Patent Application
20230364446
This application claims priority of European patent application no. 22172682.1 filed on May 11, 2022, which is incorporated herein by reference in its entirety.
The present disclosure relates to the general field of treatment of tumoral cells by irradiation with accelerated particles, such as protons. In particular, the present disclosure provides a method for setting the operating parameters of a given particle accelerating system ensuring that the given particle accelerating system can deliver within a predefined confidence level beamlets of accelerated particles fulfilling the requirements of a treatment plan (TP) comprised within an acceptable band of variation (BV). This method is advantageous in that it reduces to within the predefined confidence level the risk of having to suddenly interrupt a patient irradiation session because the given particle accelerating system failed to deliver one or more beamlets satisfying the TP within the acceptable band of variation.
Radiation therapy with particles or waves, such as protons beams, electron beams, heavy ions beams, x-rays, y-rays, and the like, has become an essential tool for treating patients with tumors.
Pencil beam scanning (PBS) is a technique consisting of steering beamlets of charged particles towards a target comprising tumoral cells defining a structure of interest. PBS reduces unnecessary radiation exposure to surrounding non-cancerous cells by shaping the area being treated to mirror the tumor geometry of the structure of interest. Pencil beam scanning can treat a tumor with a single beam composed of various beamlets or with multiple beams of different orientations each composed of various beamlets, sometimes called intensity modulated proton therapy (IMPT). Beside the geometry of the target, PBS allows local tuning of the parameters of the beamlets depending on the position within the target. The parameters can include a position and a monitor (provided as a monitor unit) of each beamlet as well as the scanning sequence of the beamlets, with starting time and end time of each beamlet.
Since both tumoral cells and healthy cells are damaged by such radiations, a major challenge in cancer treatment is to define a treatment plan (TP) ensuring that the tumoral cells are effectively destroyed or killed, while sparing as much as possible the healthy cells, in particular those adjacent to the tumoral cells. A first step of a treatment plan is the capture of images of the tumoral region by CT-scan. Based on these images, an oncologist identifies the right targets and determines the locations and doses to be deposited to kill the tumoral cells. Such plan must satisfy multiple, often competing, parameters, and is therefore quite complex. For this reason, treatment planning is generally carried out with a computer.
The treatment plan (TP) generally comprises a definition of an array of n beamlets (bi), including values of planned parameters comprising,
The treatment plan must ensure that, at the end of the treatment, a total target dose greater than or equal to a minimum target dose has been delivered to the tumoral cells forming the target effective for destroying/killing the tumoral cells. This can be defined by a target dose volume histogram (=tDVH), for the structure of interest. An example of tDVH is represented with a solid line in
Historically, treatment plans by radiation therapy included the delivery of radiation doses to the treated cells at a conventional dose deposition rate (CDR) lower than 1 Gy/s. With rare exceptions, current radiation therapy facilities deliver dose-rates around 0.1 Gy/s and most clinical protocols involve daily delivery of several target fraction doses of 2 to 15 Gy cumulated to reach the total target dose which often exceeds the tolerance limit of normal tissues located in the radiation field, thus damaging them together with the tumoral cells. Recently, it has been observed that a same dose had different effects on healthy cells but not on tumoral cells when deposited at conventional dose deposition rates (CDR) or at ultra-high dose deposition rate (HDR); HDR can be one or more orders of magnitude larger than conventional dose deposition rates (CDR) usually applied. Deposition of a charge at ultra-high dose deposition rates (HDR) is also referred to as FLASH-radiotherapy (=FLASH-RT). It has been demonstrated experimentally on animals and on various organs, that ultra-high rate dose deposition at HDR can significantly spare healthy tissues in comparison with conventional deposition of a same dose at CDR and, at the same time, tumoral cells respond same or even better to HDR deposition than to CDR deposition. For example, FLASH-RT reportedly elicits in mice a dramatic decrease of the incidence of lung fibrosis, of memory loss subsequent to brain irradiation, and of necrosis of the small intestine whilst keeping the anti-tumor efficiency unchanged. Such specific normal tissue sparing has been confirmed in large animals and a patient with cutaneous lymphoma has already been treated with FLASH-RT.
The dose rate distribution in tissue can be defined by a target dose rate volume histogram (=tDRVH), for the structure of interest. An example of tDRVH is represented with a solid line in
The dose volume histogram (DVH) and dose rate volume histogram (DRVH) are cumulative histograms. There are other ways, however, of representing the distribution of doses or dose rates. For example,
Fulfilling the planned beamlet scanning sequence may require defining a planned starting time and a planned end time for each beamlet. This is particularly useful for FLASH-RT.
A treatment plan system (TPS) defines beamlets parameters including positions of beamlets (Xj), monitoring units (MUj) and spots sequence to achieve a treatment plan (TP). A translation system (=TS) determines the operating parameters of a given particle accelerating system required for implementing the beamlets parameters, taking into account limits of the particle accelerating system. This is described, e.g., in US20200298020, EP3932482, and EP3932481. The operating parameters are determined to ensure that the beamlets delivered by the given particle accelerating system will deposit doses into the structure of interest according to the target dose (rate) distribution histogram (D(R)DH) within the acceptable bands of variation (BV). The conversion of the TP into machine operating parameters ensures that a treatment session can be completed within the treatment plan, and not interrupted because at some point, the random variability of some operating parameters leads the beamlets actually delivered by the particle accelerating system to yield a D(R)DH which falls outside of the acceptable band of variation (BV).
The nominal value of the operating parameters defined in the plan may not be precisely met by a given particle accelerating system. The operating parameters values at which the particle accelerating system will actually be operating follow instead specific statistical distributions (Tj) as illustrated in
Some particle accelerating systems are equipped with a monitoring device measuring the actual operating parameters of the beamlets as they are being delivered through a nozzle. EP2116277, EP3375484, U.S. Ser. No. 10/456,598, EP3222322, WO2020249565, and EP2833970 describe examples of devices for in situ monitoring and verification of a selection of operating parameters of the beamlets delivered by a particle accelerating system. In case the operating parameters of one or more beamlets monitored differ from the planned values (which is likely to happen), there is a risk that the corresponding dose (rate) distribution histogram (D(R)DH) extend beyond the acceptable band of variation (BV). Should this be the case, the treatment session must be stopped. This is very uncomfortable for the patient who may have to come back later to complete the treatment session, depending on the usually tight schedule of the particle accelerating system. It is therefore useful to take the distributions of the operating parameters into consideration to ensure that a treatment session can be completed within a predefined confidence level (CLj) with all beamlets yielding the dose (rate) distribution histogram (D(R)DH) in accordance with the treatment plan.
Furthermore, an actual value of an operating parameter differing from the corresponding planned value does not necessarily mean that the corresponding dose (rate) distribution histogram (D(R)DH) extends beyond the acceptable band of variation. With the calculating power of processors available to date, it is not thinkable of calculating for each measurement of the actual values of the operating parameters the corresponding calculated dose (rate) distribution histogram (D(R)DH) to decide whether or not to interrupt the treatment session because one actual value differs from the planned values of the operating parameters.
EP3498336 describes a system and method for treating a dummy (mannequin) and evaluating a dose volume histogram (DVH) by dosimetry before applying the treatment to a patient. This technique clearly reduces the risk of having to interrupt a treatment session, but it also requires blocking a particle accelerating system for the time required to make the tests, during which it cannot be used for treating a patient. Furthermore, a dosimetric tolerance can translate into different machine tolerance levels for spots at different locations or with different monitoring units (Mus) (e.g., spot on the edge of the structure of interest may have a higher constraint on position accuracy than spots at the center of the structure). It is also not obvious how to translate a dose rate tolerance into a tolerance for each spot in the spot map. In some places it is more useful to check that the irradiation is in FLASH-RT mode than in other places. The tolerances on the dose rate may also be different between positions in the tissue. For example, FLASH-RT could be required at the edges of the structure of interest, where tumoral cells are flanked by healthy cells which must be spared.
To date, most methods for ensuring that the operating parameters will yield the desired dose (rate) distribution histogram (D(R)DH) is a posteriori, i.e., by measuring the treatment properties of the beamlets delivered by the particle accelerating system or, at best, based on dosimetric tests on a dummy (mannequin) performed before treating the patient. There remains a need in the art for a method for determining by calculation, i.e., without having to use precious accelerator time and energy, a set of operating parameters of the particle accelerating system to be used, yielding within a predefined confidence level the desired dose (rate) distribution histogram (D(R)DH) within the acceptable band of variation (BV).
The present disclosure provides a method implemented by a computer for optimizing tolerance values of operating parameters of a particle accelerating system allowing a beam formed by a plurality of beamlets of particles accelerated along an irradiation axis (Z), to deposit doses by pencil beam scanning (=PBS) into a structure of interest of a patient according to a treatment plan (=TP), the computer implemented method comprising,
In accordance with some embodiments, a final statistical distribution (Tf) can be set with a final confidence level (CLf) to the corresponding operating parameters by a human operator or by a processor as follows,
In accordance with some embodiments, the new tentative statistical distributions (T(j+k)) can have lower standard deviations (σj) than the corresponding tentative statistical distributions (Tj) defined above. The tentative statistical distributions (Tj) of each of the operating parameters may be gaussian distributions, and the value of the confidence level (CLj) can be comprised between 68% and 99.7%, or between 95.5 and 99% of the tentative statistical distribution wherein a confidence level (CLj) of 68% corresponds to μj±σj, a confidence level (CLj) of 95% corresponds to μj±2σj, and a confidence level of 99.7% of the tentative statistical distribution, corresponds to μj±3σj, wherein μj is an average value and σj is the standard deviation of the corresponding tentative statistical distributions, the average values (μj) and standard deviations (σj) of each operational parameter being the same or different for each beamlet (bi).
In accordance with some embodiments, the particle accelerating system can be equipped with a cyclic checker (provided as a cyclic check module) configured to measure at different intervals or continuously actual values of the operating parameters including the monitor unit (MUai), the position (Xai), and of the starting time and end time (t0ai, t1ai) of the beamlets emitted by the particle accelerating system.
In accordance with some embodiments, the particle accelerating system can also be equipped with a processor configured to compare the actual values of the operating parameters with the corresponding confidence level (CLj), and to stop a treatment session in case one actual value of an operating parameter falls outside of the corresponding confidence level (CLj).
In accordance with some embodiments, the planned parameters also comprise a planned beamlet size (dj) and a planned beam current (Ij), whose respective values (dij, Iij) used for calculating the one or more calculated dose (rate) distribution histograms (=cD(R)DH) are randomly selected within a corresponding tentative statistical distribution (Tj) of the beamlet size (dj) and of the planned beam current (Ij).
In accordance with some embodiments, the calculated distributions (CDj) of the one or more cD(R)DH are defined by a corresponding area comprised in an envelope defined between, on the one hand, a minimum calculated dose (rate) distribution histogram (=cD(R)DHj0) and, on the other hand, a maximum calculated dose (rate) distribution histogram (=cD(R)DHj1). The cD(R)DHj0 is defined by the lowest values of cD(R)DHj calculated with the predefined confidence level (CLj) from the N randomly selected values of the monitor unit (Muij), position (Xij) of the beamlets, and starting time and end time (t0ij, t1ij) and the cD(R)DHj1 is defined by the highest values of cD(R)DHj calculated with the predefined confidence level (CLj) from the N randomly selected values of the monitor unit (Muij), position (X0i) of the beamlets, and starting time and end time (t0ij, t1ij).
In accordance with some embodiments, the treatment plan includes depositing doses into at least a portion of the structure of interest at ultra-high deposition rates (UHDR) defined as a deposition rate greater than or equal to 1 Gy/s.
The present disclosure also provides an error predictor (provided as an error predicting module) configured to implement the above method, comprising,
In accordance with some embodiments, the processor can be further configured, in case any one of the one or more of the calculated distributions (CDj) of cDDHj and cDRDHj are not comprised within the corresponding acceptable bands of variation with the pre-defined confidence level (CLj), to repeat steps (i) to (iii) with new tentative statistical distributions (T(j+k)) of the operating parameters, until the calculated distributions (CDj+k) of the one or more of cDDHj+k and cDRDHj+k are both comprised within the corresponding acceptable bands of variation with the pre-defined confidence level (CLj).
For a fuller understanding of the nature of the present disclosure, reference is made to the following detailed description taken in conjunction with the accompanying drawings in which:
The present disclosure provides a computer implemented method and an error predicting module which considerably reduce the risk of carrying out a treatment session which does not respect a corresponding treatment plan, for reasons of equipment.
The present disclosure also provides a computer implemented method for optimizing tolerance values of operating parameters of a particle accelerating system allowing a beam formed by a plurality of beamlets of accelerated particles to deposit doses by pencil beam scanning (=PBS) into a structure of interest of a patient according to a treatment plan (=TP). The method allows determining the confidence level (CLj) for a given particle accelerating system to deliver beamlets which will fulfil the TP as a function of a set of operating parameters. If the confidence level obtained is too low, an alternative set of operating parameters needs be evaluated. These and other advantages of the present disclosure are presented below.
The present disclosure provides a method implemented by a computer for optimizing tolerance values of operating parameters of a particle accelerating system allowing a beam formed by a plurality of beamlets of particles accelerated along an irradiation axis (Z), to deposit doses by pencil beam scanning (=PBS) to a patient according to a treatment plan (=TP). The particles may be protons, but they can be electrons, heavy ions beams, but also waves formed by accelerated particles interacting with a converting material, such as x-rays (or y-rays). As illustrated in
The method also receives a planned starting time (t0pi) and end time (t1pi) at which each beamlet is to be delivered. This is particularly useful in case of FLASH-RT. The starting and end times may or may not be part of the TP.
A structure of interest is defined, characterizing one or more tissues, of the patient, being traversed by or interacting with the beamlets (bi). The structure of interest comprises the target comprising the tumoral cells to be killed, but also healthy tissues traversed or somehow touched by one or more beamlets. These include, for example, the tissues located upstream from the target along the irradiation axis (Z), i.e., between the nozzle of the particle accelerating system and the target, or also the tissues adjacent to and surrounding the target.
An objective of the treatment is to yield at the end of a session, given values of a target dose (rate) distribution histogram (tD(R)DH) for the structure of interest, within an acceptable band of variation (BV) in which a target dose (rate) volume histogram tD(R)VH is allowed to vary. As shown in
First, as shown in
The selection of tentative statistical distributions (Tj) is based on the performance of the particle accelerating system and is expected to generate a beam delivering the beamlets scanning sequence and generating calculated dose volume histograms (=cDVHj) and calculated dose rate volume histograms (=cDRVHj) in the structure of interest that are comprised within the acceptable band of variation (BV). For example, a treatment plan system (TPS) can determine average values (μj) achievable by the particle accelerating system which would yield the desired cDVH and cDRVH. The actual operating parameters of the treatment machine on a specific day, however, are not restricted to the corresponding average values (μj) but are distributed generally over a Gaussian curve, varying from day to day or during the course of a day (shown in
A value (MUij) of the monitor unit, a value (Xij) of the position of the beamlet, and values (t0ij, t1ij) of starting time (t0ij) and end time (t1ij) are randomly selected from the corresponding tentative statistical distributions (Tj) within the corresponding confidence levels (CLj) as shown with the black circles in the Gaussian distribution curves (Tj) of
As shown in
If, on the other hand, any one of the cD(R)DHj-Run x calculated extends beyond the acceptable band of variation (BV), it could be concluded that there would be a risk higher than the predefined confidence level (CLj) of having to interrupt a treatment run with the operating parameters of the particle accelerating system according to the tentative statistical distributions (Tj). A new tentative statistical distribution (T(j+1)) of the operating parameters is then selected, and the cD(R)DH(j+1) is calculated in a new series of N Runs with randomly selected values of the new tentative statistical distributions (T(j+1)) at each successive Run. This operation is repeated with new tentative statistical distributions (T(j+k)) until the corresponding calculated distributions (CD(j+k)) are entirely comprised within the acceptable band of variation (BV).
Besides the planned position (Xpi), the planned monitor unit (MUpi), the planned beamlet scanning sequence over the planned positions (Xpi), and the planned starting time (t0pi) and end time (t1pi), the planned parameters can also comprise a planned beamlet size (dj) and a beam current (Ij), whose values (dij, Iij)) used for calculating the calculated dose (rate) distribution histogram (=cD(R)DHj) are randomly selected within a corresponding tentative statistical distribution (Tj) of the beamlet size (dj) and beam current (Ij).
The calculated distributions (CDj) of the cD(R)DHj can be defined by a corresponding area comprised between,
By comparing the calculated distributions (CDj) of cD(R)DHj with the acceptable band of variation (BV), it can be determined whether or not the calculated distribution (CDj) of cD(R)DHjj is comprised in the corresponding acceptable band of variation for the pre-defined confidence level (CLj).
The method of the present disclosure may conclude by setting a final statistical distribution (Tf) to the corresponding operating parameters with a final confidence level (CLf). If the N runs of calculated distributions (CDj) of cD(R)DHj are comprised within the corresponding acceptable band of variation (BV) with the pre-defined confidence level (CLj), for the given treatment plan (TP), the tentative statistical distribution (Tj) can be set as the final statistical distribution (i.e., Tf=Tj) with the final confidence level (CLf=CLj) to define the corresponding operating parameters. It can be concluded that, by implementing the final statistical distributions (Tf) of operating parameters, the particle accelerating system has a probability equal to the final confidence level (CU) of delivering the beamlets fulfilling the treatment plan (TP) with corresponding actual dose (rate) distribution histogram(aD(R)DH) comprised within the acceptable band of variation (BV), as shown e.g., in
If, on the other hand, any one of the N runs of calculated dose (rate) distribution histogram (cD(R)DHj) extends beyond the boundaries of the corresponding acceptable bands of variation (BV) with the pre-defined confidence level (CLj) as shown e.g., in
The new tentative statistical distributions (T(j+k)) defined in case the j+(k−1) preceding tentative statistical distributions did not yield calculated distributions included within the bands of variation (BV) surrounding target dose (rate) distribution histogram (tD(R)DH) for the pre-defined confidence level (CLj), can be selected as distributions having lower standard deviations (σj) (or variances (σj)2) than the corresponding preceding tentative statistical distributions (Tj+(k−1)).
Setting the final statistical distribution (Tf) for the corresponding operating parameters can be carried out by a human operator or automatically, by a processor.
The tentative statistical distributions (Tj) of each of the operating parameters may be Gaussian distributions. The values of the confidence levels (CL) may be comprised between 68% and 99.7%, or may be between 95.5 and 99% of the tentative statistical distribution. As shown in
The particle accelerating system can be equipped with a cyclic check module configured for measuring at different intervals or continuously actual values of the monitor unit (MUai), the position (Xai), and of the starting time and end time (t0ai, t1ai) of the beamlets emitted by the particle accelerating system. A processor can be configured to determine whether any actual operating parameter falls outside of the corresponding final confidence level (CU) (shown in
The fact that one or more values of the actual operating parameters fall outside of the corresponding confidence levels (CLj) does not mean that the resulting actual dose (rate) distribution histogram (aD(R)DH) necessarily falls outside of the acceptable band of variation (BV). It could therefore be over-shooting to interrupt the treatment session simply because one actual value of any operating parameter falls outside of the corresponding confidence level (CLj), as it could perfectly yield a dose (rate) distribution histogram (D(R)DH) comprised within the acceptable band of variation (BV). The processor may be configured to calculate the cD(R)DH as soon as a measured actual value of an operating parameter falls outside of the corresponding confidence level (CLj) to determine whether or not the calculated cD(R)DH are comprised within the corresponding acceptable bands of variation (BV). The calculated dose (rate) distribution histogram (cD(R)DH) can be calculated based on the actual values of the operating parameters measured on the beamlets already delivered, including the parameter falling outside of the confidence level (CLj), and on the average values (Rj) of the operating parameters according to the final statistical distribution (Tf) for the beamlets which remain to be delivered to end the treatment session. If the calculated cD(R)DH falls outside of the acceptable band of variation (BV), the treatment session is stopped. If, on the other hand, the calculated cD(R)DH is within the acceptable band of variation (BV), the treatment session can proceed further. With this functionality of the processor, the risk of stopping a treatment session is further reduced.
This function does not take excessive calculating power, as it would concern only 100%−CLj % of the actual values of the operating parameters. For a confidence level (CLj) of 95% i.e., μj±2σj, there would be a probability of only 5% that a value of an operating parameter should fall outside of the confidence level (CLj) whch the D(R)DH would have to be calculated for. For a confidence level of 99.7% of the tentative statistical distribution, i.e., μj±3σj, it would concern a probability of merely 0.3% where such calculation would be required.
In an embodiment, the treatment plan includes FLASH-RT, in that doses are to be deposited into at least a portion of the structure of interest at ultra-high deposition rates (UHDR) defined as a deposition rate greater than or equal to 1 Gy/s. In this embodiment, the planned parameters also comprise a beam current (I), whose values (Ij) used for calculating the calculated dose distribution histogram (=cDDHj) and, in particular, the calculated dose rate distribution histogram (=cDRDHj) are randomly selected within a corresponding tentative statistical distribution (Tj) of the beam current (Ij).
The present disclosure also provides an error predicting module configured to implement the method described above. The error predicting module comprises a memory comprising a plurality of tentative statistical distributions (Tj) for each operating parameter centered on a plurality of corresponding average values (Rj). It also comprises a user interface configured to enter,
The error predicting module comprises a processor configured to,
The processor can be further configured, in case the calculated distribution (CDj) of cD(R)Dj is not comprised within the corresponding acceptable band of variation with the pre-defined confidence level (CLj), to repeat the foregoing three steps with new tentative statistical distributions (T(j+k)) of the operating parameters, until the calculated distributions (CDj) of cD(R)Dj is comprised within the corresponding acceptable bands of variation with the pre-defined confidence level (CLj).
The particle accelerating system is simulated (shown in step (5) of
The method of the present disclosure can be implemented with a single calculated dose (rate) histogram (cD(R)DH) or with several histograms which must all be within the corresponding acceptable bands of variation (BV) to set the final statistical distribution (Tf). If a first tentative statistical distribution (Tj) yields one histogram (e.g., cDVH) enclosed within the corresponding acceptable band of variation (BV) but another histogram (e.g., cDRVH) extending beyond the acceptable band of variation (BV), a new tentative statistical distribution (T(j+1)) is selected and the method is carried out again, until a tentative statistical distribution (T(j+k)) is found that fits all the required histograms into the corresponding acceptable bands of variation (BV). Dose rate related histograms (e.g., DRVH, DDRH) are particularly useful for treatment plans comprising beamlets to be emitted in FLASH-mode for depositing doses at ultra-high deposition rates into selected spots of the structure of interest.
| Number | Date | Country | Kind |
|---|---|---|---|
| 22172682.1 | May 2022 | EP | regional |
