Material weighting of projection based spectral x-ray imaging

Abstract

To provide a system and method for projection-based spectral X-ray imaging such as computed tomography (CT) imaging for improving image quality such as reduction of noise, improvement of a contrast-to-noise ratio (CNR) and patient dose efficiency.SOLUTION: Provided is a system and method that includes performing a projection-based material decomposition based on spectral X-ray data to generate a set of material-based sinograms, and performing a weighted combination of at least portions of at least two material-based sinograms of the set of material-based sinograms into a reconstructed image based on material weighting in the projection domain.SELECTED DRAWING: Figure 11

Description

Technical Field

[0001] The proposed technology relates to X-ray technology and X-ray imaging, and more specifically, to reconstructing projection images. Specifically, the proposed technology relates to an X-ray imaging system, such as a computed tomography (CT) imaging system configured for reconstructing projection images using projection-based spectral X-ray imaging, and a corresponding computer program product, aiming to improve image quality.

Background Art

[0002] Radiation imaging, such as CT imaging systems and other more general X-ray imaging systems, has been used for many years in medical applications such as medical diagnosis and treatment.

[0003] A typical X-ray imaging system, such as a CT imaging system, includes an X-ray source, an X-ray detector, and an attached image processing system. The X-ray detector includes a number of detector modules each containing one or more detector elements for independent measurement of X-ray intensity. The X-ray source emits X-rays, which pass through the subject or object to be imaged and are received by the X-ray detector. The X-ray source and the X-ray detector are typically provided on a rotating member of a gantry and are configured to rotate around the subject or object. The emitted X-rays are attenuated by the subject or object as they pass through it, and the resulting transmitted X-rays are measured by the X-ray detector. The X-ray detector is coupled to a digital acquisition system (DAS), and the measured X-ray data is transferred to the image processing system to reconstruct an image of the subject or object.

[0004] Referring to FIG. 1(A), it will be useful to briefly overview an example of a general X-ray imaging system according to the prior art. This exemplary X-ray imaging system 100 includes an X-ray source 10, an X-ray detector 20, and an attached image processing system 30. Generally, the X-ray detector 20 is optionally focused by an X-ray optical system or collimator, which is an optional element, and is configured to record the radiation from the X-ray source 10 that has passed through an object, a subject, or a part thereof. The X-ray detector 20 can be connected to the image processing system 30 via a suitable readout electronic circuit that is at least partially integrated into the X-ray detector 20 to enable image processing and / or image reconstruction by the image processing system 30.

[0005] As an example, a conventional CT imaging system includes an X-ray source and an X-ray detector configured in such a way that projection images of an object or a subject can be acquired at various viewing angles over at least 180 degrees. This is most commonly achieved by mounting the source and the detector on a support member, such as a rotating member of a gantry, that is capable of rotating about the object or the subject. An image that includes projections recorded by different detector elements at different viewing angles is called a sinogram. Hereinafter, even when the detector is two-dimensional (2D) and the sinogram is converted into a three-dimensional (3D) image, the set of projections recorded by different detector elements at different viewing angles will be referred to as a sinogram.

[0006] FIG. 1(B) is a schematic diagram showing an example of an X-ray imaging system setup according to the prior art, showing the projection lines from the X-ray source through the object to the X-ray detector.

[0007] A further developed form of X-ray imaging is energy-resolved X-ray imaging, also known as spectral X-ray imaging, in which X-ray transmission is measured for several different energy levels. This technique can be achieved either by rapidly switching the radiation source between two different emission spectra, using two or more X-ray sources that emit different X-ray spectra, or using an energy-discriminating detector that measures the incident radiation at two or more energy levels. An example of such a detector is a multi-bin photon-counting detector, in which each recorded photon generates an electric current pulse, and by comparing this pulse with a set of threshold values, the number of photons incident on each of a predetermined number of energy bins is counted.

[0008] Projection images of each energy level are obtained from spectral X-ray projection measurements. As described in Phys. Med. Biol. 30, 519 (Non-Patent Document 1) by Tapiovaara and Wagner, a weighted sum of these projection images can be obtained to optimize the contrast-to-noise ratio (CNR) for a given imaging task.

[0009] Another approach enabled by energy-resolved X-ray imaging is basis material decomposition. This approach utilizes the fact that all materials composed of elements with low atomic numbers, such as human tissues, can be well approximated as a linear combination of two (or more) basis functions in terms of their energy dependence of the linear attenuation coefficient. That is, μ(E)=a 1 f 1 (E)+a 2 f 2 (E) wherein, in the formula, f 1 and f 2 are basis functions, and a 1 and a 2 are the corresponding basis coefficients. More generally, f i is a basis function, and a iis the corresponding basis coefficient, where i = 1, …, N and N is the total number of basis functions. If there is one or more elements in the imaged volume whose atomic number is large enough such that the K absorption edge is within the energy range used for imaging, then one basis function must be added for each such element. In the field of medical imaging, such K-edge elements can typically be iodine or gadolinium, substances used as contrast agents.

[0010] Basis material decomposition is described in Alvarez and Macovski, Phys Med Biol. 1976; 21(5):733 - 744 (Non - Patent Document 2). In basis material decomposition, each integral A i =∫ l a i dl (where i = 1, …, N and N is the number of basis functions) is inferred from the measurement data in each projection ray l (ell) from the source to the detector element. In one implementation, this inference is first achieved by expressing the expected number of recorded counts in each energy bin as a function of A i .

Equation

[0011] where λ i is the expected number in energy bin i, E is the energy, and S i is the response function, which depends on the spectral shape incident on the imaging object, the quantum efficiency of the detector, and the sensitivity of energy bin i to X - rays with energy E. The term energy bin is most commonly used for photon - counting detectors, but the above equation can also describe other energy - resolved X - ray imaging systems such as multi - layer detectors, kVp - switched sources, or multi - source systems.

[0012] Then, using the maximum - likelihood method, under the assumption that the number in each bin is a random variable following a Poisson distribution, A ican be estimated. This is achieved by minimizing the negative log-likelihood function. See, for example, Phys. Med. Biol. 52 (2007), 4679-4696 by Roessl and Proksa (Non-Patent Document 3).

Number

[0013] where m i is the number of measurements in energy bin i, and M b is the number of energy bins.

[0014] When the estimated basis coefficient line integrals

Number

[0015] Improving the image quality of X-ray imaging systems is undoubtedly an important area for ensuring the quality and safety of patient care, and is related to various approaches not only within the field of spectral X-ray imaging.

[0016] Therefore, there is still a wide demand for improving image quality in terms of noise reduction, improvement of contrast-to-noise ratio (CNR), and improvement of patient dose efficiency, etc.

Prior Art Documents

Patent Documents

[0017]

Patent Document 1

Non-Patent Document

[0018]

Non-Patent Document 1

Non-Patent Document 2

Non-Patent Document 3

Summary of the Invention

[0019] This summary presents concepts that are further described in more detail in the detailed description. Such description should not be used to identify essential features of the claimed subject matter or to limit the scope of the claimed subject matter.

[0020] According to one aspect, a method for projection-type spectral X-ray imaging is provided. The method includes performing projection-based (projection-based) material decomposition based on spectral X-ray data to generate a set of material basis sinograms, and performing weighted combination of at least a portion of at least two of the set of material basis sinograms based on material weighting in the projection region to form a reconstructed image.

[0021] According to another aspect, there is provided a CT imaging system including an X-ray source configured to emit X-rays, an X-ray detector configured to generate spectral X-ray data, and a processor. The processor is configured to perform projection-based material decomposition on the spectral X-ray data to generate a set of material basis sinograms. The processor is further configured to perform weighted combination of at least portions of at least two of the set of material basis sinograms based on material weighting in the projection region to form a reconstructed image.

[0022] The proposed technique enables one approach to reconstructing an image for projection-based spectral X-ray imaging. The image reconstruction is based on adaptive / dynamic material weighting in the projection region, i.e., reconstructed from a set of material basis sinograms. The proposed technique takes into account the improvement of the mapping of the material basis sinograms, thus improving the quality of the reconstructed image. It should be understood that the X-ray imaging may be CT imaging and the reconstructed image may be a reconstructed CT image. The quality improvement may be related to, for example, an improvement in CNR, noise reduction, and / or enhancement of some features of the image such as anatomical features. In addition, the proposed technique enables increasing the patient dose efficiency without suffering an unacceptable loss in image quality. Further, the proposed technique does not require any additional calibration during use, thus further facilitating the implementation of the technique. Also, it is noted that the proposed technique may be suitable for all photon-counting X-ray detectors, as well as other dual-source systems such as X-ray systems and mammography systems.

Brief Description of the Drawings

[0023] Various aspects of the present disclosure will be more fully understood upon reading the detailed description with reference to the accompanying drawings.

[0024]

Figure 1A

Figure 1B

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8A

Figure 8B

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13A

Figure 13B

Figure 14

Figure 15

Mode for Carrying Out the Invention

[0025] Hereinafter, embodiments of the present disclosure will be described as examples with reference to the drawings.

[0026] For a more thorough understanding, it will be useful to continue with a preliminary description of a non-limiting example of an overall X-ray imaging system in which data processing and transfer according to the concepts of the present invention can be implemented.

[0027] FIG. 2 is a schematic diagram showing an example of an X-ray imaging system 100 such as a CT imaging system. This system 100 includes an X-ray source 10 that emits X-rays, an X-ray detector 20 by an X-ray detector that detects the X-rays after passing through the object, an analog processing circuit 25 that processes and digitizes the raw electrical signals from the X-ray detector, a digital processing circuit 40 that can perform further processing operations such as applying corrections, temporarily storing, or filtering the measurement data, and a computer 50 that stores the processed data and can perform further post-processing and / or image reconstruction. The digital processing circuit 40 may include a digital processor. According to an example of the embodiment, all or part of the analog processing circuit 25 may be implemented in the X-ray detector 20. The X-ray source and the X-ray detector may be coupled to the rotating member of the gantry 11 of the CT imaging system 100.

[0028] The overall X-ray detector can be regarded as a combination of the X-ray detector system 20, that is, the X-ray detector 20 and the attached analog processing circuit 25.

[0029] There is an image processing system 30 in communication with and electrically coupled to the analog processing circuit 25, and the image processing system 30 can include a digital processing circuit 40 and / or a computer 50 and can be configured to perform image reconstruction based on image data from an X-ray detector. Thus, the image processing system 30 can be regarded as a computer 50, or alternatively, can be regarded as a combined system consisting of a digital processing circuit 40 and a computer 50, or possibly, if the digital processing circuit is further specialized for image processing and / or image reconstruction, the image processing system 30 can also be regarded as the digital processing circuit 40 itself.

[0030] An example of a widely used X-ray imaging system is a CT imaging system, which can include an X-ray source or X-ray tube that generates a fan beam or cone beam of X-rays and an array of opposing X-ray detectors that measure the portion of the X-rays that have passed through a patient or object. The X-ray source or X-ray tube and the X-ray detectors are mounted on a gantry 11 that can rotate about the imaging subject.

[0031] FIG. 3 schematically illustrates a CT imaging system 100 as an example for the description of an X-ray imaging system. The CT imaging system includes a computer 50 that receives commands and scanning parameters from an operator via an operation console 60. The operation console 60 may have a display 62 and some form of operator interface, such as a keyboard, mouse, joystick, touch screen, or other input device. Then, the commands and parameters supplied by the operator are used by the computer 50 to provide control signals to an X-ray controller 41, a gantry controller 42, and a table controller 43. Specifically, the X-ray controller 41 provides a power signal and a timing signal to the X-ray source 10 to control the emission of X-rays to an object or a patient located on the table 12. The gantry controller 42 controls the rotation speed and position of the gantry 11 including the X-ray source 10 and the X-ray detector 20. By way of example, the X-ray detector 20 may be a photon-counting X-ray detector. The table controller 43 controls and determines the position of the patient table 12 and the scanning range of the patient. There is also a detector controller 44 configured to control and / or receive data from the X-ray detector 20.

[0032] In one embodiment, the computer 50 also performs post-processing and image reconstruction of the image data output from the X-ray detector 20. Thereby, the computer 50 corresponds to the image processing system 30 as shown in FIGS. 1 and 2. The attached display 62 enables the operator to observe the reconstructed image and other data from the computer 50.

[0033] The X-ray source 10 disposed on the gantry 11 emits X-rays. The X-ray detector 20 may be in the form of a photon-counting X-ray detector and detects the X-rays after they have passed through an object or a patient. The X-ray detector 20 may be formed, for example, by a plurality of pixels, also called sensor or detector elements, and an attached image processing circuit such as an application-specific integrated circuit (ASIC) disposed in a detector module. The analog processing part is implemented at the pixel, and all the remaining processing is implemented, for example, in the ASIC. In one embodiment, the image processing circuit (ASIC) digitizes the analog signals from the pixels. The image processing circuit (ASIC) can also include digital processing, and the digital processing can perform further processing operations such as applying corrections to, temporarily storing, or filtering the measurement data. During the scan for acquiring X-ray projection data, the gantry and the components attached to the gantry rotate about the isocenter 13.

[0034] Today's X-ray detectors typically need to convert the incident X-rays into electrons, which typically occurs through the photoelectric effect or Compton interaction, and the resulting electrons typically generate secondary visible light until the energy of the electrons is lost, and then this light is detected by a photosensitive material. There are also semiconductor-based detectors, in which case the electrons generated by the X-rays create charges as electron-hole pairs, and these electron-hole pairs are collected through an applied electric field.

[0035] There are detectors that operate in an energy integration mode in the sense of providing an integrated signal from a large number of X-rays. The output signal is proportional to the total energy deposited by the detected X-rays.

[0036] X-ray detectors with photon-counting capabilities and energy resolution capabilities are becoming common in medical X-ray applications. Photon-counting detectors have the advantage that in principle, they can measure the energy of each individual X-ray to obtain additional information about the composition of the object. This information can be used to improve image quality and / or reduce the radiation dose.

[0037] Generally, a photon counting X-ray detector determines the energy of photons by comparing the height of the electrical pulses generated by photon interactions in the detector material with a set of comparator voltages. These comparator voltages are also called energy thresholds. Generally, the analog voltage of the comparator is set by a digital-to-analog converter (DAC). The DAC can convert the digital settings sent by the controller into an analog voltage and compare the height of the photon pulse with this analog voltage.

[0038] A photon counting detector counts the number of photons that interacted in the detector during the measurement time. New photons are generally identified by the fact that the height of the electrical pulse exceeds the comparator voltage of at least one comparator. Once a photon is identified, this event is recorded by incrementing the digital counter associated with the channel.

[0039] Using several different thresholds results in an energy-discriminating photon counting detector, in which the detected photons can be stored in energy bins corresponding to various thresholds. In some cases, this type of photon counting detector is also called a multi-bin detector. Generally, the energy information enables the creation of a new type of image, and new information becomes available to remove image artifacts inherent in the prior art. In other words, for an energy-discriminating photon counting detector, the pulse height is compared with N programmable thresholds (T1 to TN) in the comparator and classified according to the pulse height, and then the pulse height is proportional to the energy. In other words, a photon counting detector that includes more than one comparator is called a multi-bin photon counting detector here. In the case of a multi-bin photon counting detector, the number of photons is typically stored in a set of counters provided one for each energy threshold. For example, 1 count can be assigned to the highest energy threshold exceeded by the photon pulse. In another example, the counter records the number of times the photon pulse crossed each energy threshold.

[0040] As an example, an edge-on is a special non-limiting design of a photon counting detector, in which case an X-ray detector or an X-ray sensor such as a pixel is oriented edge-on (intersecting the edge) with respect to the incident X-rays.

[0041] For example, such a photon counting detector can have pixels in at least two directions, and one of the directions of the edge-on photon counting detector has a component in the direction of the X-rays. Such an edge-on photon counting detector is also referred to as a depth-segmented photon counting detector and has two or more depth segments of pixels in the incident direction of the X-rays. It should be noted that one detector element may correspond to one pixel, and / or a plurality of detector elements may correspond to one pixel and / or data signals from a plurality of detector elements may be used for one pixel.

[0042] Alternatively, each pixel may be arranged as an array (non-depth segmented) in a direction substantially orthogonal to the incident direction of the X-rays, and each of the pixels may be oriented edge-on with respect to the incident X-rays. In other words, this X-ray detector can be said to be non-depth segmented while still being arranged edge-on with respect to the incident X-rays.

[0043] By configuring the edge-on photon counting detector in an edge-on type, the absorption efficiency can be increased. In this case, the absorption depth can be selected to any length, and the edge-on photon counting detector can still be fully depleted without reaching an extremely high voltage.

[0044] The conventional mechanism for detecting X-ray photons through a direct-type semiconductor detector basically operates as follows. The energy of the X-ray interaction in the detector material is converted into electron-hole pairs inside the semiconductor detector, and the number of electron-hole pairs generally is proportional to the photon energy. The electrons and holes migrate towards (or in the opposite direction to) the detector electrodes and the back surface. During this migration, the electrons and holes induce a current in the electrodes, and this current can be measured.

[0045] As shown in FIG. 4, the signal is sent via a wiring path 26 from the detector element 22 of the X-ray detector to the input of an analog processing circuit (e.g., ASIC) 25. The term "application specific integrated circuit (ASIC)" should be understood to be broadly interpreted as any general circuit that is utilized and configured for a specific application. The ASIC processes the charge generated from each X-ray and converts it into digital data, and measurement data such as the number of photons and / or the estimated energy can be obtained using this data. The ASIC is configured to be connected to a digital processing circuit, and the digital data is sent to the digital processing circuit 40 and / or one or more memory circuits or components 45, and finally the data becomes the input to the image processing circuit 30 or the computer 50 in FIG. 2 to form a reconstructed image.

[0046] Since the number of electrons and holes from one X-ray event is proportional to the energy of the X-ray photon, the total charge of one induced current pulse is proportional to this energy. After the filtering step in the ASIC, the pulse amplitude is proportional to the total charge of the current pulse and thus proportional to the X-ray energy. Then, the pulse amplitude can be measured by comparing the value of the pulse amplitude with one or more threshold values (THR) of one or more comparators (COMP), and a counter can be introduced to record the number of cases where the pulse is greater than the threshold value. In this way, it becomes possible to count and / or record the number of X-ray photons having an energy exceeding the energy corresponding to each threshold value (THR) detected within a certain time frame.

[0047] The ASIC typically samples the analog photon pulse once per clock cycle and records the output of the comparator. The comparator (threshold value) outputs 1 or 0 depending on whether the analog signal exceeds or falls below the comparator voltage. The information available in each sample is, for example, 1 or 0 indicating whether the comparator was triggered (the photon pulse was higher than the threshold value) for each comparator.

[0048] Photon counting detectors typically have photon counting logic that determines whether a new photon has been recorded and records the photon in a counter (one or more). In the case of a multi-bin photon counting detector, typically there are several counters, for example one counter for each comparator, and the number of photons is recorded in these counters according to an estimated value of the photon energy. This logic can be implemented in several different ways. Two of the most common categories of photon counting logic are non-paralyzable counting mode and paralyzable counting mode. As other photon counting logic, there is, for example, peak detection that counts the maximum (local maximum) pulse height detected in a voltage pulse and potentially can record it.

[0049] Photon counting detectors have many advantages, including but not limited to high spatial resolution, lower sensitivity to electronic noise, good energy resolution, and material separation ability (spectral imaging ability), etc. However, energy integrating detectors have the advantage of a high count rate margin. The count rate margin stems from the fact / recognition that since the total energy of the photons is measured, if one additional photon is added, the output signal will always increase (within reasonable limits) regardless of the amount of photons currently being recorded by the detector. This advantage is one of the main reasons why energy integrating detectors are the standard in current medical CT.

[0050] FIG. 5 shows a schematic diagram of a photon counting circuit and / or device according to an example of an embodiment.

[0051] When a photon interacts with a semiconductor material, a cloud of electron-hole pairs is generated. By applying an electric field across the detector material, the charge carriers are collected by electrodes attached to the detector material. The signal is sent from the detector element to the input of a parallel processing circuit, for example an ASIC. In one example, the ASIC can process the charge such that a voltage pulse is generated with a maximum height proportional to the amount of energy deposited by the photons in the detector material.

[0052] The ASIC can include a set of comparators 302, with each comparator 302 comparing the magnitude of the voltage pulse against a reference voltage. The comparator output typically becomes zero or one (0 / 1) depending on which of the two compared voltages is greater. Here, assume that if the voltage pulse is greater than the reference voltage, the comparator output is 1, and if the reference voltage is greater than the voltage pulse, it is 0. Using a digital-to-analog converter (DAC) 301, the digital settings that can be supplied by a user or a control program can be converted into a reference voltage that can be used by the comparator 302. If the height of the voltage pulse exceeds the reference voltage of a particular comparator, this comparator is said to have received a trigger. Each comparator generally has an attached digital counter 303, and the counter 303 is incremented based on the comparator output according to photon counting logic.

[0053] As described above, the estimated basis coefficient line integral obtained for each projection line

Number

[0054] It will be recognized that the mechanisms and configurations described in this document can be implemented, combined, and reconfigured in a variety of ways.

[0055] For example, each embodiment can be implemented in hardware, or at least partially, in software for execution by a suitable image processing circuit, or in a combination thereof.

[0056] The steps, functions, procedures, and / or blocks described in this book can be implemented in hardware using any conventional technology, such as individual circuit technology or integrated circuit technology, including both general-purpose electronic circuits and application-specific circuits.

[0057] Alternatively or in addition, at least some of the steps, functions, procedures, and / or blocks described in this book can be implemented in software, such as a computer program for execution by a suitable image processing circuit, such as one or more processors or processing units.

[0058] In the following, non-limiting examples of specific detector module implementations will be discussed. More specifically, these examples refer to edge-on orientation detector modules and depth segmentation detector modules. Other forms of detectors and detector modules may also be feasible.

[0059] FIG. 6 is a schematic diagram showing an example of a semiconductor detector sub-module according to an example of an embodiment. The figure is an example of a detector module 21 including a semiconductor sensor having a plurality of detector elements or pixels 22, and each detector element (or pixel) is typically based on a diode having a charge collection electrode as a main component. X-rays enter through the edge of the detector module.

[0060] FIG. 7 is a schematic diagram showing an example of a semiconductor detector sub-module according to another example of an embodiment. In this example, the detector module 21 by the semiconductor sensor is also divided into a plurality of depth sections or detector elements 22 in the depth direction, and in this case, it is also assumed that X-rays enter through the edge of the detector module.

[0061] Typically, a detector element is an individual X-ray sensitive element of the detector. Generally, photon interactions occur in the detector element, and the charge thus generated is collected by the corresponding electrode of the detector element.

[0062] Each detector element typically measures the incident X-ray beam as a series of frames. One frame is measurement data in a predetermined time interval called the frame time.

[0063] Depending on the phase geometry of the detector, the detector elements can correspond to pixels, especially when the detector is a flat-panel detector. A depth-divided detector can be regarded as having a predetermined number of detector strips, and each strip can be regarded as having a predetermined number of depth sections. In the case of such a depth-divided detector, especially when an individual charge collection electrode for each depth section is attached to each depth section, each depth section can also be regarded as an individual detector element.

[0064] The detector strips of a depth-divided detector typically correspond to the pixels of a normal flat-panel detector and are thus also called pixel strips. However, it is also possible to regard a depth-divided detector as a three-dimensional pixel array, in which case each pixel corresponds to an individual depth section / detector element.

[0065] The semiconductor sensor can be implemented as a so-called multi-chip module (MCM) in the sense that the semiconductor sensor is used as a substrate for electrical wiring and, preferably, as a substrate for a predetermined number of application-specific integrated circuits (ASICs) attached through a so-called flip-chip technique. The wiring includes connections for signals from each pixel or detector element to the ASIC input and connections from the ASIC to external memory and / or digital data processing. Power to the ASIC can be provided through similar wiring taking into account the cross-sectional increase required for the large currents in these connections, but the power can also be provided through separate connections. The ASIC can be placed on the side of the operating sensor, which means that if an absorbing cover is placed on top, it can protect the ASIC from the incident X-rays, and also means that by placing an absorber in the lateral direction, it can also protect from scattered X-rays from the side.

[0066] FIG. 8(A) is a schematic diagram showing a detector module implemented as an MCM similar to the embodiment in U.S. Patent No. 8,183,535 (Patent Document 1). In this example, it shows how the semiconductor sensor 21 can also serve as a substrate in the MCM. The signal is sent from the detector element 22 through the wiring path 23 to the input of the parallel processing circuit 24 (e.g., ASIC) arranged adjacent to the active sensor area. The ASIC processes the charge generated from each X-ray, converts it into digital data, and can use this data to detect photons and / or estimate the energy of photons. The ASIC can have its own digital processing circuits and memories for simple tasks. Also, the ASIC may be configured to connect to digital processing circuits and / or memory circuits or components located outside the MCM, and ultimately this data is used as an input for reconstructing an image.

[0067] However, the adoption of depth partitioning also poses two significant problems for silicon-based photon counting detectors. First, a large number of ASIC channels must be used to process the data supplied from the associated detector partitions. In addition to the increase in the number of channels due to both the reduction in pixel size and the depth segmentation, the data volume is further increased by multiple energy bins. Second, since a given number of X-ray inputs are divided into smaller pixels, partitions, and energy bins, the signal in each bin becomes significantly lower, and thus detector calibration / correction requires many times more calibration data to minimize statistical uncertainties.

[0068] Naturally, as the data volume increases by several orders of magnitude, in addition to the need for larger computing resources, hard drives, memories, and central processing units (CPUs) or graphics processing units (GPUs), both data processing and preprocessing are slowed down. For example, if the data volume increases from 10 megabytes to 10 gigabytes, the data processing time and read / write times can become 1000 times longer.

[0069] One problem in all counting X-ray photon detectors is the pile-up problem. When the X-ray photon beam speed is high, problems can occur in discriminating between two consecutive charge pulses. As mentioned above, the pulse length after filtering depends on the shaping time. If this pulse length is greater than the time between two charge pulses induced by X-ray photons, the pulses may merge, making the two photons indistinguishable and potentially counted as one pulse. This is called pile-up. Thus, one way to avoid pile-up at high photon beam speeds is to use a small shaping time or depth division.

[0070] For pile-up calibration vector generation, the pile-up calibration data needs to be preprocessed for spit correction. For material decomposition vector generation, the material decomposition data should preferably be preprocessed for both spit correction and pile-up correction. For patient scan data, the data needs to be preprocessed for spit, pile-up, and material decomposition before image reconstruction. These are simplified examples for explaining preprocessing. This is because the actual preprocessing steps may include several other calibration steps as needed, such as reference normalization and air calibration. The term processing may sometimes refer only to the final step in each calibration vector generation or patient scan, but is also used interchangeably in some cases.

[0071] FIG. 8(B) is a schematic diagram showing an example of a set of tile-type detector sub-modules, where each detector sub-module is a depth-division detector sub-module, and an ASIC or corresponding circuit 24 is arranged below the detector element 22 as viewed from the X-ray incident direction, and a wiring path 23 from the detector element 22 to the parallel processing circuit 24 (e.g., ASIC) is provided in the space between the detector elements.

[0072] FIG. 9 is a schematic diagram showing an overall example of a CT imaging system. In this schematic example, the overall CT imaging system 100 includes a gantry 111 and a patient table 112 that can be inserted into the opening 114 of the gantry 111 during patient scanning and / or calibration scanning. The direction of the rotation axis of the rotating member of the gantry centered on the subject or patient being imaged is shown as the z direction. The angular direction of the CT imaging system is shown as the x direction, and the X-ray incident direction is called the y direction.

[0073] However, it should be understood that the rotating and stationary members of the gantry may not be part of the CT imaging system and may be arranged and / or configured in other ways, such as relative linear and / or translational motion without rotation. As an example, the combination of the X-ray source and detector may move in a linear and / or translational manner relative to the stationary member of the overall gantry. For example, the X-ray source and detector may move together as an integrated assembly unit along the table axis, which is generally also called the z-axis. Alternatively, relative motion is important, such as the patient table moving while the combination of the X-ray source and detector remains stationary. This configuration includes, for example, a geometric system configuration in which a patient can stand, such as in a so-called phone booth type scanner.

[0074] FIG. 10 is a schematic diagram showing an example of the overall design of an X-ray source and detector system. In this example, a schematic diagram of an X-ray detector including a plurality of detector modules and an X-ray source that emits X-rays is shown. Each detector module may have a set of detector elements that define corresponding pixels. For example, the detector module may be an edge-on detector module arranged in parallel and oriented edge-on toward the X-ray source, or may be arranged in a slightly curved overall configuration. As mentioned above, the incident direction of the X-rays is referred to as the y direction. When there are a plurality of detector pixels in the direction of the rotation axis of the gantry (referred to as the z direction), multi-slice image acquisition can be performed. When there are a plurality of detector pixels in the angular direction (referred to as the x direction), a large number of projections can be measured simultaneously in the same plane, and this configuration is applied in fan-beam / cone-beam CT. The x direction is also referred to as the channel direction. Most detectors have detector pixels in both the slice (z) direction and the angular (x) direction.

[0075] FIG. 11 is a schematic flow diagram showing an example of a method for projection-type spectral X-ray imaging. According to one embodiment, a method S0 for projection-type spectral X-ray imaging includes a first step S1 of performing projection-type material decomposition based on spectral X-ray data to generate a set of material basis sinograms. The next step S2 includes performing weighted combination of at least a part of at least two of the above-mentioned set of material basis sinograms based on material weighting in the projection region to form a reconstructed image.

[0076] The projection-type material decomposition is performed for each projection measurement from the spectral X-ray data. In other words, the material decomposition is performed in the projection region. According to an example, one or more pile-up correction techniques may be applied to the spectral X-ray data prior to performing the projection-type material decomposition. This example can improve the linearization of the X-ray detector response.

[0077] A set of material basis sinograms is obtained from projection-based material decomposition. The generated set of material basis sinograms can include two material basis sinograms or a larger number of material basis sinograms. According to one example, projection-based material decomposition generates two estimates of material basis sinograms given by the following equation. μ(E)=w 1 (E)A 1 jk +w 2 (E)A 2 jk where μ(E) is the attenuation due to the single energy of the projection measurement, and A 1 and A 2 are the material basis sinograms at bin i and pixel j for a given material, respectively, also called the material basis estimates, and w 1 and w 2 are the corresponding energy-dependent linear attenuations, also called the material basis estimate coefficients or simply the material weights. Thus, the relative weights of the material basis sinograms A 1 and A 2 are determined by the selection of the single energy E. The single energy E is also called the monochromatic energy. According to one example, the basis material image is a basis sinogram.

[0078] The weighted combination of at least a portion of the material basis sinogram is based on an adaptive material weighting. In other words, the material basis sinogram (at least a portion thereof) is weighted individually. In this way, the weights for the portions and / or the whole of the material basis sinogram are determined. Thus, for each material basis sinogram, a single energy can be selected, for example, to obtain an optimized reconstructed image that maximizes CNR and / or minimizes noise. Since noise and / or CNR vary based on the path length of the material for each projection, the single energy selected in the image region such that noise can be minimized and / or CNR can be maximized is not unique. The present invention has an advantage in that the weighted combination of the material basis sinogram based on (adaptive) material weighting is performed in the projection region. Thereby, the reconstruction of the projection into the reconstructed image can be made more versatile, flexible, and / or accurate. In this way, the proposed technique takes into account the improvement of image quality, for example, noise reduction, improvement of CNR, and improvement of signal-to-noise ratio (SNR). Furthermore, the present invention enables improvement of patient dose efficiency. According to one example, the patient dose efficiency can be further increased when the present invention is combined with mA modulation, also called tube current modulation.

[0079] FIG. 12 is a schematic flow chart showing a method for projection-type spectral X-ray imaging. The schematic flow chart of FIG. 12 has some features common to the schematic flow chart of FIG. 11, and thus FIGS. 11 and the related text are incorporated to make it easier to understand at least some of the features and / or operations in the flow chart. It should be noted that, for simplicity, the schematic flow chart of FIG. 12 may not disclose every possible combination and permutation, etc. of each step of the method S0 of the representative embodiment. Therefore, FIG. 12 should be interpreted as non-limiting within the scope of each step of the method S0 of the representative embodiment and in the possible variations and / or combinations between each step.

[0080] Illustrated as an example, step S2 of performing weighted combination of at least portions of at least two basis sinograms includes step S3 of determining at least one of a view-dependent material weight for each projection in a set of a large number of projections and a pixel-dependent material weight for each pixel in a set of a large number of pixels individually. In other words, for each projection, a corresponding weight is determined. This corresponding weight may be view-dependent and / or pixel-dependent. Thus, in addition to only one of the view-dependent material weight and the pixel-dependent material weight, combination of the view-dependent material weight and the pixel-dependent material weight for each projection can be performed. The view-dependent material weight and / or the pixel-dependent material weight are also called material weights as collective terms, and can be determined to minimize noise of the single energy attenuation μ(E) of the material basis sinogram generated by the projection. This determination can be achieved by determining the single energy that minimizes the noise in the single energy attenuation μ(E) of the projection due to the fact that the material weight is energy-dependent. Thus, the material weight can be determined using the noise characteristics for a specific projection.

[0081] Step S2 of performing weighted combination of at least portions of at least two basis sinograms may further include step S4 of combining at least portions of at least two material basis sinograms in a projection region based on at least one of the view-dependent material weight and the pixel-dependent material weight. Thus, the determined view-dependent material weight and / or pixel-dependent material weight can be applied to at least two material basis sinograms in the projection region to form a combined image. When the determined weights are applied to the material basis sinograms respectively, these material basis sinograms can be called weighted basis material sinograms or weighted sinograms. The weighted sinogram can subsequently be reconstructed as an image, i.e., a reconstructed image.

[0082] In a specific example, step S3 of determining at least one of the view-dependent substance weight and the pixel-dependent substance weight can be performed based on at least one of the variance and the covariance between at least parts of at least two substance basis sinograms.

[0083] The covariance between at least parts of at least two substance basis sinograms can be measured, and / or modeled, estimated, or calculated, etc. in the calibration step. According to one example, the covariance between at least parts of at least two substance basis sinograms is ij (A N jk ) estimated by the forward model given as, where i is the bin, and A N jk is the substance basis estimate for pixel j, view k, and substance N in sinogram A. The forward model can be determined by calculating the Cramer-Rao Lower Bound (CRLB). The required Fisher information for the case of the number of uncorrelated bins can be obtained by calculating the Fisher matrix given by the following formula.

Equation

[0084] Then, the CRLB can be determined by finding the reciprocal of the Fisher information F j . In addition, this forward model can be specific to each detector element of the X-ray imaging system. Furthermore, for an additional noise model, the same method for estimating the variance between at least parts of at least two substance basis sinograms can be applied, provided that the definition of the Fisher matrix is different. According to one example, the substance basis sinogram (at least part thereof) can be filtered in at least one of the view direction and the pixel direction before determining the covariance between the substance basis sinograms (at least part thereof).

[0085] The variance for a given single energy is w(E) TF -1 is given as w(E). According to one example, for a given F j , the variable w min can be determined by representing the minimizer of the variance for a given single energy. The variable w min represents the optimal single energy material weight for the material basis sinogram A jk . According to another example, w min can alternatively be determined by maximizing the CNR given by the following formula. [w(E) T (A 1 -A 2 )] / [w(E) T F -1 w(E)]

[0086] According to one example, the material basis sinogram (at least a part thereof) can be filtered in at least one of the view direction and the pixel direction before determining the variances of the material basis sinogram (at least a part thereof), respectively.

[0087] In a specific example, step S3 of determining at least one of the view-dependent material weight and the pixel-dependent material weight may include step S5 of maximizing or minimizing a first objective function with respect to at least one of noise, artifacts, depiction of anatomical features, pile-up, ratio, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) of the reconstructed image. Thus, by step S5 of maximizing or minimizing, i.e., optimizing, the first objective function, all or only some of the above-described image characteristics can be taken into account. The first objective function can also be optimized in such a way that a trade-off between the above-described image characteristics is achieved.

[0088] In a non-limiting example, step S3 of determining at least one of the view-dependent material weight and the pixel-dependent material weight may include step S6 of selecting a region of interest (ROI) and determining at least one of the view-dependent material weight and the pixel-dependent material weight in this ROI.

[0089] Note that the ROI selected when determining at least one of the view-dependent substance weight and the pixel-dependent substance weight may be the same ROI that is common and shared for all at least parts of at least two substance basis sinograms. The selected ROI may be a plurality of ROIs in each corresponding substance basis sinogram, where the plurality of ROIs overlap and thus form the (common) selected ROI. The view-dependent substance weight and / or the pixel-dependent substance weight can be calculated based on measurements in the selected ROI, but can then be applied to a second region of the substance basis sinogram. In other words, the weighted combination of at least parts of at least two substance basis sinograms is performed in the second region. The second region may include the selected ROI or may include only a part of the selected ROI. In some examples, the selected ROI is excluded from the second region. The second region is preferably larger than the selected ROI. According to one example, step S3 of determining the view-dependent substance weight and / or the pixel-dependent substance weight can be performed by forward projection based on measurements in the selected ROI. According to one example, the substance to be decomposed can be selected based on, for example, the selected scanning protocol. A 1 and A 2 can be selected from here based on these substances. This can be achieved, for example, by applying a forward projection model to divide the substance and obtain a divided substance basis sinogram.

[0090] According to one example, the classifier algorithm can identify at least two substances in the selected ROI. The classifier algorithm can be part of a machine learning system. The view-dependent substance weight and / or the pixel-dependent substance weight can be determined to maximize the detection performance of at least two substances in the selected ROI. According to another example, it is also possible to select a plurality of ROIs for at least a part of at least two substance basis sinograms, and these plurality of ROIs do not necessarily completely overlap. In this way, some of the selected ROIs may be the same or may be different to at least some extent. Further, according to this example, the classifier algorithm may identify a plurality of at least two substances in a plurality of selected ROIs. These substances in different ROIs may be the same or different. The view-dependent substance weight and / or the pixel-dependent substance weight can be determined to maximize the combination of the detection performance of at least two substances in the selected ROI.

[0091] In another non-limiting example, at least one of the view-dependent substance weight and the pixel-dependent substance weight may be at least one of being restricted within a predetermined reference interval based on a predetermined substance weight or being parameterized as a function of a substance or signal level. The view-dependent substance weight and / or the pixel-dependent substance weight can be parameterized in such a way that, for example, a trade-off between CNR and artifact suppression is achieved. Further, the ref predetermined reference interval referred to as w may be parameterized as a function of a substance or signal level. For example, this substance function may depend on A N jk thereof.

[0092] In a specific example, the step of determining at least one of the view-dependent substance weight and the pixel-dependent substance weight may include a step S7 of maximizing or minimizing a second objective function with respect to the substance separation of a plurality of substances. The second objective function may be the same as or different from the first objective function.

[0093] In a specific example, at least one of the view-dependent substance weight and the pixel-dependent substance weight, and the predetermined substance weight may be parameterized as a third objective function of at least one of i) an estimated gradient in at least one of the view direction and the pixel direction of the projection, and ii) an estimated gradient of the reconstructed image, in order to compensate for the bias. The third objective function may be the same as or different from the first objective function and / or the second objective function. For example, the third objective function may be an additional part and / or a perspective of the substance basis sinogram with respect to the first and / or second objective functions.

[0094] In a non-limiting example, the step of determining at least one of the view-dependent substance weight and the pixel-dependent substance weight may be performed by a first machine learning system. The first machine learning system may include one or several machine learning models such as a decision tree, a support vector machine, and a neural network. According to an example, the first machine learning system may include a machine learning architecture and / or a machine learning algorithm that may be at least partially based on a convolutional neural network model. According to another example, the first machine learning system may include a machine learning architecture and / or a machine learning algorithm that may be at least partially based on at least one of a decision tree model and a support vector machine model.

[0095] In another non-limiting example, the first machine learning system may include a trained first convolutional neural network (CNN). The step of determining at least one of the view-dependent substance weight and the pixel-dependent substance weight is performed by the first CNN. The first CNN is trained for at least one of mapping at least a part of at least two substance basis sinograms to at least one of the view-dependent substance weight and the pixel-dependent substance weight, and estimating at least one of variance and covariance based on the mapping between at least a part of at least two substance basis sinograms and a fourth objective function.

[0096] Thus, the first CNN is trained to map the material basis sinogram (at least a portion thereof) to view-dependent material weights and / or pixel-dependent material weights and / or to estimate variance and / or covariance based on the mapping between the material basis sinogram (at least a portion thereof) and a fourth objective function. The fourth objective function may be the same as or different from the first, second, and third objective functions. The fourth objective function may be the CRLB. According to one example for obtaining the trained first CNN, view-dependent material weights and / or pixel-dependent material weights for a common imaging task can be determined for several phantom images. Then, the determined view-dependent material weights and / or pixel-dependent material weights can be used to train the first CNN. The output from the first CNN can be a map of view-dependent material weights and / or pixel-dependent material weights. According to another example, the CNN can be trained to output a low-noise variance map. The low-noise variance map can be used to determine at least one of view-dependent material weights and / or pixel-dependent material weights. Utilizing the first CNN is advantageous because it can obtain further optimized material weights and thus relatively low-noise and high-resolution reconstructed images. Further, providing the first CNN to determine the material weights can alleviate or completely eliminate the need for human interaction and decision-making, etc. Further, as a result of using the first CNN to determine the material weights, the computation time can be shortened.

[0097] In a specific example, the step S2 of performing weighted combination of at least parts of at least two material basis sinograms can be performed by a second machine learning system by mapping at least parts of at least two material basis sinograms into a reconstructed image. The second machine learning system may be the same as or different from the first machine learning system. The second machine learning system may include one or several machine learning models such as decision trees, support vector machines, and neural networks. According to an example, the second machine learning system may include a machine learning architecture and / or a machine learning algorithm that may be at least partially based on a convolutional neural network model. According to another example, the second machine learning system may include a machine learning architecture and / or a machine learning algorithm that may be at least partially based on at least one of a decision tree model and a support vector machine model. According to an example, (the same) one machine learning system is configured to, in addition to step S2 of performing weighted combination of (at least parts of) material basis sinograms into a reconstructed image, determine view-dependent material weights and / or pixel-dependent material weights as in S3. It should be noted that it is equally feasible to only perform step S2 of performing weighted combination of (at least parts of) material basis sinograms by the machine learning system, that is, not to determine the material weights in S3.

[0098] In a specific example, the second machine learning system can include a trained second convolutional neural network (CNN). In this case, the step of performing weighted combination of at least two material basis sinograms is performed by training the second CNN with at least sinograms obtained from at least one of a simulation image, a phantom image, and a patient image. The second CNN may be the same as or different from the first CNN.

[0099] In another specific example, the final reconstructed image can be generated by a fourth machine learning system that includes a third trained CNN. The fourth machine learning system may be the same as or different from the first and / or second machine learning systems. The third CNN may be the same as or different from the first and / or second CNNs. The CNN can be trained on at least sinograms obtained from simulation images, phantom images, and / or patient images. This CNN can be optimized such that the final reconstructed image achieves an optimal CNR for different path lengths. This CNN can further be optimized such that the final reconstructed image achieves a minimum bias and / or edge preservation. The input to the CNN can be a combination of at least two of the parameters such as various material basis sinograms A, A S , A f and material weights w min , w ref . Alternatively, the CNN may instead be configured to generate a bias map. The generated bias map can then be combined with the already determined w min (A S jk )A jk to form the final reconstructed image.

[0100] As an example, a set of training material basis sinograms and basis material weights can be generated, and the optimal combination of these sinograms can be manually selected or selected through numerical optimization to select a weighting method that gives the optimal image quality of the reconstructed image measured through objective or subjective image quality metrics. For example, the optimal weights for several different imaging tasks can be calculated, and by forming the angle-dependent weighted average and pixel-dependent weighted average of different weight maps, the combination of these weights can be at least partially manually selected. After applying this method to obtain the optimally weighted sinogram for the training set of sinograms, the CNN can be trained to map at least two of the parameters of various material basis sinograms and material weights to the optimal combination of basis sinograms. In this way, the same image quality as the training set can be obtained using the CNN without going through a manual or optimization-based search for the optimal way to combine the sinograms.

[0101] In a non-limiting example, step S2 of performing a weighted combination of at least a portion of at least two basis sinograms may include iterative forward projection. Iterative forward projection includes step S8 of calculating the estimated forward projection of at least a portion of at least two material basis sinograms, where the weighted combination may be at least partially based on this estimated forward projection.

[0102] In another non-limiting example, method S0 may include performing at least one of filtering S9 and downsampling S10 on at least a portion of at least one of spectral X-ray data and at least a portion of at least two material basis sinograms to compensate for at least one of low signal, photon deficiency, and noise. Thus, filtering S9 and / or downsampling S10 may be performed before and / or after performing projection-based material decomposition S1. For example, material basis sinogram A jkBy means of low-pass filtering and decimation in at least a partial view direction and / or pixel direction, a low-pass filtered substance basis sinogram portion A S jk can be obtained. The low-pass filtered substance basis sinogram A S jk can vary to a relatively low degree so that the minimum noise estimate itself does not introduce additional noise. According to this example, the risk of suppressing high-frequency components is reduced. For each bin i, the minimum noise is selected for each pixel j and view k after each decimation. Then, an adaptive weighted substance basis sinogram A f jk for pixel j and view k can be obtained as w min (A S jk )A jk . According to one example, while decimation is omitted, the minimum noise is selected for each pixel and view based on A S jk .

[0103] According to one example, in order to compensate for a substance basis sinogram containing relatively low signals and / or photon deficiencies, substance basis sinograms A S jk after different levels of filtering are used, for example, the variance of A S jk can be reduced.

[0104] In another non-limiting example, the method further includes biasing at least one of the first weighted material basis sinogram and the reconstructed image to form a bias-adjusted reconstructed image. The bias adjustment is performed by performing a Fourier domain division of the first weighted sinogram and the weighted filtered sinogram in the projection region, and combining at least a portion of the first weighted sinogram and at least a portion of the Fourier domain divided weighted filtered sinogram to generate a bias-adjusted material sinogram, where the bias-adjusted sinogram is reconstructed as a bias-adjusted reconstructed image; and performing a Fourier domain division of the first reconstructed image reconstructed from the weighted sinogram and the second reconstructed image reconstructed from the weighted filtered sinogram in the image region, and combining at least a portion of the Fourier domain divided first reconstructed image and at least a portion of the Fourier domain divided second reconstructed image to form a bias-adjusted reconstructed image. It should be understood that the weighted sinogram, i.e., the weighted basis material sinogram, can be defined as a combination in the projection region of at least two material basis sinograms based on the determined material weights. The weighted filtered sinogram can be a material basis sinogram that is filtered or downsampled, for example, to compensate for low signal, photon deficiency, and / or noise, to which the material weights are applied. The material weights may be determined in different manners, as disclosed throughout this application. The reference sinogram may be different weighted sinograms and / or filtered weighted sinograms. The reconstructed image can be reconstructed from the weighted sinogram, the filtered weighted sinogram, and / or the bias-adjusted material basis sinogram. The bias-adjusted material basis sinogram may be referred to as a weighted bias-adjusted material basis sinogram. Also, it should be understood that the Fourier domain divided sinogram or the Fourier domain divided reconstructed image is a sinogram or an image on which the Fourier domain division has been performed.

[0105] Thus, different portions of the material basis sinogram can be processed in different ways, such as filtering and / or bias adjustment. For example, by performing Fourier domain partitioning on two weighted filtered or unfiltered material basis sinograms to generate a Fourier domain partitioned material basis sinogram, the weighted sinogram can be combined with the weighted filtered sinogram, and then these sinograms can be combined to generate a weighted bias adjusted material basis sinogram. It should be noted that at least a portion of the material basis sinogram may be omitted in the above-described filtering and / or bias adjustment, and in this way, initial, unprocessed, and native portions of the material basis sinogram may be obtained as a result. Initial portion A of the material basis sinogram, portion A of the material basis sinogram after filtering S , and portion A of the material basis sinogram after bias adjustment fc At least two of various portions of the material basis sinogram, such as these, may be mixed in the Fourier domain. Thus, when performing Fourier domain partitioning, any combination of the presence of the initial portion A of the material basis sinogram, the portion A after filtering S , and the portion A after bias adjustment fc is executable.

[0106] The bias adjusted material basis sinogram A fc is obtained by performing the following w min (A S jk )A jk -(w min (A S jk )-w ref )A S jk .

[0107] Thereby, the low spatial frequency from the sinogram weighted by w ref , and w minIt enables coupling with high spatial frequencies from the weighted sinogram, generating a bias-corrected weighted sinogram, which can then be reconstructed as an image. Mixing of different frequencies can be achieved by dividing the frequencies of various parts of the (weighted) material basis sinogram, thus generating a Fourier-domain segmented (weighted) material basis sinogram. Then, the Fourier-domain segmented material basis sinograms are combined. For example, relatively high frequencies can be dominated by the filtered part A of the material basis sinogram f This can reduce the presence of noise. Furthermore, relatively low frequencies can be dominated by the initial part A of the material basis sinogram. This can improve bias correction. According to one example, combining different levels of the filtered material basis sinogram A S jk can improve bias correction. This bias adjustment can be given by the following equation. b = Σ m w m (w min - w refm )A S m where m represents filtering at different levels.

[0108] According to one example, Fourier-domain segmentation can also be performed in the image domain instead of the projection domain. This can be achieved by first reconstructing the weighted material basis sinogram w min (A S jk )A jk as a weighted material basis image and reconstructing the reference material basis sinogram A ref jk as a reference weighted material basis image. A ref jk is, for example, the basis sinogram weighted by the reference weight w ref A jk , the original determined material basis sinogram A jk , or the filtered material basis sinogram A Scan be referred to. Thereafter, the reconstructed weighted substance basis image and the reference weighted basis substance image are mixed in the Fourier domain. For example, these images can be low-pass filtered, for example, by Fourier transform, weighting, and inverse transform, and thereafter a bias-corrected basis substance image is generated as follows. R[w min (A S jk )A jk -R[w min (A S jk )A jk S +R[A ref jk S In the formula, R represents a reconstruction operator, and S represents low-pass filtering.

[0109] In a specific example, step S2 of performing a weighted combination of at least two substance basis sinograms can include step S15 of determining a matrix of substance weights including at least a first substance weight and a second substance weight for each of at least one of the projection and the pixel, where the second substance weight is orthogonal to the first substance weight. Accordingly, the resulting substance basis sinogram consists of substance basis sinograms represented by vector values.

[0110] This matrix is called W and includes substance weights including at least a first substance weight w min (A S jk ) and a second substance weight w ⊥ min (A S jk ). Accordingly, the second substance weight is a unit vector orthogonal to the first substance weight. The substance basis sinogram represented by vector values is obtained by replacing w ref with W in the above formula. The next step is to multiply the above formula by a constant orthogonal matrix W ref from the left. The resulting bias-adjusted substance basis sinogram A fc ​​Finally, it can be obtained by the following formula. A fc jk =A S jk +W T ref W(A S jk )(A jk -A S jk )

[0111] W(A S jk ) and W ref Since they are orthogonal matrices, the product is a rotation matrix. Thus, this given example including the matrix W containing at least the first and second material basis sinograms can be interpreted as a transformation that rotates A S jk about A jk .

[0112] Figures 13(A) and 13(B) are graphs representing the relationship between dispersion and single energy. In Figures 13(A) and 13(B), CNR is shown on the vertical axis and single energy MonoE is shown in keV on the horizontal axis. In Figure 13(A), the dotted line L 1 represents the first relationship between dispersion and single energy for a longer path length. The dashed line L 2 represents the second relationship between dispersion and single energy for a shorter path length. CNR / noise varies based on the material thickness for each projection, meaning that the optimal single energy can vary with each view and / or pixel. E 1 opt is the optimized single energy that minimizes noise and maximizes CNR for L 1 , and E 2 opt is the optimized single energy that minimizes noise and maximizes CNR for L 2 . E 1 opt and E 2 optIn contrast, as such, the single energy that can be selected in the image region to maximize the CNR and / or minimize the noise is not single.

[0113] In FIG. 13(B), the dotted line L 3 represents the relationship between the variance and the single energy. L 3 represents the CNR of one of the inserted graphs as a function of MonoE. The dashed line L 4 represents the CNR of the reconstructed image after weighted combination in the projection region of at least two material basis sinograms according to an embodiment of the present invention.

[0114] FIG. 14 is a schematic diagram showing some relevant parts of an X-ray imaging system such as a CT imaging system. The CT imaging system 400 includes an X-ray source 410 configured to emit X-rays, an X-ray detector 420 configured to generate spectral X-ray data, and a processor 430. The processor 430 is configured to perform projection-type material decomposition on the spectral X-ray data to generate a set of material basis sinograms, and perform weighted combination of at least a part of at least two material basis sinograms among the set of material basis sinograms based on material weighting in the projection region to obtain a reconstructed image.

[0115] The processor 430 may include digital and / or analog processing circuits. The processor 430 may form part or all of the imaging processing system. To further understand the processor, reference is made to FIGS. 2 to 8 and the related text.

[0116] In a specific example, the processor 430 may be configured to determine at least one of an individual view-dependent material weight for each projection among a set of a large number of projections and an individual pixel-dependent material weight for each pixel among a set of a large number of pixels. The processor 430 is further configured to combine at least a portion of at least two material basis sinograms in a projection region based on at least one of the view-dependent material weight and the pixel-dependent material weight.

[0117] In another specific example, the processor 430 may be configured to determine at least one of the view-dependent material weight and the pixel-dependent material weight based on at least one of the variance and covariance between at least a portion of at least two material basis sinograms.

[0118] In a non-limiting example, the processor 430 may be configured to determine at least one of the view-dependent material weight and the pixel-dependent material weight based on maximizing or minimizing a fifth objective function with respect to at least one of noise, artifacts, depiction of anatomical features, pile-up, ratio, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) of a reconstructed image.

[0119] In another non-limiting example, the processor 430 may include a third machine learning system 440 configured to perform weighted combination of at least a portion of at least two material basis sinograms by mapping at least a portion of at least two material basis sinograms to a reconstructed image.

[0120] It should be understood that a CT imaging system can include any 3D X-ray medical imaging modality. For example, a CT imaging system can include a general tomography system configured to combine information from images taken at different angles to create an image volume such that any slice can be redisplayed individually. A CT imaging system can include a CT system configured to use a full 180-degree range of angles and / or a limited range of angles. In other words, a CT system can include a conventional computed tomography system that uses a full 180-degree range or a breast tomosynthesis system, also known as a mammogram, that uses a limited range of angles.

[0121] In this example, the CT system includes an X-ray source 110 and an X-ray detector 120 configured in such a way that projection images of a subject or object can be acquired at different viewing angles in the X-ray beam path. This is most commonly achieved by mounting the X-ray source 110 and the X-ray detector 120 on a rotating member of a gantry that can rotate about a support body, for example, a subject or object.

[0122] As described above, at least some of the steps, functions, procedures, and / or blocks described in this document can be implemented in software, such as a computer program, for execution by a suitable image processing circuit, such as one or more processors or processing units.

[0123] FIG. 15 is a schematic diagram showing an example of a computer implementation according to an embodiment. In this specific example, system 200 includes a processor 210 and a memory 220, the memory including instructions executable by the processor, and these instructions cause the processor to operate to perform the steps and / or operations described in this document. The instructions are typically configured as computer programs 225, 235 that can be pre-built in the memory 220 or downloaded from an external memory device 230. Optionally, system 200 includes an input / output interface 240 interconnected to the processor(s) 210 and / or the memory 220 to enable input / output of related data such as input parameter(s) and / or output parameter(s) obtained as a result.

[0124] In a specific example, the memory 220 includes a set of instructions executable by the processor, and these instructions cause the processor to operate to perform the steps and / or operations described in this document.

[0125] The term processor is to be interpreted in the general sense of any system or device capable of executing program code or computer program instructions to perform a specific processing task, decision task, or computational task.

[0126] Thus, an image processing circuit including one or more processors is configured to perform well-defined processing such as that described in this document when executing a computer program.

[0127] The image processing circuit does not necessarily have to be dedicated only to executing the above-described steps, functions, procedures, and / or blocks, and may be capable of executing other tasks.

[0128] The proposed technology also provides a computer program product including a computer-readable medium 220, 230 storing such a computer program.

[0129] By way of example, software or computer programs 225, 235 can be implemented as computer program products that are typically carried or stored on computer-readable media 220, 230, particularly non-volatile media. The computer-readable media can include one or more removable or non-removable memory devices, including, but not limited to, read-only memory (ROM), random access memory (RAM), compact disc (CD), digital versatile disc (DVD), Blu-ray disc, universal serial bus (USB) memory, hard disk drive (HDD) storage devices, flash memory, magnetic tape, or any other conventional memory device. Thus, the computer program can be loaded into the operating memory of a computer or equivalent processing device for execution by the image processing circuit of the computer or equivalent processing device.

[0130] A method flow can be regarded as a computer operation flow when executed by one or more processors. The corresponding device, system, and / or apparatus can be defined as a group of functional modules, where each step executed by the processor corresponds to (one) functional module. In this case, the functional module is implemented as a computer program running on the processor. Therefore, this device, system, and / or apparatus can alternatively be defined as a group of functional modules, in which case the functional module is implemented as a computer program running on at least one processor.

[0131] The computer program present in the memory can thus be configured as a suitable functional module that, when executed by the processor, performs at least part of the steps and / or tasks described in this document.

[0132] Alternatively, the module can be realized mainly by hardware modules or, alternatively, by hardware. The software-versus-hardware scope is purely a matter of choice in implementation.

[0133] The embodiments of the present disclosure described above and shown in the drawings are merely examples of embodiments and do not limit the scope of the claims, including any equivalent configurations included within the scope of the claims. It will be understood by those skilled in the art that various modifications, combinations, and variations can be made to each embodiment without departing from the scope of the invention defined by the claims. Any combination of features described herein that are not mutually exclusive is considered to be within the scope of the present invention. That is, the features of the described embodiments can be combined with any of the above-mentioned appropriate aspects, and the optional features of any one aspect can also be combined with any other appropriate aspect. Similarly, the features described in the dependent claims can be combined with the non-mutually exclusive features of other dependent claims, particularly when each dependent claim depends on the same independent claim. Single-item dependence may be used in some jurisdictions for implementation, but it should not be construed as meaning that each feature of the dependent claims is mutually exclusive.

[0134] Furthermore, unless otherwise specified, it should be noted that the concept of the present invention relates to all possible combinations of features. Specifically, when technically possible, the solutions of different parts in different embodiments can be combined in other configurations. [Definition of Terms]

[0135] In the following description of non-limiting examples, the following notations and definitions are mainly used in this document. Index i = bin, j = pixel, k = view. A N Native-level MD sinogram for substance N, where N = 1, 2, …. A N Is usually simply called A for a given substance. Sinogram A can be appropriately selected and optionally preprocessed for a specific region of interest. MD material decomposition. A N jk Substance basis estimate for pixel j and view k of sinogram A. The substance basis estimate is often interchangeably called the substance thickness estimate. A N jk is simply A for a given substance jk and may also be called. A S MD sinogram after low-pass filtering. A S jk Substance basis estimate after low-pass filtering for pixel j and view k. A ref jk Reference substance basis estimate for pixel j and view k, representing a pre-determined substance basis estimate. A f Adaptive weighted MD sinogram. A f jk Adaptive weighted substance basis estimate for pixel j and view k. A fc Adaptive weighted bias-corrected MD sinogram. A fc jk Adaptive weighted bias-corrected basis estimate for pixel j and view k. f ij (A jk ) Forward model for pixel j, bin i. F ij Fisher matrix for pixel j, bin i. w(E) Substance weight for single energy E. w opt (A) Function that returns the optimal weight for a given projection A and task function. w ref Reference weight representing a given reference single energy. w min (A jk ) (Ajk Optimal single-energy material weight for CRLB Cramér-Rao lower bound.

Explanation of symbols

[0136] 21 Detector module 22 Detector element (pixel) 23 Wiring path 24 Parallel processing circuit 26 Wiring path 30 Image processing system 62 Display 100 X-ray imaging system 111 Gantry 112 Patient table 114 Aperture S0 Method for projection-type spectral X-ray imaging 230 External memory device 400 CT imaging system 420 X-ray detector

Claims

1. 1. A method for projection spectral X-ray imaging, comprising: performing a projective material decomposition on the spectral X-ray data to generate a set of material basis sinograms; weightedly combining at least a portion of at least two of the set of material basis sinograms into a reconstructed image based on material weighting in the projection domain; The method according to claim 1,

2. The step of performing a weighted combination of at least portions of the at least two material basis sinograms comprises: view-dependent material weights separately for each projection in the set of multiple projections; and Pixel-dependent material weights for each pixel in a set of multiple pixels determining at least one of combining at least portions of the at least two material basis sinograms in the projection region based on the at least one of the view-dependent material weights and the pixel-dependent material weights; The method of claim 1 , comprising:

3. The method of claim 2 , wherein determining at least one of the view-dependent material weights and the pixel-dependent material weights is performed based on at least one of a variance and a covariance between at least portions of the at least two material basis sinograms.

4. 3. The method of claim 2, wherein determining at least one of the view-dependent material weights and the pixel-dependent material weights comprises maximizing or minimizing a first objective function with respect to at least one of noise, artifacts, anatomical feature depiction, pile-up, ratio, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) of the reconstructed image.

5. 3. The method of claim 2, wherein the step of determining at least one of the view-dependent material weights and the pixel-dependent material weights comprises the step of selecting a region of interest (ROI) and determining at least one of the view-dependent material weights and the pixel-dependent material weights in the ROI.

6. 2. The method of claim 1 , wherein the step of performing a weighted combination of at least portions of the at least two material basis sinograms comprises iterative forward projection, including calculating an estimated forward projection of at least portions of the at least two material basis sinograms, the weighted combination being based at least in part on the estimated forward projection.

7. At least one of the view-dependent material weights and the pixel-dependent material weights may be is constrained within a predetermined reference interval based on a predetermined substance weight; and Parameterized as a function of material or signal level The method according to claim 2, wherein the at least one of

8. The method of claim 2 , wherein determining at least one of the view-dependent material weights and the pixel-dependent material weights comprises maximizing or minimizing a second objective function with respect to material separation of a plurality of materials.

9. To compensate for at least one of low signal, photon starvation, and noise, the X-ray spectral data, and At least a portion of the at least two material based sinograms 2. The method of claim 1, further comprising the step of filtering and / or downsampling at least a portion of at least one of.

10. The method further includes bias adjusting at least one of the first weighted material basis sinogram and the reconstructed image to form a bias-adjusted reconstructed image, the bias-adjusted reconstructed image comprising: performing Fourier domain decomposition of the first weighted sinogram and the weighted filtered sinogram in the projection domain and combining at least a portion of the Fourier domain decomposed first weighted sinogram and at least a portion of the Fourier domain decomposed weighted filtered sinogram to generate a bias-adjusted material sinogram, the bias-adjusted sinogram being reconstructed as the bias-adjusted reconstructed image; performing Fourier domain decomposition in the image domain of a first reconstructed image reconstructed from a weighted sinogram and a second reconstructed image reconstructed from a weighted filtered sinogram, and combining at least a portion of the Fourier domain decomposed first reconstructed image and at least a portion of the Fourier domain decomposed second reconstructed image to form the bias adjusted reconstructed image; The method of claim 1 , wherein the method is carried out by at least one of the following:

11. 3. The method of claim 2, wherein the view-dependent material weights and / or the pixel-dependent material weights and the predetermined material weights are parameterized as an objective function of at least one of i) estimated gradients in the view direction and / or the pixel direction of the projections, and ii) estimated gradients of the reconstructed image, to compensate for bias.

12. 2. The method of claim 1, wherein the step of performing the weighted combination of at least portions of the at least two material basis sinograms is performed by mapping at least portions of the at least two material basis sinograms to the reconstructed image by a machine learning system.

13. The method of claim 2 , wherein determining at least one of the view-dependent material weights and the pixel-dependent material weights is performed by a machine learning system.

14. The machine learning system includes a trained convolutional neural network (CNN), and the determining of the view-dependent material weights and / or the pixel-dependent material weights is performed by the CNN, the CNN comprising: mapping at least portions of the at least two material basis sinograms to the at least one of the view-dependent material weights and pixel-dependent material weights; and estimating at least one of a variance and a covariance based on a mapping between at least a portion of the at least two material basis sinograms and an objective function. The method of claim 13, wherein the at least one of the following is trained:

15. 13. The method of claim 12, wherein the machine learning system includes a trained convolutional neural network (CNN), and the step of performing the weighted combination of the at least two material basis sinograms is performed by training the CNN on at least sinograms obtained from at least one of a simulation image, a phantom image, and a patient image.

16. 2. The method of claim 1, wherein the step of performing the weighted combination of at least two material basis sinograms comprises determining a matrix of material weights including at least a first material weight and a second material weight separately for each of at least one of the projections and pixels, the second material weight being orthogonal to the first material weight.

17. an x-ray source configured to emit x-rays; an x-ray detector configured to generate spectral x-ray data; 1. A processor comprising: performing a projective material decomposition on the spectral X-ray data to generate a set of material basis sinograms; a processor configured to perform a weighted combination of at least a portion of at least two of the set of material basis sinograms based on material weighting in the projection domain into a reconstructed image; A computed tomography (CT) imaging system comprising:

18. The processor, individual view-dependent material weights for each projection in the set of multiple projections; and Individual pixel-dependent material weights for each pixel in a set of multiple pixels Determine at least one of the following: combining at least portions of the at least two material basis sinograms in the projection region based on the at least one of the view-dependent material weights and the pixel-dependent material weights. It is configured as follows:

18. The CT imaging system of claim 17.

19. 20. The CT imaging system of claim 18, wherein the processor is configured to determine at least one of the view-dependent material weights and the pixel-dependent material weights based on at least one of a variance and a covariance between the at least portions of the at least two material basis sinograms.

20. 20. The CT imaging system of claim 18, wherein the processor is configured to determine at least one of the view-dependent material weights and the pixel-dependent material weights based on maximizing or minimizing an objective function with respect to at least one of noise, artifacts, anatomical feature depiction, pile-up, ratio, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) of the reconstructed image.

21. 20. The CT imaging system of claim 17, wherein the processor comprises a machine learning system configured to perform the weighted combination of at least portions of the at least two material basis sinograms by mapping at least portions of the at least two material basis sinograms into the reconstructed image.

Images