Method for determining a scatter estimate for an X-ray detector, computer program product and imaging device

Abstract

The invention relates to a computer-implemented method for determining a scattered radiation estimate for an X-ray detector (1) with detector elements (11), wherein a scattered radiation grid (2) is arranged in front of the X-ray detector (1), wherein at least two detector elements (11) are arranged in at least one direction between respectively adjacent lamellae (21) of the scattered radiation grid (2), wherein the method comprises the following steps: receiving detector data and dividing the detector data into at least two groups of detector data based on the position of the detector elements (11) from which the detector data originate, relative to the respectively nearest lamella (21) of the scattered radiation grid (2); determining a scattered radiation distribution that estimates scattering based on the detector data, in the form of a scattered radiation term for each of the groups of detector data;Convolving the scattered radiation term of each group with a convolution kernel individually adapted for each group, generating multiple convolved scattered radiation terms; merging the scattered radiation distributions according to the convolved scattered radiation terms to obtain a scattered radiation estimate.

Description

[0001] The invention relates to a computer-implemented method for determining a scattered radiation estimate for an X-ray detector having a plurality of detector elements, a method for correcting scattered radiation effects in image data from an X-ray detector having a plurality of detector elements, a corresponding computer program product or computer-readable storage medium, and a corresponding imaging device.

[0002] The prior art in this case is US 2010 / 0 272 236 A1 and US 2006 / 0 153 328 A1.

[0003] When X-rays interact with an object, such as an object being scanned in a CT scanner, scattering processes occur. In a CT scan, for example, the scattered radiation caused by these scattering processes can increase the signal in the X-ray detector used, which in turn usually leads to image artifacts. The goal is therefore to reduce the influence of scattered radiation on the detector signal. In general, there are two different options for correcting scattered radiation: minimizing the initial generation of scattered radiation, and correcting the scattered radiation signal already generated in the detector. To counteract the measurement of deflected radiation, special hardware can be used through the use of a collimator or an anti-scatter grid.

[0004] In some cases, however, such as photon counting, the ability to avoid scattered radiation is limited. Due to the smaller detector pixels, photon counting detectors require coarse anti-scatter grids, with multiple pixels located between the lamellae of the anti-scatter grid. This can result in high frequencies in the otherwise low-frequency scattered radiation, which in turn can cause sharp ring-like artifacts in the reconstructed images. On the other hand, an even finer structuring of the anti-scatter grid is often undesirable because it can reduce the geometric efficiency. Geometric efficiency is a measure of the ratio of blocked area to transmitted area of ​​photons. In particular, a reduced geometric efficiency means that fewer quanta reach the detector.If the scattered radiation grid becomes more tightly meshed, fewer photons that can be assigned to the actual signal (e.g. photons that pass directly through a patient) reach the detector.

[0005] Empirical, physical or consistency-based models can be used to correct the resulting scattered radiation, see for example the publication Ohnesorge B, Flohr T, Klingenbeck-Regn K. “Efficient object scatter correction algorithm for third and fourth generation CT scanners”, Eur Radiol. 1999;9(3):563-9. doi: 10.1007 / s003300050710. PMID: 10087134. A detailed description and a comparison of the currently available scattered radiation correction options can be found in the references Rührnschopf EP, Klingenbeck K. “A general framework and review of scatter correction methods in x-ray cone-beam computerized tomography. Part 1: Scatter compensation approaches”, Med Phys. 2011 Jul;38(7):4296-311. doi: 10.1118 / 1.3599033. Erratum in: Med Phys. 2011 Oct;38(10):5830. Erratum in: Med Phys. 2011 Oct;38(10): 5830. PMID: 21859031 as well as Hitzenschopf EP, Klingenbeck K. “A general framework and review of scatter correction methods in cone beam CT.Part 2: scatter estimation approaches“, Med Phys. 2011 Sep;38(9):5186-99. doi: 10.1118 / 1.3589140. PMID: 21978063.

[0006] Monte Carlo simulations, an image-based approach, represent the gold standard for calculating scattered radiation signals and enable very precise determination of the scattered radiation. However, such methods are very computationally intensive and therefore generally not feasible in real time. A less time-consuming and computationally intensive alternative is convolution-based scattered radiation models. In these models, the scattered radiation can be represented as a combination of the integrals of the source and the propagation of the scattering, convolved with an estimated or measured scattered radiation kernel. The source term describes the scattering, which is derived from a simplified model (e.g., only first-order scattering along the primary radiation is considered). In contrast, the propagation term represents the local scattering of the photons.By multiplying these two quantities, the scattered beam distribution for a single X-ray beam can be determined, which can be completed by integrating it across all X-rays. While such convolution-based scattered beam models are relatively computationally intensive, they often reach their limits, especially when determining scattered beams using photon-counting detectors and / or static computed tomography systems. Therefore, scattered beam artifacts cannot be adequately or satisfactorily eliminated, at least in some cases.

[0007] It is therefore an object of the present invention to find a possibility of scattered radiation estimation and / or scattered radiation correction which is less computationally intensive, in particular can be carried out in real time, and at the same time is as accurate as possible.

[0008] This object is achieved by a method according to claim 1, a method according to claim 8, a computer program product or computer-readable storage medium according to claim 9, and an imaging device according to claim 10. Further features and advantages emerge from the dependent claims, the description, and the accompanying figures. Regardless of the grammatical gender of a particular term, persons with male, female, or other gender identities are included.

[0009] According to a first aspect of the invention, a computer-implemented method for determining a scattered radiation estimate for an X-ray detector having a plurality of detector elements is provided, wherein a scattered radiation grid is arranged in front of the X-ray detector. In particular, the X-ray detector can be part of a computed tomography system. In at least one direction, at least two detector elements are arranged between adjacent lamellae of the scattered radiation grid, in particular at different positions relative to the nearest lamella of the scattered radiation grid, the method comprising the following steps: - receiving detector data and dividing the detector data into at least two groups of detector data based on the position of the detector elements from which the detector data originate relative to the nearest lamella of the anti-scatter grid; - determining a scatter distribution that estimates scattering based on the detector data, in the form of a scatter term for each of the groups of detector data; - Convolving the scattered radiation term of each of the groups with a convolution kernel individually adapted for each group, so that several convolved scattered radiation terms are generated according to the number of groups; - Merging the scattered radiation distributions according to the convolved scattered radiation terms corresponding to the position of the detector elements to obtain a scattered radiation estimate.

[0010] Advantageously, the method according to the invention also explicitly takes into account the geometry of the scatter grid by dividing the detector data into at least two corresponding groups. This allows the scatter estimate to be used for scatter correction, thereby avoiding artifacts.

[0011] In particular, high-frequency artifacts, such as ring artifacts or partial ring artifacts, which often occur without splitting, can be avoided or at least significantly reduced. It has been shown that there can be considerable differences in the intensity of the scattered radiation between adjacent detector elements, which can lead to high frequencies due to their alternating occurrence (detector elements to the left and right of a leaf). This method advantageously allows the differences in intensity to be taken into account. The method enables correct correction of the scattered radiation at the outset. This eliminates the need to subsequently suppress certain frequencies in the image space in order to reduce the artifacts. It has been shown that even when using a sharp reconstruction kernel, clinical diagnostics are adequate.A sharp reconstruction kernel emphasizes edges more, whereas a soft reconstruction kernel leads to increased smoothing. By avoiding high-frequency artifacts from the outset with this method, edges with the same frequency do not need to be suppressed in the image reconstruction if the high-frequency artifacts do not need to be subsequently attenuated.

[0012] An X-ray detector is generally a detector suitable for detecting X-rays. It can also be referred to as an "X-ray detector" or simply "detector" for short. In particular, the X-ray detector is designed to detect an X-ray intensity and / or dose. The X-ray detector can in particular be an X-ray detector of a computed tomography system, a radiography device, or a C-arm. The X-ray detector comprises a plurality of detector elements. The detector elements can also be referred to as detector pixels. For example, the detector elements can comprise an X-ray-sensitive material designed to detect the X-rays. For example, detector elements can comprise a scintillation material and / or a scintillation counter for detecting the X-rays.The scintillation material can be designed to convert X-rays into light, in particular visible light. The X-ray detector can, in particular, be a photon-counting X-ray detector. A photon-counting X-ray detector is generally designed to detect and count individual X-ray photons, in particular spatially and / or temporally resolved. Typically, a photon-counting X-ray detector comprises an X-ray converter in which incident X-rays or X-ray photons generate mobile charge carriers, in particular electron-hole pairs, which are used to read out the signal. The X-ray detector can, for example, be a flat-panel detector.

[0013] A scattered beam estimate, which can also be referred to as a scattered beam estimate within the scope of the invention, is generally understood to be an estimate of the scattered rays entering an X-ray detector. In an object being examined with X-rays, a portion of the photons is typically absorbed, while another portion is scattered in one or more scattering processes. Scattered photons that enter the X-ray detector add to the actual signal, thus increasing the detector signal and typically leading to image artifacts. A scattered beam estimate can therefore be used to subtract the scattered radiation from the total signal to obtain the actual measurement signal.

[0014] An anti-scatter grid is generally a device arranged in front of the X-ray detector and designed to reduce the incidence of scattered radiation on the X-ray detector. For this purpose, the anti-scatter grid comprises lamellae. The lamellae are preferably arranged substantially parallel to one another in at least one direction. The lamellae can be arranged in front of the X-ray detector as a rectangular grid, in particular with groups of lamellae running parallel to one another in two directions. The directions of the anti-scatter grid can be differentiated into row direction and column direction. The row direction and column direction are the directions in which the detector elements are arranged. The row direction is the direction in which the rows of the anti-scatter grid run. The column direction is the direction in which the columns of the anti-scatter grid run.The row direction can in particular correspond to a z-direction of the measuring device used, in particular a computed tomography device. In a computed tomography device or a measuring device with a gantry or an examination tunnel, the z-direction is in particular an axial direction. The column direction can in particular correspond to a phi-direction or φ-direction of the measuring device used, in particular a computed tomography device. The phi-direction or φ-direction is in particular a circumferential direction in which an X-ray detector is arranged or a circumferential direction, e.g., of a gantry. Generally, the phi-direction or φ-direction is a direction substantially perpendicular to the z-direction. In at least one direction, at least two detector elements are arranged between adjacent lamellae of the anti-scatter grid. In other words, a plurality of detector elements are located between the lamellae.This arrangement can be particularly advantageous for small detector elements, such as in the case of photon-counting detectors, in order to still receive a sufficient amount of signal. Because more than one detector element is arranged between the leaves, in many cases each detector element receives a different portion of the scattered radiation. For example, a left detector element can be closer to a left leaf and therefore potentially receive less scattered radiation coming from the left than a right detector element, which is farther away from the left leaf.

[0015] The received detector data can be provided or sent, for example, from a database or directly from the X-ray detector. The detector data is divided into at least two groups of detector data. For example, the detector data can be divided into a first group and a second group. When divided into two different groups, each of the groups can in particular comprise detector data corresponding to half the amount of the original detector data. However, more than two groups can also be provided. The detector data is divided based on the position of the detector elements from which the detector data originate. The position is in particular the position relative to the nearest lamella of the anti-scatter grid. The detector data can in particular be divided alternately according to their position.For example, detector data from detector elements that have a lamella of the anti-scatter grid immediately to their left can be divided into a first group, and detector data from detector elements that have a lamella of the anti-scatter grid immediately to their right can be divided into a second group. Detector elements that have another detector element in front of the next lamella to their left or right can be understood in this sense as not having a lamella of the anti-scatter grid immediately to their left or right. If, for example, more than two detector elements are arranged between the lamellas of the anti-scatter grid, more than two different groups could optionally be provided. However, even with more than two detector elements between the lamellas, a division into just two groups can already be advantageous.For example, a group can be provided for each relative positioning of detector elements relative to the nearest slats of the anti-scatter grid.

[0016] For each of the groups, a scattered radiation distribution in the form of a scattered radiation term is determined based on the detector data. The scattered radiation distribution can, in particular, be a distribution of the scattered radiation intensities across the detector elements. The scattered radiation terms of both groups can optionally be similar or essentially identical to one another. The scattering term of the respective group is determined, in particular, using the measured intensity data of the respective group. The scattered radiation term of each of the groups is convolved using a convolution kernel individually adapted for each group. In particular, the convolution kernels are different for at least two of the groups, preferably (with more than two groups) different for each of the groups. The convolution kernels can, in particular, each be adapted depending on the geometry of the scattered radiation grid and / or the position of the detector elements of the group relative to the slats.For example, the convolution kernel can be based on a triangular shape that is or was adapted according to the geometry. For example, a side length can be adapted to the triangular shape. For example, a height and at least one flank and / or width of the convolution kernel can be adapted, in particular to the triangular shape. It can optionally be provided to perform multiple convolutions for each of the groups, for example a first time for a first direction (e.g. the phi direction or column direction) of the scattered radiation grid and a further time for a second direction (e.g. the z direction or row direction) of the scattered radiation grid. By convolving the scattered radiation terms with the convolution kernels, a plurality of folded scattered radiation terms are generated - corresponding to the number of groups. Thus, in particular, a kernel-based scattered radiation estimation with adapted convolution kernels is provided.Advantageously, by adapting the convolution kernels, in particular a high-frequency component of the scattered radiation resulting from the geometry of the anti-scatter grid.

[0017] The scattered radiation distributions according to the convolved scattered radiation terms are combined according to the position of the detector elements. In particular, the convolved scattered radiation terms can be combined alternately according to the position of the detector elements. The resulting scattered radiation estimate is, in particular, an overall scattered radiation distribution across all detector elements. In particular, the intensity values ​​of the overall scattered radiation distributions can be arranged or distributed according to the positions of the detector elements.

[0018] According to one embodiment, the scattered radiation term comprises a source term based on a scattered radiation model of scattered radiation sources, and a propagation term that estimates local photon scattering of the scattered radiation. One possibility for calculating the source term and the propagation term is described, for example, in Ohnesorge B, Flohr T, Klingenbeck-Regn K. “Efficient object scatter correction algorithm for third and fourth generation CT scanners”, Eur Radiol. 1999;9(3):563-9. doi: 10.1007 / s003300050710. PMID: 10087134 and can optionally be adopted here analogously. The source term can, for example, describe scattering derived from a simplified model. A model can, for example, be simplified in such a way that first-order scattering along the primary radiation is taken into account. In particular, it can be provided that the source term and the propagation term are multiplied with one another for the scattered radiation term.For example, a scattered radiation term pep can be of the form. pep=pSource,⋅e−κ⋅pPropagation be, where p Quelle the source term and e -κ·pAusbreitung is the propagation term and κ is a constant. The source term can be calculated from the intensity of the signal using a logarithm, for example in the form p Quelle = - log(I / I0). It is possible to integrate the two terms or the scattered radiation term, in particular across all X-rays in a measurement. In particular, multiplying the source term and the propagation term can describe a scattered radiation distribution for a single X-ray, which can be completed by integrating it across all X-rays. Using the source term and the propagation term represents a particularly advantageous, i.e., less computationally intensive, method for determining scattered radiation.

[0019] According to one embodiment, at least one of the convolution kernels is configured asymmetrically. In particular, at least two of the convolution kernels are configured asymmetrically. For example, the at least one convolution kernel can be an asymmetric triangle, in particular with unequal legs, preferably unequal left and right legs. The asymmetry of the convolution kernel can in particular correspond to and / or be based on the asymmetry of the geometry of the respective detector elements relative to the anti-scatter grid. For example, a degree of asymmetry can be set based on the distance from the anti-scatter grid to the sides of the detector element. In the prior art, in contrast, it has been common practice to use a normalized symmetric triangular convolution kernel. The asymmetric design of the convolution kernel advantageously allows the asymmetry of the underlying problem to be taken into account more precisely. A convolution ( *) with the respective convolution kernel (h rechts / links ) can be provided, for example, according to the following formula IS,right / left=(γ⋅psource,right / left⋅e−κ⋅ppropagation,right / left)∗hright / left where I S is the estimated scattered beam intensity and the scattered beam estimate, respectively, y and κ are constants and h is the convolution kernel.

[0020] According to one embodiment, the number of detector data groups into which the detector data is divided corresponds to the number of detector elements between adjacent slats of the anti-scatter grid. In other words, the detector data can be divided into as many sub-detector data as there are detector elements between the slats, with each sub-detector data containing only detector data originating from one type of detector element position relative to the slats. Advantageously, this division allows for particularly good and precise consideration of the geometry of the X-ray detector and the anti-scatter grid.

[0021] According to one embodiment, the detector data is divided into groups of detector data such that at least one of the groups originates from detector elements that differ in terms of their position relative to the slats in the phi direction from the detector elements of at least one other group. The phi direction can in particular correspond to a column direction of the anti-scatter grid. The phi direction can preferably be the direction in which more detector elements are provided than in a further direction of the anti-scatter grid, namely in particular the z direction. Advantageously, a geometric inequality of the detector elements relative to the anti-scatter grid can thus be compensated. It has been shown that this measure can achieve significant improvements in scattered radiation estimation.

[0022] According to one embodiment, the detector data is divided into, in particular at least four, groups of detector data such that at least one of the groups originates from detector elements that differ in their position relative to the slats in the phi direction from the detector elements of at least one other group, and / or that at least one of the groups originates from detector elements that differ in their position relative to the slats in the z direction from the detector elements of at least one other group. The groups that differ in the z and phi directions of their detector elements can be the same two groups and / or different pairs of groups.For example, the detector data can be divided into groups of detector data such that at least one of the groups originates from detector elements that differ in their position relative to the slats in the phi direction and in the z direction from the detector elements of at least one other group. For example, the detector data can be divided into groups of detector data such that a first of the groups originates from detector elements that differ in their position relative to the slats in the phi direction from the detector elements of at least one second group, and / or that a third of the groups originates from detector elements that differ in their position relative to the slats in the z direction from the detector elements of at least one fourth group.For example, the detector data can be divided into groups of detector data such that a first of the groups originates from detector elements which differ in terms of their position relative to the slats in the phi direction from the detector elements of at least a second group and which differ in terms of their position relative to the slats in the z direction from the detector elements of at least a third group. Advantageously, it can thus be taken into account if several detector elements are arranged between two adjacent slats in both the phi direction and the z direction. This makes it possible to achieve a particularly precise scattered radiation estimate. The phi direction can in particular correspond to a column direction and the z direction can in particular to a row direction of the anti-scatter grid. Preferably, more detector elements can be provided in the phi direction than in the z direction.It can be provided that the number of groups into which the data is divided corresponds to the number resulting from multiplying the number of detector elements in the phi direction between adjacent slats by the number of detector elements in the z direction between adjacent slats. In particular, the groups of detector data can be divided in such a way that each of the groups originates from detector elements which, in terms of their relative position, differ from the detector elements from which the other groups originate. Different in this sense means that the relative position to the slats in the phi direction and / or the relative position to the slats in the z direction is different. It can be provided to carry out two convolutions for each of the groups, in particular one convolution in the phi direction and one convolution in the z direction.

[0023] According to one embodiment, the convolution kernels are each adapted based on a precise simulation, in particular a Monte Carlo simulation, or a measurement of the scattered radiation. For example, the convolution kernels can be adapted to the simulation or to the measurement using a fitting and / or optimization method. The adaptation can, in particular, have been carried out in advance of the actual method. For example, an adaptation can be carried out for a specific imaging device and then used for subsequent applications of this method. Advantageously, the accuracy of a simulation can thus be used as a calibration, so to speak, without having to repeat the (time-consuming) simulation. The simulation and / or the measurement of the scattered radiation can, for example, be based on a defined (theoretical or actual) phantom.For example, for precise simulation, a specific phantom can be used as a simulated measurement object.

[0024] For this simulated measurement object, the scattered radiation term(s) for the groups can then be determined, and the convolution kernels are adjusted to produce a scattered radiation estimate that corresponds as closely as possible to the scattered radiation estimate from the precise simulation. A similar approach can be used with a real phantom and measurement data from this phantom. By precisely defining a phantom, the scattered radiation component of the phantom can be more easily determined directly during a measurement with the phantom. The precise simulation can be based, for example, on a statistical or an analytical approach.

[0025] The precise measurement can, for example, be based on a Monte Carlo simulation. A Monte Carlo simulation is a statistical approach. With a Monte Carlo simulation, a scatter estimate can advantageously be calculated very precisely. In particular, the Monte Carlo simulation can be adapted to the respective X-ray detector used. The use of a Monte Carlo simulation for scatter estimation per se is known in the art and is described, for example, in Rührnschopf E.-P. and Klingenbeck K. “A general framework and review of scatter correction methods in x-ray cone-beam computerized tomography. Part 1: Scatter compensation approaches”, Med Phys. 2011 Jul;38(7):4296-311. doi: 10.1118 / 1.3599033. Erratum in: Med Phys. 2011 Oct;38(10):5830. Erratum in: Med Phys. 2011 Oct;38(10):5830. PMID: 21859031 and in Rührnschopf, E.-P. and Klingenbeck, K., “A general framework and review of scatter correction methods in cone beam CT.Part 2: scatter estimation approaches". Med Phys. 2011 Sep;38(9):5186-99. doi: 10.1118 / 1.3589140. PMID: 21978063. Due to the relatively high time expenditure of such calculations, it is advantageous, however, if this calculation is only carried out in advance in a calibration step, and then, particularly in daily use, the method according to the invention is applied. An optimization function can be used to adapt to a Monte Carlo simulation. The optimization function can be used to adjust an error in the image space caused by scattered radiation. The error in the image space can be described with a ratio of scattered radiation to primary signal ("scatter-to-primary"), so that a large ratio corresponds to a large error. The error can be calculated and adjusted accordingly for each channel of the X-ray detector.It may be planned to verify the adjustment again with a test measurement in the image space as part of a calibration. For example, a scatter-to-primary mean absolute percentage error (SPMAPE) can be used to optimize the convolution parameters. The scatter-to-primary ratio correlates directly with the error in the reconstructed image. In detector pixels or detector elements where the primary signal is low, scattered photons can lead to a large image error. Therefore, it is particularly important to correct these positions with a high scatter-to-primary ratio. The optimization can be performed using a cost function. The cost function can be described, for example, by the following formula: lSPMAPE=100N∑|IScatter,Prediction−IScatter,GTIPrimary| where I Scatter,Prediction is the estimated scattered beam intensity by the convolution approach, I Scatter,GTis the scattered beam intensity by the Monte Carlo simulation and I Primary The primary beam intensity is determined by the Monte Carlo simulation. The parameter N corresponds to the number of detector pixels or detector elements, and the sum is calculated over the number N. An analytical approach can be based, for example, on a Boltzmann transport equation.

[0026] A further aspect of the invention is a method for correcting scattered radiation effects in image data from an X-ray detector having a plurality of detector elements, wherein a scattered radiation grid is arranged in front of the X-ray detector, wherein in at least one direction at least two detector elements are arranged between respectively adjacent lamellae of the scattered radiation grid, wherein the method comprises the steps of the method for determining a scattered radiation estimate as described herein and at least the following step: correcting the image data based on the scattered radiation estimate. All advantages and features of the method for determining a scattered radiation estimate can be transferred analogously to the method for correcting scattered radiation effects and vice versa. To correct the image data, in particular the scattered radiation estimate is offset against the detector data.For example, the scatter estimate can be subtracted from the intensity signal of the detector data to obtain a scatter corrected intensity signal of the detector data.

[0027] A further aspect of the invention is a computer program product or computer-readable storage medium comprising instructions that, when executed by a computer and / or an evaluation device, cause the computer and / or an evaluation device to perform the steps of one of the methods as described herein. All advantages and features of the method for determining a scattered radiation estimate and the method for correcting scattered radiation effects can be transferred analogously to the computer program product or the computer-readable storage medium, and vice versa. The storage medium can be, for example, a hard disk, an SSD, a flash memory, an online server, etc.

[0028] A further aspect of the invention is an X-ray-based imaging device, in particular a medical imaging device, comprising an X-ray source, an X-ray detector with a plurality of detector elements, a scattered radiation grid with slats arranged in front of the X-ray detector, wherein in at least one direction of the scattered radiation grid, at least two detector elements are arranged between adjacent slats of the scattered radiation grid, and an evaluation device configured to receive detector data from the X-ray detector and to execute one of the methods described herein. All advantages and features of the method for determining a scattered radiation estimate, the method for correcting scattered radiation effects, and the computer program product or computer-readable storage medium can be transferred analogously to the imaging device, and vice versa.For the evaluation device to receive the detector data, a corresponding wireless and / or wired connection can be provided between the X-ray detector and the evaluation device. The evaluation device can have an interface for receiving the detector data. The X-ray detector can have an interface for transmitting the detector data. The imaging device can be, in particular, a computed tomography device, a C-arm device, or a radiography device for conventional X-ray imaging. The anti-scatter grid can comprise a two-dimensional grid of slats. It can be provided that a plurality of detector elements are arranged between the slats of the anti-scatter grid in two directions of the anti-scatter grid. For example, at least two detector elements can be arranged between two adjacent slats in a first direction (e.g., phi direction), and in a second direction (e.g., phi direction), at least two detector elements can be arranged between two adjacent slats.B. z-direction) at least two, in particular at least three, detector elements are arranged between two adjacent slats. The anti-scatter grid and the X-ray detector can be designed, in particular in their structure and / or their relative arrangement to one another or generally in their properties, as described herein with reference to an anti-scatter grid and / or an X-ray detector. The term X-ray source is to be interpreted broadly within the scope of this invention. It generally relates to a device that is suitable for generating X-rays. X-ray source can also be referred to as an X-ray source. The X-rays can, for example, be fan-shaped. The X-ray source can, for example, comprise at least one X-ray tube for generating the X-rays. Optionally, the X-ray source can comprise a plurality of X-ray tubes.

[0029] All embodiments described herein can be combined with one another unless explicitly stated otherwise.

[0030] Embodiments are described below with reference to the attached figures. Fig. 1 shows a flowchart of a computer-implemented method for determining a scattered beam estimate according to an embodiment of the invention, Fig. 2 shows schematically the principle of an X-ray detector with exactly one detector element between adjacent lamellae of a scattered radiation grid, Fig. 3 shows schematically the principle of an X-ray detector according to the invention with several detector elements between adjacent lamellae of a scattered radiation grid, Fig. Figure 4 shows an example of an X-ray detector according to the invention, in front of which a scattered radiation grid with several lamellas is arranged, Fig. 5 shows the incidence of X-rays on the detector elements between two adjacent lamellae of an X-ray detector according to the invention, Fig. 6 shows a measurement of an anthropomorphic thorax phantom on a computer tomography device with an energy-integrating X-ray detector, Fig. 7 shows a measurement of the anthropomorphic thorax phantom on a computer tomography device with a photon-counting X-ray detector, Fig. Fig. 8 shows schematically an arrangement of lamellae of a scattered radiation detector in front of an X-ray detector according to an embodiment of the invention, Fig. 9 shows detector data of a central 30 cm water phantom in log space, Fig. 10 shows the principle of splitting detector data from a detector according to the Fig. 8 shown X-ray detector according to an embodiment of the invention, Fig. 11 shows a scattered beam distribution of a central 30 cm water phantom based on two scattered beam terms of detector data split according to the invention, Fig. 12 shows a symmetric convolution kernel for convolution in the phi direction according to the prior art, Fig. 13 shows two asymmetric convolution kernels for convolution in the phi direction according to an embodiment of the invention, Fig. Figure 14 shows a scattered beam distribution of a central 30 cm water phantom for a folded scattered beam term of the left group of Fig. 10 according to an embodiment of the invention together with a scattered beam distribution based on a Monte Carlo simulation and a scattered beam distribution according to the prior art, Fig. Figure 15 shows a scattered beam distribution of a central 30 cm water phantom for a folded scattered beam term of the right group of Fig. 10 according to an embodiment of the invention together with a scattered beam distribution based on a Monte Carlo simulation and a scattered beam distribution according to the prior art, Fig. 16 shows a combined scattered beam estimate for a central 30 cm water phantom in the form of a scattered beam distribution according to an embodiment of the invention together with a scattered beam distribution based on a Monte Carlo simulation and a scattered beam distribution according to the prior art, Fig. 17 shows a flowchart of a method for correcting scattered radiation effects in image data from an X-ray detector having a plurality of detector elements and having a scattered radiation grid in front of the X-ray detector according to an embodiment of the invention, Fig. 18 shows the influence of high-frequency scatter effects without scatter correction in exemplary image data, Fig. 19 shows the influence of high-frequency scattered beam effects with a scattered beam correction according to the state of the art in exemplary image data, Fig. Figure 20 shows the influence of high frequency scattered beam effects with a scattered beam correction according to a method as described in Fig. 1 described methods in exemplary image data, Fig. 21 shows detector data of a decentralized 30 cm water phantom in log space, Fig. 22 shows a scattered beam distribution of a decentralized 30 cm water phantom based on two scattered beam terms of detector data split according to the invention, Fig. Figure 23 shows a scattered beam distribution of a decentralized 30 cm water phantom for a folded scattered beam term of the left group of Fig. 10 according to an embodiment of the invention together with a scattered beam distribution based on a Monte Carlo simulation and a scattered beam distribution according to the prior art, Fig. Figure 24 shows a scattered beam distribution of a decentralized 30 cm water phantom for a folded scattered beam term of the right group of Fig. 10 according to an embodiment of the invention together with a scattered beam distribution based on a Monte Carlo simulation and a scattered beam distribution according to the prior art, Fig. 25 shows a combined scattered beam estimate for a decentralized 30 cm water phantom in the form of a scattered beam distribution according to an embodiment of the invention together with a scattered beam distribution based on a Monte Carlo simulation and a scattered beam distribution according to the prior art, and Fig. 26 shows an X-ray based imaging device according to an embodiment of the invention.

[0031] Fig. Figure 1 shows a flowchart of a computer-implemented method for determining a scattered radiation estimate according to an embodiment of the invention. The scattered radiation estimate is performed for an X-ray detector 1 having a plurality of detector elements 11. A scattered radiation grid 2 is provided in front of the X-ray detector 1. In this case, at least two detector elements 11 are arranged in at least one direction between adjacent lamellae 21 of the scattered radiation grid 2. This concept of a scattered radiation grid is illustrated in Figure 2 with reference to Fig. 2 and Fig. 3 illustrates. Fig. Figure 2 schematically shows the principle of an X-ray detector 1 with exactly one detector element 11 between adjacent slats 21 of an anti-scatter grid 2. Two detector elements 11 are shown, each surrounded on each side by slats 21. Such a detector shape is typically used in energy-integrating X-ray detectors 1. Fig. Figure 3, on the other hand, schematically shows the inventive principle of an X-ray detector 1, each with a plurality of detector elements 11 between adjacent lamellae 21 of an anti-scatter grid 2. Such a detector shape is often preferred, for example, in photon-counting X-ray detectors 1. In this example, there are a plurality of detector elements 1 between the adjacent lamellae 21 in both the phi direction p (four detector elements 11) and the z direction z (six detector elements 11).

[0032] Fig. Figure 4 shows an example of an X-ray detector 1 according to the invention, in front of which a scattered radiation grid 2 with several lamellae 21 is arranged, between which, according to the principle of Fig. 3, several detector elements 11 are provided in both the phi-direction p and the z-direction z. The incidence of X-rays 31, 32, 33 on the detector elements 11 between two adjacent lamellae 21 is Fig. 5 illustrates. X-ray signal beams 31 that are just incident are not blocked by the slats 21 here and strike the detector elements directly. Some scattered rays 33 are blocked by the slats 21 of the anti-scatter grid 2, but further scattered rays 32 strike the detector elements 11. However, there are differences with regard to the incoming scattered rays 32, depending on whether the detector elements 11 are arranged to the left or right in the phi direction, directly next to one of the slats 21 or further away from the next slat 21. The tendency is for more scattered rays 32 to strike a detector element 11 from a particular side, the further the detector element 11 is arranged on this side from the next slat 21. Depending on the position of the detector elements 11, there is therefore a difference in the scattered beam intensity for the various detector elements 11, in particular also for neighboring detector elements.If this difference is not taken into account, as is currently the case in the state of the art, artifacts may arise when calculating the incoming scattered rays 32.

[0033] The Fig. 6 and Fig. 7 serve to illustrate the effect that can result from several detector elements 11 between adjacent slats. Fig. Figure 6 shows a measurement of an anthropomorphic thorax phantom on a computer tomography device with an energy-integrating X-ray detector 1, which basically has a scattered radiation grid 1 as described with reference to Fig. 2 described. Fig. Figure 7, however, shows a measurement of the anthropomorphic thorax phantom on a computer tomography device with a photon-counting X-ray detector 1, which basically has a scattered radiation grid 1 as with reference to Fig. 3. The scan parameters used in both cases are: 140 kV, 600 mA, slice thickness 5 mm, slice thickness step 5 mm, matrix size 1024 × 1024. The images each show a field of view with a side length of 40 cm; in the bottom right, a section with a field of view of 4 cm is shown enlarged. It can be seen that a high-frequency interference signal (partial rings) has been added to the photon-counting X-ray detector 1. This interference signal is due to the configuration of the detector elements 11 relative to the anti-scatter grid 2. The anti-scatter correction commonly used in the prior art does not correct this high-frequency component of the scattered radiation 32 resulting from the geometry of the anti-scatter grid 2. To avoid these artifacts, corresponding frequencies must therefore be suppressed in the image space, which, however, can be disadvantageous in itself, since it can also suppress signals.With the inventive scattered radiation estimation, such high-frequency components can be prevented or corrected at the initial stage, eliminating the need to suppress frequencies in the image space. This allows for the use of a very sharp reconstruction kernel without significantly impairing clinical diagnostics. To avoid these artifacts, certain frequencies in the image space previously had to be suppressed. By correctly correcting scattered radiation at the initial stage, no frequencies need to be suppressed, and clinical diagnostics are still possible even when using a sharp reconstruction kernel. This allows, for example, (real) edges to be displayed more clearly.

[0034] In the following, the method according to the embodiment of Fig. 1 based on the Fig. 8 to 16. The starting point is an X-ray detector 1 as shown in Fig. 8. The X-ray detector 1 has six detector elements 11 between each of the slats 21 of the anti-scatter grid 2, which are designated (0,0), (1,0), (0,1), (1,1), (0,2), and (1,2) according to their position relative to the anti-scatter grid. This corresponds to six different relative positions for the detector elements 11. In the phi direction p, there are two detector elements 11 between each of the adjacent slats 21, and in the z direction z, there are three detector elements 11 between each of the adjacent slats 21. In total, the X-ray detector 1 in this example has 1376 columns and 144 rows, i.e., 1376 detector elements 11 in the phi direction times 144 detector elements 11 in the z direction. However, an X-ray detector 1 with a different number and / or distribution of detector elements 11 can be used.

[0035] Referring to Fig. 1, detector data are received in a first method step 101. Fig. Figure 9 shows exemplary detector data, or here representatively a row (row 72 of 144 rows) of detector data that has already been preprocessed and transferred to the log space. The detector data can be described, for example, with p=-log(I / I0), essentially as the logarithm of the normalized signal intensity. The data originate from a 30 cm water phantom that was placed centrally in a computed tomography device. This detector data is divided into at least two groups of detector data. The division is based on the position of the detector elements 11, from which the detector data originate, relative to the nearest lamella 21 of the scattered radiation grid 2. In the example shown here, the detector data is divided into two groups, as shown in Fig. 10. In this case, the detector data (0, Y) originating from a detector element 11 directly to the right of a slat 21 are divided into a first group (in Fig. 10 the left group) and the detector data (1, Y) coming from a detector element 11 directly to the left of a slat 21 are divided into a second group (in Fig. 10 the right group). Thus, the detector data is divided into two groups of detector data such that the left group (in the illustration here) originates from detector elements 21 which differ in terms of their position relative to the slats 21 in the phi direction p from the detector elements 21 of the right group (in the illustration here). Data with 1376 columns resulted in two groups of data, each with 688 columns. It would also be conceivable to divide the data into more than two groups. In particular, a division in the z-direction z could also be provided. For example, based on the six different relative positions, six different groups could also result. The number of groups of detector data into which the detector data is divided would thus correspond to the number of detector elements 21 between adjacent slats 21 of the anti-scatter grid 2.However, even dividing the data into two groups is clearly advantageous. The two groups are designated p, for example. links and p rechts designated.

[0036] In a further step 102, a scatter distribution, which estimates scattering based on the detector data, is determined in the form of a scatter term for each of the groups of detector data. The scatter term may include a source term p Quelle which is based on a simplified scattered radiation model of scattered radiation sources, and a propagation term e -κ·pAusbreitung , which estimates a local photon scattering of the scattered rays. The scattered ray distribution pep can thus be estimated with a scattered ray term of the form pep=pSource,⋅e−κ⋅pPropagation for each of the groups (pep links 5 and pep rechts 6) can be calculated by integrating over all X-rays. A corresponding scattered beam distribution for both groups is shown in Fig. 11 shown.

[0037] Referring again to Fig. 1, in a further step 103, the scattered radiation term of each group is convolved with a convolution kernel individually adapted for each group, so that several convolved scattered radiation terms are generated according to the number of groups. For comparison, in Fig. 12 shows a normalized convolution kernel for convolution in the phi direction p according to the prior art. This is symmetric and not specifically adapted to the X-ray detector 1 in Fig. 8. According to the state of the art, this convolution kernel would also be applied to the entire data. Fig. Figure 13, however, shows two convolution kernels 51, 61 according to an embodiment of the invention. One convolution kernel 51 (h links ) is for the left group according to Fig. 8 and the other convolution kernel 61 (h rechts ) is for the right group according to Fig. 8. The convolution kernels 51, 61 are asymmetric and adapted to the relative position of the detector elements 11 relative to the slats. Accordingly, pep links with h links and pep rechts with h rechts The free model parameters of the convolution kernels were adjusted by comparing them with a Monte Carlo simulation. The convolution can be performed (for both left and right) according to the following formula to obtain an estimated scattered beam intensity I S,rechts / links for each of the groups (with the constants y and κ): IS,right / left=(γ⋅psource,right / left⋅e−κ⋅ppropagation,right / left)∗hright / left

[0038] If more than two detector elements 11 (e.g., six detector elements as in this case) are located between the lamellae 21 of the scattered radiation grid 2, more than two different convolution kernels could be applied (e.g., six convolution kernels corresponding to six groups). For example, for each of the groups, a convolution could first be performed in the phi direction p and then in the z direction z. The resulting scattered radiation distribution is shown in Fig. 14 for the folded scattered radiation term 51 of the left group and in Fig. 15 for the convolved scattered radiation term of the right group. Shown are a (precise) Monte Carlo simulation 7, a scattered radiation distribution 8 determined according to the method according to the invention as described here, and a scattered radiation distribution 9 determined according to the prior art with a single convolution kernel as in Fig. 12. It can be seen that the scattered radiation terms 8 convolved according to the invention are significantly closer to the (quasi-exact, but more time-consuming) Monte Carlo simulation 7 than the scattered radiation terms 9 estimated according to the state of the art.

[0039] Referring again to Fig. 1, in a further step 104, the scattered radiation distributions are alternately combined according to the folded scattered radiation terms according to the position of the detector elements 11 in the X-ray detector 1. As a result, in this example, on the horizontal axis, the detector pixels are arranged according to the arrangement of the detector elements 11 in the X-ray detector 1 in Fig. 8. Thus, a scattered beam estimate is obtained as shown in the present example in Fig. 16. The scattered radiation distribution 8, which was obtained using the method according to the invention described here, is shown again, together with a scattered radiation distribution 7 according to a Monte Carlo simulation and a scattered radiation distribution 9 created using an unadapted convolution kernel. The scattered radiation distribution 8 determined according to the invention is significantly closer to the realistic scattered radiation distribution 7 of the Monte Carlo simulation than the scattered radiation distribution 9 according to the prior art. The scattered radiation distribution 8 of the method according to the invention also takes into account the high-frequency components of the scattered radiation, which is reflected in the high-frequency oscillations of the scattered radiation distribution 8. The scattered radiation distribution 9 determined according to the prior art, in contrast, clearly does not take these high-frequency components into account. Thus, the scattered radiation estimation according to the invention allows for improved scattered radiation correction.

[0040] Fig. 17 shows a flowchart of a method for correcting scattered radiation effects in image data from an X-ray detector having a plurality of detector elements, wherein a scattered radiation grid 2 is arranged in front of the X-ray detector 1, wherein in at least one direction at least two detector elements are arranged between respectively adjacent lamellae of the scattered radiation grid, according to an embodiment of the invention. Steps 201 to 204 correspond to steps 101 to 104 of the method described with reference to Fig. 1. In a further step 205, the image data are corrected based on the determined scattered beam estimate. This is done in particular by calculating the scattered beam estimate with the input signal of the detector elements 11, for example, by subtracting it. The effect of this correction can be seen, for example, in the Fig. 18 to 20. These figures each show a photograph of an anthropomorphic head phantom with a real skull bone. Fig. 18, no scatter correction was performed. The semi-circular, high-frequency components of the scattered radiation (sharp ring artifacts) are visible. Fig. 19, a state-of-the-art scatter correction was performed using an unadapted convolution kernel. Here, too, the sharp ring artifacts are still visible in the image section shown, almost unchanged. Fig. In contrast, the method according to the example described here was applied to Figure 20. It can be seen that the sharp ring artifacts are significantly reduced.

[0041] The method according to the invention can be flexibly applied to various situations. As a further example, the Fig. 21 to 25. The Fig. 21 to 25 correspond to the Fig. 9, 11 and 14-16, this time using a 30cm eccentric water phantom instead of a central water phantom. Compared to Fig. 9 are in Fig. 21 the detector data is therefore slightly decentralized. Accordingly, it also results in Fig. 22 shows a decentralized arrangement of the scattered beam distribution according to the two scattered beam terms determined from the split detector data. The scattered beam distribution according to the convolved scattered beam terms is also correspondingly asymmetric in the Fig. 23 (for the left part) and 24 (for the right part). In Fig. Finally, the composite scattered beam estimate is shown again in Figure 25. In the Fig. 23-25, the scattered beam distribution 8 determined with the inventive method based on adapted convolution kernels is again significantly closer to the scattered beam distribution 7 of the Monte Carlo simulation than the scattered beam distribution 9 with an unadapted convolution kernel. In addition, Fig. 25 again shows the high-frequency components of the scattered radiation in the scattered radiation distribution 8 determined according to the invention.

[0042] In Fig. 26 shows an X-ray-based imaging device, namely a computed tomography device 70. The computed tomography device comprises a gantry 71, which in turn comprises an X-ray source and an X-ray detector 1 with a plurality of detector elements 11 and an anti-scatter grid 2 with slats 21. In at least one direction of the anti-scatter grid 2, at least two detector elements 11 are arranged between adjacent slats 21 of the anti-scatter grid 2. The X-ray detector 1 and the anti-scatter grid 2 can, for example, be configured as shown in the Fig. 3 to 5 or Fig. 8. The computer tomography device 70 further comprises an evaluation device 72 which is configured to receive detector data from the X-ray detector, a method according to the invention, for example the method as described in relation to Fig. 1 or Fig.17. Also shown are the phi direction p and the z direction z of the computed tomography device 70.

Claims

[1] Computer-implemented method for determining a scattered radiation estimate for an X-ray detector (1) having a plurality of detector elements (11), wherein a scattered radiation grid (2) is arranged in front of the X-ray detector (1), wherein in at least one direction at least two detector elements (11) are arranged between respectively adjacent lamellae (21) of the scattered radiation grid (2), the method comprising the following steps: - receiving detector data and dividing the detector data into at least two groups of detector data based on the position of the detector elements (11) from which the detector data originate, relative to the nearest lamella (21) of the anti-scatter grid (2); - determining a scatter distribution that estimates scattering based on the detector data, in the form of a scatter term for each of the groups of detector data; - Convolving the scattered radiation term of each of the groups with a convolution kernel individually adapted for each group, so that several convolved scattered radiation terms are generated according to the number of groups; - combining the scattered radiation distributions according to the convolved scattered radiation terms according to the position of the detector elements (11) in order to obtain a scattered radiation estimate. [2] Method according to claim 1, - where the scattered radiation term comprises a source term based on a scattered radiation model of scattered radiation sources and a propagation term estimating local photon scattering of the scattered radiation. [3] Method according to one of the preceding claims, - wherein at least one of the convolution kernels is designed asymmetrically, in particular at least two of the convolution kernels are designed asymmetrically. [4] Method according to one of the preceding claims, - wherein the number of groups of detector data into which the detector data are divided corresponds to the number of detector elements (11) between adjacent lamellae (21) of the anti-scatter grid (2). [5] Method according to one of the preceding claims, - wherein the detector data are divided into groups of detector data such that at least one of the groups originates from detector elements (11) which differ from the detector elements (11) of at least one other group with regard to their position relative to the slats (21) in the phi direction (p). [6] Method according to one of the preceding claims, - wherein the detector data are divided into groups of detector data such that at least one of the groups originates from detector elements (11) which differ from the detector elements (11) of at least one other group with regard to their position relative to the slats (21) in the phi direction (p), and / or that at least one of the groups originates from detector elements (11) which differ from the detector elements (11) of at least one other group with regard to their position relative to the slats (21) in the z direction (z). [7] Method according to one of the preceding claims, - wherein the convolution kernels are each adapted based on a precise simulation, in particular a Monte Carlo simulation, or a measurement of the scattered radiation. [8] Method for correcting scattered radiation effects in image data from an X-ray detector (1) having a plurality of detector elements (11), wherein a scattered radiation grid (2) is arranged in front of the X-ray detector (1), wherein in at least one direction at least two detector elements (11) are arranged between respectively adjacent lamellae (21) of the scattered radiation grid (2), wherein the method comprises the steps of the method according to one of the preceding claims and at least the following step: - Correcting the image data based on the scattered beam estimation. [9] Computer program product or computer-readable storage medium comprising instructions which, when executed by a computer and / or an evaluation device (72), cause the computer and / or an evaluation device (72) to carry out the steps of the method according to one of the preceding claims. [10] X-ray based imaging device, in particular medical imaging device, comprising - an X-ray source, - an X-ray detector (1) with a plurality of detector elements (11), - an anti-scatter grid (2) with slats (21) arranged in front of the X-ray detector (1), wherein in at least one direction of the anti-scatter grid (2) at least two detector elements (11) are arranged between adjacent slats (21) of the anti-scatter grid (2), and - an evaluation device (72) configured to receive detector data from the X-ray detector (1) and to carry out the method according to one of claims 1 to 8.

Images