Multi-sensor induction-type radiotherapy

Abstract

To provide a multi-sensor induction-type radiotherapy.SOLUTION: Methods for radiotherapy planning and delivery using sensor data from one or more target sensors is disclosed in the present specification. One modification of the radiotherapy planning method includes: generating a sensor characterization image based on a sensor characterization probability density function (PDF) of the target sensor; and calculating a pair of launch filters that can be applied to the sensor image generated from sensor data acquired during a radiation delivery session. In addition, a modification of the radiotherapy planning method includes: generating a plurality of sensor characterization images based on a plurality of sensor characterization PDFs for the plurality of target sensors; and calculating a plurality of pairs of launch filters for the plurality of target sensors.SELECTED DRAWING: Figure 2A

Description

[Technical Field]

[0001] (CROSS-REFERENCE TO RELATED APPLICATIONS) This application claims priority to U.S. Provisional Application No. 63 / 062,792, filed August 7, 2020, the disclosure of which is incorporated herein by reference in its entirety. [Background technology]

[0002] Radiation therapy is a non-invasive treatment that involves applying high levels of radiation to a tumor or lesion. Such high levels of radiation can slow or otherwise stop the growth of cancer cells. Some radiation treatment systems have a therapeutic radiation source that is movable around the patient so that radiation is directed at the tumor from various positions and / or angles around the patient. The patient is usually positioned on a platform that may or may not move in coordination with the therapeutic radiation source to irradiate the tumor.

[0003] Applying high levels of radiation to a tumor while limiting radiation exposure to surrounding healthy tissue can be difficult. This difficulty is compounded by the fact that radiation is invisible and tumors are typically located within a patient. Treatment plans are generated based on planning images of the patient and tumor, but treatment sessions typically occur days or weeks later, after which time the location of the tumor within the patient may have changed. Some radiation therapy systems have imaging systems that can be used to acquire images of the patient and / or tumor, thereby allowing the location of the patient and / or tumor to be determined at the start of a treatment session. Information about the location of the patient, tumor, and other patient structures can be compared to their locations in the planning images. If there is any discrepancy, the patient is moved (e.g., by moving the platform and / or adjusting the patient's posture and / or position on the platform) so that the location of the patient (and / or tumor) matches the location in the planning images. This helps ensure that the therapeutic radiation source irradiates the tumor according to the treatment plan.

[0004] However, patients may move during a treatment session, and such movement can cause the tumor to move and change its location from its original location at the start of the treatment session. In some cases, movement may be due to unavoidable physiological processes, such as breathing and digestion, and / or unexpected patient movement (e.g., the patient shifting position due to discomfort or coughing). Images acquired at the start of a treatment session cannot capture changes in tumor location after the treatment session begins (i.e., when the therapeutic radiation source begins emitting radiation to the tumor). Some radiation treatment planning methods can account for tumor motion due to predictable, cyclical physiological motion (e.g., breathing) by defining a tumor motion envelope based on planning images. This results in a treatment plan that delivers high levels of radiation to the area encompassed by the motion envelope, ensuring that ablative levels of radiation are sufficiently delivered to the tumor. However, this method may also result in irradiation of healthy tissue that happens to be located within the motion envelope.

[0005] Therefore, improved methods of radiation treatment planning and delivery are desirable to deliver tumoricidal levels of radiation to patient target areas while limiting radiation exposure to healthy tissue. Summary of the Invention

[0006] Disclosed herein are methods for radiation treatment planning and delivery using sensor data from one or more target area sensors. In some variations, the one or more target area sensors may be one or more patient sensors. The radiation treatment planning method may include generating a sensor characterization image based on a sensor characterization probability density function (PDF) of the target sensor and calculating a set of emission filters that can be applied to sensor data acquired during a radiation delivery session (e.g., a non-treatment quality assurance or QA session, or alternatively, a treatment session). The sensor characterization PDF of the target sensor may represent the noise characteristics and / or variability and / or error profile of the target sensor. The emission filters may be shift-invariant. The target sensor may include one or more position sensors, image sensors, etc. The radiation treatment delivery method described herein may include obtaining one or more sensor data readings from the target sensor, calculating a radiation fluence map for delivery by convolving an image generated from the sensor data readings with a treatment planning emission filter, and delivering radiation according to the calculated radiation fluence map. The image and / or position sensor data may be acquired frequently during a radiation delivery session (e.g., a non-treatment QA session in the absence of a patient, or alternatively, a treatment session in the presence of a patient), e.g., within a few seconds or milliseconds before a radiation beam is emitted toward a target region. In some variations, the target region may be a patient target region to which therapeutic radiation is delivered during a treatment session. The frequent acquisition of image and / or position sensor data may be used by the radiation therapy system to adjust a radiation fluence map in real time to account for patient and / or tumor motion to help direct radiation delivery to the actual location of the tumor.

[0007] In some variations, radiation treatment planning may use sensor data from multiple target area sensors to calculate multiple sets of corresponding emission filters. For example, a radiation treatment planning method may include generating a first sensor characterization image based on a first sensor characterization probability density function (PDF) of a first target sensor, generating a second sensor characterization image based on a second sensor characterization PDF of a second target sensor, calculating a first set of emission filters that can be applied to sensor data from the first target sensor acquired during a radiation delivery session (e.g., a QA or treatment session), and calculating a second set of emission filters that can be applied to sensor data from the second target sensor acquired during a radiation delivery session (e.g., a QA or treatment session). The first target sensor and the second target sensor may be the same sensor type and / or different sensor types. During a radiation delivery session, sensor data may be acquired from both the first target sensor and the second target sensor and used to generate sensor data images, which are convolved with their respective emission filters and combined to generate a radiation fluence map for delivery. In some variations, the radiation therapy delivery method may include obtaining first sensor data readings from a first target sensor, obtaining second sensor data readings from a second target sensor, calculating a radiation fluence map for delivery by summing a convolution of a first image generated from the first sensor data readings with a first emission filter and a convolution of a second image generated from the second sensor data readings with a second emission filter, and delivering radiation according to the calculated radiation fluence map. Calculating the radiation fluence for delivery using real-time acquired data from two or more target sensors may help provide a precise indication of the location of the target region and / or help facilitate accurate delivery of therapeutic radiation to the target region.

[0008] Also disclosed herein are methods for radiation therapy planning that use patient imaging data and position data to calculate an emission filter, which can be applied to images acquired during a radiation delivery session (e.g., a QA or treatment session) to calculate a radiation fluence map for delivery. In some variations, the radiation therapy planning method can include generating a sensor characterization image based on a sensor characterization PDF including a plurality of position values ​​representing the location of a center of gravity of the target region over time, and calculating an emission filter based on the sensor characterization image. The sensor characterization image can include one or more motion retention histograms of the target region. The center of gravity position values ​​can be determined from any suitable imaging modality, for example, 4D CT imaging data. A radiation therapy delivery method can include acquiring imaging data from an image sensor, generating an image from the acquired imaging data, and calculating a radiation fluence map for delivery by convolving the generated image with an emission filter. The method can then include delivering radiation to the target region according to the calculated radiation fluence map. Calculating an emission filter based on a sensor characterization PDF containing multiple position values ​​representing the location of the centroid of the target region over time can result in a treatment plan that delivers the prescribed tumoricidal dose to the target region while reducing irradiation to surrounding healthy tissue.

[0009] One variation of a method for radiation delivery includes obtaining sensor data readings from a target sensor, generating a sensor image from the sensor data readings, calculating a radiation fluence map for delivery to the target region by convolving the sensor image with a shift-invariant emission filter derived from a sensor characterization probability density function (PDF) of the target sensor, and delivering radiation to the target region according to the calculated radiation fluence map. The sensor characterization PDF may be a sensor error characterization PDF representing a sensor data error rate and / or a sensor data variability rate. For example, the sensor characterization PDF may include one or more of a 1D plot of sensor data, a 2D plot of sensor data, and / or a 3D plot of sensor data, and a histogram representing sensor data variability. In some variations, the target sensor may be a position sensor. For example, the position sensor may include a target region position sensor and / or an X-ray projector system configured to track embedded fiducials. The sensor image may be a delta function, a Gaussian function, and / or a truncated Gaussian function centered at a location corresponding to the position sensor data reading. Some methods may involve attaching optical fiducials to the patient's skin and tracking the optical fiducials using an optical imaging system.

[0010] In some variations, the target sensor may be a null position sensor whose sensor data readings are constant position values ​​representing the center of gravity of the target region, and the sensor characterization PDF of the null position sensor includes multiple position values ​​representing the location of the center of gravity of the target region over time. The sensor image may be a delta function, a Gaussian function, and / or a truncated Gaussian function centered at a location corresponding to the position sensor data reading. The sensor characterization PDF may include a motion retention histogram of the target region. The multiple position values ​​may be determined using, for example, 4D CT imaging data.

[0011] In some variations, the shift-invariant emission filter may correspond to a launch position of the therapeutic radiation source, and calculating the fluence for delivery may include calculating the fluence for delivery at the launch position by convolving a projection of the sensor image on the launch position with the shift-invariant emission filter of the launch position. The target sensor may include one or more image sensors, the sensor data readings may include imaging data, and the sensor characterization PDF of the target sensor may include an image generated from the imaging data. The one or more image sensors may include an image sensor selected from the group consisting of a PET sensor, an MRI sensor, and a CT sensor.

[0012] In some variations, the target sensor may be a first target sensor, the sensor data readings may be first sensor data readings, the image may be a first sensor data image, the shift-invariant emission filter may be a first shift-invariant emission filter, and the sensor characterization PDF may be a first sensor characterization PDF, and the method may further include obtaining second sensor data readings from a second target sensor and generating a second sensor data image from the second sensor data readings, and calculating the fluence map for delivery may include summing (a) a convolution of the first sensor data image with the first shift-invariant emission filter and (b) a convolution of the second sensor data image with a second shift-invariant emission filter derived from the second sensor characterization PDF of the second target sensor. The first target sensor data readings may include a first type of data, and the second target sensor data readings may include a second type of data different from the first type of data. The second sensor characterization PDF may be, for example, a sensor error characterization PDF. The second target sensor may be a position sensor. In some variations, the first target sensor data reading may include positron annihilation emission path data, and the second target sensor data reading may include target region location data. The first target sensor data reading may include partial imaging data, and the second target sensor data reading may include target region location data. The first target sensor data reading may include at least one of 3D PET imaging data, 2D X-ray imaging data, projection imaging data, fluoroscopic imaging data, CT imaging data, and MR imaging data, and the second target sensor data reading may include target region location data. In some variations, the shift-invariant emission filter may correspond to a launch position of the therapeutic radiation source, and calculating the fluence for delivery may include calculating the fluence for delivery at the launch position by projecting the sensor data readings onto the launch position, generating a second sensor image of the projected sensor data readings, and convolving the second sensor image with the shift-invariant emission filter for the launch position.

[0013] Variations of radiation therapy systems are also described herein. One variation of the radiation therapy system may include a patient platform, a therapeutic radiation source movable to one or more firing positions around the patient platform, a target sensor system including a target sensor that acquires sensor data, and a controller in communication with the therapeutic radiation source and the target sensor system, the controller configured to calculate a radiation fluence map for delivery to the target region by convolving an image generated from the sensor data with a shift-invariant emission filter derived from a sensor characterization PDF of the target sensor, and the controller configured to deliver radiation according to the calculated radiation fluence map. The target sensor may be a first target sensor, and the target sensor system may include a second target sensor. In some variations, at least one of the first target sensor and the second target sensor may be a position sensor configured to be coupled to a patient disposed on the patient platform. For example, the position sensor may be configured to be coupled to the target region. In some variations, the position sensor may include an optical imaging system configured to track optical fiducials attached to the patient's skin, and the target sensor system may further include an optical camera configured to detect the position of the optical fiducials. The sensor characterization PDF may be a sensor error characterization PDF representing a sensor data error rate or may represent a sensor data variability rate. For example, the sensor characterization PDF may include one or more of a 1D plot of the sensor data, a 2D plot of the sensor data, and / or a 3D plot of the sensor data, and a histogram representing the sensor data variability. The image generated from the position sensor data may be a delta function, a Gaussian function, and / or a truncated Gaussian function centered at a location corresponding to the position sensor data. Alternatively or additionally, the target sensor may include one or more image sensors, the sensor data may include imaging data, and the sensor characterization PDF of the target sensor may include an image. The one or more image sensors may include an image sensor selected from the group consisting of a PET sensor, an MRI sensor, and a CT sensor.The controller may be configured to receive a first sensor data reading from a first target sensor and a second sensor data reading from a second target sensor, the shift-invariant emission filter being a first shift-invariant emission filter of the first target sensor, and the sensor characterization PDF being a first sensor characterization PDF of the first target sensor, and the controller may be further configured to calculate a fluence map for delivery by summing (a) a convolution of the first sensor data image with the first shift-invariant emission filter and (b) a convolution of a second image generated from the second sensor data with a second shift-invariant emission filter derived from the second sensor characterization PDF of the second target sensor. The first target sensor may be a first type sensor, and the second target sensor may be a second type sensor different from the first type. The second target sensor may be a position sensor. In some variations, the first target sensor may be a positron annihilation emission path sensor, and the second target sensor may be a target area position sensor. Alternatively or additionally, the first target sensor may be an image sensor and the second target sensor may be a position sensor. The first target sensor may include at least one of a 3D PET sensor, a 2D X-ray sensor, a projection image sensor, a fluoroscopy image sensor, a CT image sensor, and an MR sensor, and the second target sensor may include a position sensor.

[0014] A method for treatment planning using target sensor data is described herein. The method for sensor-based treatment planning generates a sensor characterization image N based on the sensor characterization PDF of the target sensor for each of the i firing locations. i and the condition

[0015]

number

[0016] The methods for generating treatment plans described herein are performed in the absence of a patient. These treatment planning methods alone do not involve the delivery of therapeutic radiation to a patient.

[0017] The target sensor may be a position sensor, and the sensor characterization PDF may include one or more of a 1D plot of the position sensor data, a 2D plot of the position sensor data, a 3D plot of the position sensor data, and a histogram representing the position sensor data variability, and the sensor characterization image N i may include one or more of position sensor error data and a position sensor error histogram. The position sensor data may include coordinates in space. In some variations, the shift-invariant emission filter p i is calculated to minimize a cost function C(D,F) derived from clinician-defined constraints and objectives for the prescribed dose D and radiation fluence map F. i In some variations, the shift-invariant emission filter p i is calculated to minimize a cost function C(D,F) derived from clinician-defined constraints and objectives for the prescribed dose D and radiation fluence map F. i The method may further include iterating over values ​​for .times. ...

[0018] In some variations, the target sensor may be a null position sensor having a constant position value representing the centroid of the target area, and the sensor characterization PDF may comprise multiple position values ​​representing the location of the centroid of the target area over time, and the sensor characterization image N i Generating the sensor characterization image N may include generating a motion retention histogram of the target region. The plurality of position values ​​may be determined using 4D CT imaging data. In some treatment planning methods, the sensor characterization image N iGenerating the sensor characterization PDF may include generating an inverse motion dwell histogram of the target region. In some variations, the target sensor may be an image sensor, and the sensor characterization PDF may include multiple image sensor data, and the sensor characterization PDF may include multiple image sensor data. i Generating the sensor characterization image N may include combining a plurality of image sensor data. The plurality of image sensor data may include at least one of 3D PET imaging data, 2D X-ray imaging data, projection imaging data, fluoroscopy imaging data, CT imaging data, and MR imaging data. The target sensor may be a first target sensor, and generating the sensor characterization image N i may be a first set of sensor characterization images, and the shift-invariant emission filter p i may be a first set of shift-invariant firing filters, and the method may generate a second set of sensor characterization images M based on the sensor characterization PDF of the second target sensor at each of the i firing locations. i and generating p i and q i By iterating over the values ​​of i and a second set of sensor characterization images M i , the first set of shift-invariant emission filters p for each of the i emission positions is calculated based on i and a second set of shift-invariant emission filters q i whereby the following equation is satisfied:

[0019]

number

[0020] In some variations, the shift-invariant emission filter p i and q i is calculated to minimize a cost function C(D,F) derived from clinician-defined constraints and objectives for the prescribed dose D and radiation fluence map F. i and q iThe first target sensor may be a first position sensor, and the second target sensor may be a second position sensor. In some variations, the sensor characterization PDFs for the first position sensor and the second position sensor may each include one or more of a 1D plot of the position sensor data, a 2D plot of the position sensor data, and / or a 3D plot of the position sensor data, and a histogram representing the position sensor data variability, and the sensor characterization image N i and M i may include a motion retention histogram of the target region. The position sensor data of the first position sensor and the second position sensor may include coordinates in space. Alternatively, the first target sensor may be a first image sensor and the second target sensor may be a second image sensor. The sensor characterization PDFs of the first image sensor and the second image sensor may each include a plurality of image sensor data, and the sensor characterization image N i and M i may include combining a plurality of image sensor data from the first image sensor and the second image sensor, respectively. The plurality of image sensor data from the first image sensor and the second image sensor may The imaging data may include at least one of PET imaging data, 2D X-ray imaging data, projection imaging data, fluoroscopic imaging data, CT imaging data, and MR imaging data.

[0021] In some variations, the first target sensor may be a position sensor and the second target sensor may be an image sensor. The sensor characterization PDF of the position sensor may include one or more of a 1D plot of the position sensor data, a 2D plot of the position sensor data, and / or a 3D plot of the position sensor data, and a histogram representing the position sensor data variability, and the sensor characterization PDF of the image sensor may include a plurality of image sensor data. For example, a sensor characterization image N i may be the motion retention histogram of the target region, and the sensor characterization image M imay be a combination of multiple image sensor data, where the position sensor data may include coordinates in space, and the multiple image sensor data may include at least one of 3D PET imaging data, 2D X-ray imaging data, projection imaging data, fluoroscopy imaging data, CT imaging data, and MR imaging data. [Brief explanation of the drawings]

[0022] [Figure 1A] 1 illustrates a flowchart representation of a radiation therapy system. [Figure 1B] 1 illustrates one variation of a radiation therapy system. [Figure 1C] 1 illustrates one variation of a radiation therapy system. [Figure 1D] 1 illustrates a variation of a radiation therapy system. [Figure 2A] 1 illustrates a flowchart representation of a treatment planning method using data from sensors. [Figure 2B] 1 illustrates a sensor image of a position sensor data reading. [Figure 2C] 1 illustrates a sensor error image of a position sensor. [Figure 2D] 1 illustrates a sensor error image of a position sensor. [Figure 2E] 1 illustrates a sensor error image of a position sensor. [Figure 2F] 1 illustrates a sensor error image of a position sensor. [Figure 2G] 1 illustrates a sensor error image of a position sensor. [Figure 2H] 1 illustrates a sensor error image of a position sensor. [Figure 2I] 1 illustrates a sensor error image of a position sensor. [Figure 2J] 1 illustrates a sensor error image of a position sensor. [Figure 2K] 1 illustrates a sensor error image of a position sensor. [Figure 2L] 1 illustrates a sensor characterization image of a sensor error characterization probability distribution function of a position sensor. [Figure 3A]1 illustrates variations in the sensor image. [Figure 3B] 1 illustrates variations in the sensor image. [Figure 3C] 1 illustrates variations in the sensor image. [Figure 3D] 1 illustrates variations in the sensor image. [Figure 3E] 1 illustrates variations in the sensor image. [Figure 4A] 1 illustrates a schematic diagram of tumor motion and tumor position in tumor-perspective space and patient-perspective (static) space based on data from sensors. [Figure 4B] 1 illustrates a schematic diagram of tumor motion and tumor position in tumor-perspective space and patient-perspective (static) space based on data from sensors. [Figure 4C] 1 illustrates a schematic diagram of tumor motion and tumor position in tumor-perspective space and patient-perspective (static) space based on data from sensors. [Figure 4D] 1 illustrates a schematic diagram of tumor motion and tumor position in tumor-perspective space and patient-perspective (static) space based on data from sensors. [Figure 5] 1 illustrates a flowchart representation of one variation of a radiation delivery method. [Figure 6] 1 illustrates a flowchart representation of a treatment planning method incorporating target sensor data from multiple target sensors. [Figure 7] 1 illustrates a flowchart representation of one variation of a radiation delivery method. [Figure 8A] 1 illustrates a schematic representation of tumor motion and tumor position in tumor-viewed space and patient-viewed (static) space based on data from a constant-valued ("null") sensor. [Figure 8B] 1 illustrates a schematic representation of tumor motion and tumor position in tumor-viewed space and patient-viewed (static) space based on data from a constant-valued ("null") sensor. [Figure 8C] 1 illustrates a schematic representation of tumor motion and tumor position in tumor-viewed space and patient-viewed (static) space based on data from a constant-valued ("null") sensor. [Figure 8D]1 illustrates a schematic representation of tumor motion and tumor position in tumor-viewed space and patient-viewed (static) space based on data from a constant-valued ("null") sensor. [Figure 9] 1 illustrates a flowchart representation of a treatment planning method incorporating target sensor data from a target sensor. [Figure 10] 1 illustrates a flowchart representation of one variation of a radiation delivery method. [Figure 11A] 1 illustrates simulation results of one embodiment of a radiation treatment plan and radiation delivery method. [Figure 11B] 1 illustrates simulation results of one embodiment of a radiation treatment plan and radiation delivery method. [Figure 11C] 1 illustrates simulation results of one embodiment of a radiation treatment plan and radiation delivery method. DETAILED DESCRIPTION OF THE INVENTION

[0023] Disclosed herein are methods for radiation treatment planning and delivery using acquired target area sensor data. In some variations, the sensor data may be acquired during a treatment session in the presence of a patient, while in other non-therapeutic variations, the sensor data may be acquired during a QA session in the absence of a patient. The radiation treatment system may include one or more target area sensors (also referred to herein as target sensors), and data acquired by the one or more target sensors during radiation delivery (e.g., a QA or treatment session) may be used to calculate radiation fluence delivered to one or more target areas during a radiation delivery session. The target area sensors may include image sensors (e.g., X-ray detectors, PET detectors, and / or optical sensors) and / or position sensors (e.g., phantoms and / or fluence measurement devices such as ion chambers and / or radiographic films) that may be coupled to the patient and / or target area. The radiation treatment planning method may include calculating, for each patient target area, a shift-invariant emission filter based on the sensor characterization PDF of the one or more target sensors used during the treatment session. During a treatment session, readings from the target sensor may be used to generate a sensor image, which may be convolved with a shift-invariant emission filter to calculate the delivered radiation fluence, which may be performed within one hour (e.g., within about 60 minutes, within about 30 minutes, within about 20 minutes, within about 15 minutes, within about 10 minutes, within about 5 minutes, within about 3 minutes, within about 1 minute, within about 0.5 minutes, within about 20 seconds, within about 2 seconds, within about 1 second, within about 0.5 seconds, within about 0.25 seconds, within about 0.2 seconds, within about 0.1 seconds, etc.) of delivering therapeutic radiation to the patient target area. In some variations, the one or more target sensors may include one or more position sensors that may be associated with different anatomical structures of the patient, for example, position sensors that sense the position of one or more patient target regions, one or more organs-at-risk (OARs), and / or one or more bony structures that may be used by the radiation therapy system as anatomical landmarks and / or reference points for radiation delivery.Alternatively or additionally, the target sensor may comprise one or more image sensors, such as an array of PET detectors, an array of MV detectors, an array of kV detectors, and / or an optical detector (e.g., a camera that detects visible and / or infrared light). In these cases, the sensor data readings may already be images, so additional sensor data image generation may not be required. In some variations, the patient's position may be sensed by coupling markers to the patient and using a camera to monitor the position and / or movement of the markers.

[0024] In some variations, radiation treatment planning may use sensor characterization data from multiple target sensors to calculate multiple sets of corresponding emission filters. For example, a radiation treatment planning method may include generating a first sensor characterization image based on a first sensor characterization PDF of a first target sensor, generating a second sensor characterization image based on a second sensor characterization PDF of a second target sensor, calculating a first set of emission filters that can be applied to sensor data from the first target sensor acquired during a treatment session, and calculating a second set of emission filters that can be applied to sensor data from the second target sensor acquired during a treatment session. The first target sensor and the second target sensor may be the same sensor type and / or different sensor types. During a treatment session, sensor data may be acquired from both the first target sensor and the second target sensor and used to generate first and second sensor data images, which are then convolved with their respective emission filters to generate a radiation fluence map for delivery.

[0025] While the examples provided herein are in the context of intensity-modulated radiation therapy (IMRT), stereotactic body radiation therapy (SBRT), and / or biologically guided radiation therapy (BgRT), it should be understood that the systems and methods described herein can be used in any type of external beam radiation therapy. Furthermore, while some of the methods described herein are described in the context of a treatment session in which radiation may be delivered to a patient, the same methods can be used in quality assurance (QA) sessions that do not involve radiation delivery to a patient. That is, the methods described herein can also be used in non-therapeutic modalities, such as QA procedures or sessions, in which radiation is delivered to a phantom and / or fluence measurement device instead of a patient. Thus, methods described for delivering radiation to a patient target region can also be used to deliver radiation to a target region of a phantom located within the patient area of ​​a radiation therapy system. In some variations, the phantom can be anthropomorphic and / or optionally attached to a motion simulator to mimic patient movement. Examples of fluence measurement devices may include, but are not limited to, an ionization chamber, an array of diodes and / or thin-film transistors (TFTs), a thermoluminescent dosimeter (TLD), a dosimeter, etc.

[0026] system FIG. 1A is a block diagram representation of an external beam radiation therapy system. The radiation therapy system (100) includes one or more therapeutic radiation sources (102) and a patient platform (104). The therapeutic radiation source may include an X-ray source, an electron source, a proton source, and / or a neutron source. For example, the therapeutic radiation source (102) may include a linear accelerator (linac), a cobalt-60 source, and / or an X-ray machine. The therapeutic radiation source may be movable around the patient platform so that a radiation beam may be directed at a patient on the patient platform from multiple emission positions and / or angles. In some variations, the radiation therapy system may include one or more beam-shaping elements and / or assemblies (106) that may be positioned within the beam path of the therapeutic radiation source. For example, the radiation therapy system may include a linac (102) and a beam-shaping assembly (106) disposed in the path of the radiation beam. The beam-forming assembly may include one or more movable jaws and a multi-leaf collimator (e.g., a binary multi-leaf collimator, a 2D multi-leaf collimator, etc.). The linac and beam-forming assembly may be mounted to a gantry that includes a motion system configured to adjust the position of the linac and beam-forming assembly. The patient platform (104) may also be movable. For example, the patient platform (104) may be configured to linearly translate the patient along a single axis of motion (e.g., along the IEC-Y axis) and / or to move the patient along multiple axes of motion (e.g., two or more degrees of freedom, three or more degrees of freedom, four or more degrees of freedom, five or more degrees of freedom, etc.). In some variations, the radiation therapy system may have a 5-DOF patient platform configured to move along the IEC-Y axis, the IEC-X axis, the IEC-Z axis, as well as pitch and yaw.

[0027] In some variations, the radiation therapy system (100) may include one or more target area sensors (108). The target sensors may include image sensors (e.g., X-ray detectors, PET detectors, MR coils, and / or optical sensors, etc.) and / or position sensors (e.g., implantable beacons, surface markers, etc.) and / or combined sensors (e.g., X-ray tracked implantable fiducial systems). In some variations, the position sensors or combined sensors may be surgically implanted in the tumor or target area using any minimally invasive method (e.g., percutaneous placement using a needle) and / or may be implanted in the patient's skin adjacent to the target area. The radiation therapy system may include a target sensor system having multiple sensor elements and / or components. In some variations, the target sensor system may include an imaging system. For example, a BgRT system, such as an EGRT system configured to deliver radiation based on PET emission data, may have a target sensor system (i.e., a PET imaging system) having one or more arrays of PET detectors. The PET emission data can be used to characterize the location of one or more patient tumors and / or the patient's biological and / or physiological state. Alternatively or additionally, the radiation therapy system may have a target sensor system including a CT imaging system. The CT imaging system may include an X-ray source (e.g., a kV X-ray source) and an X-ray detector located opposite the X-ray source. Radiopaque implanted fiducials can be placed in a predetermined patient area (e.g., using a percutaneous method, needle implantation, etc.), and the CT imaging system or X-ray projector system can be configured to image and track the location of the fiducials. Optionally, the target sensor can include an X-ray detector (e.g., an MV X-ray detector) located opposite the therapeutic radiation source. The X-ray detector data can be used to characterize the patient's position. One example of a target sensor system can include an implanted or implanted reflector less than about 1 cm that can be placed in the patient target area. The target sensor system can also include a detector that uses radar signals that can be used to identify the location of the reflector.Alternatively or additionally, the target sensor system may include a photodetector configured to sense light within the visible and / or near-infrared spectrum. For example, the target sensor system may include a camera positioned such that the patient platform is within its field of view, and the camera may be configured to acquire a video stream (e.g., a series of images over time) from which the patient's location and / or movement may be determined. Optionally, to facilitate detection of the patient's location and / or movement, markers, tags, optical fiducials, and / or any visual indicators may be coupled to the patient that may facilitate detection of the patient's position and / or movement by the camera. For example, optical fiducials may be coupled to the patient's skin using adhesive and / or straps. In some variations, the marker or tag may include an optical emitter (e.g., emitting light at a preselected wavelength and / or pulsed frequency), and the camera may be configured to detect light from the emitter. Alternatively or additionally, the marker or tag may be an object having optical properties that may be distinguishable from the surrounding optical environment. For example, the marker, tag, or other such optical fiducial may have a unique solid color, a distinct visual pattern, high-contrast markings, and / or high optical reflectivity properties that may facilitate detection by the camera's image processor. That is, the marker or tag may comprise a visually distinct pattern or graphics that allows it to be easily distinguished from background features. In some variations, the marker or tag may be a pigment (e.g., a tattoo) injected into a localized area of ​​the patient's skin. The pigment may include a radiopaque material that can be detected using an X-ray (e.g., CT) imaging system.

[0028] In some variations, the target sensor may include a position sensor configured or arranged to detect the location and / or movement of the patient. In some variations, at least one position sensor may be attached to an anatomical structure proximate to the target region (e.g., a tumor or lesion) and / or an anatomical structure whose location and / or movement can be correlated with the location and / or movement of the target region.

[0029] Some variations of radiation therapy systems may include one or more target sensors of different sensing modalities. For example, a radiation therapy system may have a target sensor system including an imaging system (e.g., a PET imaging system, a CT imaging system, an MR imaging system, an optical or visible light imaging system, a fluoroscopic imaging system) and a target sensor including a position sensor that can be attached to a patient during a treatment session. The radiation therapy system may have a first target sensor system including a CT imaging system (i.e., a kV X-ray source and detector, an X-ray projector system), a second target sensor system including a PET imaging system (i.e., multiple PET arrays arranged in a continuous ring or two separate, opposing arcs), and / or a third target sensor system including a therapeutic radiation source and an MV detector located on the opposite side of the therapeutic radiation source. The radiation therapy system may have a first target sensor system including a CT imaging system and a second target sensor including a position sensor coupled to the patient at a location correlated to the location of the patient target area. Alternatively or additionally, the radiation therapy system may have multiple position sensors attached to predetermined areas of the patient's body that may provide additional information regarding the location of the patient target area.

[0030] While the description herein describes target sensors in the context of radiation delivery to a patient, it should be understood that target sensors can also include sensors in the context of radiation delivery during non-therapeutic procedures (e.g., QA procedures or sessions) in the absence of a patient. The target sensors described above can also be used in QA procedures or sessions for radiation delivery to a phantom and / or fluence measurement device, i.e., the target sensors and / or sensor systems can also be used in non-therapeutic procedures without radiation delivery to a patient. Various functions, structures, and systems of the target sensors and / or sensor systems described herein can be used during QA procedures or sessions in which a patient is replaced by a phantom. For example, the target sensor system can include an imaging system (e.g., a PET imaging system, a CT imaging system, an MR imaging system, an optical or visible light imaging system, a fluoroscopic imaging system) that can be used in a QA procedure in which radiation is delivered to a phantom instead of a patient. Position sensors, markers, tags, optical fiducials, and / or any visual indicators can be applied to the phantom as they would be applied to a patient. Any of the target sensors described herein (e.g., image sensors, and / or position sensors, and / or combined sensors, etc.) may be used in conjunction with a phantom and / or fluence measurement device during non-therapeutic treatments.

[0031] The radiation therapy system (100) also includes a controller (110) in communication with the therapeutic radiation source (102), the beam-shaping element or assembly (106), the patient platform (104), and one or more target sensors (108) (e.g., one or more target sensor systems). The controller (110) may include one or more processors and one or more machine-readable memories in communication with the one or more processors, which may be configured to execute or perform any of the methods described herein. The one or more machine-readable memories may store instructions that cause the processor to execute modules, processes, and / or functions associated with the system, such as one or more treatment plans, target sensor data (e.g., imaging data, location / position data, motion data), calculation of a radiation fluence map based on the treatment plan and / or clinical objectives, segmentation of the fluence map into radiation therapy system instructions (which may, e.g., direct the operation of the gantry, therapeutic radiation source, beamforming assembly, patient platform, and / or any other components of the radiation therapy system), and image and / or data processing associated with treatment planning and / or radiation delivery. In some variations, the memory may store treatment plan data (e.g., treatment plan emission filters, fluence map, planning images), target sensor data, instructions for processing the sensor data to derive a radiation-delivered fluence map, and instructions for delivering the derived fluence map (e.g., instructions for operating the therapeutic radiation source, beamforming assembly, and patient platform in a coordinated manner). The controller of the radiation therapy system may be connected to other systems by wired or wireless communication channels. For example, the radiation treatment system controller may communicate wired or wirelessly with the radiation treatment planning system controller so that fluence maps, emission filters, target sensor data (e.g., sensor characterization probability density functions), planning images (e.g., CT images, MRI images, PET images, 4D CT images), patient data, and other clinically relevant information can be transferred from the radiation treatment planning system to the radiation treatment system.The delivered radiation fluence, any dose calculations, and any clinically relevant information and / or data acquired during the treatment session may be transferred from the radiation treatment system to a radiation treatment planning system, which may use this information to adapt the treatment plan and / or adjust the delivery of radiation for subsequent treatment sessions.

[0032] FIG. 1B illustrates one variation of a radiation therapy system (100). The radiation therapy system (100) may include a gantry (110) rotatable about a patient treatment region (112), one or more PET detectors (108) mounted on the gantry, a therapeutic radiation source (102) mounted on the gantry, a beam-shaping module (106) disposed in a beam path of the therapeutic radiation source, and a patient platform (119) movable within the patient treatment region (112). In some variations, the gantry (110) may be a continuously rotating gantry (e.g., capable of rotating through 360 degrees and / or in an arc with an angular extent of less than about 360 degrees). The gantry (110) may be configured to rotate about 20 RPM to about 70 RPM about the patient treatment region (112). For example, the gantry (110) may be configured to rotate about 60 RPM. The beam shaping module (106) may include a movable jaw and a dynamic multi-leaf collimator (MLC). The beam shaping module may be arranged to provide a variable longitudinal collimation width of 1 cm, 2 cm, or 3 cm at the system isocenter (e.g., the center of the patient treatment region). The jaw may be located between the therapeutic radiation source and the MLC or below the MLC. Alternatively, the beam shaping module may include split jaws, with a first portion of the jaw located between the therapeutic radiation source and the MLC and a second portion of the jaw located below the MLC and coupled to the first portion of the jaw so that both portions move together. The therapeutic radiation source (102) may be configured to emit radiation at predetermined launch positions (e.g., launch angles 0° / 360° to 359°) around the patient treatment region (112). For example, in a system having a continuously rotatable gantry, there may be about 50 to about 100 firing positions (e.g., 50 firing positions, 60 firing positions, 80 firing positions, 90 firing positions, 100 firing positions, etc.) at various angular positions (e.g., firing angles) along the circle circumscribed by the therapeutic radiation source as it rotates. The firing positions may be evenly distributed such that the angular displacement between each firing position is the same.

[0033] 1C is a perspective component diagram of the radiation therapy system (100). As shown therein, the beam shaping module may further comprise a primary collimator or jaw (107) disposed above the binary MLC (122). The radiation therapy system may also comprise an MV X-ray detector (103) located opposite the therapeutic radiation source (102). Optionally, the radiation therapy system (100) may further comprise a kV CT imaging system on a rotatable ring (111) mounted on the rotatable gantry (110) such that rotating the gantry (110) also rotates the ring (111). The kV CT imaging system may The radiation therapy system (100) may include an X-ray source (109) and an X-ray detector (115) located opposite the X-ray source (109). The therapeutic radiation source or linac (102) and the PET detector (108) may be mounted on the same cross-sectional plane of the gantry (i.e., the PET detector is coplanar with the treatment plane defined by the linac and beamforming module), while the kV CT scanner and ring may be mounted on a different cross-sectional plane (i.e., not coplanar with the treatment plane). The radiation therapy system (100) of FIGS. 1B and 1C may have a first target sensor system comprising a kV CT imaging system and a second target sensor system comprising a PET detector. Optionally, a third target sensor system may comprise an MV X-ray source and an MV detector. Target sensor data acquired by one or more of these target sensor systems may include X-ray and / or PET imaging data, and the radiation therapy system controller may be configured to store the acquired target sensor data and use the target sensor data to calculate the radiation delivered fluence. In some variations, additional target sensors, such as position sensors, may be included, and the controller may be configured to receive location and / or motion data from the position sensors and combine this data with other target sensor data to calculate the radiation delivered fluence. Further description of radiation therapy systems that may be used with any of the methods described herein is provided in U.S. Patent No. 10,695,586, filed November 15, 2017.

[0034] The patient platform (104) may be movable to discrete, predetermined locations along the IEC-Y axis within the treatment region (112). These discrete, predetermined locations may be referred to as "beam stations." For example, the radiation treatment planning system may designate 200 beam stations, each spaced approximately 2 mm (e.g., 2.1 mm) from its neighboring beam station. The number of beam stations may vary from approximately 50 to approximately 500, and the spacing between each beam station may be at least approximately 0.5 mm (e.g., approximately 1 mm or more, approximately 2 mm or more, approximately 20 mm or more, approximately 50 mm or more, etc.). During a treatment session, the radiation treatment system may move the patient platform to each of the beam stations and park the platform at the beam stations while radiation is delivered to the patient. In some variations, after the platform travels in a first direction (e.g., into the bore) to each of the 200 beam stations, the platform may travel in a second direction opposite the first direction (e.g., out of the bore, in the reverse direction) to each of the 200 beam stations, and radiation is delivered to the patient while the platform is stopped at the beam station. Alternatively or additionally, after the platform travels in the first direction (e.g., into the bore) to each of the 200 beam stations, where radiation is delivered at each of the beam stations, the platform may be moved in the reverse direction back to the first beam station. No radiation may be delivered while the platform is moved back to the first beam station. The platform may then travel again in the first direction to each of the 200 beam stations for a second pass of radiation delivery. In some variations, the platform may be moved continuously while radiation is delivered to the patient and may not be stopped at a beam station. Further description of patient platforms that may be used with any of the radiation therapy systems and methods described herein is provided in U.S. Patent No. 10,702,715, filed November 15, 2017.

[0035] Figure 1D illustrates another variation of a radiation therapy system (150) that may be used to deliver radiation according to any of the methods described herein. The radiation therapy system (150) may have the components of the radiation therapy system depicted in the block diagram of Figure 1A. The radiation therapy system (150) may include a gantry (151) having a first pair of arms (152) rotatable around a patient area and a second pair of arms (154) rotatable around the patient area, an imaging system including a therapeutic radiation system having an MV radiation source (156) attached to the first arm (152a) of the first pair of arms (152) and an MV detector (158) attached to the second arm (152b) of the first pair of arms (152), and a kV radiation source (160) attached to the first arm (154a) of the second pair of arms (154) and a kV detector (162) attached to the second arm (154b) of the second pair of arms (154). The first and second arms of the first pair of arms (152) may be positioned opposite each other (e.g., on either side of a patient area, facing each other, and / or at about 180° from each other) such that the MV radiation source (156) and the MV detector (158) are positioned opposite each other (e.g., the MV detector is positioned in the beam path of the MV radiation source). The first and second arms of the second pair of arms (154) may be positioned opposite each other (e.g., on either side of a patient area, facing each other, and / or at about 180° from each other) such that the kV radiation source (160) and the kV detector (162) are positioned opposite each other (e.g., the kV detector is positioned in the beam path of the kV radiation source). In this system, the target sensor system may include a kV radiation source and a kV detector. Optionally, the second target sensor system may include a MV radiation source and a MV detector. The target sensor data may include imaging data acquired by the kV detector (and / or MV detector) after each kV X-ray source (and / or MV X-ray source) pulse. Examples of target sensor data may include X-ray projection imaging data, such as 2D projection data. The radiation therapy system controller may be configured to store the acquired target sensor data and use the target sensor data to calculate the radiation delivered fluence.In some variations, additional target sensors, such as position sensors, may be included, and the controller may be configured to receive location and / or movement data from the position sensors and combine this data with other target sensor data to calculate the radiation delivered fluence.

[0036] The MV radiation source (156) (i.e., the therapeutic radiation source) may be configured to emit radiation at predetermined launch positions around the patient area. In some variations in which the MV radiation source is moved around the patient area along a single plane, the launch positions may be between 0° / 360° and 359°, which may be referred to as the launch angle. Alternatively or additionally, the gantry and / or gantry arm may be configured to move the MV radiation source to a launch position at any coordinate in 3D space, i.e., as specified by coordinates (x, y, z). For example, the gantry arms (152, 154) may be robotic arms with articulating joints that may be configured to position and / or aim the MV radiation source at any desired launch position. The gantry may be configured to move the MV radiation source continuously through the launch positions or may be configured to step the MV radiation source to each launch position (i.e., move the MV radiation source to the launch position and remain stationary at the launch position). Alternatively or additionally, the MV radiation source may be configured to emit radiation only at predetermined firing positions, or may be configured to emit radiation continuously, even when being moved from one firing position to the next.

[0037] Radiation treatment planning method The target sensor and / or target sensor system may be configured to provide a continuous stream of sensor data in real time during a radiation delivery session (e.g., a treatment session or a QA session). The controller or radiation treatment system may be configured to calculate a radiation fluence map based on this target sensor data and deliver the calculated radiation fluence on the same day the sensor data is acquired, for example, within a few hours (e.g., less than about 2 hours, less than about 1 hour, etc.) of acquiring the target sensor data, within a few minutes (e.g., less than about 45 minutes, less than about 30 minutes, less than about 15 minutes, less than about 5 minutes, less than about 3 minutes, etc.) of acquiring the target sensor data, and / or within a few seconds (e.g., less than about 5 seconds, less than about 3 seconds, less than about 2 seconds, less than about 1 second, less than about 0.5 seconds, less than about 250 ms, less than about 100 ms, etc.) of acquiring the target sensor data. The controller may be configured to continuously acquire the target sensor data within a specified time window of about 0.1 seconds to about 10 minutes. In some variations, the controller may be configured to continuously acquire target sensor data within a short time window (e.g., about 2 seconds or less, about 1 second or less, about 0.5 seconds or less, etc.) and use the target sensor data acquired during the short time window to calculate the radiation to be delivered within a subsequent delivery window (which may, for example, be as short as the acquisition time window). For example, the sensor data acquisition time window may be approximately proportional to the delivery window. In some variations, the delivery window is the time interval during which the calculated radiation should be delivered. For example, in a radiation therapy system in which the therapeutic radiation source continuously rotates around the patient area and the dwell time per firing position / angle is about 10 ms, the delivery window may be 100 ms, during which time the therapeutic radiation source will emit radiation pulses at 10 firing positions / angles. However, sensor data acquired within a short period of time may be noisy (e.g., have a low signal-to-noise ratio) and / or may not contain much information about the patient target area. For example, PET and / or X-ray imaging data acquired over a 0.5 second time window may not be adequate to generate an image with sufficient resolution to locate a patient target region.However, if the noise characteristics and / or variability of the target sensor are known prior to a treatment session, this information can be incorporated into the treatment plan. The radiation treatment planning methods described herein may include calculating an emission filter that incorporates the noise characteristics and / or sensor variability (e.g., a sensor probability density function) of one or more target sensors activated during a treatment session (or any radiation delivery session, including a QA session). The calculated emission filter may be applied to target sensor data acquired during the treatment session. This may facilitate delivery of therapeutic radiation to a moving patient target region despite any noisy and / or incomplete target sensor data. Similarly, target sensors may be used in a QA session in which the patient is replaced by a phantom, and the phantom may include a region that simulates the location and / or characteristics of the patient target region. In some variations, the phantom may be mounted on a motion stage to simulate the movement of the patient target region during an actual treatment session. The movement may, in some variations, be based on a motion retention histogram for the target region. The calculated emission filter may be applied to the target sensor data acquired during the QA session to deliver radiation to the target region. Measurement of the delivered radiation to the phantom target region may facilitate evaluation of the treatment plan and the functionality of the radiation treatment system. In some variations, the data acquired from the target sensor during the QA session may be used to modify the treatment plan and / or adjust components of the radiation treatment system.

[0038] The radiation treatment planning methods described herein may include generating a set of shift-invariant emission filters for each target sensor modality. The shift-invariant emission filters may be linear functions or operators, or may be nonlinear functions or operators. For example, if a single position sensor is to be used to acquire real-time patient position data (e.g., location data and / or motion data) during a treatment session, the treatment planning system may use the position sensor characterization data converted to images to calculate the set of emission filters. The position sensor characterization data may include a sensor error PDF. During the treatment session, the calculated emission filters may be applied to position sensor data images (generated based on the real-time acquired position sensor data) to calculate a radiation fluence map for delivery. In some variations, the position sensor data images may be discrete 3D delta functions centered on the position sensor data readings, discretized Gaussian function images with means centered on the position sensor data readings, truncated Gaussian functions, etc. In another example, if an image sensor for a single imaging modality (e.g., a kV detector, a PET detector, an MV detector, etc.) is to be used to acquire real-time patient position data and / or patient target region position data (e.g., location data and / or motion data) during a treatment session, the treatment planning system may use planning images generated by the same image sensor (and / or generated in the same imaging modality), which may be low-noise and non-sparse, as a PDF to calculate a set of emission filters. During the treatment session, the calculated emission filters may be applied to the real-time acquired image sensor data (which may be noisy or sparse) to calculate a radiation fluence map for delivery. In one variation, two or more different target sensor modalities may be used during a treatment session to acquire real-time patient position data and / or patient target region position data (e.g., location data and / or motion data) during the treatment session.For example, a radiation therapy system may include an imaging system including one or more image sensors (e.g., PET detector, X-ray detector) configured to acquire imaging data during a treatment session. Additionally, a position sensor may be coupled to the patient and configured to acquire location data and / or motion data during a treatment session. The treatment planning system may use planning images generated by the same one or more image sensors (and / or generated with the same imaging modality) as PDFs to calculate a first set of emission filters, and use error PDFs of the position sensors converted to images (e.g., sensor characterization images) to calculate a second set of emission filters. To calculate the delivered fluence during a treatment session, the controller may apply the first set of emission filters to the real-time acquired imaging data and the second set of emission filters to the real-time acquired position data (i.e., position sensor data images generated from the acquired position data), combining them to calculate a radiation fluence map for delivery. Any combination of target sensor modalities may be used during a treatment session to acquire data about the patient and / or one or more patient target regions (e.g., location, motion, and / or biological and / or physiological state of the patient and / or target region). To that end, the radiation treatment planning method may include calculating a set of emission filters for each of the target sensor modalities. During a treatment session, the emission filters may be applied to their respective target sensor data and then combined to calculate a radiation fluence map for delivery.

[0039] The methods for generating treatment plans described herein are performed in the absence of a patient. These treatment planning methods alone do not involve the delivery of therapeutic radiation to a patient.

[0040] Radiation treatment planning method: single sensor One variation of a radiation treatment planning method incorporating target sensor data from a target sensor is illustrated in Figure 2A. The method (200) includes obtaining (202) a target sensor characterization PDF and generating a target sensor characterization image N based on the sensor characterization PDF of the target sensor for each firing position i of the therapeutic radiation source. i and generating (204) a target sensor characterization image N i and calculating (206) a shift-invariant firing filter corresponding to each of the i firing locations based on a prescribed dose criterion for the patient target area. In some variations, the target sensor characterization PDF may be provided by a previous model and / or calculation, in which case obtaining (202) the sensor characterization PDF may be optional. The sensor characterization PDF may be a sensor error PDF. The sensor characterization PDF may be derived from multiple target sensor data, i.e., expressed as:

[0041]

number

[0042] FIG. 2L is an example of an image of a position sensor characterization PDF. FIGS. 2B-2K illustrate one example of how the sensor characterization PDF image of FIG. 2L can be generated (204). A position sensor data reading can be, for example, two coordinate values ​​(x, y). This position data reading can be converted into an image or plot. For example, a position data sensor reading of (0,0) can be converted into the sensor image shown in FIG. 2B, which is a plot of a delta function centered at (0,0). Alternatively, the position data sensor reading (0,0) can be converted into a Gaussian function centered at (0,0), i.e., a Gaussian function or truncated Gaussian function centered at a "true" value, with a FWHM calculated based on the resolution of the position sensor and / or the resolution (e.g., granularity) at which the radiation therapy system can accurately deliver the radiation beam. Additional position data sensor readings can be obtained, and the sensor characterization PDF can be a list of position data sensor readings. The sensor characterization PDF can be a histogram derived from the list of position data sensor readings.Method 200 includes generating an image of the sensor characterization PDF, which in this example may include plotting the error (or difference) of each of the position sensor readings compared to the position sensor readings depicted in FIG. 2B (centered at (0,0)), where FIG. 2C is a sensor error image of the error (or difference) of 100 position sensor readings compared to the readings at (0,0), FIG. 2D is a sensor error image of the error of 200 position sensor readings compared to the readings at (0,0), FIG. 2E is a sensor error image of the error of 300 position sensor readings compared to the readings at (0,0), and FIG. 2F is a sensor error image of the error of 300 position sensor readings compared to the readings at (0,0). FIG. 2G is a sensor error image of the error of 400 position sensor readings compared to the reading at (0,0), FIG. 2H is a sensor error image of the error of 500 position sensor readings compared to the reading at (0,0), FIG. 2I is a sensor error image of the error of 700 position sensor readings compared to the reading at (0,0), FIG. 2J is a sensor error image of the error of 800 position sensor readings compared to the reading at (0,0), and FIG. 2K is a sensor error image of the error of 900 position sensor readings compared to the reading at (0,0). As more position sensor readings are obtained, each transformed into a delta function, and aggregated, a sensor image of a sensor error characterization PDF (e.g., FIG. 2L) can be generated. Similar methods can be used to generate position sensor error characterization PDFs for 1D or 3D sensor readings. Conceptually, the sensor error characterization PDF image illustrated in Figure 2L may represent the sensor's variability, or sensor error characterization PDF, relative to a "true" position value at (0,0). That is, when an object is located at (0,0), the position sensor may provide a data reading that is not (0,0), and the probability that the position sensor may provide any particular data reading depends on the density of that data value in the sensor error characterization PDF image of Figure 2L.In this example, the position sensor is more likely to output a data reading that is close to the "true" position value (i.e., the densest part of the image is near (0,0)) than it is to output a data reading whose position is on the outer edge of the error PDF (i.e., the outer edge is the least dense). In some variations, the image of the sensor characterization PDF can be a Gaussian "ball" if the sensor error / variability is normally distributed.

[0043] 3A-3E illustrate various examples of sensor images. In some variations, the sensor image may be a 2D Gaussian function (FIG. 3A), a 3D Gaussian function (FIG. 3B), or a 3D truncated Gaussian function (FIG. 3D). The sensor image may be a 2D delta function (FIG. 3C) or a 3D delta function. Some target sensors may have asymmetric noise. FIG. 3E illustrates an example of an image of a sensor characterization PDF of a target sensor with asymmetric noise and / or error and / or variability.

[0044] An image of the sensor characterization PDF(N) may be generated for each of the i firing positions of the radiation therapy system. For example, a therapeutic radiation source may have 100 firing positions representing different locations (which may be predetermined) from which the radiation beam may be emitted into the patient. The sensor characterization image N i Generating 204 may include calculating, for each launch position, a projection of a sensor characterization PDF image N (e.g., the sensor error characterization PDF image of FIG. 2L) on each of the i launch positions, such that sensor characterization image N1 is the projection of sensor characterization image N on launch position 1, sensor image N2 is the projection of sensor characterization image N on launch position 2, etc. Alternatively, an image of the sensor characterization PDF can be calculated directly at each launch position using the beam field of view space at that launch position.

[0045] Emission filter p for emission position i i is the radiation fluence map F for delivery at launch location i to the patient target region i and target sensor characterization image N iand so that F i =p i * N i (e.g., an emission filter convolved with a sensor-characterized PDF image).

[0046] In some variations, the emission filter is a matrix multiplication of the target sensor characterization image N i Radiation fluence F for delivering images i (e.g., a fluence map) i where the sensor characterization image can be linearized into a vector: F i =P i N i (the firing matrix is ​​multiplied with the vectorized sensor characterization image).

[0047] While the emission filter may be a linear function or operator in the examples described herein, in some variations the emission filter may be a non-linear function or operator (e.g., a truncated convolution function, a convolution followed by thresholding, a matrix multiplication with a softmax operator). Alternatively, or additionally, the image of the target sensor data may be pre-processed (e.g., using one or more non-linear functions) before being converted to a radiation fluence for delivery using a linear emission filter (e.g., a shift-invariant linear emission filter).

[0048] The radiation fluence F (eg, a fluence map) over multiple firing locations (eg, i firing locations) may be represented by:

[0049]

number

[0050] For example, the radiation fluence for 50 firing locations may be:

[0051]

number

[0052] During radiation treatment planning, a clinician may provide a dose prescription for a patient target region and / or organs at risk (OARs). The dose prescription may include, for example, dose goals and objectives for the patient target region and / or OARs. The radiation dose (i.e., the amount of radiation absorbed by the subject) and the radiation fluence (i.e., the amount of radiation emitted by the radiation source, typically specified by a radiation beam or beamlet) are related to each other (e.g., mapped to each other) by a dose calculation matrix A, i.e., D=A·F

[0053] The dose calculation matrix represents the dose contribution from each of a plurality of radiation beamlets to each voxel of the patient target region (and / or OAR). For example, the dose calculation matrix A may be a (k×n) matrix, where n is the number of possible radiation beamlets {b i}, where k may be the number of preselected voxels in the patient target region. The i-th column of the dose calculation matrix A (with k elements) may be the unity-weighted beamlet b i represents the dose contribution to each of the k voxels from . The dose calculation matrix A may be calculated column-by-column, for example, by ray tracing each beamlet aperture along a path through the patient target region and calculating the 1-weighted beamlet contribution of each of the k voxels. The beamlet aperture may be an MLC aperture defined by a single MLC leaf aperture (i.e., of a binary MLC or 2D MLC). Examples of algorithms for calculating dose calculation matrices that may be used in any of the methods described herein may include Monte Carlo simulation, collapsed cone convolution superposition, pencil beam convolution, etc. Each patient target region and / or OAR may have its own dose calculation matrix.

[0054] As mentioned above, the radiation fluence map F can be represented by an emission filter p convolved with the target sensor characterization image N (F=p * N). Therefore, D=A·F=A·(p * N) where A is the dose calculation matrix. The cumulative dose over i firing locations can be:

[0055]

number

[0056] In addition to defining the dose prescription, the clinician may set one or more constraints and / or cost or penalty functions C(D,F) that specify characteristics of the dose distribution and / or radiation fluence map. Examples of cost functions may include, but are not limited to, minimum dose to the target region, average or maximum dose over the OARs, and / or fluence smoothness, total radiation output, total tissue dose, treatment time, etc.

[0057] The radiation treatment planning system may be configured to calculate a radiation fluence map F such that the dose prescription and constraints C(D,F) are satisfied. The radiation treatment planning system may iterate through different radiation fluence values ​​and / or maps to find a fluence value and / or map that minimizes the cost function C(D,F) while still meeting the dose prescription requirements. To calculate 206 an emission filter (e.g., a shift-invariant emission filter) according to method 200, the radiation treatment planning system may set up an optimization problem to minimize the cost function C(D,F) given the dose calculation matrix A and the target sensor characterization image N. Computing the emission filter p may include iterating through different emission filter values ​​such that the cost function C(D,F) is minimized while still achieving the dose goals and objectives according to the dose prescription.

[0058]

number

[0059] In some variations, the above optimization problem may be rephrased by converting the convolution of the emission filter with the target sensor characterization image into a matrix multiplication:

[0060]

number

[0061]

number

[0062] By grouping the known quantities into a single matrix, the above formulation can be computationally efficient for the radiation treatment plan optimizer. For example, the matrix A GIGRTmay be calculated once at the beginning of the optimization process and need not be recalculated for each iteration of the emission filter values. The radiation treatment planning system may be configured to iterate the emission filter values ​​until one or more stopping conditions are met. Such stopping conditions may include one or more of the following: dose goals and objectives are met within acceptable tolerances; radiation fluence map values ​​converge such that the change between iterations is less than a predetermined residual criterion; cost function values ​​converge over multiple iterations; a threshold number of iterations (e.g., an upper limit on the total number of iterations) is reached; etc. The final emission filter values ​​may be stored in the memory of the radiation treatment planning system. In some variations in which the patient platform is parked at predetermined discrete platform positions (i.e., beam stations) during delivery of radiation from multiple emission positions, emission filters may be calculated for each emission position across each of the beam stations. For example, the patient platform may have 200 beam stations, each beam station being approximately 2 mm from an adjacent beam station.

[0063] Emission filters calculated using the methods described herein incorporate target sensor noise characteristics and / or variability as part of the calculation. For example, if the target sensor is a position sensor, the emission filter may be calculated based on a position sensor characterization PDF, which may be a sensor position characterization PDF or a sensor data error characterization PDF. By doing so, a delivered fluence map calculated during a treatment session by convolving the emission filter with a sensor image generated using real-time target sensor data may more accurately direct radiation to the actual location of the patient target area. In contrast, typical radiation treatment methods do not incorporate images of target sensor error and / or variability PDFs as factors during fluence map optimization. That is, typical treatment planning methods do not specify radiation fluence or target sensor noise or variability with respect to the target sensor data, and therefore, radiation delivery may not be able to accurately adapt to real-time, noisy target sensor data during a treatment session.

[0064] Radiation treatment planning using images of the target sensor characterization PDF in the above-described manner is performed fully in the tumor point-of-view (POV) frame as opposed to a static frame. The target sensor data readings are then optimized and the emission filter is calculated. Figures 4A-4D illustrate conceptual diagrams that explain the physical meaning of the target sensor data reading PDF, using a position sensor as an example. One exemplary scenario is a tumor 400 located near a bone 402, where the tumor 400 is moving and the bone 402 is stationary. A position sensor may be coupled to the tumor (e.g., by needle insertion using image guidance), for example, at its center, so that the position sensor data readings indicate the location of the tumor centroid. Figure 4A illustrates the tumor 400 and bone 402 in a tumor POV frame and the output 404 of an ideal position sensor (i.e., one that always yields true position data values ​​without any variation or error). In the tumor POV, the origin is at the tumor centroid, and the position sensor outputs a single position value 404, e.g., (0,0,0), that reflects the true location of the tumor centroid. From the tumor 400's POV, the bone 402 is moving. FIG. 4B illustrates the same tumor (400) and bone (402) in a static frame, along with ideal position sensor outputs (404a-404c) coupled to the tumor. In the static frame, the origin is the location of the tumor center of mass when the patient is initially positioned on the patient platform (e.g., positioned using CT and / or MRI imaging guidance). The ideal position sensor data readings have different values, reflecting different locations of the moving tumor center of mass. This contrasts with the position sensor data readings in the tumor POV of FIG. 4A, which always provide the same sensor data readings. This tumor motion image, generated based on the position sensor data readings, can be a tumor position histogram or tumor residence matrix. Typically, radiation treatment planning systems and methods characterize tumor position in the static frame ( FIG. 4B ) but not in the tumor POV frame ( FIG. 4A ), because the position sensor data readings in the static frame reflect the actual position of the tumor in the same frame of reference as the radiation treatment system. Furthermore, for an ideal position sensor in the tumor POV frame, the position sensor data readings are constant regardless of tumor motion.

[0065] However, in the case of a noisy position sensor, radiation treatment planning in the tumor POV may provide unexpected advantages. These advantages may include, for example, delivery of radiation that more accurately tracks the patient target region in the presence of a noisy position sensor. Another advantage may include mitigating motion-related dose artifacts that are common when treatment planning is performed in a static frame, which may reduce irradiation of surrounding non-target tissue while still delivering the prescribed dose to the patient target region. FIG. 4C illustrates a tumor (400) and bone (402) in the tumor POV frame and the output (406) of a noisy position sensor. In contrast to an ideal position sensor, a noisy position sensor may provide different position sensor data readings for the same location in space, which may be the result of sensor variability and / or error. Such sensor variability and / or error may be represented by a sensor error characterization PDF. In the tumor POV, the origin is at the tumor centroid, and the position sensor outputs multiple position values ​​(406) according to a sensor error characterization PDF centered on the true position data value. As described above, the position sensor error characterization PDF may be known (e.g., measured or derived) prior to a treatment session, particularly regardless of tumor (400) location or movement. FIG. 4D illustrates the same tumor (400) and bone (402) in a static frame, along with the outputs (408a-408c) of noisy position sensors coupled to the tumor. In the static frame, the origin may be the location of the tumor center of mass when the patient is initially positioned on the patient platform (e.g., positioned using CT and / or MRI imaging guidance). For each of the different locations of the moving tumor center of mass, there may be variation in the noisy position sensor data readings. In the example illustrated in FIG. 4D, the tumor (400) may move to three locations, and at each of those locations, the position sensor data readings may differ according to the sensor characterization PDF. That is, at a first tumor location (409a), the position sensor data reading may be one of the values ​​within a first cluster (408a) centered around the true position data value for the first location.At the second tumor location (409b), the position sensor data measurement may be any of the values ​​in a second cluster (408b) centered around the true position data value of the second location, and at the third tumor location (409c), the position sensor data measurement may be any of the values ​​in a third cluster (408c) centered around the true position data value of the third location. The cumulative image of all the position sensor data readings may depict a blurred tumor image due to the tumor being "blurred" over its motion trajectory and variability and / or error in the noisy position sensor. Additionally, because the tumor is moving and its location is being tracked with a noisy position sensor (i.e., the tumor's true position is unknown), it may be difficult to determine the position sensor error characterization PDF in a stationary frame.

[0066] Radiation treatment planning systems that calculate delivered fluence maps and / or emission filters in static frames (i.e., using blurry or "fuzzy" images of the tumor acquired during treatment planning) can result in a delivered dose that is susceptible to motion artifacts (e.g., dose peaking artifacts). Such motion artifacts can be particularly noticeable when tumor motion on the day of treatment differs from tumor motion when the treatment planning images were acquired. In contrast, radiation treatment planning methods (such as method 200 and other methods described herein) that calculate delivered fluence maps and / or emission filters based on a sensor characterization PDF in a tumor POV frame where the tumor is stationary do not require prior knowledge of tumor motion. A delivered fluence map and / or emission filter calculated in the tumor POV frame can result in a delivered dose that has fewer motion artifacts.

[0067] Although method 200 has been described above in the context of the target sensor being a position sensor, method 200 is also applicable to a target sensor system that includes an imaging system. The imaging system may include one or more image sensors, including, but not limited to, a PET detector, an MRI detector, a CT detector, an optical camera (including a camera for fluoroscopy), etc. The target sensor characterization PDF of a target sensor system that includes an imaging system may be a "complete" image having sufficient imaging data to identify the location and / or geometry of the tumor centroid and / or the contour tumor and / or OAR boundaries. This "complete" image L may be obtained by collecting multiple imaging data I during one or more diagnostic imaging sessions. j and combining (e.g., summing) them to form the treatment planning image, i.e.,

[0068]

number

[0069] Imaging data I j may include samples or image sensor readings, such as 3D PET imaging data (e.g., positron annihilation emission pathways, sometimes referred to as lines of response or LORs), 2D X-ray imaging data, projection imaging data (e.g., X-ray projections), fluoroscopic imaging data, CT imaging data, and / or MR imaging data (sub-samples in k-space from an MRI imaging pulse sequence). In some variations, the treatment planning images may be acquired using an imaging system of the radiation therapy system. For example, images used for treatment planning may be acquired using an on-board kV CT imaging system and / or MR imaging system and / or PET imaging system of the radiation therapy system. Examples of treatment planning images may include, but are not limited to, 3D PET images, 2D X-ray images, X-ray projection images, fluoroscopic images, CT images, and / or MR images. For each of the i firing positions of the therapeutic radiation source, a sensor image N iGenerating 204 may include computing a projection of the "full" image N onto each of the i firing positions. Computing 206 the shift-invariant firing filter may be performed as described above.

[0070] Radiation therapy delivery method: Single sensor 5 illustrates one variation of a radiation therapy delivery method that uses a shift-invariant emission filter calculated during treatment planning (e.g., according to method 200) and target sensor data acquired from a target sensor (or target sensor system) during a treatment session to calculate radiation fluence for delivery during that treatment session. The method 500 may include acquiring target sensor data during a treatment session (or any radiation delivery session) 502, generating a sensor image based on the acquired target sensor data 504, calculating a radiation fluence map for delivery by convolving the generated sensor image with the shift-invariant emission filter 506, and delivering radiation to a patient target area according to the calculated radiation fluence map 508. Acquiring target sensor data 502 may include acquiring position sensor data readings at the beginning of a treatment session and / or throughout the treatment session while a therapeutic radiation source delivers radiation to the patient. Acquiring target sensor data during a treatment session (502) may include acquiring target sensor data only at the beginning of the treatment session (i.e., before the therapeutic radiation source is activated for the first time) or may include acquiring target sensor data throughout the treatment session. For example, acquiring target sensor data (502) may include acquiring localized CT images and / or PET pre-scan images (and / or any of the other aforementioned imaging modalities) at the beginning of the treatment session, and may also include acquiring CT imaging data and / or PET imaging data during a portion of the treatment session during which the therapeutic radiation source is delivering radiation to the patient. In some variations, imaging data acquired during the radiation delivery portion of a treatment session may be acquired at a short or limited time interval (e.g., about 30 seconds or less, about 10 seconds or less, about 2 seconds or less, about 1 second or less, about 0.5 seconds or less, etc.). The target sensor data may be acquired immediately prior to delivery of the therapeutic radiation beam. For example, target sensor data may be acquired in less than about 5 seconds (eg, less than about 3 seconds, less than about 2 seconds, less than about 1 second, less than 0.5 seconds, less than 0.1 seconds, etc.) of radiation delivery.In some variations, target sensor data may be acquired for a firing position (e.g., firing position i) while the therapeutic radiation source is located at a previous firing position (e.g., firing position i-1) and / or while the therapeutic radiation source is moving to the firing position (e.g., during movement between firing positions). Alternatively or additionally, target sensor data may be acquired for a firing position (e.g., firing position i) when the therapeutic radiation source is located at that firing position. For example, imaging data acquired during the radiation delivery portion may be sub-samples in k-space from PET LOR, X-ray projection, and / or MRI imaging pulse sequences.

[0071] Generating 504 a sensor image based on the acquired sensor data may include, for example, aggregating the acquired imaging data to generate a map of pixel and / or voxel intensity values. Because the imaging data may be acquired over a limited time interval, the resulting sensor image (or image map) may be referred to as a partially sampled image or a limited-time sampled (LTS) image. An LTS image may be , may have a high level of noise (i.e., a high signal-to-noise ratio) and, as a result, may not provide sufficient information, when considered alone, to identify the contour and / or center of gravity of a patient target region or OAR. Alternatively or additionally, imaging data may be acquired at the beginning of a treatment session (e.g., a CT localization scan and / or a PET pre-scan). In some variations, generating a sensor image (504) may include plotting the acquired sensor data readings as a delta function or a Gaussian function centered on the sensor data readings. For example, generating a sensor image (504) of the position sensor may include plotting the acquired position sensor data readings as a delta function or a Gaussian function centered on the position sensor data readings.

[0072] Calculating 506 a radiation fluence map for delivery at firing location i may include convolving the position sensor image (e.g., a delta function) with a corresponding firing filter p generated during treatment planning using the sensor characterization PDF of that position sensor. For example, calculating a radiation fluence f for delivery at firing location i i,j To calculate δ, at time instance j, the radiation therapy system controller applies a firing filter corresponding to firing position i to the position sensor delta function image δ i,j The projection and convolution of f i,j =p i * δ i,j

[0073] Similarly, in variations in which the target sensor includes an imaging system, calculating 506 the radiation fluence map for delivery may include convolving the partial or LTS image with a corresponding emission filter p generated during treatment planning using a "complete" image generated using that imaging system (or the same imaging modality). For example, the radiation fluence f for delivery at emission location i is calculated as i,j To calculate , at time instance j, the controller of the radiation therapy system applies the firing filter corresponding to firing position i to the subimage or LTS image x to firing position i. i,j The projection and convolution of f i,j =p i * x i,j

[0074] delivered fluence f i,jare calculated in the static frame, i.e., the same reference frame as the radiation treatment system. During treatment planning, the emission filters p were calculated and optimized in the tumor POV frame. However, because the emission filters p are shift-invariant, they can be applied to target sensor data acquired in a static frame and still result in delivery of therapeutic radiation meeting the prescribed dose (e.g., dose target and objectives).

[0075] In variations in which the patient platform is parked at a series of predetermined, discrete platform positions (i.e., beam stations) during radiation delivery, method (500) may be repeated for each beam station. For example, a radiation treatment delivery method may include moving the patient platform to a first beam station, calculating a radiation fluence map for delivery based on target sensor data readings using a method described herein (e.g., method (500)), delivering radiation to a patient target area by emitting radiation from therapeutic radiation from i firing positions, then moving the patient platform to a second beam station, and repeating the radiation fluence calculation and delivery as described herein. This may be repeated for all of the beam stations defined during radiation treatment planning. For example, a patient platform may have 200 beam stations, each approximately 2 mm from an adjacent beam station. The radiation treatment planning system may be configured to calculate i firing filters (one for each of i firing positions across all beam stations) and transfer these i firing filters to a radiation treatment system controller memory. During a treatment session, the radiation therapy system controller moves the patient platform to the first beam station b1, retrieves the emission filter from the controller memory, and maps the target sensor image (x) to the emission position i at the first beam station b1. i,b1 ) projection by the emission filter p i the radiation fluence map f for delivery at launch location i by convolvingi,b1 , and so on for all firing positions at the first beam station. In some variations (e.g., when the radiation therapy system comprises a 1D MLC), a radiation fluence map f for delivery i,b1 Calculating the slice of fluence corresponding to the MLC field of view at launch position i (i.e., the projection of the target sensor image (x) on launch position i at the first beam station b1) i,b1 ) and emission filter p i Then, a radiation fluence map f for delivery may be extracted. i,b1 can be segmented into MLC apertures, and then a fluence map can be delivered by a therapeutic radiation source (e.g., a linac pulse). When the therapeutic radiation source generates a radiation fluence map f for all of the firing positions i in the first beam station b1, i,b1 After delivering , the radiation therapy system controller can then move the patient platform to a second beam station b2 and perform similar calculations and radiation delivery, repeating this for each of the firing positions at the 200 beam stations.

[0076] The method (500) may also be used in a non-therapeutic manner, for example, in a QA session in which the patient is replaced with a phantom and / or fluence measurement device. Optionally, the phantom and / or fluence measurement device may be mounted on a motion stage, which may be a mechanical device configured to move the phantom and / or fluence measurement device according to a motion trajectory that simulates the movement of the patient and / or patient target region. The movement may, in some variations, be based on a motion dwell histogram for the target region. The phantom and / or fluence measurement device may be set up with a target sensor and / or target sensor system as described above with respect to the patient. As applied to a non-therapeutic radiation delivery session (e.g., a QA session), the method (500) may include acquiring target sensor data during the QA session (502), generating a sensor image based on the acquired target sensor data (504), calculating a radiation fluence map for delivery by convolving the generated sensor image with a shift-invariant emission filter (506), and delivering radiation to a target region (e.g., a phantom target region) according to the calculated radiation fluence map (508).

[0077] Radiation treatment planning methods: more than two sensors One variation of a radiation treatment planning method incorporating target sensor data from multiple target sensors (i.e., two or more target sensors) is illustrated in FIG. 6. The method (600) may include obtaining a target sensor characterization PDF for each of the multiple target sensors (602), generating a target sensor characterization image for each of the target sensors based on the sensor's characterization PDF (604), and calculating a shift-invariant emission filter for each of the target sensors based on the respective sensor characterization image (606) to meet prescribed dose criteria for the patient target region. The number and type of target sensors in the radiation treatment plan may correspond to the number and type of target sensors used during a treatment session. The target sensors may all be of one sensing modality or a mix of different sensing modalities. For example, the multiple target sensors may comprise imaging systems of different imaging modalities (e.g., a CT imaging system and a PET imaging system, a PET imaging system and a fluoroscopy system, a PET imaging system and an optical camera, an MR imaging system and an optical camera, etc.), or multiple imaging systems of the same imaging modality (e.g., a first CT imaging system and a second CT imaging system). The multiple target sensors may include multiple position sensors, some of which may be coupled to the patient at an anatomical structure whose location and / or movement may be correlated with the location and / or movement of the patient target region and / or OAR. The multiple target sensors may include one or more imaging systems and one or more position sensors.

[0078] In some variations, the target sensor characterization PDFs for multiple target sensors may be provided by previous models and / or calculations, in which case obtaining the sensor characterization PDFs for the target sensors (602) may be optional. As previously described, the sensor characterization PDFs for each of the multiple target sensors may be derived from multiple target sensor data. An image of the sensor characterization PDF may be the sum of images of multiple target sensor characterization data. In an embodiment where there are two target sensors, an image of the sensor characterization PDF for each of the target sensors may be provided by:

[0079]

number

[0080] Generating 604 sensor characterization images may include generating a first image X of a first target sensor for each of the i firing locations and generating a second image Y of a second target sensor for each of the i firing locations. For example, to generate a radiation treatment plan for a radiation treatment system having 50 firing locations, generating 604 sensor characterization images for each of the target sensors may include generating a series of sensor characterization images X, which are projections of image X onto each of the i firing locations for the first target sensor. i and generating a series of sensor characterization images Y, which are projections of image Y onto each of the i firing positions for the second target sensor. i and generating:

[0081] Computing 606 a shift-invariant emission filter for each of the target sensors may include computing an emission filter for each of the target sensors for each emission location. That is, each target sensor has its own set of i emission filters. For example, the radiation treatment planning system may compute a first set of shift-invariant emission filters p for a first target sensor. i (over i firing locations) and calculate the shift-invariant firing filter q for the second set of target sensors. iExtending the above-described emission filter calculation method, we can calculate a radiation fluence map F for delivery to a patient target area at emission location i: i can be expressed as follows: F i =p i * X i +q i * Y i

[0082] Radiation fluence map for delivery F i may be the sum of the firing filter and target sensor data convolution, assuming that both target sensors shift in concert with the patient target area. The above formulation, along with the optimization methods described below, can be extended to include an arbitrarily large number of target sensors.

[0083] The radiation fluence F (eg, a fluence map) over multiple firing locations (eg, i firing locations) may be represented by:

[0084]

number

[0085] For example, the radiation fluence for 50 firing locations may be:

[0086]

number

[0087] Extending the dose formula above, A is the dose calculation matrix and D is the cumulative dose over i firing locations (which may be specified by the clinician when defining the dose prescription and / or target), expressed as:

[0088]

number

[0089] The dose D may be constrained by one or more cost or penalty functions C(D,F), as previously described. To calculate 606 an emission filter (e.g., a shift-invariant emission filter) according to method 600, the radiation treatment planning system may set up an optimization problem to minimize the cost function C(D,F), given the dose calculation matrix A and the target sensor characterization images X and Y. Computing the emission filters p and q may involve iterating through different emission filter values ​​such that the cost function C(D,F) is minimized while still achieving the dose goals and objectives according to the dose prescription, expressed as follows:

[0090]

number

[0091]

number

[0092] By grouping the known quantities into a single matrix, the above formulation can be computationally efficient for the radiation treatment plan optimizer. For example, the matrix A GIGRTmay be calculated once at the beginning of the optimization process and need not be recalculated for each iteration of the fire filter values. The radiation treatment planning system may be configured to iterate the fire filter values ​​p and q until one or more stopping conditions (such as any of the stopping conditions described above) are met. The final fire filter values ​​p and q may be stored in the memory of the radiation treatment planning system. In some variations in which the patient platform is stopped at predetermined discrete platform positions (i.e., beam stations) during delivery of radiation from multiple fire positions, the radiation treatment planning system may calculate fire filters across all beam stations. For example, the patient platform may have 200 beam stations, each beam station approximately 2 mm from an adjacent beam station. During radiation treatment planning, method (600) may be used to calculate two sets of i fire filters (p and q, each set having i fire filters, one for each of the i fire positions) that may be applied across the 200 beam stations.

[0093] Radiation therapy delivery methods: Two or more sensors 7 illustrates a variation of a radiation therapy delivery method that uses multiple sets of shift-invariant emission filters calculated during treatment planning (e.g., according to method 600) and target sensor data acquired from multiple target sensors (or target sensor systems) during a treatment session to calculate radiation fluence for delivery during that treatment session. Method 700 may include acquiring target sensor data from multiple target sensors during a treatment session (or any radiation delivery session) (702), generating a sensor image for each target sensor based on the target sensor data acquired for that sensor (704), convolving the generated sensor image with a respective emission filter for each target sensor (706), calculating a radiation fluence map for delivery by summing the convolutions for each target sensor (708), and delivering radiation to a patient target area according to the calculated radiation fluence map (710). The target sensor data may be acquired from any combination of target sensors, as described above. The number and / or type of target sensors and / or target sensor systems may be determined by a clinician. For example, the target sensor data may be acquired from a PET imaging system and one or more position sensors. The target sensor data may be acquired from an X-ray imaging system (e.g., a CT imaging system) or an MR imaging system and one or more position sensors. Acquiring the target sensor data (702) may include acquiring position sensor data readings from multiple target sensors at the beginning of a treatment session and / or throughout the treatment session while the therapeutic radiation source is delivering radiation to the patient. Acquiring the target sensor data (702) during a treatment session may include acquiring target sensor data from one or more of the multiple target sensors only at the beginning of the treatment session (i.e., before the therapeutic radiation source is activated for the first time) or may include acquiring target sensor data from one or more of the multiple target sensors throughout the treatment session.For example, acquiring target sensor data (702) may include acquiring a localization CT image and / or a PET pre-scan image (and / or any of the other aforementioned imaging modalities) at the beginning of a treatment session, and may also include acquiring CT imaging data and / or PET imaging data during a portion of the treatment session during which the therapeutic radiation source is delivering radiation to the patient. Acquiring target sensor data (702) may include acquiring a patient-localization CT image and / or a PET pre-scan at the beginning of a treatment session and acquiring position sensor data during the radiation delivery portion of the treatment session. In some variations, the imaging data acquired during the radiation delivery portion of the treatment session may be acquired at a short or limited time interval (e.g., about 2 seconds or less, about 1 second or less, about 0.5 seconds or less, etc.). For example, the imaging data acquired during the radiation delivery portion may be a subsample in k-space from a PET LOR, X-ray projection, and / or MRI imaging pulse sequence. The target sensor data may be acquired immediately prior to delivery of the therapeutic radiation beam. For example, the target sensor data may be acquired less than about 5 seconds (e.g., less than about 3 seconds, less than about 2 seconds, less than about 1 second, etc.) before radiation delivery. In some variations, the target sensor data may be acquired for a firing position (e.g., firing position i) while the therapeutic radiation source is located at a previous firing position (e.g., firing position i-1) and / or while the therapeutic radiation source is moving to the firing position (e.g., during movement between firing positions). Alternatively or additionally, the target sensor data may be acquired for a firing position (e.g., firing position i) when the therapeutic radiation source is located at that firing position.

[0094] In some variations, generating 704 the sensor image may include plotting the acquired sensor data readings as a delta function or a Gaussian function centered on the sensor data readings. For example, the sensor image x of the position sensor iGenerating 704 the sensor image y for the imaging system may include plotting the acquired position sensor data readings as a delta function or a Gaussian function centered on the position sensor data readings. i Generating (704) may include, for example, aggregating the acquired imaging data to generate a map of pixel and / or voxel intensity values. Because the imaging data may be acquired over a limited time interval, the resulting sensor image (or image map) may be referred to as a partially sampled image or a limited time sampling (LTS) image. LTS images may have a high level of noise (i.e., a high signal-to-noise ratio) and, as a result, may not provide sufficient information, when considered alone, to identify the contour and / or center of gravity of a patient target region or OAR. Alternatively or additionally, imaging data may be acquired at the beginning of a treatment session (e.g., a CT localization scan and / or a PET pre-scan). While the examples herein are in the context of treating a patient based on target sensor data from a position sensor (first target sensor) and an imaging system (second target sensor system), it should be understood that, as previously discussed, similar methods may be applied to any number and / or type of target sensors and combinations thereof.

[0095] Convolving the sensor image with its respective emission filter (706) may include retrieving from memory of the radiation therapy system controller an emission filter calculated during treatment planning for that particular target sensor and firing location i, calculating a projection of the sensor image onto firing location i at time instance j, and convolving the emission filter and the projected image. These steps may be repeated for each of the target sensors. That is, for the first target sensor, p i * x i,j Here p i is the firing filter for firing position i of the first target sensor, and x i,jis the projection of the first target sensor data at time instance j onto the firing position i. For example, in a variant where the first target sensor is a position sensor, the sensor image is calculated using the position sensor delta function δ j δ i,j may be the projection of the delta function image onto the firing position i. p i * δ i,j

[0096] For the second target sensor, it is expressed as: q i * y i,j where q i is the firing filter for firing position i of the second target sensor, and y i,j is the projection of the second target sensor data onto the firing location i at time instance j. For example, in a variation where the second target sensor system is an imaging system, the sensor image may be an LTS image, and the radiation therapy system controller may be configured to convolve a firing filter with the projection of the LTS image onto the firing location i, where the convolution of the firing filter with the projection of the image of the sensor data onto the firing location may be calculated for any number (e.g., all) of the target sensors used during the treatment session and for which corresponding firing filters were calculated during treatment planning.

[0097] Based on the target sensor data acquired at time instance j, a radiation fluence map f for delivery at launch location i is calculated. i,j Calculating (706) {circumflex over (x)} may include summing (704) the convolutions, and may be expressed as: f i,j =p i * x i,j +q i * y i,j

[0098] In an embodiment where the first target sensor is a position sensor and the position sensor image is a delta function, it can be expressed as: f i,j =p i * δ i,j +q i * y i,j

[0099] Similar to the delivered fluence calculation for a single target sensor, the delivered fluence f i,j are calculated in the static frame, i.e., the same reference frame as the radiation treatment system. During treatment planning, the emission filters p and q were calculated and optimized in the tumor POV frame. However, because the emission filters p are shift-invariant, they can be applied to target sensor data acquired in the static frame and still result in delivery of therapeutic radiation meeting the prescribed dose (e.g., dose target and objective) in the tumor POV frame.

[0100] In variations in which the patient platform is parked at a series of predetermined, discrete platform positions (i.e., beam stations) during radiation delivery, method (700) may be repeated for each beam station. For example, a radiation therapy delivery method may include moving the patient platform to a first beam station, calculating a radiation fluence map for delivery based on target sensor data readings using a method described herein (e.g., methods (500), (700)), delivering radiation to a patient target area by emitting radiation from therapeutic radiation from i firing positions, then moving the patient platform to a second beam station, and repeating the radiation fluence calculation and delivery as described herein. This may be repeated for all of the beam stations defined during radiation therapy planning. For example, the patient platform may have 200 beam stations, each approximately 2 mm from an adjacent beam station. In this example, the therapeutic radiation source may be configured to emit radiation at 100 firing positions i=100. If two target sensors (and / or target sensor systems) are used, the radiation treatment planning system may be configured to calculate two sets of fire filters (each set having 100 fire filters, one for each of i=100 fire locations): a first set of 100 fire filters for the first target sensor, and a second set of 100 fire filters for the second target sensor. The planning system may transfer these two sets of fire filters to the radiation treatment system controller memory. During a treatment session, the radiation treatment system controller may move the patient platform to the first beam station, retrieve the fire filters from the controller memory, and calculate a radiation fluence map for delivery at fire location i by summing the convolution of the first target sensor image with its corresponding fire filter and the convolution of the second target sensor image with its corresponding fire filter.In some variations (e.g., when the radiation therapy system includes a 1D MLC), calculating the radiation fluence map for delivery may further include extracting a slice of fluence produced by convolving a first target sensor image with its corresponding emission filter corresponding to the MLC field of view at the firing position, extracting a slice of fluence produced by convolving a second target sensor image with its corresponding emission filter corresponding to the MLC field of view at the (same) firing position, and then summing the fluence slices to obtain the radiation fluence map for delivery. After the therapeutic radiation source has delivered the radiation fluence maps for all i firing positions at the first beam station, the radiation therapy system controller may then move the patient platform to the second beam station and perform similar calculations and radiation delivery, which may be repeated for each of the firing positions at the 200 beam stations.

[0101] The method (700) may also be used in a non-therapeutic manner, for example, in a QA session in which the patient is replaced with a phantom and / or fluence measurement device. Optionally, the phantom and / or fluence measurement device may be mounted on a motion stage, which may be a mechanical device configured to move the phantom and / or fluence measurement device according to a motion trajectory that simulates the movement of the patient and / or patient target region. The movement may, in some variations, be based on a motion dwell histogram for the target region. The phantom and / or fluence measurement device may be set up with a target sensor and / or target sensor system as described above with respect to the patient. As applied to a non-therapeutic radiation delivery session (e.g., a QA session), the method (700) may include acquiring target sensor data from multiple target sensors during the QA session (702), generating a sensor image for each target sensor based on the acquired target sensor data for that sensor (704), convolving the sensor image with its respective emission filter for each target sensor (706), calculating a radiation fluence map for delivery by summing the convolutions for each target sensor (708), and delivering radiation to a target region (e.g., a phantom target region) according to the calculated radiation fluence map (710).

[0102] Radiation treatment planning method: constant value sensor Another variation of the radiation treatment planning method may incorporate data from a target sensor, where the sensor data readings are constant values ​​in a static frame (which is also the same frame of reference as the radiation treatment system and the patient). Because the sensor data readings are constant values, such sensors may be referred to as “null sensors” or constant value sensors. A constant value sensor may be a sensor whose sensor readings do not change, i.e., the sensor data readings return the same value or readout. In some variations, the constant value sensor may be a target sensor whose sensor output is read only once during a radiation delivery session, for example, only once during a treatment session (or only once per patient localization, which in some variations may occur more than once during a treatment session). One example of a constant value sensor may be a position sensor from which the radiation treatment system reads the position sensor data readings once during a radiation treatment session. In some variations, the constant value sensor system may comprise a CT imaging system and / or a PET imaging system that acquires imaging data during a patient localization phase at the beginning of a treatment session, but not during the radiation delivery phase of the treatment session. The CT localization images and / or PET pre-scan images may be used to identify an initial position of the patient target region centroid and / or an initial boundary of the patient target region. In some variations, the CT localization images and / or PET pre-scan images may be used to identify the location (in the static frame) of the center of the patient target region motion envelope (e.g., the tumor motion envelope or the center of the internal target volume or ITV).

[0103] 8A-8D illustrate conceptual diagrams illustrating the physical meaning of a constant value sensor in the tumor POV frame and the static frame, using a constant value target sensor that outputs an initial location of the tumor centroid in the static frame. The location of the tumor centroid in the static frame can be determined using any appropriate sensor modality, including, but not limited to, one or more of CT imaging, PET imaging, MR imaging, optical imaging using visible or infrared light, position sensing (e.g., using sensors attached to the tumor and / or nearby landmarks, X-ray tracking of reference points attached to the tumor and / or nearby landmarks, etc.), etc. In this example, the constant value target sensor can be a position sensor that outputs an initial portion of the target region (e.g., tumor) centroid in the static frame. Alternatively or additionally, the constant value target sensor can be a target sensor system that includes an imaging system, and the location of the tumor centroid can be calculated based on imaging data acquired by the imaging system. FIG. 8A illustrates the tumor (800) and bone (802) in a tumor POV frame, with the origin (0,0,0) at the tumor (800) center of mass, while FIG. 8B illustrates the tumor (800) and bone (802) in a static frame, with the origin (0,0,0) at the initial location of the tumor center of mass. FIG. 8C illustrates a close-up of the tumor (800) and ITV (801) in the tumor POV frame. Alternatively or additionally, the origin (0,0,0) of the tumor POV frame may be the center of mass of the ITV. The boundaries of the ITV may be defined in a static frame and may encompass the volume through which the tumor moves (e.g., a motion envelope). As shown in FIG. 8B, the constant value sensor reading (804) does not change. This remains the same regardless of tumor (800) movement (i.e., position change) and, in this example, represents the location of the tumor center of mass. However, the tumor (800) is moving relative to the constant value sensor reading (804) in the tumor POV frame where the tumor is stationary, causing the sensor reading (804) to appear to be changing, i.e., not constant.8A and 8C conceptually illustrate a stationary tumor (800) in a tumor POV frame, during which sensor readings (reflecting values ​​in the stationary frame) appear to vary across different values ​​(e.g., 804a-804c). A constant-value sensor would produce a constant reading (804) in the stationary frame regardless of actual tumor location, but in the tumor POV frame, the values ​​of the sensor readings (804a-804c) may reflect the offset of that constant sensor reading relative to the actual location of the tumor centroid. Thus, in the tumor POV frame, the sensor readings may be inverted (or negated) tumor offsets from the tumor centroid and / or ITV center. The tumor offset from the centroid may be derived from a tumor location histogram (e.g., a tumor retention matrix), which may be generated using imaging data from a set of, for example, 4D CT images, 4D PET data, 4D MR data, a patient surface sensor coupled with a motion model, or a dual-projection X-ray system coupled with a motion model. FIG. 8D is a plot of a negated tumor position histogram in the tumor POV frame. The plot in FIG. 8D can be an image of a constant-value sensor characterization PDF (806) in the tumor POV frame. That is, tumor (800) motion in the static frame can be represented in the tumor POV frame as a sensor characterization PDF of a constant-value target sensor. Tumor motion can be recast in the tumor POV frame as error, variability, or noise of the constant-value target sensor. As such, a radiation treatment planning system can be configured to calculate an emission filter using the sensor characterization PDF (806) in the tumor POV frame. In some variations, the range or span of the sensor characterization PDF (806) in the tumor POV frame can be less than the range or span of the ITV (801) defined in the static frame. While other treatment planning methods and systems optimize fluence maps and / or emission filters based on the ITV as defined in the static frame, the methods described herein optimize fluence maps and / or emission filters based on a sensor characterization PDF in tumor POV space.Calculating the emission filter based on the sensor characterization PDF (806) in the tumor POV frame instead of the ITV (801) in the static frame can help reduce irradiation to healthy tissue surrounding the tumor (800).

[0104] The sensor characterization PDF of the constant-value target sensor may be used to calculate an emission filter according to any of the methods described herein. Figure 9 illustrates an example of a radiation treatment planning method that calculates an emission filter using the sensor characterization PDF of a constant-value target sensor. The constant-value target sensor may be any sensor or sensor system that outputs the same target region (e.g., tumor) centroid location in a static frame. In some variations, the constant-value target sensor reading may be a position sensor data reading, and in other variations, the constant-value target sensor reading may be a tumor centroid location determined based on a treatment planning image. The method (900) may include determining (902) a location of a center of mass of a tumor (or any patient target region); generating (904) a sensor characterization PDF using an offset of the tumor center of mass location associated with changes in tumor position over time; generating (906) a sensor characterization image based on the sensor characterization PDF for each of the i firing locations; and calculating (908) a shift-invariant firing filter for each of the i firing locations based on the sensor characterization image to meet prescribed dose criteria for the tumor (or patient target region). Determining (902) the tumor center of mass and generating (904) the sensor characterization PDF may include obtaining tumor position data readings over time. In some variations, this may involve obtaining multiple CT images over time (e.g., 4D CT images). CT imaging), determining an initial location of the tumor centroid, and then calculating the change in centroid location over time. Alternatively, or additionally, obtaining tumor position data readings over time may include attaching a position sensor or trackable fiducial to the tumor, determining an initial location of the tumor centroid, and then measuring the change in centroid location over time. Generating a sensor characterization image (906) may include converting the spatial changes to a delta function or Gaussian function, as described above, and aggregating the delta function or Gaussian function to obtain an image of the sensor characterization PDF (i.e., sensor characterization PDF image N). In some variations, the image of spatial changes may include a tumor retention matrix (e.g., a tumor position histogram), which may be negated to obtain the image of the sensor characterization PDF ... to obtain an image of the sensor characterization PDF (i.e., sensor characterization PDF image N). i , which may include computing a projection of the sensor characterization PDF N onto each of the i firing positions to obtain a shift-invariant firing filter (p i Calculating (908) the dose at the tumor POV can be similar to the calculation methods described above for methods (200, 600). That is, the dose at the tumor POV can be expressed as:

[0105]

number

[0106] The radiation treatment planning system may be configured to iterate through the emission filter values ​​until one or more stopping conditions are met. The final emission filter values ​​may be stored in a memory of the radiation treatment planning system.

[0107] Radiation therapy delivery method: constant value sensor 10 illustrates one variation of a radiation therapy delivery method that uses a shift-invariant emission filter calculated during treatment planning (e.g., according to methods (200, 600, 900)) and target sensor data acquired from a constant-value sensor (or sensor system) during a treatment session to calculate a radiation fluence for delivery during that treatment session. The method (1000) may include defining (1002) a location of a center of gravity of a patient target area using target sensor data acquired during a treatment session (or any radiation delivery session), generating (1004) an image based on the sensor data, calculating (1006) a radiation fluence map for delivery by convolving the generated image with the emission filter, and delivering (1008) radiation according to the calculated radiation fluence map. Defining 1002 the location of the patient target area centroid and generating 1004 the sensor image may include acquiring CT imaging data and / or PET imaging data to form one or more images that can be used to identify the location (e.g., coordinates) of the patient target area centroid. Because the target sensor is a "null" or constant value sensor, the sensor data read throughout the treatment session represents the location of the patient target area centroid as initially identified during the patient localization phase of the treatment session. Because the sensor data readings (and therefore the initial location of the centroid) may be at the origin of a static frame (which is also the same frame of reference as the radiation treatment system), the image of this constant value sensor reading may be a delta function centered at the origin. This is because the emission filter is an "identity" delta function δ identity The fluence map calculation is collapsed so that it is convolved with f i =p i * δ identity =p i

[0108] The fluence map delivered at a firing position i is simply the sum of the firing filter p iThat is, as long as the patient is positioned during a treatment session so that the tumor centroid location at treatment coincides with the tumor centroid location during treatment planning, a fixed, pre-calculated fluence map can be delivered to the tumor without additional target sensor data. This form of treatment delivery can be similar to standard IMRT / SBRT radiation delivery, but because treatment planning is performed in the tumor POV frame instead of a static frame, the radiation delivered to the patient can be reduced compared to standard IMRT / SBRT methods while still meeting the same dose goals and objectives. Standard IMRT / SBRT methods perform treatment planning based on ITV boundaries defined in a static frame, which may encompass a larger region or volume than the sensor-characterized PDF. A fluence map and / or emission filter optimized based on a relatively large ITV can result in higher levels of irradiation to surrounding healthy tissue than when the fluence map and / or emission filter are optimized based on a sensor-characterized PDF.

[0109] While the above examples describe the constant value target sensor as a target sensor system including a CT imaging system and a PET imaging system, it should be understood that in other variations, the target sensor system may include one or more of any of the imaging systems previously described herein, alone or in combination with each other. For example, instead of a CT imaging system that may be used in conjunction with a PET imaging system, the target sensor system may include an MR imaging system, alone or in conjunction with a CT imaging system. In some variations, the constant value sensor may be a position sensor (such as any of the position sensors described above).

[0110] The method (1000) may also be used in a non-therapeutic manner, for example, in a QA session in which the patient is replaced with a phantom and / or fluence measurement device. Optionally, the phantom and / or fluence measurement device may be mounted on a motion stage, which may be a mechanical device configured to move the phantom and / or fluence measurement device according to a motion trajectory that simulates the movement of the patient and / or patient target region. The movement may, in some variations, be based on a motion dwell histogram for the target region. The phantom and / or fluence measurement device may be set up with a target sensor and / or target sensor system as described above with respect to the patient. As applied to a non-therapeutic radiation delivery session (e.g., a QA session), the method (1000) may include defining (1002) a location of the center of gravity of a target region (e.g., a phantom target region) using target sensor data acquired during the QA session, generating (1004) an image based on the acquired target sensor data, calculating (1006) a radiation fluence map for delivery by convolving the generated image with an emission filter, and delivering (708) radiation to the target region (e.g., the phantom target region) according to the calculated radiation fluence map.

[0111] 11A-11C illustrate simulation results of one example of radiation treatment planning and delivery to a 12 mm long patient target region (e.g., clinical target volume or CTV) moving linearly (1D) at ±6 mm. FIG. 11A illustrates a motion retention histogram of the patient target region. Based on this motion profile, in a conventional treatment planning method for IMRT / SBRT, the ITV can be defined in a static frame with a size of 24 mm (12 mm tumor length + 6 mm displacement in one direction + 6 mm displacement in the other direction). The fluence delivered based on a standard ITV-based treatment planning method can have the profile illustrated in the plot of FIG. 11B. Trace 1100 represents the delivered fluence in the static frame, and trace 1102 represents the delivered fluence in the tumor POV frame. The delivered fluence in the tumor POV frame is somewhat blurred compared to the delivered fluence in the static frame, but still covers the entire extent of the 12 mm patient target region (or CTV). Figure 11C illustrates a profile of delivered fluence based on the treatment planning method disclosed herein (i.e., based on the sensor characterization PDF in the tumor POV frame). This is based on the constant value target sensor planning and delivery method described in Figures 9 and 10. Trace (1104) represents the delivered fluence in the static frame, and trace (1106) represents the delivered fluence in the tumor POV frame. The delivered fluence in the static frame has fluence peaks at the edges of the patient target region (or CTV), resulting in an overall irregular horn-shaped profile, but still covers the entire extent of the 12 mm patient target region (or CTV). While both methods deliver the prescribed dose to the entire patient target area, the difference between the fluence profiles in Figures 11B and 11C is that treatment planning in the tumor POV frame using the methods described herein results in an overall 35% reduction in radiation exposure to the patient. In certain embodiments, for example, the following are provided: (Item 1) 1. A radiation therapy system comprising: a patient platform; a therapeutic radiation source movable to one or more firing positions about said patient platform; a target sensor system including a target sensor for acquiring sensor data; a controller in communication with the therapeutic radiation source and the target sensor system, the controller configured to calculate a radiation fluence map for delivery to a target area by convolving an image generated from sensor data with a shift-invariant emission filter derived from a sensor characterization probability density function (PDF) of the target sensor, and the controller configured to deliver radiation according to the calculated radiation fluence map. (Item 2) Item 1, wherein the target sensor is a first target sensor and the target sensor system includes a second target sensor. (Item 3) 3. The system of claim 2, wherein at least one of the first target sensor and the second target sensor is a position sensor configured to be coupled to a patient disposed on the patient platform. (Item 4) Item 4. The system of item 3, wherein the position sensor is configured to be coupled to a target area. (Item 5) Item 4. The system of item 3, wherein the position sensor comprises an X-ray source and an X-ray detector disposed opposite the X-ray source and configured to detect the position of an embedded reference point. (Item 6) Item 4. The system of item 3, wherein the position sensor comprises an optical imaging system configured to track optical reference points attached to the patient's skin, and the target sensor system further comprises an optical camera configured to detect the position of the optical reference points. (Item 7) Item 10. The system of item 1, wherein the sensor characterization PDF is a sensor error characterization PDF that represents a rate of sensor data error. (Item 8) Item 10. The system of item 1, wherein the sensor characterization PDF represents a rate of sensor data variation. (Item 9) Item 10. The system of item 1, wherein the sensor characterization PDF includes one or more of a 1D plot of sensor data, a 2D plot of sensor data, and / or a 3D plot of sensor data, and a histogram representing sensor data variability. (Item 10) Item 10. The system of item 1, wherein the image generated from the position sensor data is a delta function, a Gaussian function, and / or a truncated Gaussian function centered at a location corresponding to the position sensor data. (Item 11) Item 10. The system of item 1, wherein the target sensor includes one or more image sensors, the sensor data includes imaging data, and the sensor characterization PDF of the target sensor includes an image. (Item 12) Item 12. The system of item 11, wherein the one or more image sensors include an image sensor selected from the group consisting of a PET sensor, an MRI sensor, and a CT sensor. (Item 13) 3. The system of claim 2, wherein the controller is configured to receive first sensor data readings from the first target sensor and second sensor data readings from the second target sensor, the shift-invariant emission filter is a first shift-invariant emission filter of the first target sensor, and the sensor characterization PDF is a first sensor characterization PDF of the first target sensor, and the controller is further configured to calculate the fluence map for delivery by summing (a) the convolution of the first sensor data image with the first shift-invariant emission filter and (b) a convolution of a second image generated from the second sensor data with a second shift-invariant emission filter derived from a second sensor characterization PDF of the second target sensor. (Item 14) Item 14. The system of item 13, wherein the first target sensor is a sensor of a first type and the second target sensor is a sensor of a second type different from the first type. (Item 15) Item 15. The system of item 14, wherein the second target sensor is a position sensor. (Item 16) Item 15. The system of item 14, wherein the first target sensor is a positron annihilation emission pathway sensor and the second target sensor is a target area position sensor. (Item 17) Item 15. The system of item 14, wherein the first target sensor is an image sensor and the second target sensor is a position sensor. (Item 18) Item 15. The system of item 14, wherein the first target sensor includes at least one of a 3D PET sensor, a 2D X-ray sensor, a projection image sensor, a fluoroscopy image sensor, a CT image sensor, and an MR sensor, and the second target sensor includes a position sensor. (Item 19) 1. A method for radiation delivery, said method comprising: obtaining sensor data readings from the target sensors; generating a sensor image from the sensor data readings; calculating a radiation fluence map for delivery to a target region by convolving the sensor image with a shift-invariant emission filter derived from a sensor characterization probability density function (PDF) of the target sensor; delivering radiation to the target area according to the calculated radiation fluence map. (Item 20) 20. The method of claim 19, wherein the sensor characterization PDF is a sensor error characterization PDF that represents a rate of sensor data error. (Item 21) 20. The method of claim 19, wherein the sensor characterization PDF represents a rate of sensor data variation. (Item 22) 22. The method of claim 21, wherein the sensor characterization PDF includes one or more of a 1D plot of sensor data, a 2D plot of sensor data, and / or a 3D plot of sensor data, and a histogram representing sensor data variability. (Item 23) 23. The method according to any one of items 19 to 22, wherein the target sensor is a position sensor. (Item 24) Item 24. The method of item 23, wherein the sensor image is a delta function, a Gaussian function, and / or a truncated Gaussian function centered at a location corresponding to the position sensor data reading. (Item 25) Item 25. The method of item 24, wherein the position sensor includes a target area position sensor. (Item 26) Item 25. The method of item 24, wherein the position sensor includes an X-ray projector system configured to track an embedded reference point. (Item 27) 25. The method of claim 24, further comprising attaching optical fiducials to the patient's skin and tracking the optical fiducials using an optical imaging system. (Item 28) 20. The method of claim 19, wherein the target sensor is a null position sensor whose sensor data reading is a constant position value representing the center of gravity of the target area, and wherein the sensor characterization PDF of the null position sensor includes multiple position values ​​representing the location of the center of gravity of the target area over time. (Item 29) Item 29. The method of item 28, wherein the sensor image is a delta function, a Gaussian function, and / or a truncated Gaussian function centered at a location corresponding to the position sensor data reading. (Item 30) 29. The method of claim 28, wherein the sensor characterization PDF includes a motion retention histogram of the target region. (Item 31) 29. The method of claim 28, wherein the plurality of position values ​​are determined using 4D CT imaging data. (Item 32) 20. The method of claim 19, wherein the shift-invariant emission filter corresponds to a launch position of a therapeutic radiation source, and wherein calculating the fluence for delivery includes calculating the fluence for delivery at the launch position by convolving a projection of the sensor image on the launch position with the shift-invariant emission filter of the launch position. (Item 33) 20. The method of claim 19, wherein the target sensor includes one or more image sensors, the sensor data readings include imaging data, and the sensor characterization PDF of the target sensor includes an image generated from the imaging data. (Item 34) Item 34. The method of item 33, wherein the one or more image sensors include an image sensor selected from the group consisting of a PET sensor, an MRI sensor, and a CT sensor. (Item 35) the target sensor is a first target sensor, the sensor data readings are first sensor data readings, the image is a first sensor data image, the shift-invariant emission filter is a first shift-invariant emission filter, and the sensor characterization PDF is a first sensor characterization PDF; and the method further comprises: obtaining a second sensor data reading from a second target sensor; generating a second sensor data image from the second sensor data readings; 20. The method of claim 19, wherein calculating the fluence map for delivery includes summing (a) the convolution of the first sensor data image with the first shift-invariant emission filter and (b) a convolution of the second sensor data image with a second shift-invariant emission filter derived from a second sensor characterization PDF of the second target sensor. (Item 36) 36. The method of claim 35, wherein the first target sensor data reading includes a first type of data and the second target sensor data reading includes a second type of data that is different from the first type of data. (Item 37) Item 37. The method of item 35 or 36, wherein the second sensor characterization PDF is a sensor error characterization PDF. (Item 38) Item 37. The method of item 36, wherein the second target sensor is a position sensor. (Item 39) 37. The method of claim 36, wherein the first target sensor data reading includes positron annihilation emission path data and the second target sensor data reading includes target region location data. (Item 40) 37. The method of claim 36, wherein the first target sensor data reading includes partial imaging data and the second target sensor data reading includes target area location data. (Item 41) 37. The method of claim 36, wherein the first target sensor data readings include at least one of 3D PET imaging data, 2D X-ray imaging data, projection imaging data, fluoroscopy imaging data, CT imaging data, and MR imaging data, and the second target sensor data readings include target region location data. (Item 42) 20. The method of claim 19, wherein the shift-invariant emission filter corresponds to a launch position of a therapeutic radiation source, and wherein calculating the fluence for delivery includes calculating the fluence for delivery at the launch position by projecting the sensor data readings onto the launch position, generating a second sensor image of the projected sensor data readings, and convolving the second sensor image with the shift-invariant emission filter of the launch position. (Item 43) 1. A method for sensor-based treatment planning, the method comprising: A sensor characterization image N is generated based on the sensor characterization probability density function (PDF) of the target sensor for each of the i firing locations. i and conditions

number

number

Claims

[Claim 1] The invention described in the specification.

Images