Multi-target treatment planning and delivery and virtual position verification for radiotherapy

Abstract

This specification discloses methods for patient setup and patient target region position matching for irradiation of multiple patient target regions in a single treatment session. Virtual position matching is a method that can be used to align patient target regions without requiring the patient to be physically moved using a patient table. Instead, the planned fluence is updated to reflect the current location of the patient target region by selecting a position matching criterion in the position matching image, calculating a position matching function based on the position matching criterion point, and calculating a delivery fluence by convolving the position matching function with a shift-invariant extrusion filter. Mosaic multi-target position matching divides a planned fluence map for multiple patient target regions into sub-regions that can be individually position-matched.

Description

Background Art

[0001] Cross - reference to Related Applications This application claims priority to U.S. Provisional Patent Application No. 62 / 873,742, filed Jul. 12, 2019, the entire disclosure of which is incorporated herein by reference.

[0002] (Background Art) Radiation therapy includes obtaining planning images of a patient and one or more tumor regions, and designing a treatment plan to deliver a prescribed dose of radiation to the tumor regions. In addition to identifying the tumor regions to be irradiated, radiation therapy attempts to limit irradiation of healthy tissue. This may include identifying radiosensitive regions or radiation - avoiding regions (e.g., risk organs or OARs), and designing a treatment plan that does not irradiate these radiation - avoiding regions beyond what is prescribed. In most cases, the treatment plan is such that the absolute and / or relative positions of one or more tumor regions, OARs, and healthy tissue regions during treatment match (or approximately match) their absolute and / or relative positions at the time the planning images were obtained. Once a treatment plan that meets the dosimetric goals of the tumor regions and OARs is designed, the patient can be positioned on the patient table or couch of the radiation therapy system such that the patient is moderately comfortable on the couch and in a position suitable for radiation delivery. One or more position - verification images of the patient on the couch (e.g., fan - beam CT scans, cone - beam CT scans, and / or PET scans) can be obtained. The location of each of the one or more tumor regions, OARs, and / or other anatomical structures can be determined using the position - verification images and their absolute positions relative to the treatment radiation source, and / or their relative positions to each other can be compared to their locations from the treatment planning images. If any of the tumor regions and / or OARs are displaced, the radiation therapy system can calculate how the patient should shift to align their position during the treatment plan and adjust the couch to execute that shift, if possible.

[0003] However, patient setup and alignment of the tumor region and / or OAR with respect to the coordinates of the radiation therapy system can be time-consuming, especially since the size, shape, and / or location of the tumor region and / or OAR can change between acquisition of the planning image and the treatment session and, in some cases, continue to change during the treatment session. In addition, patient setup and alignment are typically optimized for treatment of a single target region and may not easily accommodate treatment of multiple target regions. For example, one patient setup arrangement and alignment may be suitable for one tumor region but may not enable delivery of the prescribed dose for another tumor region and / or may result in unnecessary irradiation of the OAR. Accurately positioning a patient for a plan designed to treat multiple targets simultaneously involves mapping the new actual locations of the patient's multiple targets at delivery to the initial locations of the multiple targets in the treatment plan. Since tumors can displace relative to each other, there may be no single translational and orientation shift of the patient that can align all tumors to the plan. That is, adjustment, shifting, and rotation of the patient to align one tumor to its expected location in the plan may cause the remaining tumors to shift out of alignment.

[0004] One approach for delivering radiation to multiple patient target regions involves performing a series of patent setups and registrations. For example, for a first patient target region, the couch can be adjusted to align the patient at a setup position suitable for radiation delivery to the first target region, a registration image can be acquired to register the first target region, and then radiation can be delivered to that first target region. These steps are then repeated for each target region, which can be time-consuming (especially if the patient setup involves opening a bunker door so that the clinician or technician can access the adjustment of the patient on the couch), and since each iteration is processed individually, it can result in over-irradiation of OARs and / or healthy tissue and / or under-irradiation of the target regions (e.g., so as not to over-irradiate the OARs). Therefore, improvements in radiation treatment planning, patient setup and alignment, and registration methods are desirable, particularly for the irradiation of multiple target regions. Summary of the Invention Problems to be Solved by the Invention

[0005] Disclosed herein are methods for treatment planning, patient setup, registration, and radiation delivery for the irradiation of multiple patient target regions in a single treatment session or fraction. Means for Solving the Problems

[0006] A method for treatment planning for the irradiation of multiple target regions can include selecting a planned registration fiducial point for each target region and calculating a shift-invariant projection filter for each target region based on the selected planned registration fiducial point. A treatment planning fluence map specifying the radiation fluence delivered to the multiple target regions and the radiation levels to the OARs and healthy tissue can be calculated based on the shift-invariant projection filter and the planned registration fiducial points for each target region.

[0007] Additionally or alternatively, a shift-invariant injection filter may be represented as a functional shift and orientation operator of the radiation fluence delivered to a plurality of target regions based on the planned position registration fiducial points of each target region. For example, the functional shift operator may include a function that shifts and interpolates a two-dimensional or three-dimensional map. The functional orientation operator may include, for example, a function that interpolates a two-dimensional or three-dimensional map using rotations along axes in a two-dimensional or three-dimensional space. For example, a functional orientation operator may be used to interpolate (e.g., transform) a two-dimensional (or three-dimensional) fluence map for delivery based on the location of the position registration fiducial points on the treatment day. This may include calculating the translation of the fluence map in (x,y) coordinates, and / or scaling the fluence map in (x,y,z), and / or rotating the fluence map along axes in (θ,φ,ρ). In some variations, the treatment planning method may include defining a periphery including regions of low fluence values around each patient target region in the treatment planning fluence map. The treatment planning method for a plurality of target regions may include calculating individual treatment plans including individual fluence maps for each target region, combining the individual fluence maps into a cumulative treatment planning fluence map, and iteratively modifying the cumulative treatment planning fluence map based on one or more constraints (e.g., dose constraints such as keeping high fluence regions separated from each other, meeting OAR fluence limits, meeting the dose target for each target region). The treatment planning method may include, after the cumulative treatment planning fluence map has been iteratively modified to comply with the desired constraints, defining a periphery including regions of low fluence values around each patient target region in the treatment planning fluence map, and optionally, separating the iteratively modified cumulative treatment planning fluence map into individual fluence maps for each patient target region along the defined periphery. In some variations, during optimization of the treatment plan, one or more of the functional shift operator and the orientation operator may be used to calculate a fluence map based on the planned position registration fiducial points of each target region.

[0008] Optionally, in a biologically-guided radiotherapy (BgRT) in which positron annihilation emission path data is acquired during a treatment session and used to calculate a delivery fluence, a treatment planning method may include defining, for a given range of location displacements of first and second patient target regions, a first region of interest (ROI) or radiation emission zone surrounding the first patient target region and a second ROI or radiation emission zone surrounding the second patient target region. The output of the treatment planning method described herein may include a treatment planning fluence map defining the fluence delivered to each patient target region, as well as the levels of radiation exposure of OARs and healthy tissues, and a set of shift-invariant emission filters for each planned position registration reference point associated with each patient target region. Alternatively or in addition, in the treatment planning of BgRT, the treatment planning method may output a set of shift-invariant emission filters associated with an ROI (e.g., a biological emission zone (BFZ)) defined around each patient target region. Additionally or alternatively, the set of shift-invariant emission filters may include a general function shift operator and / or an orientation operator. Optionally, the output of the treatment planning method may include a set of radiotherapy machine instructions for execution by a radiotherapy system after patient setup and patient target region registration.

[0009] In some variations, the method for position matching of the virtual target region and radiation delivery includes obtaining an image of the patient in the treatment position, identifying the patient target region in the obtained image, and selecting a position matching reference point within the obtained image, wherein the reference point corresponds to the planned position matching reference point; calculating a spatial offset based on the displacement between the position matching reference point and the planned position matching reference point; shifting the boundary of the planned region of interest based on the spatial offset, wherein the boundary of the planned region of interest surrounds the patient target region; obtaining imaging data spatially filtered by the shifted region of interest; calculating the fluence for delivery to the patient target region at each emission position of the treatment radiation source by convolving a set of emission filters with the obtained imaging data; and emitting the calculated fluence to the patient target region using the treatment radiation source.

[0010] In some variations, the boundary of the planned region of interest may include a spatial filter. In some variations, shifting the boundary of the planned region of interest may include applying rotation and shift to the planned region of interest by a roll correction coefficient φ representing the rotational translation of the position matching reference point relative to the planned position matching reference point. Calculating the fluence for delivery may include circularly convolving a set of emission filters with the roll correction coefficient φ.

[0011] In some variations, a method for treatment planning for radiation delivery to multiple patient target regions includes identifying a first location of a first patient target region and a second location of a second patient target region in a patient planning image, defining a first region of interest having a boundary surrounding the first patient target region, and defining a second region of interest having a boundary surrounding the second patient target region. The boundary of the second region of interest can be selected to surround the second patient target region for a predetermined range of location displacements of the first and second patient target regions. The first and second regions of interest can represent a spatial filter configured to select imaging data acquired during a treatment session. The method can further include calculating a treatment plan fluence map for the first patient target region and the second patient target region that specifies the fluence to be delivered if the selected imaging data indicates that the first patient target region is within the boundary of the first region of interest and the second patient target region is within the boundary of the second region of interest.

[0012] In some variations, the first patient target region may be closer to a planning structure. In some variations, the first region of interest may be smaller than a second beam zone. In some variations, the planning structure may be an organ at risk (OAR). In some variations, the first patient target region may be closer to two or more planning structures than the second patient target region. In some variations, the method may further include designating a treatment planning reference point within the first region of interest as a positioning reference point for aligning the patient at the start of a treatment session. In some variations, the selected imaging data may be used to guide radiation delivery during a treatment session. In some variations, the treatment planning reference point may be the center point of the first patient target region.

[0013] In some variations, the imaging data may include positron annihilation emission path data. In some variations, the positron annihilation emission path may be selected when the positron annihilation emission path intersects at least one of the first and second injection zones. In some variations, the second region of interest may be formed to a size that includes a location within a range of the second patient target region. In some variations, the imaging data may include one or more of positron emission tomography (PET) imaging data, computed tomography (CT) imaging data, and magnetic resonance imaging (MRI) imaging data.

[0014] Also disclosed herein are methods for patient setup for irradiation of multiple target tissue regions and patient target region localization. The methods for patient setup and target alignment / localization may include positioning the patient on a patient table or couch of a radiotherapy system, acquiring one or more localization images of the patient including one or more target regions and / or OARs, and selecting localization fiducial points within the one or more acquired localization images. The selected localization fiducial points may correspond to the planned localization fiducial points of a particular patient target region and may represent the updated location of the patient target region. A fluence map for delivery to a particular patient target region may be calculated by calculating a localization function based on the selected localization fiducial points and applying the localization function to a shift-invariant emission filter derived based on the planned localization fiducial points. Alternatively or in addition, the shift-invariant emission filter may be derived from other planning parameters. For example, a fluence map for delivery to a patient target region may be calculated by calculating a delta function (e.g., a 3D discrete delta function) based on the selected localization fiducial points and convolving the delta function with a shift-invariant emission filter derived based on the planned localization fiducial points and / or other planning parameters. In another example, a fluence map for delivery to a patient target region may be calculated by interpolating the fluence map with a shift or rotation based on the selected localization fiducial points. In such a manner, the fluence map for delivery may be updated at the current location of the patient target region and may not require adjusting the patient's position using the couch. The localization of a patient target region by calculating a fluence map based on a set of a localization function based on the selected localization fiducial points (e.g., a delta function or a Gaussian function centered on the selected localization fiducial points) and a shift-invariant emission filter may be referred to as “virtual localization”.In some variations, while the patient may be set up on the couch once, acquisition of the registration image, selection of the registration reference points, and calculation of the fluence map for delivery may be performed for each patient target region (e.g., one physical registration / setup, multiple registration image scans, multiple virtual registrations corresponding to multiple patient target regions). Alternatively, the registration image may be acquired once (e.g., when the patient is first aligned with the radiotherapy system), and selection of the registration reference points and calculation of the delivery fluence map may be performed for each patient target region (e.g., one physical setup, one registration image scan, multiple virtual registrations corresponding to multiple patient target regions using the initially acquired registration image).

[0015] In contrast, conventional treatment delivery methods involve obtaining a registration image (e.g., a registration CT image) for physically aligning the patient, and a patient target region on the patient table that coincides with the location of the patient target region determined during treatment planning. To treat multiple patient target regions in a single treatment session, it may be necessary to physically realign the patient before irradiating each patient target region, since a particular physical location of the patient that may be suitable for irradiating one patient target region may not be suitable for irradiating another patient target region. For each realignment, an additional registration image is obtained. In some variations, the alignment and realignment of the patient may be provided by the patient table. Further, some treatment planning systems may be configured to output machine instructions (e.g., leaf patterns), which may then be varied according to the registration image. However, the treatment planning systems and methods described herein are configured to output a fluence map, and / or an image of registration fiducial points and / or a delivery system's injection filter for convolution with a registration function to generate a delivery fluence map. This may enable treatment of multiple patient target regions without the need to realign the patient for each target region. The delivery fluence map may be segmented into machine instructions by the delivery system immediately prior to dose delivery. Thus, virtual registration may eliminate or reduce the number of times a registration image is obtained and the patient is physically aligned (and / or realigned) for treatment of multiple patient target regions.

[0016] Instead of or in addition to virtual position verification, a method for patient setup may include obtaining images of a first patient target region and a second patient target region. A first set of patient position shift vectors may be calculated based on the acquired image of the first patient target region and a treatment plan image. A second set of patient position shift vectors may be calculated based on the acquired image, a treatment plan image of the second patient target region, and the first set of patient position shift vectors. The patient may be aligned to a first location and / or position according to the first set of patient position shift vectors. The patient may be aligned at a second location and aligned according to the second set of patient position shift vectors. Aligning the patient according to the first set of patient position shift vectors may include moving a radiotherapy patient table and / or a treatment radiation source according to the first set of patient position shift vectors. The patient may be aligned at a second location and adjusted according to the second set of patient position shift vectors. Aligning the patient according to the second set of patient position shift vectors may include moving a radiotherapy patient table and / or a treatment radiation source according to the second set of patient position shift vectors. Moving the radiotherapy patient table may include moving the table along the X-axis, and / or Y-axis, and / or Z-axis of the table. Moving the radiotherapy patient table may include adjusting the yaw and / or pitch and / or roll of the table, and moving the treatment radiation source may include adjusting the yaw / pitch / roll of the gantry to which the treatment radiation source is coupled. The first treatment plan image and the second treatment plan image may be the same treatment plan image. The acquired image may be a PET (positron emission tomography) image. The acquired image may be a CT (computed tomography) image. The acquired image may be an MRI (magnetic resonance imaging) image. In some embodiments, the calculation of the first and second sets of patient position shift vectors may be performed before the treatment radiation source is activated. In some embodiments, a first location difference may be calculated by comparing the location of the first patient target region in the acquired image with the location of the first patient target region in the first treatment plan image.The second location difference can be calculated by comparing the location of the second patient target region in the acquired image with the location of the second patient target region in the second treatment planning image. When the first location difference or the second location difference exceeds a location difference threshold, a notification can be generated. The first patient target region and the second patient target region can include one or more tumor regions. The first patient target region can include a first portion of the tumor, and the second patient target region can include a second portion of the tumor. The first and second sets of position shift vectors can include translations of distance and direction. The translation of direction can include a tilt angle.

[0017] A method for delivering radiation fluence to a plurality of patient regions can include converting (alternatively, segmenting) a delivery fluence map into radiation therapy system machine instructions (e.g., treatment radiation source radiation emission parameters, multi-leaf collimator configurations for each injection position around the patient table, etc.), and then executing the machine instructions by emitting radiation fluence from the treatment radiation source to the patient. The delivery fluence map for a plurality of patient target regions can be a combination of individual fluence maps calculated during virtual position registration for each of the patient target regions, and the machine instructions for delivering to two or more (e.g., all) of the patient target regions can be segmented simultaneously. Alternatively or in addition, the individual fluence maps for each of the patient target regions can be segmented separately (e.g., sequentially). Optionally, in BgRT where positron annihilation emission path data is acquired during the treatment session and used to calculate the delivery fluence, the delivery fluence map can be segmented into radiation therapy system machine instructions in real time, e.g., within 500 ms or less between the acquisition of the emission path data and the delivery of the treatment radiation beam. The delivery fluence map can be a fluence map calculated from the emission path data, and the virtual position registration can be achieved by shifting or rotating the ROI.

[0018] A method for virtual target area location verification and radiation delivery may include obtaining an image of a patient in a treatment position, identifying a patient target area in the obtained image, selecting a registration reference point in the obtained image, where the registration reference point corresponds to a planned registration reference point, calculating a registration function based on the registration reference point to calculate a fluence for delivery to the patient target area at each emission position of a treatment radiation source, applying the registration function to a shift-invariant emission filter derived based on the planned registration reference point, and emitting the delivery fluence to the patient target area using the treatment radiation source. The registration function may be a delta function, and applying the registration function to the shift-invariant emission filter may include convolving the delta function with the shift-invariant emission filter. Additionally or alternatively, the shift-invariant emission filter may include one or more general function shift operators and orientation operators for the radiation fluence delivered to a plurality of target areas based on the planned registration reference points of each target area. For example, as described above, the general function shift operator may include a function for shifting and interpolating a two-dimensional or three-dimensional map. The orientation operator may include a function for interpolating a two-dimensional or three-dimensional map using rotation. For example, the interpolation may include one or more of linear interpolation, nearest neighbor interpolation, bicubic interpolation, spline interpolation, or Fourier shift interpolation. The registration reference point may be a user-selected location in the obtained image and / or may correspond to a treatment plan isocenter defined for the patient target area during treatment planning. The first patient target area may be in a first treatment area of the patient defined during treatment planning, and the second target area may be in a second treatment area of the patient defined during treatment planning. The first treatment area may include a first set of axial planes and may have an axial length of about 8 cm or less. The second treatment area may not overlap with the first treatment area and may have an axial length of about 8 cm or less and may include a second set of axial planes. The centers of the first treatment area and the second treatment area may be, for example, on the same straight line along the IEC-Y axis and / or in the same plane as the IEC-Y axis. In some variations, the first treatment area may overlap with the second treatment area.

[0019] The emission position of the therapeutic radiation source includes the position of the therapeutic radiation source relative to the position of the patient table. The therapeutic radiation source can be mounted on a gantry rotatable about a vertical axis, and the position of the therapeutic radiation source can be specified by the gantry angle about the vertical axis. The patient table can be movable to various positions along the vertical axis.

[0020] In some variations, the therapeutic radiation source may be mounted on the gantry such that radiation can be delivered to the patient in discrete contiguous arcs (e.g., non-coplanar VMAT, 4-pi VMAT). In some of these variations, these radiation delivery arcs may not be in the same plane as each other (e.g., perpendicular to the vertical axis). The emission position may be specified as discrete positions along each arc and may be indexed by the arc. The patient table may be configured to reposition to a predetermined position and orientation between each arc.

[0021] In some variations, the therapeutic radiation source may be mounted on the gantry such that radiation can be delivered at fixed points around the perimeter of a 4π or 2π hemisphere envelope. In some of these variations, a set of emission positions may be specified as discrete positions based on a two-dimensional fluence map. The patient table may be configured to move the patient to a predetermined set of positions and / or orientations between each fixed point.

[0022] The method may further include identifying a second patient target region in the acquired image, selecting a second registration fiducial point for the second patient target region within the acquired image, wherein the second registration fiducial point corresponds to a second planned registration fiducial point, calculating a second delivery fluence at each injection position of the treatment radiation source by calculating a second registration function based on the second registration fiducial point, applying the second registration function to a second shift-invariant injection filter based on the second planned registration fiducial point, and emitting the second delivery fluence to the second patient target region using the treatment radiation source. The method may optionally include identifying a second patient target region in the acquired image, selecting a second registration fiducial point for the second patient target region within the acquired image, wherein the second registration fiducial point corresponds to a second planned registration fiducial point, calculating a second delivery fluence at each injection position of the treatment radiation source by calculating a second delta function based on the second registration fiducial point, convolving the second delta function with a second shift-invariant injection filter based on the second planned registration fiducial point, and emitting the second delivery fluence to the second patient target region using the treatment radiation source. Emitting the first delivery fluence to the first patient target region and emitting the second delivery fluence to the second patient target region may be performed simultaneously and / or sequentially.

[0023] In some variations, the method for virtual target region alignment and radiation delivery may include the steps of obtaining an image of a patient in a treatment position, identifying a patient target region in the obtained image, selecting an alignment reference point within the obtained image, where the alignment reference point corresponds to a planned alignment reference point, calculating a fluence for delivery to the patient target region at each emission position of a treatment radiation source by calculating an alignment function based on the alignment reference point, applying the alignment function to a shift-invariant emission filter derived based on the planned alignment reference point, and emitting the delivery fluence to the patient target region using the treatment radiation source.

[0024] In some variations, applying the alignment function to the shift-invariant emission filter may include convolving the alignment function with the shift-invariant emission filter. In some variations, the alignment function may be one of a delta function, a Gaussian function, a cycle function, and an interpolation. In some variations, the Gaussian function may be a truncated Gaussian function. In some variations, the interpolation may be one of linear, bicubic, spline, or Fourier shift. In some variations, the obtained image may include one or more of a positron emission tomography (PET) image, one or more X-ray projection images, a computed tomography (CT) image, and a magnetic resonance imaging (MRI) image.

[0025] In some variations, this method may further include identifying a second patient target region in the acquired image and selecting a second registration reference point for the second patient target region within the acquired image. The second registration reference point may correspond to a second planned registration reference point. This method may further include calculating a second delivery fluence at each injection position of the therapeutic radiation source by calculating a second registration function based on the second registration reference point, applying the second registration function to a second shift-invariant injection filter based on the second planned registration reference point, and emitting the second delivery fluence to the second patient target region using the therapeutic radiation source.

[0026] In some variations, this method may further include identifying a second patient target region in the acquired image and selecting a second registration reference point for the second patient target region within the acquired image. The second registration reference point may correspond to a second planned registration reference point. This method may further include calculating a second delta function based on the second registration reference point and convolving the second delta function with a second shift-invariant injection filter based on the second planned registration reference point to calculate a second delivery fluence at each injection position of the therapeutic radiation source, and emitting the second delivery fluence to the second patient target region using the therapeutic radiation source. In some variations, emitting the first delivery fluence to the first patient target region and emitting the second delivery fluence to the second patient target region are performed simultaneously. In some variations, emitting the first delivery fluence to the first patient target region and emitting the second delivery fluence to the second patient target region are performed sequentially.

[0027] In some variations, the registration function is the first registration function (δ). The first shift-invariant injection filter is a first set of fluence map filters (p1, p2,..., p calculated during treatment planning for each injection position (i) of the therapeutic radiation source i) may be included. This method involves calculating the first set of projections (δ i ) of the position matching function to each injection position (i), δ i = proj i (δ) may be included. Each projection (δ i ) can be a 2D fluence distribution. Calculating the first fluence for delivery involves convolving each projection (δ i ) in the first set of projections of the first position matching function with the corresponding fluence map filter (p i ) to calculate the first delivery fluence map (f i ) for each injection position (i) of the therapeutic radiation source, f i = p i * δ i may be included. Delivering the first fluence may involve moving the therapeutic radiation source to each injection position (i) and emitting radiation to the first patient target area according to the first delivery fluence map (f i ). In some variations, each projection (δ i ) may be an m×n matrix, where m is the number of multi-leaf collimator leaves and n is a number selected during treatment planning. In some variations, n is the number of beam stations selected during treatment planning. In some variations, the first position matching function may be one of a delta function, a Gaussian function, a cycle function, and interpolation. In some variations, the Gaussian function may be a truncated Gaussian function. In some variations, the interpolation may be one of linear, bicubic, spline, and Fourier shift.

[0028] In some variations, the second shift-invariant injection filter may include a second set of fluence map filters (p_21, p_22,..., p_2 i ) calculated during treatment planning for each injection position (i). This method involves calculating the second set of projections (δ_2 i ) of the second position matching function to each injection position (i), δ_2 i = proj i (δ_2) may be included, where each projection (δ_2i ) is a 2D fluence distribution. Calculating a second fluence for delivery involves, for each projection (δ_2 i ) during the projection of a second set of second position matching functions, convolving with a corresponding fluence map filter p_2 i to calculate a second delivery fluence map (f_2 i ) for each injection position (i), f_2 i = p_2 i * δ_2 i . Delivering the second calculated fluence may involve moving a therapeutic radiation source to each injection position (i) and emitting radiation to a second patient target region according to the second delivery fluence map (f_2 i ). In some variations, each projection (δ_2 i ) may be an m×n matrix. In some variations, m is the number of multi-leaf collimator leaves and n is a number selected during treatment planning. In some variations, n is the number of beam stations selected during treatment planning. In some variations, the second position matching function may be one of a delta function, a Gaussian function, a cycle function, and an interpolation. In some variations, the Gaussian function may be a truncated Gaussian function. In some variations, the interpolation may be one of linear, bicubic, spline, and Fourier shift. In some variations, delivering the calculated fluence may include segmenting the delivery fluence map (f i , f_2 i ) into a plurality of radiation therapy system machine instructions for each injection position.

[0029] In some variations, delivering the calculated fluence may include segmenting the calculated fluence into a plurality of radiation therapy system machine instructions for each injection position. In some variations, delivering the second calculated fluence may include segmenting the second calculated fluence into a plurality of radiation therapy system machine instructions for each injection position.

[0030] In some variations, a plurality of radiation therapy system machine instructions may include one or more multi-leaf collimator configurations for each injection position. Emitting a radiation fluence may further include moving the leaves of the multi-leaf collimator to a multi-leaf collimator configuration corresponding to the location of the injection position of the therapeutic radiation source and emitting a pulse of radiation.

[0031] In some variations, a plurality of radiation therapy system machine instructions may further include therapeutic radiation source emission (e.g., pulse) parameters for each injection position. Emitting a pulse of radiation may include emitting radiation having therapeutic radiation source pulse parameters corresponding to the location of the injection position of the therapeutic radiation source. In some variations, calculating the delivery fluence map (f i , f_2 i ) may further include convolving each projection in the first or second set of projections with a corresponding shift-invariant fluence map filter and applying a virtual flattening filter correction factor (FF), (f i ) = FF · (p i * δ i ); (f_2 i ) = FF · (p_2 i*δ_2i .

[0032] In some variations, the virtual flattening filter correction factor (FF) is an m×n matrix and is the reciprocal of the flatness profile of the radiation beam emitted by the therapeutic radiation source.

[0033] In some variations, calculating the delivery fluence map (f i , f_2 i ) may further include convolving each projection in the set of projections with a shift-invariant fluence map filter and applying the virtual flattening filter correction factor (FF), and a distance compensation factor

Number

[0034] In some variations, d i represents the distance from the injection position i to the center of the patient target region defined during treatment planning,

Number

[0035] In some variations, the registration reference point may be a user - selected location within the acquired image. In some variations, the registration reference point may correspond to a treatment planning isocenter defined for the patient target region during treatment planning. In some variations, the first patient target region may be in the first treatment area of the patient defined during treatment planning, and the second target region may be in the second treatment area of the patient defined during treatment planning. In some variations, the first treatment area has an axial length of about 8 cm or less and includes the first set of axial planes, and the second treatment area has an axial length of about 8 cm or less and does not overlap with the first treatment area. The second treatment area may include the second set of axial planes. In some variations, the centers of the first treatment area and the second treatment area may be on the same straight line along the IEC - Y axis and / or in the same plane as the IEC - Y axis. In some variations, the first treatment area and the second treatment area may overlap. In some variations, the injection position of the treatment radiation source may include the location of the treatment radiation source relative to the location of the patient table. In some variations, the treatment radiation source may be mounted on a gantry rotatable about a vertical axis, and the location of the treatment radiation source may be specified by the gantry angle about the vertical axis. In some variations, the patient table may be movable to various locations along the vertical axis.

[0036] In some variations, the delivery fluence map (f i , f_2 iCalculating ( LOC and δ_2 LOC each projection during the projection of the set with the shift-invariant fluence map filters p' and p_2' convolved with delta functions with angular shifts (φ, φ_2) located away from the isocenter at locations δ i , δ_2 i and further including convolving with the delta functions δ [Number]

[0037] A method for delivering radiation to a plurality of patient target regions may include obtaining a registration image at the start of a treatment session, identifying a first target region and a second target region in the registration image, shifting a first fluence map sub-region of a treatment plan fluence map such that a non-zero high fluence region of the first fluence map sub-region co-localizes with the first target region in the registration image, shifting a second fluence map sub-region of the treatment plan fluence map such that a non-zero high fluence region of the second fluence map sub-region co-localizes with the second target region in the registration image, delivering the shifted first fluence map sub-region to the first target region, and delivering the shifted second fluence map sub-region to the second target region.

[0038] A treatment plan fluence map can be divided into first and second fluence map sub-regions by defining the boundaries of each sub-region along the non-zero low-dose region of the treatment plan fluence map. Delivering the first and second shifted fluence map sub-regions can include segmenting the shifted first and second fluence map sub-regions into a plurality of radiation therapy system machine instructions for each injection position of the treatment radiation source. The radiation therapy system machine instructions can include one or more multi-leaf collimator configurations and treatment radiation source control parameters for each injection position. The method may optionally include acquiring imaging data during a treatment session and modifying the shifted first fluence map sub-region according to the acquired imaging data before delivering fluence to the first target region. Alternatively or in addition, the method may include acquiring imaging data during a treatment session and modifying the shifted second fluence map sub-region according to the acquired imaging data before delivering fluence to the second target region.

[0039] Delivering the shifted first fluence map sub-region and delivering the shifted second fluence map sub-region can be performed simultaneously and / or sequentially. In some variations, delivering the shifted first fluence map sub-region and delivering the shifted second fluence map sub-region may be performed sequentially. The registration image can include a CT image and / or a PET image and / or an MRI image. The method may optionally include comparing the high-fluence region of the first fluence map sub-region with the location of the first target region to define a first registration correction, comparing the high-fluence region of the second fluence map sub-region with the location of the second target region to define a second registration correction, and adjusting the patient table according to the first registration correction before delivering fluence to the first target region. After delivering fluence to the first target region, the method may adjust the patient table according to the second registration correction before delivering fluence to the second target region.

[0040] In some variations, delivering the shifted first fluence map sub-region to the first target region may include selecting a registration reference point representing a location within the registration image, applying a registration function calculated from the registration reference point to a first shift-invariant emission filter calculated during treatment planning for the first target region to update the first fluence map sub-region, segmenting the updated first fluence map sub-region into a plurality of radiation therapy system instructions, and emitting radiation fluence to the first target region according to the first plurality of radiation therapy system instructions.

[0041] Applying the registration function may include convolving a first registration function (e.g., a first delta function or a first Gaussian function such as a truncated Gaussian function) calculated from the registration reference point with the first shift-invariant emission filter. Delivering the shifted second fluence map sub-region to the second target region may include obtaining imaging data of the second target region, convolving the imaging data with a second shift-invariant emission filter calculated during treatment planning for the second target region to update the second fluence map sub-region, segmenting the updated second fluence map sub-region into a second plurality of radiation therapy system instructions, and emitting radiation fluence to the second target region according to the second plurality of radiation therapy system instructions.

[0042] In some variations, the acquired imaging data may be PET imaging data. Delivering the shifted second fluence map sub-region to the second target region may include selecting a second position registration reference point located within the second target region, applying a second position registration function calculated from the second position registration to a second shift-invariant emission filter to calculate the second fluence map sub-region, segmenting the calculated second fluence map sub-region into a second plurality of radiation therapy system instructions, and emitting a radiation fluence to the first target region according to the second plurality of radiation therapy system instructions. Alternatively or in addition, delivering the shifted second fluence map sub-region to the second target region may include selecting a second position registration reference point located within the second target region, convolving a second position registration function (e.g., a first delta function or a first Gaussian function such as a truncated Gaussian function) calculated from the second position registration with a second shift-invariant emission filter to calculate the second fluence map sub-region, segmenting the calculated second fluence map sub-region into a second plurality of radiation therapy system instructions, and emitting a radiation fluence to the second target region according to the second plurality of radiation therapy system instructions. In some variations, this method may include adjusting the deliverable radiation dose to meet one or more dose constraints. In some variations, one or more dose constraints may include one or more cost functions. In some variations, one or more cost functions may include a cumulative cost function having a weight coefficient for each cost function given by:

Number

[0043] where x is the fluence to the patient target region, A is the dose calculation matrix for the patient target region, Ax cumulative is the cumulative planned dose, and x cumulative is the cumulative planned fluence.

[0044] A method for treatment planning for radiation delivery to multiple patient target regions may include identifying a first location of a first patient target region and a second location of a second patient target region in a patient planning image, and defining a first region of interest (ROI) (which may be a first emission zone) having a boundary surrounding the first patient target region. Defining a second ROI (which may be a second emission zone) having a boundary surrounding the second patient target region, and calculating a treatment planning fluence map for the first patient target region and the second patient target region that specifies the fluence to be delivered when the selected imaging data indicates that the first patient target region is within the boundary of the first region of interest and the second patient target region is within the boundary of the second region of interest. The boundary of the second region of interest may be selected to surround the second patient target region for a predetermined range of positional displacements of the first and second patient target regions, and the first and second regions of interest may represent a spatial filter configured to select imaging data acquired during a treatment session. The first patient target region may be closer to a planning structure. The first region of interest may be smaller than the second region of interest. In some variations, the planning structure may be an organ at risk (OAR). The first patient target region may be closer to two or more planning structures than the second patient target region. Some methods may further include specifying a treatment planning reference point within the first region of interest as a positioning reference point for aligning the patient at the start of a treatment session. The selected imaging data may be used to guide radiation delivery during a treatment session. The treatment planning reference point may be the center point of the first region of interest and, in some examples, may be the center point of the first patient target region. The imaging data may include positron emission tomography data (e.g., PET data), CT imaging data, and / or MRI imaging data. The positron emission tomography scan may be selected if the positron emission tomography scan intersects at least one of the first and second regions of interest. The second region of interest may be sized to include a range of locations of the second patient target region.

[0045] In some variations, a method for patient positioning and radiation delivery may include obtaining an image including a first patient target region and a second patient target region, and aligning the first patient target region within a first region of interest defined during treatment planning. If the second patient target region is located within a second region of interest defined during treatment planning, the method may include obtaining positron annihilation emission data including a plurality of lines of response (LORs) using an array of positron emission detectors. If an LOR intersects either the first region of interest or the second region of interest, the method may include calculating a delivery fluence by convolving the LOR with a shift-invariant projection filter defined during treatment planning for the corresponding region of interest. The method may further include emitting the delivery fluence using a treatment radiation source, and emitting the delivery fluence includes segmenting the delivery fluence into a plurality of radiation therapy system machine instructions for each injection position of the treatment radiation source. The radiation therapy system machine instructions may include one or more multi-leaf collimator configurations and treatment radiation source emission (e.g., pulse) parameters for each injection position. In some variations, the image may include one or more of a positron emission tomography (PET) image, a computed tomography (CT) image, and a magnetic resonance imaging (MRI) image.

[0046] A method for positioning and aligning a patient for a radiation therapy treatment session may include positioning the patient on a radiation therapy system table such that a first patient target region is aligned with the location of a treatment radiation source and is located within a first predetermined region of interest, and determining whether the location of a second patient target region is within a second predetermined region of interest. If the location of the second patient target region is within the second predetermined region of interest, the method may include proceeding with radiation delivery to the first patient target region. If the location of the second patient target region is not within the second predetermined region of interest, the method may include generating a visual and / or audible notification. The first and second predetermined regions of interest may be calculated during treatment planning, and the location of the second predetermined radiation region of interest is defined relative to the location of the first predetermined region of interest.

[0047] Optionally, some methods may further include acquiring imaging data prior to radiation delivery to the second patient target region, wherein the first and second regions of interest are spatial masks configured to suppress imaging data with spatial characteristics that do not co-localize with any of the first and second radiation regions of interest. In some variations, the method may further include determining, based on the acquired imaging data, whether the second target region is located within the second region of interest, and proceeding with radiation delivery to the second target region if the second target region is located within the second region of interest. The acquired imaging data may include positron emission tomography (PET) imaging data, CT imaging data, and / or MRI imaging data. Delivering radiation may include segmenting the delivery fluence into a plurality of radiation therapy system machine instructions for each injection position of the therapeutic radiation source. The radiation therapy system machine instructions may include one or more multi-leaf collimator configurations and therapeutic radiation source emission (e.g., pulse) parameters for each injection position.

[0048] In some variations, the method may further include adjusting the patient's position when the location of the second patient target region is not within the second predetermined region of interest. In some variations, the method may further include adjusting the radiation therapy system table to move the location of the second patient target region to be within the second predetermined region of interest. In some variations, the method may further include shifting the planned fluence for the second patient target region according to the location of the second patient target region. In some variations, the method may further include delivering radiation including segmenting the delivery fluence into a plurality of radiation therapy system machine instructions for each injection position of the therapeutic radiation source. In some variations, the method may further include the radiation therapy system machine instructions further including one or more multi-leaf collimator configurations and therapeutic radiation source pulse parameters for each injection position.

[0049] A method for radiotherapy includes obtaining an image including a first patient target region and a second patient target region; splitting a planned fluence map into a first fluence sub-map for delivering a first prescribed dose to the first patient target region and a second fluence sub-map for delivering a second prescribed dose to the second patient target region; shifting the first fluence sub-map to be aligned with the first patient target region of the obtained image; shifting the second fluence sub-map to be aligned with the second patient target region of the obtained image; calculating a delivery fluence map by combining the shifted first fluence sub-map and the shifted second fluence sub-map; and delivering radiation to the first patient target region if high-fluence regions of the shifted first sub-fluence map and the shifted second sub-fluence map do not co-localize with each other in the delivery fluence map. In some variations, the planned fluence map may be generated by identifying the first and second patient target regions in a treatment planning image; calculating a first fluence map for delivering a first prescribed dose to the first patient target region; calculating a second fluence map for delivering a second prescribed dose to the second patient target region; combining the first and second fluence maps; and adjusting the combined fluence map such that high-fluence regions of the first and second fluence maps do not co-localize with each other and the first and second patient target regions each receive corresponding first and second prescribed doses. In some variations, the first fluence sub-map may include the first fluence map, and the second fluence sub-map may include the second fluence map. In some variations, calculating the planned fluence map may include adjusting the combined fluence map based on one or more adjustment constraints such that the first and second fluence maps do not exceed a predetermined dose limit. In some variations, the method may include identifying risk organs in the treatment planning image.Calculating the planned fluence map may further include adjusting the combined fluence map such that the fluence to the risk organ does not exceed a predetermined dose limit. In some variations, delivering radiation may include segmenting the delivered fluence into a plurality of radiotherapy system machine instructions for each injection position of the therapeutic radiation source. In some variations, the radiotherapy system machine instructions may include one or more multi-leaf collimator configurations and therapeutic radiation source emission parameters for each injection position. In some variations, the method may include determining whether the delivered fluence map includes a fluence level to a risk organ that exceeds a predetermined dose limit, and delivering radiation to a first patient target region if the fluence to the risk organ does not exceed the predetermined dose limit. In some variations, the planned fluence map may be generated by identifying first and second patient target regions in the treatment planning image, calculating a first fluence map for delivering a first prescription dose to the first patient target region, combining the first and second fluence maps, and repeatedly adjusting the combined fluence map to satisfy a common set of constraints while maintaining the first and second fluence maps as separate fluence maps to calculate the planned fluence map. In some variations, the acquired image may include one or more of a positron emission tomography (PET) image, a computed tomography (CT) image, and a magnetic resonance imaging (MRI) image.

[0050] The planned fluence map can be generated by identifying first and second patient target regions in the treatment planning image, calculating a first fluence map for delivering a first prescribed dose to the first patient target region, calculating a second fluence map for delivering a second prescribed dose to the second patient target region, combining the first and second fluence maps, and adjusting the combined fluence map such that high fluence regions of the first and second fluence maps do not co-localize with each other and the first and second patient target regions each receive corresponding first and second prescribed doses. The first fluence sub-map may include the first fluence map, and the second fluence sub-map may include the second fluence map. The planned fluence map can be re-optimized using a common set of constraints while, for example, holding the first sub-fluence map and the second fluence sub-map as separate fluence maps. The method may further include identifying risk organs in the treatment planning image, and calculating the planned fluence map may further include adjusting the combined fluence map such that the fluence to the risk organs does not exceed a predetermined dose limit. Delivering radiation may include segmenting the delivery fluence into a plurality of radiation therapy system machine instructions for each injection position of the treatment radiation source. The radiation therapy system machine instructions may include one or more multi-leaf collimator configurations and treatment radiation source emission parameters for each injection position. Optionally, the method may include determining whether the delivery fluence map includes a fluence level to a risk organ that exceeds a predetermined dose limit, and delivering radiation to the first patient target region if the fluence to the risk organ does not exceed the predetermined dose limit.

[0051] In some variations, a separate treatment plan may be generated for each patient target region. Each of these plans may be individually optimized, summed, and then re-optimized based on a common set of constraints while maintaining the treatment plans as separate entities. Separate fluence maps may be generated based on each treatment plan and summed as a combined fluence map. During the treatment session, each separate fluence map may be moved independently and their combined dose volume histogram (DVH) may be checked prior to dose delivery.

[0052] In addition or alternatively, in order to ensure that the dose does not exceed a predetermined threshold in a given zone (e.g., region), in addition to a predetermined set of constraints, one or more adjustment constraints (e.g., artificial constraints in addition to tissue constraints such as structural adjustments) may be applied. For example, adding one or more predetermined dose constraint zones during the treatment plan may be useful to reduce the likelihood of dose "hot spots" that may occur when the fluence map of each patient target region is shifted independently during the treatment session, and the zone may be maintained as a "low-dose valley". In some variations, one or more adjustment constraints may help to promote the separation of radiation delivery between two or more targets. Some variations of the adjustment constraints may limit the dose delivery angle.

[0053] This specification also discloses a method for generating a visual graphic representation of radiation dose. One variation of the method for generating a visual graphic is to generate a treatment plan for irradiating one or more patient target regions, the treatment plan including at least one of at least one treatment plan fluence map, one or more planning images for each patient target region, and a corresponding nominal dose image; calculating, for each patient target region, a plurality of uncertainty data inputs representing the uncertainty of the treatment session; calculating, for each uncertainty data input, an expected dose delivered to each patient target region in the presence of the uncertainty data input to derive a plurality of dose images; combining the plurality of dose images into a visualization graphic that is a composite volume, each dose image representing a frame within the composite volume; and displaying the visualization graphic on a display device.

[0054] In some variations, the treatment plan may further include one or more shift-invariant projection filters. The plurality of data inputs may include one or more scalar values, vector values, and / or volume measurements. In some variations, the composite volume may be a 4D volume. The plurality of dose images may include 2D dose image slices for each of one or more patient target regions. The composite volume may include 3D dose volumes for each of the patient target regions, and each 3D dose volume may include 2D dose image slices for each of the patient target regions. In some variations, the method includes iterating through dose values at each point in each dose image, comparing the dose values to a first matching point in a minimum dose volume and a second matching point in a maximum dose volume to update the first and second matching points, iterating through each point in the minimum dose volume, calculating a difference between a value of the minimum dose volume and a value of a nominal dose volume, storing the difference and a frequency of the difference in two frames of a delta dose volume, iterating through each point in the maximum dose volume, calculating a difference between a value of the maximum dose volume and a value of the nominal dose volume, storing the difference and a frequency of the difference in another two frames of the delta dose volume, and rendering a surface map for each 4-frame slice of the delta dose volume where a frequency of a maximum value protrudes along +Z from a 3D plane, a frequency of a minimum value protrudes along -Z from the 3D plane, and values at points in the surface map are represented by color.

[0055] One variation of a method for generating a visualization graphic is to generate a treatment plan for irradiating one or more patient target regions, where the treatment plan can include at least one of at least one treatment plan fluence map, one or more planning images for each patient target region, and a corresponding nominal dose image, and for each patient target region, calculate a plurality of uncertainty data inputs representing the uncertainty of the treatment session, calculate for each uncertainty data input the predicted dose delivered to each patient target region in the presence of the uncertainty data input to derive a plurality of dose images, combine the plurality of dose images into a composite volume, where each dose image represents a frame within the composite volume, to form a visualization graphic, and display the visualization graphic on a display device. In some variations, the treatment plan may further include one or more shift-invariant emission filters. The plurality of data inputs can include one or more scalar values, vector values, and / or volume measurements. The composite volume can be a 4D volume. The plurality of dose images can include 2D dose image slices for each of one or more patient target regions. The composite volume can include a 3D dose volume for each of each patient target region, and each 3D dose volume can include 2D dose image slices for each of each patient target region. Optionally, the method includes iterating through the dose values at each point in each dose image, comparing the dose values to a first matching point in a minimum dose volume and a second matching point in a maximum dose volume to update the first and second matching points, iterating through each point in the minimum dose volume, calculating the difference between the value of the minimum dose volume and the value of the nominal dose volume, storing the difference and the frequency of the difference in two frames of a delta dose volume, iterating through each point in the maximum dose volume, calculating the difference between the maximum dose volume and the value of the nominal dose volume, storing the difference and the frequency of the difference in another two frames of the delta dose volume, and rendering a surface map for each 4-frame slice of the delta dose volume where the frequency of the maximum value protrudes along the +Z axis from a 3D plane, the frequency of the minimum value protrudes along the -Z axis from the 3D plane, and the value at a point in the surface map is represented by color.

[0056] This specification describes a visual graphic representation of radiation dose. A variant of the visual graphic representation may include a 3D volume representing a patient's anatomical structure, the 3D volume including a stack of 2D patient anatomical images and a 3D dose surface map corresponding to the dose distribution for each 2D patient anatomical image. The height of the 3D dose surface map may represent the dose level according to the treatment plan for the patient, and the color values across the surface map may represent the probability values of the dose levels during the treatment session.

[0057] In some variants, a graphical user interface for a multi-target treatment plan may include a first graphic representation of a plurality of treatment areas, each encompassing one or more patient target regions. Each treatment area may span different regions of the patient and may represent various patient position alignments and a second graphic representation of the dose measurement characteristics for each treatment area. The first and second graphic representations may be configured to be output simultaneously to a display device.

[0058] In some variations, the dose measurement characteristics of all of the plurality of treatment areas may be visualized simultaneously on a display device. In some variations, the dose measurement characteristics of each treatment area may be visualized individually. In some variations, the dose measurement characteristics of each treatment area may include one or more of a dose measurement target, a dose volume histogram, dose statistics, and target performance. In some variations, a third graphical representation may include a view of all patient target regions across a plurality of treatment areas and a visual indication representing the relative spatial relationship between one or more of the patient target regions. In some variations, the graphical user interface may further include a third graphical representation that includes a view of a plurality of risk organs (OARs) across a plurality of treatment areas and a visual indication representing the relative spatial relationship of each OAR to one or more treatment areas. In some variations, the graphical user interface may further include a third graphical representation that includes a view of all patient target regions across a plurality of treatment areas and a first visual indication representing the relative spatial relationship between one or more of the patient target regions, and a fourth graphical representation that includes a view of a plurality of risk organs (OARs) across a plurality of treatment areas and a second visual indication representing the relative spatial relationship of each OAR to one or more treatment areas. In some variations, the graphical user interface may include a 3D visualization representing the patient's anatomical structure. In some variations, the first graphical representation may be layered over the anatomical 3D visualization. In some variations, the first graphical representation may include bands for each of the plurality of treatment areas. Each band may have a range representing the dimensions of the corresponding treatment area. In some variations, the graphical user interface may include a first rendering mode depicting a subset of the plurality of treatment areas and a second rendering mode depicting all of the plurality of treatment areas. In some variations, the subset of the plurality of treatment areas may include the treatment areas selected by the user. The present invention provides, for example, the following. (Item 1) A method for virtual target region alignment and radiation delivery, comprising: acquiring an image of a patient in a treatment position and identifying a patient target region in the acquired image; selecting an alignment reference point within the acquired image, the alignment reference point corresponding to a planned alignment reference point; calculating an alignment function based on the alignment reference point and calculating a delivery fluence to the patient target region at each emission position of a treatment radiation source by applying the alignment function to a shift-invariant emission filter derived based on the planned alignment reference point; emitting the delivery fluence to the patient target region using the treatment radiation source. (Item 2) The method according to item 1, wherein applying the alignment function to the shift-invariant emission filter includes convolving the alignment function with the shift-invariant emission filter. (Item 3) The method according to item 2, wherein the alignment function is one of a delta function, a Gaussian function, a circular function, and interpolation. (Item 4) The method according to item 3, wherein the Gaussian function is a truncated Gaussian function. (Item 5) The method according to item 3, wherein the interpolation is one of linear, bicubic, spline, or Fourier shift. (Item 6) The method according to item 1, wherein the acquired image includes one or more of a positron emission tomography (PET) image, one or more X-ray projection images, a computed tomography (CT) image, and a magnetic resonance imaging (MRI) image. (Item 7) identifying a second patient target region in the acquired image; selecting a second alignment reference point for the second patient target region within the acquired image, the second alignment reference point corresponding to a second planned alignment reference point; calculating a second alignment function based on the second alignment reference point and calculating a second delivery fluence to the second patient target region at each emission position of the treatment radiation source by applying the second alignment function to a second shift-invariant emission filter derived based on the second planned alignment reference point; further comprising emitting the second delivery fluence to the second patient target region using the treatment radiation source. (Item 8) identifying a second patient target region in the acquired image; selecting, within the acquired image, a second position registration fiducial point for the second patient target region, the second position registration fiducial point corresponding to a second planned position registration fiducial point; calculating a second delta function based on the second position registration fiducial point and convolving the second delta function with a second shift-invariant emission filter based on the second planned position registration fiducial point to calculate a second delivery fluence at each emission position of the treatment radiation source; releasing the second delivery fluence to the second patient target region using the treatment radiation source, the method of item 2 further comprising. (item 9) releasing the first delivery fluence to the first patient target region and releasing the second delivery fluence to the second patient target region simultaneously, the method of item 7 or 8; (item 10) releasing the first delivery fluence to the first patient target region and releasing the second delivery fluence to the second patient target region sequentially, the method of item 7 or 8; (item 11) the registration function is a first registration function (δ), the first shift-invariant emission filter includes a first set of fluence map filters (p 1 、p 2 、...、p i ) calculated during treatment planning for each emission position (i) of the treatment radiation source, the method including calculating a first set of projections (δ i ) of the registration function at each emission position (i); δ i =proj i (δ) each projection (δ i ) is a 2D fluence distribution; calculating the first fluence for delivery includes convolving each projection in the first set of projections of the first registration function (δ i ) with the corresponding fluence map filter (p i ) to calculate a first delivery fluence map (f i ) for each emission position (i) of the treatment radiation source; f i =p i*δi delivering the first fluence includes moving the treatment radiation source to each emission position (i) and emitting radiation to the first patient target region according to the first delivery fluence map (f i ), the method according to any one of items 2 and 8; (item 12) each projection (δ i ) is an m×n matrix, where m is the number of multi-leaf collimator leaves and n is the number selected during the treatment planning, the method according to item 11. (Item 13) The method according to item 12, where n is the number of the beam stations selected during the treatment planning. (Item 14) The method according to item 11, where the first position matching function is one of a delta function, a Gaussian function, a circular function, and interpolation. (Item 15) The method according to item 14, where the Gaussian function is a truncated Gaussian function. (Item 16) The method according to item 14, where the interpolation is one of linear, bicubic, spline, and Fourier shift. (Item 17) The second shift-invariant emission filter includes a second set of fluence map filters (p_2 1 、p_2 2 、...、p_2 i ) calculated during the treatment planning for each emission position (i), and the method includes calculating a second set of projections (δ_2 i ) of the second position matching function to each emission position (i), δ_2 i =proj i (δ_2) Each projection (δ_2 i ) is a 2D fluence distribution, Calculating the second fluence for delivery includes convolving each projection (δ_2 i ) in the projections of the second set of the second position matching function with the corresponding fluence map filter (p_2 i ) to calculate a second delivery fluence map (f_2 i) ) for each emission position (i), f_2 i =p_2 i*δ_2i Delivering the second calculated fluence includes moving the treatment radiation source to each emission position (i) and emitting radiation to the second patient target region according to the second delivery fluence map (f_2 i ), the method according to item 11. (Item 18) The method according to item 17, where each projection (δ_2 i ) is an m×n matrix. (Item 19) The method according to item 18, where m is the number of multi-leaf collimator leaves and n is the number selected during the treatment planning. (Item 20) The method according to item 19, where n is the number of the beam stations selected during the treatment planning. (Item 21) The method according to item 17, where the second position matching function is one of a delta function, a Gaussian function, a circular function, and interpolation. (Item 22) The method according to item 21, where the Gaussian function is a truncated Gaussian function. (Item 23) The method according to item 21, where the interpolation is one of linear, bicubic, spline, and Fourier shift. (Item 24) Delivering the calculated fluence includes segmenting the delivery fluence map (f i 、f_2 i ) into a plurality of radiotherapy system machine instructions for each injection position, the method according to any one of items 11 to 17. (Item 25) Delivering the calculated fluence includes segmenting the calculated fluence into a plurality of radiotherapy system machine instructions for each injection position, the method according to item 1 or 2. (Item 26) Delivering the second calculated fluence includes segmenting the second calculated fluence into a plurality of radiotherapy system machine instructions for each injection position, the method according to item 7 or 8. (Item 27) The plurality of radiotherapy system machine instructions includes one or more multi-leaf collimator configurations for each injection position, and emitting a radiation fluence further includes moving the leaves of the multi-leaf collimator to the multi-leaf collimator configuration corresponding to the location of the injection position of the therapeutic radiation source, and emitting a pulse of radiation, the method according to any one of items 24 to 26. (Item 28) The plurality of radiotherapy system machine instructions further includes therapeutic radiation source emission (e.g., pulse) parameters for each injection position, and emitting the pulse of radiation includes emitting radiation having the therapeutic radiation source pulse parameters corresponding to the location of the injection position of the therapeutic radiation source, the method according to item 27. (Item 29) Calculating the delivery fluence map (f i 、f_2 i ) further includes convolving each projection in the first or second set of projections with the corresponding shift-invariant fluence map filter, and applying a virtual flattening filter correction factor (FF), (f i ) = FF·(p i*δi ) (f_2 i ) = FF·(p_2 i*δ_2i ) The virtual flattening filter correction factor (FF) is an m×n matrix and is the reciprocal of the flatness profile of the radiation beam emitted from the therapeutic radiation source, the method according to item 24. (Item 30) Calculating the delivery fluence map (f i 、f_2 i) further includes convolving each projection in the set of projections with the shift-invariant fluence map filter, and applying the virtual flattening filter correction factor (FF) and a distance compensation factor

Number

Number

Number

Number

Number

Brief Description of the Drawings

[0059] This application can be understood by referring to the following description in conjunction with the accompanying drawings.

[0060]

Figure 1A

Figure 1B

Figure 2

Figure 3A

Figure 3B

Figure 4

Figure 5A

Figure 5B

Figure 5C

Figure 6

Figure 7A

Figure 7B

Figure 8A

Figure 8B

Figure 9

Figure 10

Figure 11A

Figure 11B

Figure 12A

Figure 12B

Figure 12C

Figure 12D

Figure 12E

Figure 12F

Figure 13A

Figure 13B

Figure 13C

Figure 13D

Figure 13E

Figure 13F

Figure 14A

Figure 14B

Figure 14C

Figure 14D

Figure 14E

Figure 14F

Figure 15A

Figure 15B

Figure 15C

Figure 15D

Figure 15E

Figure 16A-1

Figure 16A-2

Figure 16B

Figure 17A

Figure 17B

Figure 17C

Figure 17D

Figure 17E

Figure 17F

Figure 18A

Figure 18B

Figure 18C

Figure 18D

Figure 19

Figure 20

Figure 21A

Figure 21B

Figure 21C

Figure 22A

Figure 22B

Figure 23A

Figure 23B

Figure 23C

Figure 23D

Figure 24

Figure 25

Figure 26

Figure 27

Figure 28

Figure 29

Figure 30

Figure 31

Figure 32

Figure 33

Figure 34

Figure 35A

Figure 35B

Figure 35C

Figure 36A

Figure 36B

Figure 37A

Figure 37A

Figure 37B

Figure 37C

Figure 37D

Figure 37E

Figure 37F

Figure 37G

Figure 37H

Figure 37I

Figure 37J

Figure 37K

Figure 37L

Figure 37M

Figure 37N

Figure 38A

Figure 38B

Figure 38C

Figure 38D

Figure 38E

[0061] This specification discloses methods for treatment planning, patient setup, alignment, and radiation delivery for irradiating multiple patient target regions in a single treatment session or fraction. The treatment planning method may include selecting a planned alignment reference point for each target region and calculating a shift-invariant emission filter for each target region based on the selected planned alignment reference point. For irradiating multiple target regions, a treatment planning system that executes the methods described herein may generate a set of shift-invariant emission filters associated with the planned alignment reference points for each patient target region. Optionally, the treatment planning system may generate a planned fluence map representing the radiation fluence across one or more patient target regions, one or more OARs, and / or healthy tissue. In some variations, the treatment planning system does not generate a set of radiation therapy machine instructions to be executed by a radiation therapy system after patient setup and patient target region alignment. Instead, segmentation of the delivery fluence map occurs “in real time,” e.g., during a treatment session, within about 2 hours, about 1 hour, about 30 minutes, about 10 minutes, about 5 minutes, about 1 minute, about 30 seconds, about 1 second, about 0.5 seconds, or less of calculating the delivery fluence map. By deferring machine instruction segmentation until the treatment session, the planned fluence map may be updated (e.g., by calculating the delivery fluence map) with the latest location data of the patient target regions before calculating the machine instructions. Shifting the fluence map (e.g., linear shift, fixed shift) can be easily calculated, whereas “shifting” machine instructions can result in non-linear effects that affect the dose delivered. Further, by delaying segmentation of the machine instructions at least until after patient target region alignment, “virtual alignment” is enabled in which the fluence map generated by the treatment planning system is adjusted based on the latest location data. In some variations, virtual alignment may include updating the planned fluence map by convolving the shift-invariant emission filter with an alignment function derived from the latest location data (e.g., the current location of the planned alignment reference point).By updating the position registration reference point during a treatment session and calculating a delivery fluence map based on, for example, a delta function centered on the position registration reference point and a set of shift-invariant projection filters for its target region, the planned fluence map can "follow" the patient's target region instead of physically adjusting the patient's position (e.g., using a patient table) to match the planned fluence map. By segmenting the fluence map that best reflects the actual location of the patient's target region, more accurate radiation delivery can be facilitated compared to delivering radiation based on a fluence map that has not been updated to reflect the actual location of the patient's target region. A virtual position registration method combined with the BgRT method (where the delivery fluence is further updated with imaging data acquired during the session, such as positron emission tomography data) can help improve the accuracy of dose delivery to multiple patient target regions. In the virtual position registration method described herein, the patient can be physically set up and position registered once (e.g., for one patient target region), but radiation can be applied to multiple virtual position-registered patient target regions.

[0062] Delivery of radiation to multiple patient target regions and / or target regions of irregular shape poses challenges to alignment / registration and radiation delivery. For example, multiple tumors may vary (e.g., in location, shape, and / or size) between when a treatment planning image is acquired and when the patient arrives for treatment. Some tumors may grow, while others may shrink, some tumors may shift to the right, while others may mutate to the left, etc. If the patient loses or gains weight (e.g., due to swelling), the location of the tumor relative to healthy tissue or bone landmarks can change. In addition, when treating multiple target regions in a single treatment session or fraction, the target regions treated in the later part of the treatment session may have moved from their locations during the earlier part of the treatment session (e.g., initially, during patient setup and alignment). The motion can be periodic (e.g., due to breathing) or static (e.g., during sequential treatment of multiple target regions, the alignment of the last target to be treated may be ineffective because the patient may displace at some point during delivery). Addressing changes in the patient and / or target regions between planning and treatment, as well as changes that can occur during a treatment session, is accompanied by an additional level of complexity due to the highly coupled nature of dose delivery to the patient. Radiation applied to one target region can produce scattered dose to other target regions, and in some cases, the radiation beam can intersect two target regions (or a target region and an OAR) and deliver direct dose to both regions. If these target regions move relative to each other, the treatment plan fluence map can fail to provide the prescribed dose. Thus, aligning the entire target region or a set of target regions associated with one or more tumors with a radiation therapy system for an intended (i.e., future) radiation treatment session, such as when a healthcare provider is attempting to align a single target or multiple targets at the start of a treatment session, can be difficult and can result in inaccurate dose delivery.

[0063] The treatment planning, alignment, and radiation delivery methods described herein may help reduce some of the uncertainty in dose delivery associated with the treatment of multiple patient target regions in a single treatment session. One variation of a method for radiation delivery to multiple patient target regions is to align each patient target region individually (e.g., independently of the remaining patient target regions) (e.g., virtually align and / or physically align), derive a delivery fluence map for each patient target region, and then deliver radiation to each patient target region according to the corresponding delivery fluence map. Aligning each patient target region may include dividing the planned fluence map into fluence map sub-regions for each target region along low fluence regions (by either a treatment planning system during treatment planning or a radiotherapy system at the start of a treatment session), and using a virtual alignment method, such as shifting the fluence map sub-regions, e.g., shifting the planned fluence map, to derive the delivery fluence map. Alternatively or in addition, a method for radiation delivery to multiple patient target regions may include generating the planned fluence maps for each patient target region individually, aligning each patient target region individually (e.g., independently of the remaining patient target regions) to derive a delivery fluence map, combining the individual delivery fluence maps, verifying that the cumulative fluence map meets the prescription dose target (i.e., the minimum prescription dose to the target region, the maximum dose to the OAR does not exceed, etc.), and delivering radiation to each patient target region according to the individual corresponding delivery fluence maps for each patient target region when the dose target is met.A BgRT method for radiation delivery to a plurality of patient target regions can include designating one of the plurality of patient target regions as a registration target region, defining a region of interest (ROI) (e.g., a biological falloff zone (BFZ)) around the remaining patient target regions that is large enough to encompass the range of positional displacements of those patient target regions, registering the registration target region, and delivering radiation to each patient target region according to imaging data that is spatially co-localized with the ROI of each patient target region. The ROI around the registration target region may be smaller than the ROI around the remaining patient target regions.

[0064] Optionally, at one or more stages during treatment planning and / or radiation delivery, the system controller of the treatment planning system and / or the radiation therapy system may generate and display one or more visualization graphics indicative of the dose deposition on the patient's anatomical structure. Additional changes to the radiation delivery may be included in response to the visualization graphics, as may be desired by a clinician or technician. In particular, providing visualization graphics depicting the dose distribution on the patient anatomical structure as part of, or subsequent to, a positioning / registration procedure can help a clinician or technician evaluate the interaction of the fluence calculated for each patient target region and determine whether these cumulative levels of fluence are acceptable. For example, if the scatter dose from one target region to another is relatively low or spatially uniform, it may be possible to ignore the effects of spatial variations between the planned fluence map and the delivered fluence map. The visualization graphics can help a clinician or technician identify hot spots, and / or cold spots, and / or dose binding effects (e.g., when the radiation beam intersects two or more target regions and / or OARs) that may result from the delivery of radiation to multiple target regions. Identifying these unwanted dose interactions prior to radiation delivery and providing spatial information regarding the dose distribution as feedback to the treatment planning and optimization process can help further refine the planned fluence map and / or the ejection filter.

[0065] Alternatively or in addition, the patient setup and target area alignment / matching method may include obtaining one or more images of the tumor that reflect current information regarding the location, size, and shape of the tumor prior to applying radiation during an image-guided alignment procedure. The user may compare the acquired images to treatment planning images acquired and / or prepared prior to the procedure. The user and / or the radiation therapy system may then adjust the radiation therapy session based on the acquired images such that the applied radiation can more effectively target the current shape, size, and location of one or more tumors (i.e., perform setup corrections to the patient's positioning and orientation). For example, the user may define one or more treatment areas (also referred to herein as "zones" and / or "treatment fields") within the images of one or more tumors for image-guided alignment. Image matches or position corrections may then be defined for each treatment area or region of interest. The radiation therapy system may automate the application of each position correction as the radiation therapy system proceeds with irradiating the target area (e.g., a tumor or a portion of a tumor) or multiple target areas. The radiation therapy system may provide the user with instructions indicating position corrections or adjustments corresponding to position correction actions performed by the user (e.g., moving the patient's arm or leg, or tilting and / or rotating the surface on which the patient is disposed). After the position correction is performed, the radiation therapy system may apply radiation to the real-time or updated location of the tumor or portion of the tumor in each region of interest. Additionally, rather than holding the patient in a position that enables high-quality conformal dose delivery to some tumor regions at the expense of low-quality conformal dose delivery to other tumor regions, each region of interest can be more accurately targeted by using the most effective positioning of the patient with respect to the treatment radiation source for each region of interest. This can have the advantage of providing better treatment outcomes and overall conformity to the treatment plan. In addition, since multiple targets can be treated during a single radiation therapy session, the duration of the radiation therapy session can be shortened.By reducing the time, a medical facility may be able to treat more patients per day. Additionally, reducing the duration required for treating multiple tumors or multiple tumor portions also improves the patient experience.

[0066] A method for setting up a patient for radiation therapy may include acquiring images of a first patient target region and a second patient target region. A first set of patient position shift vectors may be calculated based on the acquired image of the first patient target region and a treatment planning image. A second set of patient position shift vectors may be calculated based on the acquired image, the treatment planning image of the second patient target region, and the first set of patient position shift vectors. The patient may be aligned to a first location according to the first set of patient position shift vectors. The patient may be moved to a second location and aligned according to the second set of patient position shift vectors.

[0067] Note that the methods for patient setup, patient target region localization / alignment, and radiation delivery for irradiating multiple patient target regions in a single treatment session described herein may be used alone or in combination in intensity modulated radiation (IMRT), and / or stereotactic body radiation therapy (SBRT), and / or BgRT. For example, a clinician may determine whether all of the patient target regions should be treated using IMRT / SBRT, or whether all of the patient target regions should be treated using BgRT, or whether some of the patient target regions should be treated using SBRT / IMRT and others should be treated using BgRT. Some treatment planning systems and / or radiation therapy systems have two operating modes, a first mode being SBRT / IMRT planning and radiation irradiation, and a second mode being BgRT planning and radiation irradiation. A single treatment session may include radiation delivery using the SBRT / IMRT method for some patient target regions and radiation delivery using the BgRT method for other patient target regions, as may be desired.

[0068] Overview of the method Virtual position verification Virtual position verification of a patient target region involves modifying (e.g., shifting) a planned fluence map for a particular patient target region to reflect the current / real-time location of that patient target region. The planned fluence map includes a set of radiation beamlets and beamlet intensities to be applied to a patient, as calculated by a treatment planning system based on one or more planning images. Conceptually, this can be thought of as adjusting the fluence map as if the registration image were the planning image. In contrast to a physical patient setup where the patient's position is physically adjusted by a patient table so that the location of the patient target region during treatment matches the location of the patient target region in the treatment plan, virtual position verification may require little, if any, patient table adjustment. A physical patient setup typically involves acquiring a registration image (e.g., a registration CT image, an MRI image), comparing the registration image to the planning image to calculate an offset representing a change in the location of the patient target region, and performing a patient table adjustment and / or otherwise moving the patient based on the calculated offset so that the current location of the patient target region is in position alignment with the location of the patient target region in the planning image. In virtual position verification, instead of adjusting the patient table, the planned fluence map is adjusted (e.g., shifted or rolled) by only the calculated offset, i.e., conceptually, it is similar to moving the planning image to be in position alignment with the registration image. Other fluence corrections may also be included to help ensure that the resulting dose distribution received by the patient is shifted to the appropriate location in the registration image without significant distortion, due to, for example, the fan beam nature of the radiation beam and the increasing radiation fluence closer to the radiation source.

[0069] The virtual position registration method aims to shift the dose received by a patient by modifying the fluence map delivered by the treatment radiation source of a radiation therapy system after position registration and before or during patient treatment. This is done by calculating a shift-invariant representation of the treatment plan (e.g., a shift-invariant injection filter) during treatment planning and selecting an appropriate position registration reference point. The shift-invariant treatment plan (e.g., a shift-invariant injection filter) is applied to the actual reference point location derived from the position registration image to shift the dose received by the patient (and to derive a delivery fluence map that can direct radiation to the current location of the patient target region). An example of a shift-invariant treatment plan representation is that for each patient target region where each injection filter is a 2D fluence map image in the radiation source beam's eye view coordinate system, a set of injection filters p i for each injection position i can be included. (That is, if there are three patient target regions and 100 injection positions, in the treatment plan, three sets of injection filters will be calculated, where each set specifies 100 injection filters, one for each injection position). The injection filter p i can be calculated based on the planned position registration reference point associated with the patient target region and, in some variations, may be selected by the user. The injection filter p i is convolved with the 2D projection of a discrete 3D delta function δ centered on the planned position registration reference point in the patient coordinate system for the 2D beam's eye view at each injection position to yield a fluence f i for each injection position i, and can be calculated by solving a set of filters such that f i =p i*δi where δ i =proj i (δ).

[0070] The projected δ i can be a 2D discrete delta function (i.e., an image with all zeros except for a single pixel with a value of 1), and thus the operation p i*δi can be the 2D shift of the injection filter pi, which can be considered as a fluence map independent of location. The fluence f i can be determined for each patient target region based on dose constraints and goals specified by a clinician using, for example, a standard radiotherapy dose optimization technique that minimizes a cost function with respect to the patient dose, using, for example, gradient descent. Selecting a desired planned position verification reference point (e.g., a point at the center of the target region, a point in the treatment area including the target region), creating a δ i representing the shift to that reference point, and applying an inverse shift to f i to obtain

Equation

[0071] The delta function δ centered on the position verification reference point can represent the location of the patient target region and can thus be used to calculate a shifted fluence map that directs radiation to the current location of the patient target region. Conceptually, the delta function δ can be considered to encode the position of the patient target region specified in 3D coordinates. The delta function δ can include a 3D image of all zeros with a single voxel of dimension infinitely narrow centered on the planned position verification reference point, but the projection of the delta function onto the injection position δ i (e.g., the "beam's eye view" of the treatment radiation source or the multi-leaf collimator space) can be defined by the width of each multi-leaf collimator (MLC) leaf and / or the beam station spacing (i.e., the spacing between discrete patient couch locations where the couch stops during delivery of the treatment radiation) and can include one or more non-zero voxels. The projection δ ican be a matrix (or image) having dimensions that match the number of MLC leaves in the MLC of the radiation therapy system and the number of patient table beam stations. A schematic representation of a delta function based on the voxels in space and the projection to the injection positions of these voxels is depicted in Figure 22A. Alternatively, the delta function δ may have a finite size set during treatment planning.

[0072] In some variations, the virtual position verification may include convolving a set of filters with a position verification function that may include a delta function, a Gaussian function, and a truncated Gaussian function centered on the position verification reference point. Figures 24 - 26 depict examples of position verification functions that can be used in any of the virtual position verification treatment planning and delivery methods described herein, and their corresponding values across a set of MLC leaf indices for multiple injection angles or positions. Figure 24 depicts the values and dose profiles of a delta function, Figure 25 depicts the values and dose profiles of a Gaussian function, and Figure 26 depicts the values and dose profiles of a truncated Gaussian function. In Figures 24 - 26, the X - axis corresponds to the MLC leaf index, and the Y - axis is the value or magnitude of the position verification function projection of the one - dimensional interpolation shift (e.g., interp1 shift, set of plots on the left) and the Fourier shift (set of plots on the right). Each line represents the position verification function values for different injection angles and / or positions. For each of the dose plots, the X - axis corresponds to the distance in millimeters, and the Y - axis corresponds to the fluence (e.g., dose). The peak - to - peak dose variability is measured from the highest peak in the family of curves to the lowest peak in the family of curves.

[0073] FIG. 24 depicts the values or magnitudes of the position-matched delta function projections over four injection angles and / or positions, as well as their corresponding dose (or fluence) profiles. The peak-to-peak dose variability of the one-dimensional interpolation shift is approximately 40%, and the peak-to-peak dose variability of the Fourier shift is approximately 7.5%. FIG. 25 depicts the values or magnitudes of the position-matched Gaussian function projections over four injection angles and / or positions, as well as their corresponding dose (or fluence) profiles, while FIG. 26 depicts the values or magnitudes of the position-matched truncated Gaussian function projections over four injection angles and / or positions, as well as their corresponding dose (or fluence) profiles. As shown in FIGS. 25 and 26, the Gaussian function and the truncated Gaussian function can have better peak-to-peak characteristics compared to the delta function shown in FIG. 24. The peak-to-peak dose variability of the Gaussian function is 11% and 1.5%, and the peak-to-peak dose variability of the truncated Gaussian function is 14% and 2.1% (for linear interpolation and Fourier shift, respectively). However, the Gaussian function and the truncated Gaussian function have broader curves than the delta function curve. The broader curves correspond to more MLC leaves that must be opened to deliver the fluence. That is, for a particular injection angle or position, the value of the position-matched function spans more MLC leaves for the Gaussian function or the truncated Gaussian function than for the delta function. The truncated Gaussian function of FIG. 26 provides improved peak-to-peak characteristics over the delta function, which has a narrower curve than the Gaussian function. A low peak-to-peak variability is desirable so that shifting the fluence does not result in significant dose differences. A treatment planning method incorporating virtual position matching may include selecting a particular position-matched function that provides a dose distribution that meets the prescription goals for each patient target region and / or OAR.

[0074] In addition to selecting a particular alignment function during treatment planning, some variations of the treatment planning method may also include selecting parameters (e.g., characteristics, constants, etc.) of the alignment function. For example, the treatment planning method may include selecting a truncated Gaussian function as the alignment function and selecting the σ value (e.g., width) of the truncated Gaussian function. The μ value (e.g., mean or center) of the truncated Gaussian function may be the planned alignment reference point. Alignment functions with different σ values can result in delivery fluences with different conformality and / or distortion characteristics. For example, selecting an alignment truncated Gaussian function with a low σ value can result in a shifted dose or fluence with better conformality compared to a similar function with a high σ value. FIGS. 27-29 depict simulation color intensity plots of alignment functions that are truncated Gaussian distributions with a σ value of 2 mm. FIGS. 30-32 depict simulation plots of alignment functions that are truncated Gaussian distributions with a σ value of 4 mm. The X-axis of the plots in FIGS. 27-32 is the injection position / angle, and the Y-axis is the MLC leaf index. FIGS. 27 and 30 depict the projection of the alignment function to all injection positions for the 2 mm and 4 mm truncated Gaussian functions, respectively. FIGS. 28 and 31 depict the injection filter values for all injection positions for the 2 mm and 4 mm truncated Gaussian functions, respectively. FIGS. 29 and 32 depict the fluence values for all injection positions for the 2 mm and 4 mm truncated Gaussian functions, respectively. The projection of the alignment function in FIG. 30 (σ value of 4 mm) is somewhat wider (i.e., spans more MLC leaves) than the alignment function in FIG. 27 (σ value of 2 mm). This can result in a fluence profile for the alignment function with a σ value of 4 mm (FIG. 32) that is more blurred than the fluence profile for the alignment function with a σ value of 2 mm (FIG. 29). However, the fluence peaks across multiple injection positions can be more uniform (less distorted) for the alignment function with a σ value of 4 mm than for the alignment function with a σ value of 2 mm (more distorted).These different fluence characteristics can be adjusted during treatment planning based on the clinical and / or prescription objectives of each patient's target region and / or OAR. For example, as depicted in FIGS. 27 to 32, the treatment planning system can balance the trade-off between fluence conformity and uniformity or distortion by repeatedly evaluating different σ values of the registration function.

[0075] In some variations, the convolution of a three-dimensional or two-dimensional image with a registration function (e.g., a delta function, a Gaussian function, a truncated Gaussian function) can be equivalently represented as an operation on the three-dimensional or two-dimensional image followed by a shift by an interpolation operator. The shift using the delta function can be implemented as linear, Fourier, or bicubic interpolation of a 2D or 3D image. The operation on the three-dimensional or two-dimensional image can be filtering by a Gaussian filter or a truncated Gaussian filter.

[0076] Examples and variations of the methods described herein may refer to the use of the delta function, but it should be understood that any of the functions described above, e.g., a Gaussian function, a truncated Gaussian function, can be used instead.

[0077] The delta function described above can encode the position of the patient's target region via a registration reference point (i.e., the delta function is a function of the 3D coordinates of the registration reference point), but optionally, the virtual registration method can include a second rotation transformation function that encodes the orientation of the patient's target region. For example, the rotation function R can be a registration function for the pitch, yaw, and roll of the patient table, which can represent the orientation of the patient's target region in the patient. That is, the virtual registration method can include two registration functions, namely, a first delta function representing the reference position of the patient's target region and a second rotation transformation function representing the orientation of the patient's target region. That is,

Number

[0078] Just as the shift can be projected at each beam angle, the fluence map rotation operation can also be projected at each beam angle. The 3D rotation operation can be a 2D affine transformation of the fluence map (i.e., rotate the fluence map in 3D and then project it onto the same 2D beam's-eye view space). The 3D rotation operation implementing pitch and yaw corrections can be expressed as follows.

Number

[0079] For a ring gantry radiation therapy system, an example of roll correction can be implemented by adding a single offset to the gantry encoder position and rolling the plan relative to the patient. For multiple targets, there may be no single roll offset that can be implemented to roll each of the individual beam filters individually. Assuming that i is an index representing the beam positions around the gantry, an example of virtual roll correction that can be implemented for a ring gantry system with beam filter p i is by interpolating the filter over the entire beam angle. An example of a radiation therapy system with N fp beam positions and a rotation angle φroll is negative with respect to the z-axis,

Number

Number

Number

Number

[0080]

Number

Number

[0081] Angle φ roll After rotating only by, a new injection filter p' can be used instead of the previous filter p. This can impart an order of operations that is consistent with applying an offset to the gantry position encoder to implement roll correction. The advantage of calculating a new set of filters in this way is that any fluences calculated can be summed together. The fluences of all patient target regions can be summed because these fluences are on the same rotation grid.

[0082] In some ring gantry radiotherapy systems, patient target regions that are not at the isocenter can undergo virtual rotation of the patient target region, rotation of the reference system, and virtual shift of the fluence. Figure 33 depicts the virtual roll correction of an IMRT / SBRT treatment delivery system. Locate δ at the virtual rotation center of the patient target region.

Number

[0083] φ roll corresponds to the roll amount at a target offset from the isocenter. In some variations, the virtual roll correction may involve rotating the patient target region about and moving it based on the location of the patient target region, as depicted in FIG. 33, by rotating the entire reference system by φ roll . Δx and Δy may correspond to the difference between the original location xy of the target and the initial location after rotation of the reference system after rotation by the angle φ roll . Δx = x(1 - cos φ)+ y sin φ Δy = y(1 - cos φ)+ x sin φ

[0084] Δr θ corresponds to the amount of leaf offset at the location of each treatment radiation source (e.g., linac) to shift the fluence to the center of the new location. For a fan beam treatment radiation source, ΔW may correspond to the distance between the leaves at the target and is divided by the distance between the source of the linac and the isocenter of system D iso and is based on the ratio of the distance to target D t . θ corresponds to the angle between the emission positions around the ring gantry.

Number

[0085] In some variations, the leaf offset may be approximately related to the amount by which the patient target region has moved in both Δx and Δy within the rotation plane of the radiation therapy system. The farther the patient target region is from the source, the greater the leaf shift may be. In some variations, a more accurate solution for Δr θ may be obtained using a convolution operation that calculates the leaf offset.

Number

[0086] In this formulation, the location of the patient target region can be projected into the fan beam projection space. The location of the patient target region δ LOC after being rotated by an angle φ can be projected into the fan beam projection space at a given injection position i. By reversing the projection of the delta function and convolving it with the location of the patient target region, this operation can shift the fluence. The inverse of the projected delta function can be implemented to transpose the discrete representation of the delta function. Alternatively, the inverse of the delta function can represent the negation of the shift of the delta function. This shift with rotation implements a virtual rotation of the planned fluence centered at any patient target region location.

[0087] Here, the position registration reference point is selected in a treatment session with 3D coordinates (x, y, z), and the pitch, yaw, roll, and the positions of x, y, z reflect the orientation of the patient table / couch in the patient setup (which may correspond to a predetermined orientation of the patient table during treatment planning). Registering the patient target region both in terms of position and orientation can correspondingly shift the delivered fluence at the injection position i. f delivery,i =δ delivery,i *p i

[0088] where δ delivery,i is the projection of δ delivery on the injection position i, and p’ i is the projection of the injection filter p’ on the injection position i.

[0089] Figure 34 depicts a schematic diagram of the virtual roll correction of the BgRT treatment delivery system. For the virtual roll at the patient target region, the fluence at each injection position i can be a combination of the roll combined with a shift of the delivered fluence.

Number

[0090] The shift-invariant injection filter p calculated by the treatment planning system for a specific patient target region based on the planned position registration reference point iWhen using, the virtual position verification of the patient target area during treatment involves obtaining a position verification image of the patient in the treatment position (e.g., aligned or set up on the patient table), identifying the patient target area in the image, selecting a position verification reference point corresponding to the planned position verification reference point, calculating the delta function δ local and convolving the delta function with the shift-invariant emission filter p i to calculate the fluence for the delivery f_delivery i at each emission position, and may include. f_delivery i =p i*δ_locali wherein, δ_local i =proj i (δ local )

[0091] Convolving the emission filter with a delta function centered on the position registration reference point has the effect of moving (e.g., shifting) the planned fluence map to reflect the current location of the patient target region. After calculating the delivery fluence, the radiation therapy system controller then segments the delivery fluence into machine instructions (e.g., MLC configuration, treatment radiation source parameters, etc.) and then executes it to deliver the prescribed fluence. Virtual position registration and machine instruction segmentation can be performed during the treatment session, preferably in real time. This is in contrast to other treatment planning and radiation therapy systems where the machine instructions are calculated during the treatment planning phase rather than during the treatment session, resulting in a set of machine instructions that deliver radiation to the location of the patient target region in the planning image rather than to the current location of the patient target region. In the case of SBRT / IMRT radiation delivery, after virtual position registration and segmentation, the radiation therapy system can then proceed to deliver the prescribed fluence to the patient target region by following the newly segmented machine instructions. In the case of BgRT radiation delivery, the region of interest or ROI (e.g., biological emission zone or BFZ) can be adjusted (e.g., shifted) using virtual position registration, and the delivery fluence map may not be segmented into machine instructions until the delivery fluence map is further updated with imaging data (e.g., positron annihilation emission path data) acquired during the treatment session. For example, approximately 500 milliseconds before delivery of the treatment radiation beam, the radiation therapy system can update the delivery fluence with imaging data acquired in a 500 millisecond window and then segment the delivery fluence immediately prior to delivery. When multiple patient target regions are to be irradiated in a single session, the virtual position registration for each patient target region can be performed in a single batch at the start of the treatment session and / or sequentially throughout the treatment session (e.g., register the first patient target region, segment the delivery fluence into machine instructions, irradiate the first patient target region, register the second patient target region, segment the delivery fluence into machine instructions, irradiate the second patient target region, etc.).

[0092] When moving the fluence map to reflect the current location of the patient target area, some position registration methods can consider and compensate for one or more physical effects that shift the fluence map. Examples include the MLC leaf tongue and groove (T&G) effect, the non-flat therapy beam effect, and the effect of attenuation of the therapy beam intensity as the inverse square of the distance from the treatment radiation source. Optionally, to calculate a delivery fluence map that takes into account one or more of these effects, the convolution of a position registration function (e.g., a delta function, a Gaussian function) with an injection filter may be convolved with an additional element. To compensate for a non-flat therapy beam (i.e., a beam intensity that varies across the entire irradiation field in the IEC-X and IEC-Y directions where the central part of the field has a greater beam intensity than the edges of the field), a virtual flattening filter correction factor (FF) can be applied to each projection, and the virtual flattening filter correction factor is a matrix that is the reciprocal of the beam intensity flat profile of the treatment radiation source beam along the MLC leaf dimension. (f i ) = FF·(p i*δi )

[0093] The virtual flattening filter correction factor (FF) matrix can be a two-dimensional matrix whose number of rows matches the number of MLC leaves and whose number of columns matches the number of patient table beam stations. Figure 22B depicts an example of a virtual flattening filter correction factor (FF) for a radiotherapy system having 64 MLC leaves and 20 patient table beam stations.

[0094] Calculating a delivery fluence map to compensate for the inverse square beam intensity reduction over the distance from the treatment radiation source may include applying a distance compensation scaling factor

Number

Number

Number

[0095] The delivery fluence map considering both the non-flat treatment beam and the inverse square beam intensity reduction is as follows.

Number

[0096] The virtual position verification method described in this specification can be used for the irradiation of multiple patient target regions in a single treatment session or fraction. A method for the irradiation of multiple patient target regions in a single treatment session is to define a patient region or treatment area (e.g., defined by a treatment planning system regardless of the presence or absence of user input) for each patient target region, and to select a position verification reference point for each patient treatment area for virtual position verification according to the steps described in this specification; and then to emit a therapeutic radiation beam according to the delivered fluence map calculated by virtual position verification for each patient treatment area. The patient treatment area can include the area of the patient that is irradiated for a single physical setup and position verification. For example, the treatment area can be the area of the patient that is irradiated with respect to a specific patient position on the table (e.g., a patient position such as a location along the table, an arm position, etc.) and a set of table locations of IEC-Y. The physical setup and position verification can include the acquisition of unique laser position alignment and / or position verification images (e.g., CT or MRI or PET images) of one or more patients, the acquisition of patient positions (e.g., raising the arm, lowering the arm, abdominal compression, breath holding), and / or one or more patient table adjustments such that the patient position coincides with the position during treatment planning. The patient treatment area can span the entire area of the patient that is irradiated for a specific physical setup and position verification. Each treatment area can have its own physical setup and position verification. In some variations, the patient treatment area may be represented by a set of IEC-Y patient table positions that span the entire irradiation area of the patient. For example, the acquired image and / or the treatment planning image can be divided into a first treatment area and a second treatment area. Each treatment area can include one or more patient target regions such that one or more patient target regions in the acquired image can be compared with one or more corresponding patient target regions in the treatment planning image. Each treatment area can correspond to a part of the patient in which one or more patient target regions that are irradiated when the patient is aligned according to a specific setup during a specific part of the radiation treatment session are located.The treatment area can be mapped to a range of patient table positions or steps (e.g., beam stations) and / or the range of movement of the patient table along the longitudinal axis of the patient table where one or more target areas of the treatment area intersect with the radiation beam of the treatment radiation source. The treatment area can correspond to a particular patient position and / or table orientation. For example, a first treatment area can be associated with a first patient position and orientation for irradiation of a first patient target area within the first treatment area, and a second treatment area can be associated with a second patient position and orientation for irradiation of a second patient target area within the second treatment area. Thus, in some embodiments, a first treatment area can be associated with a first part of a radiotherapy session (e.g., movement of the patient table through a first set of beam stations), a first patient target area, and a first position of the patient, and a second treatment area can be associated with a second part of the radiotherapy session (e.g., movement of the patient table through a second set of beam stations), a second patient target area, and a second position of the patient.

[0097] Optionally, the central regions of the defined patient treatment areas can be aligned with each other (e.g., along an "isoline") such that there is no need to adjust the patient position (except for the longitudinal IEC - Y movement to each patient table beam station where IEC - X and IEC - Z are fixed) during the emission of treatment radiation for each patient treatment area. This can help reduce the incidence of the user entering the treatment bunker, which can significantly increase the overall treatment time, during the treatment session. The center of the first treatment area and the center of the second treatment area can be on the same straight line along the IEC - Y axis and / or in the same plane as the IEC - Y axis. In some variations, the first treatment area and the second treatment area may overlap, while in other variations, the treatment areas may not overlap. The centers of gravity can be along IEC - X in the same plane (e.g., along an "isoline"). The center of the treatment area does not necessarily have to be the center of the target.

[0098] The position registration reference point for each patient target area (or group of patient target areas) can be different from the central area or point of the treatment area. In some variations, the treatment area defined during treatment planning (e.g., the planned patient treatment area) may be selected based on specific clinical criteria. For example, target areas in different axial planes may be separated into different planned patient treatment areas, and / or target areas separated by more than about 5 - 10 cm from another target area may be separated into different planned patient treatment areas. So that the virtual position registration of the patient target area can be performed separately during the treatment session if necessary, during treatment planning, a set of shift-invariant emission filters can be calculated for each patient target area and / or each group of patient target areas sharing a position registration reference point using the methods described herein. The emission of the treatment radiation beam according to the delivery fluence map of each planned patient target area may be performed sequentially, and in some variations, may include a separate physical patient setup (e.g., by adjusting the orientation of the patient table) before the delivery of radiation to the patient target area of the next patient treatment area. Alternatively or in addition, the delivery fluence maps for each patient target area are delivered in parallel (rather than by moving the patient table stepwise through discrete beam station locations along the vertical axis, i.e., IEC - Y), and can be, for example, segmented together and delivered without a physical patient setup between the target areas. In the case of a treatment session in which one or more of the patient target areas are to receive BgRT radiation and these BgRT patient target areas are registered using virtual position registration, a virtual shift is applied to the ROI. A treatment planning method incorporating virtual position registration for one or more BgRT targets involves selecting a planned position registration reference point for the BgRT patient target area, defining a region of interest (ROI) (e.g., a biological fluence zone (BFZ)) around the patient target area, and calculating a shift-invariant emission filter (e.g., a radiation emission matrix (RFM) or radiation emission filter) based on the ROI including the patient target area and the planning guidance image (e.g., a planned PET image, an MRI image, etc.) iCalculating and may include. The ROI can be a spatial mask or filter that defines a patient area or region that may include the patient target area and a margin around the patient target area. For example, the ROI can include a tumor area (or any patient target area), and the margin around the tumor area can account for the location estimation error of the tumor area, and / or the movement of the tumor area, and / or the possible location of the tumor area during radiation delivery, and / or the geometric changes of the tumor area. During a treatment session, the radiation therapy system can acquire imaging data (e.g., positron annihilation emission path data or LOR data when the patient is injected with a PET tracer, SPECT data when the patient is injected with a SPECT tracer, gamma ray data, MRI data, CT data, and / or X-ray data) that intersects and / or co-localizes with the ROI for its BgRT target area. Imaging data that does not intersect the ROI or includes data related to structures outside the ROI is not included in the calculation of the delivery fluence map. In this way, the ROI functions as a spatial mask or filter that can be applied to the imaging data acquired during a treatment session.

[0099] A radiation delivery method that merges virtual position registration with BgRT delivery can include acquiring an image of the patient at a treatment position that includes the BgRT patient target area and the ROI, selecting a registration reference point within the acquired image, calculating a delta function based on the registration reference point, and adjusting the ROI by convolving the delta function with the ROI, acquiring additional image data during the session, calculating the fluence for delivery to the patient target area at each emission position of the treatment radiation source by convolving the imaging data with an emission filter, and proceeding to emit the fluence for delivery. (f i )=(p i *proj i (x·(ROI*δ i,delivery )))

[0100] Here, x is the additional imaging data obtained (e.g., partial imaging data, imaging data obtained within a time limit, positron annihilation emission path data, MRI data, SPECT data, gamma data, CT data, and / or X-ray data), and p i is an injection filter (also known as a BgRT injection filter or radiation emission matrix) that corresponds to the filter used in virtual position matching. In this case, the virtual position matching functions to shift the ROI by δ delivery . Only the imaging data co-localized with the adjusted ROI can be used to calculate the fluence for delivery. The additionally obtained imaging data can be spatially filtered (or masked) by the ROI so that the fluence for delivery can be calculated using only the imaging data co-localized with the adjusted ROI.

[0101] Optionally, for BGRT, each time the filter p i rotates, a roll correction can be applied, and the ROI can rotate by the same angle. One example is to implement a 2D rotation function that rotates the ROI mask to a predetermined position for each plane along Z. The order of operations can be important when embedding the roll correction. In this example, the roll correction is applied first, and then translated by only the virtual offset. ROI’ = δ(Δx,Δy) * ROT(ROI, φ roll ) Δx = δx(1 - cosφ) + δysinφ Δy = δy(1 - cosφ) + δxsinφ

[0102] Here, the injection filter rotates to match the new reference system. [Number]

[0103] Here, the fluence at a given injection position f i is rotated, masked by the rotated and shifted ROI, shifted by the virtual offset δ delivery , projected onto a given injection angle i, and the rotated filter p’ iIt is a projection of a set and the imaged data that has been convolved. (f i )=(p’ i *proj i (x·(ROI’*δ delivery ))) ROI’=δ(Δx,Δy)*rot(ROI,θ roll ) (f i )=(p i *proj i (x·(ROI’*δ delivery )))

[0104] Optionally, a delivery fluence map that takes into account both the non-flat treatment beam and the inverse square beam intensity reduction for the BgRT target region is as follows.

Equation

[0105] The therapeutic radiation can be delivered to the BgRT target and / or the SBRT / IMRT target, sequentially and / or in parallel, in a single session, if desired.

[0106] Optionally, a bounded dose volume histogram curve (bDVH) showing the minimum and maximum dose values over the entire patient target region volume, taking into account the possible dose variability due to patient target region displacement, may be calculated during treatment planning after the shift-invariant emission filter has been calculated. Outputting a visualization graphic, such as a bDVH or any of the visualization graphics described herein, during treatment planning and / or during patient setup and / or target region positioning can help the user evaluate whether the treatment plan is appropriate for the patient (e.g., with respect to the day of treatment).

[0107] After multi-target virtual position matching, each of the patient target regions can have a dose component that is combined between different patient treatment areas. This dose combination can be the result of the potential for extra fluence beams to interact with some of the patient target regions. Given a set of virtual position matches, the dose to the patient target regions, OARs, and / or treatment areas can be calculated. A set of normalization factors can be applied to each patient target region and / or treatment area to ensure that a specific dose measurement goal is met or, alternatively, that a specific OAR dose is constrained. An example of implementing these factors for each IMRT / SBRT virtual multi-position match j from a set of N position matches is to solve for the factors for each position match β j to normalize the combined effect of all virtual position matches (e.g., all virtual position matches in the entire treatment plan, all virtual position matches in each treatment area). An example of such a combined solution is to minimize the root mean square of the dose measurement scalar D j associated with each patient target region and / or treatment area.

Number

[0108] For IMRT / SBRT,

Number

[0109] For BgRT,

Number

[0110] The dose measurement scalar D j can be any of the following, namely, the mean dose of the PTV, the percentile dose of the PTV, the maximum value of the PTV. Alternatively, this dose measurement scalar D jcan be obtained based on the function of the dose to the OAR linked to the target region. Instead of the mean square error such as the weighted least squares method, other convex minimization algorithms can be used.

[0111] Some of the virtual position registration methods described above and in this specification use a delta function derived from a registration reference point and convolve this delta function with a shift-invariant projection filter, but it should be understood that the virtual position registration method includes other registration functions that encode the orientation of the patient target region based on the pitch, yaw, and roll of the patient table or couch. For example, the virtual position registration method described herein may include convolving a set of filters with a registration function including a delta function, a Gaussian function, and a truncated Gaussian function on a registration reference point that can be selected during the treatment session.

[0112] Mosaic multi-target registration As briefly described above, one of the challenges in treating multiple patient target regions in a single treatment session is that the positioning / registration suitable for one patient target region may not be suitable for the remaining patient target regions. Methods for positioning / registering multiple patient target regions may include registering each separately (e.g., using a virtual registration method where a planned full fluence map is moved, and / or a physical registration method where the patient is moved using a table). For example, a treatment session may include one physical positioning / registration for a patient target region and one or more virtual registrations for the remaining patient target regions. Using virtual registration, multiple patient target regions can be irradiated without the need to physically move and / or reposition the patient to register each target region. In a variation where the treatment plan can be calculated and optimized for all of the patient target regions, i.e., a single treatment plan that specifies the 3D fluence for every patient target region, the “global” treatment plan (e.g., the “global” fluence map) can be divided into discrete sub-regions. The targets of each sub-region can be registered independently based on a single registration image or aligned in a multi-target treatment plan. Typically, the treatment plan fluence map may include high fluence regions that are located (e.g., co-localized) over target region low fluence regions in almost all other locations. These low fluence regions may sometimes be referred to as low fluence bands or boundary regions. Methods for treatment planning may include generating a global treatment plan or global fluence map having high fluence regions that are surrounded by or bordered by low fluence regions or low fluence bands and co-localized with target regions, and may further include dividing the global treatment plan or fluence into sub-regions that are surrounded by or bordered by low fluence regions. Conceptually, the global treatment plan can be “cut” along the centers of these low fluence bands and separated into discrete regions around each target region.These treatment plan fluence map sub-regions or “mosaics” can be independently shifted and oriented around each target using a positioning image at the start of a treatment session. The planned fluence map sub-region shifts can result in fluence overlap at the low-fluence boundary regions of adjacent sub-regions (i.e., the border regions can receive fluence during irradiation of the target regions of two or more adjacent sub-regions), and the low-fluence levels of these boundary regions can be specified to be low enough during treatment planning so that this fluence overlap is not clinically problematic. During treatment planning, the low-fluence boundary regions can be defined wide enough so that the amount of overlap that can be tolerated is large enough to address all sources of positioning / registration error. In some variations, adjustment structures and / or constraints may be added during treatment planning to achieve the desired low-fluence boundary regions. After registration of the individual patient target regions of the independent fluence map sub-regions, the delivered fluence map for each sub-region can be segmented into machine instructions for execution by a radiation therapy system. Radiation can be delivered continuously and / or simultaneously to multiple registered patient target regions or fluence map sub-regions. For example, some patient target regions or fluence map sub-regions can be irradiated sequentially while other patient target regions or fluence map sub-regions can be irradiated simultaneously with other patient target regions or fluence map sub-regions. A single positioning image including all of the patient target regions can be used to register / align each patient target region, or multiple positioning images (e.g., one positioning image for each patient target region or fluence map sub-region) can be used. For example, during a treatment session, there can be one physical registration for each treatment area (as defined during planning), and one or more virtual registrations for one or more patient target regions within the treatment area. The mosaic registration method can be used for SBRT / IMRT delivery and / or BgRT delivery, and optionally, in combination with other registration methods and patient setup methods (e.g., in combination with a rigid multi-registration method where a “best fit” shift based on a positioning image is applied to the patient via the table and movement of the patient table, a virtual registration method).

[0113] Hybrid mosaic multi-target position alignment Optionally, during treatment planning for mosaic position matching when a global planned fluence map is divided into planned fluence map sub - regions, this method may include specifying whether the patient target region of each fluence map sub - region should be treated using the SBRT / IMRT method (e.g., designated as an SBRT / IMRT target region) or the BgRT method (e.g., designated as a BgRT target region). The shift - invariant beam filter can be calculated based on the planned position - matching reference points selected for all target regions as described herein for virtual position matching. Optionally, a bDVH curve may be calculated for each patient target region, which may include generating a series of simulations of all potential shifts for the BgRT target regions and / or SBRT / IMRT target regions. The bDVH curve can be reviewed and approved by the clinician at the time of planning. A method for hybrid mosaic multi - target position matching may include obtaining a large - volume CT (e.g., a single large - volume CT) of multiple SBRT and BgRT target regions, depicting each set of patient couch adjustments (e.g., 6DOF corrections) for each patient target region based on the CT position - matching image, and selecting a position - matching reference point for each patient target region. To align / register a particular patient target region, the patient couch is first moved according to the depicted set of patient couch adjustments, and the delivery fluence map can be calculated using virtual position - matching techniques applied to the selected position - matching reference point for that patient target region. For BgRT delivery, a PET pre - scan can be obtained in conjunction with the acquisition of the position - matching image. Based on the PET pre - scan, a predicted dose to the BgRT target can be calculated immediately prior to the activation of the treatment radiation source for irradiation of the BgRT target region, which can help ensure that the actual position - matching and PET pre - scan images are within a pre - approved dose range. The real - time delivery fluence maps for the BgRT and SBRT / IMRT target regions can be smoothly joined together and segmented into machine commands on - the - fly (i.e., minutes or seconds before delivery).In beam station delivery where the patient table remains stationary in the vertical position while the therapeutic radiation source moves around the patient to deliver the calculated fluence, the delivered fluence maps for BgRT and SBRT / IMRT target regions at a particular beam station can be added together and can be divided among machine instructions while the table is stopped at the beam station. In this way, the delivered fluence for multiple BgRT and SBRT / IMRT target regions can be delivered simultaneously.

[0114] Separate multi-target alignment A method for treating multiple patient target regions in a single treatment session can include calculating individual treatment plans (i.e., individual planned fluence maps that specify 3D fluence for individual patient target regions) for each patient target region instead of calculating a global treatment plan (i.e., a global fluence map that specifies 3D fluence for all patient target regions). In some variations, the method for treatment planning for separate alignment calculates individual treatment plans for each target region regardless of the presence of the remaining target regions, taking into account the presence and location of OARs and / or other critical structures, combining the individual treatment plans with each other to form a combined treatment plan, and constraining the combined plan such that high fluence regions from the individual plans do not overlap with other high fluence regions and also meet the original goals such as dose coverage and dose limits for the OARs while retaining the individual plans as separate entities. Constraining the combined treatment plan so that the treatment plan can accommodate independent movement of the target regions can create a planned global fluence map having a low fluence region or band surrounding each patient target region. In some variations, adjustment structures and / or constraints may be added to the treatment plan to achieve the desired low fluence boundary region. The combined treatment plan (i.e., the planned global fluence map) can be reviewed and approved by a clinician prior to treatment.

[0115] An example of a treatment planning method for separated multi-target alignment is to generate a treatment plan for irradiating a first patient target in a first treatment area and a second patient target area in a second treatment area, and to limit the irradiation of the OAR to less than a first dose amount while irradiating the first patient target area in the first treatment area with a first prescription dose. A first treatment plan, a second treatment plan for irradiating the second patient target area in the second treatment area with a second prescription dose while limiting the irradiation of the OAR to less than a second dose amount, dividing into, combining the first treatment plan and the second treatment plan to generate a combined treatment plan, and repeatedly modifying the combined treatment plan to meet the desired dose goals and constraints. Examples of dose goals and constraints may include preventing hot spots or cold spots in the combined treatment plan. The first treatment plan may include an optimized radiation beam that delivers a first prescription dose to the first patient target area, and the second treatment plan may include an optimized radiation beam that delivers a second prescription dose to the second patient target area. The dose amount or level to the OAR as a result of the individual (e.g., first and second) treatment plans may be set to a specific percentage of the maximum dose amount. For example, the first dose amount to the OAR (e.g., an OAR located between or spanning both the first and second treatment areas) from the first treatment plan may be 35% of the maximum dose level, and the second dose to the OAR from the second treatment plan may be 65% of the maximum dose level. The specific contribution values of the first and second treatment plans to the OAR dose may be dynamically set based on the evaluation of the goals in the combined treatment plan based on iterative calculations within the co-optimization, or through a series of separate optimizations at different contribution levels until an acceptable plan is achieved. After the combined treatment plan is modified to meet the desired dose goals and constraints, the combined treatment plan may be divided into the first and second treatment plans and may also have optimized beams and / or fluences for delivery.In some examples, the combined treatment plan may include a planned fluence map, the first treatment plan may include a first fluence sub-map, the second treatment plan may include a second fluence sub-map, and the first and second fluence sub-maps are combined to form the planned fluence map.

[0116] Optionally, in some variations, the user may select one patient target area as having a higher priority than the other. For example, if the user defines the irradiation of the first patient target area as having a higher priority than the irradiation of the second patient target area, the treatment planning method divides the treatment plan into a first treatment plan for irradiating the first patient target area of the first treatment area with a first prescription dose while restricting the irradiation of the OAR to less than the selected dose amount, and a second treatment plan for irradiating the second patient target area of the second treatment area with a second prescription dose, combines the first and second treatment plans to generate a combined treatment plan, and may include repeatedly modifying the combined treatment plan to meet the desired goals and constraints. In this method, the first treatment plan may have fewer dose constraints (e.g., no OAR dose constraint) than the second treatment plan such that the first treatment plan includes beams optimized to deliver the prescription dose to the first patient target area with little or no consideration of the dose to the OAR. Then, the second treatment plan may consider the dose delivered by the first treatment plan as a "pre-dose" that constrains the beam optimization of the second treatment plan for dose delivery to the second patient target area.

[0117] During a treatment session, individual patient target regions can be individually aligned using one or more alignment images (e.g., using typical alignment methods such as adjusting the patient table or changing radiation therapy machine commands and / or using the virtual alignment methods described herein). After all of the patient target regions have been aligned, the delivery fluence maps for each patient can be combined with each other to check for "overlapping" dose regions that exceed any constraints. If the desired dose constraints are met, the radiation therapy system can proceed to divide the delivery fluence map into machine commands and deliver therapeutic radiation to the patient target regions. Delaying the segmentation or translation of the delivery fluence map into radiation therapy system commands as close as possible to delivery allows the radiation therapy system to update the delivery fluence to reflect the actual location of the patient target regions (and / or any changes in location), resulting in a more accurate delivery of the overall prescribed dose.

[0118] Biologically Induced Radiation Therapy (BgRT) Alignment Briefly stated, BgRT is a radiation therapy method that updates the irradiation fluence using imaging data acquired during a treatment session and delivers therapeutic radiation within minutes or seconds of acquiring the imaging data. Due to the low latency of BgRT, the imaging data used to update the delivery fluence is relatively sparse (e.g., "partial image data" that is collected over a short or limited time window and is insufficient for a complete image reconstruction), and / or may contain a lot of noise (e.g., having a signal-to-noise ratio that does not allow for a reliable determination of the centroid of a patient target region). For example, in BgRT, the imaging data used to update or calculate a delivery fluence map may include a partial PET image that includes one or more lines of response (LOR), a partial MRI image that includes subsampling of k-space, and / or one or more x-ray projection images. Although the examples herein are described in the context of the PET imaging modality, it should be understood that any other imaging modality can be used alone or in combination with the PET imaging modality. The imaging data can be acquired over a limited time window, e.g., about 3 seconds or less, about 2 seconds or less, about 1 second or less, about 500 ms or less, about 300 ms or less, 200 ms or less, etc. Because the time from image acquisition to radiation delivery is low latency (e.g., about 10 seconds or less, about 5 seconds or less, about 3 seconds or less, about 1 second or less, about 500 ms or less), radiation can be delivered to a target region before the target region moves. A ROI (e.g., a biological emission zone or a biological target zone) can be defined around each BgRT target region that represents a volume used to determine which LORs to use to direct fluence to the target. That is, a ROI can be a spatial filter or mask that can be applied to LORs, and if a detected LOR intersects the ROI, this LOR is used in the BgRT delivery algorithm to update the delivery fluence, otherwise this LOR is ignored. It may be preferred for the ROI to be small, as this can reduce the likelihood that non-target PET-avid tissue is included in the ROI (any PET-avid tissue in the ROI is treated as if it were the target).

[0119] The BgRT-based multi-target alignment method may include designating one of the BgRT target regions as the alignment target region and defining an ROI around the remaining ones that is large enough to encompass the range of location shifts of those patient target regions. In some variations, the ROI of the alignment target region may be smaller than the ROIs of the remaining BgRT target regions. In some variations, the alignment target region may be the BgRT target region closest to a critical structure or OAR (e.g., a PET-avid OAR such as the heart). During the treatment session, as long as the ROI of the alignment target region is aligned / registered and the remaining BgRT target regions are within the ROI (the ROI may be larger or expanded to account for the expected spatial arrangement changes relative to the alignment target region), the radiation therapy system may proceed to deliver radiation using the BgRT method (i.e., the delivery fluence is calculated by convolving the acquired spatially filtered imaging data with a shift-invariant radiation projection matrix or an emission filter and then segmented into machine instructions for immediate execution). The impact on the dose conformality of the larger ROI is calculated, reviewed during treatment planning, and in some cases shaped or designed to help reduce the likelihood of including additional PET-avid tissue while helping to increase the probability of accommodating spatial differences between planning and delivery. BgRT-based alignment may optionally be used in combination with other alignment methods and patient setup methods (e.g., in combination with patient couch movement, virtual alignment methods, rigid multi-alignment methods where an "optimal fit" shift based on alignment images is applied to the patient via the couch).

[0120] Any of the multi-target alignment and radiation delivery methods described herein can be used alone or in combination with alignment and delivery methods typically used for single patient target irradiation and / or irradiation of multiple target regions over multiple treatment sessions. For example, any of the planning, alignment, and delivery methods described herein can be used in combination with physical patient setup (e.g., one physical patient setup per patient target region or treatment area in a sequential manner), rigid body multi-alignment, treatment plan deformation, robust treatment planning (creating a plan that takes into account the uncertainty of position during the optimization process and does not vary significantly as a function of position change), and / or online treatment plan adaptation (generating a new treatment plan based on alignment images). In sequential alignment and treatment, each patient target region is aligned and treated independently, one after the other. Immediate treatment after alignment can help reduce the probability of static patient displacement during treatment, but may require acquisition of multiple alignment images (e.g., multiple kVCT images) for each alignment instance. The treatment plan for sequential alignment and treatment can include jointly optimizing the fluence map over multiple patient target regions and / or can be optimized to be robust to the individual displacements of each tumor when individually aligned. Rigid body multi-alignment can include acquiring alignment images and then calculating a single shift of the alignment images relative to the planning images that minimizes the function of the total alignment error over all target regions and / or OARs (i.e., performing a "rigid body" alignment). Taking these expected alignment errors into account, the margins around each target region of the initial treatment plan can be expanded. The targets and OARs within the treatment plan can be shifted and oriented as rigid bodies to find the "best" mapping to the actual positions of the targets and OARs in the alignment images.The user may choose to consider all alignment errors of equal weight and minimize the total alignment "error volume", or instead, the user may weight some target regions or OARs as more important than others and sacrifice an increase in alignment error elsewhere or an increase in the tumor margin in order to better align those targets.

[0121] For example, mosaic multi-target registration and BgRT-based registration can be combined with one or more of the methods described above. Some of the targets in a multi-target treatment plan may be expected to move more as a rigid body (e.g., target regions associated with the same bone) and can thus be grouped together and aligned using a rigid body method, while the remaining patient target regions can be aligned using mosaic multi-target or BgRT-based registration. A robustness algorithm can be used during treatment planning to help ensure that spatial adjustments made to each treatment plan or fluence map sub-region introduce a minimal amount of dose variability. Also, a robust plan can be used in combination with BgRT-based virtual registration to optimize the shape and placement of the ROI for each patient target region. Also, it may be desirable to align patient target regions using mosaic multi-target or BgRT-based registration techniques and then apply a soft tissue deformation method to each sub-region or target region. This can help reduce the amount of tissue deformation that must be considered and thus can help reduce the dose variability introduced by the soft tissue method. Alternatively or in addition, a mosaic multi-target registration method can be used to reduce the computational load of online treatment plan adaptation. For example, patient target regions may be aligned based on a planned fluence map using a mosaic multi-target registration method, and then the fluence map for each patient target region may be adapted individually. In yet another example, a BgRT treatment plan may incorporate treatment plan steps from a mosaic multi-target registration method, which can help reduce the size of the ROI for each target region.

[0122] In another variant, the treatment plan for a plurality of patient target regions may include a plurality of treatment areas, and the patient may be physically set up / positioned for each treatment area. Each treatment area may have one or more tumors or patient target regions (e.g., ROI). During a treatment session, the patient may be physically set up / positioned for a first treatment area, and the radiation therapy system may virtually position each of the patient target regions within the first treatment area. For example, physical positioning may provide proper positioning for a first patient target region of the first treatment area, and then the radiation therapy system may use the virtual positioning methods described herein to adjust the fluence for the remaining patient target regions of the treatment area. The fluence for delivery may be calculated or adjusted according to the positioning reference points selected during the treatment session from the positioning images. The irradiation of the patient target regions of the first treatment area may be performed continuously and / or in parallel. After the first treatment area has been irradiated, the patient may then be physically set up / positioned for a second treatment area (e.g., moving the patient table along the IEC-Y), and the radiation therapy system may virtually position each of the patient target regions within the second treatment area. The patient target regions of the second treatment area may be irradiated as described above for the first treatment area. The first and second treatment areas may overlap in some variants and may not overlap in other variants.

[0123] Methods for multi-target treatment planning, registration, and radiation delivery may optionally include generating one or more visualization graphics that can be output to a display device (e.g., a monitor). The visualization graphics may include a dose-volume histogram (DVH) or a bounded dose-volume histogram (bDVH) for each of a plurality of patient target regions, and the user may be informed that dose can be delivered to the patient target regions. However, the DVH and bDVH curves do not provide information regarding the spatial interaction between fluence maps of different target regions. For example, when the fluence map is shifted and / or updated during registration, fluence regions that did not previously overlap during treatment planning may overlap during the treatment session. In mosaic registration, fluence map sub-regions can be registered and shifted individually, and in separate registrations of different patient target regions, the individual fluence maps of different patient target regions can be registered and shifted individually. Some overlapping areas may include low fluence areas of two or more individual fluence maps or fluence map sub-regions, so the cumulative fluence of the overlapping areas may not exceed the dose safety threshold of non-target tissue, but such fluence overlap can exceed the dose safety threshold and / or co-localize on an OAR. The visualization graphics are generated based on registration data (e.g., shifted fluence maps or fluence map sub-regions, registration fiducial points, registration images, etc.) and output to a display device to help the user evaluate the dose impact of separate registrations of multiple patient target regions and / or to help identify any unwanted dosimetric effects resulting from these registered fluence map shifts. The visualization graphics can help the user better understand the interaction between targets and encode spatial information for evaluating the uncertainty of dose delivery.

[0124] Some visualization graphics can be calculated based on the bDVH curve and calculations, the boundary of the bDVH curve, i.e., the nominal dose (D n ), the potential minimum dose (D pmin) and the potential maximum dose (D pmax ) can be referred to. The visualization graphic can depict the expression of the specific dose distribution and the probability of levels on a specific anatomical structure in 2D slices and / or 3D renderings. Movies or short animations can also depict the minimum, nominal, and maximum dose levels so that visual comparisons can be made between these levels. The differences or variations therefrom from the nominal dose (e.g., delta dose) can provide the user with a visual tool for identifying hot spots and cold spots.

[0125] Multi-target treatment planning method Treatment planning methods for treating multiple patient target regions in a single treatment session may, if desired, calculate various fluence maps (e.g., global fluence maps for all patient target regions and / or OARs, fluence map sub-regions including subsets of target regions and / or OARs), injection filters (e.g., shift-invariant injection filters), and spatial filters (e.g., for spatial filtering of imaging data, regions of interest or ROIs, biological injection zones or BFZs, biological target zones). Treatment planning methods for treating multiple patient target regions in a single treatment session may include calculating one or more shift-invariant injection filters for each patient target region for virtual position alignment, and / or calculating fluence map sub-regions divided by low fluence bands for mosaic multi-target position alignment, and / or generating separate treatment plans or fluence maps for each patient target region, and constraining the cumulative fluence map for separate multi-target position alignment. A BgRT treatment planning method for treating one or more BgRT patient target regions (alone or in combination with one or more SBRT / IMRT patient target regions) may include designating one BgRT patient target region as the alignment target region and defining the ROIs of the remaining BgRT patient target regions to correspond to the movement of the alignment target region and the movement of the remaining target regions. A treatment planning method for treating both SBRT / IMRT and BgRT patient target regions (e.g., hybrid mosaic multi-target position alignment) may further include specifying whether the patient target region is a BgRT target region or an SBRT / IMRT target region.

[0126] The treatment planning methods described herein can be used separately or in combination to develop a multi-target treatment plan suitable for a particular patient. Some clinicians or clinics may configure a treatment planning system to calculate treatment planning parameters that can accommodate virtual position registration and mosaic multi-target position registration of multiple patient target regions with continuous or parallel radiation delivery to the patient target regions. Alternatively or in addition, the treatment planning system can be configured to calculate treatment planning parameters that can accommodate virtual position registration and separate multi-target position registration of multiple patient target regions with continuous or parallel radiation delivery to the patient target regions. For example, some treatment planning systems can calculate a shift-invariant beam filter for all patient target regions, calculate a beam filter for any BgRT patient target region, and / or be configured to divide a treatment plan or a planned fluence map based on the relative positions of the patient target regions, and / or OARs, and / or low fluence areas of the planned fluence map. The BgRT patient target regions can be irradiated during a first treatment session, and the SBRT / IMRT patient target regions can be irradiated during a second treatment session (i.e., on a different day than the first treatment session, if the first and second treatment sessions do not temporally overlap, at different time intervals on the same day, etc.). Alternatively, the BgRT patient target regions and the SBRT / IMRT patient target regions can be sequentially irradiated during the same treatment session. For example, the BgRT patient target regions can be irradiated during a first shuttle pass, and the SBRT / IMRT patient target regions can be irradiated during a second shuttle pass.

[0127] Figures 1A and 1B are flowchart representations of respective SBRT and BgRT variants of a treatment planning method that uses virtual registration for registering the positions of single or multiple patient target regions during a treatment session. Figure 1A depicts one variant of a treatment planning method incorporating virtual registration for an SBRT / IMRT patient target region. The method (120) can include obtaining (122) one or more treatment plan images of one or more patient target regions, defining (124) the outer contours of one or more patient target regions and / or risk organs (OARs) and corresponding dose constraints, selecting (126) planned registration reference points for each of the one or more patient target regions and / or treatment areas, calculating (128) corresponding registration functions for each registration reference point, determining a set of treatment radiation source emission positions and calculating projections of the registration functions to each emission position (129), and calculating (130) a set of shift-invariant emission filters for each emission position based on the planned registration reference points and registration function projections for each patient target region. Defining (124) the outer contours of one or more patient target regions can include defining a group of patient target regions that can be linked or associated (e.g., attached to the same bone structure) such that the positions and / or movements of these patient target regions are likely to be correlated. The method (120) can include selecting (128) planned registration reference points for the entire group of patient target regions and calculating (130) a set of shift-invariant emission filters based on the planned registration reference points for the group of patient target regions. The calculated set of shift-invariant emission filters, and the planned registration reference points for each patient target region, group of patient target regions, and / or treatment area can be transferred to a radiation therapy system for registration (along with other treatment planning parameters) during treatment. In some variants, the calculated shift-invariant emission filters can represent a planned delivery fluence based on (e.g., centered on) the planned registration reference points.For example, for a particular patient target region (or group of patient target regions), the registration function can be a delta function centered on a planned registration reference point relative to that patient target region (e.g., the tumor centroid). The shift-invariant emission filter for a particular patient target region can represent a fluence map centered on the planned registration reference point (e.g., the planned fluence map).

[0128] FIG. 1B depicts a variant of a treatment planning method that incorporates virtual registration for BgRT patient target regions (i.e., patient target regions where the fluence delivered to those target regions is calculated based on acquired imaging and / or biological data such as PET emission data). The method (140) depicted in FIG. 1B includes obtaining (142) one or more treatment plan images of one or more patient target regions, defining (144) the outer contours of the one or more patient target regions and / or risk organs (OARs) and corresponding dose constraints, defining (146) for each patient target region a region of interest that includes a margin around the outer contour of the patient target region, selecting (148) a planned registration reference point for each of the patient target regions, obtaining (150) imaging data spatially masked by the region of interest, determining a set of emission positions of a treatment radiation source and calculating the projection of the acquired imaging data to each emission position (152), and calculating (154) for each patient target region a set of shift-invariant emission filters based on the planned registration reference point and the imaging data projection. In some variants, the boundaries of the planned regions of interest can include spatial filters. Defining (146) the outer contours of the one or more patient target regions can include defining a group of patient target regions that can be linked or associated (e.g., attached to the same bone structure) such that the positions and / or movements of these patient target regions are likely to be correlated. The method (140) can include selecting (148) a planned registration reference point for the entire group of patient target regions and calculating (154) a set of shift-invariant emission filters based on the planned registration reference point for the group of patient target regions.

[0129] Optionally, some treatment planning methods may include designating some patient target regions as SBRT / IMRT target regions and other patient target regions as BgRT target regions. Generating treatment plans for both the SBRT / IMRT target regions and the BgRT target regions may also include acquiring a PET image. For each BgRT patient target region, the method may include calculating a shift-invariant emission filter (e.g., a radiation emission matrix (RFM)) based on an ROI, one or more planning images of the BgRT patient target region, and the prescribed dose to the BgRT patient target region. Optionally, some treatment planning methods may include determining a set of treatment radiation source emission positions and calculating projections of position registration functions such as delta functions, Gaussian functions, truncated Gaussian functions, etc. for each emission position. Optionally, some treatment planning methods may include calculating a cost function from a position registration reference point for each patient target region.

[0130] The radiation projection matrix (RFM) or the delivery filter can be a matrix that specifies the conversion from a partial image to a fluence map that may include the beamlet pattern and / or beamlet intensity applied to a patient during a treatment session. The delivery filter or RFM can represent the relationship between a fluence map F for radiation delivery to a patient region and an image X of that patient region. That is, the radiation projection matrix or delivery filter P can be any matrix such that F = P·X. The delivery filter or RFM can be calculated for each patient target region during a treatment planning session in conjunction with calculating a fluence map that minimizes a cost function C(D,F), which is formed based on one or more cost functions, e.g., radiation dose constraints and goals and optional constraints on F, of the resulting dose distribution D and fluence F. Examples of cost functions can include, but are not limited to, the minimum dose to a target region, the average or maximum dose on an OAR, and / or the smoothness of the fluence, the total radiation dose, the total tissue dose, the treatment time, etc. In some variations, generating the radiation projection matrix P can include setting up an optimization problem to minimize the cost function C(D,F) and repeating through different sets of P while the following conditions are met such that the cost function C(D,F) is minimized. F = P·X and D = A·F = A·P·X where D is the predicted dose distribution, A is a pre - calculated dose calculation matrix, F is the predicted total delivered radiation fluence, and X is a known complete image (e.g., an image obtained during an imaging session and / or a previous treatment). The predicted dose distribution D and the predicted radiation fluence F can be calculated using dose constraints, patient target volumes, and / or OAR data, as well as the patient planning CT image. An example of the dose calculation matrix A is a (k x n) matrix, where n can be the number of candidate beamlets {b i}, and k can be the number of pre - selected voxels for an ROI. The i - th column (having k elements) of the dose calculation matrix A represents the dose contribution from a single weighted pencil beam b i to each of the k voxels.

[0131] The dose calculation matrix A can be calculated column - by - column, for example, by ray - tracing the aperture of each beamlet along a path through the ROI or the patient volume and calculating the contribution of the beamlets weighted by 1 to each of the k voxels. The beamlet aperture can be an MLC aperture defined by a single MLC leaf opening (i.e., binary MLC or 2D MLC). Some examples of dose calculation algorithms that can be used in any of the methods described herein include Monte Carlo simulation, collapsed cone convolution superposition, pencil beam convolution, and the like.

[0132] The radiation emission matrix P (also known as the RFM or emission filter) can be a matrix that, when multiplied by the complete image X, yields a predicted or desired irradiation fluence F that minimizes a cost function. The cost function can be convex, allowing the use of well - known convex optimization algorithms such as gradient descent, fast proximal gradient method, or interior point method. The calculated radiation emission matrix P can represent the magnification that relates the fluence F to the complete image X. This relationship can be used during the treatment session to update the fluence f i based on the partial image x i obtained at the same time point by multiplying the radiation emission matrix P by the partial image at a particular time (e.g., f i = P·x i ). Additional details regarding the BgRT treatment planning and delivery method can be found in U.S. Patent No. 15 / 993,325, filed on May 30, 2018, which is hereby incorporated by reference in its entirety.

[0133] Optionally, some methods (120) may include determining a set of treatment radiation source injection positions (132) and calculating the projection of the injection filter onto each of the injection positions. For example, a radiation therapy system having a rotatable gantry may have 100 circumferentially distributed injection positions or angles around the bore where the patient table can be advanced. The method (120) may include calculating the projection of each of the injection filters for each of the 100 injection positions. For an injection filter convolved with a Gaussian function or a truncated Gaussian function, the width (σ) of the Gaussian function centered on the mean value (μ) at the planned position registration reference point can be selected during treatment planning. For example, the width (σ) of the Gaussian function or the truncated Gaussian function may be based on the width of the MLC leaf.

[0134] FIG. 2 depicts a flowchart representation of a variant of a treatment planning method that uses mosaic multi-target registration for the registration of multiple patient target regions during a treatment session. The method (220) may include obtaining one or more treatment plan images of one or more patient target regions (222), defining the outlines of one or more patient target regions and / or OARs and the corresponding dose constraints (224), selecting the planned position registration reference points for each of the one or more patient target regions and / or OARs (228), and calculating a set of shift-invariant injection filters based on the planned position registration reference points for each patient target region and / or OAR (230). The method (220) may include generating a treatment plan fluence map based on the injection filters and the one or more treatment plan images (234), and defining a periphery including a low fluence value region around each patient target region in the treatment plan fluence map (236). The calculated set of shift-invariant injection filters, the planned position registration reference points for each patient target region and / or OAR, and the treatment plan fluence map having the defined periphery may be transferred to a radiation therapy system (along with other treatment plan parameters) for registration during treatment.

[0135] Optionally, some treatment planning methods (220) may include designating (226) each of the patient target regions as an SBRT / IMRT target region or a BgRT target region. If a patient target region is designated as a BgRT target region, the method (220) may include calculating a ROI for that patient target region. For each BgRT patient target region, the method (220) may also include calculating a shift-invariant emission filter (e.g., RFM) as described above.

[0136] Optionally, some methods (220) may include determining (232) a set of treatment radiation source emission positions and calculating a projection of an emission filter to each of the emission positions. For example, a radiotherapy system having a rotatable gantry may have 100 circumferentially distributed emission positions or angles around a bore in which a patient table can be advanced. The method (220) may include calculating a projection of each of the emission filters for each of the 100 emission positions.

[0137] Optionally, some treatment planning methods (220) may include calculating a bounded dose volume histogram (bDVH) curve for each patient target region and / or OAR (238), and displaying the bDVH curve and / or dose calculation data for each patient target region and / or OAR on a display device (240). The bDVH curve can be calculated by calculating the dose per unit volume for each possible location of the patient target region within the motion envelope (or ROI in the case of a BgRT target) and / or based on biological activity and / or physiological and / or anatomical data acquired before or during the planning session to generate a family of dose volume curves. For example, a family of bounded DVH curves can be calculated for each OAR, ROI, and / or patient target region based on an injection filter by performing a fixed shift of the PET image of the patient target region within the ROI and calculating the corresponding dose to the OAR, ROI, and / or patient target region for that particular shifted target region position. For example, for a patient having a patient target region and one OAR within the ROI, a family of DVH curves for the patient target region can be calculated for each shifted position of the patient target region within the ROI. The upper threshold boundary of the bDVH curve may include the rightmost point of the family of dose volume curves, and the lower threshold boundary of the bDVH curve may include the leftmost point of the family of dose volume curves. Additional details and variations of methods for calculating bounded DVH curves are described in U.S. Patent Application No. 16 / 016,272, filed Jun. 22, 2018, which is hereby incorporated by reference in its entirety.

[0138] Figures 3A and 3B depict a flowchart representation of one variation of a treatment planning method that uses separate multi - target registration for aligning a plurality of patient target regions during a treatment session. The method (350) can include obtaining one or more treatment planning images of one or more patient target regions (352), defining the contours of one or more patient target regions and / or OARs and corresponding dose constraints (354), selecting planned registration reference points for each of the one or more patient target regions (358), and calculating a set of shift - invariant beam filters based on the planned registration reference points for each patient target region (360). The method (350) can include generating a fluence map for each patient target region based on the beam filters and the one or more treatment planning images (362), and combining all of the patient target region fluence maps into a cumulative treatment planning fluence map (364). Each patient target region can have a separate planned registration reference point if the patient target regions are physically separated from each other (e.g., do not overlap). This allows each patient target region to be moved independently. The corresponding patient target region fluence maps for each patient target region can be added together (since dose delivery is linear) to calculate the total dose. For example, if the locations of the first and second planned registration reference points change from the planning session to the treatment session, the cumulative treatment planning fluence map can be recalculated using the new registration reference point locations. Separate treatment planning fluence maps for a plurality of patient target regions allow the plurality of planned registration reference points to be moved independently of each other, thereby allowing the plurality (e.g., two or more) of fluence maps to be shifted relative to each other to obtain a more optimized dose coverage. Additionally or alternatively, each respective patient target region fluence map for each patient target region may be individually constrained.

[0139] Next, method (350) may include repeatedly modifying the cumulative treatment plan fluence map (366) based on one or more dose constraints, such as (a) the high fluence regions remain separated from each other, (b) the OAR constraints are satisfied, (c) the initial dose constraint / target is satisfied, among one or more of them.

[0140] In some variations, each of the constraints on the cumulative treatment plan fluence map may be weighted by a linear coefficient that defines or approximates their relative importance. In some variations, the fluence map for the target region may be described as variable x. The sum of the combined fluences may be described as x cumulative as described. The dose calculation matrix for the patient may be defined as A. The dose for a particular target region may be defined as Ax. Further, the dose for the entire patient may be defined as Ax cumulative as described. For example, the dose constraint may include one or more cost functions, and optionally, each cost function may be weighted by an individual scaling coefficient. The prescription dose requirement or constraint (C) may include one or more cost functions, such as a cost function C(x) for the radiation fluence (x) and / or a cost function C(Ax) for the dose in the treatment area, and / or a cost function C(Ax cumulative ), and / or a cost function C(x cumulative ) for the total fluence, among one or more of them. Each of these may optionally be weighted by an individual scaling coefficient (w i , w k , w m , w n) may be weighted. For example, a cost function for the cumulative fluence map can be used to optimize the treatment time in the context of co-delivery. For example, a cost function for the dose to each treatment area can be optimized to ensure that the minimum dose is delivered. This ensures that the fluence from different treatment areas does not irradiate most of the dose to a given target. This significantly reduces the coupling between treatment areas, and thus increases the robustness of the treatment plan to relative displacements between treatment areas. For example, D cumulative A cost function for can be used to limit the combined dose average to the combined heart from all treatment areas. [Number]

[0141] Additional details regarding cost functions and other constraints can be found in U.S. Patent Application No. 62 / 966,997, filed on January 28, 2020, which is hereby incorporated by reference in its entirety.

[0142] In addition or alternatively, in order to ensure that the dose does not exceed a predetermined threshold in a given zone (e.g., region), in addition to a predetermined set of dose constraints, one or more adjustment constraints (e.g., artificial constraints in addition to tissue constraints, adjustment of structures) may be applied. In some variations, one or more adjustment constraints may enable separation of radiation delivery between two or more target regions.

[0143] After the cumulative treatment plan fluence map is modified, method (350) may include defining a periphery including regions of low fluence values around each patient target region in the cumulative treatment plan fluence map (e.g., adjustment constraints, cost function) (368). Separating the cumulative treatment plan fluence map along the periphery into individual fluence maps for each patient target region (370), and for each patient target region, calculating a second set of shift-invariant beam filters based on the planned alignment reference points and fluence maps for that patient target region (372). The second set of shift-invariant beam filters, the planned alignment reference points for each patient target region and / or OAR, and the individual treatment plan fluence maps for each patient target region may be transferred to the radiation therapy system for alignment during treatment (along with other treatment plan parameters).

[0144] In some variations, such as in the case of SBRT / IMRT delivery, the treatment planning method may include calculating the delivery fluence for the patient target region and / or treatment area (e.g., based on a prescription dose, dose constraints, and / or dose calculation matrix), and the delivery fluence is linked (e.g., fixed) to the planned alignment reference points. The calculated delivery fluence may then be segmented into machine instructions for delivery (by either the treatment planning system and / or the radiation therapy system). During the treatment session, the calculated delivery fluence may be updated based on the updated location of the alignment reference points (e.g., selected by the user from alignment images acquired during the treatment session). In this variation, it may not be necessary for the beam filters to be calculated by the treatment planning system.

[0145] Optionally, some treatment planning methods (350) may include designating (356) each of the patient target regions as an SBRT / IMRT target region or a BgRT target region. If a patient target region is designated as a BgRT target region, the method (350) may include calculating an ROI for that patient target region. If the patient target regions are physically separated from each other to enable independent movement of each patient target region, each patient target region has a separate planned setup reference point, whereby it may be possible to shift the fluence map relative to the others for optimized dose coverage. For each BgRT patient target region, the method (350) may also include calculating a shift-invariant emission filter as described above.

[0146] Optionally, the method (350) may include displaying (374) a visualization graphic of the cumulative treatment plan fluence map on a display device. Optionally, some treatment planning methods (350) may include calculating (376) a bDVH curve for each patient target region and / or OAR, and displaying (378) the bDVH curve and / or dose calculation data for each patient target region and / or OAR on a display device. The bDVH curve for a patient target region may be calculated as described above.

[0147] FIG. 4 depicts a flowchart representation of a variant of a treatment planning method that uses BgRT-based multi-target alignment for the alignment of multiple patient target regions during a treatment session. Method (450) may include obtaining one or more treatment plan images of one or more patient target regions (452), defining the contours and corresponding dose constraints of one or more patient target regions and / or OARs (454), designating one of the multiple patient target regions as the alignment target region (456), and defining the contours of spatial masks or filters (e.g., ROIs, beamlet zones) for each of the remaining patient target regions (i.e., those not the alignment target region) to have sizes and shapes corresponding to the movement of the alignment target region and the movement of the remaining patient target regions (458). Method (450) may include calculating a set of shift-invariant beam filters based on the planned alignment reference points selected for each patient target region and / or the ROI of the alignment target region (462), generating a treatment plan fluence map based on the beam filters and one or more treatment plan images (466), and calculating the dose to the alignment target region and the remaining patient target regions of all the patient target regions (468). Method (450) may also include calculating shift-invariant beam filters such as beam filters for each ROI of each patient target region, which may be calculated as described above, and the treatment plan fluence map may be generated based on the calculated beam filters together with one or more treatment plan images (466). The shift-invariant beam filters for virtual alignment of ROIs and / or shift-invariant beam filters for BgRT delivery (e.g., RFM), the planned alignment reference points for each patient target region and / or OAR, and / or any treatment plan fluence map and / or dose calculation may be transferred to a radiation therapy system for alignment and radiation delivery (along with other treatment planning parameters).

[0148] Optionally, some treatment planning methods (450) may include defining (460) the outer shape of a registration target region's spatial mask or filter (e.g., the injection zone, ROI) to be smaller than that of the spatial mask or filter of the remaining patient target region. For example, the ROI of the registration target region may be smaller than the ROI of the remaining patient target region. In some variations, the patient target region selected to be the registration target region may be the patient target region closest to a critical structure such as a highly radiosensitive organ, and / or a PET-avid critical structure, and / or a bone structure. In some variations, the patient target region selected to be the registration target region may be the target region closest to a planned structure (e.g., calculated or user-selected) such as a potential dose area joining between two patient target regions. In some variations, the patient target region selected to be the registration target region may be the region most sensitive to uncertainty, such as a region with a high dose gradient, a critical dose measurement goal or constraint, inconsistent PET activity (e.g., regions of hypoxia or perfusion), or a variable motion trajectory near the edge of the ROI.

[0149] Optionally, some methods (450) may include determining (464) a set of treatment radiation source injection positions and calculating the projection of the injection filter onto each of the injection positions. For example, a radiation therapy system having a rotatable gantry may have 100 circumferentially distributed injection positions or angles around the bore that can advance the patient table. The method (450) may include calculating the projection of each injection filter for each of the 100 injection positions. Optionally, the method (450) may include calculating (470) a bDVH curve for each patient target region and / or OAR and displaying (472) the bDVH curve and / or dose calculation data for each patient target region and / or OAR on a display device. The bDVH curve for the patient target region can be calculated as described above.

[0150] As shown previously, any of the treatment planning methods for multi-target irradiation may optionally include calculating bDVH curves for each patient target region and / or OAR, and displaying DVH and other dose data to the clinician for review and / or approval. In some variations, such dose data may be used as feedback (i.e., as additional constraints) to further refine the treatment planning fluence map and / or the injection filter, which may be useful for improving dose measurement landmarks and compliance with the prescription. FIGS. 5A-5C depict a flowchart representation of a variation of a treatment planning method for modifying a planned fluence map and / or injection filter based on a user-selected modification of the bDVH confidence interval and / or a shift in the dose distribution for a patient target region.

[0151] bDVH Confidence Interval Editing bDVH confidence interval editing can provide a mechanism that may reduce or eliminate spatial uncertainty, and the bDVH calculation can typically be calculated at the 95% confidence interval, and the patient is represented within these boundaries on the treatment day. Reducing the width of the boundaries in the bDVH results in a potential minimum dose D pmin , and a potential maximum dose D n closer to the nominal dose D pmax . Methods for reducing the difference from the nominal dose may include reducing the confidence interval. For example, an 80% confidence interval can result in a narrower bDVH curve where D pmax and D pmin can be closer to D n . The trade-off is that, for example, instead of expecting 1 in 20 patients to fail the bDVH safety check during treatment, 1 in 5 patients may fail the bDVH safety check and treatment with a non-deterministic treatment method (i.e., BgRT) may be rejected. Similarly, if the bDVH of the target region responds as being significantly compressed during the initial 95% CI calculation, the bDVH can be recalculated at a higher confidence interval (e.g., 99% CI), in which case the bDVH widens and D pmin and D pmax are Dn Further separated from (i.e., widening / spreading the bDVH curve). Using this functionality, clinicians may be able to reduce or eliminate hot spots and cold spots within existing boundaries.

[0152] Figure 5A depicts a treatment planning method including bDVH confidence interval editing. The method (550) may include formulating a treatment plan using one or more specified uncertainty vectors (e.g., biological activity, setup position error, etc.) (551), and selecting a confidence interval (CI) of the uncertainty vector that may be determined based on the type of uncertainty (552). The method (550) may include optimizing the treatment plan fluence map with uncertainty (553), calculating a bDVH that describes the likelihood of treatment daily doses conforming within the bDVH boundary using the CI, and evaluating the bDVH to determine whether the boundary is acceptable (554). For example, the bDVH may be output to a display device and viewed by a user. If the boundary is considered acceptable, the method (550) may include calculating a final dose for the treatment plan (556) and approving the boundary for the treatment plan (557). The method (550) may include determining a method for adjusting the boundary (558) if the boundary is not acceptable. In some variations, the method (550) may include widening the boundary to increase acceptable uncertainty and setting a new CI that is larger than the original CI (559). In some variations, the method (550) may include compressing the boundary to decrease acceptable uncertainty and setting a new CI that is less than the original CI (560). After the CI is adjusted, the method (550) may include optimizing the treatment plan with uncertainty (561), calculating a new bDVH that describes the likelihood of treatment daily doses conforming within the bDVH boundary using the new CI, and then evaluating the new bDVH to determine whether the boundary is acceptable (554). The evaluation and calculation of the bDVH (554 - 561) may be repeated until a bDVH with an acceptable boundary is calculated.

[0153] In some variations, the confidence interval editing may include determining whether it is necessary to re-optimize the treatment plan fluence map after the CI has been adjusted. FIG. 5B depicts a treatment planning method that includes a determination regarding whether to re-optimize after bDVH confidence interval editing. Method (570) may include formulating a treatment plan using one or more specified uncertainty vectors (e.g., biological activity, setup position error, etc.) (571), and selecting a confidence interval (CI) of the uncertainty vector that can be determined based on the type of uncertainty (572). Method (570) may include optimizing the treatment plan fluence map with uncertainty (573), calculating a bDVH that describes the likelihood of the treatment daily dose conforming within the bDVH boundaries using the CI, and evaluating the bDVH and determining whether the boundaries are acceptable (574). For example, the bDVH may be output to a display device and viewed by a user. If the boundaries are considered acceptable, method (570) may include calculating the final dose for the treatment plan (576) and approving the boundaries for the treatment plan (577). Method (570) may include determining a method for adjusting the boundaries (578) if the boundaries are not acceptable. In some variations, method (570) may include widening the boundaries to increase the acceptable uncertainty and setting a new CI that is larger than the original CI (579). In some variations, method (570) may include compressing the boundaries to decrease the acceptable uncertainty and setting a new CI that is less than the original CI (580). After the CI has been adjusted, method (570) may include determining whether a new optimization is necessary. If so, method (570) may include optimizing the treatment plan with uncertainty (581), calculating a new bDVH that describes the likelihood of the treatment daily dose conforming within the bDVH boundaries using the new CI, and then evaluating the new bDVH and determining whether the boundaries are acceptable (574). If it is determined that no optimization is necessary, method (570) may proceed directly from the CI adjustment to evaluating the new bDVH (574).The evaluation and calculation of the bDVH (574 - 581) can be repeated until a bDVH with acceptable boundaries is calculated. In some variations, a visualization graphic may optionally be included to represent the effect on the bDVH due to a change in the CI.

[0154] Some treatment planning methods may include displaying a treatment planning dose distribution (e.g., calculated from a treatment planning fluence map) overlaid on a patient anatomical image, including each of the patient target regions to be treated in a treatment session, receiving clinician input related to a dose shift, and recalculating the treatment planning fluence map and / or the injection filter based on the dose shift input by the clinician. In some variations, the dose shift may be a fixed shift of the dose by about 0.1 mm to about 1 mm in any direction. These small dose shifts can help avoid unnecessary fluence delivery to the spine and / or reduce hot spots caused by two interacting fluence maps between two in-plane target regions. For example, when fluence map sub-regions are separately registered in a mosaic multi-target registration, and / or when individual fluence maps of different patient target regions are separately registered in a separate multi-target registration, the cumulative delivered fluence map (i.e., the sum of all delivered fluence map sub-regions and / or the sum of all individual fluence maps for each patient target region) may have regions of unwanted fluence, such as high fluence levels in an OAR or healthy tissue region. These regions of unwanted fluence may not have been present in the planned fluence map, but when changes (e.g., shifts) to the fluence map are performed during registration, unwanted fluence map artifacts can appear in the delivered fluence map. The radiation therapy system can generate a visualization graphic that reflects these fluence map changes (and / or dose changes) with the patient's anatomical image overlaid so that the user can determine whether any of the delivered fluence maps can be shifted to reduce any unwanted dose or fluence. Alternatively or in addition, in some variations, the dose shift may be performed after fluence map optimization in the treatment planning, which can help facilitate the generation of a treatment plan and / or a planned fluence map that meets the dose measurement target.After the dose shift, an additional fluence map optimization step may be optional (i.e., re-optimization may not be necessary). The updated fluence map and / or the injection filter may be transmitted to the radiotherapy system for positioning and treatment.

[0155] FIG. 5C depicts a flowchart of a variant of a treatment planning method (590), the method comprising calculating (591) a treatment planning fluence map (and / or an injection filter such as any of those described herein) for all of a patient's target regions, calculating (592) a dose distribution map based on the treatment planning fluence map, generating (593) a composite image including the dose distribution map overlaid on a patient anatomical image including one or more of the patient's target regions, receiving (594) user (e.g., clinician, dosimetrist) input regarding shifting the dose on the dose distribution map, and calculating (595) an updated treatment planning fluence map based on the dose shift input by the user. Optimization of the fluence map after the dose shift may be optional.

[0156] In some treatment planning methods, dose dispersion (aka, delta dose) may be used as a constraint for optimizing the fluence map of a treatment plan. A minimum or maximum acceptable dose dispersion may be selected as a constraint. For example, setting a maximum dose dispersion as a treatment planning fluence map optimization constraint may help reduce the width of the bDVH while maintaining the default confidence interval (e.g., CI = 95%). FIG. 6 depicts a treatment planning method that includes setting dose dispersion as a constraint for generating a treatment planning fluence map. Method (670) may include formulating (671) a treatment plan using one or more specified uncertainty vectors (e.g., biological activity, setup position error, etc.), and selecting (672) a maximum acceptable dose dispersion D of the uncertainty vector, which may be determined based on the type of uncertainty. Method (670) may include the maximum acceptable dose dispersion D plan to be selected. Method (670) may include the maximum acceptable dose dispersion D planOptimizing a treatment plan fluence map (673), calculating a bDVH that describes the likelihood of treatment day dose conformity within the bDVH boundaries using CI, evaluating the dose dispersion D actual resulting as a user outcome, and determining whether D actual is acceptable (674), and may include. Method (670) is D actual If it is considered acceptable, calculating the final dose for the treatment plan (676) and approving the dose dispersion for the treatment plan (677), and may include. Method (670) is D actual If not acceptable, may include determining a method for adjusting the dose dispersion (678). In some variations, method (670) increases the acceptable dose dispersion and sets a new dose dispersion D plan greater than the initial D plan ' (559). In some variations, method (670) decreases the acceptable dose dispersion and sets a new dose dispersion D plan greater than the initial D plan ' (580). Method (670) is D plan After being adjusted, optimizing the treatment plan with the updated dose dispersion D plan ' constraint, selecting geometries that may be somewhat affected by the dose dispersion based on the D plan ' constraint value, and then evaluating the new D actual and determining whether the dose dispersion is acceptable (674), and may include. The evaluation and calculation of the dose dispersion (674 - 681) can be repeated until an acceptable dose dispersion is calculated.

[0157] Multi - target positioning verification method A method for virtual position verification shifts a planned fluence map for a patient target region to reflect the current / real-time location of that patient target region during treatment. During treatment planning, a set of shift-invariant beam filters can be calculated based on the planned position verification reference points and desired dose measurement targets, and during the treatment session, the shift-invariant beam filters can be used to calculate a delivery fluence that results in the delivery of dose to the treatment region of that patient. A method for virtual position verification can include selecting a position verification reference point corresponding to the planned position verification reference point. For example, if the planned position verification reference point is the center of the patient target region, during position verification, the position verification reference point should also be selected as the center of the patient target region so as to be reflected in the position verification image. Then, a position verification function (such as a delta function, Gaussian function, truncated Gaussian function, etc.) based on the selected position verification reference point can be calculated. For example, the position verification function can be a delta function that can be an impulse function centered on the position verification reference point, or the position verification function can be a Gaussian function having a width (σ) selected during treatment planning, with the mean value (μ) centered on the position verification reference point. The delivery fluence for any beam position of the treatment radiation source (which can be predetermined by the treatment plan) can be calculated by convolving the projection of the delta function at that beam position with the projection of the shift-invariant beam filter at that beam. Then, the delivery fluence for any beam position of that patient target region can be segmented into radiotherapy system machine instructions for radiation delivery.

[0158] FIG. 7A depicts a flowchart representation of one variation of a method for virtual positioning verification for SBRT / IMRT delivery. This method can be used for virtual positioning verification of a single patient target area or multiple patient target areas, and can be used at the start of a treatment session or multiple times throughout a treatment session. In some variations, the treatment plan may include a plurality of treatment areas, each having one or more patient target areas. The patient can be physically set up / position verified once for a treatment area, and each of the patient target areas within the treatment area can be virtually position verified. In some variations where a group of patient target areas is defined during the treatment plan, each group shares a single positioning reference point and can be position verified together. The method (700) can include obtaining (702) one or more positioning images of one or more patient target areas, selecting (704) a positioning reference point within the obtained images for each of the patient target areas, where the positioning reference point corresponds to a planned positioning reference point selected during the treatment plan, calculating a positioning function (e.g., delta, Gaussian, truncated Gaussian, etc.) from the positioning reference points for each patient target area and calculating the projection of the positioning function onto each treatment radiation source, calculating (708) the fluence for delivery to each patient target area for each injection position by convolving the corresponding positioning function projection with a shift-invariant emission filter derived based on the planned positioning reference point. As described above, in some variations, the fluence for delivery for each injection position of the treatment radiation source can be calculated by convolving the projection of the positioning function at each injection position with the emission filter. Optionally, the method (700) can include proceeding with radiation delivery (i.e., using the treatment radiation source to emit the delivery fluence to each patient target area) (710). The shift-invariant emission filter can be calculated during the treatment plan based on (e.g., centered on) the planned positioning reference point. In some variations, the positioning function can be a delta function, and the shift-invariant emission filter calculated during the treatment plan can represent a fluence map (e.g., a planned fluence map) centered on the planned positioning reference point.In this example, calculating the fluence for delivery (708) includes convolving the calculated delta function (centered on the selected registration fiducial point) with a shift-invariant projection filter, thereby resulting in a delivery fluence map that is a shift of the planned fluence map. In some variations, the radiation therapy system may segment the delivery fluence map into machine instructions in real time, i.e., minutes or seconds before radiation delivery. The registered patient target region may be included in a single registration image or in multiple registration images acquired at the start of the treatment session and / or throughout the session. For example, a first patient target region may be identified in a first image and a second patient target region may be identified in a second image. These images may be separate images acquired at separate times or at the same time, or may be sub-regions of a single image. The planned registration fiducial point may be the treatment isocenter, but in some variations, the planned registration fiducial point may not be the treatment isocenter and may be any point selected by the user. The registration fiducial point may be specified by coordinates. The registration function used to register the patient target region may be an impulse function or a peak or pixel centered on the selected registration fiducial point, as described above. The method (700) may be used with any of the treatment planning methods of FIGS. 1-4, for example, with the methods of FIGS. 1A, 1B.

[0159] FIG. 7B depicts a flowchart representation of a variant of a method for virtual position alignment for delivery of BgRT to one or more patient target regions. Virtual position alignment of the BgRT patient target region involves adjusting the location ROI based on the offset between a planned alignment reference point and an alignment reference point selected during the treatment session. The method (720) can include obtaining (722) one or more alignment images of one or more patient target regions, selecting (724) an alignment reference point within the obtained image for each of the patient target regions, where the alignment reference point corresponds to a planned alignment reference point selected during treatment planning, calculating (726) a spatial offset based on the shift between the selected alignment reference and the planned alignment reference positioning, and shifting (728) the boundaries of the planned region of interest (ROI) based on the calculated spatial offset. In some variants, shifting (728) the boundaries of the planned region of interest includes applying rotation and shift to the planned region of interest by a roll correction factor φ representing the rotational translation of the alignment reference point relative to the planned alignment reference point. The method (720) can further include obtaining imaging data and spatially masking (e.g., filtering) the imaging data using the shifted region of interest (step 730), calculating the fluence for delivery to each patient target region at each emission position of the treatment radiation source by convolving the projection of the obtained imaging data with a corresponding shift-invariant emission filter (step 732), and advancing the radiation delivery (i.e., emitting the delivery fluence to each patient target region using the treatment radiation source) (step 734). The obtained imaging data (730) can include PET imaging data (e.g., positron annihilation emission paths, LOR), X-ray imaging data, MRI imaging data, ultrasonic imaging data, optical imaging data (e.g., from a camera), etc.Calculating the delivery fluence for a specific injection position (732) can include projecting the acquired spatially masked imaging data onto that injection position and convolving the imaging data projection with an injection filter for that injection position and the target region. The injection filter can be one of the injection filters calculated during treatment planning (e.g., using the method of FIG. 1B). In a variation where a roll correction factor is applied to the planned region of interest, calculating the fluence for delivery (732) may include convolving a set of injection filters with the roll correction factor φ, and in some variations, may include shifting the above-described ROI (e.g., as described and depicted in FIG. 34). The method (720) can be used with any of the treatment planning methods of FIGS. 1-4, e.g., with the methods of FIGS. 1A, 1B.

[0160] FIG. 8A depicts a flowchart representation of a variation of a method for mosaic multi-target position alignment. This method can be used at the start of a treatment session (e.g., to align / register all of the patient target regions) or multiple times throughout the treatment session (e.g., to sequentially align / register the patient target regions). The method (800) can include acquiring (802) one or more alignment images of one or more patient target regions and dividing (804) the treatment plan fluence map into one or more fluence map sub-regions according to the periphery around each patient target region defined during treatment planning, and shifting (806) each of the fluence map sub-regions such that the high fluence regions co-localize with the corresponding patient target regions of this high fluence region. Shifting (806) the fluence map sub-regions can be performed using the virtual alignment method described herein (e.g., selecting an alignment reference point for each of the patient target regions, calculating a delta function based on the alignment reference point, and convolving the delta function with a shift-invariant injection filter). Alternatively or in addition, shifting (806) the fluence map sub-regions can be performed by shifting radiation therapy system machine commands (e.g., MLC, gantry, linac, jaws, etc.).

[0161] Optionally, some variations of method (800) include calculating the dose for each of the patient target regions based on the shifted fluence map sub-regions, comparing the calculated dose with the boundary dose volume histogram curve calculated during the treatment plan, and / or displaying on a display device the boundary dose volume histogram curves and / or dose calculation data for each patient target region and / or OAR (808). Optionally, method (800) may include proceeding with radiation delivery (i.e., using a treatment radiation source to emit a delivery fluence to each patient target region) (810). In some variations, the radiation therapy system may segment the delivery fluence map into machine instructions in real time, i.e., several minutes or seconds before radiation delivery. The patient target regions to be position-matched may be included in a single position-matching image or, as described above, in a plurality of position-matching images acquired at the start of the treatment session and / or throughout the session. Method (800) may be used in conjunction with any of the treatment planning methods of FIGS. 1-4, for example, with the method of FIG. 2.

[0162] FIG. 8B depicts a flowchart representation of a variant of a method for mosaic multi-target registration, including shifting or adjusting a patient table. In some methods, patient setup may include adjusting the orientation of the patient table (e.g., roll, pitch, yaw, xyz translation, etc.) based on registration images and treatment plan images. This method can be used at the start of a treatment session (e.g., to align / register all of a patient's target regions) or multiple times throughout a treatment session (e.g., to sequentially align / register a patient's target regions). The method (820) can include obtaining one or more registration images of one or more patient target regions (822), dividing a treatment plan fluence map into one or more fluence map sub-regions according to the periphery around each patient target region defined during treatment planning (824), calculating a patient table position shift vector for each fluence map sub-region based on the registration images and treatment plan images (826), shifting the fluence map sub-regions such that high fluence regions co-localize with corresponding patient target regions for each patient table position (828), and calculating the fluence to each patient target region at each patient table position shift vector (830). Shifting the fluence map sub-regions (828) can be performed using the virtual registration method described herein (e.g., selecting a registration reference point for each of the patient target regions, calculating a delta function based on the registration reference point, and convolving the delta function with a shift-invariant projection filter). Alternatively or in addition, shifting the fluence map sub-regions (828) may be performed by shifting radiation therapy system machine commands (e.g., MLC, gantry, linac, jaw, etc.).

[0163] Optionally, some variations of method (820) may include calculating a dose for each of the patient target regions based on the shifted fluence map sub-regions, comparing the calculated dose to the bounded dose volume histogram curve calculated during the treatment plan, and / or displaying on a display device the bounded dose volume histogram curves and / or dose calculation data for each patient target region and / or OAR (832). Optionally, method (820) may include proceeding with radiation delivery (i.e., emitting a delivery fluence to each patient target region using a treatment radiation source) (834). In some variations, the radiation therapy system may segment the delivery fluence map into machine instructions in real-time, i.e., minutes or seconds before radiation delivery. The patient target regions to be positionally matched may be included in a single positional match image or, as described above, in a plurality of positional match images acquired at the start of the treatment session and / or throughout the session. Method (820) may be used with any of the treatment planning methods of FIGS. 1-4, for example, with the method of FIG. 2.

[0164] FIG. 9 depicts a flowchart representation of one variation of a method for separated multi-target alignment. Method (900) includes obtaining (902) one or more alignment images of one or more patient target regions, where each patient target region has a separate individual treatment plan that includes a set of treatment plan parameters including a set of shift-invariant emission filters, and selecting, for each patient target region, an alignment reference point and using a virtual alignment method (e.g., calculating a delta function based on the alignment reference point and convolving the delta function with the shift-invariant emission filters to obtain a delivery fluence map for each patient target region), calculating (904) a delivery fluence map for each target region, calculating a cumulative fluence map by combining the delivery fluence maps for all patient target regions, and calculating a dose for each of the patient target regions (906). Method (900) may include evaluating (908) whether dose targets for all patient target regions are met and / or whether high-fluence areas of the individual delivery fluence maps are sufficiently separated from each other (i.e., do not co-localize with other high-fluence areas). Method (900) may include proceeding with radiation delivery (912), which may include segmenting (914) one or more of the following, i.e., (a) the individual fluence maps for each patient target region and / or (b) the cumulative fluence map, into radiation therapy system instructions if dose targets for all patient target regions are met and / or the high-fluence areas of the individual delivery fluence maps are sufficiently separated from each other. Method (900) may include generating (916) a visual and / or audible notification if dose targets for all patient target regions are not met and / or the high-fluence areas of the individual delivery fluence maps are not sufficiently separated from each other. Optionally, dose calculation data and / or DVH curves may be presented to the user such that the user may evaluate the next course of action. In some variations, the radiation therapy system may segment the delivery fluence map into machine instructions in real time, i.e., minutes or seconds before radiation delivery.The patient target region to be registered can be included in a single registration image or, as described above, in a plurality of registration images acquired at the start of the treatment session and / or throughout the session. The method (900) can be used with any of the treatment planning methods of FIGS. 1-4, for example, with the method of FIGS. 3A-3B.

[0165] FIG. 10 depicts a flowchart representation of a variant of a method for isolated multi-target alignment. The method (1000) can include obtaining (1002) one or more alignment images of one or more patient target regions, positioning (1004) a patient target region designated as an alignment target region during treatment planning using a treatment isocenter (e.g., the central portion of a treatment radiation beam), and determining (1006) whether the remaining patient target regions are located within the boundaries of a spatial emission filter (e.g., an emission zone, ROI) designated during treatment planning. If the remaining patient target regions are located within the boundaries of a spatial mask or filter (e.g., within an ROI), the method (1000) can include proceeding with radiation delivery (1002). If the remaining patient target regions are not located within the boundaries of a spatial mask or filter (e.g., not within an ROI), the method (1000) can include generating (1010) a visual and / or audible notification. Optionally, the user may be presented with location data of the patient target regions related to the associated ROI of the patient target regions so that the user can optionally evaluate the next course of action. Optionally, one or more ROIs of these misaligned target regions may be aligned using the virtual alignment method described herein. Positioning (1004) the alignment target region can use one or more setup and / or alignment methods. For example, positioning (1004) the alignment target region can include one or more of the following: (a) shifting radiation therapy system machine commands (e.g., MLC, gantry, linac), (b) selecting an alignment reference point for the ROI of the alignment target region, calculating a delta function based on the alignment reference point, and convolving a function (e.g., delta, Gaussian, truncated Gaussian) with the ROI, and / or (c) calculating a patient table position shift vector and adjusting the patient table accordingly. In some variants, radiation delivery may include segmenting one or more of the following: (a) an individual fluence map for each patient target region, and / or (b) a cumulative fluence map, into radiation therapy system commands.In some variations, the radiation therapy system may segment the delivery fluence map into machine instructions in real time, i.e., minutes or seconds before radiation delivery. The patient target region to be aligned may be included in a single alignment image or, as described above, may be included in a plurality of alignment images acquired at the start of the treatment session and / or throughout the session. The method (1000) may be used with any of the treatment planning methods of FIGS. 1-4, for example, with the method of FIG. 4. For example, each of the plurality of patient target regions may have a separate set of planned alignment reference points and emission filters. During the treatment session, the location of the alignment reference points for each of the plurality of patient target regions may be updated based on the acquired alignment images (e.g., the alignment reference points may be selected by the user). Separate and / or independent virtual alignments of each patient target region based on the updated alignment reference points effectively shift the delivery fluence map to the real-time locations of the plurality of patient target regions. In such a manner, the fluence maps for each of the plurality of patient target regions may move relative to each other. For example, in a treatment session involving two or more patient target regions, the radiation delivery method may include performing one or more (e.g., two or more) virtual alignments and shifting the planned fluence maps for each of the two or more patient target regions.

[0166] FIG. 11A is a flowchart illustrating a method (100). The method (100) can be used to align a patient for radiation therapy. As shown in FIG. 11A, the method (100) includes obtaining (102) images of a first patient treatment area and a second patient treatment area. In some embodiments, the first patient treatment area and the second patient treatment area can include one or more tumor regions (e.g., patient target regions). For example, in some embodiments, the first patient treatment area can include a first portion of a tumor, and the second patient treatment area can include a second portion of the tumor. In some embodiments, the first patient treatment area and the second patient treatment area can each include one or more discrete tumors located in different regions of the patient's body. The acquired images can be, for example, images acquired via one or more of PET, CT, MRI, ultrasound, optical imaging modalities, combinations thereof (e.g., surface tracking / mapping, optical CT for predicting internal locations based on outer surface data), and / or any other suitable method. In some embodiments, for example, one or more images can be obtained using a CT system before or at the start of a treatment session. In some embodiments, the CT system can be attached to the same gantry as the radiation source intended for delivery of radiation therapy during the treatment session. In some embodiments, the CT system can be attached to a gantry separate from the gantry supporting the radiation source. In some embodiments where radiation is delivered at discrete patient couch locations or positions (i.e., beam stations), the location of the first patient treatment area can correspond to the location of the first beam station of the treatment radiation source, and the location of the second patient treatment area can correspond to the location of the second beam station. In some variations, the first and second treatment areas can overlap, while in other variations, the first and second treatment areas do not overlap.

[0167] Based on the acquired image of the first patient target area and the treatment plan image (also referred to herein as the "first treatment plan image"), a first set of patient position shift vectors can be calculated (104). The location of the first treatment area and / or the patient target area in the acquired image can be compared with the location of the first treatment area and / or the patient target area in the treatment plan image in 2D and / or 3D. For example, in some embodiments, the patient target area can be divided into sub-areas (e.g., sub-volumes) that can be represented by voxels, and the coordinates of each voxel in the acquired image can be compared with the coordinates of the corresponding voxel of the patient target area in the treatment plan image. Each of the vectors of the first set of patient position shift vectors can represent the difference in distance and direction between the coordinates of the voxels in the acquired image and the coordinates of the corresponding voxels in the treatment plan image. In some embodiments, the difference in direction can include a tilt angle. In some embodiments, the first set of patient position shift vectors can be calculated by moving the acquired image relative to the treatment plan image so as to align or register the first patient target area of the treatment plan image with the first patient target area of the acquired image. For example, each voxel of the acquired image can be translated along and / or centered on the X, Y, and / or Z axes, and the relative positions of the voxels of the acquired image with respect to each other can be maintained during the movement of the acquired image until the acquired image and the treatment plan image are substantially aligned with respect to the first patient target area (e.g., are aligned within a predetermined acceptable tolerance range or margin such that there is an acceptable area or ratio of overlap between the first patient target areas in the images). The first set of patient position shift vectors can then be calculated based on the location difference and / or position difference between the acquired images before and after being moved to enhance alignment with the treatment plan image. The first set of patient position shift vectors can reflect the distance and / or direction by which each voxel of the acquired image is to be translated for acceptable alignment between the first patient target area of the acquired image and the first patient target area of the treatment plan image.Furthermore, the first set of patient position shift vectors may include, or correspond to, instructions related to the first patient position and / or the first orientation of the patient (e.g., the first position and / or the first orientation of the surface on which the patient is disposed) such that the first patient target region of the patient is substantially positioned at the location where the first patient target region was when the treatment planning image was acquired. This may facilitate the delivery of therapeutic radiation to a patient target region closer to the treatment plan. Additionally, the first set of patient position shift vectors may include information regarding any tilt, pitch, yaw, and / or roll corrections that need to be implemented via the positioning of the patient table and / or via adjusting the roll of the gantry to which the radiation source (e.g., the therapeutic radiation source) is coupled (e.g., to correct the gantry emission position) such that the location of the first patient target region approximates the location of the first patient target region in the first treatment planning image.

[0168] In some embodiments, the acquired images of the first treatment area and / or the patient target area, and the treatment planning images of the first patient target area can each be two-dimensional images. By comparing the acquired two-dimensional images with the two-dimensional treatment planning images, a first set of patient position shift vectors can be calculated. In some embodiments, the plurality of two-dimensional images of the first treatment area and / or the patient target area can be acquired along different orientations or planes (e.g., three or more orientations or planes), and the images can be compared with the corresponding treatment planning images of the first patient target area acquired along the same orientation or plane. Using the changes in the position of the first patient target area along each image plane in each orientation, a first set of patient position shift vectors can be calculated. For example, the acquired images can include the acquired axial image, the acquired sagittal image, and the acquired coronal image of the first patient target area of the patient. When the patient is lying supine on a surface such as a patient table or treatment couch and is oriented such that the patient first encounters the treatment radiation source head, the acquired axial image can be taken along the axial plane of the patient (e.g., the plane that divides the body into upper and lower parts), the acquired sagittal image can be taken along the sagittal plane of the patient (e.g., the plane that divides the patient into right and left parts), and the acquired coronal image can be taken along the coronal plane of the patient (e.g., the plane that divides the patient into ventral and dorsal parts). The axial plane is arranged perpendicular to the sagittal plane, and the coronal plane is arranged perpendicular to both the axial plane and the sagittal plane. The treatment planning images can include the treatment planning axial image, the treatment planning sagittal image, and the treatment planning coronal image of the first patient target area of the patient. Thus, the acquired axial image corresponds to the treatment planning axial image, the acquired sagittal image corresponds to the treatment planning sagittal image, and the acquired coronal image corresponds to the treatment planning coronal image.

[0169] The amount by which the patient position and / or orientation should be adjusted along or about each of the X, Y, and Z axes (e.g., via movement of the patient surface and / or rotation of the radiation source) for treatment of a first patient target region can be reflected by a first set of patient position shift vectors based on the difference between the treatment planning image and each of the acquired images taken within each of the axial, sagittal, and coronal planes of the patient. In some embodiments, the X axis of the patient surface can be parallel to the intersection of the sagittal and coronal planes of the patient and can be disposed in the sagittal plane. The Y axis of the patient surface can be parallel to the intersection of the axial and coronal planes of the patient and can be disposed in the axial plane. The Z axis of the patient surface can be parallel to the intersection of the sagittal and axial planes of the patient and can be disposed in the sagittal or axial plane of the patient. To determine the first set of patient position shift vectors, each of the acquired images of the first patient target region can be compared to the respective treatment planning image of the first patient target region taken in the same plane to determine distance correction and / or rotational correction (e.g., via movement of the patient surface and / or rotation of the radiation source) in the same plane. One or more vectors of the first set of patient position shift vectors can represent the magnitude and direction by which the patient surface would be moved (e.g., shifted and / or rotated) such that the first patient target region approximates the location of the first patient target region in the treatment planning image. One or more vectors of the first set of patient position shift vectors can also represent the magnitude and direction by which the radiation source could be moved (e.g., rotated on the gantry) such that the location of the first patient target region approximates the location of the first patient target region in the treatment planning image. In some embodiments, the first set of patient position shift vectors can reflect up to six corrections (i.e., amounts of adjustment along or about each of the X, Y, and Z axes of the patient table). In some embodiments, a DICOM Spatial Registration Object (SRO) can be used to determine each of the six corrections (i.e., amounts of adjustment along or about each of the X, Y, and Z axes of the patient table) associated with the first set of patient position shift vectors.

[0170] In some embodiments, the first set of patient position shift vectors can be calculated by moving (e.g., shifting and / or rotating) each of the acquired images to align or register the first patient target area and / or patient target region of the treatment plan image with the first patient target region of the acquired image relative to each respective treatment plan image of the acquired images. Each of the acquired images can be shifted a specific distance and / or rotated a specific amount to improve the alignment between the first patient target region of the treatment plan image and the first patient target region of the acquired image. For example, an image acquired along the axial plane during treatment can be compared with a treatment plan image acquired along the patient's axial plane to determine the amount of correction needed along the horizontal axis (i.e., the Y-axis) and the vertical axis (i.e., the Z-axis) of the patient surface (e.g., the table or couch) within the axial plane, and the amount of correction centered around the required gantry roll axis (e.g., centered around the X-axis). An image acquired along the sagittal plane during treatment can be compared with a treatment plan image acquired along the patient's sagittal plane to determine the amount of correction needed along the longitudinal axis (i.e., the X-axis) and the vertical axis, and the amount of pitch correction (e.g., rotation centered around the Y-axis) of the patient surface (e.g., the table or couch) within the sagittal plane. An image acquired along the coronal plane during treatment can be compared with a treatment plan image acquired along the patient's coronal plane to determine the amount of correction needed along the horizontal axis and the longitudinal axis, and the amount of yaw correction (i.e., centered around the Z-axis) of the patient surface (e.g., the table or couch) within the coronal plane.

[0171] The first set of patient position shift vectors can improve the positional alignment between the first patient target region and the location of the first patient target region at the time of acquisition of the treatment plan image by reflecting the amount of correction of the patient surface required along or around each of the X, Y, and Z axes. Thus, in some embodiments, the first set of patient position shift vectors includes at least one patient position shift vector that reflects an adjustment along the X-axis of the patient surface based on a longitudinal difference determined from a comparison of the acquired sagittal image and the treatment plan sagittal image and / or a comparison of the acquired coronal image and the treatment plan coronal image. The first set of patient position shift vectors can include at least one patient position shift vector that reflects an adjustment along the Y-axis of the patient surface based on a lateral difference determined from a comparison of the acquired axial image and the treatment plan axial image and / or a comparison of the acquired coronal image and the treatment plan coronal image. The first set of patient position shift vectors can include at least one patient position shift vector that reflects an adjustment along the Z-axis of the patient surface based on a vertical difference determined from a comparison of the acquired axial image and the treatment plan axial image and / or a comparison of the acquired sagittal image and the treatment plan sagittal image. In some embodiments, the adjustment along the Z-axis of the patient surface included in the first set of patient position shift vectors can be an average of a lateral difference determined from a comparison of the acquired axial image and the treatment plan axial image and a lateral difference determined from a comparison of the acquired sagittal image and the treatment plan sagittal image.

[0172] Regarding rotation of the patient surface about the axis of the patient surface and / or rotation of the radiation source about the axis of the radiation source, the first set of patient position shift vectors includes rotation correction of the patient surface about the Y-axis of the patient surface based on comparison of the acquired sagittal image and the treatment plan sagittal image, and / or rotation correction of the patient surface about the Z-axis of the patient surface based on comparison of the acquired coronal image and the treatment plan coronal image, and / or rotation correction of the radiation source about the X-axis of the radiation source (e.g., the X-axis of the gantry having the same extent as the X-axis of the patient surface) based on comparison of the acquired axial image and the treatment plan axial image. For example, shifting or panning the acquired sagittal or coronal image to match the treatment plan sagittal or coronal image can result in a set of patient position shift vectors representing shifts in the X, Y, and Z axes. Tilting the acquired sagittal or coronal image to match the treatment plan sagittal or coronal image can result in a set of patient position shift vectors representing pitch and / or roll position corrections. Thus, the amounts by which the patient position and / or orientation should be adjusted along or about each of the X, Y, and Z axes can be reflected by the first set of patient position shift vectors based on the differences between the treatment plan image and each of the acquired images taken within each of the axial, sagittal, and coronal planes of the patient.

[0173] Based on the acquired image of the second treatment area and / or patient target area, the treatment planning image (also referred to herein as the "second treatment planning image"), and the first set of patient position shift vectors, a second set of patient position shift vectors can be calculated (106). In some embodiments, the second set of position shift vectors can include translational distances and / or directions. In some embodiments, the translational direction can include a tilt angle. In some embodiments, similar to that described above with respect to the first set of patient position shift vectors, the second set of patient position shift vectors can be calculated by moving the acquired image relative to the treatment planning image so as to positionally align or register the second patient target area of the treatment planning image with the second patient target area of the acquired image. For example, each voxel of the acquired image can be translated along and / or centered about the X, Y, and / or Z axes, maintaining the relative positions of each voxel of the acquired image with respect to each other during the movement of the acquired image until the acquired image and the treatment planning image improve the positional alignment with respect to the second patient target image (e.g., the overlap between the second patient target area in the image increases or is optimized). The second set of patient position shift vectors can then be calculated based on the location and / or position differences between the acquired images before and after being moved to enhance the positional alignment with the treatment planning image. The second set of patient position shift vectors can reflect the distances and / or directions by which each voxel of the acquired image had to be transformed to improve the positional alignment between the second patient target area of the acquired image and the second patient target area of the treatment planning image. In some embodiments, the second set of position shift vectors can be calculated based on the first set of patient position shift vectors such that the distance and / or direction information included in the second set of position shift vectors is relative to the first set of patient position shift vectors rather than the initial coordinates of the voxels of the acquired image.Thus, the position and / or orientation instructions based on the second set of patient position shift vectors may include, or may correspond to, instructions enumerating the modifications to the patient's position and / or orientation relative to the patient's first position and / or first orientation based on the first set of patient position shift vectors such that the patient assumes a second position and / or orientation for irradiation of the second treatment area and / or the patient target area.

[0174] The patient may be aligned according to the second set of patient position shift vectors such that the patient is transitioned from a first location to a second location. For example, the treatment radiation source may be stopped so that the first patient target area is not irradiated by the treatment radiation source, and then the patient may be aligned according to the second set of patient position shift vectors. Aligning the patient according to the second set of patient position shift vectors may include moving the radiotherapy patient table on which the patient is disposed relative to the treatment radiation source and / or adjusting the rotational position of the radiation source according to the second set of patient position shift vectors. The treatment radiation source may then be activated so that the second patient target area is irradiated. In some embodiments, for beam station-based delivery, the patient may be aligned according to the first set of patient position shift vectors at a first location associated with a first beam station during the period of irradiation, and the patient may be aligned according to the second set of patient position shift vectors during the period of transition of the patient table between a first location associated with the first beam station and a second location associated with a second beam station (e.g., a second beam station adjacent to the first beam station).

[0175] In some embodiments, the method can include defining not only a first patient target region and a second patient target region, but also any suitable number of target regions, obtaining an image of each of the target regions, and aligning the patient based on the position shift vectors as described above based on the first patient target region and the second patient target region. For example, for beam station-based delivery, the method can include defining a plurality of patient target regions, each of which can span a plurality of beam stations. In some variations, there can be the same number of patient target regions as beam stations. An image can be obtained from each of the patient target regions, and a set of position shift vectors associated with each defined patient target region can be generated as described above. Each patient target region and the associated set of position shift vectors can be associated with a particular beam station. As the patient and / or patient table transition to the locations associated with each respective beam station, the patient can be aligned based on the set of position shift vectors associated with that beam station. In some embodiments, the defined target regions (and the associated sets of patient position shift vectors) can be associated with a plurality of beam stations such that the patient is aligned based on the set of position shift vectors and does not need to be realigned as the patient table passes through a series of two or more beam stations. Thus, the number of defined target regions (and each respective set of associated patient position shift vectors) can be less than or equal to the number of beam stations.

[0176] In addition, the second set of patient position shift vectors can include information regarding any tilt, pitch, yaw, and roll corrections that need to be implemented via movement of the patient table and / or adjustment of the roll of the gantry to which the radiation source is coupled (e.g., correcting the gantry emission position) such that the second patient target region improves its positional alignment with the location of the second patient target region in the second treatment plan image.

[0177] In some embodiments, similar to that described above with respect to the first set of patient position shift vectors, a plurality of two-dimensional images of the second treatment area and / or the patient target area can be acquired along different orientations or planes (e.g., three or more orientations or planes), and the images can be compared to corresponding treatment planning images of the second patient target area acquired along the same orientation or plane. In some embodiments, the plurality of two-dimensional images of the second patient target area can be the same two-dimensional images acquired from the first patient target area to calculate the first set of patient position shift vectors. Using the changes in the position of the second patient target area along each image plane in each orientation, a second set of patient position shift vectors can be calculated. For example, the acquired images can include the acquired axial image, the acquired sagittal image, and the acquired coronal image of the second patient target area of the patient. When the patient is lying supine on a surface such as a patient table or treatment couch and is oriented such that the patient first encounters the treatment radiation source head, the acquired axial image can be taken along the axial plane of the patient (e.g., the plane that divides the body into an upper part and a lower part), the acquired sagittal image can be taken along the sagittal plane of the patient (e.g., the plane that divides the patient into a right part and a left part), and the acquired coronal image can be taken along the coronal plane of the patient (e.g., the plane that divides the patient into a ventral part and a dorsal part). The axial plane is arranged perpendicular to the sagittal plane, and the coronal plane is arranged perpendicular to both the axial plane and the sagittal plane. The treatment planning images can include the treatment planning axial image, the treatment planning sagittal image, and the treatment planning coronal image of the second patient target area of the patient. Thus, the acquired axial image corresponds to the treatment planning axial image, the acquired sagittal image corresponds to the treatment planning sagittal image, and the acquired coronal image corresponds to the treatment planning coronal image.

[0178] The amount by which the patient's position and / or orientation should be adjusted along and / or about each of the X, Y, and Z axes (e.g., via movement of the patient surface and / or rotation of the radiation source) for treatment of a second patient target region can be reflected by a second set of patient position shift vectors based on the difference between the treatment planning image of the second patient target region and each of the acquired images of the second patient target region taken within each of the axial, sagittal, and coronal planes o...

Claims

1. A system for virtual target area registration and radiation delivery, the system comprising: a therapeutic radiation source; one or more processors; and wherein the one or more processors are configured to: receive an image of a patient in a treatment position and identify a patient target area in the received image; select a reference point for registration of a fluence map within the received image, the fluence map including radiation beamlet intensities across a patient area, the fluence map representing a radiation fluence across the patient target area, the reference point corresponding to a planned registration reference point; determine a function for registration of the fluence map, and calculate a delivery fluence of therapeutic radiation to the patient target area at each emission position of the therapeutic radiation source by applying the function to a shift-invariant representation of the fluence map derived based on the planned registration reference point, the function encoding the position of the patient target area as a Gaussian function centered on the reference point, and the shift-invariant representation of the fluence map being a matrix specifying a transformation from the received image to the fluence map; wherein the therapeutic radiation source is configured to emit the delivery fluence to the patient target area, and the emission position is the position at which the delivery fluence is emitted from the therapeutic radiation source.

2. The system of claim 1, wherein applying the function to the shift-invariant representation of the fluence map includes convolving the function with the shift-invariant representation of the fluence map.

3. The system of claim 2, wherein the Gaussian function is a truncated Gaussian function.

4. The system of claim 2, wherein applying the function includes shifting the shift-invariant representation of the fluence map by an interpolation algorithm, the interpolation algorithm being one of a linear shift, a bicubic shift, a spline shift, or a Fourier shift.

5. The system of claim 1, wherein the received image includes one or more of a positron emission tomography (PET) image, one or more X-ray projection images, a computed tomography (CT) image, or a magnetic resonance imaging (MRI) image.

6. ​ ​ The patient target area is a first patient target area, the one or more processors are configured to identify a second patient target area in the received image, select a second reference point for registering the position of the fluence map to the second patient target area within the received image, wherein the second reference point corresponds to a second planned registration reference point, determine a second function for registering the position of the fluence map, and calculate a second delivered fluence at each emission position of the therapeutic radiation source by applying the second function to a second shift-invariant representation of the treatment plan based on the second planned registration reference point, wherein the second function encodes the position of the second patient target area as a Gaussian function centered on the second reference point, and the second shift-invariant representation of the fluence map is a matrix specifying the transformation from the received image to the fluence map, and are configured to perform the above, and the therapeutic radiation source is further configured to emit the second delivered fluence to the second patient target area. The system according to claim 1. **Claim 7** The patient target area is a first patient target area, the one or more processors are configured to identify a second patient target area in the received image, select a second reference point for registering the position of the fluence map to the second patient target area within the received image, wherein the second reference point corresponds to a second planned registration reference point, determine a second Gaussian function centered on the second reference point, and calculate a second delivered fluence at each emission position of the therapeutic radiation source by convolving the second Gaussian function with a second shift-invariant representation of the treatment plan based on the second planned registration reference point, wherein the second shift-invariant representation of the fluence map is a matrix specifying the transformation from the received image to the fluence map, and are configured to perform the above, and the therapeutic radiation source is further configured to emit the second delivered fluence to the second patient target area. The system according to claim 2. **Claim 8** The therapeutic radiation source is configured to simultaneously emit the first delivery fluence to the first patient target region and emit the second delivery fluence to the second patient target region, the system according to any one of claims 6 or 7.

9. The therapeutic radiation source is configured to sequentially emit the first delivery fluence to the first patient target region and emit the second delivery fluence to the second patient target region, the system according to any one of claims 6 or 7.

10. The function is a first function (δ) for position matching of the fluence map, and the first shift-invariant representation of the fluence map is a first set of planned fluence maps (p １ , p ２ , ···, p ｉ ) calculated during treatment planning for each injection position (i) of the therapeutic radiation source, and the one or more processors are further configured to calculate a first set of projections (δ ｉ ) of the first function for position matching of the fluence map to each injection position (i). δ ｉ = proj ｉ (δ), and Each projection (δ ｉ ) is a 2D fluence distribution, Calculating the first delivery fluence involves convolving each projection in the projection of the first set of the first function (δ ｉ ), with the corresponding planned fluence map (p ｉ ) to calculate a first fluence map (f ｉ ) for delivering therapeutic radiation to each injection position (i) of the therapeutic radiation source, f ｉ = p ｉ * δ ｉ where The system further comprises a motion system configured to move the therapeutic radiation source to each injection position (i), and the therapeutic radiation source is further configured to emit radiation to the first patient target region according to the first fluence map (f ｉ ) of claim 7, which is further configured to emit radiation to the first patient target region according to the first fluence map (f

11. Each projection (δ ｉ ) is an m×n matrix, where m is the number of multi-leaf collimator leaves and n is the number selected during treatment planning, the system of claim 10.

12. n is the number of beam stations selected during the treatment plan, the system according to claim 11.

13. The Gaussian function is a truncated Gaussian function, the system according to claim 10.

14. Applying the function includes shifting the shift-invariant representation of the fluence map by an interpolation algorithm, and the interpolation algorithm is one of a linear shift, a bicubic shift, a spline shift, and a Fourier shift, the system according to claim 10.

15. The second shift-invariant representation of the fluence map includes a second set of planned fluence maps (p_2 １ , p_2 ２ , ···, p_2 ｉ ) calculated during treatment planning for each injection position (i), and the one or more processors are further configured to calculate a second set of projections (δ_2 ｉ ) of the second function for position registration of the fluence map to each injection position (i). δ_2 ｉ = proj ｉ (δ_2), and Each projection (δ_2 ｉ ) is a 2D fluence distribution, Calculating the second delivery fluence includes convolving each projection (δ_2 ｉ ), during the projection of the second set of the second function for position matching of the fluence map, with the corresponding planned fluence map (p_2 ｉ ) to calculate a second fluence map (f_2 ｉ） ) for delivering therapeutic radiation to each injection position (i). f_2 ｉ = p_2 ｉ * δ_2 ｉ where The system further comprises a motion system, the motion system being further configured to move the therapeutic radiation source to each injection position (i), and the therapeutic radiation source is for delivering therapeutic radiation according to the second fluence map (f_2 ｉ ) and is further configured to emit radiation to the second patient target region, the system according to claim 6.

16. Each of said projections (δ_2 ｉ ) is an m×n matrix, the system according to claim 15.

17. m is the number of multi-leaf collimator leaves, and n is a number selected during the treatment plan, the system according to claim 16.

18. n is the number of beam stations selected during the treatment plan, the system according to claim 17.

19. The Gaussian function is a truncated Gaussian function, the system according to claim 15.

20. Applying the second function includes shifting the shift-invariant representation of the fluence map by an interpolation algorithm, and the interpolation algorithm is one of a linear shift, a bicubic shift, a spline shift, and a Fourier shift, the system according to claim 15.

21. The one or more processors are further configured to convert the fluence map (f ｉ , f_2 ｉ ) into a plurality of instructions for a radiation therapy system for each injection position, the system according to any one of claims 10 to 15.

22. The one or more processors are further configured to convert the calculated delivery fluence into a plurality of instructions for a radiation therapy system at each injection position, the system according to any one of claims 1 or 2.

23. The system according to any one of claims 6 or 7, wherein the one or more processors are further configured to convert the second calculated delivery fluence into a plurality of instructions for a radiation therapy system for each injection position.

24. The plurality of instructions for the radiation therapy system include one or more configurations for a multi-leaf collimator for each injection position, and emitting a radiation fluence further includes moving a plurality of leaves of the multi-leaf collimator according to the configuration of the multi-leaf collimator corresponding to the location of the injection position of the therapeutic radiation source, and emitting a pulse of radiation. The system according to any one of claims 21 to 23.

25. The plurality of instructions for the radiation therapy system further include therapeutic radiation source pulse parameters for each injection position, and emitting the pulse of radiation includes emitting radiation having the therapeutic radiation source pulse parameters corresponding to the location of the injection position of the therapeutic radiation source. The system according to claim 24.

26. The system according to claim 1, wherein the reference point is a location selected by a user within the received image.

27. The system according to claim 1, wherein the reference point corresponds to a treatment planning isocenter defined for the patient target region during treatment planning.

28. The system according to claim 6, wherein the first patient target region is in a first treatment area of the patient defined during treatment planning, and the second patient target region is in a second treatment area of the patient defined during treatment planning.

29. The first treatment area has an axial length of 8 cm or less and includes a first set of axial planes, the second treatment area has an axial length of 8 cm or less and does not overlap with the first treatment area, and the second treatment area includes a second set of axial planes. The system according to claim 28.

30. The centers of the first treatment area and the second treatment area are on the same straight line along the IEC-Y axis and / or in the same plane as the IEC-Y axis. The system according to claim 29.

31. The system according to claim 28, wherein the first treatment area and the second treatment area overlap.

32. The system according to claim 1, further comprising a patient table, wherein the emission position of the therapeutic radiation source includes the location of the therapeutic radiation source relative to the location of the patient table.

33. The system according to claim 32, further comprising a gantry rotatable about a vertical axis, wherein the therapeutic radiation source is mounted on the gantry, and the location of the therapeutic radiation source is specified by a gantry angle about the vertical axis.

34. The system according to claim 33, wherein the patient table is movable to a plurality of different locations along the vertical axis.

35. The system according to claim 34, wherein the boundary of the planned region of interest surrounds the patient target region, and the boundary of the planned region of interest includes a spatial filter.

36. The one or more processors are configured to shift the boundary of the planned region of interest by applying rotation and shift to the planned region of interest by means of a roll correction coefficient φ representing the rotational translation of the reference point relative to the planned registration reference point, and the one or more processors are configured to calculate the delivery fluence by circularly convolving the shift-invariant representation of the treatment plan with the roll correction coefficient φ. The system according to claim 35.

37. said fluence map (f ｉ , f_2 ｉ ) for delivering therapeutic radiation is calculated by convolving each projection of said set of projections with a delta function and a delta function δ ＬＯＣ and δ_2 ＬＯＣ that are located away from the isocenter by angular shifts (φ, φ_2) at locations δ ｉ , δ_2 ｉ and further includes convolving with said planned fluence maps p', p_2' 【Number 4】 The system according to claim 21.

Images