Method and system for projection binning for image reconstruction

Abstract

Methods and systems for projection binning for image reconstruction are disclosed herein where areas of projections are binned based on a comparison of a characteristic indicative of how a physiological cycle of the patient impacts anatomy represented in said projection area at the instant in time at which the projection was acquired, with a reference value for that characteristic. Each bin corresponds to a portion of the physiological cycle and has a reference value of the characteristic for the portion of the physiological cycle relating to that bin, wherein each projection area is binned based on a similarity of its corresponding characteristic to the reference value.

Description

Field of the Invention Embodiments of the present invention described herein relate to methods and systems for projection binning for image reconstruction. More specifically, the present invention relates to a computer-implemented method for projection binning for image reconstruction, and data processing apparatuses, computer programs, and non-transitory computer-readable storage mediums configured to execute methods for projection binning for image reconstruction. Background of the Invention Radiotherapy (RT) is one of the cornerstones of cancer treatment, using ionizing radiation to eradicate tumor cells. A total radiation dose for a patient is typically divided into 1 to 30 daily fractions to optimize its effect. As the surrounding normal tissue is also sensitive to radiation, highly accurate delivery is a key part of effective RT. Imaging plays a crucial role in adaptive radiotherapy by providing on-line and real-time information on tumour motion and anatomical changes, guiding adaptive planning and replanning strategies, and verifying treatment delivery accuracy throughout the course of treatment. This dynamic approach improves treatment outcomes by maximizing tumour control while minimizing radiation exposure to healthy tissues. One example of imaging used for RT is 4D Cone Beam Computed Tomography (4D CBCT). 4D CBCT reconstruction is a technique used to reconstruct volumetric images over time from a series of 2D projection images acquired during a physiological dynamic imaging process, e.g. a respiratory or cardiac cycle. Unlike traditional 3D CBCT, which reconstructs a static volume from a single snapshot of projections, 4D CBCT reconstruction captures temporal changes in the patient's anatomy. During a 4D CBCT scan, a series of 2D projection images are acquired over multiple time points as the X-ray source and detector rotate around the patient. These projection images are captured at different phases of the respiratory or cardiac cycle, depending on the specific imaging protocol. The acquired projection data is organised into different bins corresponding to specific points in the motion cycle (e.g. a respiratory motion cycle or a cardiac motion cycle). For example, for respiratory motion, the projection data may be binned based on respiratory phase information obtained from respiratory motion surrogates. Typically, the respiratory signal is divided into a predetermined number of temporal bins representing different points in the respiratory cycle (e.g., end-inhalation, end-exhalation). Each projection is assigned to the appropriate bin based on its temporal alignment with the respiratory signal. This binning process ensures that projection images acquired at similar phases of the motion cycle are grouped together for reconstruction. Each bin of sorted projection images is reconstructed independently to generate a 3D volumetric image corresponding to that specific temporal phase. Various reconstruction algorithms can be used, such as filtered back projection (FBP), iterative reconstruction, or motion compensated methods, to reconstruct the 3D volume from the sorted projections. After reconstructing the 3D volumes for each temporal phase, these volumes are integrated or combined to create a cohesive 4D dataset representing the patient's anatomy over time. Summary of the Invention It is an aim of the present disclosure to at least partially address one or more of the challenges mentioned above. The invention is defined in the independent claims, to which reference should now be made. Further features are set out in the dependent claims. According to an aspect of the invention, there is provided a method for projection binning for image reconstruction. The method comprises receiving projection data comprising a plurality of two-dimensional (2D) projections of a patient volume, each of the plurality of 2D projections being acquired at an instant in time. The method further comprises, for each of the plurality of 2D projections, defining a projection area of the 2D projection, wherein the projection area is a variable parameter which can range from a single pixel of the 2D projection to the entirety of the 2D projection; and extracting at least one characteristic from the projection area, the characteristic being indicative of how a physiological cycle of the patient impacts anatomy represented in the projection area at the instant in time at which the 2D projection was acquired, such that each projection area has at least one corresponding characteristic. The method further comprises binning the projection areas into a plurality of bins, each bin corresponding to a portion of the physiological cycle, each bin having a reference value (or a range of reference values) of the characteristic for the portion of the physiological cycle relating to that bin, wherein each projection area is binned based on a similarity of its corresponding characteristic to the reference value. Several advantages are obtained from embodiments according to the above-described aspect. Embodiments of the present invention allow for more projection data to be included in each bin, whilst ensuring that the projections included within a given bin do not have adverse effects on the reconstruction. Embodiments of the present invention may reduce aliasing artifacts and improve reconstruction quality compared to known reconstruction binning approaches. Embodiments of the present invention may also be advantageous for motion compensated (MoCo) single or multiple phase reconstruction techniques. Through increased reconstruction quality, embodiments of the present invention may allow for the estimation of better motion models, which can then be used for improved MoCo reconstruction. In some embodiments, if the projection area is a portion of the 2D projection, repeating the extracting step for further portion(s) of the 2D projection, until every pixel of the 2D projection has been included in at least one portion. In some embodiments, the reference value for each bin is a predicted value for what the characteristic would be if the projection area was acquired during the portion of the physiological cycle corresponding to that bin. In some embodiments, each projection area is binned in: (i) none of the plurality of bins; (ii) one or more of the plurality of bins; or (iii) each of the plurality of bins. In some embodiments, the projection area comprises a single pixel, a plurality of pixels, or every pixel of the 2D projection. In some embodiments, the binning of each projection area is weighted based on the similarity, such that projection data with a higher similarity has a higher weight than projection data with a lower similarity. In some embodiments, one or more of the projection areas are weighted such that they do not contribute to at least one bin of the plurality of bins. In some embodiments, the projection areas are weighted using binary weights. In some embodiments, the at least one characteristic comprises a motion surrogate. In some embodiments, the motion surrogate is obtained by: defining a projection stack comprising the plurality of 2D projections; selecting a plurality of sub-volumes of the projection stack; extracting a candidate motion surrogate signal from projection data contained within each sub-volume of the projection stack to create a plurality of candidate motion surrogate signals; and combining the plurality of candidate motion surrogate signals into a single final motion surrogate signal to obtain the motion surrogate. In some embodiments, the physiological cycle is a respiratory cycle or a cardiac cycle. In some embodiments, the physiological cycle is a respiratory cycle and the plurality of bins are arranged assuming symmetry for exhalation and inhalation states of the respiratory cycle. In some embodiments, the plurality of bins are arranged such that two or more of the plurality of bins overlap. In some embodiments, the overlap of the two or more of the plurality of bins is controlled by an overlap parameter. In some embodiments, the method further comprises reconstructing a volumetric image using the projection data, based on the binning. In some embodiments, the projection data serves for cone-beam computed tomography, CBCT, reconstruction. In some embodiments, the method may be applied in adaptive radiotherapy. Embodiments of another aspect include a data processing apparatus comprising a memory storing computer-readable instructions and a processor. The processor (or controller circuitry) is configured to execute the instructions to carry out the method for projection binning for image reconstruction. Embodiments of another aspect include a computer program comprising instructions, which, when executed by computer, causes the computer to execute the method for projection binning for image reconstruction. Embodiments of another aspect include a non-transitory computer-readable storage medium comprising instructions, which, when executed by a computer, cause the computer to execute the method for projection binning for image reconstruction. Other features of the disclosure are described below and recited in the appended claims. The invention may be implemented in digital electronic circuitry, or in computer hardware, firmware, software, or in combinations thereof. The invention may be implemented as a computer program or a computer program product, i.e., a computer program tangibly embodied in a non-transitory information carrier, e.g., in a machine-readable storage device or in a propagated signal, for execution by, or to control the operation of, one or more hardware modules. A computer program may be in the form of a stand-alone program, a computer program portion, or more than one computer program, and may be written in any form of programming language, including compiled or interpreted languages, and it may be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a data processing environment. The invention is described in terms of particular embodiments. Other embodiments are within the scope of the following claims. For example, the steps of the invention may be performed in a different order and still achieve desirable results. Elements of the invention have been described using the terms “processor”, “input device” etc. The skilled person will appreciate that such functional terms and their equivalents may refer to parts of the system that are spatially separate but combine to serve the function defined. Equally, the same physical parts of the system may provide two or more of the functions defined. For example, separately defined means may be implemented using the same memory and / or processor as appropriate. Brief Description of the Drawings Embodiments of the invention will now be further described by way of example only and with reference to the accompanying drawings, wherein like reference numerals refer to like parts, and wherein: Figure 1 is a flow chart of a method for projection binning for image reconstruction, according to an embodiment; Figure 2 illustrates an example of how respiratory motion can impact projections; Figure 3 illustrates the benefit of assuming symmetry in the breathing motion; Figure 4 illustrates the effect of overlapping bins; Figure 5 illustrates the effect of per-pixel weighting; Figure 6 is a diagram of the geometric arrangement for cone beam projection scanning and reconstruction; Figure 7 is a diagram of a cone beam CT scanner suitable for obtaining projection data for use in a method for projection binning for image reconstruction, according to an embodiment; and Figure 8 is a radiotherapy system, suitable for using the method for image reconstruction, according to embodiments. Detailed Description Overview Embodiments of the present invention improve upon known binning processes for image reconstruction, e.g. 4D image reconstruction, motion compensated image reconstruction. Embodiments of the present invention harness the idea that the approach of sorting projections into discrete ‘bins’ (each bin corresponding to a particular phase of a motion cycle) can be considerably relaxed and generalized. In embodiments of the present invention, every pixel of every projection can be included in every bin, and its contribution to the reconstruction of that bin can be weighted. Instead of binning projections purely based on the time (within the motion cycle) at which they were acquired (i.e. temporal binning), embodiments of the present invention bin projections (or parts of projections) based on their similarity to a prediction of what the projection (or the part of the projection) should look like if it were acquired at a certain point in the motion cycle. This is a weaker requirement than temporal binning, and thus allows more projection data to contribute to each bin. Further, the contribution of each projection (or part of a projection) to the reconstruction can be weighted based on their determined similarity. Various aspects and details of these principal concepts will be described below with reference to Figures 1 to 8. Figure 1 is a flow chart illustrating a method for projection binning for image reconstruction 100 according to embodiments of the present invention. Method 100 comprises steps 102, 104, 106 and 108. Step 102 comprises receiving projection data comprising a plurality of 2D projections of a patient volume, each of the plurality of 2D projections being acquired at an instant in time. The 2D projections may be acquired sequentially and continuously during a scan. The projection data may be acquired or received from external computational means, such as a data storage server, or directly from an imaging or treatment system. The projection data comprises projections, which are representative of measured ray intensities corresponding to attenuated rays of radiation. Beams of rays are emitted from some radiation source, passed through a region of the patient, and subsequently detected at a detector. The detector and the radiation source may be configured on or in a gantry and may be rotated about an axis of the patient. Equivalently this rotation is about a radiation isocentre, which is the point (or centre of mass of a small bound region) in space through which rays intersect when the detector and the radiation source are rotated during beam-on. The techniques herein are equally applicable to static detectors and static radiation sources, arranged circumferentially about an axis of the patient. Similarly, the techniques herein are equally applicable to a combination of a static detector and rotating radiation sources, or a rotating detector and static radiation sources. For each of the plurality of 2D projections, steps 104 and 106 are performed. Step 104 comprises defining a projection area of the 2D projection, wherein the projection area is a variable parameter which can range from a single pixel of the 2D projection to the entirety of the 2D projection. In other words, the projection area may comprise any number of pixels of the 2D projection. The projection area may comprise a single pixel, a plurality of pixels, or every pixel of the 2D projection. The projection area may be chosen based on a user input (e.g. selecting an area of interest, or selecting the areas expected to have significant motion). In other cases, the projection area may be automatically chosen (e.g. the air around the patient may be considered one projection area and extracted by thresholding, the table may be considered another projection area). Alternatively, motion surrogates which provide some spatial localization of the movement could be used to define regions according to some criteria. Generally, it’s better to make the projection areas as small as possible, but richer prior knowledge of the motion is required to do this. It is advantageous to make the projection areas as small as possible given the specific knowledge of the motion available. Step 106 comprises extracting at least one characteristic (e.g. a motion surrogate) from the projection area. The characteristic is indicative of how a physiological cycle (e.g. a respiratory cycle, a cardiac cycle, etc.) of the patient impacts anatomy represented in the projection area at the instant in time at which the 2D projection was acquired. For example, the respiratory cycle may impact the position of the anatomy due to breathing motion, or the cardiac cycle may impact the position of the anatomy due to cardiac motion. Each projection area has at least one corresponding characteristic. In some cases, the impact may be zero (in these cases, the projection area may be advantageously binned into every bin). For example, some areas of anatomy are not affected (or minimally affected) by breathing motion (or cardiac motion), such as the back area. Step 108 comprises binning the projection areas into a plurality of bins, each bin corresponding to a portion of the physiological cycle and having a reference value (or a range of reference values) of the characteristic for the portion of the physiological cycle relating to that bin. Each projection area is binned based on a similarity of its corresponding characteristic to the reference value. The reference value quantifies what you would expect the characteristic to be for that portion of the cycle. For example, the reference value may be a predicted or idealized or optimal value of the characteristic at that portion. In some examples, the reference value may be based on historical data or simulated data. In other examples, the reference value may be based on the projection data, for example, a projection may be selected and the surrogate value of that projection may have value x. A bin could then be defined with reference value x, and that bin should then contain all the projections taken at around the same point in the motion cycle as the originally selected projection. In other words, a physiological cycle (e.g. cardiac or respiratory) is divided into portions. These portions may be equally spaced (e.g. the cycle may be divided equally into 10 bins, each bin representing 10% of the cycle). These portions may be based on specific phases of the cycle. For example, for the respiratory cycle, the cycle may be divided into specific phases such as inhalation, exhalation, and intermediate phases. Each portion is then assigned a bin. Each bin has a corresponding reference value of the characteristic. For example, if a bin is for the inhalation phase of a respiratory cycle, the reference value may be a predicted / idealized / optimal value for the characteristic during the inhalation phase. The projection area’s characteristic value is then compared to this reference value, and if it is above a similarity threshold, the projection area is included in the bin. If the projection area’s characteristic value is below the similarity threshold, the projection area is not included in the bin. This process is then repeated for every bin and for every projection area. It should be noted that the similarity threshold may vary for each bin. As such, it is possible for any projection area to end up in any bin. A projection area may be binned into none of the bins, one or more of the bins, or every bin. In some cases, the projection area may be a portion of the 2D projection, i.e. the projection area is not the entirety of the projection. In this case, a further projection area of the 2D projection may be defined 104 (the further projection area being a further portion of the 2D projection). The extraction step 106 is then repeated for the further projection area such that the further projection area has a corresponding characteristic. This process may be repeated until every pixel of the 2D projection has been included in at least one portion / projection area. In some cases, one pixel may be included in more than one projection area (i.e. the projection areas may overlap). Projection areas may be binned using weights (i.e. rather than being in a bin or not, a projection area can be e.g. 30% in a bin). In some embodiments, the binning of each projection area is weighted based on the similarity, such that projection data with a higher similarity to the reference value has a higher weight than projection data with a lower similarity to the reference value. This weight may be different in different bins for the same projection area (e.g. the same projection area may be included in all bins with different weight in each). In some cases, one or more of the projection areas may be weighted such that they do not contribute to at least one bin of the plurality of bins. In other words, the weighting could be zero and that projection area will not be used when reconstructing from that bin. In some embodiments, the projection areas may be weighted using binary weights, i.e., the projection areas are either included in a bin or they are not. There is no weighted contribution. In other words, a subset of pixels of the 2D projection may be selected for each bin. In some embodiments where the characteristic is a motion surrogate, the motion surrogate may be obtained via the following steps: 1. Defining a projection stack comprising the plurality of 2D projections. 2. Selecting a plurality of sub-volumes of the projection stack. 3. Extracting a candidate motion surrogate signal from projection data contained within each sub-volume of the projection stack to create a plurality of candidate motion surrogate signals. 4. Combining the plurality of candidate motion surrogate signals into a single final motion surrogate signal to obtain the motion surrogate. This approach of obtaining a motion surrogate from projection data uses different (but potentially overlapping) regions (also referred to as sub-volumes) of the data to obtain multiple (inherently noisy) candidate surrogate signals. A sub-volume of a projection stack refers to a subset (or all) of the acquired 2D projection images that cover a fixed sub-region of the detector. These candidate surrogates contain various levels of information about the breathing or cardiac motion, and various degrees of noise. These candidate signals are then advantageously combined into a single (clean) final surrogate signal. The combination may be achieved using a weighted sum of the candidates. The weight may be calculated using a heuristic which assigns higher weights to less noisy and more regular candidates. Zero weight may be assigned to candidates deemed too poor by these metrics, to avoid these degrading the final surrogate. Fundamentally, this approach may be thought of as a method to extract and combine multiple noisy (and easy to obtain) motion surrogates into a single clean motion surrogate. Unlike previous methods, this method does not require specific anatomy to be imaged, nor a specific field of view (FOV). Further detail of this approach may be found in Elekta Limited’s co-filed application titled “Method and System for Obtaining a Motion Surrogate Signal” with Agent Reference P149642GB00 and Application No. GB2406438.8. In embodiments where the physiological cycle is a respiratory cycle, the plurality of bins may be arranged assuming symmetry for exhalation and inhalation states of the respiratory cycle. This is described in detail in Example 2 below. In some embodiments, the plurality of bins are arranged such that two or more of the plurality of bins overlap, i.e. there may be some particular values of the characteristic that fall within the range of more than one bin. This overlap may be controlled by an overlap parameter. This is described in detail in Example 3 below. The method 100 may further comprise reconstructing a volumetric image using the projection data, based on the binning. Each bin of projection data may be reconstructed independently to generate a 3D volumetric image corresponding to the bin’s portion of the cycle. Examples of known suitable reconstruction techniques include a known Feldkamp, Davis, and Kress (FDK) reconstruction method, an iterative reconstruction method, an Al reconstruction method, a deep-learning based reconstruction method and a Polyquant reconstruction method. After reconstructing the 3D volumes for each bin, these volumes are integrated or combined to create a cohesive 4D dataset representing the patient's anatomy over time. The projection data may serve for CBCT reconstruction. The method 100 may be used in adaptive radiotherapy, which enables a patient’s treatment to be changed or adapted to respond to a signal that additional information is known about the patient or that the patient has changed from the original state at the time of planning. Examples of how the concept of the present invention may be implemented are described in detail below, but at a high level: • All pixels in projections that are in the correct phase for the bin can be weighted highly. • All pixels in all projections that correspond to rays passing through (approximately) static regions of the anatomy and / or the table can be weighted highly (referred to below as “safe projections”). • All remaining pixels can be weighted based on how close their values are expected to be to the values that would be seen if the projection was acquired in the correct phase for the bin. This can be variously estimated, but may be summarized as: o Projections from phases that are closer together in the breathing cycle tend to also be more geometrically similar than those from more separated phases. o Pixels corresponding to rays passing through anatomy that moves more should generally be weighted lower than those that correspond to rays that pass-through anatomy that moves less. The weighting for some pixels may be zero and binary weights may be used - selecting a subset of projection pixels for each bin. This concept of relaxation / generalization may be applied to a lesser or greater degree. For example, one could consider full projections (rather than individual pixels) but allow each projection to have a weight. There is a spectrum of options from the strictest (the default discrete binning) to the most general (all pixels for all projections individually weighted in every bin), and a suitable balance can be selected based on a particular use case, desired outcomes and constraints. Embodiments of the present invention may require some ability to know how similar a projection (or region of a projection) is to an imagined ‘ideal’ projection (or projection region) acquired at the same angle, but with the patient in a different motion state. In other words, estimating “how similar is this projection to what it would have been if it was acquired when the patient was in the desired / ideal motion state?”. This similarity may be estimated using, for example: • A motion surrogate, which discretely or continuously assigns a motion state to each projection. With this information we can assume that motion states that are temporally close together are also geometrically similar. This can be further refined by exploiting more detailed knowledge, such as of the rhythm of the breathing cycle, i.e., at which stages in the cycle lung movement is faster or slower. o This motion surrogate could come from many sources, including an additional signal acquired during acquisition via some form of motion sensor, or extracted as a post-processing step from the projection data itself. • A patient specific motion model, for example, created from a 4D CT scan of the same patient. From such a model, measures of similarity can be directly calculated, e.g. by simulating projections and measuring the differences. • Prior knowledge can also be leveraged, for example that the bones generally move less than soft tissue under breathing motion, with most motion occurring toward the front of the patient, and that the table is static, etc. Detailed Examples To illustrate embodiments of the present invention, four specific examples are set out below. These four examples may be used individually or combined with one another. Example 1: Including safe projections in all bins The reason for binning projection data is to keep projections corresponding to different motion states separate, so that the distinct reconstructions produce volumes with the anatomy in the different motion states, and so that artifacts resulting from motion during acquisition are minimized in each reconstruction. If a scanner is held fixed at acquisition angle X and continuously acquires projections over a breathing cycle, the variation in the acquired projections over the breathing cycle can be measured. This variation is called the ‘variation intensity’. For every acquisition angle X there is similarly a corresponding variation intensity. The variation intensity depends on the angle / FOV of the scan, the shape of the patient, the anatomical region being imaged, and the nature of the motion. For certain fixed acquisition angles the variation intensity will be lower than for others. The proposed idea here is that projections coming from angles with particularly low variation intensity are potentially safe and beneficial to include in all projection bins. As a specific example, some acquisition angles in a MFOV pelvis scan may only capture the table and anatomy below the central coronal slice of the patient, which is essentially unmoving during breathing motion. These projections may then beneficially be included in all bins. Figure 2 shows an example of ‘safe projections’ that can be used, for example, in all bins of a binned reconstruction. Subfigure (A) shows a sinogram-view slice from the projection volume 202, showing the horizontal axis of the projections (x) and the projection (i.e., time) axis (t). The projections have been split into three groups 204, 206, 208 along the time axis. In groups 1 and 3 the respiratory motion has little or no impact on the projections, whereas in group 2 respiratory motion does impact the projections. Evidence for this can be seen in the vertical stripe patterns and spikey shapes within region 2 of the sinogram-view slice, which are not present in regions 1 and 3. Subfigures (B), (C) and (D) show example projections 216, 218, 220 from regions 1, 2 and 3 respectively (the sketch 210, 212, 214 above each projection visualizes the approximate area of the patient that projection captures). As can be seen, the projection 218 from region 2 captures the patient’s stomach, which moves during breathing. In contrast, projections 210, 214 from regions 1 and 3 capture mainly the patient’s back, which is almost static under normal breathing motion. Thus, projections 216, 220 in regions 1 and 3 can be considered safe and may be included in all bins of the reconstruction. Example 2: Assuming symmetry in breathing motion The breathing motion cycles from maximum exhalation (ME) through inhalation states to maximum inhalation (Ml), then through exhalation states back to ME. In general, although the start / end points are shared, the motion states when transitioning from ME to Ml (the inhalation states) are not necessarily the same as those passed through when transitioning from Ml back to ME (the exhalation states). For this reason, bins are usually created spanning the full breathing cycle. However, even if not identical, exhalation and inhalation states are similar (and this similarity is higher still when restricting attention to certain anatomical regions). The proposed idea here is to assume sufficient similarity between inhalation and exhalation states for there to be benefit in combining them into shared bins. As an explicit example, assume the motion transition from ME to Ml is symmetric with that from Ml to ME. A transform can be applied to the motion surrogate values that captures this symmetry. For example, if originally the surrogate values, v, increase from 0 to 1 over the full cycle (with 0.5 = Ml), those values can be mapped to 0.5 - |v - 0.5| so that the surrogate values increase from 0 to 0.5 while transitioning from ME to Ml, then decreases from 0.5 to 0 when transitioning from Ml to ME. Now binning can be performed as normal, with the resulting bins assuming the symmetry. In practice, the number of bins is generally in the range of 6-10. In this case, assuming one bin for Ml and one bin for ME, there are normally k bins covering the inhalation states and k bins covering the exhalation states, resulting in a total of 2+2k bins (a typical value of k is 2 or 4). However, in the assumed symmetry case the k bins for inhalation states and the k bins for exhalation states become a single set of k bins, and thus there are 2+k bins in total. This results in a considerable increase in projections-per-bin. Figure 3 shows a demonstration of the benefit of assuming symmetry in the breathing motion. Subfigure (B) shows a standard binning approach, where the 996 projection scan is split into 6 bins (310, 312, 314, 316, 318, 320), each containing 166 projections. When symmetric binning is used instead (as seen in Subfigure (A)), 4 bins (302, 304, 306, 308) are reconstructed, each with 249 projections, and then bins 3 (306) and 2 (304) are repeated to provide images for a whole motion cycle. (Note that due to the way the bins are created here, bin 1 (310) with standard binning falls between bin 1 (302) and 2 (304) of the symmetric binning, and so on. In subfigure (C) a larger version of the third bin from each approach (which capture approximately the same motion state) is shown, with the standard binning approach on the left 322 and the symmetric binning strategy on the right 324. As can be seen, there is an improvement in image quality resulting from the symmetric assumption. Example 3: Using overlapping bins Using non-overlapping bins ensures that each bin captures a distinct motion state. However, as the number of bins increases, or the number of projections in the scan of interest decreases, the number of projections per bin becomes lower and aliasing artifacts in the reconstructions become more severe. The number of bins used is generally in the order of 6 to 10 and, for a given scan, is the only parameter determining the number of projections per bin. By introducing a second parameter specifying how much the bins should overlap, the number of projections per bin can be increased without altering the number of bins. This can help to alleviate artifacts whilst only minimally reducing the motion resolution. This may be thought of as decoupling how many bins you create and what makes a projection similar enough to be included within a bin. Standardly, for discrete bins, having more bins means tighter restrictions on the requirement for being in a bin. However, the number of bins created and the similarity constraint can actually be adjusted separately. Finally, it should be noted that a projection area’s inclusion in a bin may be weighted. For example, less similar projections may still be included in a bin, but with lower weight. Further, the inclusion criteria do not have to be uniform across all bins but could be tuned separately for each bin. Figure 4: Subfigure (B) shows a reconstruction 402 from a 996 projection scan split into 10 bins (about 100 projections per bin). Subfigure (A) shows the same scan 404 but reconstructed with 10 bins, each of 180 projections using overlapping bins. Subfigure (C) shows the absolute difference image 406. Reduced noise and sharper bone edges can be observed, at the cost of slight blurring of the diaphragm. Example 4: Weighting projection pixels Some anatomical regions, such as the lungs, diaphragm and surrounding soft tissue move considerably during the breathing cycle. Other anatomy, such as the spine, hips, and back muscles (as well as the table itself) are effectively static during normal breathing. The underlying concept here is that any projection pixels corresponding to rays that pass though only anatomy that is static during breathing can be included in all bins. This can be generalised to weighting each pixel based on how static the anatomy its corresponding ray passed through is. Figure 5 shows a demonstration of the benefits of per-pixel weighting. Subfigure (A) shows a reconstruction 502 using all projections, as can be seen the diaphragm is blurred (pointed to by arrow 508). Subfigure (B) shows a reconstruction 504 of a single bin (1 / 6th of the projections) of the same data. In this image the diaphragm is sharp, but the image quality degrades overall (e.g., in the shoulders and on the bones). Subfigure (C) shows the result 506 of per-pixel weighting, which uses all projections, but weights each pixel in each projection. This results in a sharp diaphragm while maintaining the overall image quality. As Examples 1 to 4 set out above show, improved reconstruction quality is achievable by relaxing the standard projection binning approach in various ways, in line with the overarching principal of the present invention. In practice, these benefits can be achieved with motion surrogates, patient motion modelling, or even well-chosen heuristics (e.g., “anatomy close to the table tending to be more static”). Cone Beam Projection Acquisition Figure 6 illustrates the geometry of a typical cone beam projection acquisition setup 600, as may be used to obtain projection data for use in embodiments. For simplicity, only the x-y plane is illustrated; the skilled reader will appreciate that the beam of rays extends also into the z-axis, orthogonal to the x-y plane. In this example, the acquisition setup captures projection data of object 602. X-ray source 604 generates and emits X-rays 606 towards object 602. In CBCT, X-rays may be considered as a beam of rays, emitted from a point source. Detector (or detectors) 608 capture projections, which are sets of line integrals along paths that radiate from the source 604. Multiple projections of the image may be acquired from different angles by rotating the source 604 and detector 608 around the centre of the image 610. In the present example, the source 604 and detector 608 may be rotated arcuately along orbital path 612. In this way, the x-y plane may be rotated counterclockwise around the point of origin (or centre of the image 610) in a manner that keeps the mutual positional relationship between source 604 and detector 608 when passing through the orbital path 612. Other configurations of imaging systems are, of course, feasible; for instance, the source may be configured to rotate and a complete ring of detectors may be configured to capture projections, or there may be multiple sources arranged circumferentially around a complete ring of detectors. Further, the imaging system may rotate the source along a helical path, so as to capture projection data along the axis of the object 602 of interest (along the z-axis in the present example). The attenuation of the intensity of the rays that pass through the object 602 may be measured by processing signals received from the detector 608. By making projective measurements at a series of different projection angles through the object 602, a sinogram may be constructed from the projection data, mapping the spatial dimension of the detector array to the projection angle dimension. The intensity attenuation resulting from a particular volume within the object will trace out a sine wave for the spatial dimension along the detector perpendicular to the rotation axis of the system. Volumes of the object farther from the centre of rotation correspond to sine waves with greater amplitudes than those corresponding to volumes nearer the centre of rotation. The phase of each sine wave in the sinogram corresponds to the relative angular positions with respect to the rotation axis. By performing an image reconstruction technique (such as an inverse Radon transform) on the projection data in the sinogram, one may reconstruct an image, where the reconstructed image corresponds to a cross-sectional slice of the object 602. For the reconstruction, the projection data may be binned in accordance with the method for projection binning for image reconstruction described herein. The detector 608 may comprise underlying detector elements (such as pixels or bins of pixels), rather than a single active region. The measured ray intensities may then be taken as a function of underlying intensity signals acquired at detector elements of the detector. This is advantageous in that one reduces the impact of random noise, which may plague an intensity (particularly a low intensity) measured at a single detector element. For example, the measured ray intensities may be an average (mean, mode or median) of underlying single detector element intensities. Figure 7 illustrates the geometry of a cone beam projection acquisition setup, as incorporated into a medical CT scanner 700. Radiation sources 702 emit beams of X-ray radiation, which may pass through the patient supported on a couch within the CT scanner (not depicted). Detectors 704 capture attenuated X-rays. The radiation sources 702 and detectors 704 may be configured to rotate within a gantry of the CT scanner, so as to acquire projective measurements at a series of different projection angles. As shown, the angular separation between measurement orientations need not be equal. Referring back to the method of the disclosure, the received projection data may be acquired using the set up described above in relation to Figures 6 and 7. Radiotherapy System Figure 8 is a block diagram of an implementation of a radiotherapy system 800, suitable for executing methods according to embodiments. The example radiotherapy system 800 comprises a computing system 810 within which a set of instructions, for causing the computing system 810 to perform the method (or steps thereof) discussed herein, may be executed. The computing system 810 may implement an image reconstruction system. The image reconstruction system may implement methods described herein to bin projection data prior to reconstructing. The image reconstruction system may use any conventional image reconstruction technique suitable for use with the binned projection data. Examples of known suitable reconstruction techniques include a known Feldkamp, Davis, and Kress (FDK) reconstruction method, an iterative reconstruction method, an Al reconstruction method, a deep-learning based reconstruction method and a Polyquant reconstruction method. The computing system 810 may also be referred to as a computer. In particular, the methods described herein may be implemented by a processor or controller circuitry 811 of the computing system 810. The computing system 810 shall be taken to include any number or collection of machines, e.g., computing device(s), that individually or jointly execute a set (or multiple sets) of instructions to perform any one or more of the methods discussed herein. That is, hardware and / or software may be provided in a single computing device, or distributed across a plurality of computing devices in the computing system. In some implementations, one or more elements of the computing system may be connected (e.g., networked) to other machines, for example in a Local Area Network (LAN), an intranet, an extranet, or the Internet. One or more elements of the computing system may operate in the capacity of a server or a client machine in a client-server network environment, or as a peer machine in a peer-to-peer (or distributed) network environment. One or more elements of the computing system may be a personal computer (PC), a tablet computer, a set-top box (STB), a Personal Digital Assistant (PDA), a cellular telephone, a web appliance, a server, a network router, switch or bridge, or any machine capable of executing a set of instructions (sequential or otherwise) that specify actions to be taken by that machine. The computing system 810 includes controller circuitry 811 and a memory 813 (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM) such as synchronous DRAM (SDRAM) or Rambus DRAM (RDRAM), etc.). The memory 813 may comprise a static memory (e.g., flash memory, static random access memory (SRAM), etc.), and / or a secondary memory (e.g., a data storage device), which communicate with each other via a bus (not shown). Memory 813 may be used to store or buffer projection data until required for image processing. Controller circuitry 811 represents one or more general-purpose processors such as a microprocessor, central processing unit, accelerated processing units, or the like. More particularly, the controller circuitry 811 may comprise a complex instruction set computing (CISC) microprocessor, reduced instruction set computing (RISC) microprocessor, very long instruction word (VLIW) microprocessor, processor implementing other instruction sets, or processors implementing a combination of instruction sets. Controller circuitry 811 may also include one or more special-purpose processing devices such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), network processor, or the like. One or more processors of the controller circuitry may have a multicore design. Controller circuitry 811 is configured to execute the processing logic for performing the operations and steps discussed herein. The computing system 810 may further include a network interface circuitry 815. The computing system 810 may be communicatively coupled to an input device 820 and / or an output device 830, via input / output circuitry 816. In some implementations, the input device 820 and / or the output device 830 may be elements of the computing system 810. The input device 820 may include an alphanumeric input device (e.g., a keyboard or touchscreen), a cursor control device (e.g., a mouse or touchscreen), an audio device such as a microphone, and / or a haptic input device. The output device 830 may include an audio device such as a speaker, a video display unit (e.g., a liquid crystal display (LCD) or a cathode ray tube (CRT)), and / or a haptic output device. In some implementations, the input device 820 and the output device 830 may be provided as a single device, or as separate devices. In some implementations, the computing system 810 may comprise image processing circuitry 814. Image processing circuitry 814 may be configured to process image data 870 (e.g., images, imaging data, projections, projection data), such as medical images obtained from one or more imaging data sources, a treatment device 850 and / or an image acquisition device 840. Image processing circuitry 814 may be configured to process, or pre-process, image data 870. For example, image processing circuitry 814 may convert received image data into a particular format, size, resolution or the like. Image processing circuitry 814 may be configured to perform methods described herein. Image processing circuitry 814 may be configured to perform image reconstruction. In some implementations, image processing circuitry 814 may be combined with controller circuitry 811. In some implementations, the radiotherapy system 800 may further comprise an image acquisition device 840 and / or a treatment device 850. The image acquisition device 840 and the treatment device 850 may be provided as a single device. In some implementations, treatment device 850 is configured to perform imaging, for example in addition to providing treatment and / or during treatment. Image acquisition device 840 may be configured to perform positron emission tomography (PET), computed tomography (CT), magnetic resonance imaging (MRI), single positron emission computed tomography (SPECT), X-ray, and the like. Image acquisition device 840 may be configured to output image data 870, which may be accessed by computing system 810. Treatment device 850 may be configured to output treatment data 860, which may be accessed by computing system 810. Treatment data 860 may be obtained from an internal data source (e.g., from memory 813) or from an external data source, such as treatment device 850 or an external database. The various methods described above may be implemented by a computer program. The computer program may include computer code (e.g., instructions) arranged to instruct a computer to perform the functions of one or more of the various methods described above. For example, the steps of the methods described in relation to any of Figures 1 to 5 may be performed by the computer code. The steps of the methods described above may be performed in any suitable order. The computer program and / or the code for performing such methods may be provided to an apparatus, such as a computer, on one or more computer readable media or, more generally, a computer program product. The computer readable media may be transitory or non-transitory. The one or more computer readable media could be, for example, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, or a propagation medium for data transmission, for example for downloading the code over the Internet. Alternatively, the one or more computer readable media could take the form of one or more physical computer readable media such as semiconductor or solid-state memory, magnetic tape, a removable computer diskette, a random access memory (RAM), a read-only memory (ROM), a rigid magnetic disc, and an optical disk, such as a CD-ROM, CD-R / W or DVD. The instructions may also reside, completely or at least partially, within the memory 813 and / or within the controller circuitry 811 during execution thereof by the computing system 810, the memory 813 and the controller circuitry 811 also constituting computer-readable storage media. In an implementation, the modules, components and other features described herein may be implemented as discrete components or integrated in the functionality of hardware components such as ASICS, FPGAs, DSPs or similar devices. A “hardware component” is a tangible (e.g., non-transitory) physical component (e.g., a set of one or more processors) capable of performing certain operations and may be configured or arranged in a certain physical manner. A hardware component may include dedicated circuitry or logic that is permanently configured to perform certain operations. A hardware component may comprise a special-purpose processor, such as an FPGA or an ASIC. A hardware component may also include programmable logic or circuitry that is temporarily configured by software to perform certain operations. In addition, the modules and components may be implemented as firmware or functional circuitry within hardware devices. Further, the modules and components may be implemented in any combination of hardware devices and software components, or only in software (e.g., code stored or otherwise embodied in a machine-readable medium or in a transmission medium). Unless specifically stated otherwise, as apparent from the following discussion, it is appreciated that throughout the description, discussions utilizing terms such as “receiving”, “defining”, “extracting”, “binning”, “determining”, “comparing ”, “enabling”, “maintaining”, “identifying”, “obtaining”, “accessing”, or the like, refer to the actions and processes of a computer system, or similar electronic computing device, that manipulates and transforms data represented as physical (electronic) quantities within the computer system's registers and memories into other data similarly represented as physical quantities within the computer system memories or registers or other such information storage, transmission or display devices. While certain embodiments have been described, these embodiments have been presented by way of example only, and are not intended to limit the scope of the inventions. Indeed, the novel methods and apparatuses described herein may be embodied in a variety of other forms; 5 furthermore, various omissions, substitutions and changes in the form of methods and apparatus described herein may be made.

Claims

1. A method for projection binning for image reconstruction comprising:receiving projection data comprising a plurality of two-dimensional, 2D, projections of a patient volume, each of the plurality of 2D projections being acquired at an instant in time;for each of the plurality of 2D projections:defining a projection area of the 2D projection, wherein the projection area is a variable parameter which can range from a single pixel of the 2D projection to the entirety of the 2D projection; andextracting at least one characteristic from the projection area, the characteristic being indicative of how a physiological cycle of the patient impacts anatomy represented in the projection area at the instant in time at which the 2D projection was acquired, such that each projection area has at least one corresponding characteristic; andbinning the projection areas into a plurality of bins, each bin corresponding to a portion of the physiological cycle, each bin having a reference value of the characteristic for the portion of the physiological cycle relating to that bin, wherein each projection area is binned based on a similarity of its corresponding characteristic to the reference value.

2. The method of claim 1, wherein if the projection area is a portion of the 2D projection, repeating the extracting step for further portion(s) of the 2D projection, until every pixel of the 2D projection has been included in at least one portion.

3. The method of claim 1 or 2, wherein the reference value for each bin is a predicted value for what the characteristic would be if the projection area was acquired during the portion of the physiological cycle corresponding to that bin.

4. The method of any preceding claim, wherein each projection area is binned in:(i) none of the plurality of bins;(ii) one or more of the plurality of bins; or(iii) each of the plurality of bins.

5. The method of any preceding claim, wherein the projection area comprises a single pixel, a plurality of pixels, or every pixel of the 2D projection.

6. The method of any preceding claim, wherein the binning of each projection area is weighted based on the similarity, such that projection data with a higher similarity has a higher weight than projection data with a lower similarity.

7. The method of claim 6, wherein one or more of the projection areas are weighted such that they do not contribute to at least one bin of the plurality of bins.

8. The method of claim 6 or 7, wherein the projection areas are weighted using binary weights.

9. The method of any preceding claim, wherein the at least one characteristic comprises a motion surrogate.

10. The method of claim 9, wherein the motion surrogate is obtained by:defining a projection stack comprising the plurality of 2D projections;selecting a plurality of sub-volumes of the projection stack;extracting a candidate motion surrogate signal from projection data contained within each sub-volume of the projection stack to create a plurality of candidate motion surrogate signals; andcombining the plurality of candidate motion surrogate signals into a single final motion surrogate signal to obtain the motion surrogate.

11. The method of any preceding claim, wherein the physiological cycle is a respiratory cycle or a cardiac cycle.

12. The method of claim 11, wherein the physiological cycle is a respiratory cycle and the plurality of bins are arranged assuming symmetry for exhalation and inhalation states of the respiratory cycle.

13. The method of any preceding claim, wherein the plurality of bins are arranged such that two or more of the plurality of bins overlap.

14. The method of claim 13, wherein the overlap of the two or more of the plurality of bins is controlled by an overlap parameter.

15. The method of any preceding claim, further comprising reconstructing a volumetric image using the projection data, based on the binning.

16. The method of any of the preceding claims, wherein the projection data serves for conebeam computed tomography, CBCT, reconstruction.5 17. The method of any of the preceding claims, for use in adaptive radiotherapy.

18. A data processing apparatus comprising a memory storing computer-executable instructions, and a processor configured to execute the instructions to carry out the method of any of the preceding claims.1019. A computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of any of claims 1 to 17.

20. A computer-readable medium comprising instructions which, when executed by a 15 computer, cause the computer to carry out the method of any of claims 1 to 17.

Images