Method for treatment of multiple brain metastases based on iso-dose line prescriptions

Abstract

Disclosed is a computer-implemented method of determining a treatment plan, encompassing acquiring patient image data, acquiring target data describing targets, acquiring position data describing control points which define one or more arcs, and determining target projection data which describes outlines of the target in a beam's-eye view. Margin data is acquired. For the outlines, margins are applied to determine auxiliary outlines. Beam shaping device data is determined describing configurations of the collimator leaves so that irradiation of the auxiliary outlines is enabled. Based on these configurations, the irradiation amount is simulated for voxels of the patient image data. Constraints to be fulfilled by the treatment plan may be set. Configurations of blockings, arc-weights and margins are proposed. Only different combinations of these parameters are proposed while additional possible parameters are neglected. An optimization algorithm is used to minimize an objective function. The best configuration is selected as the treatment plan.

Claims

1. A computer-implemented method of determining a treatment for treating at least one target by emitting irradiation by a beam source through a beam shaping device in an irradiation direction movable around a movable patient support device, the method comprising the following steps: acquiring patient image data which describes one or more anatomical body parts of a patient; acquiring target data which specifies at least one of the one or more anatomical body parts as the at least one target for irradiation; acquiring position data describing at least one position of the moveable patient support device in relation to the irradiation direction; determining target projection data based on the target data and the position data, wherein the target projection data is determined for the at least one target and for the at least one position of the moveable patient support device in relation to the irradiation direction, and wherein the target projection data describes outlines of the at least one target each projected into a plane perpendicular to a corresponding simulated beam direction, wherein the corresponding simulated beam direction is specified by a corresponding position of the at least one position of the moveable patient support device in relation to the irradiation direction; acquiring margin data describing one or more margins for the at least one target, wherein each of the one or more margins is a distance of a corresponding outline of the at least one projected target to a corresponding auxiliary outline correlated with the at least one target; determining auxiliary outline data based on the target projection data and the margin data, wherein the auxiliary outline data is determined for the at least one target, for the at least one position of the moveable patient support device in relation to the irradiation direction and for the one or more margins, and wherein the auxiliary outline data describes one or more auxiliary outlines correlated with the at least one target, for the one or more margins; determining beam shaping device data based on the auxiliary outline data, wherein the beam shaping device data describes one or more configurations of the beam shaping device which enable irradiation of one or more irradiation areas specified by the one or more auxiliary outlines; determining irradiation data based on the patient image data and the beam shaping device data, wherein the irradiation data is determined for at least one voxel of the patient image data and for the one or more configurations of the beam shaping device, and wherein the irradiation data describes a simulated irradiation dose received by the at least one voxel, for each of the one or more configurations of the beam shaping device described by the beam shaping device data; acquiring constraint data describing criteria to be fulfilled by the treatment, the treatment specifying one or more paths along which the irradiation direction and/or the moveable patient support device shall move during irradiation, wherein the one or more paths are each specified by one or more control points being one or more of the at least one position of the moveable patient support device in relation to the irradiation direction described by the position data, specifying, for each of the one or more paths, monitor units to be emitted by the beam source during movement along each of the one or more paths and specifying a corresponding configuration of the beam shaping device for each of the one or more control points; determining the treatment based on the irradiation data and the constraint data, wherein an arc weight is defined as a sum of the monitor units to be emitted by the beam source during movement along one of the one or more paths, a blocking is defined as a configuration of the beam shaping device preventing irradiation to an irradiation area at a corresponding control point, and wherein the treatment is determined based on different combinations of: the one or more margins, arc weights for the one or more paths, and blockings for the one or more configurations of the beam shaping device; and causing the irradiation to be emitted by the beam source according to the treatment.

2. The method of claim 1, further comprising the following step: acquiring path data describing the one or more paths each specified by the one or more control points being one or more of the at least one position of the moveable patient support device in relation to the irradiation direction described by the position data, wherein the one or more paths specified by the treatment are the one or more paths described by the path data.

3. The method of claim 1, further comprising the following step: acquiring blocking data describing the blockings for the one or more paths specified by the treatment.

4. The method according to claim 1, wherein the position data describes only the one or more control points specifying the one or more paths specified by the treatment.

5. The method according to claim 1, wherein the constraint data describes at least one of the following: a lower dose prescription limit specifying a minimum value of a sum of all simulated irradiation doses received by a first predetermined volumetric percentage of a target, of the at least one target, when following the treatment; and an upper dose prescription limit specifying a minimum value of the sum of all simulated irradiation doses received by a second predetermined volumetric percentage of the target when following the treatment.

6. The method according to claim 5, wherein the treatment is determined by performing the following steps: generating a plurality of auxiliary treatment steps, based on the different combinations of the one or more margins, the arc weights and the blockings; and for each of the plurality of auxiliary treatment steps, determining target dose data based on the irradiation data, wherein the target dose data is determined for the at least one target, and wherein the target dose data describes the sum of all simulated irradiation doses received by the at least one target when following the plurality of auxiliary treatment steps, and determining rating data at least based on the target dose data and the constraint data, wherein the rating data describes a degree to which the plurality of auxiliary treatment steps match the criteria to be fulfilled by the treatment; and selecting one of the plurality of auxiliary treatment steps as the treatment based on the rating data.

7. The method according to claim 6, further comprising the following steps performed for each of the plurality of auxiliary treatment steps after having determined the target dose data: determining first difference data based on the constraint data and the target dose data, wherein the first difference data is determined for the at least one target, wherein the first difference data describes a first difference between a lower dose prescription limit and the sum of all simulated irradiation doses received by the at least one target when following the plurality of auxiliary treatment steps; determining second difference data based on the constraint data and the target dose data, wherein the second difference data is determined for the at least one target, wherein the second difference data describes a second difference between an upper dose description limit and the sum of all simulated irradiation doses received by the at least one target when following the plurality of auxiliary treatment steps; and determining total arc weight data based on the monitor units to be emitted by the beam source during movement along the one or more paths specified by the treatment having the plurality of auxiliary treatment steps, wherein the total arc weight data describes a sum of all arc weights of all paths specified by the plurality of auxiliary treatment steps, wherein a degree to which the plurality of auxiliary treatment steps match the criteria to be fulfilled by the treatment is specified at least by the first difference described by the first difference data, the second difference described by the second difference data and the sum of all arc weights described by the total arc weight data.

8. The method according to claim 6, further comprising the following step performed for each of the plurality of auxiliary treatment steps after having determined the target dose data: determining normal tissue dose data based on the irradiation data, wherein the normal tissue dose data is determined for the at least one target, wherein the normal tissue dose data describes the sum of all simulated irradiation doses received by normal tissue when following the plurality of auxiliary treatment steps, wherein the normal tissue includes all of the one or more anatomical body parts of the patient described by the patient image data and wherein the rating data is determined furthermore based on the normal tissue dose data.

9. The method according to claim 8, further comprising the following steps performed for each of the plurality of auxiliary treatment steps after having determined the normal tissue dose data: determining gradient index data based on the normal tissue dose data, wherein the gradient index data describes a relation between a first volume of the normal tissue which receives at least a first predetermined sum of all simulated irradiation doses when following the plurality of auxiliary treatment steps and a second volume of the normal tissue which receives at least a second predetermined sum of all simulated irradiation doses when following the plurality of auxiliary treatment steps, wherein a degree to which the plurality of auxiliary treatment steps matches the criteria to be fulfilled by the treatment is specified furthermore by the relation between the first volume and the second volume described by the gradient index data.

10. The method according to claim 6, further comprising the following step: acquiring risk structure data which designates at least one of the one or more anatomical body parts as at least one risk structure for irradiation, wherein the constraint data describes at least one risk structure dose limit specifying a simulated irradiation dose received by the at least one risk structure.

11. The method according to claim 10, further comprising the following step performed for each of the plurality of auxiliary treatment steps before determining the rating data: determining risk structure dose data based on the irradiation data and the risk structure data, wherein the risk structure dose data is determined for the at least one risk structure, wherein the risk structure dose data describes the sum of all simulated irradiation doses received by the at least one risk structure when following the plurality of auxiliary treatment steps, and wherein the rating data is determined furthermore based on the risk structure dose data.

12. The method according to claim 11, further comprising the following step performed for each of the plurality of auxiliary treatment steps after having determined the risk structure dose data: determining difference data based on the constraint data and the risk structure dose data, wherein the difference data is determined for the at least one risk structure wherein the difference data describes a difference between the at least one risk structure dose limit and the sum of all simulated irradiation doses received by the at least one risk structure when following the plurality of auxiliary treatment steps, wherein a degree to which the plurality of auxiliary treatment steps match the criteria to be fulfilled by the treatment is specified furthermore by the difference described by the difference data.

13. A non-transient computer program which, when running on a computer or when loaded onto a computer, causes the computer to perform the method steps of the method according to claim 6.

14. A medical system, comprising: at least one computer having at least one processor operable to execute the steps of a computer program comprising: acquiring patient image data which describes one or more anatomical body parts of a patient; acquiring target data which specifies at least one of the one or more anatomical body parts as at least one target for irradiation; acquiring position data describing at least one position of a patient support device in relation to an irradiation direction; determining target projection data based on the target data and the position data, wherein the target projection data is determined for the at least one target and for the at least one position of the patient support device in relation to the irradiation direction, wherein the target projection data describes outlines of the at least one target projected into a plane perpendicular to a corresponding simulated beam direction, and wherein the corresponding simulated beam direction is specified by a corresponding position of the at least one position of the patient support device in relation to the irradiation direction; acquiring margin data describing one or more margins for the at least one target, wherein a margin, of the one or more margins, is a distance of a corresponding outline of the at least one projected target to a corresponding auxiliary outline correlated with the at least one target; determining auxiliary outline data based on the target projection data and the margin data, wherein the auxiliary outline data is determined for the at least one target, for the at least one position of the patient support device in relation to the irradiation direction and for the one or more margins, and wherein the auxiliary outline data describes one or more auxiliary outlines correlated with the at least one target, that respectively corresponds to an outline of the at least one projected target; determining beam shaping device data based on the auxiliary outline data, wherein the beam shaping device data describes one or more configurations of a beam shaping device which enable irradiation of one or more irradiation areas specified by the one or more auxiliary outlines; determining irradiation data based on the patient image data and the beam shaping device data, wherein the irradiation data is determined for at least one voxel of one or more images of the patient image data and for the one or more configurations of the beam shaping device, and wherein the irradiation data describes a simulated irradiation dose received by the at least one voxel, for each of the one or more configurations of the beam shaping device described by the beam shaping device data; acquiring constraint data describing criteria to be fulfilled by a treatment, the treatment specifying one or more paths along which the irradiation direction and/or the patient support device shall move during irradiation, wherein the one or more paths are each specified by one or more control points being one or more of the at least one position of the patient support device in relation to the irradiation direction described by the position data, specifying, for each of the one or more paths, monitor units to be emitted by a beam source during movement along each of the one or more paths and specifying a corresponding configuration of the beam shaping device for each of the one or more control points; and determining the treatment based on the irradiation data and the constraint data, wherein an arc weight is defined as a sum of the monitor units to be emitted by the beam source during movement along one of the one or more paths, a blocking is defined as a configuration of the beam shaping device preventing irradiation to an irradiation area at a corresponding control point, of the one or more control points, and wherein the treatment is determined based on different combinations of: the one or more margins, arc weights for the one or more paths, and blockings for the one or more configurations of the beam shaping device; at least one electronic data storage device storing at least the patient image data; and a medical device for carrying out a medical procedure on the patient, wherein the at least one computer is operably coupled to the at least one electronic data storage device to acquire, from the at least one electronic data storage device, at least the patient image data, and wherein the at least one computer is operably coupled to the medical device to issue a control signal to the medical device to control an operation of the medical device based on the treatment.

15. The system according to claim 14, wherein the medical device comprises: a radiation treatment apparatus comprising the beam source, the beam shaping device and the patient support device, wherein the at least one computer is operably coupled to the radiation treatment apparatus for issuing a control signal to the radiation treatment apparatus for controlling, based on the treatment, at least one of: an operation of the beam source, an operation of the beam shaping device, or the at least one position of the patient support device.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) In the following, the invention is described with reference to the appended figures which give background explanations and represent specific embodiments of the invention. The scope of the invention is however not limited to the specific features disclosed in the context of the figures, wherein

(2) FIG. 1 illustrates the basic steps of the method according to a first aspect;

(3) FIG. 2 shows a schematic illustration of the system according to the fifth aspect;

(4) FIG. 3 shows spatial dose distributions;

(5) FIG. 4 shows a cumulative dose-volume histogram;

(6) FIG. 5 shows a configuration of the beam shaping device for a given outline;

(7) FIG. 6 shows a configuration of the beam shaping device for a given auxiliary outline;

(8) FIG. 7 shows a configuration of the beam shaping device for another given auxiliary outline;

(9) FIG. 8 is a schematic illustration of the system according to the fifth aspect; and

(10) FIG. 9 illustrates the steps of an example of the method according to the first aspect.

DESCRIPTION OF EMBODIMENTS

(11) FIG. 1 illustrates the basic steps of the method according to the first aspect, in which step S1.1 encompasses acquiring patient image data, step S1.2 encompasses acquiring target data, S1.3 encompasses acquiring position data, S1.4 encompasses determining target projection data, S1.5 encompasses acquiring margin data, S1.6 encompasses determining auxiliary outline data, S1.7 encompasses determining beam shaping device data, S1.8 encompasses determining irradiation data, S1.9 encompasses acquiring constraint data and S1.10 encompasses determining the treatment plan.

(12) FIG. 2 shows a schematic illustration of the system according to the fifth aspect. The beam shaping device 5 may emit an irradiation beam B. The irradiation beam B may be a simulated irradiation beam when conducting the method for determining the treatment plan. A patient 4 may be positioned on the patient support device 6. The patient may be in a position defined in a first reference system R1, the patient support device may be in a position defined by a second reference system R2 and the irradiation direction may be defined by a third reference system R3. The patient image data may also be defined in the first reference system R1, for example by registering the patient to the patient image data using a commonly known image registration method. The irradiation beam B may irradiate a target 7 which is an anatomical body part of the patient 4 and in the shown example located in the head of the patient (e.g. a brain tumor). The beam shaping device 5 may be movable around the patient support device 6. The patient support device 6 itself may also be movable. Transformations between R1, R2 and R3 are for example known. These transformations may be predetermined, determined using image registration or optical tracking means, specified by a user or else. The irradiation direction may be equal to the position of the beam shaping device 5 in the third reference system.

(13) FIG. 3 shows spatial dose distributions for two different beam shaping device configurations. The dose profiles for the two different beam shaping device configurations, for example corresponding to two different auxiliary outlines for two different margins of a given outline, are very different. To compare the profiles, they are both normalized to fulfill a minimum dose of D1. The spatial dose distributions are represented as 1D dose profiles. The primary (horizontal) axis denotes spatial position and the secondary (vertical) axis denotes a received irradiation dose at the respective spatial positions. The area of a target to be irradiated is indicated in the primary (horizontal) axis with a dashed bar. A first spatial dose distribution for a first configuration of the beam shaping device is shown as H1. A second spatial dose distribution for a second configuration of the beam shaping device is shown as H2. The minimum dose received by the target in the case of H1 and H2 is indicated by D1 (e.g. normalized as noted above). The maximum dose received by the target in the case of H2 is indicated by D2. It is clear that the maximum dose received by the target is higher for H2 than for H1, which means that the target receives a higher irradiation dose in the case of H2.

(14) Furthermore, the areas outside the target region indicated by the dashed bar receive different amounts of irradiation depending on the configuration of the beam shaping device. In the case of H1, a larger area (volume) of tissue outside the target receives irradiation. Consequently, tissue which is not to be treated (outside the target) receives large irradiation doses according to H1. In this example, the gradient index of H2 is closer to a value of “1” than the gradient index of H1.

(15) As noted above, the shape of the functions H1 and H2 (the spatial irradiation dose distribution) depends on the configuration of the beam shaping device. In the case of a collimator blocking parts of the irradiation emitted by the beam source, physical effects such as scattering of the irradiation have to be taken into account. Not only the shape of a mask through which irradiation passes determines the shape of the functions H1 and H2, but also the absolute dimensions of the mask. In general, the spatial irradiation dose distribution will look more like H1 in case a larger hole is used in the mask allowing for more irradiation to pass. The smaller the hole in the mask (e.g. a collimator), the more will the spatial irradiation dose distribution look like H2. The absolute dose values can be adjusted by increasing/decreasing the amount of irradiation emitted by the beam source, while the spatial irradiation dose distribution will not change (shape of the curve (e.g. H1 or H2) stays the same).

(16) FIG. 4 shows a cumulative dose-volume histogram. The primary (horizontal) axis denotes an irradiation dose and the secondary (vertical) axis denotes a volumetric percentage (e.g. a volumetric percentage of a target or a volumetric percentage of all anatomic body parts described by the patient image data) of tissue which is to receive the respective irradiation dose.

(17) For example, the curve H3 indicates a planned volumetric irradiation dose distribution for a given target. A lower prescription limit P3a is defined as a minimum amount D3a Gy to be received by V3a Vol.-% of the given target. For example, the lower prescription limit may define a minimum amount of 16 Gy to be received by 95 Vol.-% of the given target. An upper prescription limit P3b is defined as a minimum amount of D3b Gy received by V3b Vol.-% of the given target. For example, the upper prescription limit may define a minimum amount of 20 Gy to be received by 1 Vol.-% of the given target.

(18) For example, the curve H4 indicates a planned irradiation dose for a risk structure. An irradiation dose limit P4 is defined as a maximum amount of D4 Gy to be received by V4 Vol.-% of the risk structure. For example, the irradiation dose limit may define a maximum amount of 5 Gy to be received by 10 Vol.-% of the risk structure, e.g. by the 10% of the risk structure which receive the highest irradiation dose. Instead of a risk structure, curve H4 may be defined for the normal tissue.

(19) Both curves H3 and H4 may be influenced by a user, for example by defining one or more points through which the curves shall run (e.g. a lower dose prescription limit P3a and/or an upper dose prescription limit P3b). Note that the curves do not necessarily run through the points. For example, a user may specify a maximum dose limit to be received by a risk structure as P4. The curve H4 may run through the point P4, but may also lie below the point P4. That is, the points defined by the user restrict the curve at a given dose to be equal to or lower than a set value. The points defined by the user may alternatively restrict the curve at a given dose to be equal to or higher than a set value. The treatment plan may be determined based on this user input, i.e. a lower point prescription limit, an upper dose prescription limit, a normal tissue dose limit and/or a risk structure dose limit. A perfect treatment plan would result in an irradiation which perfectly fits all the constraints (boundary conditions) specified by the user (e.g. perfectly fits the lower and upper dose prescription limit specified as P3a and P3b). However, the determination of such a perfect treatment plan is very cumbersome—in some cases even impossible—and also may result in several disadvantages (longer treatment time, higher energy consumption, necessary patient re-alignment etc.). Therefore, a compromise between the given boundary conditions (constraints) and these disadvantages has to be found in order to determine the treatment plan.

(20) As described above with respect to FIG. 1, patient image data is acquired, for example x-ray, ultrasonic, CT, MR or CBCT image data. Then, target data is acquired. For example, a user specifies the at least one target (the geometry (shape and position)) in the patient image data. Afterwards, position data is acquired. The position data describes several control points for several paths (arcs). For at least some of the control points (for example for all of the control points), projections of the at least one target (outlines) are determined. One example of a determined outline of a single target is shown as outline 2 in FIGS. 5 to 7.

(21) Then, margin data is acquired. Based on the margins described by the margin data and the determined outlines, auxiliary outlines are determined. One example of an auxiliary outline correlated with the single target used for determining the outline 2 is shown as 2a and 2b in FIGS. 6 and 7. In FIG. 6, the auxiliary outline 2a lies within the outline 2 (is enclosed in the outline 2). In this case, a negative margin (e.g. −1 mm) was used to determine the auxiliary outline 2a from the outline 2. In FIG. 7, the auxiliary outline 2b encloses the outline 2. In this case, a positive margin (e.g. 0.5 mm) was used to determine the auxiliary outline 2b from the outline 2. As can be seen from both FIGS. 6 and 7, the auxiliary outlines 2a and 2b each have a homogeneous distance to the outline 2. These distances are each defined by the margin (e.g. −1 mm and 0.5 mm) used to generate the auxiliary outlines 2a and 2b, respectively. The auxiliary outline may be regarded as enlarged or scaled-up (using size and/or shape adjustment) in the case of FIG. 7 and as shrunk or scaled-down (using size and/or shape adjustment) in the case of FIG. 6.

(22) After having determined the auxiliary outline data, the method proceeds with determining beam shaping device data. The beam shaping device data describes configurations of the beam shaping device which enable irradiation of one or more irradiation areas specified by the one or more auxiliary outlines.

(23) In FIGS. 5 to 7, the beam shaping device and the target are shown in a beam's-eye-view for a given control point. The projection of the target 1 into the plane perpendicular to the simulated beam direction (which here corresponds to the viewing direction) results in a two-dimensional shape of the target 1 indicated by a dotted area in FIGS. 5 to 7. In the example of FIGS. 5 to 7, a multi-leaf collimator is shown as the beam shaping device. It should be clear that other beam shaping devices are possible. In FIGS. 5 to 7, positions of collimator leaves 3 are specified by the configuration of the beam shaping device described by the beam shaping device data.

(24) FIG. 5 shows a configuration of the beam shaping device for a given auxiliary outline determined from an outline 2 for a margin equal to zero. In this case, as noted above, the auxiliary outline is equal to the outline 2. In the example shown in FIG. 5, the collimator leaves do not cross the auxiliary outline which is equal to the outline 2. In particular, the collimator leaves are adapted to/fit to the auxiliary outline which is equas to the outline 2. The area through which irradiation can pass is defined by the collimator leaves. This area is minimized as much as possible whilst the collimator leaves do not cross the auxiliary outline.

(25) FIG. 6 shows a configuration of the beam shaping device for a given auxiliary outline 2a determined from an outline 2 for a margin smaller than zero. In this case, as noted above, the auxiliary outline 2a lies inside the outline 2. In the example shown in FIG. 6, the collimator leaves cross the outline 2, but do not cross the auxiliary outline 2a. In particular, the collimator leaves are adapted to/fit to the auxiliary outline 2a. Also in this case, the area through which irradiation can pass is defined by the collimator leaves and minimized as much as possible whilst the collimator leaves do not cross the auxiliary outline 2a.

(26) FIG. 7 shows a configuration of the beam shaping device for a given auxiliary outline 2b determined from an outline 2 for a margin larger than zero. In this case, as noted above, the auxiliary outline 2b encloses the outline 2. In the example shown in FIG. 7, the collimator leaves do not cross the outline 2 and also do not cross the auxiliary outline 2b. In particular, the collimator leaves are adapted to/fit to the auxiliary outline 2b. Also in this case, the area through which irradiation can pass is defined by the collimator leaves and minimized as much as possible whilst the collimator leaves do not cross the auxiliary outline 2a.

(27) After having determined the beam shaping device data, the method proceeds with determining irradiation data for the configurations of the beam shaping device. Then, constraint data is acquired and the treatment plan is determined.

(28) For example, a plurality of auxiliary treatment plans is determined. After the selection of one of the plurality of auxiliary treatment plans, the method continues with a step of generating a plurality of secondary auxiliary treatment plans. For example, one auxiliary treatment plan is generated using a margin equal to zero, no blockings and arc-weights which enable the lower and/or upper dose prescription limit (e.g. only the lower dose prescription limit) to be met. Other auxiliary treatment plans may then be generated based on this one auxiliary treatment plan by changing one (or more) of the margins, arc-weights and blockings of the one auxiliary treatment plan, using heuristic, stochastic and/or gradient-based exploration. The generation of the auxiliary treatment plans and a subsequent selection of one of the secondary auxiliary treatment plans is in other words performed using heuristic, stochastic and/or gradient-based exploration. Multiple iterations of generating (secondary, tertiary, quaternary . . . ) auxiliary treatment plans and subsequently selecting one of these auxiliary treatment plans may be performed until the method converges (i.e. until the selected auxiliary treatment plan is only little improved with respect to a previously selected auxiliary treatment plan). The “little improvement” may be determined using an objective function. For example, the objective function assigns a rating value to each of the auxiliary treatment plans. In case the rating value of the selected auxiliary treatment plan differs from the rating value of the previously selected auxiliary treatment plan less than a predetermined convergence threshold, the method is considered as converged. In this case, the currently selected auxiliary treatment plan is selected as the treatment plan. The generation of the (secondary, tertiary, quaternary . . . ) auxiliary treatment plans and the subsequent selection of one of these auxiliary treatment plans may be performed using heuristic, stochastic and/or gradient-based exploration, for example based on the aforementioned objective function. A detailed example will be given below.

(29) FIG. 8 is a schematic illustration of the medical system 8 according to the fifth aspect. The system is in its entirety identified by reference sign 8 and comprises a computer 9, an electronic data storage device (such as a hard disc) 10 for storing at least the patient image data and a medical device 11 (such as a radiation treatment apparatus). The components of the medical system 8 have the functionalities and properties explained above with regard to the fifth aspect of this disclosure.

(30) The invention also relates to the exemplary method as described in the following.

(31) The method aims provide advanced treatment planning for multiple brain metastases including iso-dose line (IDL) prescriptions, allowing the operator to control dose homogeneity/inhomogeneity by prescribing a range of dose values per treated metastasis, and risk structure sparing, allowing the operator to reduce dose in specified volumes of interest.

(32) One can differentiate between three tissue types: target volumes (e.g. the volumes of interest containing the brain metastases which are selected for treatment by irradiation), normal tissue (e.g. the volume of the patient's head surrounding the target) and risk structures (e.g. pre-defined volumes of interest, typically corresponding to vital organs such as brainstem, eye and optical nerve).

(33) The aim of treatment planning is to find an irradiation plan which delivers the prescribed dose values to the target volumes, while minimizing the dose to surrounding normal tissue. Moreover, dose limits can be set for risk structures, to constrain dose to the respective localities.

(34) The method produces treatment plans consisting of dynamic conformal arcs (a treatment modality for linac-based radiation therapy in which the linac head rotates around a patient, utilizing a gantry) with a single iso-center. Fields are collimated dynamically using a multi-leaf collimator while the gantry of the linac rotates around the patient's head. The fields are shaped according to projections of the metastases (outlines) for a finite set of gantry angles (control points). For each control point, a projected shape (outline) can be either opened or blocked to alter the dose contribution to the irradiation are defined by the projected shape (outline). Moreover, a (e.g. negative or positive) margin can be added to the projected shape (outline) to influence the dose profile. Finally, (a number of) monitor units (arc-weights) must be set per arc (single rotation of the gantry). Monitor units are a measure of beam source (e.g. LINAC) output and influence treatment time/efficiency.

(35) For example, the outline is a feature defining dynamic conformal arcs (DCA). This separates the method from other approaches, for example from volumetric arc therapy (VMAT) approaches. In a sense, VMAT is more sophisticated than DCA (Dynamic Conformal Arc), as it contains a superset of degrees of freedom. At the same time this makes finding a good solution (a treatment plan which sufficiently fits the criteria) intractable. Moreover, VMAT fields tend to be discontinuous and hence potentially decrease the dose calculation accuracy.

(36) It follows that for dynamic conformal arc treatment plan optimization, several degrees of freedom are available: (1) Distribution of metastases (targets) to arcs (paths) (2) Arc-weights (sum of all (numbers of) monitor units of one arc/path) (3) Opening or closing (blocking) of a projected shape (outline) per control point (4) Margin per metastasis (target) per arc (path)

(37) Previous solutions for multiple brain metastases provide a solution to this optimization problem by mainly focusing on degrees of freedom (1), (2) and (3). The respective algorithm is tailored to the optimization of a single dose prescription point per metastasis and its application cannot be extended to include feature (A) and (B) as defined below. This patent application proposes a completely novel solution to solve the optimization problem, incorporating all four degrees of freedom (1), (2), (3) and (4). This allows the implementation of the following features (A) and (B).

(38) (A) Iso-Dose Line Prescription Optimization

(39) For each individual metastasis (target), a prescription range can be configured by the operator and may be defined by a lower and higher dose prescription point (lower dose prescription limit and upper dose prescription limit) as follows: The lower point (lower dose prescription limit) corresponds to the minimum dose which should be received in the metastasis (target) under treatment (when following the treatment plan). It is usually prescribed to a volume of 98%-100% of the target volume. The upper dose prescription point (upper dose prescription limit) is a surrogate for the maximum dose, which is usually expressed as the minimum dose received by the 1%-5% of the target volume receiving the highest dose values. The definition of the dose range (e.g. by defining the lower dose prescription limit and upper dose prescription limit) allows clinicians (users) to carefully design iso-dose line prescriptions utilizing homogeneous/inhomogeneous dose distributions. In other words, a user may specify the shape of the function H3 shown in FIG. 4 for one or more targets by defining the points P3a and P3b for each of the one or more targets. As explained above, the use of margins (degree of freedom (4)) influences the spatial dose distribution as shown in FIG. 3. The spatial dose distribution influences the shape of the respective function in the cumulative dose-volume histogram (e.g. the shape of function H3 shown in FIG. 4). Therefore, the treatment plan is determined by optimizing the margins (degree of freedom (4)) to drive homogeneity/inhomogeneity of the dose distribution.

(40) (B) Risk-Structure Sparing Optimization

(41) For each identified risk-structure, a risk structure dose limit can be specified by the operator. A stochastic optimizer iteratively explores fields which can be blocked (i.e. irradiation areas which can be blocked) in order to achieve the provided risk structure dose limit, while satisfying the prescription doses (upper and/or lower dose prescription limit) to the metastases (targets) as reasonably achievable.

(42) An instance of the degrees of freedom (3) and (4) is called an arc configuration in the remainder of this document.

(43) An objective function is used to express the “goodness” (the degree to which an auxiliary treatment plan matches the criteria to be fulfilled by the treatment plan) of a given dose distribution (of an auxiliary treatment plan) during all optimization stages. The function is composed (e.g. as weighted sum) of the following factors: i. (e.g. quadratic) deviation of lower dose prescription point (per metastasis) (first difference as described by the first difference data) ii. (e.g. quadratic) deviation of upper dose prescription point (per metastasis) (second difference as described by the second difference data) iii. Gradient index (e.g. determined based on A simple dose gradient measurement tool to complement the conformity index (Ian Paddick, M. Sc., and Bodo Lippitz, M. D., in J Neurosurg (Suppl) 105:194-201, 2006)), defined as the relative volume of normal tissue dose outside of the treated target volume exceeding a dose level (per metastasis) (relation between first volume and second volume as described by the gradient index data) iv. (e.g. quadratic) deviation of risk structure constraint (per risk structure constraint) (third difference as described by the third difference data) v. Total monitor units (sum of all arc weights as described by the total arc weight data)

(44) Note that several targets and/or risk structure may need to be assessed. This may be done by minimizing an objective function for each of the targets and/or risk structures individually or by combining each of the targets and risk structures in the objective function at the same time. For example, factors i. and ii. may be weighted between all targets and factor iv. may be weighted between all risk structures. The objective function is to be minimized. For example, the value of the objective function for a first auxiliary treatment plan is lower than the value of the objective function of a second treatment plan. In this case, the first auxiliary treatment plan has a higher degree of “goodness” (e.g. has a higher value of the degree to which the auxiliary treatment plan matches the criteria to be fulfilled by the treatment plan).

(45) Factors i. and ii. articulate the deviation to the homogeneous/inhomogeneous prescription. For example, factor i. defines the first difference as described by the first difference data whilst factor ii. defines the second difference as described by the second difference data.

(46) Factor iii. indicates the normal tissue volume exceeding a threshold dose (e.g. in the form of a gradient index). A meaningful threshold may contain a range of high dose-values, which may be defined relatively to the lower dose prescription point (the lower dose prescription limit). A threshold of 50-90%, for example 80%, of the lower dose prescription limit can be chosen as the threshold dose. As described above, a first predetermined sum and a second predetermined sum may be used to determine the relation between the first volume and the second volume which is described by the gradient index data (correlated with factor iii.). The first predetermined sum can be equal to 50-90%, for example 80%, of the lower dose prescription limit.

(47) Factor iv. penalizes violated risk-structure constraints. For example, the deviation of risk structure constraint is expressed as third difference as described by the third difference data.

(48) Factor v. influences treatment time and/or efficiency. For example, a beam source (e.g. LINAC) needs more time to deliver more (a greater number of) monitor units. Treatment time and (number of) monitor units are (very roughly) proportional. The total (number of) monitor units are for example the sum of all arc weights as described by the total arc weight data of the auxiliary treatment plan.

(49) At least the factors ii., iii. and iv. were not included in the previous solutions for multiple brain metastases, but are required to implement the features (A) and (B) mentioned above.

(50) The degrees of freedom (1) to (4) described above result in a large search space. Many combinations of these degrees of freedom are possible for a treatment plan. To make searching (determining the treatment plan, e.g. by selecting one of the auxiliary treatment plans as the treatment plan) feasible, a fast arc-weight optimization algorithm is used to optimize degree of freedom (2) independently for a given arc configuration (for a given instance of the degrees of freedom (3) and (4)): At several arc configuration optimization stages (each stage defined by a generated set of auxiliary treatment plans, e.g. a first optimization stage for the auxiliary treatment plans, a second optimization stage for the secondary auxiliary treatment plans and so on), the individual arc-weights (the arc-weights of the individual paths (arcs) of the (secondary, tertiary, . . . ) auxiliary treatment plan; the arc-weights are for example each expressed as a sum of (the number of) monitor units) are optimized stochastically by minimization of the aforementioned objective function.

(51) To enable fast dose computation, a dose-influence approach is used. It is assumed that the total dose (e.g. described by the total irradiation dose data) can be composed linearly as sum of all individual subfield doses (the sum of all simulated irradiation doses received by (one or more of/all of) the voxels of (generated from) the patient image data (e.g. the voxels of a target) when following the (secondary, tertiary . . . ) auxiliary treatment plan, i.e. the sum of all simulated irradiation doses for all control points of the (secondary, tertiary . . . ) auxiliary treatment plan). This is related to the “beamlet approach”, which is widely used for treatment plan optimization. However, instead of dose computation for rectangular subfields, the dose contribution is computed for a set of target volume projections (irradiation areas based on auxiliary outlines) for various margins.

(52) Prior to the arc configuration optimization, dose contributions are pre-computed (the irradiation data is determined) for each metastasis (target), for each margin setting to be explored (margins), for each control point of an arc (path). The optimization algorithm allows for evaluation of full dose distributions for any arc configuration by fast addition of a set of these dose contributions. The algorithm is able to compute doses (determine the irradiation data) and evaluate the objective function (e.g. determine the rating data and select one of the auxiliary treatment plans) thousands of times per second by exploiting parallel computing.

(53) During arc configuration optimization, regular recalibration is advantageous to maintain a stable optimization result. This is for example implemented by computing the dose for a whole arc and comparing it to the approximated dose contribution.

(54) FIG. 9 illustrates the steps of an example of the method according to the first aspect. The algorithm for multiple metastases treatment plan optimization (determining the treatment plan) according to this example of the method is summarized below with reference to FIG. 9.

(55) In step S9.1, all metastases (targets) are distributed to a preset number of arcs (paths). The algorithm for this target distribution can be adapted from commonly known methods. The metastases (targets) are for example selected such that unnecessary leaf gaps (distances between collimator leaves and the (auxiliary) outlines) are avoided to reduce the normal tissue dose. The metastases (targets) are for example selected such that the number of patient support device angles (relative positions between the patient support device and the irradiation direction) per metastasis (target) is maximized for optimal dose conformity.

(56) Instead of a brute-force approach, a stochastic search strategy is used to make the arc configuration optimization feasible for a large number of metastases (targets) and/or arcs (paths).

(57) After determination of the metastasis-to-arc distribution, the irradiation data can be determined for individual control points of the arcs and for several margins as described above (step S9.2). To improve computation speed, this process is e.g. parallelized over control points. The result is for example stored in a main memory (e.g. a transitory or a non-transitory storage medium).

(58) An arc configuration optimization loop explores opening and closing (blocking) of fields per metastasis (target) per control point and various margins (step S9.3 comprising sub-steps S9.3a, S9.3b and S9.3c).

(59) After determination of the optimal blockings/margins and arc-weights, final arcs are constructed and the respective dose contributions (irradiation doses, e.g. as described by the sum of all simulated irradiation doses received by the at least one voxel of the patient image data for each individual arc when following the auxiliary treatment plan/when using the determined optimal blockings, margins and arc-weights) are computed (S9.4). For example, the total irradiation dose data is determined for each of the arcs (paths) individually in this step, i.e. the total irradiation dose data describes the sum of all simulated irradiation doses received by at least one voxel of the patient image when using the selected treatment plan, for each of the paths of the selected treatment plan individually (dose per arc).

(60) Subsequently, a final arc-weight optimization is performed (step S9.5) to fine-tune the final dose distribution, using the same search algorithm and objective function as above. During this step, the arc-weights are optimized once more. However, the leaves (and other machine parameters) are not changed (i.e. the blockings and margins are kept constant). This ensures that the dose determined in step S9.4 remains accurate during this step.

(61) Finally, the treatment plan is generated in step S9.6 (e.g. determined) and available for evaluation, modification, saving and export.

(62) Finding the optimal clipping and margin configuration to optimize the arc configuration (S9.3) can be formulated as a stochastic optimization problem, which iteratively proposes random configurations based on the previous best result.

(63) To improve the convergence (and hence runtime) of the algorithm, new configurations can be proposed in sub-step S9.3a heuristically based on deviations from the lower and upper dose prescription points (lower and upper dose prescription limits): If the dose (as described by the target dose data) in the lower and upper prescription point (the lower and the upper dose prescription limits) of a metastasis (target) is exceeded (e.g. by a threshold dose tolerance t1), then a field (an irradiation area of a given control point) can be blocked (blocking). If dose (as described by the target dose data) in the upper prescription point (upper dose prescription limit) of a metastasis (target) is exceeded (by a threshold dose tolerance t2) and the lower prescription point (lower dose prescription limit) is satisfied (within a threshold dose tolerance t3), then a margin can be increased (a larger margin can be used for some or all of the control points, e.g. when generating the next auxiliary treatment plans). If dose (as described by the target dose data) in the upper prescription point (the upper dose prescription limit) is less than prescribed (by a threshold dose tolerance t4) and the lower prescription point (lower dose prescription limit) is satisfied (within a threshold dose tolerance t5), then a margin can be decreased (a smaller margin can be used for some or all of the control points, e.g. when generating the next auxiliary treatment plans). If dose (as described by the target dose data) in the lower prescription point (lower dose prescription limit) of a metastasis (target) is exceeded (by a threshold dose tolerance t6) and the upper prescription point (upper dose prescription limit) is satisfied (within a threshold dose tolerance t7), then a field (an irradiation area of a given control point) can be blocked (blocking) and/or a margin can be decreased (a smaller margin can be used for some or all of the control points, e.g. when generating the next auxiliary treatment plans).

(64) A value of t1, t2, t3, t4, t5, t6 and/or t7 may be picked from the range 1-5%, for example 2%. This value represents a trade-off between good treatment plans and optimization time. Only considering heuristic configurations may result in the optimizer getting stuck in local minima. Therefore, configurations based on random margin changes might be proposed in addition to the heuristic configurations.

(65) Changing an arc configuration requires re-optimization of the arc-weights. To take advantage of multi-core CPU architectures, a new algorithm is used to optimize arc-weights quickly for several proposed configurations in parallel (sub-step S9.3b). The proposed configurations from sub-step S9.3a are sorted by dose deviation (e.g. using the first difference, the second difference) and added to a priority queue.

(66) This queue is subdivided in batches of a fraction of the number of threads. For N threads and M proposed heuristic configurations, another N-M random configurations are added. The ratio between heuristically and randomly proposed configurations depends on the total number of threads: for example at least one random configuration is proposed. For the initial phase of the arc configuration optimization, a 1:1 ratio between heuristically and randomly proposed configurations ratio can be expected to yield good results. However, other ratios may be used.

(67) Arc-weight optimization is started in parallel for resulting batches of N configurations based on rating data (e.g. rating data is determined for the proposed configurations which represent the auxiliary treatment plans, for example the rating data is determined based on the aforementioned objective function). For example, the target dose data is determined and the rating data is determined on the first difference and/or the second difference. For example, the normal tissue dose data is determined and the rating is determined on the relation described by the gradient index data (which is for example determined based on the normal tissue dose data). For example, risk structure dose data is determined and the rating data is determined based on the third difference.

(68) If an objective function improvement is found in a batch, the rest of the priority queue is neglected. If not, the rest of the subdivided queue is optimized consecutively. For example, if no overall improvement of the objective function could be found, the optimization problem is considered converged.

(69) After the algorithm converges (the optimization algorithm of step S9.3 including the sub-steps S9.3a, S9.3b and S9.3c), it is restarted (not indicated in FIG. 9) with a different ratio (refinement step). The converging of the optimization algorithm may be determined by selecting the proposed configuration with the overall smallest objective function value from the proposed configurations (e.g. selection one of the auxiliary treatment plans). In case this objective function value changes from one iteration to the other less than a predetermined threshold, it is determined that the algorithm has converged.

(70) As no heuristic configurations can be proposed (optimization algorithm has converged, i.e. no better configurations can be proposed heuristically), a 100% randomly changed configuration batch is started. This might introduce new deviations from the lower and upper dose prescription point (lower and upper dose prescription limit) and hence new heuristic configuration changes become available. As it is expected that the number of heuristic configurations is small at this stage, the ratio can be increased in favor of the randomly proposed configurations (e.g. to a ratio of 1:3 between heuristically and randomly proposed configurations). The number of restarts is for example restricted to limit optimization time. For example, the number of restarts is restricted to 2 to yield good results.

(71) For each configuration, the respective pre-computed irradiation doses described by the irradiation data are summed over each of the arcs (paths) of the determined treatment plan (the selected configuration) in sub-step S9.3b in a cache friendly manner. Arc-weights can afterwards be optimized stochastically in sub-step S9.3b. For each arc-weight combination under consideration, the dose in the prescription points is computed (e.g. the target dose), along with normal tissue dose and all terms of the objective function (e.g. risk structure dose).

(72) The determined irradiation data is independent on the lower/upper dose prescription limits and must be computed only once. Therefore, after initial optimization, the operator can interact with the optimization result: the operator is given the flexibility to explore several prescriptions in terms of dose and volume. In this case steps S9.1 and S9.2 are omitted. A new graphical user interface solution can be implemented to support an interactive planning workflow.