Radiation therapy treatment plan evaluation and improvement

Abstract

A radiotherapy treatment plan can be evaluated based on a combination of two or more metrics by simulating the outcome of the delivery of the plan, obtaining values for at least a first metric and a second metric from the simulation, performing a multivariate analysis of the values and determining the quality of the plan based on the outcome of the multivariate analysis. The metrics can include at least one of dose, LET, trajectory endpoint, RBE, alpha / beta ratio, and statistical uncertainty of the plan.

Description

Technical Field

[0001] The present invention relates to a computer-based method for evaluating radiotherapy treatment plans, a computer program product, and a computer system for performing the method. Background Technology

[0002] When developing a radiotherapy treatment plan for a patient, various factors are considered, such as the patient's anatomy and the type of radiation. The radiotherapy treatment plan must also be evaluated before it can be delivered to the patient to determine if it will produce the desired effect.

[0003] To evaluate a radiotherapy treatment plan, it is important to determine the dose distribution resulting from that plan, and particularly the dose to regions of interest (ROIs), such as the target and one or more organs at risk (OARs). Dose-volume histograms (DVHs) are typically used to indicate the fraction of a region receiving at least a certain dose (cumulative histogram) or a dose within a certain limit (differential histogram) against the range of doses received. For example, this can be used to ensure that the plan will result in the entire target receiving a dose substantially above a certain minimum target dose and / or substantially all OARs receiving a dose below a maximum target dose, ensuring that damage to the OARs is limited.

[0004] However, besides dosage, other factors can also affect the effectiveness of a treatment plan on a patient. These factors include linear energy transfer (LET) and trajectory endpoints. Treatment planning systems are increasingly able to monitor factors other than dosage.

[0005] The purpose of this invention is to provide an improved method for evaluating radiotherapy treatment plans. Summary of the Invention

[0006] This disclosure relates to a computer-implemented method for evaluating radiotherapy treatment plans. The method includes:

[0007] • The results of the simulation plan delivery

[0008] • Obtain the values ​​of at least the first and second metrics from the simulation.

[0009] • Multivariate analysis of execution values, and

[0010] • Determine the quality of the plan based on the results of this multivariate analysis.

[0011] For example, a metric may include any combination of two or more of the following:

[0012] Physical dose

[0013] • RBE dosage

[0014] • LET, for example in the form of weighted averages, such as dose-weighted LET or flux-weighted LET.

[0015] • RBE factor

[0016] • Dose rate

[0017] • Track endpoints

[0018] • Statistical uncertainty of predicted dose

[0019] • The alpha / beta ratio reflects the tissue's sensitivity to radiation.

[0020] The assessment results can be used to determine whether the plan is suitable for delivery to patients. It can also be used to guide improvements to the plan.

[0021] This invention is based on the understanding of the effects of treatment on patients through the interaction of different metrics. For example, the damage caused by a particular dose will vary depending on the level of lung injury (LET). Therefore, for example, if the LET is low, a higher dose to the organs at risk is acceptable compared to if the LET is high.

[0022] Multivariate analysis can be bivariate, where the relationship between two factors is determined. Preferred, but not necessarily, is dosage as one of the factors considered. A particularly useful combination is determining the relationship between dosage and LET.

[0023] Alternatively, three or more factors can be considered together in a more detailed manner, as described below.

[0024] The method preferably includes a step of visualizing the relationship between two or more measures on a user's display. For bivariate analysis, this step can be performed, for example, in a heatmap or a 2D histogram in the form of a 3D surface.

[0025] Subsequent steps can be performed based on the determined quality. These subsequent steps may include whether the plan should be delivered to the patient, or deciding on improvements to the plan to enhance its quality.

[0026] The method may also include steps of setting upper and / or lower limits for one or two measures and identifying any voxels that fall within these limits for both measures. In this way, combinations of high values ​​for two or more measures that negatively impact each other can be avoided. By applying the method sequentially with limits spanning a range of all possible values, it is possible to organize the information in a way that facilitates further analysis. For example, the data could be organized as a two-dimensional volume histogram, where each bin corresponds to a combination of limits for the two measures, and the bin value corresponds to the number of voxels (and, by extension, the proportion of volume) satisfying these limits in a given region. This process can be repeated for different sets of upper and lower limits for the measures.

[0027] This disclosure also relates to a computer program product comprising a computer-readable code means, which, when executed in a processor, causes the processor to perform the method according to any one of the preceding claims. The code means may be stored in a memory unit such as a non-transitory memory unit.

[0028] This disclosure also relates to a computer system including a processor and a program memory, the program memory holding a computer program product according to the above description for execution in the processor. Attached Figure Description

[0029] The invention will now be described in more detail by way of examples and with reference to the accompanying drawings.

[0030] Figure 1a and Figure 1b The cumulative histogram and the difference histogram are shown respectively.

[0031] Figure 2a , Figure 2b and Figure 2c Different 2D representations of combinations of values ​​are shown.

[0032] Figure 3 This is a flowchart of the method according to the present invention.

[0033] Figure 4 The illustration shows possible uses of the analysis performed according to an embodiment of the present invention.

[0034] Figure 5 This is a schematic diagram of a computer capable of executing the method of the present invention. Detailed Implementation

[0035] As described above, this disclosure relates to correlation values ​​of two or more relevant metrics used to evaluate radiotherapy treatment plans. This is advantageous because it acknowledges that different metrics may influence each other, and therefore the combined effect of two or more metrics can be considered. Preferably, the metrics include two or more of the following:

[0036] • Dosage. The dosage to different areas of the patient is an important factor in the treatment plan. Physical dose or RBE-weighted dose may be considered.

[0037] • Dose rate. The dose rate can also affect the outcome of the treatment. In particular, in some cases, very high dose rates have been found to be advantageous.

[0038] • Linear energy transfer (LET). LET is the amount of energy transferred per unit distance by an ionized particle through the material it passes through. Therefore, high LET is often also associated with increased radiative effects.

[0039] • Relative bioeffectiveness (RBE) is an indicator of the bioeffectiveness of the type of radiation used in the program.

[0040] • Trajectory endpoints are the locations where individual particles deposit their final energy and stop. LET is higher near trajectory endpoints, meaning that a large number of trajectory endpoints in Monte Carlo-based dose calculations implies high LET statistical uncertainty (or Monte Carlo uncertainty). High statistical uncertainty in low-dose regions is tolerable, while high statistical uncertainty in clinically relevant regions may require repeated dose calculations with higher precision. Users may wish to avoid routinely performing this operation as it is time-consuming. For example, clinically relevant regions might be areas with high dose values ​​or other metrics of interest.

[0041] • Alpha / Beta ratio. This ratio indicates an organ's sensitivity to radiation.

[0042] Including dose as a metric is particularly suitable, and preferred embodiments involve bivariate analysis of dose and LET. Other suitable combinations include the following:

[0043] • Dosage and MC uncertainty in MC calculations. The relationship between dosage and uncertainty can be significant, as high uncertainty may be acceptable at low doses, but becomes more problematic at high doses.

[0044] • Dosage and “dirty” dosage. Dirty dosage is the dosage of LET deposited above a certain threshold.

[0045] • The dose calculated using one RBE model and the dose calculated using different RBE models. In this case, the analysis can identify areas where RBE distortion may be problematic, for example, if a simple RBE model is used, and it is also desirable to consider more refined models. This is especially important for organs at risk.

[0046] • Dose rate and oxygenation level. Oxygenation level is important for the flash effect and should be considered in particular for organs at risk.

[0047] Figure 1a This is a schematic example of a cumulative dose-volume histogram (DVH), where volume is along the Y-axis and dose is along the X-axis. The curve indicates the fraction of the volume that has received at least the corresponding dose. It can be seen that 100% of the volume has received at least 0 Gy, and the volume decreases as the dose increases.

[0048] Figure 1b This is a schematic example of differential DVH, where the volume is along the Y-axis and the dose is along the X-axis. In this case, the dose is divided into dose bins, each 10 Gy, and the histogram shows the fraction of the volume that has received the dose within each bin.

[0049] As will be understood, the same type of histogram can be obtained for other measures, such as LET. Each histogram is one-dimensional and displays volume data for a single measure. It should be noted that such histograms only show the fraction or size of the volume, not the actual distribution within the volume, i.e., which part of the volume has received which dose.

[0050] Figure 2a This is an example of a simplified cumulative 2D histogram, designed to display a fraction of the volume for each combination of x and y values. In this example, the size of the region is displayed as a series of patterns; in a practical implementation, color gradations would typically be used. The blocks on the right side of the graph indicate the region size corresponding to each pattern. It can be seen that combinations of x and y values ​​applied to large volumes or portions of the region of interest have brighter patterns, and decreasing fractions are matched by progressively darker patterns. Specifically, almost 100% of the volume has values ​​for both dose and LET, which are above 5 Gy and 20 keV / μm, respectively, while very small fractions of the volume have values ​​close to the maximum values ​​for both dose and LET.

[0051] exist Figure 2a In the example shown, the dose is displayed along the Y-axis and LET along the X-axis, but this type of graph can be provided for any combination of the two variables.

[0052] Figure 2b This is a simplified example of a difference 2D histogram, designed to display the size of a volume with values ​​representing the dose on the X-axis and the LET on the Y-axis, within specific ranges, e.g., x in the range [a, a') and y in the range [b, b'). In this example, as... Figure 2a As shown, the volume size within each range is displayed as a pattern sequence; in practical implementations, color gradations are typically used. The blocks on the right side of the figure indicate the volume size corresponding to each pattern. It can be seen that the regions with both high dose and high LET are relatively small.

[0053] Figure 2c This is a different example of a cumulative 2D histogram, where a rotatable 3D surface is used to indicate the size or fraction of a region. The height of the 3D surface at any (x, y) point represents the size of the region with the combination of x and y values. Difference 2D histograms can also be displayed as surfaces. How to do this is known in itself.

[0054] Figure 2a , Figure 2b and Figure 2cEach figure in is of a well-known type, and in this case helps a user identify whether there is a large portion of volume in which both of the two shown metrics have unfavorable values, for example, both dose and LET are high.

[0055] Figure 3 is a flowchart of an overall method according to an embodiment of the present invention. The input data S31 of the method includes a treatment plan that has been obtained in a suitable manner, for example by optimization. In step S32, simulation of the plan is performed to determine a predicted outcome of the delivery of the plan, the predicted outcome including values of at least a first metric and a second metric. In step S33, the values of at least the first metric and the second metric are obtained from the predicted outcome, and in step S34, the metrics are analyzed by a multivariate analysis method. If two different metrics are used, the multivariate analysis is bivariate analysis. In an optional step S35, the result of the multivariate analysis is displayed, for example in the form of one or more figures, as Figure 2a , 2b and 2c illustrate.

[0056] In step S36, the result of the analysis is used to determine the quality of the plan, which plan can in turn be used in step S37 to determine a subsequent action. If the quality is determined to be sufficiently high, the action may be delivery of the plan to an intended patient, or discarding the plan and obtaining a new plan. The action may also be improving an existing plan, and possibly re-evaluating the improved plan.

[0057] The display of the metrics in the figures as described above is preferably enhanced by the possibility of selecting one or more regions in the figure to identify corresponding voxels in the dose view. This means that an operator is able, for example, to select a region indicating an unfavorable combination of metrics in the figure, and the system will identify voxels having this combination in the dose view. Figure 4 Taking a heat map as an example, this function is visualized. Figure 4 on the left in is the Figure 2a 2D heat map shown. An operator has identified a region in the heat map where a<A<a' and b<B<b'. Figure 4 on the right in is a corresponding dose view, in which only voxels that meet this condition are shown. This enables the user to identify the region in the patient's body that will have the unfavorable combination of metrics in this example, which makes it easier to assess the importance of this to the plan quality. Although the visualization shown in Figure 4 is helpful for an operator, the graph-based analysis can be performed without creating a graph, and can be used for automatic assessment by the system.

[0058] It is worth noting that for a cumulative view of 2D-VH, taking a "cross section" of the histogram along the A axis at B=0 will result in a regular one-dimensional volume histogram of metric A (i.e., regular DVH if the metric is dose). Similarly, taking a "cross section" along the B axis at A=a' results in a one-dimensional volume histogram of metric B, which is constrained by A>a' (e.g., for LET-VH with dose>a'). Preferably, the system should allow the user to take such a cross section (shown as a regular one-dimensional histogram) at any point of interest. If a one-dimensional volume histogram is selected, these one-dimensional volume histograms can be plotted on the same graph as a three-dimensional surface.

[0059] Consider plotting more than two measures, and multivariate analysis may also be performed. Cutouts for one or more additional relevant measures can be introduced, and then 2D-VH and any “cross-sections” of interest can be plotted, considering only voxels through these cutouts. For example, one might want to study 2D-VH for dose and LET for only certain regions of alpha / beta.

[0060] It should also be possible to simultaneously render multiple 3D surfaces, for example, where different 2D-VH surfaces may originate from:

[0061] • Different regions are selected through cutouts related to a third metric (e.g., high, medium, and low alpha / beta).

[0062] • Display 2D-VH surfaces only within a specific range of values ​​for the third metric.

[0063] • Different plans that users want to compare.

[0064] • In the context of robust assessment, different scenarios allow users to visualize the range of deformations between them.

[0065] The visibility of different surfaces can be turned on / off for easier inspection.

[0066] It should also be possible to extract clinical goals from 2D-VH. In the examples of dose and LET-2D-VH, some samples of such possible clinical goals might be:

[0067] • First, determine the volumetric dose level (e.g., D2, D50, D98), and then set clinical targets for LET for doses above / below that dose level.

[0068] • The requirement that the maximum volume percentage exceeding certain thresholds of dose and LET is allowed, for example, the maximum X% of volume may be higher than both dose>a and LET>b.

[0069] • Requirements for maximum dose and LET at a certain percentage of volume, such as maximum X dose and maximum Y LET at v% of volume.

[0070] This system can be deployed to determine whether each objective has been achieved.

[0071] Based on a bivariate analysis involving dose and LET, an example of a method for evaluating the plan will be described below: First, the user selects a low dose threshold below which LET is not considered. Then, the user selects a limit on the maximum acceptable LET. In response, the system displays the doses of voxels whose doses are above the dose threshold and whose LETs are also above the limit. The system returns the volumes corresponding to the voxels whose values ​​are above both the threshold and the limit, or these voxels are indicated in an image in an appropriate manner, such as by using color. Any combination of the measures listed above can be used instead of dose and LET, including the following:

[0072] • The dosage of RBE factor and at-risk organs can be considered, as the combination of high RBE factor and high dosage will result in high biological dose, which is undesirable in at-risk organs.

[0073] • High dose rates and doses to high-risk organs can be considered for FLASH therapy to ensure that the dose rate is high enough to achieve the FLASH effect.

[0074] • High EPD gamma and high doses in organs at risk. High deviation combined with high dose is undesirable. If the dose is low, the gamma value is less important.

[0075] Figure 5 This is a schematic overview of a computer system in which the optimized version of the invention can be executed. Computer 41 includes a processor 43, data memory 45, and program memory 46. Preferably, one or more user input devices 48, 49 are also present in the form of a keyboard, mouse, joystick, voice recognition device, and / or any other available user input device. The user input devices may also be arranged to receive data from external memory units. The data memory includes relevant metrics and / or input data required to calculate such values. This typically includes patient data and characteristics of the delivery machine. The program memory includes computer programs arranged to perform methods according to embodiments of the invention, such as those in conjunction with… Figure 3 As is common in the art, one or more data memories and one or more program memories may be used, and all or some of the data and / or program may be found in an external unit and obtained as needed via conventional communication devices.

Claims

1. A computer-based method for evaluating radiotherapy treatment plans, comprising: The result of simulating (S32) the delivery of the plan, From the simulation, obtain (S33) at least the values ​​of the first and second metrics. Create a histogram representing the combinations of the values ​​of at least the first and second metrics. Perform multivariate analysis of the values ​​described in (S34) based on histogram values, and The quality of the plan is determined based on the results of the multivariate analysis (S36).

2. The method according to claim 1, wherein, The at least first metric and the second metric include at least one of the following: a. dosage; b. LET; c. Trajectory endpoints; d. RBE; E. Alpha / Beta ratio; F. Statistical uncertainty.

3. The method according to claim 1 or 2, wherein, The values ​​of the first and second measures are obtained, and the multivariate analysis is a bivariate analysis.

4. The method according to any one of the preceding claims, wherein, One of the first and second metrics is dose-related.

5. The method according to any one of the preceding claims, comprising a subsequent step based on the determined quality determination (S37), said subsequent step being one of the following steps. Decide whether to deliver the plan to the patient. Decide on improvements to the proposed plan.

6. The method according to any one of the preceding claims, comprising the following steps: Define a first interval for one metric and a second interval for another metric, wherein the first interval has a first upper limit and / or a first lower limit, and the second interval has a second lower limit and / or a second upper limit, and identify any voxels that have a metric located within both the first interval and the second interval.

7. The method according to claim 6, wherein, The step of identifying voxels is repeated for different values ​​of one or more of the first upper limit and the first lower limit and the second upper limit and the second lower limit.

8. The method of claim 6, further comprising the step of visualizing the identified voxels in an image.

9. A computer program product including a computer-readable code means, which, when executed in a processor, causes the processor to perform the method according to any one of the preceding claims.

10. A computer system (41) comprising a processor (43) and a program memory (46) wherein the program memory holds a computer program product according to claim 9 for execution in the processor (43).

Images