Systems and methods for photon count data compression

Abstract

Methods and systems are provided for compressing detector data in a computed tomography imaging system. In an example, a method for a photon counting computed tomography (PCCT) system includes, during a scan of an imaging subject, obtaining detector data from a photon counting detector of the PCCT system, the detector data comprising, for each pixel or detector element of the photon counting detector, photon counts partitioned into a plurality of energy bins based on an energy imparted by each photon on the photon counting detector, compressing the detector data into a respective set of eigenbin measurements for each pixel using one or more sets of eigenvectors identified during calibration of the PCCT system, decompressing each respective set of eigenbin measurements using the one or more sets of eigenvectors to form decompressed detector data, and reconstructing one or more images from the decompressed detector data.

Description

SYSTEMS AND METHODS FOR PHOTON COUNT DATA COMPRESSIONCROSS REFERENCE TO RELATED APPLICATIONS

[0001] The present application claims priority to U.S. Provisional Application No. 63 / 602,103, titled “SYSTEMS AND METHODS FOR PHOTON COUNT DATA COMPRESSION,'’ and filed November 22, 2023, the entire contents of which is hereby incorporated by reference for all purposes.TECHNICAL FIELD

[0002] Embodiments of the subject matter disclosed herein relate to imaging systems and methods, and more particularly, to data compression in computed tomography (CT) imaging systems.BACKGROUND

[0003] In computed tomography (CT) imaging systems, an electron beam generated by a cathode is directed towards a target within an X-ray source or X-ray tube. A fan-shaped or cone-shaped beam of X-rays produced by electrons colliding with the target is directed towards a subject, such as a patient. After being attenuated by the object, the X-rays impinge upon an array of X-ray detectors, generating an image. A quality of a CT image may be increased by using Photon Counting CT (PCCT). where the X-ray detectors are photon counting detectors, and photons are counted to provide spectral information.SUMMARY

[0004] In an example, a method for a photon counting computed tomography (PCCT) system includes, during a scan of an imaging subject, obtaining detector data from a photon counting detector of the PCCT system, the detector data comprising, for each pixel or detector element of the photon counting detector, photon counts partitioned into a plurality of energy7bins based on an energy imparted by each photon on the photon counting detector, compressing the detector data into a respective set of eigenbin measurements for each pixel using one or more sets of eigenvectors identified during calibration of the PCCT system, decompressing each respective set of eigenbin measurements using the one or more sets of eigenvectors to form decompressed detector data, and reconstructing one or more images from the decompressed detector data.

[0005] The above advantages and other advantages and features of the present description will be readily apparent from the following Detailed Description when taken alone or in connection with the accompanying drawings. It should be understood that the summary above is provided to introduce in simplified form a selection of concepts that are further described in the detailed description. It is not meant to identify key or essential features of the claimed subject matter, the scope of which is defined uniquely by the claims that follow the detailed description. Furthermore, the claimed subject matter is not limited to implementations that solve any disadvantages noted above or in any part of this disclosure.BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Various aspects of this disclosure may be better understood upon reading the following detailed description and upon reference to the drawings in which:

[0007] FIG. 1 show s a pictorial view of a computed tomography (CT) imaging system, in accordance with one or more embodiments of the present disclosure;

[0008] FIG. 2 shows a block schematic diagram of an example CT imaging system, in accordance with one or more embodiments of the present disclosure;

[0009] FIG. 3 is a schematic diagram of an exemplary detector array of a PCCT system, in accordance with one or more embodiments of the present disclosure;

[0010] FIG. 4 is a flow chart illustrating a high-level method for photon count data compression, according to embodiments of the present disclosure;

[0011] FIG. 5 is a flow chart illustrating a first example method for generating eigenvectors during calibration;

[0012] FIG. 6 is a flow chart illustrating a second example method for generating eigenvectors during calibration;

[0013] FIG. 7 is a flow chart illustrating an example method for compressing photon counts using one or more sets of eigenvectors;

[0014] FIG. 8 schematically shows a process for identifying eigenvectors during calibration;

[0015] FIG. 9 schematically shows a process for compressing photon counts using the eigenvectors of FIG. 8;

[0016] FIGS. 10-12 show example images, generated from detector data compressed and decompressed using pixel-specific eigenvectors, and quantitative analysis of the images;

[0017] FIGS. 13-15 show example images, generated from detector data compressed and decompressed using pixel-general eigenvectors, and quantitative analysis of the images; and

[0018] FIG. 16 is a comparison of the mean absolute percent error between energy-bin sinogram values in the original eight energy bins and the energy -bins after eigenbin compression and decompression using both pixel-specific and pixel-general eigenvector identification methods.DETAILED DESCRIPTION

[0019] This description and embodiments of the subject matter disclosed herein relate to methods and systems for compressing data acquired via a photon counting computed tomography (PCCT) system. In computed tomography (CT) imaging systems, an X-ray source or X-ray tube emits an X-ray beam towards an object, such as a patient, and X-rays attenuated by the subject are detected by one or more detectors (e.g., a detector array) to generate projection data that is used to reconstruct one or more images. The X-ray detector or detector array typically includes a collimator for collimating X-ray beams received at the detector, a scintillator disposed adjacent to the collimator for converting X-rays to light energy, and photodiodes for receiving the light energy from the adjacent scintillator and producing electrical signals therefrom. An intensity of the attenuated X-ray beam radiation received at the detector array is typically dependent upon the attenuation of the X-ray beam by the patient. Each detector element of a detector array produces a separate electrical signal indicative of the attenuated beam received by each detector element. The electrical signals are transmitted to a data processing system for analysis. The data processing system processes the electrical signals to facilitate generation of an image. Generally, in CT systems the X-ray source and the detector array are rotated about a gantry within an imaging plane and around the patient, and images are generated from projection data at a plurality of views at different view angles. For example, for one rotation of the X-ray source, 1000 views may be generated by the CT system.

[0020] Such conventional CT imaging systems utilize detectors that convert radiographic energy into current signals that are integrated over a time period, then measured and ultimately digitized. However, a drawback of such detectors is their inability to provide data or feedback as to the number and / or energy of photons detected. That is, the light emitted by the scintillator is a function of both a number of X-rays impinged and an energy level of the X-rays. The photodiodes may not be capable of discriminating between the energy level or the photon count from the scintillation. For example, two scintillators may illuminate with equivalent intensity and, as such, provide equivalent output to their respective photodiodes. Yet. despite yielding an equivalent light output, the number of X-rays received by each scintillator may be different, and an intensity of the X-rays may be different.

[0021] In contrast, PCCT detectors may provide photon counting and / or energy discriminating feedback with high spatial resolution. PCCT detectors can be caused to operate in an X-ray counting mode, and also in an energy measurement mode of each X-ray event. While a number of materials may be used in the construction of a direct conversion energy discriminating detector, semiconductors have been shown to be one preferred material. Typical materials for such use include Cadmium Zinc Telluride (CZT), Cadmium Telluride (CdTe) and Silicon (Si), which have a plurality of pixilated anodes attached thereto.

[0022] A drawback of photon counting detectors, however, is that they generate a relatively large amount of data due to the increased number of pixels and energy bins. For example, PCCT systems may have a plurality of energy bins (e.g., 5 or 8) that are determined by a comparator that may be part of a readout of a data acquisition system (DAS). A system having many energy bins may be formed with multiple comparators in the readout DAS. Each comparator may be set to trigger above a set level of energy that results in accumulation on a register of the number of photons above a corresponding X-ray energy level. Because the detector arrays rotate, the collected data are transmitted to a data processing computer for image reconstruction through a slip ring, which has a limited bandwidth. Thus, initial image reconstruction may be delayed due to the relatively long amount of time it may take to transmit the data to the data processing computer (e.g., across the slip ring). In other examples, the data may be transmitted to a data processing computer through another mechanism, including wired and / or wireless data transfer protocols.

[0023] Thus, systems and methods are proposed herein that compress the detector data into fewer bins, e.g., 2, 3, or 4 bins rather than the full 5 or 8 bins, thereby reducing the amount of data that is transmitted across the slip ring or via other data transfer protocols / mechanisms (e.g., via a rotary transformer). By compressing the detector data, images may be reconstructed and presented for review relatively quickly, which may aid clinical decision-making, particularly in clinical environments where speed may be prioritized, such as emergency rooms and / or when evaluating a patient in an acute stroke condition. The detector data may be compressed by performing principal component analysis (PCA) to transform the data acquired with N energy bins (e.g., 8 energy bins) to a smaller number of bins that carries the largest possible amount of information for that reduced number of bins. PCA specifically organizes data (e.g., of N energy bins) into N principal components, with the first principal component carrying the most information (representing the most variance) followed by the second, third, etc. This allows later principal components (e.g.. the fifth, sixth, etc.) to be cut without significant information loss. The compressed bin counts may be sent to the data processingcomputer, where the compressed bin counts may be decompressed and used to generate virtual monoenergetic images (VMI) and / or perform material decomposition (MD) and / or generate reconstructed energy bin images.

[0024] An example of a PCCT system including a PCCT scanner that may be used to perform imaging scans in accordance with the present techniques is provided in FIGS. 1 and 2. FIG. 3 shows an example detector array of the PCCT scanner, where photons of X-rays directed at a subject by an X-ray source are counted by detectors of the detector array. The counted photons may be partitioned into bins and the bins may be compressed according to the method shown in FIG. 4, which includes a calibration process (show n schematically in FIG. 8) whereby calibration x-ray projection measurements through different thicknesses and combinations of materials are obtained and used to generate eigenvectors for each pixel, as shown by the method of FIG. 5, or for all pixels collectively, as shown by the method of FIG. 6. A linear transformation provided by the eigenvectors is then applied to the counted photons (e.g., bin counts) to compress the bin counts to fewer bins / measurements, as shown by the method of FIG. 7 and schematically in FIG. 9. The compressed bin counts may be decompressed off the scanner (e.g., by a data processing computer of the PCCT system) and used to generate VMIs and / or MD images and / or energy bin images. Example MD images generated from compressed and then decompressed bin counts using pixel-specific eigenvectors and pixel-general eigenvectors are shown in FIGS. 10 and 13, respectively. Example VMIs generated from compressed and then decompressed bin counts using pixelspecific eigenvectors and pixel-general eigenvectors are shown in FIGS. 11 and 14, respectively. A quantitative analysis of the generated MD images and VMIs are shown in FIGS. 12 and 15. A comparison of mean absolute percent error betw een energy-bin sinogram values in original eight energy bins and the energy bins after eigenbin compression and decompression using both the pixel-specific method of FIG. 5 and the pixel -general method of FIG. 6.

[0025] FIG. 1 illustrates an exemplary PCCT system 100 (also referred to as a photon counting X-ray imaging system) configured for CT imaging with photon counting detectors. Particularly, the PCCT system 100 is configured to image a subject 112 such as a patient, an inanimate object, one or more manufactured parts, and / or foreign objects such as dental implants, stents, and / or contrast agents present within the body. The PCCT system 100 includes a gantry 102, which in turn, may further include at least one X-ray source 104 configured to project a beam of X-ray radiation 106 (see FIG. 2) for use in imaging the subject 112 laying on a table 114. Specifically, the X-ray source 104 is configured to project the X- ray radiation beams 106 towards a detector array 108 positioned on the opposite side of thegantry 102. Although FIG. 1 depicts a single X-ray source 104, in certain embodiments, multiple X-ray sources and detectors may be employed to project a plurality’ of X-ray radiation beams for acquiring projection data at the same or different energy levels corresponding to the patient. In some embodiments, the X-ray source 104 may enable dual-energy' gemstone spectral imaging (GSI) by rapid peak kilovoltage (kVp) switching. In the embodiments described herein, the X-ray detector employed is a photon counting detector which is capable of differentiating X-ray photons of different energies.

[0026] In certain embodiments, the PCCT system 100 further includes an image processor unit 110 configured to reconstruct images of a target volume of the subject 112 using an iterative or analytic image reconstruction method. For example, the image processor unit 110 may use an analytic image reconstruction approach such as filtered back projection (FBP) to reconstruct images of a target volume of the patient. As another example, the image processor unit 110 may use an iterative image reconstruction approach such as advanced statistical iterative reconstruction (ASIR), conjugate gradient (CG), maximum likelihood expectation maximization (MLEM), model-based iterative reconstruction (MBIR), and so on to reconstruct images of a target volume of the subject 112. In some examples the image processor unit 110 may use an analytic image reconstruction approach such as FBP in addition to an iterative image reconstruction approach.

[0027] In some CT imaging system configurations, an X-ray source projects a cone- shaped X-ray radiation beam which is defined with respect to an X-Y-Z Cartesian coordinate system and generally referred to as an "imaging volume." The X-ray radiation beam passes through an object being imaged, such as the patient or subject. The X-ray radiation beam, after being attenuated by the object, impinges upon an array of detector elements. The intensity' of the attenuated X-ray radiation beam received at the detector array is dependent upon the attenuation of an X-ray radiation beam by the object. Each detector element of the array produces a separate electrical signal that is a measurement of the X-ray beam attenuation at the detector location. The attenuation measurements from all the detector elements are acquired separately to produce a transmission profile.

[0028] In some CT systems, the X-ray source and the detector array are rotated with a gantry within the imaging volume and around the object to be imaged such that an angle at which the X-ray beam intersects the object constantly changes. A group of X-ray radiation attenuation measurements, e.g., projection data, from the detector array at one gantry' angle is referred to as a "view." A "scan" of the object includes a set of views made at different gantry angles, or view angles, during one revolution of the X-ray source and detector.

[0029] FIG. 2 illustrates an exemplary imaging system 200 similar to the PCCT system 100 of FIG. 1. In accordance with aspects of the present disclosure, the imaging system 200 is configured for imaging a subject 204 (e.g., the subject 1 12 of FIG. 1). In one embodiment the imaging system 200 includes the detector array 108 (see FIG. 1). The detector array 108 further includes a plurality' of detector elements 202 that together sense the X-ray radiation beam 106 (see FIG. 2) that passes through the subject 204 (such as a patient) to acquire corresponding projection data. In some embodiments, the detector array 108 may be fabricated in a multislice configuration including the plurality of rows of cells or detector elements 202, where one or more additional rows of the detector elements 202 are arranged in a parallel configuration for acquiring the projection data. The detector elements 202 may also be referred to as pixels or detector pixels.

[0030] In certain embodiments, the imaging system 200 is configured to traverse different angular positions around the subject 204 for acquiring desired projection data. Accordingly, the gantry' 102 and the components mounted thereon may be configured to rotate about a center of rotation 206 for acquiring the projection data, for example, at different energy levels. Alternatively, in embodiments where the projection angle relative to the subject 204 varies as a function of time, the mounted components may be configured to move along a general curve rather than along a segment of a circle.

[0031] As the X-ray source 104 and the detector array 108 rotate, the detector array 108 collects data of the attenuated X-ray beams. The data collected by the detector array 108 undergoes pre-processing and calibration to condition the data to represent the line integrals of the attenuation coefficients of the scanned subject 204. The processed data are commonly called projections. In some examples, the individual detectors or detector elements 202 of the detector array 108 may include photon counting detectors which register the interactions of individual photons into one or more energy bins.

[0032] The acquired sets of projection data may be used for basis material decomposition (BMD). During BMD, the measured projections are converted to a set of material -density' projections. The material-density projections may be reconstructed to form a set of materialdensity maps or images of each respective basis material, such as bone, soft tissue, and / or contrast agent maps. The density7maps or images may be, in turn, associated to form a 3D volumetric image of the basis material, for example, bone, soft tissue, and / or contrast agent, in the imaged volume.

[0033] Once reconstructed, the basis material image produced by the imaging system 200 reveals internal features of the subject 204, expressed in the densities of two basis materials.The density image may be displayed to show these features. In traditional approaches to diagnosis of medical conditions, such as disease states, and more generally of medical events, a radiologist or physician would consider a hard copy or display of the density image to discern characteristic features of interest. Such features might include lesions, sizes and shapes of particular anatomies or organs, and other features that would be discernable in the image based upon the skill and knowledge of the individual practitioner.

[0034] In one embodiment, the imaging system 200 includes a control mechanism 208 to control movement of the components such as rotation of the gantry 102 and the operation of the X-ray source 104. In certain embodiments, the control mechanism 208 further includes an X-ray controller 210 configured to provide power and timing signals to the X-ray source 104. Additionally, the control mechanism 208 includes a gantry motor controller 212 configured to control a rotational speed and / or position of the gantry 102 based on imaging requirements.

[0035] In certain embodiments, the control mechanism 208 further includes a data acquisition system (DAS) 214 configured to sample analog data received from the detector elements 202 and convert the analog data to digital signals for subsequent processing. The DAS 214 may be further configured to selectively aggregate data from a subset of the detector elements 202 into so-called macro-detectors. The data sampled and digitized by the DAS 214 is transmitted to a computer or computing device 216 via a slip ring 213. In one example, the computing device 216 stores the data in a storage device or mass storage 218. The storage device 218, for example, may be any type of non-transitory memory and may include a hard disk drive, a floppy disk drive, a compact disk-read / write (CD-R / W) drive, a Digital Versatile Disc (DVD) drive, a flash drive, and / or a solid-state storage drive.

[0036] Additionally, the computing device 216 provides commands and parameters to one or more of the DAS 214, the X-ray controller 210, and the gantry motor controller 212 for controlling system operations such as data acquisition and / or processing. In certain embodiments, the computing device 216 controls system operations based on operator input. The computing device 216 receives the operator input, for example, including commands and / or scanning parameters via an operator console 220 operatively coupled to the computing device 216. The operator console 220 may include a keyboard (not shown) or a touchscreen to allow the operator to specify the commands and / or scanning parameters.

[0037] Although FIG. 2 illustrates one operator console 220, more than one operator console may be coupled to the imaging system 200, for example, for inputting or outputting system parameters, requesting examinations, plotting data, and / or viewing images. Further, in certain embodiments, the imaging system 200 may be coupled to multiple displays, printers,workstations, and / or similar devices located either locally or remotely, for example, within an institution or hospital, or in an entirely different location via one or more configurable wired and / or wireless networks such as the Internet and / or virtual private networks, wireless telephone networks, wireless local area networks, wired local area networks, wireless wide area networks, wired wide area networks, etc.

[0038] In one embodiment, for example, the imaging system 200 either includes, or is coupled to. a picture archiving and communications system (PACS) 224. In an exemplary implementation, the PACS 224 is further coupled to a remote system such as a radiology department information system, hospital information system, and / or to an internal or external network (not shown) to allow operators at different locations to supply commands and parameters and / or gain access to the image data.

[0039] The computing device 216 uses the operator-supplied and / or system-defined commands and parameters to operate a table motor controller 226, which in turn, may control a table 114 which may be a motorized table. Specifically, the table motor controller 226 maymove the table 114 for appropriately positioning the subject 204 in the gantry 102 for acquiring projection data corresponding to the target volume of the subject 204.

[0040] As previously noted, the DAS 214 samples and digitizes the projection data acquired by the detector elements 202. Subsequently, an image reconstructor 230 uses the sampled and digitized X-ray data to perform high-speed reconstruction. Although FIG. 2 illustrates the image reconstructor 230 as a separate entity-, in certain embodiments, the image reconstructor 230 may form part of the computing device 216. Alternatively, the image reconstructor 230 may be absent from the imaging system 200 and instead the computing device 216 may perform one or more functions of the image reconstructor 230. Moreover, the image reconstructor 230 may be located locally or remotely, and may be operatively connected to the imaging system 200 using a wired or wireless network. Particularly, one exemplary embodiment may use computing resources in a "cloud" network cluster for the image reconstructor 230.

[0041] In one embodiment, the image reconstructor 230 stores the images reconstructed in the storage device 218. Alternatively, the image reconstructor 230 may transmit the reconstructed images to the computing device 216 to generate useful patient information for diagnosis and evaluation. In certain embodiments, the computing device 216 may transmit the reconstructed images and / or the patient information to a display or display device 232 communicatively coupled to the computing device 216 and / or the image reconstructor 230. In some embodiments, the reconstructed images may be transmitted from the computing device216 or the image reconstructor 230 to the storage device 218 for short-term or long-term storage.

[0042] Information may be transmitted between the components residing in the gantry 102 and external devices (such as the computing device 216 and / or image reconstructor 230) via the slip ring 213, which facilitates electronic communication across the rotating gantry'. In some examples, the gantry and internal components (e.g., the control mechanism 208, X-ray source 104, the detector array 108) may be collectively defined as a PCCT scanner, and as such the computing device 216 and image reconstructor 230 may reside off the scanner.

[0043] Referring now to FIG. 3, a PCCT photon counting detector array 300 is shown, which may be a non-limiting example of detector array 108 of FIG. 2. Detector array 300 includes rails 304 having collimating blades or plates 306 placed therebetween. Plates 306 are positioned to collimate X-rays 302 before such beams impinge upon a plurality of detector modules 308 of detector array 300, which may be arranged between the plates 306. As an example, detector array 300 may include 57 detector modules 308, each detector module 308 having an array size of 64x16 of detector elements (e.g.. pixels). As a result, detector array 300 would have 64 rows and 912 columns (16 pixels x 57 detector modules), allowing for 64 simultaneous slices of data to be collected with each gantry rotation (e.g., the gantry 102 of FIG. l).h

[0044] As described above, each detector element of each detector module 308 may be configured to directly convert radiographic energy to electrical signals containing energy discriminatory or photon count data. For example, when a photon impinges upon a detector element of a detector module 308, a charge may be generated within a semiconductor layer of the detector element that is proportional to the energy of the photon. A comparator may compare the voltage of the generated charge to one or more thresholds and increment a count of a bin (of a plurality of bins) based on the voltage relative to the one or more thresholds. The plurality of bins may include 8 bins, for example, with energy thresholds configured for optimal material decomposition performance.

[0045] Thus, the X-ray beam may generate multiple photon counts (e.g., one or more counts for each energy bin) for each detector element, resulting in substantially more data than that generated by integrating detectors. Generating images from the data collected with photon counting detectors may be time-consuming, which may delay image review. Further, transmitting the data from the DAS to an external computing device for image reconstruction may be challenging, as the example including a rotating gantry necessitates a mechanism such as a slip ring to transmit the data, which has a limited bandwidth and thus may limit the amountof data that can be transmitted. Accordingly, compression methods are provided herein to compress the amount of data to be transmitted and / or used in image reconstruction.

[0046] The compression may include a calibration procedure and accompanying algorithms that compress (decrease the size of) photon counting CT data to prior to transfer and then decompress the data off of the scanner. The algorithm is based on principal component analysis to transform the acquired data with N energy bins to a smaller number of measurements that carry the largest possible amount of information for that reduced number of channels. The compression process first acquires calibration X-ray projection measurements through different thicknesses and combinations of materials. The calibration materials are selected to span and sample the X-ray attenuation that may be seen in the body. Consider for example a photon counting CT system that acquires N spectral energy bins for each ray measurement. Each measurement is a point in an N-dimensional space. The collection of measurements for all of the calibration materials forms a cloud in this N-dimensional space. The compression process uses an algorithm to perform principal component analysis, also known as eigenvector decomposition, on the full set of calibration data, which finds the N orthonormal vectors where the first vector is the direction that represents the greatest variance in the data, the second vector is orthogonal to the first and is the direction that represents the next greatest variance, and so on.

[0047] The eigenvectors determined from the calibration data provide a linear transformation that can be applied to any N-dimensional measurement acquired by the scanner. After determining the eigenvectors during the calibration step, the linear transformation provided by these vectors can be applied to any N-dimensional measurement made by the scanner. The data size is reduced by selecting only the X eigenvectors corresponding the X largest directions of variations (the vectors with the X greatest eigenvalues). During compression, the X eigenvectors are arranged as columns of an N x X matrix, which is then multiplied with the vector containing an N-dimensional photon counting measurement. The resulting X values are the coefficients representing the amount of each eigenvector in the measurement. Only these X coefficients need to be transferred off of the gantry. These X measurements may be referred to as the eigenbin measurements. Because the X eigenvectors selected represent the greatest directions of variance, the eigenvectors that are not included should not contain much of the information that describes the differences between calibration measurements. Using this approach, the importance of the lost information is minimized. After transfer of the X coefficients from the gantry, an estimate of the original N bins can be obtainedby a simple mathematical operation using the X coefficients and the N x X array of eigenvectors. This is the decompression or decoding step.

[0048] The decompressed data may be applied to perform material decomposition and / or generate virtual monoenergetic images and / or reconstruct energy-bin images. The photon count data compression described herein may significantly reduce the data size that needs to be transmitted through the slip ring without heavy computational loads or substantial loss of spectral information, which may result in faster processing / reconstruction times.

[0049] In some examples, during calibration, the set of N eigenvectors is calculated for each detector pixel in the 2D detector array, which is referred to as a pixel-specific eigenbin method. This demands that an N x X matrix containing the eigenvectors for each pixel be stored in the scanner, which may have practical challenges. To address this issue, another approach may include combining the calibration data for all detector pixels and calculating one set of eigenvectors for the entire detector array. This method, referred to as a pixel-general eigenbin method, demands storage of only one N x X matrix of eigenvectors that is used for all pixels in the detector array.

[0050] Turning now to FIG. 4, it shows a method 400 for performing a scan using photon count data compression. Method 400 may be carried out according to instructions stored in memory of one or more controllers or computing devices included as part of and / or operatively coupled to a CT imaging system, such as DAS 214, X-ray controller 210, image reconstructor 230, and / or computing device 216.

[0051] At 402, method 400 includes performing one or more calibration scans to generate and store a plurality of eigenvectors. Briefly, during the calibration scan(s), detector data may be obtained while scanning object(s) of known size and composition, referred to as phantoms. The detector data may include photon counts partitioned into bins based on the energy of each photon, such as eight or five energy bins depending on the configuration of the detector. A principal component analysis (PCA) may be performed on the photon counts / energy bins obtained during the one or more calibration scans to generate the eigenvectors. In some examples, pixel-specific eigenvectors are generated during the calibration, as indicated at 404 and explained in more detail below with respect to FIG. 5. The pixel-specific eigenvectors may include a set of eigenvectors for each pixel of the detector array. Each set of eigenvectors includes N eigenvectors, where N corresponds to the number of bins that the photon counts are partitioned into, or fewer than N eigenvectors, depending on the configuration of the PCCT system (as explained in more detail below). In other examples, pixel-general eigenvectors are generated during the calibration, as indicated at 406 and explained in more detail below withrespect to FIG. 6. The pixel-general eigenvectors may include one common set of eigenvectors (e.g., N eigenvectors) for all pixels of the detector array.

[0052] At 408, method 400 includes scanning a patient according to a selected scan protocol to obtain detector data including bin counts. The detector data may include, for each pixel of the photon counting detector (e.g., for each detector element 202), photon counts partitioned into a plurality of energy bins based on an energy imparted by each photon on the photon counting detector, which is referred to as bin counts herein. During the patient scan, the X-ray source (e.g.. X-ray source 104 of FIGS. 1 and 2) of the CT imaging system may be controlled to emit X-rays according to a scan prescription set forth by the selected scan protocol (e.g., with a set X-ray source or X-ray tube current and voltage). Detector data may be obtained from the detector array (e.g., detector array 108 of FIGS. 1 and 2) as the X-ray source and detector array rotate around the patient, resulting in a plurality of views of detector data being obtained. For each view, a photon count for each energy bin for each pixel or detector element of the detector array is generated (e.g., during readout of the detector array by DAS 214). For example, in detector configurations with eight energy bins, eight photon counts may be generated for each pixel or detector element of the detector array and for each view. In this way, the output of the detector array may be referred to as the bin counts, as the photon counts are partitioned into a plurality of energy7bins based on the energy7of each photon that impinges on the detector array. The number of energy bins may be based on the configuration of the detector. For example, silicon detectors may be configured to differentiate photon energy into 8 energy bins, while cadmium telluride detectors may be configured to differentiate photon energy into 5 bins. The energy thresholds that define the energy bins may be determined during a calibration phase and / or may be based on the specific scan protocol. For example, the energy thresholds may be determined to optimize material basis decomposition and / or to maximize detected spectral information for a given incident spectrum emitted by the X-ray source.

[0053] At 410, the bin counts are compressed using the eigenvectors generated during the calibration. Additional details about the compression of the bin counts are provided below with respect to FIG. 7. Briefly, a subset of the one or more sets eigenvectors is selected (e.g., a subset of the common set of eigenvectors, or a subset of each pixel-specific set of eigenvectors) and for each pixel, the bin counts for that pixel are weighted based on the selected eigenvectors and, for each eigenvector, the weighted bins are summed. Thus, if a given pixel takes measurements that are partitioned into 8 energy7bins and four eigenvectors are selected, the energy bin counts are multiplied by a first eigenvector of the selected eigenvectors and the resultant products (e.g., weighted energy bins) are summed to form a first compressed bincount; the energy bin counts are multiplied by a second eigenvector of the selected eigenvectors and the resultant products (e.g.. weighted energy bins) are summed to form a second compressed bin count; and so forth for each selected eigenvector to form a total of four compressed bin counts (also referred to as eigenbin measurements). Thus, the 8 bin counts may be compressed to four bin counts.

[0054] At 412, the compressed bin counts are sent to an image reconstructor or another suitable computing device configured to reconstruct one or more images from the downsampled bin counts. The compressed bin counts may be sent from the DAS of the CT scanner to the image reconstructor via a mechanism such as a slip ring of the CT scanner.

[0055] At 414, the compressed bin counts are decompressed (e.g., on the image reconstructor or the other suitable computing device) by applying the eigenvectors generated during the calibration. For example, the first of the compressed bin counts (e.g., a first eiginbin measurement) explained above may be multiplied by the first eigenvector to generate 8 initial decompressed bin counts; the second of the compressed bin counts (e.g., a second eiginbin measurement) may be multiplied by the second eigenvector to generate another 8 initial decompressed bin counts; and the process is repeated for each remaining compressed bin count / eigenbin measurement using the remaining two eigenvectors to form four sets of initial decompressed bin counts; and then the decompressed bin counts for each bin may be summed to form a final 8 decompressed bin counts. In this way, the decompressed bin counts include a value, which may be referred to as a decompressed measurement value, for each energy bin of the plurality of energy bins for each pixel of the detector array.

[0056] At 416, method 400 includes reconstructing one or more images from the decompressed bin counts. The images may be reconstructed by the image reconstructor or the other suitable computing device. Reconstructing the one or more images may include reconstructing one or more grayscale (e.g., virtual monoenergetic) images from the decompressed bin counts, as indicated at 418. To reconstruct a VMI, the decompressed bin counts may be decomposed into selected basis materials or attenuation components (e.g., Compton and photoelectric) using maximum likelihood estimation (MLE), least-squared, polynomial fitting, neural networks, or other suitable decomposition methods, and then material decomposition (MD) images may be reconstructed using filtered backprojection or another suitable reconstruction technique. A VMI at a selected energy (e.g., 68 keV) may be formed by linear combination of the reconstructed images.

[0057] Additionally or alternatively, reconstructing the one or more images may include reconstructing one or more MD images, as indicated at 420. To reconstruct MD images, thedecompressed bin counts may be decomposed into selected basis materials (e.g.. calcium and water) using a suitable decomposition method as explained above (e.g., MLE or least-squared) and then MD images may be reconstructed using a suitable reconstruction technique such as filtered backproj ection.

[0058] Additionally or alternatively, reconstructing the one or more images may include reconstructing one or more energy bin images, as indicated at 422. The one or more energy bin images may include one or more images reconstructed from data for a single respective energy bin. For example, the decompressed data from a first energy bin, for all pixels, may be used to reconstruct a first energy bin image (without any data from any other energy bins used in the reconstruction process); the decompressed data from a second energy bin, for all pixels, may be used to reconstruct a second energy bin image (without any data from any other energy bins used in the reconstruction process); and so forth. In some examples, a respective energy bin image may be reconstructed for each energy bin (e.g., resulting in 8 energy bin images), while in other examples, a respective energy bin image may be reconstructed from only selected energy bins (e.g., resulting in fewer than 8 energy bin images).

[0059] At 424, the reconstructed image(s) are displayed on a display device and / or saved in memory (e.g., in a PACS as part of a patient exam). In some examples, the reconstructed image(s) may be used internally by the CT imaging system (e.g., to set parameters for a subsequent scan or view acquisition) and thus the reconstructed image(s) may only be stored locally and temporarily and not displayed on a display device.

[0060] Thus, a plurality of energy bin counts (such as 5 or 8 energy bins) may be compressed into a reduced number of energy' bins (such as 2 or 3) and sent for image reconstruction, which may expedite the reconstruction of at least initial images during a CT exam. Once the compressed detector data (e.g., the reduced number of energy bins) is sent to the image reconstructor, the full amount of collected detector data (e.g., all energy bin counts) may be sent to the image reconstructor as well, which may allow for additional images to be reconstructed using all the available (e.g., non-compressed) detector data.

[0061] FIG. 5 illustrates a method 500 for generating pixel-specific eigenvectors. Method 500 may be carried out according to instructions stored in memory of one or more controllers or computing devices included as part of and / or operatively coupled to a CT imaging system, such as DAS 214, X-ray controller 210, image reconstructor 230, and / or computing device 216. In some examples, method 500 may be carried out as part of method 400. for example method 500 may be carried out at 402 and 404 of method 400.

[0062] At 502, method 500 includes performing a calibration scan using one or more phantoms. As explained previously, the calibration scan may be performed to calibrate the specific CT imaging system, such as to determine the response of each detector element, identify optimal energy bin thresholds, etc. During the calibration scan, one or more phantoms are scanned. The phantoms may be composed of one or more suitable materials, such as polyvinyl chloride (PVC) or poly ethylene (PE). In some examples, the phantoms may include regions of water, iodide, calcium, blood, adipose, and / or other materials (e.g., other contrast agents). During the calibration scan, the CT imaging system may be controlled so that the X- ray source emits X-rays which are detected by the detector of the CT imaging system after attenuation by each scanned phantom. A plurality of views of the one or more phantoms may be acquired. Thus, with the one or more phantoms including regions of different material and a plurality of views being obtained, detector measurements through a variety of materials and combinations of materials may be obtained to replicate the X-ray attenuation properties of materials imaged during diagnostic / patient scanning. In doing so, the eigenvectors identified during calibration may apply to all possible anatomy, which may reduce memory usage by allowing the same eigenvectors to be applied during all scans.

[0063] At 504, the full bin counts obtained during scanning of the phantom(s) during the calibration scan are obtained from the detector. The full bin counts may be similar to the full bin counts described above with respect to FIG. 4, e.g., photon counts partitioned into the full number of energy bins allowed by the configuration of the detector, such as 5 or 8 bins. Bin counts (e.g., photon counts partitioned into energy bins) may be obtained for each pixel of the detector and for each view obtained during the scan of each phantom. The full bin counts may be sent from the DAS 214 to the image reconstructor 230 and / or the computing device 216.

[0064] At 506, eigenvectors are generated for each pixel from the full bin counts. For each pixel, the full bin counts may include 8 bin counts for each view or measurement obtained during the calibration scan. If 78 measurements were obtained during the calibration scan, for example, the full bin counts may include an array of 8x78 measurements (e.g., photon counts) for each pixel. As indicated at 508, a set of calibration measurements is obtained for a first pixel, where the set of calibration measurements includes the bin counts for the first pixel for each of a plurality of measurements (e.g., for each of the 78 measurements discussed above). PCA is performed on the set of calibration measurements for the first pixel, which includes identification of a first vector through N-dimensional space (where N corresponds to the number of energy bins) that has a direction that represents the greatest variance in the set of calibration measurements for the first pixel. Thus, as indicated at 510, for the first pixel, a setof orthonormal eigenvectors (e.g., N vectors) is identified from the set of calibration measurements for the first pixel, where the set of orthonormal eigenvectors includes the first eigenvector that has a direction that corresponds to the greatest variability in the set of calibration measurements in the N-dimensional space. The set of orthonormal eigenvectors may further include a second eigenvector that is orthonormal to the first eigenvector and has a direction that corresponds to the next greatest variability (second highest variability) in the measurements; a third eigenvector that is orthonormal to the first eigenvector and the second eigenvector and has a direction that corresponds to the next greatest variability (third highest variability); and so forth. The identified eigenvectors for the first pixel are stored at 512, and may be stored in order based on variability (e.g., most to least). The above-described process is repeated for each pixel, as indicated at 514.

[0065] Thus, a set of eigenvectors is generated for each pixel of the detector array and the eigenvectors for each pixel are saved for data compression during scanning, as indicated at 516. The eigenvectors may be saved on the scanner (e.g., on the DAS) in order to compress the detector data before the detector data is sent off the scanner. The eigenvectors may also be saved on the image reconstructor or another suitable computing device (e.g., computing device 216) in order to decompress the compressed data prior to image reconstruction. In some examples, all N eigenvectors, for each set of eigenvectors, may be stored. Storing all N eigenvectors may allow more flexible workflows, such that some scans / workflows may use half of the eigenvectors (e.g., four) to compress the detector data two-fold while other scans / workflows may use a quarter of the eigenvectors (e.g., two) to compress the detector data four-fold. However, in other examples, only a subset of the N eigenvectors corresponding to the X greatest variability may be stored (e.g., four eigenvectors corresponding to the four greatest variabilities). In some examples, prior to generating the set of eigenvectors, the mean across all photon counting calibration measurements may be calculated and subtracted from each calibration measurement, which may translate the cluster of calibration measurements to be centered at the origin. Method 500 then ends.

[0066] FIG. 8 schematically shows a process 800 for generating eigenvectors from a measurement matrix 802 (also denoted as A in FIG. 8), which may be performed to generate the eigenvectors according to method 500 explained above. The measurement matrix 802 may include an NxM vector of calibration measurements (e.g., photon counts) for one pixel, where N is the number of bins (e.g., 8) and M is the number of measurements performed for the pixel during calibration (e.g., the number of views or different measurements at one view angle obtained, such as 78). Eigenvector decomposition is performed on the measurement matrix 802to identify a plurality of eigenvectors, where each eigenvector is an axis in an N-dimensional space (shown in plot 804) and is orthonormal to each other axis. The first eigenvector (V i) has a direction that corresponds to the largest variance in the measurement matrix 802, the second eigenvector (V2) has a direction that corresponds to the second largest variance in the measurement matrix 802, etc. As appreciated by plot 804, the directions of maximum variance of the measurements are not aligned with the coordinate system defined by the horizontal and vertical energy bin axes. Thus, the identification of the eigenvectors provides an orthogonal coordinate system that is aligned with the directions of variation in the measurements.

[0067] The set of eigenvectors 806 (also denoted as V in FIG. 8) identified via the eigenvector decomposition includes an NxN matrix where each column represents an orthonormal axis in the transformed space. Also shown in FIG. 8 is a set of eigenvalues 808 (also denoted as D in FIG. 8), which represent coefficients or weights of each eigenvector that represent the amount of variation in each eigenvector direction. The eigenvector with the greatest variance is circled in FIG. 8 along with its corresponding eigenvalue.

[0068] Thus, the measurement matrix A is composed of M columns (the number of measurements) and N rows (the number of bins). The matrix of eigenvectors V is derived from the measurement matrix A. Regardless of how many columns the measurement matrix has (e.g., regardless of how many measurements are in the measurement matrix), the resulting matrix of eigenvectors will be NxN (8x8 in the example of FIG. 8), because one of the steps in calculating the principal components of A involves calculating the covariance matrix of A. The covariance matrix of A is calculated by multiplying A by A’ (A transpose), according to C =AA’, where C is the covariance matrix of A.

[0069] A is an NxM matrix (N rows x M columns) and the transpose of A (A’) is an MxN matrix (M rows x N columns), and the resultant matrix from multiplication of two matrices has the same number of rows as the first matrix (so N rows inherited from A) and the same number of columns as the second matrix (so N columns inherited from A’), and thus the covariance matrix C is also an NxN matrix.

[0070] The eigenvectors and eigenvalues of C are calculated based on the formula CV = VD (where C is the covariance matrix, V is the matrix of eigenvectors, and D is the matrix of eigenvalues). The size of the eigenvalues in D conveys how much variance occurs along the direction of each respective eigenvector, which enables selection of the top X eigenvectors to represent the original data (A) in a compressed form which is now encoded using only X values per measurement instead of N values (specifically in the example shown in FIG. 8, from 8 bins per pixel to 4 eigenbin measurements per pixel). The matrix of eigenvectors (V) is stored inmemory' and retrieved to compress bin counts during scanning, as explained in more detail below with respect to FIG. 7.

[0071] FIG. 6 illustrates a method 600 for generating pixel-general eigenvectors. Method 600 may be carried out according to instructions stored in memory of one or more controllers or computing devices included as part of and / or operatively coupled to a CT imaging system, such as DAS 214, X-ray controller 210, image reconstructor 230, and / or computing device 216. In some examples, method 600 may be carried out as part of method 400. for example method 600 may be carried out at 402 and 406 of method 400.

[0072] At 602, method 600 includes performing a calibration scan using one or more phantoms. The calibration scan may be carried out as explained above with respect to 502 of FIG. 5. At 604, the full bin counts obtained during scanning of the phantom(s) during the calibration scan are obtained from the detector. The full bin counts may be similar to the full bin counts described above with respect to FIG. 4, e.g., photon counts partitioned into the full number of energy' bins allowed by the configuration of the detector, such as 5 or 8 bins. Bin counts (e.g., photon counts partitioned into energy’ bins) may be obtained for each pixel of the detector and for each view or measurement obtained during the scan of each phantom. The full bin counts may7be sent from the DAS 214 to the image reconstructor 230 and / or the computing device 216.

[0073] At 606, eigenvectors are generated for all pixels from the full bin counts. For each pixel, the full bin counts may include 8 bin counts for each view / measurement obtained during the calibration scan. If 78 measurements were obtained during the calibration scan, for example, the full bin counts may include an array of 8x78 measurements (e.g., photon counts) for each pixel. To generate the eigenvectors, a set of calibration measurements is obtained for all pixels, as indicated at 608. If 78 measurements were obtained during the calibration scan for each of 1000 pixels, for example, the full bin counts may include an array of 8 x 78,000 measurements (e.g., photon counts). The set of calibration measurements may' be the full bin counts for all pixels obtained at 604. PCA is performed on the set of calibration measurements, which includes identification of a set of orthonormal eigenvectors, as indicated at 610. The set of orthonormal eigenvectors may include a first eigenvector through an N-dimensional space that has a direction that represents the greatest variance in the measurements. The set of orthonormal eigenvectors may include additional eigenvectors that are orthonormal to the first eigenvector, such that N vectors are identified. For example, the set of eigenvectors may further include a second eigenvector that is orthonormal to the first eigenvector and has a direction that corresponds to the next greatest variability (second highest variability) in themeasurements; a third eigenvector that is orthonormal to the first eigenvector and the second eigenvector and has a direction that corresponds to the next greatest variability (third highest variability); and so forth.

[0074] Thus, a set of eigenvectors is generated for all pixels of the detector array and the eigenvectors are saved for data compression during scanning, as indicated at 612. The eigenvectors may be saved on the scanner (e.g., on the DAS) in order to compress the detector data before the detector data is sent off the scanner. The eigenvectors may also be saved on the image reconstructor or another suitable computing device (e g., computing device 216) in order to decompress the compressed data prior to image reconstruction. In some examples, all N eigenvectors may be stored. Storing all N eigenvectors may allow more flexible workflows, such that some scans / workflows may use half of the eigenvectors (e.g., four) to compress the detector data two-fold while other scans / workflows may use a quarter of the eigenvectors (e.g., two) to compress the detector data four-fold. However, in other examples, only a subset of the N eigenvectors corresponding to the X greatest variability' may be stored (e.g., four eigenvectors corresponding to the four greatest variabilities). In some examples, prior to generating the set of eigenvectors, the mean across all photon counting calibration measurements may be calculated and subtracted from each calibration measurement, which may translate the cluster of calibration measurements to be centered at the origin. Method 600 then ends.

[0075] Thus, both method 500 and method 600 may be performed to generate eigenvectors that are stored for subsequent compression of bin counts obtained during a subsequent scan. Method 500 generates a set of eigenvectors for each pixel of the detector array, while method 600 generates one common set of eigenvectors for all pixels of the detector array. The pixelspecific approach of method 500 may result in more accurate data compression and thus less noise or bias in the resultant reconstructed images. However, storing a set of eigenvectors for each pixel may be impractical (e.g., demand a large memory that may not be available on the scanner) and thus the pixel-general approach of method 600 may facilitate implementation on the scanner. It is to be appreciated that method 600 may be carried out similarly to process 800 of FIG. 8, with the PCA / eigenvector decomposition being performed on the collective set of calibration measurements for all pixels, and the identification of the eigenvectors may be performed in the same manner as explained above with respect to FIG. 8. Further, in some examples, an intermediate approach to generating the eigenvectors may be performed that includes determining a set of eigenvectors for each group of a plurality of groups of pixels. For example, the pixels of the detector array may be grouped into groups of 2, 4, 8, 10, or morepixels. A set of eigenvectors may be generated for each pixel group. The intermediate approach may provide a compromise between the increased accuracy of the pixel-specific approach of method 500 and the reduced memory demand of the pixel-general approach of method 600.

[0076] FIG. 7 illustrates a method 700 for compressing detector bin counts. Method 700 may be carried out according to instructions stored in memory' of one or more controllers or computing devices included as part of and / or operatively coupled to a CT imaging system, such as DAS 214. X-ray controller 210, image reconstructor 230, and / or computing device 216. In some examples, method 700 may' be carried out as part of method 400, for example as the compression of bin counts at 410 of method 400.

[0077] At 702, method 700 includes determining scan parameters to be applied during the scan of the imaging subject. The scan parameters may be determined based on a selected scan protocol and / or based on user input received at a computing device of the CT imaging system (e.g., computing device 216). The scan parameters may' include the scan prescription (e.g., X- ray source voltage and current, slice thickness, gantry table speed, etc.), the anatomy being scanned, whether one or more contrast agents have been administered to the imaging subject, and other parameters. The scan parameters may also include the diagnostic goal of the scan, which may' dictate that images be reconstructed as quickly7as possible or within a particular time range to facilitate rapid diagnosis (e.g., during stroke care) or adjustment of subsequent scan parameters, as well as a dictated relative quality of the reconstructed images.

[0078] At 704, method 700 determines if the calibration performed at 402 of method 400 was a pixel-specific calibration, and thus if a set of eigenvectors for each pixel of the detector array is stored for use in compressing the bin counts. If the calibration was a pixel-specific calibration, method 700 proceeds to 706 to select a subset of eigenvectors from each respective set of eigenvectors based on the variability of each eigenvector and the scan parameters. As discussed previously, the eigenvectors generated for each pixel may correspond to different directions of diminishing variation (e.g., the first eigenvector for a given pixel corresponds to the largest direction of variation, the second eigenvector for the given pixel corresponds to the second largest direction of variation, etc.). Each subset is selected such that only the X eigenvectors corresponding to the X largest directions of variations are included in the subset, where X is based on the scan parameters (at least in some examples). For example, the scan parameters may dictate that the bin counts be compressed two-fold, from 8 bins to 4 bins. Thus, four eigenvectors may be selected from each set of eigenvectors, with the four selected eigenvectors corresponding to the four largest directions of variation. Some scan parameters may dictate that the bin counts be compressed to two bins, three bins, or five bins, in whichcase the X eigenvectors selected for each pixel may be two eigenvectors, three eigenvectors, or five eigenvectors. As a non-limiting example, scan parameters that dictate images be reconstructed as quickly as possible, even if the images include increased noise and / or bias, may result in fewer eigenvectors being selected (to compress the data into fewer bins), while scan parameters that dictate images be reconstructed quickly, but with less noise and / or bias, may result in more eigenvectors being selected (to compress the data into more bins). How ever, in other examples, the number of eigenvectors selected may be uniform across all scans / scan parameters. In still further examples, only a selected subset of the N possible eigenvectors for each set of eigenvectors may be stored and all the stored eigenvectors may be selected.

[0079] At 708, the bin counts for each pixel are combined using the selected eigenvectors for each pixel to form eigenbin measurements. For example, for a first pixel, the bin counts for the first pixel are combined using the selected eigenvectors for the first pixel (e.g., the eigenvectors generated during calibration may include a set of eigenvectors, such as 8 eigenvectors, for the first pixel, and four of the eigenvectors for the first pixel may be selected and applied to combine the bin counts for the first pixel). Combining the bin counts for a given pixel may include, for each eigenvector of the selected eigenvectors, multiplying the bin counts for that pixel by the eigenvector and summing the resultant products (e.g., weighted bin counts), as indicated at 709.

[0080] The process of forming the eigenbin measurements is shown schematically in FIG. 9. The selected eigenvectors (e.g.. the X selected eigenvectors) are arranged as columns of an NxX matrix 904 (denoted V-i in FIG. 9, with the unselected eigenvectors blacked out), which is then multiplied with a vector 902 (denoted m in FIG. 9) containing an N-dimensional photon counting measurement (e.g., the 8 energy bins from one pixel of the detector array obtained during a scan). This multiplication results in a matrix of weighted bin counts, with each bin count weighted four times (based on each selected eigenvector). Each column of the matrix is then summed to form four coefficients (m'V4), referred to as the eigenbin measurements, that represent the amount of each eigenvector in the measurement. Only these coefficients need to be transferred off of the gantry. Because the X eigenvectors selected represent the greatest directions of variance, the eigenvectors that are not included should not contain much of the information that describes the differences between calibration measurements (e.g., the coordinates corresponding to the eigenvectors with the least amount of variance may represent stochastic noise). Using this approach, the importance of the lost information is minimized. The four coefficients are decoded / decompressed back to 8 bins using the four eigenvectors, e.g., according to m = V^Cm'l^)'.

[0081] Returning to FIG. 7, at 710, the compressed bin counts (e.g., the eigenbin measurements) are sent to the image reconstructor or other suitable computing device for use in reconstructing one or more images, as explained above with respect to FIG. 4. In some examples, method 700 includes, at 712, sending the full bin counts to the image reconstructor or other suitable computing device. In this way, initial images may be generated relatively quickly using the eigenbin measurements, while additional images may be generated using the original / full bin counts after a delay period (as the original / full bin counts may take more time to be sent to the image reconstructor), if desired. It is to be appreciated that a similar approach may be taken in examples where a set of eigenvectors is determined for pixel groups (e.g., for each group of pixels of a plurality of groups of pixels, as described above) rather than individual pixels. In such examples, the eigenvectors for each pixel group may be selected based on variability and scan parameters, as described above at 706. For each pixel, the counts may be combined using the selected eigenvectors for that pixel to form the eigenbin measurements, as described above at 708, and the compressed bin counts sent to the image reconstructor. In this way, each pixel is a given pixel group may utilize the same eigenvectors to compress the bin counts but different pixel groups may use different eigenvectors.

[0082] Returning to 704, it is determined that the calibration performed was not a pixelspecific calibration, but instead a pixel-general calibration, method 700 proceeds to 714 to select eigenvectors for all pixels based on the variability of the eigenvectors and optionally the scan parameters, similarly to the selection of the eigenvectors performed at 706, to form a subset of eigenvectors. For example, the pixel-general calibration may result in one set of eigenvectors being generated and stored. The number of eigenvectors is selected from the set such that only the X eigenvectors corresponding to the X largest directions of variations are included in the subset. At 716, the bin counts for each pixel are combined using the selected eigenvectors to form eigenbin measurements. For example, for a first pixel, the bin counts for the first pixel are combined using the selected eigenvectors (e.g., the eigenvectors generated during calibration may include a set of eigenvectors, such as 8 eigenvectors, and four of the eigenvectors may be selected and applied to combine the bin counts for the first pixel). Combining the bin counts for a given pixel may include, for each eigenvector of the selected eigenvectors, multiplying the bin counts for that pixel by the eigenvector and summing the resultant products (e.g., weighted bin counts), as indicated at 717.

[0083] Thus, energy bins may be compressed according to the methods disclosed herein in order to compress the amount of data sent from the CT machine to an off-board computing device / image reconstructor. The energy bins may be compressed using PCA / eigenvectorsgenerated during calibration to form eigenbin measurements. As explained above, because the X eigenvectors selected represent the greatest directions of variance, the eigenvectors that are not included should not contain much of the information that describes the differences between calibration measurements. Using this approach, the importance of the lost information is minimized. After transfer of the eigenbin measurements from the gantry, an estimate of the original N bins can be obtained by a simple mathematical operation using the eigenbin measurements and the NxX array of eigenvectors (e.g., the selected eigenvectors). For example, the eigenbin vector m can be decompressed or decoded to estimate the original N measurements as n = Vrn Conceptually, the compression step projects each measurement onto the coordinate system spanned by the M eigenvectors. The decompression step weights each eigenvector by the corresponding eigenbin coefficient and then sums the resulting weighted vectors to estimate the coordinates of the original measurement in the N-dimensional Cartesian space. Further, while FIG. 7 selected eigenvectors based on whether pixel-specific or pixelgeneral calibration was performed, it is to be appreciated that (for both pixel-specific and pixelgeneral calibration methods) different sets of calibration measurements may be obtained under various different settings, such as tube voltage and / or bowtie filter, with sets of eigenvectors identified for each set of calibration measurements. During scanning, the appropriate set of eigenvectors may be selected based on the settings of that scan (e.g., the tube voltage and / or bowtie filter used).

[0084] Accordingly, the methods disclosed herein include, during a scan of an imaging subject, obtaining detector data from a photon counting detector, the detector data comprising, for each pixel of the photon counting detector, photon counts partitioned into a plurality of energy bins based on an energy imparted by each photon on the photon counting detector; compressing the detector data into a respective set of eigenbin measurements for each pixel using one or more sets of eigenvectors identified during calibration; decompressing each respective set of eigenbin measurements using the one or more sets of eigenvectors to form decompressed detector data; and reconstructing one or more images from the decompressed detector data.

[0085] The one or more sets of eigenvectors identified during calibration may include a single set of eigenvectors that applies to all pixels of the detector (pixel-general), a respective set of eigenvectors for each pixel of the detector (pixel-specific), or a respective set of eigenvectors for each pixel group of the detector. The calibration may include obtaining calibration detector data from a calibration scan of a phantom, the calibration detector data comprising, for each pixel of the photon counting detector, calibration photon countspartitioned into the plurality of energy bins based on the energy of each photon that impinges on the detector. When a single set of eigenvectors is identified for all pixels, the set of eigenvectors is identified via PCA / eigen decomposition on the calibration photon counts for all pixels collectively (e.g., the decomposition may be performed on all the photon / bin counts for all pixels, and all views, in one common N-dimensional space). When a respective set of eigenvectors is identified for each pixel (or pixel group), each set of eigenvectors is identified via PCA / eigen decomposition on the calibration photon counts for each pixel (or pixel group) individually (e.g., the decomposition may be performed on the photon / bin counts for each pixel (and all views of that pixel) or pixel group in individual N-dimensional spaces).

[0086] For the pixel-general method, identifying the one or more sets of eigenvectors based on the calibration detector data comprises identifying a set of eigenvectors for all pixels of the photon counting detector from the calibration detector data from all pixels in N- dimensional space, including identifying N eigenvectors from the calibration detector data, and selecting the set of eigenvectors from the N eigenvectors. The set of eigenvectors includes X eigenvectors and the X eigenvectors have the X greatest eigenvalues from the N eigenvectors, N corresponds to the number of energy bins in the plurality of energy bins, and X is less than N. Further, compressing the detector data into the respective set of eigenbin measurements for each pixel using the one or more sets of eigenvectors identified during calibration comprises compressing the detector data using the set of eigenvectors, including compressing detector data from each pixel of the photon counting detector using the set of eigenvectors. For example, for first detector data from a first pixel, the compressing may include multiplying the first detector data by each eigenvector of the set of eigenvectors, and summing the resultant products of each eigenvector to form a first set of eigenbin measurements. A similar process is performed for each remaining pixel, such that for second detector data from a second pixel, the compressing may include multiplying the second detector data by each eigenvector of the set of eigenvectors, and summing the resultant products of each eigenvector to form a second set of eigenbin measurements. Thus, a respective set of eigenbin measurements is generated for each pixel, wherein each set of eigenbin measurements includes fewer measuremen ts / counts than the original, full number of bin counts for that pixel (e.g., the full bin counts may include 8 bin counts for each pixel and the eigenbin measurements may include four eigenbin measurements for each pixel). The eigenbin measurements for each pixel are sent off the scanner and then decompressed, using the set of eigenvectors, back to the original number of bins, and the one or more images are reconstructed from the decompressed data.

[0087] For the pixel-specific method, identifying the one or more sets of eigenvectors based on the calibration detector data comprises identifying a respective set of eigenvectors for the first pixel of the photon counting detector from calibration detector data from only a first pixel of the photon counting detector in N-dimensional space, including identifying N eigenvectors from the calibration detector data from only the first pixel, and selecting a first set of eigenvectors from the N eigenvectors. The first set of eigenvectors includes X eigenvectors and the X eigenvectors have the X greatest eigenvalues from the N eigenvectors. N corresponds to the number of energy bins in the plurality of energy bins, and X is less than N. Further, identify ing the respective set of eigenvectors for each pixel of the photon counting detector comprises separately analyzing calibration detector data from each additional pixel of the photon counting detector in N-dimensional space to identify respective N eigenvectors from the calibration detector data from each pixel, and selecting a respective set of eigenvectors from each respective N eigenvectors. The compression may include compressing the detector data into the respective set of eigenbin measurements for each pixel by, for first detector data from a first pixel, multiplying the first detector data by each eigenvector of a first set of eigenvectors corresponding to the first pixel, and summing the resultant products of each eigenvector to form a first set of eigenbin measurements. A similar process is performed for each additional pixel, such that, for second detector data from a second pixel, the second detector data is multiplied by each eigenvector of a second set of eigenvectors corresponding to the second pixel, and the resultant products of each eigenvector are summed to form a second set of eigenbin measurements. Thus, a respective set of eigenbin measurements is generated for each pixel, wherein each set of eigenbin measurements includes fewer measurements / counts than the original, full number of bin counts for that pixel (e.g., the full bin counts may include 8 bin counts for each pixel and the eigenbin measurements may include four eigenbin measurements for each pixel). The eigenbin measurements for each pixel are sent off the scanner and then decompressed, using the set of eigenvectors, back to the original number of bins, and the one or more images are reconstructed from the decompressed data.

[0088] FIG. 10 shows a first set of MD basis images 1000 of a phantom reconstructed from full bin data (e.g., 8 bins) and compressed bin data at different levels of compression, where the bin counts are compressed using pixel-specific eigenvectors as described herein. For example, the sinogram data (e.g., bin counts for each pixel of the detector array) were compressed into 4, 3, and 2 eigenbins with pixel-specific eigenvectors and then decompressed back into 8 bins before image reconstruction. Material decomposition may be performed with material decomposition models generated using calibration data that also went through thesame eigenbin processing as the data used to form the first set of MD basis images 1000. A first image 1002 shows a water MD basis image using the full 8 bins (e.g. , no data compression) and a second image 1004 shows an iodine MD basis image using the full 8 bins. A third image 1006 and a fourth image 1008 show a water MD basis image and an iodine MD basis image, respectively, with the data compressed into four eigenbin measurements (e.g., four bins). A fifth image 1010 and a sixth image 1012 show a water MD basis image and an iodine MD basis image, respectively, with the data compressed into three eigenbin measurements (e.g., three bins). A seventh image 1014 and an eighth image 1016 show a water MD basis image and an iodine MD basis image, respectively, with the data compressed into two eigenbin measurements (e.g., two bins).

[0089] FIG. 11 shows a first set of VMI images 1100 at 70 keV reconstructed from the same sinogram / full bin data as the images in FIG. 10. The first set of VMI images 1100 includes a first image 1102 reconstructed from the full 8 bins (e.g., without data compression), and a second image 1104, a third image 1106, and a fourth image 1106 reconstructed from compressed-decompressed data (e.g., compressed with the pixel-specific eigenvectors and then decompressed), with the data compressed into four eigenbin measurements (e.g.. four bins), three eigenbin measurements (e.g., three bins), and two eigenbin measurements (e.g., two bins), respectively.

[0090] FIG. 12 illustrates a quantitative evaluation 1200 of the images shown in FIGS. 10 and 11. wherein the quantitative evaluation 1200 includes a first graph 1202, a second graph 1204, and a third graph 1206. The first graph 1202 depicts a mean reconstructed value for various ROIs in the iodine basis images shown in FIG. 10. The mean reconstructed values are iodine basis values calculated from the ROIs of each image, with the ROIs including a region of adipose, a region of calcium at 10 mg / ml, a region of iodine at 2 mg / ml, a region of iodine at 5 mg / ml, a region of iodine at 10 mg / ml, and a region of iodine at 15 mg / ml. The mean reconstructed values for the ROIs from the 8 bin image (e.g., the second image 1004) are shown by the blue bars, the mean reconstructed values for the ROIs from the 4 bin image (e.g., the fourth image 1008) are shown by the orange bars, the mean reconstructed values for the ROIs from the 3 bin image (e.g.. the sixth image 1012) are shown by the yellow bars, and the mean reconstructed values for the ROIs from the 2 bin image (e.g., the eighth image 1016) are shown by the purple bars.

[0091] The second graph 1204 depicts a mean reconstructed value for various ROIs in the water basis images shown in FIG. 10. The mean reconstructed values are water basis values calculated from the ROIs of each image, with the ROIs including the same ROIs as the firstgraph 1202. The mean reconstructed values for the ROIs from the 8 bin image (e.g., the first image 1002) are shown by the blue bars, the mean reconstructed values for the ROIs from the 4 bin image (e.g., the third image 1006) are shown by the orange bars, the mean reconstructed values for the ROIs from the 3 bin image (e.g., the fifth image 1010) are shown by the yellow bars, and the mean reconstructed values for the ROIs from the 2 bin image (e.g., the seventh image 1014) are shown by the purple bars.

[0092] Each of the compressed images (e.g.. the images reconstructed from compressed, and then decompressed, bin counts) has a mean reconstructed value, for each of the ROIs, wi thin a threshold of the mean reconstructed values for the corresponding ROI of the full (noncompressed data) image for both the iodine and water basis images (e.g., differs from the full, 8-bin image by less than 10%). Even the images reconstructed from the data compressed into two bins (e.g., the seventh image 1014 and the eighth image 1016) demonstrated a less than 5% change in the iodine basis values relative to the image reconstructed from the full, noncompressed data, a less than 2% change in the water basis values relative to the image reconstructed from the full, non-compressed data, and showed a less than 20% increase in MD noise.

[0093] The third graph 1206 depicts a mean Hounsfield Unit (HU) for various ROIs in the 70 keV VMI images shown in FIG. 11. The mean HUs are calculated from the ROIs of each image, with the ROIs including the same ROIs as the first graph 1202 and second graph 1204. The mean HUs for the ROIs from the 8 bin image (e.g., the first image 1102) are shown by the blue bars, the mean HUs for the ROIs from the 4 bin image (e.g., the second image 1104) are shown by the orange bars, the mean HUs for the ROIs from the 3 bin image (e.g., the third image 1106) are shown by the yellow bars, and the mean HUs for the ROIs from the 2 bin image (e.g., the fourth image 1108) are shown by the purple bars. The mean HUs for the compressed images are within a threshold of the mean HUs for the non-compressed image, for all of the ROIs (e.g., each compressed image differs from the non-compressed image by less than 10%). For example, the image from the data compressed into two bins (e.g., the fourth image 1108) demonstrated a less than 5 HU change in the VMI values relative to the image reconstructed from the full, non-compressed data, and showed a less than 10% increase in VMI noise.

[0094] Similar MD images, VMI images, and quantitative evaluation are shown for detector data compressed with pixel-general eigenvectors, at FIGS. 13-15. FIG. 13 show s a second set of MD basis images 1300 of a phantom reconstructed from full bin data (e.g.. 8 bins) and compressed bin data at different levels of compression, where the bin counts arecompressed using pixel-general eigenvectors as described herein. For example, the sinogram data (e.g., bin counts for each pixel of the detector array) were compressed into 4 and 3 eigenbins with pixel-general eigenvectors and then decompressed back into 8 bins before image reconstruction. The sinogram data may be the same sinogram data used to reconstruct the images shown in FIGS. 10 and 11, but compressed with the pixel-general eigenvectors rather than the pixel-specific eigenvectors. Material decomposition may be performed with material decomposition models generated using calibration data that also went through the same eigenbin processing as the data used to form the second set of MD basis images 1300. A first image 1302 shows a water MD basis image using the full 8 bins (e.g., no data compression) and a second image 1304 shows an iodine MD basis image using the full 8 bins. A third image 1306 and a fourth image 1308 show a water MD basis image and an iodine MD basis image, respectively, with the data compressed into four eigenbin measurements (e.g., four bins). A fifth image 1310 and a sixth image 1312 show a w ater MD basis image and an iodine MD basis image, respectively, with the data compressed into three eigenbin measurements (e.g., three bins).

[0095] FIG. 14 shows a second set of VMI images 1400 at 70 keV reconstructed from the same sinogram / full bin data as the images in FIG. 13 (e.g., using pixel-general eigenvectors). The second set of VMI images 1400 includes a first image 1402 reconstructed from the full 8 bins (e.g., without data compression), and a second image 1404 and a third image 1406 reconstructed from compressed-decompressed data (e.g., compressed with the pixel-general eigenvectors and then decompressed), with the data compressed into four eigenbin measurements (e.g., four bins) and three eigenbin measurements (e.g., three bins), respectively.

[0096] FIG. 15 illustrates a quantitative evaluation 1500 of the images shown in FIGS. 13 and 14. wherein the quantitative evaluation includes a first graph 1502, a second graph 1504, and a third graph 1506. The first graph 1502 depicts a mean reconstructed value for various ROIs in the iodine basis images shown in FIG. 13. The mean reconstructed values are iodine basis values calculated from the ROIs of each image, with the ROIs including a region of adipose, a region of calcium at 10 mg / ml, a region of iodine at 2 mg / ml, a region of iodine at 5 mg / ml. a region of iodine at 10 mg / ml. and a region of iodine at 15 mg / ml. The mean reconstructed values for the ROIs from the 8 bin image (e g., the second image 1304) are shown by the blue bars, the mean reconstructed values for the ROIs from the 4 bin image (e.g., the fourth image 1308) are shown by the orange bars, and the mean reconstructed values for the ROIs from the 3 bin image (e.g.. the sixth image 1312) are shown by the yellow bars.

[0097] A second graph 1504 depicts a mean reconstructed value for various ROIs in the water basis images shown in FIG. 13. The mean reconstructed values are water basis values calculated from the ROIs of each image, with the ROIs including the same ROIs as the first graph 1502. The mean reconstructed values for the ROIs from the 8 bin image (e.g., the first image 1302) are show n by the blue bars, the mean reconstructed values for the ROIs from the 4 bin image (e.g., the third image 1306) are shown by the orange bars, and the mean reconstructed values for the ROIs from the 3 bin image (e.g.. the fifth image 1310) are shown by the yellow bars.

[0098] Each of the compressed images (e.g., the images reconstructed from compressed, and then decompressed, bin counts) has a mean reconstructed value, for each of the ROIs, within a threshold of the mean reconstructed values for the corresponding ROI of the full (noncompressed data) image for both the iodine and water basis images. Even the images reconstructed from the data compressed into three bins demonstrated a less than 5% change in the iodine basis values relative to the image reconstructed from the full, non-compressed data, a less than 5% change in the water basis values relative to the image reconstructed from the full, non-compressed data, and show ed a less than 9% increase in MD noise.

[0099] A third graph 1506 depicts a mean Hounsfield Unit (HU) for various ROIs in the 70 keV VMI images shown in FIG. 14. The mean HUs are calculated from the ROIs of each image, with the ROIs including the same ROIs as the first graph 1502 and second graph 1504. The mean HUs for the ROIs from the 8 bin image (e.g., the first image 1402) are shown by the blue bars, the mean HUs for the ROIs from the 4 bin image (e.g., the second image 1404) are shown by the orange bars, and the mean HUs for the ROIs from the 3 bin image (e.g., the third image 1406) are shown by the yellow bars. The mean HUs for the compressed images are within a threshold of the mean HUs for the non-compressed image, for all of the ROIs. For example, the image from the data compressed into three bins (e.g., the third image 1406) demonstrated a less than 12 HU change in the VMI values relative to the image reconstructed from the full, non-compressed data, and showed a less than 5% increase in VMI noise.

[0100] FIG. 16 shows a plot 1600 comparing the mean absolute percent error between the original eight energy-bin sinograms and the energy-bin sinograms after eigenbin compression and decompression. Results are plotted for both the pixel-specific and pixel-general eigenbin methods. For the pixel-specific eigenbin method, the error in the energy -bin sinogram values was 12% for the highest energy' bin and below- 6% for most other energy -bins. The pixelgeneral eigenbin method resulted in errors below 16% when using three and four eigenbins, with errors as high as 60% for when using two pixel-general eigenbins.

[0101] Thus, the pixel-specific eigenbin compression (e.g., using pixel-specific eigenvectors) can reduce the amount of data sent from the scanner by a factor of 4. with a less than 5% change in quantitative values and less than 20% noise increase. The pixel-general eigenbin compression (e.g., using pixel -general eigenvectors) can reduce the amount of data sent from the scanner by a factor of 2.67 with a less than 5% change in quantitative values and less than 10% noise increase. It is to be appreciated that further data reduction with the pixelgeneral eigenbin compression may cause a greater than 5% change in quantitative values and thus may reduce image quality to an unsatisfactory level. Accordingly, eigenbin compression may be useful for applications that demand rapid generation of MD images, with the full amount of data still available for subsequent, slower reconstructions.

[0102] The technical effect of compressing energy binned photon counts from a full amount of energy bins to a reduced number of energy bins is that less data is transmitted to an external computing device for reconstruction (e.g., across the slip ring of the CT system), thereby expediting the process of reconstructing images from the photon counts. Another technical effect of compressing the photon counts using the eigenvectors described herein is that the compressed / reduced number of energy bins may be decompressed back to the full amount of energy bins using the eigenvectors by the external computing device, which may allow image reconstruction to occur using the standard workflow(s) for reconstructing images already available on the external computing device. For example, the MD basis and / or VMI reconstruction process described herein may be carried out on the full, 8-bin photon count data using a standard workflow configured for 8-bin data. When the 8-bin data is compressed to four eigenbin measurements, for example, and then decompressed back to 8 bins, the MD basis and / or VMI and / or energy -bin images may still be reconstructed using the standard workflow configured for 8-bin data, obviating a need for a new workflow to handle 4-bin data, for example. Further, decompressing the data back to 8-bin data allows for reconstruction of energy bin images that otherwise not be possible if the data was maintained in the compressed state.

[0103] Though a photon counting computed tomography (PCCT) system is described by way of example, it should be understood that the present techniques may also be useful when applied to other X-ray imaging modalities having photon counting detectors, such as X-ray angiography systems, X-ray tomosynthesis systems, X-ray mammography systems, X-ray fluoroscopy systems, X-ray interventional systems, X-ray C-arm systems, etc. The present discussion of a PCCT imaging modality is provided merely as an example of one suitable imaging modality.

[0104] The disclosure also provides support for a method for a photon counting computed tomography (PCCT) system, the method comprising: during a scan of an imaging subject, obtaining detector data from a photon counting detector of the PCCT system, the detector data comprising, for each pixel of the photon counting detector, photon counts partitioned into a plurality of energy' bins based on an energy' imparted by each photon on the photon counting detector, compressing the detector data into a respective set of eigenbin measurements for each pixel using one or more sets of eigenvectors identified during calibration of the PCCT system, decompressing each respective set of eigenbin measurements using the one or more sets of eigenvectors to form decompressed detector data, and reconstructing one or more images from the decompressed detector data. In a first example of the method, the method further comprises: identifying the one or more sets of eigenvectors based on calibration detector data obtained from a calibration scan of a phantom, the calibration detector data comprising, for each pixel of the photon counting detector, calibration photon counts partitioned into the plurality' of energy bins based on the energy of each photon that impinges on the detector. In a second example of the method, optionally including the first example, identifying the one or more sets of eigenvectors based on the calibration detector data comprises identifying a set of eigenvectors for all pixels of the photon counting detector by identify ing N eigenvectors from calibration detector data from all pixels, and selecting the set of eigenvectors from the N eigenvectors, wherein the set of eigenvectors includes X eigenvectors and the X eigenvectors have the X greatest eigenvalues from the N eigenvectors, and wherein N corresponds to the number of energy bins in the plurality of energy' bins and X is less than N. In a third example of the method, optionally including one or both of the first and second examples, compressing the detector data into the respective set of eigenbin measurements for each pixel using the one or more sets of eigenvectors identified during calibration of the PCCT system comprises compressing the detector data using the set of eigenvectors, including compressing detector data from each pixel of the photon counting detector using the set of eigenvectors. In a fourth example of the method, optionally including one or more or each of the first through third examples, compressing the detector data from each pixel of the photon counting detector using the set of eigenvectors comprises, for first detector data from a first pixel: multiplying the first detector data by each eigenvector of the set of eigenvectors, and summing the resultant products of each eigenvector to form a first set of eigenbin measurements. In a fifth example of the method, optionally including one or more or each of the first through fourth examples, identifying the one or more sets of eigenvectors based on the calibration detector data comprises identifying a respective set of eigenvectors for each pixel of the photon countingdetector by identifying N eigenvectors from calibration detector data from only a first pixel of the photon counting detector, and selecting a first set of eigenvectors from the N eigenvectors, wherein the first set of eigenvectors includes X eigenvectors and the X eigenvectors have the X greatest eigenvalues from the N eigenvectors, and wherein N corresponds to the number of energy7bins in the plurality of energy' bins and X is less than N. In a sixth example of the method, optionally including one or more or each of the first through fifth examples, identifying the respective set of eigenvectors for each pixel of the photon counting detector comprises separately identifying respective N eigenvectors from calibration detector data from each additional pixel of the photon counting detector, and selecting a respective set of eigenvectors from each respective N eigenvectors. In a seventh example of the method, optionally including one or more or each of the first through sixth examples, compressing the detector data into the respective set of eigenbin measurements for each pixel using the one or more sets of eigenvectors identified during calibration of the PCCT system comprises compressing the detector data into the respective set of eigenbin measurements for each pixel by, for first detector data from a first pixel: multiplying the first detector data by each eigenvector of a first set of eigenvectors corresponding to the first pixel, and summing the resultant products of each eigenvector to form a first set of eigenbin measurements. In an eighth example of the method, optionally including one or more or each of the first through seventh examples, the plurality of energy bins comprises eight energy bins and each respective set of eigenbin measurements comprises four or fewer eigenbin measurements.

[0105] The disclosure also provides support for a photon counting computed tomography (PCCT) system, comprising: an X-ray source that emits a beam of X-rays toward a subject to be imaged, a photon counting detector that receives the beam of X-rays attenuated by the subject, and a data acquisition system (DAS) operably connected to the photon counting detector and configured to: during a scan of an imaging subject, obtain detector data from the photon counting detector, the detector data comprising, for each pixel of the photon counting detector, photon counts partitioned into a plurality' of energy' bins based on an energy' imparted by each photon on the photon counting detector to form a set of bin counts for each pixel, compress each set of bin counts into a respective set of eigenbin measurements using one or more sets of eigenvectors identified during calibration of the PCCT system, send each compressed set of eigenbin measurements to a computer comprising non-transitory memory and operably connected to the DAS, wherein the computer is configured with instructions in the non-transitory memory that when executed cause the computer to decompress each respective set of eigenbin measurements using the one or more sets of eigenvectors to formdecompressed detector data and reconstruct one or more images from the decompressed detector data. In a first example of the system, the one or more sets of eigenvectors identified during calibration of the PCCT system includes a single set of eigenvectors corresponding to all pixels of the photon counting detector, wherein compressing each set of bin counts into a respective set of eigenbin measurements using the one or more sets of eigenvectors comprises compressing, for each pixel, that set of bin counts using the set of eigenvectors to form a respective set of eigenbin measurements for that pixel. In a second example of the system, optionally including the first example, the one or more sets of eigenvectors identified during calibration of the PCCT system includes a plurality of sets of eigenvectors, each corresponding to a respective pixel of the photon counting detector, and wherein compressing each set of bin counts into a respective set of eigenbin measurements using the one or more sets of eigenvectors comprises compressing, for each pixel, that set of bin counts using a respective set of eigenvectors corresponding to that pixel to form a respective set of eigenbin measurements for that pixel. In a third example of the sy stem, optionally including one or both of the first and second examples, the plurality of energy’ bins comprises eight energy bins and each respective set of eigenbin measurements comprises four or fewer eigenbin measurements. In a fourth example of the system, optionally including one or more or each of the first through third examples, each compressed set of eigenbin measurements is sent to the computer via a slip ring of the PCCT system.

[0106] The disclosure also provides support for a method for a photon counting X-ray imaging system, the method comprising: during a calibration scan of a phantom, obtaining calibration detector data from a photon counting detector of the X-ray imaging system, the calibration detector data comprising, for each pixel of the photon counting detector, calibration photon counts partitioned into a plurality of energy bins based on an energy imparted by each photon on the photon counting detector, identifying one or more sets of eigenvectors from the calibration detector data, during a scan of an imaging subject, receiving a respective set of eigenbin measurements for each pixel of the photon counting detector, decompressing each respective set of eigenbin measurements using the one or more sets of eigenvectors to form decompressed detector data, and reconstructing one or more images from the decompressed detector data. In a first example of the method, the calibration detector data includes a respective set of calibration measurements for each pixel of the photon counting detector, and wherein identifying one or more sets of eigenvectors from the calibration detector data comprises identifying a respective set of eigenvectors from each set of calibration measurements, such that a respective set of eigenvectors is identified for each pixel of thephoton counting detector. In a second example of the method, optionally including the first example, the calibration detector data includes a respective set of calibration measurements for each pixel of the photon counting detector, and wherein identifying one or more sets of eigenvectors from the calibration detector data comprises identifying a set of eigenvectors from all sets of calibration measurements, such that one set of eigenvectors is identified for all pixels of the photon counting detector. In a third example of the method, optionally including one or both of the first and second examples, reconstructing one or more images from the decompressed detector data comprises reconstructing one or more virtual monoenergetic images, one or more material basis images, and / or one or more energy bin images. In a fourth example of the method, optionally including one or more or each of the first through third examples, the decompressed data includes a decompressed measurement value for each energy bin of the plurality of energy bins for each pixel of the photon counting detector.

[0107] When introducing elements of various embodiments of the present disclosure, the articles “a,” “an,” and “the” are intended to mean that there are one or more of the elements. The terms “first,” “second.” and the like, do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. The terms “comprising,” “including,” and “having” are intended to be inclusive and mean that there may be additional elements other than the listed elements. As the terms “connected to,” “coupled to,” etc. are used herein, one object (e.g., a material, element, structure, member, etc.) can be connected to or coupled to another object regardless of whether the one object is directly connected or coupled to the other object or whether there are one or more intervening objects between the one object and the other object. In addition, it should be understood that references to “one embodiment” or “an embodiment” of the present disclosure are not intended to be interpreted as excluding the existence of additional embodiments that also incorporate the recited features.

[0108] In addition to any previously indicated modification, numerous other variations and alternative arrangements may be devised by those skilled in the art without departing from the spirit and scope of this description, and appended claims are intended to cover such modifications and arrangements. Thus, while the information has been described above with particularity and detail in connection with what is presently deemed to be the most practical and preferred aspects, it will be apparent to those of ordinary skill in the art that numerous modifications, including, but not limited to, form, function, manner of operation and use may be made without departing from the principles and concepts set forth herein. Also, as used herein, the examples and embodiments, in all respects, are meant to be illustrative only and should not be construed to be limiting in any manner.

Claims

CLAIMS:

1. A method for a photon counting computed tomography (PCCT) system, the method comprising: during a scan of an imaging subject, obtaining detector data from a photon counting detector of the PCCT system, the detector data comprising, for each pixel of the photon counting detector, photon counts partitioned into a plurality of energy bins based on an energy imparted by each photon on the photon counting detector; compressing the detector data into a respective set of eigenbin measurements for each pixel using one or more sets of eigenvectors identified during calibration of the PCCT system; decompressing each respective set of eigenbin measurements using the one or more sets of eigenvectors to form decompressed detector data; and reconstructing one or more images from the decompressed detector data.

2. The method of claim 1, further comprising identifying the one or more sets of eigenvectors based on calibration detector data obtained from a calibration scan of a phantom, the calibration detector data comprising, for each pixel of the photon counting detector, calibration photon counts partitioned into the plurality of energy7bins based on the energy7of each photon that impinges on the photon counting detector.

3. The method of claim 2, wherein identifying the one or more sets of eigenvectors based on the calibration detector data comprises identify ing a set of eigenvectors for all pixels of the photon counting detector by identifying N eigenvectors from calibration detector data from all pixels, and selecting the set of eigenvectors from the N eigenvectors, wherein the set of eigenvectors includes X eigenvectors and the X eigenvectors have the X greatest eigenvalues from the N eigenvectors, and wherein N corresponds to the number of energy bins in the plurality of energy7bins and X is less than N.

4. The method of claim 3, wherein compressing the detector data into the respective set of eigenbin measurements for each pixel using the one or more sets of eigenvectors identified during calibration of the PCCT system comprises compressing the detector data using the set of eigenvectors, including compressing detector data from each pixel of the photon counting detector using the set of eigenvectors.

5. The method of claim 4, wherein compressing the detector data from each pixel of the photon counting detector using the set of eigenvectors comprises, for first detector data from a first pixel: multiplying the first detector data by each eigenvector of the set of eigenvectors, and summing the resultant products of each eigenvector to form a first set of eigenbin measurements.

6. The method of claim 2, wherein identifying the one or more sets of eigenvectors based on the calibration detector data comprises identifying a respective set of eigenvectors for each pixel of the photon counting detector by identifying N eigenvectors from calibration detector data from only a first pixel of the photon counting detector, and selecting a first set of eigenvectors from the N eigenvectors, wherein the first set of eigenvectors includes X eigenvectors and the X eigenvectors have the X greatest eigenvalues from the N eigenvectors, and wherein N corresponds to the number of energy bins in the plurality7of energy bins and X is less than N.

7. The method of claim 6, wherein identify ing the respective set of eigenvectors for each pixel of the photon counting detector comprises separately identifying respective N eigenvectors from calibration detector data from each additional pixel of the photon counting detector, and selecting a respective set of eigenvectors from each respective N eigenvectors.

8. The method of claim 6, wherein compressing the detector data into the respective set of eigenbin measurements for each pixel using the one or more sets of eigenvectors identified during calibration of the PCCT system comprises compressing the detector data into the respective set of eigenbin measurements for each pixel by, for first detector data from a first pixel: multiplying the first detector data by each eigenvector of a first set of eigenvectors corresponding to the first pixel, and summing the resultant products of each eigenvector to form a first set of eigenbin measurements.

9. The method of claim 1, wherein the plurality of energy bins comprises eight energy bins and each respective set of eigenbin measurements comprises four or fewer eigenbin measurements.

10. A photon counting computed tomography (PCCT) system, comprising: an X-ray source that emits a beam of X-rays toward a subject to be imaged; a photon counting detector that receives the beam of X-rays attenuated by the subject; and a data acquisition system (DAS) operably connected to the photon counting detector and configured to: during a scan of an imaging subject, obtain detector data from the photon counting detector, the detector data comprising, for each pixel of the photon counting detector, photon counts partitioned into a plurality of energy bins based on an energy' imparted by each photon on the photon counting detector to form a set of bin counts for each pixel; compress each set of bin counts into a respective set of eigenbin measurements using one or more sets of eigenvectors identified during calibration of the PCCT system; send each compressed set of eigenbin measurements to a computer comprising non-transitory memory and operably connected to the DAS, wherein the computer is configured with instructions in the non-transitory memory that when executed cause the computer to decompress each respective set of eigenbin measurements using the one or more sets of eigenvectors to form decompressed detector data and reconstruct one or more images from the decompressed detector data.1 1 . The system of claim 10, wherein the one or more sets of eigenvectors identified during calibration of the PCCT system includes a single set of eigenvectors corresponding to all pixels of the photon counting detector, wherein compressing each set of bin counts into a respective set of eigenbin measurements using the one or more sets of eigenvectors comprises compressing, for each pixel, that set of bin counts using the set of eigenvectors to form a respective set of eigenbin measurements for that pixel.

12. The system of claim 10, wherein the one or more sets of eigenvectors identified during calibration of the PCCT system includes a plurality of sets of eigenvectors, each corresponding to a respective pixel of the photon counting detector, and wherein compressing each set of bin counts into a respective set of eigenbin measurements using the one or more sets of eigenvectors comprises compressing, for each pixel, that set of bin counts using a respective set of eigenvectors corresponding to that pixel to form a respective set of eigenbin measurements for that pixel.

13. The system of claim 10, wherein the plurality of energy bins comprises eight energy bins and each respective set of eigenbin measurements comprises four or fewer eigenbin measurements.

14. The system of claim 10, wherein each compressed set of eigenbin measurements is sent to the computer via a slip ring of the PCCT system.

15. A method for a photon counting X-ray imaging system, the method comprising: during a calibration scan of a phantom, obtaining calibration detector data from a photon counting detector of the X-ray imaging system, the calibration detector data comprising, for each pixel of the photon counting detector, calibration photon counts partitioned into a plurality of energy bins based on an energy imparted by each photon on the photon counting detector; identifying one or more sets of eigenvectors from the calibration detector data; during a scan of an imaging subject, receiving a respective set of eigenbin measurements for each pixel of the photon counting detector; decompressing each respective set of eigenbin measurements using the one or more sets of eigenvectors to form decompressed detector data; and reconstructing one or more images from the decompressed detector data.

16. The method of claim 15, wherein the calibration detector data includes a respective set of calibration measurements for each pixel of the photon counting detector, and wherein identifying one or more sets of eigenvectors from the calibration detector data comprises identify ing a respective set of eigenvectors from each set of calibration measurements, such that a respective set of eigenvectors is identified for each pixel of the photon counting detector.

17. The method of claim 15, wherein the calibration detector data includes a respective set of calibration measurements for each pixel of the photon counting detector, and wherein identifying one or more sets of eigenvectors from the calibration detector data comprises identify ing a set of eigenvectors from all sets of calibration measurements, such that one set of eigenvectors is identified for all pixels of the photon counting detector.

18. The method of claim 15, wherein reconstructing one or more images from the decompressed detector data comprises reconstructing one or more virtual monoenergetic images, one or more material basis images, and / or one or more energy bin images.

19. The method of claim 15, wherein the decompressed data includes a decompressed measurement value for each energy bin of the plurality of energy bins for each pixel of the photon counting detector.

Images