Methods for real-time target and organs-at-risk monitoring for radiotherapy

Abstract

Described herein are variations of methods for monitoring and / or tracking a region of interest within a patient a radiotherapy session. For example, a method for monitoring a region of interest may include comparing an image of the target region acquired during real-time delivery of radiation to a reference image of same target region acquired before treatment delivery to determine a difference between the two images. If the treatment image is determined to be unlike the reference image, radiation delivery may be stopped so that radiation is not delivered to healthy tissue as a result of anatomical shifting relative to the target region. Also described herein are systems which may be used to implement such methods.

Description

METHODS FOR REAL-TIME TARGET AND ORGANS- AT-RISK MONITORINGFOR RADIOTHERAPYCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U.S. Provisional Patent Application Serial No. 63 / 493,488 filed March 31, 2023, the disclosure of which is hereby incorporated by reference in its entirety.TECHNICAL FIELD

[0002] The disclosures herein relate generally to systems and methods for real-time monitoring of a target region of interest and / or organ-at-risk during a radiotherapy session and control of treatment delivery. The methods may be used for emission-guided high-energy photon delivery, such as biology-guided radiation therapy.BACKGROUND

[0003] Radiotherapy, or radiation therapy, uses high-energy photons or other particles to treat a variety of diseases. For example, radiotherapy is commonly used to treat cancerous tumors. During treatment, a gantry having a radiation source mounted thereon, may move around a patient and emit radiation beams to a region of interest within the patient from several positions. Accurately tracking the region of interest (e.g., tumor position) is an important factor when maximizing radiation dose to the target region and limiting radiation exposure to a patient’s normal tissue. Additionally, accurately tracking organs near the target region (i.e., organs-at- risk), which are particularly vulnerable to radiotherapy-associated toxicity, can be used to help inform how to limit radiation exposure to the patient’s healthy tissue. Some radiotherapy systems use image guidance to track a region of interest and facilitate delivery of radiation to the region. However, current image-guided radiotherapy systems and methods require high-quality and / or complete images of the region of interest to guide treatment delivery. Such images take time to acquire and reconstruct and do not typically characterize the region of interest in realtime. Additionally, current radiotherapy systems do not currently provide methods to track the region of interest in real time within a treatment zone (e.g., due to anatomical shifting) and stop the treatment if the region of interest is no longer within a treatment zone. Accordingly, there is a need for new and improved methods of real-time monitoring of radiotherapy regions of interestto reduce the risk of radiation toxicity for a patient by monitoring and responding to real-time changes of the region of interest.BRIEF SUMMARY

[0004] In some variations, a method for real-time radiotherapy treatment includes generating a set of pre-scan images of a region of interest (ROI) within a patient, acquiring a set of real-time images of the ROI, comparing the set of real-time images to the set of pre-scan images, and terminating a radiotherapy treatment session if the set of real-time images deviates from the set of pre-scan images by a predetermined threshold. In some variations, the method includes generating a graphical or other notification of a deviation status of the set of real-time images from the set of pre-scan images. In some variations, acquiring the set of real-time images includes acquiring one or more subsets of real-time images at a control point. In some variations, the method includes delivering radiotherapy treatment to the patient from a control point if the set of real-time images does not deviate from the set of pre-scan images by the predetermined threshold. In some variations, the method is executed by a first module (e.g., controller or processor) within a radiotherapy treatment system and terminating the radiotherapy treatment session includes transmitting instructions from the first module to a second module (e.g., controller or processor) configured to control hardware of the radiotherapy treatment system. In some variations, the method includes generating one or more of a visual notification, an audio notification, or a haptic notification indicative of a deviation status of the set of real-time images from the set of pre-scan images. In some variations, terminating the radiotherapy treatment session includes automatically saving a record of radiotherapy treatment delivered to a patient and generating an updated set of pre-scan images comprising the ROI. In some variations, each of the set of pre-scan images and the set of real-time images includes 3D images of at least a portion of the ROI. In some variations, the method includes defining the predetermined threshold based on the set of pre-scan images, and the predetermined threshold includes one or more predetermined thresholds. In some variations, each of the set of pre-scan images and the set of real-time images includes one of: PET images, SPECT images, CT images, MRI images X-ray images, ultrasound images, or combinations thereof. In some variations, generating the set of pre-scan images includes acquiring a subset of pre-scan images at each of a plurality of control points and combining the subsets of pre-scan images. In some variations, the method includes combining a plurality of subsets of pre-scan images corresponding to the plurality ofcontrol points to generate a summed pre-scan image of the ROI. In some variations, each of the plurality of control points includes a beam station within a radiotherapy treatment system which may be a patient platform position selected for radiotherapy delivery to a portion of the ROI within the patient. In some variations, the ROI includes an inner ROI within an outer ROI. In some variations, the method includes identifying the inner ROI prior to the real-time radiotherapy treatment and localizing the outer ROI prior to the real-time radiotherapy treatment.

[0005] In some variations, the method includes calculating a pre-scan image metric and a realtime image metric, and comparing the set of real-time images of the ROI to the set of pre-scan images of the ROI includes comparing the pre-scan image metric and the real-time image metric. In some variations, each of the pre-scan image metric and the real-time image metric includes two or more image metrics.

[0006] In some variations, the set of pre-scan images includes a summed pre-scan image of a plurality of subsets of pre-scan images, for example images taken at each control point or beam station, where calculating the pre-scan image metric includes calculating a self-cross correlation between each subset of the plurality of subsets of pre-scan images and the summed pre-scan image, and where calculating the real-time image metric includes calculating a local cross correlation between the set of real-time images and the summed pre-scan image.

[0007] In some variations, each of the real-time image metric and the pre-scan image metric includes a rate of change of a maximum signal intensity. In some variations, the set of pre-scan images includes a summed pre-scan image having a plurality of subsets of pre-scan images, and the method includes: calculating the rate of change of the maximum signal intensity for the plurality of subsets of pre-scan images to determine the pre-scan image metric, generating a hybrid image of the ROI by substituting the set of real-time images for a corresponding subset of pre-scan images within the summed pre-scan image, and calculating the rate of change of the maximum signal intensity for the hybrid image to determine the real-time image metric. In some variations, the ROI includes an inner ROI within an outer ROI, and the maximum signal intensity includes a maximum signal intensity within the inner ROI. In some variations, the rate of change of the maximum signal intensity is indicative of movement of the inner ROI with respect to the outer ROI.

[0008] In some variations, each of the real-time image metric and the pre-scan image metric includes a rate of change of a standard deviation of an image signal. In some variations, the set of pre-scan images includes a summed pre-scan image comprising plurality of subsets of prescan images, and the method includes: calculating the rate of change of the standard deviation of the signal for each of the plurality of subsets of pre-scan images to determine the pre-scan image metric, generating a hybrid image comprising the ROI by substituting the set of real-time images for a corresponding subset of pre-scan images within the summed pre-scan image, and calculating the standard deviation of the signal for the hybrid image to determine the real-time image metric. In some variations, the pre-scan image metric comprises a maximum rate of change of the standard deviation of the signal intensity among the plurality of subsets of prescan images. In some variations, the ROI includes an inner ROI within an outer ROI, and calculating the standard deviation of the signal includes calculating the standard deviation of the signal within the outer ROI, excluding the signal within the inner ROI. In some variations, the rate of change of the standard deviation of the signal is indicative of movement of anatomical object into the outer ROI if the set of real-time images deviates from the set of pre-scan images by the predetermined threshold.

[0009] In some variations, the ROI includes a volumetric shell comprising an inner ROI within an outer ROI, and each of the real-time image metric and the pre-scan image metric include a mean intensity of the ROI. In some variations, the method may include adjusting a signal intensity of the inner ROI, calculating the mean intensity of the ROI using the adjusted signal of the inner ROI to determine the pre-scan image metric, and calculating the mean intensity of the ROI for the set of real-time images to determine the real-time image metric. In some variations, adjusting the signal intensity of the inner ROI may include duplicating a signal of the inner ROI and shifting the duplicated signal to a plurality of positions within the outer ROI, and calculating the pre-scan image metric includes calculating the mean intensity of the ROI at each of the plurality of positions. In some variations, the duplicated signal of the inner ROI may be used to simulate a presence of an anatomical object within the ROI. In some variations, calculating the mean intensity of the ROI for the set of real-time images includes applying a weight factor to a calculation of the mean intensity of the ROI.

[0010] In some variations, a method for monitoring an ROI within a patient during a radiotherapy treatment session includes acquiring a plurality of subsets of pre-scan imagescomprising at least a portion of the ROI at a plurality of corresponding beam stations and combining a plurality of subsets of the pre-scan images corresponding to the plurality of beam stations to generate a summed pre-scan image of the ROI. For each of the plurality of beam stations, the method includes combining one or more sets of sequential pre-scan images to generate a subset pre-scan image of the summed pre-scan image and calculating a self-cross correlation between the subset image corresponding to the beam station and the summed prescan image. The method further includes acquiring a set of real-time images at the beam station, combining the set of real-time images with a remainder of one or more sets of sequential realtime images to generate a subset real-time image of the ROI, where the one or more sets of sequential real-time images correspond to the one or more sets of sequential pre-scan images for the beam station, calculating a local cross correlation between the subset real-time image of the ROI and the summed pre-scan image of the ROI, comparing the local cross correlation to the self-cross correlation to determine if the local cross correlation is within an acceptable range, generating a graphical notification to indicate whether the local cross correlation is within the acceptable range, and optionally modifying a radiation treatment for the patient if the local cross correlation is not within the acceptable range.

[0011] In some variations, a method for monitoring an ROI within a patient during a radiotherapy treatment session includes acquiring a set of pre-scan images of at least a portion of the ROI at a plurality of corresponding beam stations and combining a plurality of subsets of pre-scan images corresponding to the plurality of beam stations to generate a summed pre-scan image of the ROI. For each of the plurality of beam stations, the method further includes combining one or more sets of sequential pre-scan images to generate a subset pre-scan image of the summed pre-scan image, calculating a rate of change of a maximum pre-scan image signal for the summed pre-scan image of the ROI using a plurality of subset pre-scan images corresponding to the plurality of beam stations, acquiring a set of real-time images comprising at least a portion of the ROI, calculating a maximum real-time image signal for the set of real-time images, generating a hybrid image of the ROI by substituting the set of real-time images for a corresponding set of pre-scan images within the summed pre-scan image, calculating the rate of change of the maximum real-time image signal for the hybrid image, comparing the rate of change of the maximum real-time image signal to the rate of change of the maximum pre-scan image signal, generating a graphical notification to indicate whether the maximum real-timeimage signal is within an acceptable range, and optionally modifying a radiotherapy treatment for the patient if the maximum real-time image signal is not within the acceptable range.

[0012] In some variations, a method for monitoring an ROI within a patient during a radiotherapy treatment session includes acquiring a plurality of subsets of pre-scan images of at least a portion of the ROI at a plurality of corresponding beam stations and generating a summed pre-scan image of the ROI. At each of the plurality of beam stations, the method further includes calculating a pre-scan minimum signal activity metric, determining an overall minimum signal activity threshold, acquiring a set of real-time images of the ROI, calculating a real-time signal activity metric, and comparing the real-time signal activity metric to the overall minimum signal activity threshold. If the real-time signal activity metric is greater than the overall minimum signal activity threshold, the method includes registering the set of real-time images with a set of corresponding pre-scan images and calculating a normalized mutual information metric (NMI) using the registered set of real-time and corresponding set of pre-scan images. If the NMI metric is above a predetermined NMI threshold, the method includes calculating a z-score to compare the set of real-time images to the summed pre-scan image, generating a graphical notification to indicate whether the z-score is within an acceptable range, and optionally modifying a radiotherapy treatment for the patient if the z-score is not within the acceptable range.BRIEF DESCRIPTION OF THE DRAWINGS

[0013] FIG. 1 A depicts an illustrative schematic of a radiotherapy patient having three regions of interest to be monitored during a radiotherapy session.

[0014] FIG. IB depicts an illustrative schematic of a variation of a region of interest for monitoring during a radiotherapy session.

[0015] FIG. 2 depicts a conceptual diagram of a variation of a method for acquiring images of a region of interest during a radiotherapy session.

[0016] FIGS. 3 A and 3B depict conceptual diagrams of a variation of a method for generating images of a region of interest during a radiotherapy session.

[0017] FIG. 4 depicts a schematic of an illustrative radiotherapy treatment system for use with the methods described herein.

[0018] FIG. 5 depicts an illustrative flowchart representation of a variation of a method for monitoring a region of interest during a radiation delivery session.

[0019] FIG. 6 depicts an illustrative flowchart representation of an example of a method for monitoring a region of interest during a radiation delivery session.

[0020] FIG. 7 depicts a simulation result of processing image data to determine an imaging monitoring metric according to an example method described herein.

[0021] FIG. 8 depicts an illustrative flowchart representation of an example of a method for monitoring a region of interest during a radiation delivery session.

[0022] FIG. 9A depicts three images of a region of interest monitored during a radiation delivery session using an example method described herein.

[0023] FIG. 9B depicts graphical representations of image monitoring metrics calculated during a radiation delivery session using an example method described herein.

[0024] FIG. 10 depicts an illustrative flowchart representation of an example of a method for monitoring a region of interest during a radiation delivery session.

[0025] FIG. 11 depicts two graphical representations of a distribution plot used to determine an image monitoring metric as described herein.DETAILED DESCRIPTION

[0026] Nonlimiting examples of various aspects and variations of the invention are described herein and illustrated in the accompanying drawings.

[0027] Described herein are methods for real-time monitoring and / or tracking of a region of interest within a treatment zone during a radiation delivery session, which may be, in some variations, a radiotherapy treatment session. In general, the methods provide various ways to determine whether a region of interest remains within the treatment zone during the treatment session, and stopping the treatment if the region of interest moves outside of the treatment zone, for example due to anatomical movement, movement of the patient, etc. The methods described herein are also useful in the real-time monitoring and / or tracking of organs at risk relative to atreatment zone. In general, the methods may help determine whether an organ at risk has moved into or within a treatment zone during a radiotherapy session and stop the radiotherapy session if the organ at risk has moved into the treatment zone beyond a predetermined threshold. In this way, the methods provided herein provide a safety mechanism to stop the radiotherapy treatment to limit potential radiation exposure to healthy tissues and organs at risk. While the methods herein are described in the context of a treatment session, it should be understood that these methods may be used in the context of a quality assurance or calibration session, not a treatment session, where radiation may be delivered to a phantom and / or to radiation sensors instead of a patient. In such variations, the treatment zone may be referred to as a radiation-firing zone, and the treatment zone may be referred to as a target region or region of interest. If a target region has moved outside the radiation-firing zone and / or or a non-target region has moved into the radiation-firing zone, a notification may be generated to notify the user and / or a command signal may be sent to the radiotherapy system to cease radiation delivery. The methods described herein may be used with any type of radiotherapy, for example, image-guided radiotherapy (e.g., intensity modulated radiation therapy IMRT, stereotactic body radiation therapy IMRT, biology- guided radiotherapy BgRT).

[0028] In general, the methods described herein include generating a set of pre-scan images of a region of interest within a treatment zone within a patient, acquiring a set of real-time images comprising the region of interest, comparing the set of real-time images to the set of pre-scan images, and terminating the radiotherapy treatment session if the set of real-time images deviates from the set of pre-scan images by a predetermined threshold. For example, radiation delivery may be terminated or paused if a characteristic (e.g., intensity range, location of high-intensity region(s), noise level(s), frequency or spectral characteristics, etc.) of the real-time images differs from the pre-scan images by a predetermined threshold. Various modalities may be used to generate the pre-scan and real-time images (e.g., PET, CT, X-ray, MRI, SPECT, etc.), and various methods of generating and comparing the images are described herein throughout.

[0029] The region of interest (RO I) is the region targeted for radiation delivery, and typically contains one or more tumors (i.e., targets, target volumes) to be treated during a radiotherapy treatment session. As described herein throughout, the ROI may be determined and / or localized prior to delivery of radiotherapy using medical imaging (e.g., PET, CT, X-ray, MRI, SPECT, etc.). In some variations, the ROI may include more than one anatomical area. In somevariations, the intended radiation dose to the ROI may be higher than the dose delivered to regions surrounding the ROI.

[0030] As described herein, the methods of real-time imaging of an ROI may include a determination of an outer ROI zone (e.g., a biology-tracking zone BTZ) and an inner ROI zone (e.g., a planning target volume PTV). For example, as shown in FIG. 1 A, a radiotherapy patient 100 has a first ROI 102 including a BTZ 104 encompassing a PTV 106. As described herein throughout, the PTV 106 may be an inner ROI zone and the BTZ may be an outer ROI zone. The outer ROI zone may encompass the entire ROI and be used to help define a safe zone for delivery of radiation, i.e., radiation may be delivered to the inner ROI zone (e.g., a moving tumor or target) anywhere inside the outer ROI zone (e.g., BTZ). The inner ROI zone (e.g., PTV) may encompass the gross volume of tumor or tissue to be treated (the treatment volume) within the ROI and may also include a surrounding margin to compensate for uncertainties in planning or treatment delivery. For example, the surrounding margin may include an internal margin to account for tumor motion and a setup margin to account for other uncertainties. Alternatively, in some variations, the inner ROI zone may encompass the tumor (e.g., gross tumor volume) without a margin for motion and / or setup uncertainties. While radiation dose delivery may be prescribed for the inner ROI zone, radiation delivery may occur anywhere within the outer ROI zone during radiation delivery. For example, a tumor may be within a target volume having a surrounding margin, which together define a treatment volume (i.e., an inner ROI) within a biology-tracking zone (i.e., an outer ROI). In conventional SBRT, the outer ROI zone may be a planning target volume (PTV) and the inner ROI zone may be a clinical target volume (CTV) encompassing a target for radiation therapy (e.g., tumor, lesion). In some variations, an internal volume (ITV) may be a margin added to the CTV to compensate for internal physiological movements and variation in size, shape, and position of the CTV during therapy. In BgRT, the outer ROI zone may be the BTZ and the inner ROI zone may be the PTV. The BTZ may be defined by adding a margin (to account for, as an example, tracking errors) to the ITV. The PTV may be defined by adding a margin (which may be called a biological guidance margin) to the gross tumor volume GTV. Thus, as shown in FIG. IB, the PTV 126, defined in BgRT, may encompass the ITV 124, the CTV 122, and the GTV 120. The methods described herein may be used with one, or more than one, ROI (e.g., two, three, four, or more ROIs, which may each comprise an outer ROI zone and an inner ROI zone).

[0031] Referring again to FIG. 1, the patient 100 has an organ at risk (OAR) 108, which may optionally be monitored or tracked using the same methods as for monitoring or tracking an ROI. An OAR may be an anatomical object especially vulnerable to radiation toxicity and for which radiation exposure should be limited. In some variations, treatment plan constraints may comprise a maximum OAR dose, and the plan may be constrained such that the radiation exposure of the OAR does not exceed the maximum OAR dose. During a radiation delivery session, the OAR may be monitored or tracked using any of the methods described herein to help ensure that the OAR is not irradiated beyond the specified maximum OAR dose. For example, movement of patient 100 (due to, e.g., respiration, peristalsis, etc.) may cause an OAR 108 to enter ROI 102 (e.g., BTZ 104) during treatment, potentially exposing OAR 108 to radiation prescribed for PTV 106. Accordingly, one or more OARs may be monitored or tracked during a radiotherapy session to account for movements of healthy tissue relative to a safe zone for treatment delivery (e.g., BTZ 104).

[0032] According to some variations of the methods described herein, a volume or set of prescan images and a volume or set of real-time images are generated or acquired and then compared. Each volume of images comprises a plurality of images generated or acquired to monitor a ROI and / or OAR. Pre-scan images may be acquired during an imaging session prior to a radiation treatment delivery session (e.g., immediately before the treatment session, or at some extended point in time before the treatment session, or days or weeks before the treatment session). Real-time images may be acquired during the radiation treatment delivery session. Each volume or set of images may have one or more subsets of images, and the subset of images may be acquired at and correspond to various control points (e.g., particular locations or positions of a patient platform within the radiotherapy system). In some variations of the methods described herein, a patient may be moved to each control point within a series of control points where medical imaging and / or treatment (e.g., radiation) delivery may occur. A volume of image subsets may include more than one subset of images corresponding to more than one control points of the radiotherapy treatment system. That is, each control point may define a subset or volume of image subsets of at least a portion of the ROI, which may be treated at the control point. In some variations, a number of subsets of images to be acquired at a control point may depend on an amount of image data needed for reconstruction and further processing. This number of subsets of images may be referred to as an index of limited time sampled images (i.e., an LTS index or LTS volume). Further, each image within a subset, volume of imagesubsets, or set of images (e.g., a set including more than one volume of image subsets) may provide an image of a portion of the ROI. The images may be summed or combined to form a larger or complete image of the ROI. As an example, FIG. 2 shows a plurality of control points 202 as numbered vertical lines (1-16) intersecting an ROI including BTZ 214, PTV 212, and GTV 210. Here, each LTS volume 204 includes an image of an ROI which is a combination of four subsets of images, each subset corresponding to one of the numbered control points 202. In other variations, each LTS volume may include any number of subsets of images for any number of control points, e.g., 3 subsets of images for 3 control points, 5 subsets of images for 5 control points, etc. As shown, each LTS volume (e.g., LTS 1 (204)) is a unique volume of image subsets because each includes a first subset of images within the combination of subsets of images that corresponds to a unique control point. For example, LTS volume 204 labeled “LTS 1” includes an image that is a combination of the subset of images acquired at control point 1, the subset of images acquired at control point 2, the subset of images acquired at control point 3, and the subset of images acquired at control point 4. In some variations of the methods described herein, the methods may include summing acquired images to form a larger or complete image of the ROI. For example, referring again to FIG. 2, each of the image subsets shown in 202 (i.e., subsets 1-16) may be stitched 208 together to generate summed image 206, which is a complete image that may include the entire ROI (i.e., BTZ 210 encompassing PTV 212 and GTV 210). In some variations, a method of forming a larger or complete image of a ROI may comprise acquiring a subset of images of the ROI at a plurality of control points, generating a plurality of LTS volumes where each LTS volume comprises a combination of two or more of the acquired subsets of images, and generating or forming a larger image of the ROI by summing (i.e., combining) two or more of the plurality of LTS volumes.

[0033] In some variations of the methods described herein, the methods may include generating a hybrid image that includes imaging data acquired at different times. In some variations, a hybrid image may comprise imaging data of the same region or volume of an ROI and / or OAR, but acquired at different time points. For example, a hybrid image may be a combination of imaging data acquired at the beginning of a radiation delivery session (e.g., prescan imaging data) and imaging data acquired during radiation delivery (e.g., real-time imaging data). This hybrid image may be used to inform further image acquisition and / or radiation delivery. For example, a hybrid image may be generated during a radiation delivery treatment session and may be a combination of the total available real-time images and the remaining pre-scan images or remaining real-time images from the previous treatment pass. In one variation, a method of generating a hybrid image may comprise generating a reference image by combining multiple LTS volumes (also referred to as LTS indices), acquiring real-time (e.g., new) images comprising multiple real-time LTS volumes, and generating a hybrid image by substituting one or more of the LTS volumes of the reference image by the corresponding one or more real-time LTS volumes. The reference image may be the combination of imaging data acquired at multiple control points, e.g., multiple subsets of images from multiple control points. A real-time LTS volume may correspond to a reference image LTS volume if it is comprised of imaging data from the same control points. A hybrid image may comprise a plurality of LTS volumes, where some of the LTS volumes are from a reference image and the other LTS volumes are generated from real-time (e.g., current, new) imaging data. FIGS. 3 A and 3B illustrate an example of how a hybrid image may be generated. Referring to FIG. 3A, a reference or pre-scan image 302, which includes seven LTS indices of images (each corresponding to a control point Pl -P7), is used during a first treatment pass of a radiation therapy session to create hybrid images 304 and 306, and ultimately a summed real-time image 308 including a combination of all real-time images acquired during the first pass. The hybrid image may be updated at each control point when a complete LTS volume of real-time images is acquired and substituted into the reference image (pre-scan image 302) for a corresponding LTS volume of pre-scan images. That is, as shown in FIG. 3B, a hybrid image may be a summation of real-time and pre-scan images, where each LTS volume of real-time images, like real-time LTS volume 320, may be a combination of images (e.g., R1 and R2) and each LTS volume of pre-scan images, like pre-scan LTS volume 322, may be a combination of images (e.g., P3 through P7). Referring again to FIG. 3 A, after the first treatment pass is complete (i.e., real-time images for all LTS volumes have been acquired), the summed real-time image 308 may be used as the reference image (i.e., replaces the pre-scan image 302) for the second treatment pass, and newer real-time images acquired during the second pass (e.g., those included in LTS volume RL) may replace the corresponding images collected during the first treatment pass to create another hybrid image. This process may repeat for all treatment passes of a radiation delivery treatment session.

[0034] In some variations of the methods described herein, the methods may further include calculating a metric (which may be referred to herein throughout as an image metric or an image monitoring metric) that represents a characteristic of the image, for example, a pre-scan image metric, a real-time image metric, or both. The metric may be any suitable metric useful forcomparing images or for helping to inform the radiotherapy treatment session. For example, an image metric may be calculated to represent a state or status of an object (e.g., ROI) being imaged. An image metric may be a signal intensity value of one or more pixels or voxels within an image. That is, an image is a spatial map of pixels (2D) or voxels (3D) that each have a signal intensity value. Thus, calculations may be performed on the pixel or voxel intensity or signal values captured by images. For example, a signal intensity of a pixel / voxel or a group of pixels / voxels within an image (e.g., a real-time image) may be determined and compared to a signal intensity of a corresponding pixel / voxel or a group of pixels / voxels in a corresponding image (e.g., a pre-scan image) to determine a similarity of the two images. In some variations, a signal intensity value may be a statistic of a signal intensity value such as a mean signal intensity, maximum signal intensity, minimum signal intensity, standard deviation of a signal intensity (e.g., of a mean, maximum, or minimum signal intensity), or a rate of change of a signal intensity value (or statistic thereof). A rate of change of a signal intensity value may be calculated for a set of images by recording the signal intensity value for each subset of images within the set (e.g., at each control point or for each LTS index) and comparing each recorded signal intensity value (e.g., via linear modeling). As another example, because images are spatial maps, an image metric may be a location (having coordinates on the x-, y-, and / or z-axes of an acquired or generated image) of a signal intensity value or a statistic of a signal intensity value of one or more pixels or voxels within the image. As an example, an image metric may be a location of a maximum signal intensity value of one or more pixels or voxels within an image. In some variations, an image metric may be used in calculations to determine the state or status of the object being imaged. As described in detail below, the image metrics may be independently monitored to inform a real-time location status (e.g., shifting of the ROI) and / or a real-time obstruction status (e.g., via adjacent healthy tissue) of a ROI within a patient. One or more image metrics may be determined prior to delivery of radiation (e.g., pre-scan image metric that may be determined during an imaging-only or treatment planning session and / or during a pre-scan on the same day as the treatment session), and again during a radiation treatment delivery session (e.g., real-time image metric that may be determined while the radiation is being delivered). For example, a real-time image metric value that is similar to the pre-scan image metric value may indicate that the real-time status of the ROI is similar to its status during the pre-scan. This may, for instance, indicate that the ROI has not moved or changed since the pre-scan. In some variations, one or both of the pre-scan and real-time image metrics may be used to calculate athreshold or acceptable range of values to compare to the real-time image metric during a treatment delivery session.

[0035] When metrics are used and compared to one another they are attempting to proximate movement and / or obstruction of the ROI and / or OARs. When the metric of the real-time image is within an acceptable tolerance of the pre-scan image, its metric value may be above a predetermined lower threshold value and / or below a predetermined upper threshold value. In one variation, when the comparison of the pre-scan image metric and the real-time image metric exceeds a predetermined threshold during treatment delivery, that may indicate that the ROI is not in the same location it was in during pre-scan and / or that an OAR has entered the ROI. In this way, real-time modifications to the treatment plan may be made, or the treatment plan may be terminated for safety purposes.

[0036] In some variations, pre-scan images and / or real-time images may undergo processing to determine a pre-scan and / or real-time image metric. Processing the images may include applying filters to the images (e.g., shift-invariant filters, enhancement filters, background noise reduction filters, convolution filters, spatial filters, bit masks, etc.), shifting the images (e.g., shifting at least a portion of the images relative to a reference point), extracting data from the images, categorizing the images, visualizing the images (e.g., graphically), applying weights to the images, and the like. In some variations, the image comparison methods described herein may rely on a contrast between a high-intensity foreground region (e.g., a PTV emitting intense signal) and a low-intensity background region (e.g., a BTZ excluding the PTV (i.e., BTZ-PTV) and not emitting intense signal) to determine image metrics and thresholds used to determine if: (1) the foreground region (e.g., PTV) has shifted relative to the background region (e.g., BTZ), and / or (2) a nearby anatomical object (e.g., an OAR) has shifted into the background region.

[0037] For example, the image monitoring metrics may be used to determine if a planned treatment volume (PTV) has shifted out of a biology-tracking zone (BTZ) (e.g., a tumor excursion metric), and may include a maximum correlation position, a maximum correlation intensity, and / or a BTZ maximum value rate of change. As another example, the image metrics may be used to determine if an OAR has shifted into the BTZ (e.g., an OAR incursion metric) and may include a BTZ-PTV signal intensity change and / or BTZ signal activity standard deviation rate of change. In some variations, a BTZ-PTV signal intensity change may be a change in the signal intensity of the region within the BTZ but excluding the PTV region (BTZregion minus the PTV region). The methods described herein may include modifying a therapeutic radiation treatment for the patient if an image metric indicates significant tumor excursion from and / or significant OAR incursion to the ROI (i.e., if a calculated real-time image metric exceeds a predetermined image metric threshold). For example, a radiotherapy treatment system for use with the methods described herein may generate a notification indicating whether the real-time image or image metric is within a predetermined threshold and / or acceptable range (e.g., below a predetermined upper threshold and / or above a predetermined lower threshold) determined using the pre-scan image and may enter a locked state (e.g., beam interlock) if the real-time image or image metric is not within a safe range or below and / or above a safety threshold (e.g., delivery of radiation to the patient may be stopped or paused). For example, in a gross motion event where the real-time image metrics indicate that the target volume is no longer in the ROI (e.g., BTZ), or an OAR has entered the ROI, the system may trigger a beam- off interlock. When the interlock is triggered, the system may automatically save a record of the treatment delivered and may create a new “partial” treatment plan for the remainder of the fraction. In some variations, a treatment session may then be restarted with a new treatment plan, e.g., the new “partial” treatment plan. In some variations, one or more tumor excursion monitoring indices and / or one or more OAR incursion monitoring indices may be used simultaneously to determine whether it is safe to continue the treatment delivery session. Accordingly, comparison of pre-scan and real-time images and / or metrics associated therewith, may help to guide radiation delivery such that a prescribed radiation dose is delivered to the actual, real-time location of a region of interest over the course of the treatment session, and to reduce the cumulative toxicity of radiation therapy to a patient. The methods described herein may be used with compatible radiotherapy treatment systems described in further detail below.Systems

[0038] Any suitable radiotherapy treatment system may be used with the methods described herein.

[0039] For example, the radiotherapy system may include an integrated imaging system for any suitable imaging modality (e.g., PET, CT, ultrasound, X-ray, MRI, SPECT, etc.). For example, the imaging system may include at least one PET detector supporting an imaging mode to deliver biology-guided radiation therapy. In some variations, a radiotherapy system may comprise a gantry to which the imaging system may be mounted. A therapeutic radiation sourceand one or more beam-shaping components of the radiotherapy system may be mounted on the gantry. The beam-shaping components may be configured to change configurations in real-time in accordance with any real-time changes to the target position and / or a treatment plan. The treatment plan may be optimized prior to treatment delivery and may include a radiation-firing matrix and a planned fluence map, where the fluence map may include a set of beamlets and beamlet intensities to be applied to a patient. In some variations, the treatment plan may include multi-leaf collimator (MLC) instructions and / or therapeutic radiation source firing instructions for a sequence of control points. The radiation therapy system may deliver radiation to the patient in accordance with a segmented fluence map for corresponding control points, where each control point may include multiple firing angles or positions. For a BgRT radiotherapy treatment session, the radiation therapy system may deliver radiation to the patient in accordance with a fluence map generated in real-time by convolving the radiation-firing matrix (e.g., one or more firing filters) with the imaging data acquired by the PET detectors. This fluence map may be segmented in real-time into radiotherapy system instructions (e.g., MLC instructions and / or therapeutic radiation source firing instructions) for delivery.

[0040] In some variations, the imaging system may be mounted on a circular or substantially circular gantry configured to rotate around a patient area at a speed of about 1 RPM or more (e.g., about 10 RPM, about 40 RPM, about 60 RPM, about 70 RPM, etc.). A BgRT radiotherapy system comprising one or more arrays of PET detectors (e.g., two opposing PET arcs) may be configured to rotate about 60 RPM or more. Additionally, or alternatively, the imaging system may be capable of acquiring tomographic imaging data without any rotation. In some variations, the imaging system may be configured to continuously acquire real-time images of at least a portion of the ROI throughout a treatment delivery session. These real-time image subsets or combined image subsets may be rapidly processed (e.g., within milliseconds) by a controller into machine instructions that control the radiation treatment beam. As an example, in BgRT delivery, the real-time image subsets may be filtered and convolved with a radiation-firing matrix (e.g., firing filters) to generate a delivery fluence map, which is then segmented in realtime into radiotherapy system instructions that control the treatment beam.

[0041] FIG. 4 depicts one variation of a radiotherapy system that may be used for real-time monitoring of an ROI during a radiotherapy session. The radiation therapy system 400 may include gantry 402, rotatable about patient area 404, patient platform 412 capable of movingwithin the patient area, one or more PET detectors 406 (e.g., PET-CT, PET-MRI) mounted on the gantry, therapeutic radiation source 408 mounted on the gantry, and dynamic multi-leaf collimator 410 disposed in the beam path of the therapeutic radiation source. In some variations, the radiation therapy system may comprise a first array of PET detectors 406a and a second array of PET detectors 406Z> disposed across from the first array, a linear accelerator 408 (i.e., a LINAC), and a dynamic binary multi-leaf collimator 410. In some variations, the PET detectors 406 may have sufficient timing resolution and / or detector sensitivity to acquire time-of-flight (TOF) PET data. A patient disposed within the patient area 404 may have been injected with a PET tracer that emits positrons, and the PET tracer may accumulate at particular regions of the patient (e.g., irradiation-target regions such as tumor regions). The annihilation of a positron with a nearby electron may result in the emission of two photons traveling in opposite directions to define a line. One or more acquired pre-scan or real-time image or detected image data may include one or more positron annihilation emission paths (i.e., lines of response or LORs, emission paths). In some variations, the PET detectors may be time-of-flight PET detectors, which may help to identify the location of the positron annihilation event. In some variations, the PET detectors may support image reconstruction modes for supporting PET -guided radiotherapy. For example, real-time image reconstruction may be used for PET-guided radiotherapy. In some variations, a previously calculated treatment plan may be updated in accordance with real-time images and / or image data acquired by the PET detectors to update the fluence map such that the LINAC and MLC leaf configuration / beamlet selection account for target volume movement within an ROI (e.g., the BTZ). The fluence map may be updated using real-time images or image data as the patient is moved through the patient area (e.g., translated or otherwise moved through the gantry bore).

[0042] The patient may be moved to each of a sequence of control points for imaging and / or treatment delivery. For example, patient platform 412 may be a platform configured to move relative the gantry 402. For example, the platform may include translational movement capabilities in the x-, y-, and z-axes and / or rotational movement capabilities in pitch, yaw, and roll rotations. In some variations, platform 412 may translate along a y-axis of the treatment plane and pause at one or more discrete control points 414 (i.e., beam stations), where images of the ROI may be acquired and / or where radiation treatment may be delivered. The one or more control points 414 may be separated by a distance of between 0.5 mm and 5.0 mm, such as a distance of between about 1 mm and about 4 mm, between about 1.5 mm and about 3 mm, orbetween about 2 mm and about 2.5 mm. For example, each control point within sequence of may be separated from a neighboring control point by a distance of about 2.1 mm. Radiation may be delivered to a patient at each control point using single- or multi-pass motions of the platform 412, where the term “pass” refers to movement of the platform 212 to each of the control points 414 during treatment delivery such that an ROI (e.g., a PTV) passes through the treatment plane (e.g., the irradiation field or beamlets emitted by the LINAC) once. In some variations, a multipass motion may include 2, 3, 4, or more passes. For example, a treatment session may include four passes including two back-and-forth motions of the platform. In some variations, treatment delivery from the radiotherapy system may be delivered across a sequence of the control points 414, where delivery is further delivered across one or more firing positions or firing angles (platform angles subgroups) of the LINAC at each of the control points.

[0043] Optionally, radiation therapy system 400 may include a CT and / or X-ray imaging system mounted on the same gantry as the therapeutic radiation source or mounted on a separate gantry. For example, a kVCT imaging system may be mounted to the gantry to acquire 3D CT fan-beam images for localizing and aligning the patient for treatment delivery. In some variations, the imaging plane of the kVCT imaging system may be separate from the treatment plane of the LINAC in some variations, while in other variations, the kVCT imaging plane may be coplanar with the LINAC treatment plane. It should be understood that other imaging modalities (e.g., MRI, SPECT, etc.) may additionally or alternatively be used (or integrated) with an analogous radiotherapy system to implement the methods described in further detail below.

[0044] The methods described herein may be executed by a radiotherapy system via software, hardware, or a combination thereof. For example, the system may include a first software node to control and deliver radiation treatment delivery instructions to a second software node which implements the radiation treatment delivery instructions by controlling the delivery hardware (e.g., MLC 410 of FIG. 4). The delivery instructions transmitted to the software node for controlling delivery hardware may cause the radiotherapy system to lock, permanently or temporarily terminating the treatment, for example. As another example, a combination of software and hardware may be used to develop and update a radiotherapy treatment plan for a patient in consideration of a deviation in location or obstruction of a real-time ROI from a prescan ROI. Hardware modules may include, for example, a general-purpose processor (ormicroprocessor or microcontroller), a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), or the like, along with corresponding machine-readable memory components. Software modules (executed on hardware) may be expressed in a variety of software languages (e.g., computer code), including MATLAB, C, C++, Java®, Python, Ruby, Visual Basic®, and / or other object-oriented, procedural, or other programming language and development tools. Examples of computer code include, but are not limited to, micro-code or micro-instructions, machine instructions, such as produced by a compiler, code used to produce a web service, and files containing higher-level instructions that are executed by a computer using an interpreter. Additional examples of computer code include, but are not limited to, control signals, encrypted code, and compressed code.

[0045] Further, a radiotherapy system may include a controller communicably coupled to one or more of the gantry, PET detectors, LINAC, and / or MLC, where the controller has a memory that may store treatment plans, fluence maps, and system instructions / commands, and the like. The controller may also include a processor to execute the calculations and methods described herein and may be configured to receive and store data. The processor may be any suitable processing device configured to run and / or execute a set of instructions or code stored in a machine-readable memory component and may include one or more data processors, image processors, graphics processing units, physics processing units, digital signal processors, and / or central processing units. In some variations, the controller may include more than one processor. In some variations, the controller may include a display capable of supporting a user interface which may allow a medical professional to view, modify, or otherwise interact with collected radiotherapy treatment data (e.g., treatment plan, pre-scan and / or real-time images) and / or controls. In some variations, the user interface may display recommendations for therapy modifications. The user interface may be displayed on a suitable computing device which may interface with one or more medical databases (e.g., EMR) stored in one or more machine- readable memory components. In some variations, the controller may provide treatment information via the user interface (e.g., display, audio feedback, visual feedback, etc.).Methods

[0046] The methods described herein may be used in real-time radiotherapy or any radiation delivery session, including radiation delivery sessions where radiation is emitted to a phantom and / or radiation detectors instead of a patient (e.g., a quality assurance or calibration radiationdelivery session). In some variations, the methods may be used to deliver a prescribed dose of radiation to an ROI within a patient. FIG. 5 is a flowchart representation of one variation of a method for monitoring a region of interest during a radiation delivery session, e.g., a radiotherapy session. While FIG. 5 shows that each step of method 500 occurs one time, it should be understood that method 500 may be in part a continuous process having feedback loops between steps (e.g., a feedback loop may exist between steps 510 and 508), may include optional steps (i.e., step 502 and / or 512), and / or may include additional steps.

[0047] In one variation, method 500 may be used to deliver prescribed radiation to a patient upon a real-time determination that treatment delivery is safe, and to trigger a radiation beaminterlock upon a real-time determination that treatment delivery may be unsafe. Any of the various imaging comparison methods described throughout may be used. Method 500 may include conducting an optional imaging-only session 502 in which a reference image of an ROI within a patient may be generated or acquired. The reference image may be acquired of the patient in the treatment position, using the imaging system of the radiotherapy system (e.g., the PET imaging system of a BgRT radiotherapy system) and / or a diagnostic imaging system (e.g., a diagnostic PET-CT and / or PET-MRI imaging system). The reference image may be used to generate a treatment plan 504 which may include instructions for a radiotherapy system to deliver a prescribed dose of radiation to the patient. Once a treatment plan is generated, a radiation therapy session may begin in which, prior to delivery of radiation to the patient, a prescan image of the patient’s ROI is acquired 506 and used to calculate pre-scan image metric values and thresholds. Subsequently, a treatment delivery session may begin in which one or more chosen image metrics may be monitored 508 (e.g., compared to predetermined thresholds or ranges) and / or used in calculations to conclude whether treatment delivery is safe to continue in accordance with the treatment plan in real-time. In some variations, monitoring 508 image metrics may comprise acquiring real-time images during radiation delivery and calculating the value(s) of the image metric(s) based on the real-time images. In some variations, method 500 may include delivering 510 therapeutic radiation to the ROI concurrently or sequentially with monitoring 306 the image metric or metrics. A conclusion as to whether real-time treatment delivery is safe may be made by a user in conjunction with the information provided by the radiotherapy system, which may determine the real-time status of the ROI based on the one or more image metrics. For example, a notification may optionally be generated to indicate 512 a significant change of a real-time image of the ROI or real-time image metric relative to the pre-scan image of the ROI, pre-scan image metric, or pre-scan metric threshold. If a such the change exceeds an acceptable range or threshold, treatment delivery may be unsafe for the patient. Accordingly, a radiation beam-interlock may optionally be triggered 514, thereby permanently or temporarily terminating the treatment delivery. Variations of steps 502-514 are described in further detail below.Optionally conduct imaging only session 502

[0048] An imaging-only session may be conducted prior to a radiotherapy treatment session (including steps 506-510). During an imaging-only session, a representative reference image of the patient (e.g., a PET volumetric 3D image), including any ROIs, may be obtained. The image from the imaging-only session and simulation image data (e.g., CT simulation image data) may be provided as inputs to a treatment planning system to generate a treatment plan. In some variations, the simulation image data may include defined and / or localized contours (e.g., BTZ, PTV, OAR). In some variations of the method, the imaging-only session may be omitted and existing patient images (e.g., diagnostic PET images, MR images, ultrasound, and / or CT images) may be used for treatment planning.

[0049] To begin the imaging-only session, a localization image (e.g., a CT image) may be obtained and registered to a simulation image (e.g., a CT image) with defined and / or localized contours (e.g., BTZ, PTV, OAR). The image registration may be used to verify that an ROI is at an expected location (e.g., a tumor volume is within a PTV) and may facilitate mapping of subsequently acquired image data (e.g., PET images) to contours in the simulation image reference frame. The images may be acquired across multiple control points using one or more passes of a patient platform, for example. The number of control points within a pass may depend on the size of the ROI. That is, as described with respect to FIG. 4, the control points may be separated by a fixed distance (e.g., about 2.1 mm), so as a size of the ROI increases, the number of control points necessary to image the ROI at and deliver treatment from may increase.

[0050] To acquire the images, a patient may be loaded onto a platform (e.g., patient platform 412 in FIG. 4). In some variations, the patient may receive an injection comprising a radiotracer (e.g., fluorodeoxyglucose (FDG)) for PET-guided radiotherapy (e.g., BgRT), and / or any relevant PET tracer). Next, the patient may be moved via the platform to one or more control points where PET images of the ROI may be acquired. For example, the PET detectors 406a, b shownin FIG. 4 may be used to detect PET emissions from a treatment volume that has taken up the radiotracer. At each control point, a volume of images may be acquired. That is, each control point may define a subset or a volume of image subsets of at least a portion of the ROI to be treated at a corresponding control point. In some variations, a number of subsets of images to be acquired at a control point may depend on an amount of image data needed for reconstruction and further processing. This number of subsets of images may be referred to as an index of limited time sampled images (i.e., an LTS index or LTS volume). In some variations, the LTS volume at one given control point may include a combination of subsets of images from more than one control point to reconstruct and process (e.g., evaluate an image metric). For example, an LTS volume may include a subset of images corresponding to the given control point and at least a portion of a subset of images corresponding to a neighboring control point. In some variations, the LTS volume may include a combination of several subsets of images acquired at several corresponding control points (e.g., a combination of subsets of images corresponding to 4, 8, 10, or 12 control points). For example, the LTS volume may include the subset of images acquired at a given control point plus the subsets of images acquired at two adjacent control points. As another example and as shown in FIG. 2, the LTS volume (labeled as LTS1, LTS2, LTS3) may include four subsets of images corresponding to four control points, where the four subsets include the image subset for the current control point and the image subsets from three more other control points (e.g., three preceding control points, three succeeding control points, or combinations thereof). Further, the LTS volume may include a rolling subset of control points from which images are acquired such that the index collects a unique volume of images for each control point. In some variations, the set of pre-scan images may include a combination of LTS volumes (e.g., a combination of the LTS volume for the current control point plus one or more LTS volumes corresponding to other control points). In some variations, the LTS volume may include overlapping images which may be averaged (e.g., the signal intensity for each pixel or voxel within the overlapping region(s) may be averaged during image reconstruction). For example, the LTS volume for a given control point may include the subset of images for the control point, plus a portion of a subset of images for a preceding control point, plus the LTS volumes for two preceding adjacent control points. That is, in these variations, the portion of the subset of images for the preceding control point may overlap with the unique LTS volume for that preceding control point, and so on. Any overlapping portions of images within an LTS volume may be averaged for image reconstruction. Similarly, overlapping portions of combinedimages within a complete summed image of an entire ROI (e.g., overlapping portions of images within multiple LTS volumes used to generate a summed pre-scan image) may be averaged (e.g., the signal intensity for each pixel or voxel within the overlapping region(s) may be averaged during image reconstruction).

[0051] Reconstructed images acquired during the imaging-only session may be used to create a treatment plan for a patient. In some variations, reconstructing the images may include using a computer to reconstruct three-dimensional images showing a distribution of radiotracer concentration within the patient. In some variations, the reconstructed images acquired at each control point within a series of control points may be stitched together to yield a summed or complete image representative of the patient’s ROI. As described further throughout, images (e.g., summed image, subset of images, volume of image subsets) and / or image data collected during the imaging-only session and / or the pre-scan acquisition session may be used to calculate or define thresholds or acceptable ranges for image monitoring metrics (e.g., a maximum correlation position, a maximum correlation value, a BTZ maximum value (e.g. intensity value) rate of change, a BTZ-PTV shell intensity change, a BTZ activity standard deviation rate of change, etc.). Real-time images and / or image metrics may then be compared to these thresholds or acceptable ranges to help determine whether it is safe to continue treatment, or whether the treatment should be permanently or temporarily terminated.Generate treatment plan 504

[0052] Once images and / or image data are output from the imaging-only session, a treatment plan for the patient may be generated and / or optimized. In some variations, a number of control points and a number of treatment passes for a planned treatment volume may be prescribed depending on the size of the treatment volume and / or the number of target regions. In some variations, a treatment plan may define a number of subsets of images for an LTS volume to determine the amount of image data required to calculate the image metrics and thresholds described herein. For example, the treatment plan may define a number of subsets of images for an LTS volume to determine the minimum amount of image data needed to calculate image metrics and thresholds. In some variations, a prescribed dose of radiation may be divided into several smaller doses (i.e., fractions) over more than one radiotherapy session (e.g., several radiotherapy sessions over a period of several days). Accordingly, the methods described hereinmay be repeated for each fraction and / or until the full dose of radiation has been delivered to the patient.

[0053] In some variations, a treatment plan may include a radiation dose distribution for the ROI, such as a dose map for the PTV. In some variations, a treatment plan may also define radiation-sensitive regions of a patient (OARs) to which radiation exposure should be limited or minimized. Optionally, for BgRT, one or more firing filters (e.g., a radiation-firing matrix) and / or a fluence map may be generated as part of the treatment plan. For example, a fluence map may be defined by the firing filter acting on (e.g., convolving with) a matrix of target volume PET projections, where data that is used to generate the radiation-firing matrix (e.g., firing filters) may be calculated from imaging data obtained during a PET imaging-only session. Because the tumor PET projection is a fixed quantity, a cost function may be used to find an optimal firing filter to attain the prescribed dose or fluence distribution. One or more simulation CT images, contours defined by the simulation images (e.g., BTZ, PTV, OARs), PET planning images, and dose objectives may be inputs for optimizing the firing matrix using a cost function optimization process. Without wishing to be bound by theory, the principle of superposition allows for the firing filters (e.g., radiation-firing matrix) to be applied to real-time LTS images acquired during a treatment session to generate limited fluences, where the sum of limited fluences is the intended total fluence. This principle underpins the “real-time” nature of the monitoring methods described herein because it allows delivery of radiotherapy beamlets to be modified in response to rapidly collected packets of real-time imaging data (e.g., PET emissions). Alternatively, for image-guided radiotherapy (e.g., intensity modulated radiation therapy IMRT, stereotactic body radiation therapy SBRT), the cost function optimization process may include iterating through different fluence maps to in order to attain the desired prescribed dose while minimizing (or optimizing) the cost function. Developing the treatment plan may comprise segmenting the resultant fluence map into radiotherapy system instructions, such as an MLC and / or therapeutic radiation source firing instructions for each control point.Acquire pre-scan image to calculate image metric values and thresholds 506

[0054] A pre-scan image of the patient’s ROI may be acquired 506 or generated after generating 504 a treatment plan development and delivery 510 of therapeutic radiation. In some variations, the pre-scan image may be acquired or generated just before the patient’s treatment delivery session begins (e.g., 5 hours before, 1 hour before, 30 minutes before, 10 minutesbefore, etc.). In some variations, the pre-scan image may be acquired or generated on the day of a treatment delivery session, after the patient has been injected with a PET tracer. That is, acquiring 506 a pre-scan image may be a first step during a radiation therapy treatment session. For example, on the day of treatment method 500 may include acquiring or generating 506 a prescan image of the patient in the treatment position (e.g., in a supine position on a patient platform). For example, for a BgRT treatment session, a patient may be injected with a PET tracer, positioned in the BgRT radiotherapy system in the treatment position, and a PET pre-scan may be conducted to acquire PET imaging data. In some variations, conducting a pre-scan image session may include (1) choosing one or more image metrics to be used to help determine the safety of real-time delivery of radiation to the patient and / or (2) using the pre-scan images to calculate thresholds and / or acceptable ranges of values for the one or more image metrics to indicate a level of treatment safety. The image metrics, described in further detail below, may include: a maximum correlation position, a maximum correlation, a BTZ maximum value rate of change, a BTZ-PTV shell intensity change, a BTZ activity standard deviation rate of change, or combinations thereof.

[0055] That is, the pre-scan imaging data may be analyzed and / or processed to determine a pre-scan image metric for comparing real-time and pre-scan images to guide a radiotherapy treatment session. Processing the pre-scan images may include one or more: applying filters to the images (e.g., shift-invariant filters, enhancement filters, background noise reduction filters, convolution filters, spatial filters, bit masks, etc.), shifting the images (e.g., shifting at least a portion of the images relative to a reference point), extracting pre-scan image data from the images, categorizing the images, visualizing the images (e.g., graphically via histograms), applying weights to the images, registering the images, and the like. For example, an enhancement filter may be applied to a pre-scan and / or real-time image to generate a signal- enhanced pre-scan and / or real-time image. As another example, a mask filter may be applied to a pre-scan image to mask certain areas of an ROI (e.g., a BTZ mask may be used to hide the PTV and analyze only the BTZ-PTV shell region). In some variations, a bitmask may be used to filter the acquired PET imaging data to remove PET imaging data from areas outside the ROI.

[0056] An image metric (e.g., pre-scan image metric or real-time image metric) may be a characteristic or value of the image and may be any suitable metric useful for comparing images or for helping to inform the radiotherapy treatment session. In particular, an image metric maybe a signal intensity value of one or more pixels or voxels within an image. For example, a signal intensity of a pixel or a group of pixels within an image (e.g., a real-time image) may be determined and compared to a signal intensity of a corresponding pixel or group of pixels in a corresponding image (e.g., a pre-scan image) to determine a similarity of the two images. In some variations, a signal intensity value may be a statistic of a signal intensity value such as a mean signal intensity, maximum signal intensity, minimum signal intensity, standard deviation of a signal intensity (e.g., of a mean, maximum, or minimum signal intensity), or a rate of change of a signal intensity value (or statistic thereof). A rate of change of a signal intensity value may be calculated for a set of images by recording the signal intensity value for each subset of images within the set (e.g., at each control point or for each LTS volume) and comparing each recorded signal intensity value (e.g., via linear modeling). As another example, an image metric may be a location (along the x-, y-, and / or z-axes of an acquired or generated image) of a signal intensity value or a statistic of a signal intensity value of one or more pixels or voxels within the image. As an example, an image metric may be a location of a maximum signal intensity value of one or more pixels or voxels within an image. In some variations, an image metric may be used in calculations to determine the state or status of the object being imaged. As described in detail below, the image metrics may be independently monitored to inform a real-time location status (e.g., shifting of the ROI) and / or a real-time obstruction status (e.g., via adjacent healthy tissue) of a ROI within a patient.

[0057] In some variations of method 500, an acceptable tolerance may be determined for the pre-scan image metric(s) to represent acceptable movement and / or obstruction of an ROI and / or OARs. For example, acceptable movement of a PTV relative to a BTZ may be movement of the PTV within the BTZ such that in the BTZ still encompasses a majority of the PTV (e.g., encompasses between about 50% and about 100% of the PTV). Similarly, an example of acceptable movement of an OAR relative to a BTZ may be movement of the OAR resulting in the BTZ encompassing a minority of the volume of the OAR (e.g., encompassing between about 0% and about 50% of the OAR); in other words, less than about 50% of the OAR is within the BTZ. The acceptable tolerance may be an acceptable range of image metric values (e.g., signal intensity values and / or signal intensity location values) or one of an upper or lower threshold for acceptable image metric values to compare to a real-time image metric. In some variations, the pre-scan image metric(s) calculated for each control point within a series of control points may be compared to define an overall acceptable tolerance of image metric values to compare to allreal-time images acquired during treatment. For example, an acceptable range of image metric values may be used for image comparison at each control point and may be a range of values between an overall maximum signal intensity value (e.g., the maximum signal intensity among a pixel / voxel or group of pixel s / voxels across all image subsets or combined image subsets within a pre-scan image) and overall minimum signal intensity (e.g., the minimum signal intensity among a pixel / voxel or group of pixels / voxels across all image subsets or combined image subsets within a pre-scan image). In this example, at each control point within a series of control points visited during a treatment delivery session, each calculated real-time image metric may be compared to the range of acceptable image metric values for the same series of control points as determined during pre-scan acquisition. In some variations, an overall pre-scan image metric to compare to all real-time images may be using only the summed pre-scan image (i.e., an image metric may be determined by analyzing the summed pre-scan image and without comparing prescan image metrics for each control point).

[0058] In some variations, the pre-scan image metric(s) calculated for images corresponding to each control point may be used to define a local acceptable tolerance of image metric values (e.g., values determined using a subset of images or volume of image subsets for an LTS index for the control point) to compare to real-time images corresponding to the control point during treatment. As another example, an acceptable range of image metric values may be unique to each control point and may be, for example, a range of values between a local maximum signal intensity value and a local minimum signal intensity value. In this example, a real-time image metric may be compared to a unique range of acceptable image metric values corresponding to the given control point. In some variations, when the comparison of the real-time image metric value to the acceptable tolerance of pre-scan image metric values finds that the real-time image metric value is not within (i.e., the value is above or below) the acceptable tolerance, that may indicate that an ROI is not in the same location it was in during pre-scan and / or that an OAR has entered the ROI.

[0059] In some variations, defining acceptable ranges and / or thresholds for the image metric(s) may include conducting a calibration or simulation procedure in which pre-scan images (e.g., the summed pre-scan image, the images within an LTS volume) may be manipulated to simulate an event in which a target volume has moved out of the BTZ by a significant amount and / or an event in which an OAR has entered the BTZ by a significantamount. For example, the pre-scan and real-time images of the ROI may depict a foreground region (e.g., PTV) that emits more signal than, and is highly contrasted relative to, a background region (e.g., BTZ-PTV). Thus, a real-time image of the ROI having a lower signal intensity than that of a corresponding pre-scan image of the ROI may indicate that at least a portion of the PTV has shifted outside of the BTZ. It may be unsafe to deliver radiation to a PTV that is not in an expected location because it may increase the risk of radiation toxicity to the patient. Similarly, an image of an obstructed ROI (e.g., an OAR or non-target anatomical object has entered the ROI) compared to an unobstructed ROI may have a greater signal value due to the additional signal contributed by the obstructive object. If a real-time image of the ROI shows more signal than the corresponding pre-scan image of the ROI, at least a portion of an OAR or anatomical object may have shifted into the BTZ. It may be unsafe to deliver radiation to the PTV if the BTZ (the “safe zone” for firing) is obstructed by healthy tissue due to an increased risk of radiation toxicity to the patient. In both examples, delivery of radiation should be delayed until the PTV is within the BTZ. Accordingly, a simulation procedure using pre-scan images may represent either a PTV excursion event or an OAR incursion event and may inform a real-time treatment session so that a dose of radiation may be stopped or delayed if necessary. A PTV excursion event may indicate that a percentage of a volume or a percentage of a linear dimension of the PTV has moved out of the BTZ. A PTV excursion threshold may represent about a 10% tumor excursion from the BTZ, about 30% tumor excursion from the BTZ, about 50% tumor excursion from the BTZ, about 70% tumor excursion from the BTZ, or about 90% tumor excursion from the BTZ. Similarly, an OAR incursion event may indicate that a percentage of a volume or a percentage of a linear dimension of the imaged OAR has moved into BTZ. An OAR incursion threshold may represent about 10% OAR incursion to the BTZ, about 30% OAR incursion to the BTZ, about 50% OAR incursion to the BTZ, about 70% OAR incursion to the BTZ, or about a 90% OAR incursion to the BTZ.

[0060] A PTV excursion calibration procedure may include shifting a pre-scan image such that an overlap between the PTV and the BTZ is reduced (e.g., between about 5% and about 95% reduced, such as between about 20% and about 75% reduced, between about 30% and about 65% percent reduced, or between about 40% and about 60% reduced). For example, the pre-scan image may be shifted such that there is at least 50% overlap between the PTV and the BTZ. An image metric or threshold may be calculated using the shifted pre-scan image and used as athreshold during treatment delivery (i.e., beam interlock may be triggered if the signal intensity of the real-time image is at or below the predetermined threshold).

[0061] An OAR incursion simulation procedure may include simulating signal characteristics that may be representative of an OAR into the BTZ of the pre-scan images and reevaluating the images with the simulated signal characteristics. For example, an OAR incursion simulation may include increasing the signal intensity in the BTZ of pre-scan images and reevaluating the images that now have increased signal intensity. An image metric or threshold may be calculated based on this simulation and used during treatment delivery (e.g., beam interlock may be triggered if a signal of a real-time image is at or above the predetermined threshold). In some variations, processing the pre-scan image of the ROI may include using the PTV signal to mimic an OAR. That is, a total PTV signal (e.g., the summed signal intensity for each pixel or voxel within the PTV) may be adjusted (i.e., amplified or duplicated) relative to the BTZ to simulate at least a portion of an OAR within the BTZ. For example, in some variations, the simulated OAR within the BTZ may be assumed to be the same size and intensity as the imaged target volume. Accordingly, the target volume signal may be duplicated and at least a portion of it may be overlapped with the BTZ by an amount (e.g., half of the duplicated signal may overlap with the BTZ) to determine an OAR incursion threshold.Monitor image metrics(s) 508

[0062] Method 500 may include monitoring 508 one or more image metrics during a radiation treatment delivery session. In some variations, monitoring 508 the chosen image metrics over the course of a radiotherapy session may include determining a real-time image metric to compare to a threshold defined by the pre-scan metrics or uses (further calculations) thereof. The continuously acquired real-time images of the ROI may be analyzed and / or processed to determine one or more real-time image metrics, which may be compared to one or more pre-scan image metric values and / or metric thresholds calculated based on the pre-scan imaging data. In some variations, a volume of real-time images within an LTS index may be analyzed and / or processed to determine one or more image metrics.

[0063] As described below, determining a real-time image metric may be similar to determining a corresponding pre-scan image metric. A real-time image metric may be calculated for each control point within a series of control points. In some variations, the real-time imagemetric may be compared to a corresponding pre-scan image metric for the same control point(s) (e.g., a threshold and / or acceptable range of metric values was determined for each control point based on pre-scan imaging data). In some variations, the real-time image metric may be compared to a pre-scan image metric that is the same for all control points (i.e., a threshold and / or acceptable range of metric values was determined for use at all control points).

[0064] In some variations, a real-time image may undergo processing to determine a real-time image metric. Processing real-time images may include one or more of the following: applying filters to the images (e.g., shift-invariant filters, enhancement filters, convolution filters, spatial filters, bitmask, etc.), shifting the images (e.g., shifting at least a portion of the images relative to a reference point), extracting real-time image data from the images, categorizing the images, visualizing the images (e.g., graphically via histograms), applying weights to the images, registering images, and the like. For example, an acquired image may include foreground signal (e.g., the target volume signal) and background signal (e.g., the BTZ signal), and a mask may be applied to the image (e.g., a BTZ mask) such that only the background signal is considered in further processing. As another example, an enhancing convolution filter such as a Gaussian filter may be applied to acquired pre-scan images and / or real-time images to smooth an image and / or reduce image noise. As yet another example, in some variations, corresponding pre-scan and real-time images may be registered such that one or more common features of the images (e.g., BTZ, PTV, OAR(s)) may be compared. Generally, image registration is a process of finding a spatial transformation to make corresponding points of different images reach the same spatial position and anatomical position. The images may be obtained by different equipment or different imaging modalities. In some variations, the normalized mutual information between the images may be measured and / or the percent overlap between the images may be calculated as part of the image registration process. In some variations, nonrigid image registration may be performed using a combination of normalized mutual information and spatial information to consider the spatial information of the image neighborhood and / or increase the weight of the spatial information within the images. For example, adaptive Gaussian filtering may be introduced into a local image structure tensor and used to extract spatial information from the images while normalized mutual information is distributed to each pixel or voxel, and the discrete normalized mutual information may be multiplied with a weighting term to obtain a new measure (of normalized mutual information). Registered images having a maximized normalized mutual information metric and / or a high percent overlap may be determined to be similar.Accordingly, registering pre-scan and real-time images, and comparing the registered images may provide an indication of the similarity between the pre-scan and real-time images. A high degree of similarity may support a conclusion that a prescribed dose of radiation is safe to deliver to the patient. Generally, registering a pre-scan image (e.g., a volume of pre-scan image subsets corresponding to a control point or a summed pre-scan image of the ROI) and a real-time image (e.g., a volume of real-time image subsets corresponding to a control point) may include designating one image as the reference image, which may also be called the fixed image, and applying geometric transformations or local displacements to the other image so that they align with the reference image. In some variations, the reference image may be the pre-scan image, and the real-time image may be aligned with the pre-scan image.

[0065] In some variations, a pre-scan image and / or a real-time image acquired during a previous treatment pass may be used in conjunction with real-time images acquired during a current treatment pass to determine a real-time image metric. For example, in some variations, the real-time images acquired to inform treatment at a given control point may replace corresponding reference images (e.g., pre-scan images, acquired real-time images from a preceding treatment pass) within a summed reference image to generate a hybrid image. The hybrid image may be used to calculate a real-time image metric and may be dynamically updated as treatment progresses. For example, as shown in FIG. 3 A, at a first control point within a sequence of control points, a first LTS index of real-time images (e.g., R1 in hybrid image 304 or R1 ’ in hybrid image 310) may replace a first LTS index of reference images within the summed reference image (e.g., Pl in summed pre-scan image 302 may be replaced by Rl, R1 in summed real-time image 308 may be replaced by RL and so on). Subsequently, at a second control point within the sequence of control points, a second LTS index of real-time images (e.g., R2 in hybrid image 306) may replace a second LTS index of corresponding reference images (e.g., P2 in pre-scan image 302 and hybrid image 304) within the hybrid image. As represented by summed real-time image 308 in FIG. 3 A, after the first treatment pass, all pre-scan images may be replaced by corresponding real-time images. For a subsequent treatment pass, newly acquired real-time images may replace corresponding real-time images acquired during the preceding treatment pass.

[0066] Described below are example variations of image metrics and thresholds which may be calculated for pre-scan imaging data and for real-time imaging data, where comparing the real-time image metric value(s) with the corresponding pre-scan image metric value(s) during a radiation treatment delivery session may be used to determine in real-time whether there has been significant deviation in location and / or obstruction of an ROI. In some variations, an image metric may be determined or calculated without the use of a pre-scan image metric.Maximum correlation position

[0067] A real-time image that highly correlates with a corresponding pre-scan image may indicate that the real-time ROI is similar to the ROI as captured by the pre-scan image. For example, a high level of correlation between a real-time image with the pre-scan image may indicate that the ROI in the real-time image may be at the same location and / or orientation, and / or may have the same shape as the ROI in the pre-scan image. A maximum correlation metric may represent an amount of overlap between the cross correlated images via the degree to which the intensity of the real-time image correlates with the signal intensity of the pre-scan image. This may indicate, for example, that the real-time location of the ROI is similar to the location of the ROI during the pre-scan. Thus, a maximum correlation position metric be used to monitor any movement or shift (in the x-, y-, or z-axes) of an inner ROI relative to an outer ROI (e.g., movement or excursion of the PTV from the BTZ). The maximum correlation position metric may be a location (defined by a coordinate on one or more of the x-, y-, and z-axes) at which a maximum signal intensity resulting from a cross correlation calculated between two images occurs. At a given control point, a set of pre-scan images may be autocorrelated (i.e., the signal may be cross correlated with itself) with a summed pre-scan image of the ROI and evaluated to determine a pre-scan maximum correlation position. The autocorrelation may represent a zero- or low-shift event of the PTV relative to the BTZ because the resultant maximum signal intensity will occur at a lag or displacement of zero on the coordinate system. In some variations, a set of shifted pre-scan images corresponding to a given control point may be autocorrelated with the summed pre-scan image of the ROI. The shifted pre-scan images may be processed such that only about 50%, for example, of the PTV overlaps with the BTZ (in one or more of the x-, y-, and z-axes). Accordingly, a shifted pre-scan maximum correlation position may be determined using the shifted pre-scan autocorrelation, which may represent a large-shift even of the PTV relative to the BTZ because the resultant maximum signal intensity will occur at a lag of above zero on the coordinate system. This may simulate possible value(s) of the maximum correlation position metric if the PTV were to move relative to the BTZ.

[0068] To determine a real-time maximum correlation position, a local cross correlation may be calculated between a set of real-time images acquired at a control point and a summed prescan image of the ROI. A shift (in the x-, y-, or z-axes) of the real-time images from a corresponding location index for the control point may be identified. A shift that exceeds a predetermined threshold (e.g., the threshold determined based on the pre-scan imaging data) may indicate a significant excursion of the target volume out of the BTZ. A notification may be generated to inform the user of the target volume excursion.Maximum correlation intensity

[0069] Additionally, or alternatively, a maximum correlation intensity metric may be used to monitor target volume excursion from the ROI. The maximum correlation intensity metric may be used to monitor any movement or shift (in the x-, y-, or z-axes) of an inner ROI relative to an outer ROI (e.g., excursion of the PTV from the BTZ). The maximum correlation intensity metric may be a maximum signal intensity (among a plurality of pixel or voxel signal intensities analyzed) of a cross correlation between two images. At a given control point, a set of pre-scan images may be autocorrelated (i.e., the signal may be cross correlated with itself) with a summed pre-scan image of the ROI and evaluated to determine a pre-scan maximum correlation intensity. The autocorrelation may represent a zero- or low-shift event of the PTV relative to the BTZ because the maximum signal intensity value resulting from the correlation will be large (relative to the maximum signal intensity resulting from cross correlating shifted pre-scan images with the summed pre-scan image, for example). In some variations, a set of shifted pre-scan images corresponding to a given control point may be autocorrelated with the summed pre-scan image of the ROI. The shifted pre-scan images may be processed such that only about 50%, for example, of the PTV overlaps with the BTZ (in one or more of the x-, y-, and z-axes).Accordingly, a shifted pre-scan maximum correlation intensity may be determined using the shifted pre-scan autocorrelation, which may represent a large-shift even of the PTV relative to the BTZ because the maximum signal intensity value resulting from the correlation will be low (relative to the maximum signal intensity resulting from the autocorrelation).

[0070] To determine a real-time maximum correlation intensity metric, a local cross correlation may be calculated between a volume of real-time images acquired at a control point and a summed pre-scan image of the ROI. Then, a maximum signal intensity value of the local cross-correlation may be determined and used to evaluate a change in the maximum signalintensity relative to the pre-scan image data. In some variations, a weight factor may be applied to the maximum signal intensity of the local cross-correlation to reduce a detection sensitivity for treatment sessions using low-intensity imaging sessions and to strengthen the detection sensitivity for high-intensity imaging sessions. A change that exceeds a predetermined threshold (e.g., the threshold determined based on the pre-scan imaging data) may indicate a significant excursion of the target volume out of the ROI. A notification may be generated to inform the user of the target volume excursion.BTZ maximum value rate of change

[0071] Certain imaging modalities (e.g., PET) may show that the PTV accounts for most or substantially all of the signal activity within the BTZ. Thus, if a portion of the PTV exits the BTZ, a peak value of the signal activity within the BTZ may decrease. Therefore, monitoring a rate of change of a maximum value of imaging signal activity within the ROI as a function of control points may inform any excursion of an inner ROI from an outer ROI. The pre-scan rate of change of the maximum BTZ signal activity may be calculated per control point by modeling (e.g., simulating) the total maximum BTZ signal activity for a set of shifted pre-scan images corresponding to the control point. For example, the maximum BTZ signal activity for a volume of shifted pre-scan image subsets within LTS index corresponding to the control point may be calculated for each of the multiple subsets included. The calculation may include analyzing and / or processing each pixel / voxel or each of a plurality of groups of pixels / voxels to determine a maximum signal intensity value. Subsequently, the values for each subset within the volume of shifted pre-scan image subsets may be modeled such that the maximum BTZ signal activity values are fit linearly. Accordingly, a slope of the maximum BTZ signal activity may be determined for each control point. Further, a threshold for the rate of change of the maximum BTZ signal activity may be calculated by averaging the slopes of the maximum BTZ signal activity for all control points. As described further throughout, this BTZ maximum value rate of change threshold may be used for all control points during treatment delivery to determine if a real-time BTZ maximum value rate of change should trigger beam-interlock.

[0072] In some variations, to calculate a real-time maximum BTZ value rate of change, hybrid images (a combination of the total available real-time images and the remaining pre-scan images or remaining real-time images from the previous treatment pass, as described above) may be generated during the treatment delivery session. The real-time maximum value of signal activitywithin the BTZ may be modeled (e.g., using linear fitting) with any previously calculated realtime maximum values of BTZ activity for any preceding control points to determine the realtime maximum value rate of change. If this metric exceeds a predetermined threshold for a control point (e.g., the threshold determined based on the pre-scan imaging data), significant excursion of the target volume from the BTZ may have occurred in real-time. A notification may be generated to inform the user of the target volume excursion.BTZ-PTV intensity change

[0073] As noted above, an anatomical object entering the BTZ-PTV (a shell region of the BTZ excluding the PTV) may cause a mean value of activity (e.g., a mean value of the signal intensity value within each pixel / voxel or each group of a plurality of groupings of pixel s / voxels of interest) within the region to increase. In some examples, this increase may be detected as an increase in the mean image intensity value within the region. Thus, the BTZ-PTV signal intensity change metric may inform any incursion of the BTZ by an OAR and / or any other PET- avid, non-target region. In some variations, a calibration procedure as described above may be used to simulate incursion of the BTZ by an OAR. Next, the mean BTZ-PTV intensity may be calculated and used as a threshold for a significant change indicative of OAR incursion during treatment delivery.

[0074] At each control point during treatment delivery, a real-time mean BTZ-PTV signal intensity value may be calculated and compared to a pre-scan mean BTZ-PTV intensity threshold. Some methods may comprise identifying the region of the imaging data (e.g., pixels or voxels) that is likely to be the PTV, and the shell region may then be determined by subtracting the identified PTV region from the BTZ region. In some variations, a weight factor may be applied to the real-time mean BTZ shell intensity to reduce a detection sensitivity for low-imaging treatment session and to strengthen the detection sensitivity for high-imaging treatment sessions. In some variations, the pre-scan mean BTZ-PTV intensity may be constant and used for comparison at all control points. A calculated real-time mean BTZ-PTV intensity value rate of change that exceeds the pre-scan mean BTZ shell intensity threshold (e.g., the threshold determined based on the pre-scan imaging data) may indicate a significant incursion of an OAR to the ROI. A notification may be generated to inform the user of the OAR incursion into the BTZ.BTZ activity standard deviation rate of change

[0075] If an anatomical object enters the BTZ, a standard deviation of activity within the BTZ may change, e.g., increase. In some examples where the standard deviation of activity increases, this increase may be detected as an increase in the standard deviation of the signal activity within the BTZ (e.g., the standard deviation of the signal intensity value within each pixel / voxel or each group of a plurality of groupings of pixels / voxels of interest within the BTZ). Thus, measuring the BTZ activity standard deviation as a function of control points may provide an indication that an OAR and / or non-target region may be entering the BTZ. In some variations, a calibration procedure (as described above) may be used with a set of pre-scan images to simulate incursion of the BTZ by an OAR. The pre-scan rate of change of the BTZ activity standard deviation may be calculated per control point by modeling the BTZ activity standard deviation values calculated for each subset of images within a volume of shifted pre-scan image subsets within an LTS index corresponding to the control point. For example, the BTZ activity standard deviation for a volume of shifted pre-scan image subsets of an LTS index corresponding to the control point may be calculated for each of the multiple subsets included. The calculation may include analyzing and / or processing each pixel / voxel or each of a plurality of groups of pixels / voxels of interest to determine a BTZ activity standard deviation value. Subsequently, the values for each subset within the volume of shifted pre-scan image subsets (for a control point) may be modeled such that the BTZ activity standard deviation values are fit linearly. Accordingly, a slope of the BTZ signal activity standard deviation may be determined for each control point. Further, a threshold for the rate of change of the maximum BTZ signal activity may be calculated by averaging the slopes of the BTZ activity standard deviation values for all control points. As described further throughout, this BTZ activity standard deviation threshold may be used for all control points during treatment delivery to determine if a real-time BTZ maximum value rate of change should trigger beam-interlock.

[0076] In some variations, to calculate a real-time BTZ activity standard deviation rate of change, hybrid images (a combination of the total available real-time images and the remaining pre-scan images or remaining real-time images from the previous treatment pass, as described above) may be generated during the treatment delivery session. A real-time value of BTZ signal activity standard deviation may be determined for the real-time images corresponding to a given control point. To calculate the real-time BTZ activity standard deviation rate of change, the real-time BTZ signal activity standard deviation may be modeled (e.g., using linear fitting) in combination any available real-time BTZ activity standard deviations for preceding control points. A calculated real-time BTZ activity standard deviation rate of change that exceeds a predetermined threshold (e.g., an OAR incursion threshold that has been calculated using prescan imaging data) for the control point may indicate a significant incursion of an OAR into the BTZ. A notification may be generated to inform the user of the OAR incursion into the BTZ.BTZ peak activity

[0077] As described above with respect to the preceding image metrics, calculating and recording a pre-scan image metric representing signal activity (e.g., image pixel intensities) within the BTZ may be useful in analyzing real-time signal activity within the BTZ to qualify any tumor excursion from and / or OAR incursion to the BTZ. Accordingly, sets of pre-scan images for each of a plurality of control points may be evaluated to determine a peak signal activity of the BTZ, the PTV within the BTZ, and / or the BTZ-PTV (i.e., a shell region of the BTZ excluding the PTV). In some variations, a mask may be applied to the pre-scan image so that the BTZ-PTV may be evaluated. In some variations, a calculated pre-scan peak PTV signal activity (e.g., a maximum signal intensity of a pixel or group of pixels showing the PTV) and a calculated pre-scan peak BTZ-PTV activity (e.g., a maximum signal intensity of a pixel or group of pixels showing the BTZ-PTV) may be compared to determine a threshold for use in treatment delivery. In some variations, a ratio of the pre-scan peak PTV activity to the pre-scan peak BTZ- PTV activity may be determined. The pre-scan peak activity ratio may be categorized as meaningful or significant if it exceeds a threshold (e.g., greater than 1, such as, for example, between about 1.0 and about 2.0, between about 1.03 and about 1.8, between about 1.06 and about 1.6, between about 1.09 and about 1.4, or between about 1.12 and about 1.2). In some variations, the pre-scan peak activity ratio may be defined as meaningful if it is greater than 1.15. In some variations, the lowest meaningful pre-scan PTV peak activity value determined (where one pre-scan PTV peak activity is calculated per control point) may be set as a minimum activity threshold to compare to a real-time PTV peak activity.

[0078] During a treatment session, the real-time PTV peak activity value calculated using a volume of real-time images at each control point may be compared to the minimum activity threshold (e.g., the threshold determined based on the pre-scan imaging data) to determine if thePTV is currently within the BTZ. A notification may be generated to inform the user of the target volume excursion.Normalized mutual information

[0079] A normalized mutual information metric may be evaluated during real-time treatment delivery and used to spatially register the real-time images to the corresponding pre-scan images (or summed pre-scan image) for real-time tracking of the PTV. The normalized mutual information may be measured to indicate a level of similarity between registered real-time images and corresponding pre-scan images or a summed pre-scan image. In some variations, the mutual information between images is maximized when the entropy of their joint probability distribution (JPD) is minimized. Accordingly, in some variations, a JPD comparing the set of real-time images to the set of pre-scan images may be generated and analyzed. In some variations, the images may be scanned and clusters of signal types within each may be identified (e.g., organized into histogram bins) such that the JPD may show the distribution of each cluster. For example, three histogram bins may be used to categorize the signal within the pixels or voxels of each set of images as “background” (low signal), “foreground” (high signal), and “inbetween” (neither low nor high signal). If the JPD plot shows that entropy is minimized for all clusters, the real-time image may be determined to have converged to its registered location with respect to the pre-scan image. This measure may be useful in tracking the PTV during treatment delivery. If the PTV is determined to lack BTZ coverage by more than a certain amount (e.g., 50% or more, 75% or more) during tracking then a beam-interlock may be triggered.Percent overlap

[0080] The percent overlap metric may be calculated during real-time treatment delivery and compared to a predetermined threshold (e.g., a threshold chosen by medical professionals or automatically determined by a radiotherapy system) to determine an amount of overlap between the real-time imaged PTV and the pre-scan BTZ. The percent overlap may be measured to indicate a level of similarity between images after a set of real-time images is registered with a corresponding set of pre-scan images or a summed pre-scan image. In some variations, the percent overlap between two registered images may be calculated by scanning one of the images (e.g., the set of real-time images), counting a number of pixels or voxels having a certain signal type (e.g., having background (low) signal), scanning the other image (e.g., the set of pre-scanimages), counting a number of pixels or voxels having the certain signal type, and comparing the counts. If the percent overlap is greater than a predetermined threshold (e.g., greater than about 50% overlap), the real-time image of the PTV may be determined to adequately overlap with the pre-scan image of the BTZ. If the percent overlap is less than a predetermined threshold (e.g., less than 50% overlap which may indicate that the PTV has mostly escaped the BTZ), then the beam-interlock may be triggered. In some variations, the threshold percent overlap may be set at 100% if it is determined that the entirety of the PTV must be within the BTZ for radiation delivery. The percentage thresholds may be selected by a clinician based on one of more factors, including but not limited to: breathing motion, anatomic location, radiation-sensitivity of surrounding non-target tissues, and / or clinic practice guidelines.Z-score

[0081] A z-test may be carried out to calculate a z-score indicative of how well a real-time image of the ROI matches a corresponding pre-scan image of the ROI. A z-test is a statistical test to determine whether two population means are different (with known variances and a large sample size) and the resultant z-score is a number representing the result of the z-test. The higher the z-score, the further from the norm the data can be considered to be. Accordingly, a z-test may be used to calculate a z-score indicative of the degree to which a real-time image metric (e.g., real-time BTZ peak signal activity) deviates from a corresponding pre-scan image metric (e.g., pre-scan BTZ peak signal activity). For example, using the BTZ peak signal activity as the image metric, the z-score may inform whether an OAR has entered the BTZ and whether the signal within real-time BTZ matches the signal within pre-scan BTZ. That is, the higher the z- score, the more likely that at least a portion of an OAR has entered the BTZ and / or that at least a portion of a PTV has exited the BTZ. A range of acceptable z-score values may be determined during acquisition 506 of the pre-scan image, and a calculated z-score outside of the predefined range may signify a significant dissimilarity between the real-time image and pre-scan image. In some variations, the acceptable range of z-score value may be about -1.5 to 1.5, about -2.0 to 2.0, about -2.5 to 2.5, about -3.0 to 3.0, about -3.5 to 3.5, or about -4.0 to 4.0.Voxel or pixel count

[0082] In some variations, a number of voxels or pixels having values of signal intensity exceeding a minimum signal activity threshold (e.g., a threshold calculated using the pre-scanBTZ peak activity metric) may be counted and compared to the pixel or voxel count threshold (e.g., a voxel count threshold determined based on pre-scan imaging data) to inform any OAR incursion to the BTZ. For example, a number of pixels or voxels within a BTZ-PTV shell having signal activity value greater than the minimum signal activity threshold value may be relatively low absent any OAR incursion to the shell because the BTZ-PTV constitutes a background (low signal) region of an image of an ROI. Thus, if the number of pixels or voxels within the shell having a signal activity value greater than the minimum signal activity threshold value surpasses the pixels or voxel count threshold as compared to the pre-scan imaging data, the increased signal in the background region may suggest an OAR incursion into the BTZ. Accordingly, beam-interlock may be triggered if the real-time voxel count exceeds a predetermined voxel count threshold. A threshold for voxel or pixel count may be predetermined and the voxel or pixel count of a real-time image acquired during treatment delivery may be compared to the threshold to determine if an OAR has entered the BTZ region. In particular, the voxel or pixel count threshold may be a maximum number of voxels within an ROI (e.g., the BTZ-PTV) having signal activity exceeding a minimum signal activity threshold (e.g., a threshold calculated using the pre-scan BTZ peak activity metric). For example, the voxel count threshold may be between about 60 voxels and about 200 voxels, such as about 90 voxels, about 100 voxels, about 150 voxels, etc. In some variations, the voxel count threshold may be determined in part based upon one or more of the following factors: the volume of the BTZ, the size or volume of nearby OAR(s), the statistical noise distribution, and / or the strength of the PTV (e.g., foreground) signal relative to the signal in the BTZ-PTV (background), etc.Deliver therapeutic radiation 510

[0083] Treatment during a radiotherapy treatment session (or radiation delivery during a quality assurance or calibration procedure) may involve the delivery of a prescribed dose of radiation to the patient (e.g., target volume(s) within the patient) over a number of beam stations, a number of treatment passes, and in some variations, over a number of treatment fractions. Treatment delivery and acquisition of real-time images of the ROI may occur simultaneously or substantially simultaneously, where the real-time images may be used to inform whether the treatment should be modified or permanently or temporarily terminated in real-time. As noted throughout, a number of control points may be used for treatment, and they may vary at least in part based on a size or volume of a target volume. In some variations, the treatment session mayinclude revisiting each control point (and in some variations, each firing position for the control point) more than once to deliver a total prescribed dose of fluence to the treatment volume. For example, a treatment delivery session may include about 2, about 3, about 4, about 5, or about 6 passes at each control point to treat a patient.

[0084] During treatment delivery, e.g., BgRT treatment delivery, real-time images of the ROI may be continuously acquired and reconstructed to reveal a biological signature of the target volume. In some variations, real-time images for a predefined LTS index may be continuously acquired and updated at a current control point where treatment is delivered. At each prescribed control point, radiation delivery may be further broken (i.e., partitioned, divided, binned) into subgroups of gantry angles, or firing position groups, each of which may include a plurality of firing positions. At each firing position group, a location status and / or obstruction status of the ROI may be updated using real-time reconstructed images of an LTS index. In some variations, the ROI status may be updated for a given control point using real-time PET images reconstructed from the last 100-1000 ms (e.g., about 250 ms, about 500 ms, or about 750 ms) of PET data, which may be updated every 1-500 ms (e.g., about 10 ms, about 50 ms, about 100 ms, or about 150 ms). A radiation fluence may be calculated for the updated ROI status using a firing filter and normalized to match planned fluence (which may be determined during develop treatment plan step 504). The fluence may then be segmented to machine deliverable fluence which may include of a number of beam pulses and / or an MLC leaf pattern at each firing position within the group. Thus, the LTS PET images may be used to redirect the fluence to a tumor that has moved within the BTZ. As described herein throughout, if a comparison of the LTS PET images and / or image data to the pre-scan images and / or image data indicates that an inner ROI has shifted outside an outer ROI by a significant amount (e.g., the real-time image metric values exceeds a predetermined threshold) or that an OAR has entered the ROI by a significant amount (e.g., the real-time image metric value exceeds a predetermine threshold), the radiotherapy system may enter a state of interlock.Indicate real-time status of ROI 512

[0085] A signal indicative of a real-time status of the patient’s ROI may be transmitted to a display and / or to a user interface on a display. The real-time status may include an ROI excursion status (e.g., location status of PTV relative to the BTZ) and / or an ROI incursion status (e.g., obstruction status of the BTZ relative to one or more OARs). The notification may begraphical and may be interpretable to a medical professional. In some variations, the image metrics and / or thresholds described herein may be used to generate a notification of unsafe and / or safe levels of tumor excursion from and / or OAR incursion to the ROI. The notification may inform a medical professional about a real-time status of the ROI via digital, visual, aural, or haptic outputs (or combinations thereof), and may be produced via a controller or a device communicably coupled to the controller (e.g., a computer). In some variations, a radiotherapy treatment system may automatically trigger a beam and / or system interlock if a real-time image metric exceeds a predefined threshold. That is, a real-time image metric that exceeds a predefined threshold for the metric may suggest that the real-time ROI - which may be imaged and reconstructed continuously during a treatment session - does not match the pre-scan ROI, and that delivering radiation to the real-time ROI may increase a risk of complications from radiotherapy treatment (e.g., toxicity) for the patient. Accordingly, generation and / or emission of a radiation beam from a compatible radiotherapy treatment system may be permanently or temporarily terminated based on the indication of the real-time ROI status. In some variations, a notification indicative of a significant difference between a real-time image and a corresponding pre-scan image (i.e., a significant real-time tumor excursion from and / or OAR incursion to the ROI considering pre-scan image data) may differ from a notification indicative of a nonsignificant difference between a real-time image and a corresponding pre-scan image (i.e., a nonsignificant real-time tumor excursion from and / or OAR incursion to the ROI considering pre-scan image data). For example, a notification that a real-time image deviates significantly from a pre-scan image may be represented by the activation of a first LED on one or more elements of a radiotherapy treatment system (e.g., a controller, a computer, a gantry, etc.) while a notification that a real-time image does not deviate significantly from a pre-scan image may be represented by activation of a second LED on the one or more elements of a radiotherapy treatment system. The first LED may indicate to a medical professional that it is unsafe to deliver radiotherapy, while the second LED may have a different color from the first LED to indicate that it is safe to deliver the prescribed radiation to the patient.Optionally trigger beam interlock 514

[0086] As noted above, if a comparison of a real-time image and corresponding a pre-scan image indicates that treatment delivery may be unsafe for the patient, a beam interlock may be triggered to prevent or halt generation of a prescribed radiation dose for the patient. Theradiotherapy systems described herein may be used to implement an interlocked state of the radiotherapy treatment system. For example, a notification indicating it is unsafe to deliver radiation to the ROI may cause a first software node of the treatment system to transmit delivery instructions to a second software node which controls the delivery hardware to execute the transmitted commands. Thus, the second software node may execute a command to pause or shut down at least part of the radiotherapy treatment system (e.g., to close all the leaves of a collimator, halt radiation generation from the LINAC) in response to a notification indicating treatment is not safe to deliver to the patient. In some variations, a treatment plan may be updated during beam-interlock considering a real-time location and / or obstruction status of an ROI. Subsequently, a treatment delivery session may restart using the updated treatment plan.Examples

[0087] Various aspects of example variations of methods of monitoring an ROI during a radiotherapy treatment session are described in further detail below.Example 1

[0088] A variation of a method of monitoring an ROI during a radiotherapy session using maximum correlation location and intensity metrics is summarized as steps 602-618 of method 600 in FIG. 6 and described in detail below. In some variations, some or all of steps 602-618 (e.g., steps 608-618) of the method 600 may be repeated for all treatment passes in a treatment session.

[0089] In one variation, method 600 may comprise acquiring 602 pre-scan PET images of the ROI with the patient in the treatment position. The pre-scan PET images may include a plurality of volumes of PET images which may be acquired at each of a plurality of corresponding beam stations (where j denotes the jth beam station). In some variations, each image volume may include combination of pre-scan images within an LTS volume for the current beam station and the volumes of pre-scan images within LTS indices for two immediately preceding beam stations. A summed pre-scan image may be generated by combining the plurality of subsets of PET images as a single reconstructed image of the entire ROI. Method 600 may include calculating 604 true and shifted pre-scan image metrics using cross correlations (as described in detail above) and determining 606 beam-interlock thresholds using the pre-scan image metricsand an OAR incursion simulation procedure (as described in detail above). For example, in one variation, to calculate 604 a true pre-scan image metric for a beam station, a volume of pre-scan image subsets (corresponding to the beam station) may be cross correlated with the summed prescan image, and maximum signal location (Lo(j)) and maximum signal intensity (Io(j)) may be recorded for each beam station. In one variation, to calculate 604 the shifted pre-scan image metric for a beam station, a volume of pre-scan image subsets may be shifted such that a 50% overlap exists between the PTV and the BTZ and may subsequently be cross correlated with the summed pre-scan image. The maximum shifted signal location (Ls(j)) may be determined 606 as the shift threshold for each beam station. Additionally, the pre-scan mean signal in the BTZ-PTV shell (Mo(j)) may be noted for each beam station, and a simulation method as described above may be performed to determine 606 an OAR incursion threshold (R(j), an increased mean signal) for each beam station (where the PTV may be used to mimic an OAR having 50% overlap with the BTZ). Further, in some variations, the mean maximum intensity correlation (lo(mean)) and the minimum intensity correlation (lo(min)) may be determined considering all pre-scan images, and the following signal intensity auto-interlock trigger reference may be determined 606 for use at all beam stations during treatment delivery: [[lo(mean) - Io(min)] / Io(mean)]. In other variations, the signal intensity auto-interlock trigger reference may be determined 606 by the user and / or calculated based on other characteristics of the pre-scan image.

[0090] Method 600 may include acquiring 608 real-time images of the ROI at each of a plurality of beam stations during treatment delivery. Next, at each beam station, the real-time image metrics may be calculated 610 using cross correlations. For example, the real-time maximum correlation location (Li(j )) and maximum correlation intensity (Ii(i)) metrics may be calculated 610 using volumes of real-time PET images corresponding to the volumes of pre-scan images. In some variations a weight factor may be applied to Ii(i)) to reduce a detection sensitivity for low-imaging treatment session and to strengthen the detection sensitivity for high- imaging treatment sessions. FIG. 7 depicts images from a simulation of method 600 and provides an example of how real-time maximum correlation location and intensity were identified per LTS for a treatment pass. Each LTS (1, 2, and 3 as pictured) was acquired and cross correlated locally with the summed Pre-scan image shown. The cross correlations produced signals 702 in the correlation domain which were analyzed to identify a maximum cross correlation per beam station (Cmax(j)), which includes both a location and an intensity) atpoints 704, 706, 708. Li(j)and Ii(j) were determined from Cmax(j) and recorded for use steps 612-618 of method 600.

[0091] Method 600 may include performing 612 calculations using the pre-scan and real-time image metrics and comparing 614 the results to beam-interlock thresholds determined in step 606. For example, in some variations, the calculation [Li(j)-Lo(j)] may be performed 612, and the result compared 614 to Ls(j). If the result is greater than Ls(j), a notification of beam interlock may be indicated 616 and beam interlock may be triggered 618. In some variations, the calculation [Ii(i) / Io(mean)] may be performed 612, and the result compared 614 to the signal intensity auto-trigger reference. If the result is greater than the signal intensity auto-trigger reference, a notification of beam interlock may be indicated 616 and beam interlock may be triggered 618. In some variations, the calculation [Mi(j)-Mo(j)] may be performed 612 and the result compared 614 to R(j). If the result is greater than or equal to R(j), a notification of a realtime status of the ROI may be indicated 616 and beam interlock may be triggered 618. Optionally, if the result of the comparison 614 indicates that the real-time image metrics are within the acceptable range(s) and / or threshold(s), a notification may be generated indicating to the user that the monitoring of the ROI has completed successfully and that it is safe to continue radiation delivery.Example 2

[0092] A variation of a method of monitoring an ROI during a radiotherapy session using the BTZ maximum value rate of change and BTZ activity standard deviation rate of change metrics is summarized in steps 802-822 of method 800 in FIG. 8 and described in detail below. In some variations, some or all of steps 802-822 of the method 800 may be repeated for all treatment passes prescribed to the patient.

[0093] Method 800 may include generating 802 shifted pre-scan images of a patient’s ROI (e.g., BTZ and PTV) by acquiring a plurality of subsets of pre-scan PET images at each of a plurality of corresponding beam stations, where each subset may be a combined volume of images within an LTS index for a corresponding beam station. Generating 802 the shifted prescan images may include shifting each subset of images such that the imaged PTV may have half overlap with the BTZ. Method 800 may include stitching 804 the shifted subset of pre-scan images with any available (preceding) shifted subsets of images making up the LTS index forthe beam station. Then, statistics describing the signal intensity within the voxels of the volume of current images within the LTS index may be calculated 806. In some variations, the statistics may include, for example, the BTZ maximum signal activity metric and / or BTZ standard deviation of signal activity metric, which may be determined using all available stitched shifted subsets of pre-scan images for the current beam station. The method 800 may include simulating or modeling 808 all calculated statistics for a beam station (e.g., the BTZ maximum signal activity metric and / or BTZ standard deviation of signal activity metric for each subset of prescan images within a full LTS index for the beam station). Modeling may be, for example, linear fitting of the statistics such that pre-scan image metrics may be calculated 808 based on their fit (e.g., slope values of the statistics may be any image metric, as described above). Once all prescan image metrics are calculated 808 for a beam station (e.g., metrics for each subset of prescan images within an LTS index for the beam station), they may be averaged 810 to obtain beam-interlock thresholds. Such thresholds may be, for example, a PTV excursion threshold (e.g., the average slope of BTZ maximum signal activity value, as described above) and / or an OAR incursion threshold (e.g., the average slope of BTZ standard deviation of signal activity, as described above) for the given beam station. In some variations, the plurality of subsets of prescan images may be combined to generate a summed pre-scan image for use in generating and analyzing real-time hybrid images of the ROI.

[0094] The method 800 may include acquiring 812 set of real-time PET images corresponding to the subset of pre-scan images for each of the plurality of beam stations during a radiation delivery session (e.g., treatment delivery session). The set of real-time images may be stitched 814 to generate a dynamic hybrid image (i.e., it may replace a corresponding subset of pre-scan images within the summed pre-scan image to update a hybrid image, as described above). Additionally, statistics describing the signal intensity within the voxels of the subset of real-time images may be calculated in order to calculate 816 real-time image metrics of the ROI. For example, the real-time BTZ maximum signal activity value and / or the BTZ standard deviation of signal activity value may be determined, and once all subsets of real-time images for an LTS index are acquired and added to the hybrid image, linear fitting may be performed to calculate 816 the real-time BTZ maximum signal activity value rate of change and real-time BTZ standard deviation of signal activity rate of change were calculated for a current beam station (as described above). The method 800 may include comparing 818 the real-time images metrics to the beam-interlock threshold(s) calculated in step 810. If the comparison 818 finds that the real-time image metric (e.g., real-time BTZ maximum signal activity value rate of change and / or real-time BTZ standard deviation of signal activity rate of change) exceeds the beam station beam-interlock thresholds (e.g., the PTV excursion threshold and / or the OAR incursion threshold) then a notification of a real-time status of the ROI may be indicated 820 and beam interlock may be triggered 822.

[0095] FIGS. 9A-9B depict plots and results from a simulation of method 800. FIG. 9A depicts data from three treatment passes of a radiotherapy treatment delivery session in the simulation. FIG. 9B depicts simulation data from treatment pass 2 (904 in FIG. 9A) showing that beam interlock was triggered during the pass. In particular, FIG. 9A shows summed pre-scan image 900, summed image of first treatment pass 902 (a sum of the images which were acquired in real-time during the first treatment pass), and summed image of second treatment pass 904 (a sum of the images which were acquired in real-time during the second treatment pass). PTV 920 is shown as a circle about in the center of each image. As shown, a high-signal target volume remained substantially within PTV 920 during the first treatment pass 902 but did not remain within PTV 920 during second treatment pass 904. In this example, beam-interlock was triggered at the 51stbeam station - about halfway through the second treatment pass 904, where PTV 920 is depicted - as a result of comparing a PTV image metric to a PTV excursion threshold. FIG. 9B depicts two graphical representations of the PTV excursion and OAR incursion metrics described above and calculated for treatment pass 904 in FIG. 9A. Each graph 906, 908 shows how the respective monitoring metric varied as a function of beam stations during the second treatment pass 904. PTV excursion threshold 930 and OAR incursion threshold 932 for beam station 51 are also denoted. As shown, the PTV monitor metric surpassed the PTV excursion threshold at beam station 51, triggering beam-interlock. The OAR incursion threshold 932 for beam station 51 was not surpassed during this treatment pass, which aligns with the information observed in summed image 904 of FIG. 9A, i.e., the OAR did not enter the BTZ.Example 3

[0096] One variation of a method of monitoring an ROI during a radiotherapy session may use the BTZ peak activity monitoring metric, the minimum signal activity threshold, the normalized mutual information metric, the z-score threshold, and the voxel count threshold. The BTZ peak activity monitoring metric, the minimum signal activity threshold, the normalized mutualinformation metric, and the z-score threshold may be used to track PTV characteristics and excursion from the BTZ during treatment delivery. The voxel count threshold may be used to determine OAR incursion into the BTZ during treatment delivery. The method is summarized as steps 1002-1022 of method 1000 in FIG. 10 and described in detail below. In some variations, some or all of steps 1002-1022 of the method 1000 may be repeated for all treatment passes prescribed to the patient.

[0097] Method 1000 may include acquiring 1002 a plurality of subsets of PET pre-scan images at each of a plurality of corresponding beam stations, where each subset may be included in one or more beam station LTS index (e.g., each LTS index may include four subsets of prescan images corresponding to four consecutive beam stations). In some variations, an enhancement filter may be applied to the subset of pre-scan images. Pre-scan image metrics may be calculated 1004 using the (enhanced) pre-scan images. In some variations, the pre-scan image metrics may include, e.g., the PTV peak activity value and the BTZ-PTV peak activity value for each volume of images within an LTS index. Method 1000 may comprise calculating 1006 a signal threshold using the pre-scan image metrics. For example, a foreground signal-to- background signal (PTV peak activity to BTZ-PTV peak activity) ratio may be calculated and categorized as meaningful if it is greater than a predefined threshold (e.g., equal to or greater than about 1.15). Then, the lowest meaningful PTV peak activity value (i.e., the lowest PTV peak activity value out of the group of PTV peak activity values used to calculate a meaningful foreground signal-to-background signal ratio) may be set 1006 as a signal threshold (e.g., a minimum signal activity threshold) for use in evaluating real-time images. Additionally, or alternatively, a summed pre-scan image may be generated and / or thresholds for normalized mutual information, z-score, and / or voxel count may be determined for use in evaluating realtime images.

[0098] The method 1000 may include acquiring 1008 an LTS volume of real-time images of the ROI at each of the plurality of corresponding beam stations. The real-time images may be acquired, for example, during radiation delivery. In some variations, acquiring the volume of real-time images may include enhancing the images (e.g., with an enhancement filter) to facilitate calculating 1010 real-time image metrics (e.g., real-time PTV peak activity and / or BTZ-PTV peak activity) using the real-time images (as described with respect to step 1016 below). In some variations, calculating 1010 the real-time image metrics may include using theimage metrics to determine a representative signal value. For example, a real-time foreground signal-to-background signal ratio may be calculated using real-time PTV peak activity and BTZ- PTV peak activity (as described above). In some variations, ratio may be categorized as meaningful (i.e., enough signal is present that the method 1000 can proceed) if it is, for example, equal to or greater than about 1.15. Next, a representative signal value may be compared to a signal threshold 1012. For example, the signal threshold may be the minimum activity threshold calculated in step 1006. If the real-time PTV peak activity value is less than the minimum activity threshold, real-time image acquisition and / or treatment delivery may continue. If the real-time PTV peak activity is equal to or greater than the minimum activity threshold, the current set of real-time images may be enhanced 1014 (e.g., with an enhancement a mask) such that fewer than all voxels in the BTZ may be considered in the following steps of method 1000. The volume of enhanced real-time images may be registered 1014 with a corresponding volume of pre-scan images such that the pre-scan PTV (the PTV as reconstructed in the summed prescan image) may be mapped to the current beam station frame of reference. In some variations, normalized mutual information may be used to register 1014 the image volumes using a joint probability distribution including three clusters of voxels - voxels having foreground (PTV) signal levels, voxels having background (BTZ-PTV) signal levels, and voxels having signal levels in between the foreground and background thresholds - and the following formula: NMI(X;Y) = [[I(X; Y)] / [max(H(X), H(Y))]], where I(X; Y) is mutual information and max(H(X), H(Y)) is the maximum entropy of the real-time image (X) and pre-scan image (Y). FIG. 11 depicts examples of a plot that represents misaligned images (left panel) and a plot that represents aligned images (right panel). The X and Y axes of the plots in FIG. 11 represent the intensity bins of two images, A and B. For a given pixel in image A the intensity is plotted in the appropriate bin along the X axis and the intensity bin for the corresponding (same) pixel in image B is plotted along the Y axis. The result is a scatter plot of image pixel pairs which is representative of a JPD. The increased variance exists for the cluster distributions in the misaligned plot 1102 relative to the aligned plot 1104. The relatively decreased variance (entropy) on plot 1104 indicates that the images compared have more mutual information than those compared with plot 1102. Method 1000 may include calculating 1016 a real-time image metric using the registered images (e.g., a measurement of normalized mutual information) and comparing 1018 the real-time metric to a predetermined beam-interlock threshold (e.g., a measurement of NMI that is about equal to or less than 0.5). Simultaneously, a notification of areal-time status of the ROI may be indicated 1018 and beam interlock may be triggered 1018 if the real-time image metric is below the predetermined beam-interlock threshold. The method 1000 may include performing 1020 further calculations using real-time image metrics calculated in previous steps (e.g., using real-time PTV peak activity and / or real-time BTZ-PTV peak activity values calculated in step 1010). For example, the real-time PTV signal activity value may be used to calculate 1020 the real-time foreground (PTV) signal mean and standard deviation for each of the set of real-time images and the corresponding volume of pre-scan images, and a z-test may be used to compare 1022 the similarity of the PTV activity in the realtime images and pre-scan images via a z-score. That is, a low z-score may indicate that the realtime images match the pre-scan images. A notification of a real-time status of the ROI may be indicated and beam interlock may be triggered if the results of the calculation using the real-time image metric is outside of a predetermined beam-interlock range (e.g., is outside of an acceptable range of z-scores). As another example, the method 1000 may optionally include performing 1020 further calculations using the real-time images acquired in step 1008. For example, a number of voxels within the BTZ-PTV region of the volume of real-time images having signal values larger than the minimum activity threshold may be counted. Next, the voxel count may be compared to a predetermined voxel count threshold to determine if the total voxel count is low enough to deliver treatment (i.e., no OAR has entered the BTZ, as described above). Simultaneously, a notification of a real-time status of the ROI may be indicated 1022 and beam interlock may be triggered 1022 if the results of the calculation using the real-time images is outside of a predetermined beam-interlock range (e.g., if the total voxel count is too high, beaminterlock may be triggered).

[0099] While certain variations are described and illustrated herein, those of ordinary skill in the art will readily envision a variety of other means and / or structures for performing the function and / or obtaining the results and / or one or more of the advantages described herein, and each of such variations and / or modifications is deemed to be within the scope of the inventive variations described herein. More generally, those skilled in the art will readily appreciate that all parameters, dimensions, materials, and configurations described herein are meant to be exemplary and that the actual parameters, dimensions, materials, and / or configurations will depend upon the specific application or applications for which the inventive teachings is / are used. Those skilled in the art will recognize or be able to ascertain using no more than routine experimentation, many equivalents to the specific inventive variations described herein. It is,therefore, to be understood that the foregoing variations are presented by way of example only and that, within the scope of the appended claims and equivalents thereto; inventive variations may be practiced otherwise than as specifically described and claimed. Inventive variations of the present disclosure are directed to each individual feature and / or method described herein. In addition, any combination of two or more such features and / or methods, if such features and / or methods are not mutually inconsistent, is included within the inventive scope of the present disclosure.

[0100] The above-described methods can be implemented in any of numerous ways. For example, at least some methods of the present technology may be implemented using hardware, firmware, software, or a combination thereof. When implemented in firmware and / or software, the firmware and / or software code can be executed on any suitable processor or collection of logic components, whether provided in a single device or distributed among multiple devices.

[0101] In this respect, various aspects described herein may be embodied as a computer readable storage medium (or multiple computer readable storage media) (e.g., a computer memory, one or more floppy discs, compact discs, optical discs, magnetic tapes, flash memories, circuit configurations in Field Programmable Gate Arrays or other semiconductor devices, or other non-transitory medium or tangible computer storage medium) encoded with one or more programs that, when executed on one or more computers or other processors, perform methods that implement the various embodiments of the invention discussed above. The computer readable medium or media can be transportable, such that the program or programs stored thereon can be loaded onto one or more different computers or other processors to implement various aspects of the present invention as discussed above.

[0102] The terms “program” or “software” are used herein in a generic sense to refer to any type of computer code or set of computer-executable instructions that can be employed to program a computer or other processor to implement various aspects of embodiments as discussed above. Additionally, it should be appreciated that according to one aspect, one or more computer programs that when executed perform methods disclosed herein need not reside on a single computer or processor but may be distributed in a modular fashion amongst a number of different computers or processors to implement various aspects of the inventions disclosed herein.

[0103] Computer-executable instructions may be in many forms, such as program modules, executed by one or more computers or other devices. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Typically, the functionality of the program modules may be combined or distributed as desired in different variations.

[0104] Also, data structures may be stored in computer-readable media in any suitable form. For simplicity of illustration, data structures may be shown to have fields that are related through location in the data structure. Such relationships may likewise be achieved by assigning storage for the fields with locations in a computer-readable medium that may convey the one or more relationships between the fields. However, any suitable mechanism may be used to establish a relationship between information in fields of a data structure, including through the use of pointers, tags or other mechanisms that establish relationship between data elements.

[0105] Also, the acts performed as part of the method may be ordered in any suitable way. Accordingly, various methods may be constructed in which acts are performed in an order different than illustrated, which may include performing some acts simultaneously, even though shown as sequential acts in illustrative examples.

Claims

CLAIMS1. A method for real-time radiotherapy treatment, comprising: generating a set of pre-scan images comprising a region of interest (ROI) within a patient; acquiring a set of real-time images comprising the ROI; comparing the set of real-time images to the set of pre-scan images; and terminating a radiotherapy treatment session if the set of real-time images deviates from the set of pre-scan images by a predetermined threshold.

2. The method of claim 1, further comprising calculating a pre-scan image metric, and calculating a real-time image metric, wherein comparing the set of real-time images of the ROI to the set of pre-scan images of the ROI comprises comparing the pre-scan image metric and the real-time image metric.

3. The method of claim 1, further comprising generating a graphical or other notification of a deviation status of the set of real-time images from the set of pre-scan images.

4. The method of claim 1, wherein generating the set of pre-scan images comprises: acquiring a subset of pre-scan images at each of a plurality of control points, and combining the subsets of pre-scan images.

5. The method of claim 4, further comprising combining a plurality of subsets of pre-scan images corresponding to the plurality of control points to generate a summed pre-scan image of the ROI.

6. The method of claim 4, wherein each of the plurality of control points comprises a beam station within a radiotherapy treatment system configured to deliver radiotherapy to a portion of the ROI within the patient.

7. The method of claim 1, wherein acquiring the set of real-time images comprises acquiring one or more subsets of real-time images at a control point.

8. The method of claim 1, further comprising delivering radiotherapy treatment to the patient from a control point if the set of real-time images does not deviate from the set of pre-scan images by the predetermined threshold.

9. The method of claim 1, wherein the ROI comprises an inner ROI within an outer ROI.

10. The method of claim 9, further comprising: identifying the inner ROI prior to the real-time radiotherapy treatment, and localizing the outer ROI prior to the real-time radiotherapy treatment.

11. The method of claim 2, wherein the set of pre-scan images comprises a summed pre-scan image of a plurality of subsets of pre-scan images, and wherein calculating the pre-scan image metric comprises calculating a self-cross correlation between the summed pre-scan image each of the plurality of subsets of pre-scan images, and wherein calculating the real-time image metric comprises calculating a local cross correlation between the set of real-time images and the summed pre-scan image.

12. The method of claim 11, wherein each of the pre-scan image metric and the real-time image metric comprises one or more of a maximum cross correlation signal intensity for the control point and a location of the maximum cross correlation signal intensity for the control point.

13. The method of claim 2, wherein each of the real-time image metric and the pre-scan image metric comprise a rate of change of a maximum signal intensity.

14. The method of claim 13, wherein the set of pre-scan images comprises a summed pre-scan image comprising a plurality of subsets of pre-scan images, the method further comprising: calculating the rate of change of the maximum signal intensity for the plurality of subsets of pre-scan images to determine the pre-scan image metric, generating a hybrid image comprising the ROI by substituting the set of real-time images for a corresponding subset of pre-scan images within the summed pre-scan image, andcalculating the rate of change of the maximum signal intensity for the hybrid image to determine the real-time image metric.

15. The method of claim 13, wherein the ROI comprises an inner ROI within an outer ROI, and wherein the maximum signal intensity comprises a maximum signal intensity within the inner ROI.

16. The method of claim 15, wherein the rate of change of the maximum signal intensity is indicative of movement of the inner ROI with respect to the outer ROI.

17. The method of claim 2, wherein each of the real-time image metric and the pre-scan image metric comprise a rate of change of a standard deviation of a signal.

18. The method of claim 17, wherein the set of pre-scan images comprises a summed pre-scan image comprising plurality of subsets of pre-scan images, the method further comprising: calculating the rate of change of the standard deviation of the signal for each of the plurality of subsets of pre-scan images to determine the pre-scan image metric, generating a hybrid image comprising the ROI by substituting the set of real-time images for a corresponding subset of pre-scan images within the summed pre-scan image, and calculating the standard deviation of the signal for the hybrid image to determine the real-time image metric.

19. The method of claim 18, wherein the pre-scan image metric comprises a maximum rate of change of the standard deviation of the signal intensity among the plurality of subsets of prescan images.

20. The method of claim 17, wherein the ROI comprises an inner ROI within an outer ROI, and wherein calculating the standard deviation of the signal comprises calculating the standard deviation of the signal within the outer ROI, excluding the signal within the inner ROI.

21. The method of claim 20, wherein the rate of change of the standard deviation of the signal is indicative of movement of anatomical object into the outer ROI if the set of real-time images deviates from the set of pre-scan images by the predetermined threshold.

22. The method of claim 2, wherein the ROI comprises a volumetric shell comprising an inner ROI within an outer ROI, and wherein each of the real-time image metric and the pre-scan image metric comprise a mean intensity of the ROI.

23. The method of claim 22, method further comprising: adjusting a signal intensity of the inner ROI; calculating the mean intensity of the ROI using the adjusted signal of the inner ROI to determine the pre-scan image metric, and calculating the mean intensity of the ROI for the set of real-time images to determine the real-time image metric.

24. The method of claim 23, wherein adjusting the signal intensity of the inner ROI comprises: duplicating a signal of the inner ROI, and shifting the duplicated signal to a plurality of positions within the outer ROI, wherein calculating the pre-scan image metric comprises calculating the mean intensity of the ROI at each of the plurality of positions.

25. The method of claim 24, wherein the duplicated signal of the inner ROI is used to simulate a presence of an anatomical object within the ROI.

26. The method of claim 23, wherein calculating the mean intensity of the ROI for the set of real-time images comprises applying a weight factor to a calculation of the mean intensity of the ROI.

27. The method of claim 2, wherein each of the pre-scan image metric and the real-time image metric comprises two or more image metrics.

28. The method of claim 1, wherein the method is executed by a first module within a radiotherapy treatment system, and wherein terminating the radiotherapy treatment session comprises transmitting instructions from the first module to a second module configured to control hardware of the radiotherapy treatment system.

29. The method of claim 1, further comprising generating one or more of a visual notification, an audio notification, or a haptic notification indicative of a deviation status of the set of real-time images from the set of pre-scan images.

30. The method of claim 1, wherein terminating the radiotherapy treatment session comprises: automatically saving a record of radiotherapy treatment delivered to a patient, and generating an updated set of pre-scan images comprising the ROI.

31. The method of claim 1, wherein each of the set of pre-scan images and the set of real-time images comprises 3D images of at least a portion of the ROI.

32. The method of claim 1, further comprising defining the predetermined threshold based on the set of pre-scan images, wherein the predetermined threshold comprises one or more predetermined thresholds.

33. The method of claim 1, wherein each of the set of pre-scan images and the set of real-time images comprises one of: PET images, SPECT images, CT images, MRI images X-ray images, ultrasound images, or combinations thereof.

34. A method for monitoring a region of interest (ROI) within a patient during a radiotherapy treatment session, comprising: acquiring a plurality of subsets of pre-scan images comprising at least a portion of the ROI at a plurality of corresponding beam stations; combining a plurality of subsets of the pre-scan images corresponding to the plurality of beam stations to generate a summed pre-scan image of the ROI; for each of the plurality of beam stations:combining one or more sets of sequential pre-scan images to generate a subset pre-scan image of the summed pre-scan image, and calculating a self-cross correlation between the subset image corresponding to the beam station and the summed pre-scan image, acquiring a set of real-time images at the beam station, combining the set of real-time images with a remainder of one or more sets of sequential real-time images to generate a subset real-time image of the ROI, wherein the one or more sets of sequential real-time images correspond to the one or more sets of sequential pre-scan images for the beam station, calculating a local cross correlation between the subset real-time image of the ROI and the summed pre-scan image of the ROI, comparing the local cross correlation to the self-cross correlation to determine if the local cross correlation is within an acceptable range, generating a graphical notification to indicate whether the local cross correlation is within the acceptable range, and optionally modifying a radiation treatment for the patient when the local cross correlation is not within the acceptable range.

35. A method for monitoring a region of interest (ROI) within a patient during a radiotherapy treatment session, comprising: acquiring a plurality of subsets of pre-scan images comprising at least a portion of the ROI at a plurality of corresponding beam stations; combining the plurality of sets of pre-scan images corresponding to the plurality of beam stations to generate a summed pre-scan image of the ROI; for each of the plurality of beam stations, combining one or more sets of sequential prescan images to generate a subset pre-scan image of the summed pre-scan image; calculating a rate of change of a maximum pre-scan image signal for the summed prescan image of the ROI using a plurality of subset pre-scan images corresponding to the plurality of beam stations; acquiring a set of real-time images comprising at least a portion of the ROI; calculating a maximum real-time image signal for the set of real-time images; generating a hybrid image of the ROI by substituting the set of real-time images for acorresponding set of pre-scan images within the summed pre-scan image; calculating the rate of change of the maximum real-time image signal for the hybrid image; comparing the rate of change of the maximum real-time image signal to the rate of change of the maximum pre-scan image signal; generating a graphical notification to indicate whether the maximum real-time image signal is within an acceptable range; and optionally modifying a radiotherapy treatment for the patient if the maximum real-time image signal is not within the acceptable range.

36. A method for monitoring a region of interest (ROI) within a patient during a radiotherapy treatment session, comprising: acquiring a plurality of subsets of pre-scan images comprising at least a portion of the ROI at a plurality of corresponding beam stations; generating a summed pre-scan image of the ROI; at each of the plurality of beam stations, calculating a pre-scan minimum signal activity metric; determining an overall minimum signal activity threshold; acquiring a set of real-time images of the ROI; calculating a real-time signal activity metric; comparing the real-time signal activity metric to the overall minimum signal activity threshold; if the real-time signal activity metric is greater than the overall minimum signal activity threshold, registering the set of real-time images with a set of corresponding pre-scan images; calculating a normalized mutual information metric (NMI) using the registered set of real-time and corresponding set of pre-scan images; if the NMI metric is above a predetermined NMI threshold, calculating a z-score to compare the set of real-time images to the summed pre-scan image; generating a graphical notification to indicate whether the z-score is within an acceptable range; andoptionally modifying a radiotherapy treatment for the patient if the z-score is not within the acceptable range.

Images