Energy weighting of photon counts for conventional imaging

Abstract

An image processing system comprises an input interface (IN) for receiving energy-resolved count data generated by a photon-counting detector (PCD) having at least 2 energy bins, but preferably more than two energy bins. A count data transformer (DT) of the system is configured to perform a transformation operation to transform the energy-resolved count data into transformed count data, the operation comprising applying one or more weights to the count data, the weights being obtained previously based on i) calibration image data for the photon-counting detector or a photon-counting detector and ii) calibration image data for an energy-integrating detector. A data converter component (CC) is configured to convert the transformed count data into path length data through a material. The system allows to emulate an energy-integrating detector.

Description

Technical Field

[0001] This invention relates to systems, imaging arrangements, image processing methods, calibration methods, computer program units, and computer-readable media for image processing. Background Technology

[0002] X-ray imaging is an important tool for learning the internal structural details of an object in a non-destructive manner. Its primary application is in the medical field. For example, CT (computed tomography) imaging allows for obtaining detailed cross-sectional images of a patient's anatomy.

[0003] X-ray imaging apparatus (e.g., CT scanners) includes detectors configured to detect X-ray radiation after it has passed through the object or patient to be imaged. One type of X-ray detector includes an energy integration detector. Energy integration detectors have been used for decades and represent the standard of practice followed by generations of medical professionals during their training. In X-ray imaging, an X-ray beam, comprising X-ray photons of varying energies, passes through the patient or object and is detected at the detector. The energy integration detector measures the total energy deposited at the detector pixel after the X-ray photons have passed through the patient's tissue. In conventional CT, the detected data can be converted to Hounsfield unit values. While useful in many cases, energy integration detectors typically do not allow for sufficiently high sensitivity to distinguish soft tissue (which may be necessary in many medical applications, such as oncology), and the Hounsfield unit scale is not without its problems, as its interpretation can depend on the surrounding anatomy.

[0004] To address these issues, a new generation of detectors known as energy-resolved photon counting detectors has emerged. These detectors can analyze incoming X-ray photons into energy levels. Specifically, these detectors count the number of photons falling into each energy level. In other words, these detectors can record an approximate result of the spectrum of the transmitted X-ray beam. The precision of the approximation depends on the number of energy levels considered. This counting data recorded by the photon counting detector forms spectral information, and this spectral information can be collected in various ways to generate new types of images. These images include material-specific maps, virtual monochromatic images, and more.

[0005] However, the medical community’s acceptance of photon counting detectors may be relatively low because, as mentioned above, generations of medical students have been trained on Hounsfield unit-type images derived from energy integral detectors.

[0006] US 9693743 B2 describes a photon counting CT apparatus. In an embodiment, a normalization parameter is used to estimate an integral signal based on a measurement count signal output by a PCCT detector.

[0007] WO 2018 / 002226 A1 discloses a photon counting computed tomography mechanism. Calibration data is used to convert photon counting projection data into associated path lengths. Summary of the Invention

[0008] Therefore, it may be necessary to improve the comparability between images obtainable using a photon counting detector imager and images obtainable using an energy integration detector imager.

[0009] The objectives of this invention are achieved by the subject matter of the independent claims, wherein further embodiments are incorporated in the dependent claims. It should be noted that the aspects described below are equally applicable to imaging arrangements, image processing methods, calibration methods, computer program units, and computer-readable media.

[0010] According to a first aspect of the present invention, a system for image processing is provided, comprising:

[0011] An input interface for receiving energy-resolved counting data generated by a photon counting detector;

[0012] A counting data transformer is configured to perform a transformation operation to transform the energy-resolved counting data into transformed counting data. The operation includes applying one or more weights to the counting data, the weights being previously obtained based on: i) calibration image data for the photon counting detector or a photon counting detector, and ii) calibration data for an energy integration detector; and

[0013] An output interface for outputting the transformed count data.

[0014] According to one embodiment, the system includes a data converter component configured to convert the count data into path length data through a material. In other words, the count for each pixel is converted into a corresponding path length through a reference material. More than one material type can be used. According to one embodiment, the reference material used for the conversion component to convert the material path length can be based on water or (one or more) materials with similar radiometric densities.

[0015] Preferably, the weights are calculated in advance.

[0016] Calibration data for a photon counting detector (“PCD”) may include projection count data obtained during the calibration process using the PCD.

[0017] Calibration data for the energy integration detector (“EID”) may include projection data, or preferably image data reconstructed from it. This data is obtained during the calibration process using the EID.

[0018] These two calibration procedures may include imaging a test subject, such as an imaging phantom, using the two detector types respectively.

[0019] Alternatively, the weights can be obtained theoretically, for example, through simulation.

[0020] According to one embodiment, the system includes a reconstructor configured to reconstruct image data from the path length data through the material. This image data can be displayed on a display device.

[0021] The proposed computerized system allows for cross-detector image adjustment and comparability. The system allows for the “simulation” of images obtainable with a given energy-integrating detector using a photon-counting detector. The ability to reconstruct images from projection data of two different detector types enhances comparability. For example, in a CT embodiment, the converted image data simulates Hounsfield unit (HU)-based images obtainable when using an energy-integrating detector.

[0022] According to one embodiment, the system includes a user interface configured to allow a user to change at least one of the weights.

[0023] According to one embodiment, the imaging arrangement includes the system according to any one of the preceding claims and an imaging device having the photon counting detector.

[0024] According to another aspect, an image processing method is provided, comprising:

[0025] Receive energy-resolved counting data generated by a photon counting detector;

[0026] The energy-resolved counting data is transformed into transformed counting data. This operation includes applying one or more weights to the counting data, the weights being previously obtained based on: i) calibration image data for the photon counting detector or a photon counting detector, and ii) calibration data for the energy integration detector; and

[0027] Output the transformed count data.

[0028] According to another aspect, a calibration method is provided, comprising:

[0029] Receive energy-resolved calibration counting data for the photon counting detector;

[0030] Receive calibration image data for the energy integration detector; and

[0031] Adjust one or more weights applicable to the energy-resolved count data so that image data derived from the weighted count data is suitable for the calibration image data for the energy integral detector; and

[0032] Output the one or more weights.

[0033] The calibration image data and calibration count data involve the same phantom or test specimen. In the automated optimization process, weights are preferably (but necessary) iteratively adjusted until the image reconstructed from the count data corresponds to (i.e., "looks like") the calibration image data for the energy integration detector. This correspondence can be quantified using a similarity metric. The calibration image data for the energy integration detector includes image data reconstructed from projection data recorded by the EID.

[0034] As an alternative or supplement to automatically determining weights through such optimization, in the embodiments there exists a user interface that allows users to change / adjust the weights for fine-tuning or to manually perform adjustments.

[0035] According to another aspect, a computerized system configured to perform calibration methods is provided.

[0036] According to one embodiment, the energy-resolved calibration count data and / or the calibration image data are obtained based on exposing the sample body to X-ray radiation.

[0037] According to another aspect, a computer program unit is provided, which, when run by at least one processing unit, is adapted to cause the processing unit to perform a method according to any of the embodiments described above.

[0038] According to another aspect, a computer-readable medium having program units stored thereon is provided. Attached Figure Description

[0039] Exemplary embodiments of the invention will now be described with reference to the following drawings, wherein the same reference numerals denote the same parts, wherein:

[0040] Figure 1 A schematic block diagram of an imaging arrangement using a photon counting detector is shown;

[0041] Figure 2 A schematic diagram of a photon counting circuit coupled to a detector pixel is shown.

[0042] Figure 3 A schematic block diagram of components of an image processing system configured to process counting data generated by a photon counting detector is shown.

[0043] Figure 4An arrangement for calculating weights based on test images or data received from a photon counting detector and an energy integration detector is shown.

[0044] Figure 5 Exemplary spectral data related to a photon counting detector and an energy integration detector are shown;

[0045] Figure 6 It shows from Figure 5 The exemplary ratio and energy relationship obtained from the exemplary spectrum;

[0046] Figure 7 A flowchart of the image processing method is shown; and

[0047] Figure 8 A flowchart of the calibration method is shown. Detailed Implementation

[0048] refer to Figure 1 It shows a schematic block diagram of an imaging arrangement IAR contemplated in the embodiments herein.

[0049] The imaging setup IAR includes an X-ray imaging device IA1 configured for spectral imaging. More specifically, the spectral imaging capability is detector-based. More specifically, the imager IA1 includes a photon-counting detector (PCD). Broadly speaking, the PCD allows the acquisition of projection data at different energy levels. In particular, as explained in more detail below, the projection data includes count data generated by the PCD. Compared to conventional energy-integral-based detectors, the count data can be processed into a variety of different image types to better aid diagnosis. In particular, monochrome images and material-specific maps (including material decomposition) can be obtained for better identification of soft tissues. Additionally, the projection count data can also be processed by the IPS into images similar to the energy-integral images obtainable by conventional energy-integral detectors.

[0050] In a broad sense, an imaging setup (IAR) includes an image processing system (IPS), which specifically includes a projection data processor (PDP). The PDP is configured to apply specially configured weights w to the counted projection data π received from a photon-counting detector (PCD). Applying these weights allows the PDP to "simulate" a given energy-integrating detector of another imaging device with high fidelity. In other words, the PDP allows for better cross-modal comparison between images based on the PCD and images based on the energy-integrating detector. Images processed by the proposed PDP have the "look and feel" of more conventional images obtained from the energy-integrating detector. The proposed IPS system could be particularly beneficial for medical or research settings that simultaneously maintain both photon-counting and energy-integrating detectors. Furthermore, the proposed PDP is more efficient due to the high-fidelity simulation of images based on the energy-integrating detector, as it allows for the complete abandonment of conventional energy-integrating detectors.

[0051] The components of the image processing system IPS, including the projection data processor, can be arranged as software modules running on one or more computing units (PUs). The computing units can be general-purpose and located remotely from the X-ray imaging apparatus IA1. Specifically, the image processing system can be located remotely and externally to the housing of the detector PCD. The computing units (PUs) can be appropriately interfaced to receive projection counting data π generated by the photon counting detector via a suitable wired or wireless communication network. Some or all of the image processing system components can be stored in a single central memory (MEM) and run by one or more processing computing units (PUs). Alternatively, the image processing system IPS can be maintained such that the components are distributed across multiple memory (MEMs) in a computing environment spanning multiple computing systems (logically or geographically distributed). Alternatively, some or all of the image processing system components can be arranged in hardware, such as in a suitably programmed FPGA, ASCII, or other device. In this embodiment, instead of arranging the image processing system externally and remotely to the housing detector PCD, some or all of the components can be integrated into the detector PCD itself as an onboard post-processing stage.

[0052] Before explaining the components of the image processing system IPS more comprehensively, let’s first discuss the X-ray imaging device IA1, which includes the photon counting detector PCD, in more detail.

[0053] A photon-counting detector is arranged opposite an X-ray source XR1, thus defining a portion of the space between them (referred to herein as the examination area). The patient or object to be imaged, or at least its region of interest (ROI), resides in the examination area during imaging. Specifically, the X-ray source typically generates polychromatic X-ray radiation in the form of an X-ray beam XB. The X-ray beam is emitted from the X-ray source XR1 during use. Specifically, the X-ray beam passes through the examination area and thus through the ROI, and then propagates along the principal direction q toward the photon-counting detector PCD for registration at the photon-counting detector PCD. After the X-rays pass through the object or patient, they appear distally in a modified form due to interaction with the tissue or material in the ROI. The modifications the beam undergoes encode information about the internal structure of the ROI, which is precisely the information that is desired to be seen. During imaging, the patient or object to be imaged may lean against the examination table TB or other support, but this is not necessary in all embodiments, as in some embodiments the patient may be standing during imaging.

[0054] In embodiments used for tomographic imaging to perform spectral photon counting computed tomography (SPCCT), an X-ray imaging apparatus IA1 is particularly contemplated. The imaging apparatus may be arranged as a C-arm or U-arm imager or a CT scanner with a gantry containing a chamber in which the patient is located during imaging. While tomographic imaging is primarily contemplated herein, planar projection-based radiographic systems are also contemplated in alternative embodiments.

[0055] Turning more specifically to the tomography embodiment, the X-ray source rotates, but not necessarily, a full circle, around the patient during imaging. In this embodiment, the detector DCP rotates with the source XR1 and acquires projection-counted images π from different directions q. However, not all embodiments require the X-ray detector to rotate with the X-ray source, as is the case in fourth-generation CT scanners, for example. All generations of CT scanners (from generation 1 to generation 4) are envisioned herein, with third-generation scanners being the preferred embodiment.

[0056] As described above, an X-ray beam with an initial spectral intensity exiting the X-ray source interacts with tissue material in the patient and appears at the distal end of the patient as a modified beam. As a result of the interaction with the tissue, the modified beam has an altered spectrum compared to the initial spectrum. This interacting beam is then registered at the photon counting detector PCD for each projection direction q.

[0057] Turning more specifically to the operation of the image processing system IPS, the projection counting data π acquired from different directions q is output at the detector interface and forwarded to the image processing system IPS. Specially configured weights, designed to simulate energy integration behavior, are stored locally or remotely in a weight database WM. The weight data, which will be described more fully below, may be collectively referred to as "w" herein, and should be understood as typically comprising multiple numbers, particularly one or more sets of numbers. The projection data processor PDP accesses the weights w and applies them (preferably for each direction q) to the counting projection data.

[0058] The weighted projection count data can then be passed to the Reconstructor module RECON, which implements a tomographic reconstruction algorithm to generate axial cross-sectional image data. Axial cross-sectional image data can represent the internal structure of the patient or object being imaged within a plane. In other words, the Reconstructor module RECON maps weighted projection data from the projection domain to the image domain. The image domain comprises the various planes in the examination area that intersect with the patient. The projection domain is located at the detector PCD. The above process can be repeated for different image domain planes along the patient's longitudinal axis Z to obtain multiple cross-sectional image data at various different locations z on the axis Z. The cross-sectional images form a 3D image volume. Different sets of projection data for different z locations can be acquired by moving the source and / or support TB (with the patient resting on the support TB) along the axis Z during imaging. This longitudinal movement can be achieved, for example, by using a helical X-ray scanner.

[0059] Image data generated by the reconstructor at its output can be passed to a visualizer (VIZ). The visualizer maps the image data to a range of image values ​​(e.g., a range of Hounsfield units (HU) or other suitable imaging scales). These mapped image values ​​can then be mapped to color or grayscale values, and then visualized as a cross-sectional image on a display device (DD). Users, such as medical personnel, can then view the image for diagnostic or other purposes.

[0060] Before explaining the operation of the projection data processor (PDP) and how to obtain the specially configured weights w for energy integration simulation, the photon counting detector (PCD) will be explained in more detail.

[0061] The photon counting detector PCD envisioned in the embodiments herein comprises a conversion layer CL and a photon counting circuit PCC communicatively coupled thereto. Broadly speaking, the conversion layer CL converts incident photons in X-ray radiation into electrical signals (specifically pulses). The photon counting circuit PCC then processes the electrical pulses into counting data based on pulse height. The height is proportional to the energy of the individual photon that caused the pulse when it interacts with the individual photon in the conversion layer. These pulses are then processed for multiple energy thresholds to classify the detected electrical pulses according to their energy. Two, three, four, five, or more such energy thresholds can be used. For each energy range (sometimes referred to as a “bin”), projected counting data measures the photon flux (i.e., the number of photons recorded per unit time in that bin during acquisition). Such counting data is recorded in multiple bins for each projection direction q.

[0062] Turning more specifically to the conversion layer CL, the conversion layer CL can be arranged as a direct conversion layer or an indirect conversion layer. A direct conversion layer is preferred. In an embodiment, the direct conversion layer is arranged as a semiconductor (e.g., CZT, CdTe, or silicon or other materials) or other crystalline layer. Preferably, the conversion layer is configured as a plurality of detector pixels PX. The electrical pulses generated by each pixel PX are then processed by the photon counting circuit PCC as described above. In an embodiment, each individual detector pixel is coupled to the photon counting circuit via a communication line to transmit the electrical pulses to the photon counting circuit.

[0063] Now for reference Figure 2 , Figure 2 A photon counting circuit PCC is illustrated in more detail and according to one embodiment in a schematic manner. Only one portion of the photon counting circuit PCC is shown in association with a single detector pixel PX. It should be understood that similar circuitry is used for each of the other detector pixels. In an alternative embodiment, not each detector pixel may have its own photon counting circuit, as the photon counting circuit can be switched between multiple such detector pixels. However, preferably, as shown, each pixel PX has its own portion of the photon counting circuit.

[0064] More specifically, the exemplary pixel PX generates the electrical pulse described above. The amplitude (“height”) of the pulse largely corresponds to the energy of the impacting photon. The higher the energy of the photon, the higher the pulse amplitude that can be detected at the corresponding pixel PX. Figure 2 As shown on the right, each pixel electrode is coupled to the photon counting circuit by an independent signal line (or "(pixel) readout channel") CH.

[0065] According to one embodiment, the electrical pulses generated at pixel PX are processed by the photon counting circuit PCC in the following manner:

[0066] Optional conditioning circuitry includes a preamplifier 320, which amplifies each electrical signal generated by the corresponding pixel PX.

[0067] Optional conditioning circuitry may also include a pulse shaper (not shown) for processing the amplified electrical signal for the detected photon and generating a corresponding simulated signal, which includes a pulse height, such as voltage / current or other pulses indicating the detected photon. The pulse thus generated has a predefined shape or profile. In this example, the pulse has a peak amplitude indicating the energy of the detected photon.

[0068] Energy discriminator 325 performs energy discrimination on simulated pulses. In this example, energy discriminator 325 includes multiple comparators 330, each of which compares the amplitude of the simulated signal to a corresponding threshold corresponding to a specific energy level. Adjacent thresholds define energy bins. In other words, discriminator 325 operates to determine the “height” of the incoming pulse generated by the shaper. More specifically, each comparator 330 generates an output count signal indicating whether the pulse amplitude exceeds its threshold. In this example, the output signal of each comparator generates a digital signal that includes a low-to-high (or high-to-low) transition as the pulse amplitude increases and crosses its threshold, and a high-to-low (or low-to-high) transition as the pulse amplitude decreases and crosses its threshold.

[0069] In an exemplary comparator embodiment, the output of each comparator changes from low to high when the amplitude increases and crosses its threshold; and the output of each comparator changes from high to low when the pulse amplitude decreases and crosses its threshold.

[0070] Counter 335 counts the rising (or falling) edges for each threshold. For each threshold, counter 335 may include a single counter or independent sub-counters. Optionally, in the case of only two-sided binning, there are two-sided energy bins (not shown) that perform energy binning of the counts or assign the counts to energy ranges corresponding to the range between energy thresholds. In fact, in the preferred embodiment with high throughput, there is no two-sided binning operation of binning to multiple ranges, but simply registering the counts of threshold intersections (i.e., one-sided binning). However, by taking the difference between the data in adjacent bins, the count data for one-sided binning can be transformed into two-sided binning data, as shown in the following embodiment.

[0071] The photon counting circuit provides multiple counts recorded per unit time for each bin at its output for each pixel PX. The projected (photon) count data π includes the photon count rate for each bin and pixel. For each pixel, the photon count data is the count rate for each bin. c k vector C i =(c1,…,c K ) i Where i represents the corresponding pixel, and 1 corresponds to the number of energy bins used. Exemplary values ​​for K are 3, 5, or other values. In rotating systems such as CT or C-arms, the recorded count rate may differ for different projection directions q, thus additional indexing of the projection direction can be used to supplement the above concept. In the latter case, C forms a sine curve (of the count data) in the CT embodiment. Further indexing can be applied to the corresponding position z on the imaging axis Z.

[0072] The projection count data C, quantized in this way, can then be processed by the image processor IPS. i Now we will refer to Figure 3 To explain this in more detail, Figure 3 A block diagram of the projection data processor (PDP) is shown.

[0073] The projection data processor (PDP) receives projection count data π generated by the photon counting circuit (PCC) of the photon counting detector (PCD). The components of the projection data processor (PDP) include a conversion component (CC).

[0074] The conversion component CC allows the count data π to be converted into projected data that approximates or resembles the line integral (intensity) value that can typically be registered by an energy integral detector.

[0075] The operation of the conversion unit CC is based on calibration data stored in memory. This calibration data may be referred to herein as Count to Path Length (C2PL) calibration data. More specifically, in the calibration procedure (referred to herein as C2PL calibration to distinguish it from the hereinafter particularly...), Figure 8 During another type of calibration discussed herein, certain configurations of calibration materials are placed within the X-ray beam XB of the imager IA1. Suitable materials (“phantoms”) are water, (one or more) polymers (e.g., The material may be either Teflon or a metal (e.g., tin, aluminum, K-edge materials (e.g., AU, Bi, Pb) or other materials). The material should have at least two different thicknesses. A single exposure may be sufficient, or multiple articles of the material may be stacked to achieve different thicknesses, and two more exposures may be required, as reported, for example, in R. Alvarez’s “Estimator for photon counting energy selective X-ray imaging with multi-binpulse height analysis” (published in Med. Phys., Vol. 38, No. 5, May 2011). In practice, due to pulse stacking effects, the relationship between path length and photon flux is non-linear, thus enabling a wide range of different material thicknesses, see, for example, Glenn F. Knoll, John Wiley & Sons’ “Radiation Detection and Measurement” (4th edition, September 24, 2010).

[0076] The calibration operation produces pairs of data points (c, l) calib In other words, the length l of each known path calib (Phantom thickness) is associated with the count data c measured during the phantom calibration process.

[0077] The operation of the conversion component CC according to one embodiment can now be explained in more detail, assuming calibration data (c,l) calib The count rate c is known and stored in memory. The count rate c is expressed logarithmically as log count. It will be very convenient. The conversion unit CC is configured to calculate the path length l for the “optimal” interpretation, such as the path length l for a given photon count measured for different energy bins per pixel. The calculation of the optimal path length l can be carried out by formulating a minimization problem based on a cost function that quantifies how well the length interprets the count rate. The cost function envisioned in the embodiment is based on the least squares method, specifically the weighted least squares (WLS) method. In WLS, the “weights” are independent of the simulator weights, which will be discussed later. In the WLS method, a covariance matrix with cross-error terms is used. Applying WLS for this purpose, the optimization can be expressed in the embodiment as:

[0078]

[0079] Among them, COV -1 It is a photon count for each pixel. The covariance matrix COV is the reciprocal of the projection of the path length l for a specific measurement. Each off-diagonal entry i,j in COV represents the correlation between counts in bins i,j, where the diagonal entries represent the corresponding statistical variance for each bin. If a two-sided bin counter is used, the off-diagonal entries are zero (in the absence of pulse stacking). The number of rows and columns in COV and COV... -1 The number of rows and columns in the table equals the number of boxes.

[0080] Only the path length l is known calib Calibration data at the location To minimize (l), we need to know the counts at all path lengths l. It is possible to use linear or cubic spline interpolation based on known data points calculate By doing so, even in areas with strong accumulation, it is possible to estimate the correct (e.g., water) equivalent path length. The calibration data C2PL can be arranged and stored in a lookup table (LUT).

[0081] Phantom-based C2PL calibration aims to replace air normalization, water jet hardening correction, channel correction, and Hounsfield correction currently used in conventional CT preprocessing chains. It should be understood that the WLS method is only one exemplary embodiment for solving the C2PL problem. Other cost functions can also be used. Moreover, the minimization problem can be rewritten as a dual problem as a maximization problem to maximize the utility function. Solving (1) does not necessarily produce a global minimum, but rather a local minimum. If the optimization problem for C2PL calibration is solved iteratively, a stopping condition can be defined.

[0082] The conversion unit allows the photon counting imager IA1 to be operated in "normal mode," that is, as if it were an imager with a conventional energy integration imager.

[0083] However, the applicant has observed that simply applying this conversion from count data to path length may not faithfully simulate an energy integral detector. For example, the applicant has found that there may be local discrepancies of up to 30% HU units between images reconstructed from count data via path length conversion and corresponding images of the same object obtained using an energy integral detector. This discrepancy can be confusing for clinicians.

[0084] To generate more faithful energy integral images, the projection data processor (PDP) includes, in addition to the conversion unit (CC), another component referred to herein as the counting data transformer (DT). The operation of the counting data transformer (DT) is based on the aforementioned weights w (referred to herein as simulator weights w).

[0085] The counting data transformer (DT) operates by applying simulator weights to the projected counting data π received from the photon counting detector at the input port IN. Preferably, the weights are applied before processing by the data transformer (DT). The application of weights to the projected counting data can be done multiplicatively on data from one bin or preferably on data from both bins. However, this may not preclude other application modes, such as applying weights additively or by using a more complex functional expression F (the π function). F may be referred to herein as the weight application operator. In a preferred embodiment, a simple multiplicative application is used, where the corresponding weight w is multiplied by the counting data for each pixel, and, if applicable, also by the counting data for each imaging direction q and / or z position. The weight w may comprise one or more sets of numbers. Each set can be stored and processed as a vector. Here, the length K of each such "weight vector" is equal to the number of bins. The weight vector can depend on pixel i. This is achieved by counting c in each bin. k Multiply by the corresponding weight The corresponding weight vector w i The corresponding counting vector C applied to a given pixel position i i Alternatively, in this embodiment, the same weight vector can be used for all pixels i, thus the weight vector is pixel-independent. The same one or more weight vectors can be used for all projected data, or it can further depend on the scan direction q.

[0086] Preferably, before applying weights to the counting data π, and if the photon counting circuit PCC has not yet used two-sided bins, it may be necessary to first transform the counting data based on one-sided (OS) bins into two-sided (TS) bin data before applying weights. The process of obtaining the weights will be explained in more detail below.

[0087] The operation of the counting data transformer DT will now be explained in more detail, this component being implemented by applying simulator weights. Modify the count using a high-fidelity simulated energy integral detector. The aforementioned binning transformation requirement from optional one-sided counting to two-sided counting can be implemented as follows:

[0088] c TS (boxing) = c OS (boxing + 1) - c OS (Separate boxes) (2a)

[0089] c TS (Separate boxes) max ) = c OS (Separate boxes) max (2b)

[0090] Then the weights can be applied to each bin (e.g., multiplicatively) as follows:

[0091]

[0092] Optionally, the weighted counts are transmitted back to one side of the bin and then fed into the conversion unit CC required to calculate the path length of the material (e.g., water). The same weight w used by the data transformer DT for the scan projection data π is also applied to the count data c in the C2PL calibration data. This ensures the consistency and comparability of the count data in the C2LP calibration data and the count data in the scan projection data π after transformation by the data transformer DT.

[0093] For a detector with 3 energy bins and a weight vector w = (w1, w2, w3), the bin counting transformation from two sides to one side can be compactly written and implemented using matrix operations as follows:

[0094]

[0095] Emulator weights may typically be different or different for each pixel location. However, in a preferred embodiment, constant weights are used for some or all pixels and the acquisition direction q and / or z position. An array structure is used to store the corresponding simulator weights for each given pixel location. Weights can also be further differentiated based on the acquisition direction q and z position; in this case, a higher-dimensional matrix structure can be used to store the weights associated with the pixel location and any one or more of the acquisition projection directions q and z.

[0096] To summarize the operation of the projection data processor, in use, after acquiring the projection count data π, the projection count data π is received at the input port IN and passed to the counting data transformer DT. The simulator weights can be retrieved from the weight memory WM. The counting data transformer DT converts the data into transformed weighted count data based on the retrieved simulator weights. The transformation unit CC then maps the thus transformed count data onto the path length data using C2PL calibration data (which is also transformed into weighted count calibration data) stored in the calibration memory (not shown). The C2PL data and simulator weights can be stored together in a single memory, or the two datasets can be stored in different memories.

[0097] The PDP outputs the approximate path length for each pixel at the output interface OUT, and, if applicable, the approximate path length for each projection direction q and / or position z. The reconstructor RECON can then use this path length data provided at the output interface OUT to reconstruct the HU image in the image domain. This reconstructed image can be considered a simulation result of the image reconstructed from the data acquired by the energy integration detector. The visualizer can map the simulation image onto a suitable color or grayscale scheme. The video circuitry then presents the associated image color or grayscale values ​​as a visualization result on the display device DD.

[0098] The proposed IPS system may also include a user interface (UI) that allows users to interactively change stored weights previously calculated in the automatic optimization scheme, as explained in more detail below. The user interface (e.g., a graphical user interface) includes suitable input widgets, such as sliders, which allow users to change the values ​​of individual weights pixel-by-pixel or in subgroups. The user interface can be communicatively coupled to the reconstructor. Each user-requested change to the weights triggers the reconstructor to re-reconstruct, and this process can continue until the user is satisfied that the image now appears as if it were obtained from an energy integral detector.

[0099] Now for reference Figure 4 , Figure 4 A block diagram of a computerized system configured to obtain the aforementioned simulator weights W is shown. The described setup uses an existing imaging device IA2 with an energy integration detector EID combined with a photon counting imager IA1 as described above. This system allows the output of the photon counting detector PCD to be tuned to the imaging properties of the existing energy integration detector EID.

[0100] The experimental procedure based on this system can be performed as follows: Place the phantom PH in the examination area of ​​the conventional imaging device IA2. Then operate the X-ray source XR2 of the imaging device IA2 to generate a beam that crosses the phantom and allow the energy integration detector EID to detect the conventional energy integration projection data π. E Energy integral projection data π E This includes measurements of intensity per pixel and line integrals, each representation relating to the integrated energy deposited by incident photons of different energies after passing through the phantom PH. The phantom PH is the test object, preferably made of one or more materials (e.g., water).

[0101] If a tomographic setup is envisioned, the X-ray source XR2 will rotate on a rotating orbit around the phantom object PH, as described above regarding the photon counting imager IA1, while the detector EID collects projection data π from different directions. EThe regular projection data π collected by the Energy Integration Detector (EID) is then processed by the Reconstructor (RECON). E To obtain one or more cross-sectional image data associated with the conventional energy integration detector AI2. E For the purposes of this document, this image data is IM. E This can be referred to as energy integration test or target image IM. E Or calibration images. Calibration images (IM) can be acquired at a relatively high X-ray dose. E To reduce image noise. In this regard, "high" is understood to mean a higher dose than that used for imaging the patient.

[0102] Then, the same or similar phantoms are imaged by the spectral imager IA1 with a photon counting detector PCD in the manner described above to obtain the projection counting data π as described above, which is now referred to herein as π. C .

[0103] Then calibrate the image IM E and counting data π C The input is forwarded to the data fitter DF. The data fitter DF calculates the simulator weights based on the input. In an embodiment, the data fitter may implement an iterative optimization algorithm to adjust the initial weight set w0, such that it can calculate the simulator weights based on the count data π. C The reconstructed image, after being converted to path length, corresponds to the test image IM. E This conversion can be accomplished by the data fitter DF using a component similar to the conversion component CC discussed above regarding the projection data processor PDP.

[0104] Two images (i.e., according to π) C Reconstructed image vs. test image IM E The correspondence or matching between them can be achieved through a suitable similarity metric μ (e.g., root mean square distance) or through at least a squared function or by using any other suitable similarity metric (e.g., L). pThe norm (where p is not 1 / 2) is used for quantification. The metric μ can be evaluated across the entire image or only within one or more ROIs and can be formulated as an objective function F, depending on w0. The initial weight set w0 can be adjusted using an iterative optimization scheme (e.g., Newton-Raphson, conjugate gradient, Nelder-Mead, or any other scheme). The optimization algorithm is configured to adjust the initial weight set over one or more iterations to minimize the objective function. It should be understood that optimization does not necessarily produce a global minimum but may return an approximation of a local minimum; however, this may be sufficient for most purposes. A stopping condition can be defined based on a predefined threshold. The weights are adjusted until the objective function returns a value below the predefined threshold. Once the stopping condition is met, the data fitter DF outputs the current weight set as output. This output weight set is the simulator weights. The simulator weights can be stored in a weight memory WM.

[0105] Instead of using the experimental methods based on actual measurements described above, the weights can also be calculated theoretically through modeling and simulation. In this embodiment, this can be accomplished by calculating the spectra or more specifically the count data of the two detectors' PCD and EID based on the corresponding detector specifications, the initial spectrum of the phantom, and material properties. Detector specifications may include detector absorption efficiency. Phantom material properties may include one or more Z atomic numbers, material density, etc. The initial spectrum can be calculated based on the specifications of the X-ray source (including cathode current, tube voltage, and anode material). Approximate results of the output spectrum in response to exposure to the two detectors can then be calculated by running, for example, Monte Carlo simulations, where both scattering and absorption events can be accounted for using other known suitable models for the corresponding cross-sections (e.g., the Klein-Nishina model for the scattering cross-section), while the photoelectric absorption cross-section can be calculated using the familiar Z / E ratio. 3 The law is used for modeling. Monte Carlo (MC) simulations can be implemented using the probabilistic Metropolis-Rosenbluth-Teller algorithm, described by N. Metropolis et al. in "Equation of State Calculations by Fast Computing Machines" (J. Chem. Phys., Vol. 21, p. 1087, 1953). Other implementations and / or non-MC simulations are also envisioned.

[0106] Exemplary spectra (intensity versus energy curves) for the two detectors can be obtained through the calculations described above, such as... Figure 5As shown. Specifically, the spectrum of photon counting (including the spectral response function) and the spectrum for the energy integration detector are illustrated with respect to attenuation over 40 cm of water. The curve with two peaks is the spectrum for the energy integration detector.

[0107] From the spectrum calculated in this way, the ratio r of each energy can be obtained, such as Figure 6 An exemplary ratio curve is shown. Interestingly, the ratio curve is not strongly dependent on the phantom thickness. Figure 6 Four exemplary curves are shown, representing water thicknesses of 10, 20, 30, and 40 cm (from top). Therefore, a given phantom with a fixed thickness (e.g., 40 cm) can be simulated. For each energy value E, the ratio r can be formulated as r(E) = I e / I p , among which, I e It refers to the intensity for each energy range of the energy integral detector, and I p This refers to the intensity for each energy range of the photon counting detector. The simulator weights can then be derived as averages over r(E). Specifically, the count data in a given bin i for a given energy range ERN can be obtained by multiplying the count rate in bin i by the average ratio. The average ratio can be obtained by averaging the ratios r(E) of all energy values ​​E within a given range ERN captured by bin i. In other words, the simulation weights are the average of the ratio curves of the energy included within a given energy range of a bin.

[0108] Now for reference Figure 7 , Figure 7 A flowchart of an image processing method as envisioned in the embodiments described herein is shown. The method can be implemented using the steps described below. Figure 1-4 The above-mentioned systems, however, can also be understood as their own teachings, and not necessarily as... Figure 1-4 The architecture of the system discussed in the text is related.

[0109] At step S710, energy-resolved counting data generated by the photon generation detector (PCD) is received. Preferably, the photon counting detector has at least two image bins, but more preferably it has two or more (e.g., three, four, five or even more) energy bins.

[0110] In step S720, the counting data is transformed into transformed counting data. This transformation is implemented by applying one or more weights to the counting data. These weights are previously obtained based on: i) data for the photon counting detector or different photon counting detectors, and additionally ii) data related to the energy integration detector. Data for the photon counting detector may include projected counting data acquired during imaging of a test object (e.g., a phantom comprising one or more suitable materials, such as water, aluminum, Derwent, etc.). In embodiments, this data may be calculated through simulation based on detector specifications, X-ray source specifications, and the material properties of the virtual test object. Data related to the energy integration detector may include projected data acquired during imaging of the test object phantom, or preferably include a test image reconstructed from such projected data. Any reconstruction algorithm may be used for this purpose, such as FBP or iterative reconstruction, algebraic reconstruction, or other reconstruction methods.

[0111] The energy integral detector can also be referred to as the target because what people want to simulate by applying weights is the imaging properties of the energy integral detector.

[0112] At step S730, the transformed count data is then converted into material path length data. This path length data refers to the conceptual thickness of a reference material (e.g., water or a water-equivalent material, or other materials with similar radiometric density). In this way, each count for each pixel is associated with a corresponding path length. The proposition is that, for a given pixel location, if a material of a specified thickness is exposed to the X-ray radiation used to acquire the count data, one will observe approximately the same count as received for that pixel at step S710. The path length data can be further converted into line integrals or other attenuation-based scales.

[0113] At step S740, the obtained path length data or line integral data through the material is then reconstructed into image data using any reconstruction algorithm in a known manner. The image data can be mapped to HU values. The image data thus obtained is considered a simulation of the image obtainable by the real or virtual energy integral detector involved in the weight calculation. Preferably, the reconstruction algorithm used in step S740 is the same as the reconstruction algorithm used in the weight calculation, as described below. Figure 8 The lieutenant general further explained.

[0114] At optional step S750, the reconstructed image data is then output, for example, displayed on a display device, stored in memory, or otherwise processed.

[0115] In optional step S760, the applied weights can be changed in response to a user request. In this case, the changed weights are reapplied and the above method steps are repeated starting from step S730 to generate a new image at step S740, which is preferably displayed in step S760. In this way, the user can use trial and error to fine-tune the weights to obtain an image that, according to the user's opinion, looks like an image obtained using an energy integration detector.

[0116] Now refer to the calibration method shown. Figure 8 The flowchart. In particular, the calibration method described below can be used to calculate in Figure 7 The simulator weights used in the method.

[0117] In step S810A, energy resolution calibration data from the photon counter detector is received.

[0118] At step S810B, calibration image data for the energy integration detector is received.

[0119] In step S820, a data fitting operation is performed by adjusting the initial set of weights to be applied to the energy-resolved data, so that image data IM can be reconstructed from the weighted data. p Corresponding to the calibration image data IM for the energy integration detector e .

[0120] An automatic optimization scheme can be used based on a suitable objective function that quantizes the two reconstructed images IM. p IM e The difference between them. Then, the simulator weights are obtained by adjusting the initial weights in one or more iterations to minimize the objective function f, which is a function of the weights, until the stopping condition is met. This optimization problem can be formally written as:

[0121] f w =μ[R(cc(F(π)] p ,w)),R(π e (5)

[0122] argmin w f w (6)

[0123] Where R(·) is the reconstruction algorithm;

[0124] μ(·) is the corresponding measure of the degree of matching of the reconstructed image;

[0125] cc(·) is the operator for converting counts to path lengths;

[0126] F(·,·) is the weight application operator.

[0127] In this embodiment, the metric μ is a least-squares function of the pixel-wise difference between the two images. However, other suitable distance metrics, such as RMS and L, can be used alternatively. p Norm (p is not 1 / 2).

[0128] At step S830, the weights or simulator weights are then output. The simulator weights w can be stored in memory WM.

[0129] If one wishes to simulate the imaging capabilities of an energy integral detector, then when imaging using a photon counting detector, weights can be retrieved to be applied to the counting projection date, as described above. Figure 3 and Figure 7 As explained in the text.

[0130] It should be understood that the above-described method for generating weights can be experimentally performed by imaging a test object such as a phantom, and the data received at input steps S810a and S810b is then derived according to the calibration imaging procedure. Preferably, the phantom imaged to obtain the input data received at receiving steps S810a and S810b can comprise a variety of different materials. In embodiments, the phantom comprises different polymers containing different densities (e.g., polystyrene, etc.). The module could be any one or more of the following: an air shell and a water bag. However, the method used to generate the weights could also be theoretically practiced through computational simulations based on the specifications of the X-ray source, the specifications of the two detectors, and the material specifications of the current virtual phantom (e.g., Monte Carlo methods or other methods).

[0131] One or more features disclosed herein may be configured or implemented as circuitry encoded in a computer-readable medium, or in conjunction with or implemented with circuitry encoded in a computer-readable medium, and / or combinations thereof. Circuitry may include discrete circuitry and / or integrated circuits, application-specific integrated circuits (ASICs), systems-on-a-chip (SoCs) and combinations thereof, machines, computer systems, processors and memories, and computer programs.

[0132] In another exemplary embodiment of the present invention, a computer program or computer program unit is provided, characterized in that it is adapted to run method steps of a method according to an embodiment of the foregoing embodiments on a suitable system.

[0133] Therefore, a computer program unit can be stored in a computer unit, and this computer program unit can also be part of an embodiment of the present invention. The computing unit can be adapted to perform or cause the execution of steps of the above-described method. Furthermore, the computing unit can be adapted to operate components of the above-described apparatus. The computing unit is capable of automatically operating and / or executing user commands. The computer program can be loaded into the working memory of a data processor. Therefore, a data processor can be equipped to execute the method of the present invention.

[0134] This exemplary embodiment of the invention covers both computer programs that use the invention from the outset and computer programs that use the invention by means of updating existing programs.

[0135] In addition, the computer program unit may be able to provide all the necessary steps to complete the process of the exemplary embodiments of the method described above.

[0136] According to another exemplary embodiment of the present invention, a computer-readable medium, such as a CD-ROM, is provided, wherein the computer-readable medium has computer program units stored on the computer-readable medium, the computer program units being described in the preceding sections.

[0137] Computer programs can be stored and / or distributed on suitable media (in particular, but not necessarily non-transient media), such as optical storage media or solid-state media supplied together with or as part of other hardware, but can also be distributed in other forms, such as via the Internet or other wired or wireless telecommunications systems.

[0138] However, computer programs can also exist on networks (such as the World Wide Web) and can be downloaded from such networks into the working memory of a data processor. According to another exemplary embodiment of the invention, a medium is provided for making computer program units available for download, said computer program units being arranged to perform a method according to an embodiment of the previously described embodiments of the invention.

[0139] It should be noted that embodiments of the present invention are described with reference to different subjects. In particular, some embodiments are described with reference to method claims, while others are described with reference to apparatus claims. However, unless otherwise stated, those skilled in the art will infer from the above and below that any combination of features relating to different subjects, in addition to any combination of features belonging to one type of subject matter, is also considered to be disclosed in this application. However, all features can be combined to provide synergistic effects beyond the simple addition of features.

[0140] Although the invention has been illustrated and described in detail in the accompanying drawings and the foregoing description, such illustrations and descriptions should be considered illustrative or exemplary, and not restrictive. The invention is not limited to the disclosed embodiments. Those skilled in the art, through studying the drawings, the disclosure, and the claims, will understand and implement other variations of the disclosed embodiments in practicing the claimed invention.

[0141] In the claims, the word "comprising" does not exclude other elements or steps, and the words "a" or "an" do not exclude a plurality. A single processor or other unit can perform the functions of several items recited in the claims. Although certain measures are recited in different dependent claims, this does not indicate that combinations of these measures cannot be advantageously used. No reference numerals in the claims should be construed as limiting the scope.

Claims

1. An image processing system for simulating an energy integration detector (EID) using a photon counting detector (PCD), comprising: The input interface (IN) is used to receive energy-resolved counting data generated by the photon counting detector (PCD); A counting data converter (DT) is configured to perform a transformation operation to transform the energy-resolved counting data into transformed counting data, the operation including: Retrieve one or more weights from the weight memory (WM), which were previously obtained based on: i) calibration image data for the photon counting detector or different photon counting detectors and ii) calibration image data for the energy integration detector; Apply the one or more weights to the energy-resolved count data; A data converter component (CC) configured to convert transformed counting data into path length data through the material; and The output (OUT) interface is used to output the path length data through the material.

2. The system according to claim 1, wherein, The data conversion unit (CC) is configured to convert the transformed count data into path length data through the material based on one or more weights that are the same as the weights applied to the energy-resolved count data.

3. The system according to any one of the preceding claims, wherein, The one or more weights are the same for all pixels of the photon counting detector.

4. The system of claim 1 or 2, further comprising a reconstructor (RECON) configured to reconstruct the path length data through the material into image data.

5. The system according to claim 1 or 2, wherein, The material includes water or a material with a similar radiometric density.

6. The system of claim 1 or 2, further comprising a user interface (UI) configured to allow a user to change at least one of the weights.

7. An imaging arrangement (IAR) comprising the system according to any one of the preceding claims and an imaging device (IA) having the photon counting detector (PCD).

8. An image processing method for simulating an energy integration detector (EID) using a photon counting detector (PCD), comprising: Receive (S710) energy-resolved counting data generated by the photon counting detector (PCD); The energy-resolution counting data is transformed (S720) into transformed counting data, the transformation operation including: Retrieve one or more weights from the weight memory (WM), which were previously obtained based on: i) calibration image data for the photon counting detector or different photon counting detectors and ii) calibration image data for the energy integration detector; Apply the one or more weights to the energy-resolved count data; The transformed count data is converted (S730) into path length data through the material; and Output (S750) the path length data through the material.

9. A computer program product, when run by at least one processing unit (PU), adapted to cause said processing unit (PU) to perform the method according to claim 8.

10. A computer-readable medium having thereon a computer program product according to claim 9.

Images