WO2016029917A1

WO2016029917A1 - Method for real-time dose reconstruction during radiotherapy

Info

Publication number: WO2016029917A1
Application number: PCT/DK2015/050251
Authority: WO
Inventors: Thomas RAVKILDE; Per Rugaard Poulsen; Paul Keall
Original assignee: Aarhus Universitet; The University Of Sydney
Priority date: 2014-08-29
Filing date: 2015-08-28
Publication date: 2016-03-03

Abstract

A method for real-time dose reconstruction during radiotherapy or particle therapy, such as proton therapy, from a radiotherapy/particle therapy beam source, comprising the steps: a) selecting one or more points in the space exposed to radiotherapy/particle therapy; b) projecting at least one of the points to at least one plane intersected by the central axis of a radiotherapy/particle therapy beam projecting from the radiotherapy/particle therapy beam source; c) calculating dose in the plane(s) generated by the radiotherapy/particle therapy beam; d) depth scaling the dose in the plane from the projected points to the selected points, thereby obtaining delivered doses in the selected points; and e) iteratively repeating the above steps b)-d).

Description

Method for real-time dose reconstruction during radiotherapy

The present invention relates to a computer implemented method for real-time dose reconstruction during radiotherapy or particle therapy and a method for radiotherapy or particle therapy treatment using the method for real-time dose reconstruction. The invention allows dosimetric errors during radiotherapy/particle therapy to be detected and quantified in real-time and enables adaptation methods to ameliorate the impact of these errors. Background of invention

Radiotherapy is therapy using ionizing radiation, generally as part of cancer treatment to control or kill malignant cells. Radiotherapy is commonly applied to the cancerous tumor because of its ability to control cell growth. Half of all cancer patients are treated with radiotherapy. In radiotherapy, precise delivery of the treatment dose is crucial in order to maximize the ratio between tumor dose and normal tissue dose to effectively cure the patient with minimal side effects. Dose reconstruction is the process of estimating radiation doses received by objects or individuals. During radiotherapy the targeted object and the surrounding areas may move, rotate or deform for different reasons. As an example lung tumors may move with respiration. It should be noted that some motions/rotations/deformations are difficult to predict, and thus difficult to model in advance. In current radiotherapy of moving tumors/organs large volumes of healthy tissue are irradiated in order to ensure adequate treatment of the tumor. Organ motion during treatment delivery can lead to deterioration of the dose distribution. Therefore, an active research topic in recent years has been organ/tumor motion management. There are different concepts for managing the dose delivery based on organ/tumor motion. For example, multileaf collimator (MLC) tracking is a method that adapts the MLC aperture based on the location of the object to be treated.

The inclusion of motion/rotation/deformation in radiotherapy severely challenges the quality assurance (QA) protocols, including dose reconstruction, for the delivered doses. There are algorithms available that are capable of precise dose reconstruction, however, these algorithms are either too slow to be used in real-time due to computation time and/or are not capable of taking into account

motion/rotation/deformation of the targeted object.

In one example, a pencil beam convolution based dose absorption algorithm by Storchi and collaborators (Storchi et al 1999, Storchi and Woudstra 1996, 1995), the dose in a voxelized water equivalent block is calculated in at least five planes at different depths by convolution of single pencil beam 2D scatter kernels with a normalized fluence, and converted to other depths by interpolation of percentage depth dose along the ray lines and by inverse square law scaling.

Summary of invention

The present invention relates to a method for real-time dose reconstruction during radiotherapy. A computer implemented version of the present invention is sufficiently fast to calculate radiation doses for an object exposed to radiotherapy, while taking into account motion/rotation/deformation of the object. The inventors have realized that by selecting a number of points in space, for which the dose is calculated, rather than calculating the dose to a full volume (3D grid of voxels) and by calculating the dose distribution to a limited (1 , 2, 3 or 4) number of planes, rather than at least five planes, real-time dose reconstruction can be achieved. Since no ray tracing is needed in the method, and the amount of convolutions is limited such that the convolutions can be handled in real-time by the computer implemented method, real-time dose

reconstruction for moving/rotating/deforming anatomy can be achieved. The method comprises the steps: a) selecting one or more points; b) projecting at least one of the points to at least one plane intersected by the central axis of a radiotherapy beam; c) calculating dose in the plane generated by the radiotherapy beam projecting from the radiotherapy beam source; d) depth scaling the dose in the plane from the projected points to the selected points, thereby obtaining delivered doses in the selected points; and e) iteratively repeating the above steps b)-d). A consequence of performing the abovementioned steps is that dose reconstruction can be performed in real-time for moving/rotating/deforming anatomy.

As stated above dose reconstruction is the process of estimating radiation doses received by objects or individuals. However, dose reconstruction can also be forward looking, i.e. using past data and estimating future data to estimate dose reconstruction for a treatment. Dose reconstruction can be performed during a radiotherapy session to ensure that a correct dose is delivered, or for verification after a session, or modeled in advance. It should be noted that although the method is capable of real-time dose reconstruction it does not necessarily have to be applied real-time. It can also be used prior to a treatment, and/or after a treatment based on collected data.

The selected points may be defined by the user. By only selecting points having particular relevance for the application of the radiotherapy, calculations are reduced and still a sufficiently accurate delivered dose can be extracted. The presently disclosed invention includes the possibility to adjust the position of the calculation points individually during the radiotherapy, enabling efficient mimicking of motion, rotation and/or deformation. In the calculation, the doses for a number of selected points can be based on the calculated doses in 1 , 2, 3, or 4 planes. Thus, introducing additional points is possible without affecting the convolutions that are responsible for calculating the plane doses.

The real-time dose reconstruction is particularly suitable for comparison against a planned dose and real-time verification. One advantage with the presently disclosed method is that for a radiotherapy treatment plan, including a planned delivery dose for an area or object, real-time motion-induced dose error reconstruction is possible. In one embodiment of the present invention the dose and dose errors are monitored in real-time during radiotherapy to improve patient safety by intervention to avoid erroneous dose delivery. The dose reconstruction calculations may also be used to gate the radiotherapy dose for moving targets or assist in choosing best MLC apertures based on already delivered doses or to calculate dose volume histograms (DVHs).

Description of Drawings The invention will in the following be described in greater detail with reference to the drawings. The drawings are exemplary and are intended to illustrate some of the features of the present method and unit and are not to be construed as limiting to the presently disclosed invention. Fig. 1 shows an example of a schematic representation of a model of radiotherapy of an object.

Fig. 2 shows a comparison of reconstructed and measured dose for volumetric modulated arc therapy (VMAT) of a moving object.

Fig. 3-4 show comparisons of reconstructed and measured 3%/3 mm γ failure rates for VMAT of moving objects.

Detailed description of the invention As stated the presently disclosed invention relates to a method for real-time dose reconstruction during radiotherapy or particle therapy from a radiotherapy/particle therapy beam source. The method comprises the steps: a) selecting one or more points in the space exposed to radiotherapy/particle therapy; b) projecting at least one of the points to at least one plane intersected by the central axis of a

radiotherapy/particle therapy beam projecting from the radiotherapy/particle therapy beam source; c) calculating dose in the plane(s) generated by the radiotherapy/particle therapy beam; d) depth scaling the dose in the plane from the projected points to the selected points, thereby obtaining delivered doses in the selected points; and e) iteratively repeating the above steps b)-d). By iteratively repeating steps b)-d) the dose is constructed in real-time. The method is preferably computer implemented.

The presently disclosed method for real-time dose reconstruction can be construed broadly as covering both radiotherapy and particle therapy, such as proton therapy. Therefore, in one embodiment of the presently disclosed invention, the dose reconstruction method is used for reconstruction of a delivered dose of particles in particle therapy, such as proton therapy. Particle therapy uses a beam of protons or ions (carbon, oxygen, etc.) instead of photons as in regular radiotherapy. Therefore, one embodiment the presently disclosed method relates to a method for real-time dose reconstruction during particle therapy, such as proton therapy, from a particle therapy beam source, comprising the steps: a) selecting one or more points in the space exposed to particle therapy; b) projecting at least one of the points to at least one plane intersected by the central axis of a particle therapy beam projecting from the particle therapy beam source; c) calculating dose in the plane(s) generated by the particle therapy beam; d) depth scaling the dose in the plane from the projected points to the selected points, thereby obtaining delivered doses in the selected points; and d) iteratively repeating the above steps b)-d). The described sub-features and details explained in the present disclosure can be considered to cover both radiotherapy and particle therapy.

Selection of points

By selecting one or more points in space (step a) in the method) rather than calculating the dose to a full volume (3D grid of voxels), the number of calculations can be restricted substantially. By projecting the selected points onto a plane intersected by the central axis of the beam, interpolation along the beam lines during ray tracing is avoided. By utilizing the fact that the target of the radiotherapy can be regarded as homogeneous during QA, the inventors have realized that no ray tracing for inhomogeneity correction is needed to calculate the dose. This assumption of homogeneity is in line with common QA procedures. As a consequence, the depth scaling step becomes less computationally intensive for the calculations of the dose in the selected points. Fig. 1 shows an example of a schematic representation of a model of radiotherapy of an object. In fig.1 the calculation point with coordinates (x,y,z) is a point for which the dose can be calculated according to the presently disclosed method. The idea is that by choosing the points in step a) such that they cover specific points, dose reconstruction values can be obtained, which can be used to provide sufficiently accurate estimations of dose errors while restricting calculations

significantly. The points may be selected adjacent to or inside an object that is the target of the radiotherapy, or some other region of interest, e.g. an organ at risk of too high dose, but the points could also be basically any points that could be exposed to radiation.

The selected points may be selected arbitrarily. It is also possible to indicate the motion and positions of the selected points. In one embodiment of the present invention transponders indicate the motion of the selected points, and, more specifically, in one embodiment the transponders are electromagnetic. Transponders in general and electromagnetic transponders in particular may provide precise information about the current positions of the selected points with negligible delay. However, other possible means for indicating the positions of the selected points can be imagined. Alternatively, electronic images and/or x-ray images or any other image modality may also indicate the location of the selected points.

A further aspect of the invention relates to the position of the individual points being adjusted individually during the radiotherapy. In the scenario of e.g. a deforming target, comprising points that indicate the position of a number of parts of the object, the individual distances between the points are changed automatically. However, an additional option is to change the distances between the points actively e.g. for mimicking motion, rotation and/or deformation.

Plane(s)

In step b) the points are projected to at least one plane intersected by the central axis of a radiotherapy beam. As mentioned above one advantage related the projection to (a) plane(s) is that no ray tracing is needed to calculate the dose. In fig.1 this corresponds to point (x,y,z) being projected to a plane, resulting in the projected point {xp' , y_p, z_iso). The projection of the point on the plane can be expressed as

(x_p' , y_p, z_iso) = (^x^^_> y^¾_> ¾₀) where the source axis distance SAD in fig. 1 is the distance between the beam source and the plane and the source-surface distance SSD in fig.1 is the distance between the source and the targeted object.

One aspect of the presently disclosed invention relates to projecting the selected points to 1 , 2, 3 or 4 planes. By limiting the number of planes, the convolutions of electron scatter kernels, such as single pencil beam 2D scatter kernels according to the Storchi algorithm, which are computational intensive, are limited. In one embodiment, only one plane intersected by the central axis of the radiotherapy beam is used. The inventors have realized that this limitation does not necessarily imply much loss of accuracy in the dose calculations, or at least the calculations may still be used for a number of purposes, but the computation times are reduced enough to be able to re-calculate the doses continuously in real-time such that it can handle relevant scenarios of moving targets of the radiotherapy, i.e. calculating the doses with a certain degree of accuracy.

In one embodiment of the invention at least one of the plane(s) intersected by the central axis of a radiotherapy beam is a plane perpendicular to the central axis of the radiotherapy beam. In one embodiment of the invention, the plane is the isoplane. The isocenter in this context requires a circular rotation of a gantry around the target of the radiotherapy. The isocenter is defined as the point in the center of the circle. The isoplane is defined as the plane perpendicular to the central axis of the radiotherapy beam, wherein the isocenter is part of the plane. One advantage of using the isoplane as the plane to which the selected points are projected is that it has a fixed distance to the beam source. In a further embodiment, the plane is a plane perpendicular to the central axis of the radiotherapy beam, wherein a specific point in the target is part of the plane. In this embodiment the distance to the beam source is not fixed but the lateral electron scatter computed by convolution of the plane with the scatter is more accurate.

In one embodiment of the presently disclosed method, the depth scaling for at least two selected points is based on the calculated dose in 1 , 2, 3 or 4 planes. This means that step c) in the method, calculating (a) dose(s) in the plane(s) generated by the radiotherapy beam, only has to be performed once for every complete iteration of steps b)-d). Step d), depth scaling the dose in the plane from the projected points to the selected points, thereby obtaining delivered doses in the selected points, may then be calculated for all selected points based on the same plane calculations, which saves computation time compared to executing steps b)-d) for each selected point. A consequence of this approach is that more points can be handled within a certain predefined time; hence spatial resolution of the dose reconstruction is increased.

Real-time reconstruction

Real-time in the present invention has the meaning that a dose reconstruction should be able to finish fast enough to be repeated such that dose reconstruction can be performed with sufficient accuracy and sufficiently high time resolution. Typically steps b)-d) are repeated continuously or based on events; however the points may also be changed (step a), selecting points) during radiotherapy. The time that should be considered real-time in this context depends on e.g. how fast the target moves. A typical scenario of a moving target is a tumor moving with the respiration of the patient. Typical human resting respirations are 10 to 18 per minute, which provides an example of a reference to take into account when deciding how frequently the calculations have to be formed. In one embodiment of the present invention, steps b)-d) are iteratively repeated within a time interval of between 0.1 milliseconds and 5 seconds, such as between 0.1 ms and 1 ms, or between 0.1 ms and 100 ms, or between 0.1 ms and 300 ms, or between 0.1 ms and 500 ms, or between 0.1 ms and 1 s, or between 0.1 ms and 3 s, for example 0.1 ms, 0.2 ms, 0.3 ms, 0.4 ms, 0.5 ms, 0.6 ms, 0.7 ms, 0.8 ms, 0.9 ms, 1 ms, 2 ms, 3 ms, 4 ms, 5ms, 10 ms, 20 ms, 30 ms, 40 ms, 50 ms, 100 ms, 200 ms, 300 ms, 400 ms, 500 ms, 1 s, 2 s, 3 s, 4 s or 5 s. It should be noted that real-time in this context is not necessarily a fixed number, but rather a consideration of which update frequency that is needed in order to obtain sufficiently accurate dose reconstruction for a moving/rotating/deforming target.

In one embodiment of the present invention, the execution of steps b)-d) is triggered by an event. In another embodiment of the present invention, the event is based on patient motion and/or generated by a change in the radiotherapy equipment and/or an indication that the previous dose reconstruction has finished. For example, after one round of calculation has finished, it is possible to check whether new measurements of the positions of the selected points, or the positions of the MLC leaves, are available. Another possible approach is to base the triggering of the execution of steps b)-d) on for example sudden motion of the target of the radiotherapy. The event that triggers the execution of steps b)-d) may also be manual interaction e.g. that an operator of the radiotherapy system pushes a button, or any other external signal. Alternatively, in a further embodiment, a temporal resolution or a minimum temporal resolution of the execution of steps b)-d) is defined by the user. In such an embodiment the user can for example set the minimum resolution to be 500 ms. This means that if the program is capable of calculating doses in 100 ms, the intervals of calculated doses will be 100 ms. If the program is only capable of calculating doses in 800 ms, the intervals of calculated doses will be 500 ms, i.e. the user defined minimum resolution. This ensures a desired temporal resolution at the expense of an increased latency of the results. The resolution may be dynamic in the sense that it adapts to the performance of the program in terms of actual doses calculation time.

Real-time may also refer to being capable of finishing calculations in time. If the dose reconstruction should be used as a means to warn for areas being exposed to unhealthy levels of radiation or assisting in dose adjustments based on the calculations, it implies requirements on the dose reconstruction computation time. If the reconstruction cannot be performed in real-time, there is a risk of exposure of unhealthy levels of radiation due to machine errors or human errors. Real-time dose reconstruction may be used to interrupt the radiation if such exposure is observed.

In one embodiment the invention uses one or more application specific integrated circuits (ASIC) or field-programmable gate arrays (FPGA) or graphical processing unit (GPU) to calculate the dose in the plane and/or the absolute depth dose and/or the dose delivered in the selected points. By using these types of circuits, computation times can be further improved. This means that the method is capable of handling even faster motion of a target by re-calculating the dose more frequently and/or add accuracy in the calculations e.g. by adding more points or more planes.

Dose calculation in a plane

Step c) according to the presently disclosed method is the step: calculating (a) dose(s) in the plane(s) generated by the radiotherapy beam. This step includes convolution of the normalized fluence with the scatter kernel in the plane. In the presently disclosed invention the calculated dose in the plane may be a relative dose, meaning the calculated value is dimensionless. Furthermore, typically, but not necessarily, the dose in the plane is calculated as a function of time.

The dose in the plane may be calculated by several methods. As an example it may be calculated for each point in time by convolution of a field intensity function with a pencil beam kernel, where the dose is expressed as R{x_p, y_p, z_iso; t) = F(x_p, y_p; t) *

K(x_p, y_p), where F(x_p, y_p; t) is the field intensity function and K{x_p, y_p) is the kernel matrix.The field intensity function describes the field shape and blocking in the field at time t. An example of values in a field intensity function, using a multileaf collimator, is 1 in an open beam, 0 behind collimator jaws, and 0.019 for points shielded only by MLC leaves from a 6 megavolt radiation beam.

The kernel matrix is designed to yield the lateral electron scatter of the approaching radiation beam. In one embodiment the kernel matrix is constructed from a rotation symmetric single pencil beam kernel. The inventors have realized that by truncating the extent of the scatter kernel, significantly faster computation of the doses can be achieved. The idea is that the kernel is truncated such that there is only a minor loss of doses in the calculation while saving a significant amount of computations. For example, the radius of a rotation symmetric scatter kernel, such as a single pencil beam, may be truncated to for example 2.5 cm, 2.6 cm, 2.7 cm, 2.8 cm, 2.9 cm, 3.0 cm, 3.2 cm, 3.5 cm, 4 cm, 5 cm, 6 cm, 7 cm, 8 cm, 9 cm, 10 cm, 12 cm, 14 cm, 16 cm, 18 cm, or 20 cm or any other suitable radius.

Depth scaling Step d) according to the presently disclosed method is the depth scaling of the dose in the plane from the projected points to the selected points, thereby obtaining delivered doses in the selected points.

In one embodiment the depth scaling is performed by executing the steps:

- calculating absolute depth dose on the central axis of the radiotherapy beam as a function of the distance from the beam source;

calculating the dose delivered in the selected points based on the dose in the plane and the absolute depth dose.

The central axis of the radiotherapy beam in fig. 1 can serve as an example of a central axis for this purpose.

By using a function (of the distance z from the beam source) for the absolute depth dose on the central axis of the radiotherapy beam, continuous dose interpolation along the beam lines (ray tracing) is avoided, which saves a significant amount of

computation time. In a further embodiment, interpolation is performed at discrete points along the beam lines for increased accuracy with negligible performance penalty.

Specifically, the absolute depth dose on the central axis of the radiotherapy beam may be calculated based on a measured percentage depth dose. In one embodiment the absolute depth dose is calculated based on a single measured percentage depth dose in water. Alternatively, the absolute depth dose function can be calculated based on a set of percentage depth doses, which increases the accuracy of the disclosed method compared to using a single measured percentage depth dose. More specifically, in one embodiment, the absolute depth dose on the central axis of the radiotherapy beam is expressed as D_a(z; t) = PDD_norm(z)^■ c_M(z; t) ^■ Cj (t) where c_M(z; t) is Mayneord's F factor and Cj(t) is an inverse square law correction factor. Preferably, the absolute depth dose on the central axis of the radiotherapy beam is calculated as a function of time. The dose delivered in the selected points can then be calculated based on the absolute depth dose function and the calculated dose in the plane. The absolute depth dose function is typically expressed as a function of time and the distance from the beam source. In one embodiment the dose delivered in the selected points is calculated as D(x, y, z; t) = R{x_p' , yp, z_iso; t) ^■ D_a(z; t)^■ D dt ^■ c_d(z) ^■ c₀ (optional) where

R(x_p, y_p, z_iso; t) is the dose in the plane, D_a(z; t) is the absolute depth dose on the central axis of the radiotherapy beam, D is the dose rate, c_d(z) is a density correction factor, and c₉ is a gantry dependent attenuation correction factor (wherein c₀ is an optional parameter). Application/system

The present invention can be used in a number of situations. As mentioned the disclosed method is particularly useful for a moving target in radiotherapy. In one embodiment of the invention the target of the radiotherapy is a tumor. In alternative embodiments the target of the radiotherapy may be any biological structure or tissue.

A further aspect of the invention relates to the calculation of dose errors (difference between planned dose and delivered dose). The error in dose is typically the quantity of interest during irradiation. The dose error can be calculated with sufficient accuracy by computing differences between planned dose and delivered dose at any point in time. The inventors have realized that in certain cases the accuracy of the absolute doses may be of less importance than the difference between planned and delivered dose. Therefore, one embodiment of the invention further comprises the step of verifying the calculated doses and/or the accumulated calculated doses in the selected points against planned treatment doses.

In some scenarios it may also be useful to calculate the failure rate of the dose in a number of selected points. If the difference between the planned dose and the delivered dose for a point is greater than a predefined value the delivered dose for the point is considered to be failing. Furthermore, in some applications it is useful to accumulate the calculated doses in the selected points and, possibly, also calculate the error of such accumulated doses.

In one embodiment of the invention, the delivered dose is calculated for more than one radiotherapy beam source. The beam sources may move independently of each other. In this scenario the dose has to be calculated from the two or more beam sources separately. The selected points can be the same for the beam sources, thereby enabling the possibility to add the dose from the different beam sources. However, steps b)-d) have to be performed individually for each beam source.

Not only the target of the radiotherapy may be moving, but also the beam source. Therefore, in the presently disclosed method the beam source may be applied from different angles and/or moving. A special case of a moving beam source is the beam source rotating around the targeted object in a circle. The beam source may also rotate around the targeted object in other non-circular trajectories. Other objects, e.g. the targeted object or other objects in regions of interest, may also move. One example of the target and other objects moving is couch tracking, i.e. using a supporting couch to track a moving target of radiotherapy. It may be useful to monitor the calculated values in real-time, preferably graphically, to be able to react if for example an undesired dose or failure rate occurs in the selected points. Therefore, in one embodiment of the present invention the delivered doses and/or cumulated delivered doses and/or dose errors (both of transient and cumulative doses) and/or failure rates (both of transient and cumulative doses) are monitored in real-time. Other applications of the method are during simulation of radiotherapy and for testing and verifying the functioning and correctness of radiotherapy equipment. For example, the method can be used in simulation to evaluate a treatment plan or to evaluate how well the tracking performs, or would perform for a given target motion and treatment plan. The method may also be used for independent verification of a planned dose. The method does not necessarily have to be applied on radiotherapy of a real target.

The present disclosure furthermore relates to a system comprising means for executing the abovementioned method for real-time dose reconstruction during radiotherapy. Such a system may be a radiotherapy system, and, since the method is computer implemented, it may comprise a non-transitive, computer-readable storage device for storing instructions that, when executed by a processor, performs the method for realtime dose reconstruction during radiotherapy. The radiotherapy system may comprise a multileaf collimator to provide conformal shaping of the radiotherapy beam, and the MLC aperture may be continuously adapted based on the position of the targeted object (MLC motion tracking). During MLC tracking there is a risk that optimal leaf adaptation is not achieved. The

abovementioned method for real-time dose reconstruction can be used to highlight dose distribution errors.

Method for radiotherapy treatment

A further aspect of the presently disclosed invention relates to a method for radiotherapy treatment, comprising the steps: treating a patient with radiotherapy; reconstructing the delivered dose in selected points in real-time using the method as described above. As mentioned, the method for real-time dose reconstruction during radiotherapy can be used for several purposes. In one scenario a patient is treated with radiotherapy. During the radiotherapy session, the method for real-time dose reconstruction is used continuously or during parts of the treatment. Preferably the reconstructed doses in the selected points are compared against a planned treatment dose. This enables the possibility of e.g. monitoring of reconstructed delivered dose in real-time. Since the method is capable of taking into account both moving targets and moving beam sources, in one embodiment the radiotherapy is performed using volumetric modulated arc therapy (VMAT). VMAT is a radiotherapy technique in which the beam source rotates around the target in single or series of arcs, thereby generating a beam from constantly changing angles. This method is said to precisely conform the radiation and reduce treatment time.

The method for real-time dose reconstruction during radiotherapy may also be used to gate the radiotherapy dose based on the reconstructed delivered dose. Furthermore, the dose reconstruction calculations may also be used to gate the radiotherapy dose for moving targets or assist in choosing best MLC apertures based on already delivered doses or to calculate dose volume histograms (DVHs).

As mentioned above, the points for which doses are reconstructed may be adjacent to or inside the object that is the target of the radiotherapy. For radiotherapy of a real patient, this means that the selected points may be located in the patient.

Examples

Experiments have been conducted to test the performance of the method. The experiments were conducted using a standard linear accelerator with prototype MLC tracking software guided by an electromagnetic transponder system. Radiotherapy treatment plans were delivered to a phantom. A three-axis motion stage reproduced eight representative tumor trajectories; four lung and four prostate. Low and high modulation 6 MV single-arc VMAT treatment plans were delivered for the different trajectories of the phantom with and without MLC tracking, as well as without motion, for reference. Temporally resolved doses received by the phantom were measured during all treatments using a biplanar dosimeter. The dose delivered to each of 1069 diodes in the dosimeter was reconstructed with 500 ms temporal resolution. The accuracy of the method for real-time dose reconstruction deliveries was quantified. The results have also been confirmed using a dynamic temporal resolution with an average resolution of approximately 180 ms, executed in real-time. Example 1 - Summary of dose reconstruction deviations

Table 1 summarizes the dose deviations comparing reconstructed and measured time- resolved dose distributions. The values are averages for all trajectories, diodes, and time points. The values in the tables are mean dose deviations and, within

parentheses, standard deviations.

Low/high modulation refers to the intensity modulation complexity for the MLC. Low modulation means that there are restrictions on how the positions of the individual leaves of the MLC may change, whereas with high modulation there is more freedom for changing the positions of the leaves. Plan Static No tracking Tracking All

Transient

Low modulation 4.8 (5.8) 4.8 (5.9) 4.8 (5.9)

4.5 (6.7)

High modulation 3.9 (5.9) 3.9 (5.9) 3.9 (6.0)

Cumulative

Low modulation 3.8 (2.6) 3.9 (2.7) 3.9 (2.7)

3.9 (3.9)

High modulation 3.3 (3.2) 3.2 (3.1) 3.2 (3.0)

Accumulated

Low modulation 3.5 (2.2) 3.5 (2.1) 3.5 (2.1)

3.9 (3.6)

High modulation 3.6 (2.9) 3.5 (2.7) 3.5 (2.7)

Table 1 : Mean (and standard deviation) of dose deviations (%) comparing reconstructed and measured time-resolved dose distributions.

Table 2 shows similar data using a slightly different metric and summarizes the dose differences comparing reconstructed and measured time-resolved dose distributions.

Plan Static No tracking Tracking All

Transient

Low modulation -0.2 (7.7) -0.2 (8.6) -0.22 (8.0)

-0.6 (8.1)

High modulation -0.8 (8.0) -0.9 (8.0) -0.8 (8.2)

Cumulative

Low modulation 0.2 (6.0) 0.3 (5.9) 0.3 (6.1)

-0.5 (5.5)

High modulation -1.0 (5.8) -1.0 (5.0) -0.9 (5.0)

Accumulated

Low modulation -0.4 (5.4) -0.4 (5.4) -0.4 (5.4)

-1.1 (5.2)

High modulation -1.7 (5.7) -1.8 (4.8) -1.5 (4.8) Table 2: Mean (and standard deviation) of dose differences comparing

reconstructed and measured time-resolved dose distributions. Unit: % of max dose. Example 2 - Comparison of reconstructed and measured dose

Fig. 2 shows measured (thick curves) and reconstructed (thin curves) doses for three sample diodes in the phantom for high modulation VMAT of a lung tumor travelling along the Baseline shifts motion trajectory without MLC tracking; (A) the diode in the center of phantom, (B) a diode in a high dose gradient, and (C) a diode in the periphery of the phantom. Transient doses are shown in the left column and the resulting cumulative doses in the right column.

In this example the mean computation time was 35.9 ms for each transient dose distribution calculation in Matlab (1069 diode points) including overhead on a standard laptop equipped with an Intel® Core™ i5-M460 CPU running at 2.53 GHz.

Example 3 - Motion-induced dose error for lung tumor Fig. 3 shows a comparison of measured (thick curves) and reconstructed (thin curves) 3%/3 mm γ failure rates during the high modulation VMAT deliveries to a lung tumor travelling along each of the lung tumor motion trajectories without (top panel) and with (bottom panel) tracking. The γ failure rate is shown for both transient doses (Rows (A) and (C)) and cumulative doses (Rows (B) and (D)). Note the different y-scales between rows.

Example 4 - Motion-induced dose error for prostate tumor

Fig. 4 shows a comparison of measured (thick curves) and reconstructed (thin curves) 3%/3 mm γ failure rates during the high modulation VMAT deliveries to a prostate tumor travelling along each of the prostate motion trajectories without (top panel) and with (bottom panel) tracking. The γ failure rate is shown for both transient doses (Rows (A) and (C)) and cumulative doses (Rows (B) and (D)). Note the different y-scales between rows.

Claims

A method for real-time dose reconstruction during radiotherapy or particle therapy, such as proton therapy, from a radiotherapy/particle therapy beam source, comprising the steps:

a) selecting one or more points in the space exposed to radiotherapy/particle therapy;

b) projecting at least one of the points to at least one plane intersected by the central axis of a radiotherapy/particle therapy beam projecting from the radiotherapy/particle therapy beam source;

c) calculating dose in the plane(s) generated by the radiotherapy/particle

therapy beam;

d) depth scaling the dose in the plane from the projected points to the selected points, thereby obtaining delivered doses in the selected points; and e) iteratively repeating the above steps b)-d).

The method according to any of the preceding claims, wherein steps b)-d) are iteratively repeated within a time interval of between 0.1 milliseconds and 5 seconds.

The method according to any of the preceding claims, wherein the execution of steps b)-d) is triggered by an event.

The method according to claim 3, wherein the event is based on at least one of patient motion, a change in the radiotherapy equipment, and an indication that the previous dose reconstruction has finished.

The method according to any of the preceding claims, wherein an object is the target of the radiotherapy.

6. The method according to claim 5, wherein the object is a tumor.

7. The method according to claim 5-6, wherein the selected points are adjacent to or inside the object.

8. The method according to any of the preceding claims, wherein the position of the individual points are adjusted individually during the radiotherapy.

The method according to any of the preceding claims, wherein at least one of the planes intersected by the central axis of a radiotherapy beam is a plane perpendicular to the central axis of the radiotherapy beam.

0. The method according to any of the preceding claims, wherein the selected points are projected to 1 , 2, 3 or 4 planes.

1. The method according to any of the preceding claims, wherein the depth scaling for at least two selected points is based on the calculated dose in 1 , 2, 3 or 4 planes.

12. The method according to any of the preceding claims, further comprising the step of accumulating the calculated doses in the selected points.

13. The method according to any of the preceding claims, further comprising the step of verifying the calculated doses and/or the accumulated calculated doses in the selected points against planned treatment doses.

14. The method according to any of the preceding claims, further comprising the step of calculating a failure rate of the selected points, wherein a difference between the calculated dose and the planned dose outside a predefined limit represents a failure for a point.

15. The method according to any of the preceding claims, wherein transponders indicate the motion of the selected points.

16. The method according to claim 15, wherein the transponders are

electromagnetic.

17. The method according to any of the preceding claims, wherein electronic

images such as magnetic resonance imaging (MRI) and/or x-ray images indicate the location of the selected points.

18. The method according to any of the preceding claims, wherein the targeted object and/or additional object(s) is/are moving and/or is rotating and/or is deforming during the radiotherapy.

19. The method according to any of the preceding claims, wherein the delivered dose is calculated for more than one radiotherapy beam source.

20. The method according to any of the preceding claims, wherein the beam source is applied from different angles.

21. The method according to any of the preceding claims, wherein the beam source and/or the targeted object and/or additional objects move(s) such that their relative positions are changed.

22. The method according to any of the preceding claims, wherein the beam source rotates around the targeted object in a circle.

23. The method according to claim 22, wherein the plane is the isoplane.

24. The method according to any of the preceding claims, wherein the projected point on the plane is expressed as (x_p' , y_p, z_iso) = (^x^^7 _> y^^7 _> ¾₀) where the source axis distance SAD is the distance between the beam source and the plane and the source-surface distance SSD is the distance between the source and the targeted object.

25. The method according to any of the preceding claims, wherein the calculated dose in the plane is relative.

26. The method according to any of the preceding claims, wherein the dose in the plane is calculated as a function of time.

27. The method according to any of the preceding claims, wherein the dose in the plane is calculated for each point in time by convolution of a field intensity function with a pencil beam kernel.

28. The method according to claim 27, wherein the radius of the pencil beam kernel is truncated.

29. The method according to any of claims 27-28, wherein the dose in the plane is calculated for each point in time by convolution of a field intensity function with a pencil beam kernel, expressed as R{x_p, y_p, z_iso; t) = F(x_p,y_p; t) * K{x_p, y_p) where F(x_p,y_p; t) is the field intensity function and K{x_p, y_p) is the kernel matrix.

30. The method according to any of claims 27-29, wherein the field intensity function describes the field shape and blocking in the field at time t and takes values of 1 in an open beam, 0 behind collimator jaws, and 0.019 for points shielded only by MLC leaves.

31. The method according to any of claims 27-30, wherein the kernel matrix is constructed from a rotation symmetric single pencil beam kernel.

32. The method according to any of the preceding claims, wherein the depth

scaling is performed by executing the steps:

calculating absolute depth dose on the central axis of the radiotherapy beam as a function of the distance from the beam source;

33. The method according to claim 32, wherein the absolute depth dose on the central axis of the radiotherapy beam is calculated as a function of time.

34. The method according to any of claims 32-33, wherein the absolute depth dose on the central axis of the radiotherapy beam is calculated based on a measured percentage depth dose.

35. The method according to claim 34, wherein the measured percentage depth dose is a single measured percentage depth dose in water.

36. The method according to claim 34, wherein the measured percentage depth dose is a set of percentage depth doses.

37. The method according to any of claims 32-36, wherein the absolute depth dose on the central axis of the radiotherapy beam is expressed as D_a(z; t) = PDD_norm (z) - c_M(z; t) ^■ Cj (t) where c_M(z; t) is Mayneord's F factor and Cj (t) is an inverse square law correction factor.

38. The method according to any of the preceding claims, wherein the dose

delivered in the selected points is calculated based on the dose in the plane and a depth dose based on the distance from the beam source.

39. The method according to any of the preceding claims, wherein the dose delivered in the selected points is calculated as D(x, y, z; t) = R(x_p, y_p, z_iso; t) ^■ D_a(z; t) ^■ D dt ^■ c_d(z) ^■ c₀ (optional) where R(x_p, y_p, z_iso; t) is the dose in the plane, D_a(z; t) is the absolute depth dose on the central axis of the radiotherapy beam, D is the dose rate, c_d(z) is a density correction factor, and c₉ (optional parameter) is a gantry dependent damping correction factor.

40. The method according to any of the preceding claims, wherein the delivered doses and/or cumulated delivered doses and/or dose errors and/or failure rates are monitored in real-time.

41. The method according to any of the preceding claims, wherein the method is applied on simulation of radiotherapy.

42. The method according to any of the preceding claims, wherein the method is used for testing and verifying the functioning and correctness of radiotherapy equipment.

43. The method according to any of the preceding claims, using one or more

application specific integrated circuits (ASIC) or field-programmable gate arrays

(FPGA) or graphical processing unit (GPU) to calculate the dose in the plane and/or the absolute depth dose and/or the dose delivered in the selected points.

44. The method according to any of the preceding claims, wherein the dose

reconstruction is forward looking.

45. A system comprising means for executing the method for real-time dose

reconstruction during radiotherapy according to any of claims 1-43.

46. The system according to claim 45, wherein the system is a radiotherapy

system.

47. The system according to any of claims 45-46, comprising a non-transitive,

computer-readable storage device for storing instructions that, when executed by a processor, performs the method for real-time dose reconstruction during radiotherapy according to any of claims 1-43.

48. The system according to any of claims 45-47, further comprising a multileaf collimator (MLC).

49. The system according to claim 48, wherein the multileaf collimator is configured such that the MLC aperture is continuously adapted based on the position of the targeted object

50. A method for radiotherapy treatment, comprising the steps:

treating a patient with radiotherapy;

reconstructing the delivered dose in selected points in real-time using the method according to claims 1-43.

51. The method according to claim 50, further comprising the step of comparing the reconstructed dose in the selected points against a planned treatment dose.

52. The method according to claim 51 , wherein the radiotherapy treatment is

interrupted based on the comparison of the reconstructed dose and the planned treatment dose.

53. The method according to any of claims 50-52, wherein the radiotherapy is

performed using volumetric modulated arc therapy (VMAT).

54. The method according to any of claims 50-53, wherein the radiotherapy is

assisted by multileaf collimator (MLC) tracking.

55. The method according to any of claims 50-53, wherein the radiotherapy dose is gated based on the reconstructed delivered dose.

56. The method according to any of claims 50-55, wherein the radiotherapy dose is adjusted and/or corrected based on the reconstructed delivered doses.

57. The method according to any of claims 50-56, wherein the selected points are located in the patient.