limited to any particular dependency structure between voxels. Referring here to Figures 3a and 3b, an example of one embodiment is shown in which at least two different image registrations are used to map doses in at least one second image to a first image. Figure 3a shows a first image including point A and four arrows representing the mapping of four points A', A'', A''', and A'''' in the second image shown in Figure 3b to point A. Each of the four arrows is a vector in the corresponding deformed vector field resulting from four different image registrations.
[0036] A set of mapped doses can be considered as a distribution of mapped doses. More reliable estimations of dose mapping uncertainty can be achieved by increasing the number of image registrations, i.e., by using several different image registration algorithms, or by using the same algorithm with different settings or by adding different perturbations.
[0037] Referring now to Figures 4a and 4b, another embodiment of the method for generating the mapped dose distribution is shown, where each vector in the deformed vector field is considered to have an error. Figure 4a shows a first image including two points A and B, and vectors assigned to each point, representing the mapping of points A' and B' in the second image shown in Figure 4b to points A and B. The estimation of the error for the two vectors in Figure 4a is given by radius ε, respectively. A and ε B These are represented by circles centered at points A' and B', respectively. Then, using this error estimate, multiple plausible recognitions for each vector are generated to produce a mapped dose distribution. More specifically, the mapped dose distribution can be based on at least two mapped doses resulting from a deformed vector field perturbed according to the error estimate using at least two different perturbations. One of the perturbations may be a zero perturbation. That is, the deformed vector field obtained from the image registration is considered without perturbing the vectors.
[0038] The method described herein may be used in different planning situations. In one embodiment, the method is used to generate a radiotherapy retreatment plan or to adapt an existing radiotherapy plan. In such a case, the dose defined in at least one second image is the dose from a previous treatment, part of a treatment, or part of a partial treatment, which is also referred to herein as the background dose. The total dose is then the sum of the background dose and the new dose to be delivered in the retreatment plan or adapted plan.
[0039] In one embodiment, the method is used to generate a four-dimensional radiotherapy plan in which the treatment volume undergoes periodic spatial deformation, for example, during different phases of the patient's respiratory cycle. In such a case, the dose defined in at least one second image is the partial beam dose delivered during a particular phase. Then the total dose is the sum of the partial beam doses that will be delivered in the four-dimensional radiotherapy plan.
[0040] The optimization function may be included as a constraint on the optimization. Alternatively, the optimization function may be included as a component of the objective function in the optimization. Typically, goals are set for a treatment, and these goals are used to define the components of the objective function, constraints, or combinations thereof. The components of the objective function are the desired goals that the optimization should aim for, or should achieve as far as possible, while the constraints are strict goals or states that must be exactly satisfied, such as a minimum dose to the tumor, a maximum dose to the OAR, or boundaries of variables that control the objective function.
[0041] Various types of optimization methods for achieving robustness can be used in conjunction with the methods according to this disclosure. For example, minimax (or "composite worst-case") optimization can be used, where the worst-case scenario is optimized for the composite objective function. In that case, the optimization problem is formulated as follows:
number
[0042] Another type of optimization method that achieves robustness is expected value optimization, where the expected value is optimized for uncertainty. The optimization problem is formulated as follows:
number
[0043] A third alternative is a voxel-based worst-case optimization method. In this method, two artificial worst-case dose distributions are used, d high and d low However, it is calculated based on the mapped dose. Here, d high This is calculated as the maximum mapped dose for each voxel, which is considered individually for each scenario, and d low This is calculated as the minimum mapped dose for each voxel considered individually for each scenario, i.e., as follows:
number
number
[0044] At that time, the optimization problem was formulated as follows:
number
[0045] Another alternative is to minimize an objective function h(x) that is not necessarily related to (but may be related to) the full set of scenarios S, and to include the following constraint on the function f(x;s) for all s in S.
number
number
[0046] Other methods, such as the stochastic minimax method which is a combination of composite worst-case optimization and expected value optimization, can also be used, and these are known in the art.
[0047] Referring here to Figure 5, a simplified schematic diagram of a computer-based system 100 for generating a radiotherapy plan 114 according to this disclosure is shown. The computer-based system 100 includes a memory or database 110 which stores computer programs 116 for generating the radiotherapy plan 114 and an improved radiotherapy plan 118. The memory 110 is any volatile or non-volatile memory device, which may be, for example, a flash drive, a hard drive, an optical drive, dynamic random access memory (DRAM), static random access memory (SRAM), and any other suitable device for storing information and for retrieving and using information for subsequent data processing. Furthermore, the system 100 includes one or more hardware processors 120 for performing data processing which are capable of accessing the memory 110. The hardware processor 120 may be fabricated from one or more of the following: a central processing unit (CPU), a digital signal processor (DSP), a reduced instruction set computer (RISC), an application-specific integrated circuit (ASIC), a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a parallel processor system, or a combination of these different hardware processor types.
[0048] Computer program 116 is written in computer-readable instructions, which can be moved to a hardware processor 120 and executed by the hardware processor 120. When the computer-readable instructions are executed by the hardware processor 120, they implement a method for generating an improved radiotherapy plan 118. When computer program 116 is executed, the results of the processing performed by the hardware processor 120, such as the improved radiotherapy plan 118 and related data, can be stored in memory 110. The hardware processor 120 can access memory 110 via direct memory access (DMA), or it can use cache memory to store temporary processing results. The computer program 116 can also be stored on non-temporary computer-readable media 130, such as general-purpose serial bus (USB) flash drives, optical data media such as CD-ROMs, DVD-ROMs, and Blu-ray discs, floppy disks, swappable hardware drives, USB external hard drives (HDDs), or any other portable information storage devices. As a result, the computer program 116 can be moved to different computing systems and even loaded into the memory 110 of system 100. This can be done by connecting the computer-readable media 130, such as an optical drive or USB interface, to system 100 via a data reader / writer 140.
[0049] Furthermore, system 100 also includes a display unit 150 having a display driver that enables visualization of the results of data processing. This unit visualizes, for example, a three-dimensional (3D) representation of a patient's target volume, 3D contour data, or a two-dimensional (2D) slice representation of the biological effects (e.g., probability of injury / cell death / side effects) of both the target volume and the organs at risk, with respect to various cross-directions and LET distributions. For example, a 3D computer reconstruction of a CT scan may be displayed. The display unit 150 can also display a dose-volume histogram (DVH) summarizing the 3D dose distribution using a graphic 2D format. For example, the display unit 150 is configured to display comparative DVH diagrams for a patient's volume showing the dose contribution of radiotherapy plan 114 and for the same volume of an optimized or improved radiotherapy plan 118, which also allows for a visual comparison of LET distributions.
[0050] The display unit 150 is used to display 3D scans of the patient created before, during, or after treatment. For example, a 3D computer reconstruction of a CT scan may be displayed. The display unit 150 can also display LET, dose, and / or DVH summarizing the 3D dose distribution, either by using a graphic 2D format or by using a numerical format. For example, the display unit 150 is configured to display a comparative LET diagram to the patient's volume showing the cancer cell breakdown or dose contribution of the radiotherapy plan 114. This is shown and compared for the same volume of optimized or improved radiotherapy plans so that improvements can be visually compared. It is also possible that the display unit 150 has touchscreen functionality to display a graphical user interface for operating the system 100.
[0051] In addition, the computer system 100 has a system bus 160, which connects a hardware processor 120, memory 110, data reader 140, touchscreen, and various other data input / output interfaces, as well as peripheral devices not shown. For example, the computer system 100 may be connected to a keyboard 170 for user data input, and may also be connected to an external radiotherapy planning device 180, such as a powerful dedicated computer, which has constructed a radiotherapy plan. The system 100 may also be connected to a CT scanner not shown. For example, the external device 180 which has constructed a radiotherapy plan 114 can develop doses, and an LET distribution calculation algorithm coded in the software has access to radiation data, machine calibration data, and patient-specific information regarding the dose distribution, as well as the patient's target volume and organs at risk. The external device 180 can then provide the radiotherapy plan 114 to the computer system 100 for evaluation, visualization, creation of new plans, and improvement of existing plans, taking the LET distribution into consideration. However, the computer program 116 can also operate on the external device itself, thereby generating not only the radiotherapy plan 114 but also an improved radiotherapy plan 118.
[0052] Furthermore, a computer program product is introduced to perform parameter optimization. The computer program product 130 includes computer-readable code means, which, when run on a computer, implements the method described above.
[0053] Example 1: A retreatment plan robust to background dose uncertainty. Certain patients who have received radiation therapy may experience tumor recurrence near the previously treated volume. In these cases, a second course of radiation therapy, called retreatment or re-irradiation, may be considered. At that time, a first image is taken in the form of a planning image, which is the target of the planned retreatment. A second image, in the form of an older planning image from the previous treatment, is also considered. To avoid healthy tissue exceeding a certain total partial corrected dose, the retreatment plan must take into account the dose already irradiated present in the second image. The irradiated dose can be mapped from the second image to the first image following image registration and will be used as the background dose. However, image registration, and therefore the background dose, can be uncertain or unclear. Different image registrations may result in different background doses.
[0054] Question: Which mapped doses are reliable for use as background doses in a total dose-based retreatment plan?
[0055] Proposed solution: To account for dose mapping uncertainty, generate at least two background dose distributions and optimize the total dose robustly against OAR. - Define a new treatment plan for the first image received. - Define at least one objective for the total dose to at least one OAR. This objective may be robustly optimized against uncertainties in dose mapping. Examples of such objectives are as follows: • Maximum total physical dose relative to OAR • Maximum total biological equivalent dose (BED) relative to OAR, e.g., EQD2 - Define at least one objective for the retreatment dose to the target. -Optionally, define additional error scenarios that represent a specific recognition of one or more additional relevant parameters for radiotherapy planning, such as setting error or range error. - The second image is received from the initial treatment plan, and image registration is performed between the first and second images to map the irradiated dose in the second image to the first image. - Generate a mapped dose distribution and, based on that distribution, define a scenario in one of the following two ways, for example: Perform at least one additional image registration using an alternative algorithm or alternative algorithm parameters. Define a scenario for each resulting mapped dose. • Calculate error estimates for image registration and use them to characterize the mapped dose distribution. Define scenarios based on artificial worst-case dose distributions, such as the stochastic minimax method described above. - Optimize the retreatment plan using a preferred, robust optimization framework.
[0056] Example 2: Adaptive treatment plan with robustness to background dose uncertainty. During the course of radiotherapy, it may be desirable to adapt the original treatment plan to account for changes in the patient's anatomical structure or previous errors in the treatment irradiation. In this case, a first image in the form of a replanned image is acquired. At least one second image in the form of a partial image of the irradiated area is also considered. For use as background dose, the irradiated dose may be mapped from at least one second image to the first image following image registration. However, image registration, and therefore background dose, may be uncertain or unclear. Different image registration techniques may result in different background doses.
[0057] Question: Which mapped doses are reliable for use as background doses in adaptive treatment planning based on total dose?
[0058] Proposed solution: To account for dose mapping uncertainty, generate at least two background dose distributions and optimize the total dose robustly against OAR. - Define a new, adapted treatment plan for the first image received. - Define at least one objective for the total dose. This objective may be robustly optimized against uncertainties in dose mapping. Examples of such objectives may include: • Maximum total physical dose relative to OAR • Maximum total BED for OAR • Minimum total physical dose relative to the target volume -Optionally, define additional error scenarios that represent a specific recognition of one or more additional relevant parameters for radiotherapy planning, such as setting error or range error. -Receive at least one second image from a previously irradiated area, perform image registration between the first image and at least one second image to map the dose irradiated in at least one second image to the first image. For each of the second images, generate the mapped dose distribution in the first image using, for example, one of the following two methods: • Perform at least one additional image registration using an alternative algorithm or alternative algorithm parameters. • Calculate an error estimate for the image registration and use it to characterize the mapped dose distribution. - Define scenarios based on the recognition of uncertainty in each of the mapped dose distributions from at least one second image. - Optimize the retreatment plan using a preferred, robust optimization framework.
[0059] Example 3: A 4D treatment plan robust to dose mapping uncertainty. When treating tumors in body parts that undergo periodic motion, it is sometimes desirable to be robust to the timing of dose delivery within the motion cycle. For this purpose, multiple images from various phases of the motion cycle may be acquired. In this case, the optimized dose is a function of the dose in each phase image, which is mapped to a common reference image by DIR.
[0060] Question: Which dose mapping is reliable for use as phase dose in 4D optimized treatment planning?
[0061] Suggested solution: Generate at least two mapped doses from at least one phase image to a reference image, and optimize the mapped doses to be robust to dose mapping uncertainty. - Define 4D treatment plans for a moving patient volume. - Define at least one objective for dose in the reference phase image. This objective may be robustly optimized against uncertainty in dose mapping. Examples of such objectives are as follows: • Maximum total physical dose relative to OAR • Minimum total physical dose relative to the target volume -Optionally, define additional error scenarios that represent a specific recognition of one or more additional relevant parameters for radiotherapy planning, such as setting error or range error. -Receive at least one second image, which is a phase image from a different phase in the motion cycle, and perform image registration between the first image used and at least one second image to map the dose in at least one second image to the first image. -For each of the second images, a mapped dose distribution is generated in the first image. In this case, the mapped dose will be a function of the optimization variable and its change in each iteration of the optimization. The mapped dose distribution can be generated, for example, by one of the following methods: • Perform at least one additional image registration using an alternative algorithm or alternative algorithm parameters. • Calculate an error estimate for the image registration and use it to characterize the mapped dose distribution. - Define scenarios based on the recognition of uncertainty in each of the mapped dose distributions from at least one second image. - Optimize the retreatment plan using a preferred, robust optimization framework.
[0062] Preferred embodiments of a method and system for generating a radiotherapy plan have been disclosed above. However, those skilled in the art will understand that these can be modified within the scope of the appended "Claims" without departing from the concept of the present invention.
[0063] All of the alternative embodiments described above, or parts of one embodiment, can be freely combined or used separately without departing from the concept of the present invention, provided that the combination is inconsistent. The following abbreviations are used: BED (Biological Equivalent Dose) CT (Computed Tomography) CTV Clinical Tumor / Target Volume DICOM: Digital imaging and communication in medicine DVH dose-volume histogram EHR Electronic Health Record System EQD equivalent standard partial dose Equivalent dose in the EQD2 2Gy portion SDD Equivalent Uniform Distribution eMIX Electronic Medical Information Exchange System GUI (Graphical User Interface) GTV Macroscopic Tumor / Target Volume HIS Hospital Information System HIM Health Information Management System IGRT (Image-Guided Radiation Therapy) IMRT (Intensity-Modulated Radiation Therapy) LET Linear Energy Transfer MLC Multi-Leaf Collimator MRI (Magnetic Resonance Imaging) System MU Monitor Unit NTCP Normal Tissue Damage Probability OAR-risk organs PBS Pencil Beam Scan PET positron emission tomography PTV Planning Tumor / Target Volume QA (Quality Assurance) QC quality control US Ultrasound Examination RBE (Relativistic Biological Effect) ROI Area of Interest RVS Recording and Verification System SPECT Single-Photon Positron Emission Tomography TCP tumor control probability
Claims
1. A method for generating a robust radiotherapy plan for a subject's treatment volume, wherein the treatment volume is defined using a plurality of voxels, and the method is: Receiving the first image of the aforementioned treatment volume, Receiving at least one second image of the aforementioned treatment volume, By mapping the dose defined in at least one second image to the first image using image registration, a distribution of the mapped dose in the first image is generated. Defining an optimization problem using at least one optimization function for the total dose related to the radiotherapy, wherein the total dose is a function of the dose defined in the first image and the dose defined in at least one second image, Calculate at least one optimization function value based on at least two mapped doses in the mapped dose distribution in the first image, A method comprising generating a radiotherapy plan by optimizing the at least one optimization function value, which is evaluated by considering the at least two mapped doses in the mapped dose distribution.
2. The method according to claim 1, wherein mapping the dose in at least one second image to the first image is performed using non-rigid image registration.
3. The method according to claim 1, wherein mapping the dose in at least one second image to the first image is performed using at least two different image registrations, and the distribution of the mapped dose is the change between the at least two mapped doses.
4. Mapping the dose in at least one second image to the first image is performed by estimating the error of each vector in the deformation vector field generated from the image registration. The method according to claim 1, wherein the mapped dose distribution is based on at least two mapped doses resulting from the deformed vector field perturbed according to the error estimation using at least two different perturbations.
5. The method according to claim 1, wherein the dose in at least one second image is a dose from a previous treatment, part of a treatment, or part of a partial treatment, and generating the radiotherapy plan includes generating a radiotherapy retreatment plan or adapting an existing radiotherapy plan.
6. The method according to claim 1, wherein the dose of at least one second image is a partial beam dose of a four-dimensional radiotherapy plan.
7. The method according to claim 1, wherein the optimization problem includes constraints that define parameters maintained during the optimization.
8. The method according to claim 1, wherein the optimization problem includes a biological goal or a physical goal.
9. The method according to claim 8, wherein the physical objectives include dose limits to target and risk organs in the therapeutic volume, dose-volume histogram limits, linear energy delivery limits, locations where particles stop, and / or homogeneity and consistency indices.
10. The aforementioned optimization is, The method according to claim 1, comprising: a probabilistic programming method that minimizes the expected value of the optimization function for the at least two mapped doses in the mapped dose distribution; a minimax method that minimizes the maximum value of the optimization function for the at least two mapped doses in the mapped dose distribution; or any combination of two commonly referred to as minimax probabilistic programming methods; or a voxel worst-case method that optimizes a worst-case dose to each separately considered voxel.
11. The at least two mapped doses in the mapped dose distribution are further combined with additional error scenarios. The method according to claim 1, wherein the additional error scenarios represent a specific recognition of uncertainty in one or more parameters related to the treatment plan, the parameters including particle range, spatial location of the treatment volume, radiotherapy device settings, density of irradiated tissue, interaction effects, organ motion, and / or biological model parameter values.
12. A computer program product (130), Including non-temporary computer-readable media, The non-temporary computer-readable medium includes a computer program product (130) which, when executed on a computer, causes the computer to perform the method according to claim 1.
13. A computer system (100) comprising a processor (120) coupled to a memory (110) for storing computer-readable instructions, wherein when the computer-readable instructions are executed in the processor, the computer system (100) causes the processor to execute the method according to claim 1.
14. A radiotherapy planning system comprising the computer system (100) described in claim 13.