Method for determining individual scatter radiation images for X-ray image points to be considered in an X-ray image

Abstract

The invention relates to a computer-implemented method for determining individual scatter radiation images for X-ray image points to be considered in an X-ray image, comprising the steps: - Obtaining an X-ray image, wherein the X-ray image is captured using an X-ray detector and comprises a plurality of X-ray image points, each of which is assigned to a detector area of ​​the X-ray detector, where several or all of the X-ray image points are X-ray image points to be considered; - Providing an estimation algorithm which, based on the X-ray image, determines a single scatter radiation image assigned to each of the X-ray image points to be considered, wherein each individual scatter radiation image has a multitude of scatter radiation image points, each of which is assigned to a detector area of ​​the X-ray detector, where the image values ​​of the scattered radiation image points of the single scattered radiation image correlate with the intensity of the scattered radiation which, starting from the primary radiation incident on a detector area assigned to the respective X-ray image point to be considered, is scattered onto the detector area assigned to the respective scattered radiation image point; - Determining an individual scatter radiation image for each of the X-ray image points to be considered by applying the estimation algorithm to the X-ray image.

Description

[0001] Regardless of the grammatical gender of a particular term, persons with male, female or other gender identities are included.

[0002] The invention relates to a computer-implemented method for determining individual scatter radiation images for X-ray image points to be considered in an X-ray image. The invention also relates to a computer-implemented method for training at least one trained model, a data processing device, a computer program, and a data storage medium.

[0003] In the field of X-ray imaging, scattered radiation has a significant impact on the image quality that can be achieved. For example, in computed tomography, the presence of scattered radiation can lead to fringe artifacts, blurring, or low-frequency distortion of image contrast.

[0004] To reduce the influence of scattered radiation, an anti-scatter grating can be placed in front of the X-ray detector of an imaging modality, thus reducing the occurrence of scattered radiation artifacts. However, anti-scatter gratings also absorb some of the primary radiation, which can lead to a deterioration in image quality for a given X-ray dose. Conversely, this may necessitate the use of higher X-ray doses to achieve a given image quality.

[0005] In some applications, such as neurovascular imaging, for example when examining aneurysms or embolic strokes in the brain, the use of anti-scatter gratings is therefore typically avoided today. In this case, scatter artifacts can be reduced by software-based anti-scatter correction.

[0006] Software-based scatter radiation correction can, in principle, be achieved by simulating X-ray imaging, for example by solving the Boltzmann transport equation or using a Monte Carlo method. However, due to the high computational effort required for the simulation, such corrections are not very suitable when low-latency imaging, i.e., at least approximately real-time imaging, is required, for example, to accompany a medical procedure.

[0007] The same problem arises with so-called empirical methods, which involve iterative optimization of projection images, which is also very computationally intensive and therefore not suitable for real-time use.

[0008] Some approaches use scatter correction via a machine learning-trained model for image optimization to minimize the computational effort required. For example, such a model can directly generate a scatter-corrected image from a projection image. However, the results often exhibit insufficient image quality. This is particularly true when the model is trained only for a specific imaging configuration, such as a particular imaging geometry, specific imaging parameters, and / or a specific patient group. Generalization is not readily possible because the image processing in this approach is heavily influenced by the image datasets used for training.

[0009] DE 10 2004 029 009 A1 discloses a device for scatter radiation correction in computed tomography (CT) scans. DE 10 2004 029 010 A1 also discloses a device configured for scatter radiation correction. DE 10 2021 205 294 B3 discloses a CT scanner with a control unit for controlling the operation of the CT scanner. DE 10 2023 200 426 A1 discloses a method for correcting initial images from an imaging device.

[0010] The invention is therefore based on the objective of providing an improved approach for enhancing the image quality of X-ray and CT scans. In particular, the provided approach should also enable interventional imaging and / or imaging with lower latency. Furthermore, the approach should be flexibly applicable to various imaging tasks.

[0011] The problem is solved according to the invention by a method and a device according to the independent claims. Preferred embodiments are specified in the dependent claims.

[0012] According to one aspect of the invention, a computer-implemented method for determining individual scatter radiation images for X-ray image points to be considered in an X-ray image is provided. The method comprises at least the following steps: One step involves obtaining an X-ray image, which is captured by an X-ray detector and comprises a multitude of X-ray image points. Each X-ray image point is assigned to a detector area of ​​the X-ray detector. Several or all of the X-ray image points are the X-ray image points to be examined.

[0013] One step involves providing an estimation algorithm which, based on the X-ray image, determines a single scatter radiation image for each X-ray image point under consideration. Each individual scatter radiation image contains a multitude of scatter radiation image points, each corresponding to a detector area of ​​the X-ray detector. The image values ​​of the scatter radiation image points of the individual scatter radiation image correlate, preferably exclusively, with the intensity of the scatter radiation that is scattered onto the detector area corresponding to the respective scatter radiation image point, originating from the primary radiation incident on a detector area associated with that X-ray image point.

[0014] One step involves determining an individual scatter radiation image for each of the X-ray image points to be considered by applying the estimation algorithm to the X-ray image.

[0015] An X-ray image can be obtained, for example, using an imaging system. The X-ray image, whether two-dimensional or three-dimensional, can be based on X-ray imaging or a simulation of X-ray imaging. The imaging system can include an X-ray source, an X-ray detector, and / or a collimator that shapes the primary X-ray radiation. The imaging device can be, for example, an X-ray imaging device in which the images are acquired using X-rays. In particular, this can be a CT (computed tomography) scanner, such as a C-arm scanner.

[0016] In the context of this description, an X-ray image comprises a plurality of X-ray image points. Preferably, image values ​​of X-ray image points describe the X-ray intensity or dose received in a detector area associated with the respective image point. In one embodiment, the X-ray image can be an attenuation image or be based on an attenuation image. In this case, image values ​​of X-ray image points describe the attenuation that primary radiation, i.e., an X-ray beam, experiences before striking a detector area associated with the respective image point.

[0017] Regardless of whether used in the context of describing X-ray image points or the scatter radiation image points described below, the detector area can be a single detector pixel or comprise multiple detector pixels. Combining several detector pixels, for example, 2x2 pixels, into a single image point is called "binning." Binning offers advantages in terms of efficiency when processing X-ray images.

[0018] In one embodiment, the X-ray image is a two-dimensional image. In this case, the X-ray image points can be configured as pixels. Alternatively, the X-ray image can be a three-dimensional image, in which the X-ray image points can be configured as voxels. The same applies to the scatter radiation image points described below and, more generally, to image points of a cross-sectional image, which can be determined, for example, by reconstruction based on X-ray images and / or based on scatter radiation images. The X-ray image, a single or total scatter radiation image, or, more generally, a cross-sectional image, can represent a complete image of an object under investigation or a section of a larger image.

[0019] Some, several, or all of the X-ray image points are classified as X-ray image points to be considered. The X-ray image points to be considered can represent the entire X-ray image or a partial area of ​​it. In one embodiment, the partial area defined by the X-ray image points to be considered is a contiguous area. Alternatively, the X-ray image points to be considered can define several spaced-apart partial areas of the X-ray image.

[0020] The X-ray image points to be considered can be selected, for example, by a preselection rule or algorithm. For instance, only those X-ray image points whose intensity falls below and / or exceeds a predefined threshold can be considered. This can lead to a more efficient generation of individual scatter radiation images because it eliminates unnecessary calculations that contribute little to the overall understanding.

[0021] The invention encompasses the pre-processing of the X-ray image prior to processing by the estimation algorithm. For example, X-ray image points can be grouped or summarized so that, after pre-processing, the X-ray image points represent a summary of the intensity or dose information originally contained within them. For instance, the X-ray image points can be grouped into 2x2 groups for processing by the estimation algorithm, so that the image value of a single X-ray image point is determined from the image values ​​of several original X-ray image points after pre-processing. In other words, additional "binning" can be performed at the X-ray image point level to further improve the efficiency of X-ray image processing.

[0022] The statements made in connection with X-ray images and X-ray image points are also applicable to adapted X-ray images and the corresponding X-ray image points of adapted X-ray images.

[0023] The statements also apply to medical images in general, adapted medical images, and cross-sectional images.

[0024] The term "medical images" is broadly defined and encompasses images acquired for medical purposes using an imaging modality. Specifically, it includes X-ray images (i.e., projection images) and cross-sectional images obtained by reconstructing projection images. Such cross-sectional images are obtained, for example, in a CT scan by reconstructing the X-ray / projection images. These cross-sectional images may, for instance, be oriented in the transverse plane.

[0025] The intensity of the X-ray radiation measured at a specific area of ​​the X-ray detector has two components: Firstly, the intensity reflects the primary radiation, i.e., radiation emitted by the X-ray source that did not interact with the object under investigation and therefore reaches the specific detector area without deflection. Secondly, the measured intensity is caused by scattered radiation, i.e., radiation that was originally directed towards another area of ​​the X-ray detector but was deflected on its way to reaching the specific area of ​​the X-ray detector.

[0026] The primary radiation can be approximated as a straight X-ray beam or as X-rays. The X-ray beam experiences attenuation on its way from the X-ray source to the X-ray detector, and this attenuation can be described by an overall attenuation parameter. The overall attenuation parameter can be a line integral that describes the attenuation of the X-ray beam. The overall attenuation parameter can be an overall attenuation coefficient. The X-ray beam can have an initial intensity, i.e., an initial number of photons, I0, and according to the Beer-Lambert law, I = I0e-1. Σxµtot(x)x are attenuated, whereby µ tot (x) is a total attenuation coefficient of the object under investigation at location x, which describes the total attenuation of the X-ray beam at location x.

[0027] The scattered radiation that strikes a certain area of ​​the X-ray detector originates from surrounding X-rays whose primary radiation does not strike that specific area of ​​the X-ray detector, but rather surrounding areas.

[0028] Consequently, an X-ray beam directed at a specific area of ​​the X-ray detector produces primary radiation, creating intensity in that specific area. Scatter radiation causes measurable intensity in other, surrounding areas of the X-ray detector. The areas of the X-ray detector are assigned to specific X-ray image points, so an X-ray beam emits primary radiation to a specific X-ray image point (or points) assigned to it and scatter radiation to other X-ray image points.

[0029] By applying the estimation algorithm to an X-ray image, the corresponding distribution of scattered radiation emanating from the X-rays under consideration can be calculated, generating a corresponding individual scattered radiation image for each X-ray image point. In other words, by providing an X-ray image as an input variable for the estimation algorithm and executing the algorithm, an individual scattered radiation image corresponding to each X-ray image point under consideration is generated. For example, if the X-ray image comprises 100 x 100 X-ray image points, and all X-ray image points are X-ray image points under consideration, a total of 100 x 100, i.e., 10,000, individual scattered radiation images are generated. Each individual scattered radiation image is assigned to a specific X-ray image point under consideration.

[0030] A single scatter radiation image describes, for a given X-ray image point under consideration, the distribution of scatter radiation emanating from the primary radiation that strikes a detector area associated with that X-ray image point. In other words, for an X-ray beam directed at a given X-ray image point, the corresponding single scatter radiation image describes how the scatter radiation emanating from the X-ray beam is distributed across the detector.

[0031] The individual scatter radiation image comprises a plurality of scatter radiation image points, each corresponding to a detector area of ​​the X-ray detector. In one embodiment, the number and arrangement of the scatter radiation image points correspond to the number and arrangement of the X-ray image points to be considered, or to the number and arrangement of the X-ray image points themselves. Preferably, the dimensions of the individual scatter radiation images correspond to the dimensions of the X-ray image, for example, 2D or 3D. More preferably, the sizes of the individual scatter radiation images correspond to the sizes of the X-ray image, e.g., the number of pixels or voxels. This simplifies the processing of the individual scatter radiation images, for example, when the individual scatter radiation images are to be combined or compared with the X-ray images.

[0032] The image values ​​of scatter radiation image points correlate with the intensity of the scatter radiation that, originating from the primary radiation associated with the respective X-ray image point under consideration, strikes the detector area corresponding to that scatter radiation image point. The image values ​​displayed in a single scatter radiation image thus correlate with the intensity of the scatter radiation emanating from a single, straight X-ray beam, namely the X-ray beam associated with the X-ray image point under consideration. The single scatter radiation image therefore describes the distribution of scatter radiation for the individual X-ray image point associated with it.

[0033] The image values ​​of scatter radiation image points preferably correlate exclusively with the intensity of the scatter radiation that, originating from the primary radiation associated with the respective X-ray image point under consideration, strikes the detector area associated with that scatter radiation image point. This means, in particular, that the image values ​​of the scatter radiation image point in the individual scatter radiation image do not depend on or correlate with any other primary or scatter radiation. Only the scatter radiation emanating from the primary radiation or X-ray beam associated with the corresponding X-ray image point influences the image values ​​of the scatter radiation image points in the individual scatter radiation image.

[0034] The acquired individual scatter radiation images can be used in various ways. In one embodiment of the invention, the individual scatter radiation images are provided, displayed, output, and / or transmitted. Optimizing the X-ray image based on the individual scatter radiation images is particularly preferred.

[0035] By generating an individual scatter radiation image for each X-ray image point under consideration, the determination of scatter radiation contained in an X-ray image is simplified. The computational effort required to determine the scatter radiation is reduced, while simultaneously improving the accuracy of the scatter radiation calculation. The resulting estimate of scatter radiation can be used, for example, to optimize the X-ray image, particularly to correct imaging artifacts or improve image contrast. Generating individual scatter radiation images requires less computational effort than generating an overall estimate of scatter radiation.

[0036] As will be explained in more detail below, the acquisition of individual scatter radiation images allows the scatter radiation to be used to adjust the display of image values ​​in medical images, thus creating additional evaluation possibilities. In particular, the image values ​​of X-ray image points can be adjusted, thereby creating additional evaluation options for X-ray images. Specifically, the contrast of an X-ray image can be adjusted, or the material properties of examination objects can be determined more precisely. The same advantage can be achieved when the information contained in the individual scatter radiation images is used to adjust the display of cross-sectional images.

[0037] In a preferred embodiment, the method may include a further step. This further step comprises determining, for at least one X-ray image point and based on the individual scattered radiation images, an overall absorption characteristic, in particular an overall absorption coefficient. The overall absorption characteristic can be a line integral that describes the total absorption of the primary radiation along the entire path from the X-ray source to the detector area of ​​the X-ray detector associated with the respective X-ray image point. The overall absorption coefficient describes the total absorption of the primary radiation at a certain point on the object under investigation. In other words, the overall absorption coefficient can be considered a specific form of an overall absorption characteristic.

[0038] The overall absorption coefficient, and in particular the total absorption coefficient, is influenced, as already indicated above, at least by the interaction of the primary radiation with the object under investigation and by the scattering of the primary radiation. Various scattering effects, such as classical scattering and / or Compton scattering, can be taken into account.

[0039] The determination of the total absorption coefficient(s) is therefore primarily based on the individual scatter radiation images obtained. For example, the information contained in the individual scatter radiation images can be rearranged to determine the proportion of the intensity of an X-ray image point caused by scatter radiation. This determined scatter radiation component incident on an X-ray image point can then be subtracted from the total intensity to obtain the portion of the intensity corresponding to the primary radiation associated with that X-ray image point. When evaluating CT scans, this rearrangement of the radiation components can be performed particularly in the projection area.

[0040] The obtained intensity value can be converted into an absorption coefficient. During the evaluation of a CT scan, the absorption coefficient(s) can be determined, particularly in a cross-sectional area that has been reconstructed based on the projection area. For this purpose, the Beer-Lambert law, explained above, can be used, for example. The initial intensity can be determined. The determined intensity of the primary radiation can be divided by the initial intensity. The logarithm of the result can be calculated, and subsequently, a spatially dependent overall absorption coefficient can be determined through reconstruction.

[0041] In other words, for a given point in the X-ray image, the proportion of incident radiation attributable to scattered radiation from other points in the image can be determined. This scattered radiation, originating from primary radiation sources other than that associated with the X-ray point, can be neglected when calculating the intensity of that point. In other words, the total absorption coefficient indicates the attenuation of the X-ray beam associated with the point in question—that is, the primary radiation associated with that point—caused by interaction with the object under investigation and / or by scattering of the X-ray beam.

[0042] Optionally, a customized medical image, in particular a customized X-ray image or a customized projection image, can be generated based on the determined total absorption parameters, especially the determined total absorption coefficient. For example, the intensity of the X-ray image points in a customized X-ray image can correlate with the total absorption parameter. Alternatively or additionally, the intensity of the image points in a cross-sectional image can correlate with the total absorption coefficient at the respective location of the image point. In particular, the intensity of the image points of the customized medical image can correlate solely with the total absorption parameter corresponding to the image point; that is, no other technical information is incorporated into the intensity value of the X-ray image points. Alternatively or additionally, an X-ray attenuation image based on the total absorption parameters can be generated.

[0043] The resulting adapted medical image, based on total absorption parameters, exhibits altered contrast ratios compared to the original medical image. These altered contrast ratios, in particular, enable a reduction in scatter radiation artifacts and other image defects caused by scatter radiation, such as blurring or insufficient contrast.

[0044] In a preferred embodiment, the method further comprises the step of determining an overall scattered radiation image based on the individual scattered radiation images of the respective X-ray image points.

[0045] The overall scatter radiation image contains a multitude of pixels, where the image values ​​of these pixels correspond to the intensity of the scatter radiation that strikes a detector area of ​​the X-ray detector associated with that pixel. In particular, the source of the scatter radiation is irrelevant. The overall scatter radiation image thus describes how the scatter radiation is distributed throughout the X-ray image.

[0046] The overall scatter radiation image can be generated, for example, by superposition, i.e., by combining the individual scatter radiation images. However, other methods for determining an overall scatter radiation image are also conceivable; for example, a weighted sum of the individual radiation images can be calculated.

[0047] An adapted medical image can be determined based on the total scatter radiation image. The total scatter radiation image can be used in determining the adapted medical image in such a way that the adapted medical image is essentially free of scatter radiation. Specifically, the adapted medical image can be obtained by calculating the difference between the image values ​​of the original medical image and the total scatter radiation image. By considering the total scatter radiation when determining the adapted medical image, scatter radiation artifacts can be efficiently reduced, resulting in improved image quality.

[0048] In a preferred embodiment, the method can include the step of determining, for at least one X-ray image point and based on the individual scattered radiation images, a photoelectric absorption characteristic, in particular a photoelectric absorption coefficient. The photoelectric absorption characteristic describes the absorption of the primary radiation by photoelectric absorption on the path from the X-ray source to the detector area of ​​the X-ray detector associated with the respective X-ray image point. If the photoelectric absorption characteristic is a line integral, it describes the photoelectric absorption over the entire path from the X-ray source to the X-ray detector. If the photoelectric absorption characteristic is a photoelectric absorption coefficient, it describes the absorption of the primary radiation by photoelectric absorption at a certain location on the object under investigation.In other words, the information contained in the individual scattered radiation images about the attenuation of an X-ray beam's intensity due to scattering effects is recycled, and the intensity "lost" through scattering is added to the intensity of the primary radiation. This "recycling" process thus eliminates the attenuation of intensity due to scattered radiation, so that only the attenuation of the X-ray beam due to photoelectric absorption is considered.

[0049] In particular, the information about the scattering of an X-ray beam contained in the individual scatter radiation images can be redistributed such that the energy contained in the scattered radiation of the X-ray beam is added to the primary energy of the X-ray beam. In one embodiment, the sum of the intensities of the scatter radiation image points is calculated, and this sum is added to the intensity of the primary radiation. This approach requires little computational effort; however, due to the simple calculations, not all radiation effects can be represented. For example, Compton radiation or the attenuation of the scattered radiation in the object under investigation can only be represented to a limited extent.

[0050] In another embodiment, a weighted sum of the intensities of the scattered radiation image points is therefore additionally or alternatively calculated. Preferably, the intensity of photons that strike the detector further away from the X-ray image point under consideration is weighted more heavily, for example, because they are partially absorbed after scattering as they pass through the object under investigation, or because part of the photon energy is absorbed by (further) Compton scattering.

[0051] Determining the photoelectric absorption coefficient, in particular the photoelectric absorption coefficient, can be performed before, after, or simultaneously with determining the total absorption coefficient. Similar to determining the total absorption coefficient, the line integral describing the photoelectric absorption coefficient in the projection area can be calculated when evaluating a CT scan. The photoelectric absorption coefficients at a specific location on the object under investigation, especially for a particular pixel in a projection image, can be obtained by reconstruction based on the information in the projection area.

[0052] Optionally, a customized medical image can be determined based on the individual scatter radiation images or on the photoelectric absorption coefficients. The image values ​​of the pixels in the customized medical image can be based on the intensity of the primary radiation incident on the corresponding detector area as well as on the intensity of the scatter radiation emitted by the primary radiation. Alternatively or additionally, the image values ​​of the medical image can be based on the photoelectric absorption coefficient. In particular, the image values ​​can be based solely on the photoelectric absorption coefficient. The intensity of the X-ray beam can be expressed as I = I0e Σxµphoto(x)x expressed where I0 is the initial intensity of the X-ray beam and µ photo (x) is the photoelectric absorption coefficient at position x.

[0053] In other words, the X-ray beam associated with a pixel is attenuated on its way to the detector both by interaction with the object being examined and by scatter radiation. The attenuation of the X-ray beam by scatter radiation correlates directly with the information depicted in the individual scatter radiation image of the corresponding pixel. Thus, the primary radiation of the X-ray beam belonging to the pixel, deflected by scattering, can be re-linked to the pixel. In particular, the image value and / or intensity of a pixel in the adjusted medical image can be modified so that the image value and / or intensity correlates with the primary radiation associated with the pixel as well as with the scatter radiation emitted by the primary radiation associated with the pixel.

[0054] The resulting modified medical image, which is based primarily on the difference between the total absorption coefficient and the scatter absorption coefficient, exhibits altered contrast characteristics compared to the original X-ray image. These altered contrast characteristics allow for improved detection and differentiation of objects or parts of objects with high CT values, such as bones or contrast agents.

[0055] As described above, the photoelectric absorption coefficient µ photo (x) determined based on the total attenuation and the attenuation due to scattered radiation, in particular by subtracting the attenuation of the scattered radiation from the total attenuation or by adding the corresponding intensities.

[0056] In a preferred embodiment, the method therefore further comprises determining, based on the determined individual scatter radiation images, the determined total absorption coefficient and / or the determined photoeffect absorption coefficient, a scattering absorption characteristic, in particular a scattering absorption coefficient, for the primary radiation.

[0057] If the scattering absorption coefficient is expressed as a line integral, the line integral describes what fraction of the total X-ray radiation emitted by the X-ray source was scattered on its way from the X-ray source to the detector area associated with the image point, such that the scattered fraction does not reach the associated detector area. The scattering absorption coefficient of an image point under consideration, in particular an image point of a cross-sectional image, describes what fraction of the X-ray radiation emitted by the X-ray source was scattered at the location of the image point, such that the scattered fraction does not reach the detector area associated with the corresponding X-ray beam.In the underlying X-ray image, this portion of the X-ray radiation does not contribute to the image value of the X-ray image point; in particular, the intensity of the X-ray image point in the original X-ray image does not reflect the intensity of the associated deflected scattered radiation.

[0058] A corresponding scattering absorption coefficient can be determined for all X-ray image points under investigation, i.e., for all X-ray image points for which an individual scattered radiation image is determined. However, this is not strictly necessary, so a scattering absorption coefficient can also be determined for additional or fewer X-ray image points. The same applies to the photoabsorption coefficient and the total absorption coefficient.

[0059] In a CT scan, the scattering absorption coefficient can be determined in the projection area or, after reconstruction, in the cross-sectional area. In particular, the scattering absorption coefficient can be calculated directly based on the line integrals.

[0060] Alternatively, a reconstruction can first be performed, after which the correlation µ described above can be used to determine the location-dependent dispersion absorption coefficient. photo = µ tot - µ sc can be used and the scattering absorption coefficient of an X-ray beam can be based on µ photo and µ tot can be determined, for example, by µ sc = µ tot - µ photo is determined.

[0061] Scattered radiation on the way to an X-ray image point, which does not find its way into the image value of the corresponding X-ray image point, can be interpreted as misdirected primary radiation, which can allow more precise conclusions to be drawn about the nature of the object under investigation.

[0062] In a further preferred embodiment, the method therefore also includes the step of determining at least one adapted medical image based on the determined scattering absorption coefficients.

[0063] After estimating the scattered radiation and determining a scattered radiation absorption parameter, in particular a scattered radiation absorption coefficient, a customized medical image can be generated. In this customized medical image, especially in a customized cross-sectional image, the scattered radiation absorption coefficient corresponding to a pixel can be incorporated into the image values ​​of the respective pixels. This allows the image information contained in the scattered radiation to be assigned to the pixel whose primary radiation is the origin of the scattered radiation.

[0064] For example, the intensity of the pixels in the adjusted medical image can correlate with the scatter radiation caused by the primary radiation associated with the X-ray image pixel. Specifically, the intensity of the pixels in the adjusted medical image can correlate solely with the scatter radiation associated with the pixel; that is, no other technical information is incorporated into the pixel intensity value. The intensity of the pixels in the adjusted medical image can also or alternatively correlate with the respective scattering absorption coefficients.

[0065] The resulting adapted medical image exhibits altered contrast ratios compared to conventional medical images. These altered contrast ratios enable, in particular, improved detection and differentiation of materials with low CT values / CT numbers / Hounsfield values, such as soft tissue.

[0066] Preferably, the method further comprises the step of determining another adapted medical image, wherein the adapted medical images are determined based on different absorption characteristics or absorption coefficients and / or are determined based on the individual scattered radiation images, so that the adapted medical images represent different intensity and / or absorption values.

[0067] This allows for the provision of multiple customized medical images, each containing different image information. For example, one customized medical image can be determined based on, and in particular solely based on, the dispersion absorption coefficient. Another customized medical image can be determined based on, and in particular solely based on, the photoelectric absorption coefficient. This provides medical professionals with different viewing modes for examining a patient without requiring multiple X-ray or CT scans of the same patient with different settings. When evaluating the medical images, the medical professional can compare different viewing modes, thus simplifying the identification and / or classification of patients.

[0068] The estimation algorithm can be designed in different ways. In one embodiment of the invention, the estimation algorithm comprises scatter kernel superposition and / or a Monte Carlo simulation.

[0069] In an additional or alternative embodiment, the estimation algorithm comprises or is a trained model. The model is trained to determine, based on an X-ray image with a plurality of X-ray image points, for each X-ray image point under consideration, a single scatter radiation image with a plurality of scatter radiation image points associated with that point. The image values ​​of the scatter radiation image points correlate, preferably exclusively, with the scatter radiation that is scattered onto the detector area associated with the respective X-ray image point, originating from the primary radiation incident on that detector area.

[0070] In general, a trained model mimics the cognitive functions that people associate with the thought processes of others. Through training based on training data, the trained model is particularly able to adapt to new circumstances and to recognize and extrapolate patterns. Another term for "trained model" is "trained function."

[0071] Although trained models can learn complex relationships, the application of even complex trained models is possible, at least approximately, in real time. During training, the trained model can, for example, learn to provide results that essentially correspond to the results of a more complex computational method used during training, such as a solution of the Boltzmann transport equation or a Monte Carlo method. Thus, using a trained model can achieve an estimation of similar accuracy to that obtained through considerably more complex calculations. In this way, despite good estimation quality, a quasi-real-time capability for estimating scattered radiation and, in particular, for correcting the X-ray image based on this estimation can be achieved.Thus, the described procedure can, for example, significantly improve the quality of scatter radiation correction in interventional imaging.

[0072] The estimation algorithm is physically motivated by the determination of separate individual scatter radiation images for different X-ray image points to be considered, which allows intermediate results of the estimation algorithm, such as the determined individual scatter radiation images, to be used during training and / or directly during the application of the estimation algorithm in order to, for example, recognize and discard physically unrealistic results or to suppress the generation of such results during training.

[0073] The trained model can be based on a machine learning model. In general, the parameters of a machine learning model can be adjusted through training to provide the trained model. The training can, in particular, be performed upstream of the method according to the invention and thus not be part of the method itself. Alternatively, however, it would also be possible to perform the training as additional upstream process steps within the method according to the invention.

[0074] In particular, supervised training, semi-supervised training, unsupervised training, reinforcement learning, and / or active learning can be used. Furthermore, representational learning, also known as feature learning, can be employed. Specifically, the parameters of the machine learning models can be iteratively adjusted through multiple training steps. In particular, a specific cost function can be minimized during training. Specifically, the backpropagation algorithm can be used when training, for example, a neural network.

[0075] A machine learning model can, in particular, comprise a neural network, a support vector machine, a decision tree, a Bayesian network, and / or a transformer, and / or the machine learning model can be based on k-means clustering, Q-learning, genetic algorithms, and / or association rules. A neural network can, in particular, be a deep neural network, a convolutional neural network, or a convolutional deep neural network. Furthermore, a neural network can be an adversarial network, a deep adversarial network, and / or a generative adversarial network. The machine learning model can have a U-Net architecture and / or a vision transformer architecture.

[0076] In a preferred embodiment, the method further comprises the step of compressing the determined individual scatter radiation images by approximating a parametric function or a parametric curve to the image values ​​of the scatter radiation image points.

[0077] The parametric curve can be, for example, B-splines, a polynomial, or another basis function. The scattered radiation depicted by the individual scattered radiation images has a low frequency, so that only a small amount of information is lost when approximating it with a parametric function. At the same time, the storage space required for the individual scattered radiation images is reduced, and the processing of the scattered radiation images is made more efficient.

[0078] When determining individual scattering patterns, only first-order scattering can be considered. This results in fewer calculations being required by the estimation algorithm or when generating the training data for a trained model. Consequently, the method requires less computing power during execution.

[0079] When determining the individual scattering images, scattering of a higher order than first order can also be taken into account. Likewise, when training the trained model, scattering of a higher order than first order can be considered. In this case, the underlying physical relationships are represented more accurately. This leads to an increase in the accuracy of the generated individual scattering images.

[0080] Another aspect of the invention relates to a computer-implemented method for training at least one trained model. The trained model is, for example, intended for use in an estimation algorithm in the computer-implemented method according to one of the preceding embodiments.

[0081] The process for training the model includes the step of obtaining training data, where the training data comprises several training datasets of source data and target results. The source data each consists of an X-ray image with a multitude of X-ray image points. The respective target result comprises, for each X-ray image point to be considered, a single scatter radiation image assigned to that individual X-ray image point. The single scatter radiation image is to be provided when the trained model processes the source data of the respective training dataset. Image values ​​of scatter radiation image points in the single scatter radiation image correlate with the scatter radiation that is scattered from the primary radiation incident on a detector area assigned to the respective X-ray image point to that detector area.

[0082] The procedure also includes training the trained model through supervised learning based on the training data sets.

[0083] The procedure also includes providing the trained model.

[0084] The trained model can therefore be based on supervised training with training datasets, each comprising source data and several target results. The source data can serve as input data for the estimation algorithm directly or after preprocessing. The corresponding target result, in particular a single scattered radiation image, should be provided when the respective source data is processed by the estimation algorithm. At least the target results can be based on a simulation of X-ray imaging, especially the scattering of X-rays.

[0085] The simulation of X-ray imaging or the scattering of X-rays can be a preliminary step in the process for training a trained model and / or the process for determining individual scattered radiation images. Such a simulation of X-ray imaging can therefore constitute an additional preliminary step in the procedures described here.

[0086] X-ray imaging and / or the scattering of X-rays can be simulated, for example, by solving the Boltzmann transport equation, using a Monte Carlo method, and / or raycasting. The simulation can utilize known parameters of a simulated X-ray imaging system and a three-dimensional model of the object or patient to be imaged. This three-dimensional object can, in particular, describe the spatial distribution of different materials or the scattering and interaction properties of the imaged matter. In the simplest case, such a three-dimensional model can be synthetically generated, for example, based on an anatomical atlas and known scattering and interaction properties of different body components, or the model or parts of the model can be generated manually.

[0087] In particular, the source data of the respective training dataset, such as an X-ray image, can also be generated, at least partially, within such a simulation. Additionally or alternatively, X-ray images based on actual X-ray imaging can be used as source data or as the basis for source data for at least some or all of the training datasets. In this case, the three-dimensional model of the imaged object or patient can be generated based on the image data from the same X-ray imaging in order to determine, as explained above, the individual scatter radiation images associated with the respective X-ray image as target results through simulation of the X-ray imaging.

[0088] For example, the X-ray image of the respective training dataset can be acquired as part of a computed tomography scan, meaning that the X-ray imaging data describes a three-dimensional representation of the object or patient. By classifying different image areas, for example using a trained algorithm or by employing an anatomical atlas to which the three-dimensional representation is linked, the different areas of the three-dimensional image can then be classified, for example, as tissue, bone, etc., thus allowing the interaction and scattering properties of the respective imaged material, which are at least approximately known, to be assigned to them.For the sake of completeness, it should be noted that in principle the interaction and scattering properties of a three-dimensional object can also be estimated from individual X-ray images, whereby additional information about a known imaging geometry and the object can be used in particular.

[0089] In a preferred embodiment, the respective target result can further comprise a composite scatter radiation image, which is determined based on the individual scatter radiation images assigned to the respective X-ray image points. Including a composite scatter radiation image in the training data, in addition to the individual scatter radiation images provided as the target result, improves the training of the trained model because more information is available for evaluating the training results. In particular, the trained model can provide more accurate and consistent results.

[0090] Preferably, training the trained model includes optimizing at least one cost function, wherein the value of the cost function depends on the image values ​​of the scattered radiation image points and / or on the image values ​​of the image points of the overall scattered radiation image. Particularly preferably, training the trained model includes optimizing a plurality of cost functions, wherein a cost function is determined and optimized for each X-ray image point to be considered, i.e., for each individual scattered radiation image to be generated. Additionally, a further cost function can be determined for an overall scattered radiation image. As mentioned above, the overall scattered radiation image can be based on the individual scattered radiation images, or it can be generated independently of the individual scattered radiation images by the trained model.In one example, for nxm X-ray image points to be considered, cost functions nxm + 1 are optimized as part of the training of the trained model.

[0091] Additionally or alternatively, the optimization can be performed under a constraint that evaluates the image values ​​of the scatter radiation image points and / or the image values ​​of the image points of the overall scatter radiation image.

[0092] Further features, which are explained in the context of the computer-implemented method according to the invention for determining individual scatter radiation images for X-ray image points to be considered, can also be transferred to the computer-implemented method for training at least one trained model with the advantages mentioned therein, and vice versa.

[0093] Furthermore, the invention relates to a data processing device configured for carrying out the computer-implemented method according to one of the embodiments described above. The data processing device can, for example, be designed as a suitably programmed data processing device, or the aforementioned functionality can be implemented, at least in part, in hardware. The data processing device can be integrated into a medical imaging device, in particular an X-ray device, or be designed separately from it. It can, for example, be implemented as a workstation, server, or cloud solution.

[0094] Furthermore, the invention relates to an X-ray device comprising an X-ray source and an imaging X-ray detector, wherein the X-ray device includes a data processing device according to the invention. By integrating a data processing device according to the invention, and thus by implementing the method according to the invention in an X-ray device, the described scatter radiation estimation or a scatter radiation correction based thereon can be performed directly during data acquisition or visualization by the X-ray device for a user.

[0095] The invention also relates to a computer program with instructions designed to carry out the computer-implemented method according to the invention when executed on a data processing device.

[0096] Furthermore, the invention relates to a computer-readable data carrier with instructions that are configured to execute the computer-implemented method when implemented by a data processing device.

[0097] Features discussed in the context of one of the methods or devices according to the invention can also be transferred to the other open subject matter of the invention with the aforementioned advantages.

[0098] Further advantages and details of the invention will become apparent from the following exemplary embodiments and the accompanying drawings.

[0099] This schematically illustrates: Fig. 1 an embodiment of an imaging system comprising an embodiment of a processing device, Fig. 2a a visualization of the scattered radiation emitted by an X-ray beam, Fig. 2b a visualization of the scattered radiation emitted by another X-ray beam, Fig. 3. A flowchart of a computer-implemented procedure for training a trained model and a flowchart of a computer-implemented procedure for determining single-scatter radiation images. Fig. 4a a schematic representation of an adapted X-ray image based on the photoelectric absorption coefficient, Fig. 4b a schematic representation of an adapted X-ray image based on the scattering absorption coefficient, and Fig. 4c a schematic representation of an adapted X-ray image based on the total absorption coefficient.

[0100] The exemplary embodiments described in more detail below represent preferred embodiments of the present invention.

[0101] In Fig. Figure 1 schematically depicts an exemplary embodiment of an imaging system 1 according to the improved concept, which is designed, for example, as an X-ray imaging system. In the example of the Fig. Figure 1 shows a construction of the X-ray imaging system or device based on the principle of a C-arm device with a rotatable and movable C-arm 6, which can be rotated and moved accordingly to image an object 4 to be imaged or an examination object 4 from different directions, i.e. with different recording angles.

[0102] An imaging device 1 according to the improved concept can also be built according to other designs. In particular, the improved concept is not fundamentally limited to X-ray-based imaging methods.

[0103] The imaging device 1 of the Fig. The imaging system 1 includes, for example, an X-ray source 2, which is configured to generate X-rays and emit them towards the object under investigation 4. On the opposite side of the object under investigation 4 from the X-ray source 2, an X-ray detector 3 of the imaging system 1 is arranged. This detector 3 contains, for example, a detector array of photodiodes to detect X-ray quanta penetrating the object under investigation 4. The detector 3 can then transmit the corresponding detector signals, for example, to a control or processing unit 5 of the imaging system 1 for further processing. An image processing device 9 can receive captured images from the control or processing unit 5. The image processing device 9 can then make any necessary corrections to the captured images.

[0104] Imaging system 1 can be configured, for example, to perform a rotational angiography procedure, such as one based on the principle of subtraction angiography. In this case, processing unit 5 can, for example, generate a large number of two-dimensional projections (also called source images) acquired from different angles, and processing unit 5 can then calculate a three-dimensional reconstruction from these projections.

[0105] The X-ray radiation, attenuated by the matter of the object of investigation 4, here a patient, thus strikes the X-ray detector 3, which comprises several detector areas 12, whereby the X-ray intensity or X-ray dose striking the respective detector area 12 is recorded in order to specify a respective image value of a respective X-ray image point of an X-ray image.

[0106] Since the interaction of the X-rays with the material of the object under investigation 4 is not limited to pure absorption of the X-rays, but also involves scattering, scattered radiation also reaches the X-ray detector 3, which can interfere with the imaging. In addition to or as an alternative to reducing the detected scattered radiation by other methods, such as using anti-scatter gratings, it is therefore advantageous to estimate the scattered radiation distribution in the X-ray image. This is particularly useful in order to subsequently correct the X-ray image based on the estimated distribution, for example, by subtracting the estimated distribution or by scaling or iteratively multiplying the detected X-ray intensities based on the estimated distribution, and thus provide an adapted X-ray image.

[0107] Therefore, the processing unit 9, which in this example is integrated into the imaging system 1 but can in principle also be designed separately from it, for example as a workstation, server, or cloud solution, implements a method for determining individual scattered radiation images. An exemplary configuration of such a method will be described below with reference to Fig. 3 will be explained.

[0108] First, however, the following will be used as a guide Fig. 2a and Fig. 2b briefly explains the principle underlying the invention. As already mentioned in connection with Fig. As explained in Figure 1, the imaging system 1 comprises an X-ray source 2 by means of which X-rays are emitted onto a test object 4. Behind the test object 4, viewed from the X-ray source 2, a detector 3 is arranged, which has several detector areas 12. The intensity of the X-rays measured at a specific detector area 12.x comprises two essential components: intensity 14 due to primary radiation PR and intensity 16 due to scattered radiation SR. Thus, primary radiation PR is measured at the specific detector area 12.x, which, after passing through the test object 4, reaches the detector 3 (see Figure 1). Fig. 2a).

[0109] The primary radiation PR can be approximated as a rectilinear X-ray beam, which is emitted from the X-ray source with a certain initial intensity and is attenuated on its way to the detector 3. This attenuation occurs firstly through photoelectric absorption of the energy by the object under investigation 4. Secondly, attenuation occurs through scattering of the X-ray beam, whereby a portion of the energy of the X-ray beam is deflected from its rectilinear path due to the scattered radiation. In the Fig. 2a and Fig. 2b shows the scattered radiation represented by the dashed lines. Consequently, in the specific detector range 12.x, in addition to the primary radiation PR, the scattered radiation SR, which is thrown onto the specific detector range 12.x from other X-rays, is measured (see Fig. 2b).

[0110] The scatter radiation SR emanating from a primary radiation PR or an X-ray beam can therefore be considered as deflected primary radiation PR, which contains information about the structure of the object under investigation, but does not strike the detector area 12 that actually belongs to the primary radiation PR. Accordingly, it is worthwhile to determine the scatter radiation emanating from the associated X-ray beams for individual detector areas 12 or for the corresponding X-ray image points of an X-ray image in order to utilize the information contained in the scatter radiation distribution for the optimization or adaptation of the X-ray image.

[0111] The previously mentioned method for determining individual scatter radiation images is implemented by a computer program, which is stored, for example, in the memory of the image processing device 9 and is executed by its freely programmable processor.

[0112] Fig. Figure 3 presents a flowchart of an exemplary computer-implemented procedure for training a trained model and a flowchart of an exemplary computer-implemented procedure for determining single scatter radiation images.

[0113] In step S1, an X-ray image is obtained. As described above, the X-ray image is based on the X-ray radiation incident on an X-ray detector 3. The X-ray image has a multitude of X-ray image points, each of which is assigned to a detector area 12 of the X-ray detector.

[0114] In step S2, an estimation algorithm is provided, which is designed to determine individual scatter radiation images based on the obtained X-ray image. A separate individual scatter radiation image is generated for each X-ray image point to be considered. Consequently, if there are n × m × z X-ray image points to be considered, n × m × z individual scatter radiation images are generated. As explained above, an individual scatter radiation image for a given X-ray image point describes the distribution of the scatter radiation SR emanating from an X-ray beam whose primary radiation PR strikes the X-ray image point.

[0115] The single scatter radiation image exhibits a multitude of scatter radiation image points, each assigned to detector regions 12. The image values ​​of the scatter radiation image points correspond to the intensity of the scatter radiation SR, which originates from the X-ray beam under consideration and strikes the respective detector region 12.

[0116] In step S3, the provided estimation algorithm is applied to the provided X-ray image to determine a large number of individual scatter radiation images.

[0117] Optionally, in step S4, a total absorption coefficient µ can be determined for at least one X-ray image point. tot The total absorption coefficient µ can be determined. tot For example, in the case of a CT scan, it can be determined directly based on the scatter radiation images, whereby, optionally, a reconstruction can then be carried out so that the determined total absorption coefficient µ totpixels of cross-sectional images can be assigned. In particular, the total absorption coefficient µ can be used. tot To describe the absorption that an X-ray beam experiences at a specific point on its path from the X-ray source 2 to the X-ray detector 3. Each pixel of a projection image can be assigned a specific total absorption coefficient µ. tot be assigned. Additionally or alternatively, an X-ray beam can be assigned a total absorption coefficient, particularly an average one. To determine the total absorption coefficient µ tot Based on the individual scatter radiation images, it can be determined what proportion of the total intensity of an X-ray image point is due to the primary radiation PR of the associated X-ray beam and what proportion of the total intensity of the X-ray image point is caused by scatter radiation SR emanating from other X-ray beams.

[0118] Optionally, a photoelectric absorption coefficient µ can be determined in step S6. photo can be determined. Similar to the total absorption coefficient µ. tot can the photoelectric absorption coefficient µ photo can be determined directly based on the individual scattered radiation images, whereby, optionally, a reconstruction can then be carried out so that the determined photoelectric absorption coefficients µ photo pixels of cross-sectional images can be assigned. The photoelectric absorption coefficient µ photo This describes how much an X-ray beam would be attenuated if only the photoelectric absorption of the X-ray beam were considered, but not the energy loss due to scattering effects. The determination of the photoelectric absorption coefficient µ photo This is done based on the individual scatter radiation images and, optionally, based on the total absorption coefficient µ. totBy analyzing the information contained in the individual scatter radiation images about the intensity and distribution of the scatter radiation, it is possible to determine from the individual scatter radiation images how much energy of the associated X-ray beam was deflected by scattering effects.

[0119] Optionally, a scattering absorption coefficient µ can be determined in step S7. Sc The scattering absorption coefficient µ can be determined. Sc can be based in particular on the previously determined total absorption coefficient µ tot and the photoelectric absorption coefficient µ photo can be determined. In particular, the scattering absorption coefficient µ can be determined. Sc by subtracting the previously determined photoelectric absorption coefficient µ photo of the total absorption coefficient µ tot This can be done in the cross-sectional area, i.e., after a reconstruction step. Alternatively or additionally, the scattering absorption coefficient µ can be used.Sc However, it can also be determined based on the individual scattered radiation images. If the total absorption coefficient µ has already been determined, tot and the photoelectric absorption coefficient µ photo Once determined, the scattering absorption coefficient µ can be calculated. Sc simply by forming the difference from the photoelectric absorption coefficient µ photo and the total absorption coefficient µ tot to be determined. The calculation of the scattering absorption coefficient µ Sc This can generally be done before or after the reconstruction of cross-sectional images from the X-ray images, although reconstruction is preferably performed first. In other words, the scattering absorption coefficient µ can be Sc in the projection area, whereby a reconstruction of cross-sectional images can then take place, or a reconstruction can be carried out first, and then the scattering radiation coefficient µ. Scbased on the total absorption coefficient µ tot and the photoelectric absorption coefficient µ photo to be determined.

[0120] Optionally, based on at least one of the determined absorption coefficients in step S5, a customized medical image, in particular a customized X-ray image and / or a customized cross-sectional image, can be generated. Preferably, in step S8, another customized medical image, in particular a customized X-ray image and / or a customized cross-sectional image, is generated based on a different one of the determined absorption coefficients, so that the customized medical images are comparable with each other.

[0121] The core of the procedure lies in the use of the estimation algorithm, which preferably includes or is implemented as a trained model. The trained model can be obtained, for example, by a procedure for training a trained model, as described by steps ST1, ST2, and ST3. These steps can be performed as a standalone procedure or as part of the procedure for determining single-scatter radiation images. However, for the procedure for determining single-scatter radiation images, steps ST1, ST2, and ST3 are optional.

[0122] In step ST1, training data is obtained. This preferably includes generating training data. The training data comprises source data and target results, with source data and multiple target results preferably forming a training dataset for each instance. The source data is used as input for the trained model. The source data includes, in particular, an X-ray image, which can be structured as described above. The target results include, in particular, for each pixel of the X-ray image to be considered, an individual scatter radiation image assigned to that pixel, which indicates the distribution of scatter radiation emanating from an X-ray beam assigned to that pixel. Furthermore, the target results preferably include an overall scatter radiation image, which is determined by superimposing the individual scatter radiation images.

[0123] Individual scatter radiation images, and optionally the overall scatter radiation image, can be determined, for example, by simulating the scattering of X-rays during X-ray imaging. Starting with a fictitious object whose material properties are known, it is possible to determine how the X-rays are scattered by the object. Such a simulation of scatter radiation can be performed, for example, using the Monte Carlo method, taking into account first-order and / or higher-order scatter radiation.

[0124] In this way, for a given X-ray image, which in one embodiment could also be a simulated X-ray image, a single scatter radiation image can be simulated for each X-ray image point to be considered, reflecting the distribution of the scatter radiation emanating from an X-ray beam assigned to the X-ray image point to be considered.

[0125] Additionally or alternatively, the single scatter radiation images used in the training datasets could be generated at least partially by applying or based on the slot scanning technique, which has the advantage that the single scatter radiation images are based on real measurements and no model of an object under investigation is required.

[0126] In step ST2, the trained model is trained using a preferably supervised learning method based on the training datasets. In an iterative process, X-ray images are presented to the initially untrained model as source data. Based on this source data, the model generates individual scatter radiation images and, optionally, a composite scatter radiation image. A cost function is provided for each individual scatter radiation image and for the optional composite scatter radiation image. This cost function describes how similar the generated individual scatter radiation images and composite scatter radiation images are to the target results of the training dataset. Based on these cost functions, the model's weights are adjusted, aiming to minimize the cost functions. A backpropagation learning algorithm is therefore preferably used.For example, after a certain number of learning iterations or when the value of the cost functions falls below a certain threshold, the training is complete.

[0127] In step ST3, the trained model is then made available, in particular for use in a method for determining single scatter radiation images, as described above.

[0128] The Fig. 4a, Fig. 4b and Fig. Figure 4c presents a comparison of schematically represented, adapted medical images, specifically adapted cross-sectional images, of the same object 4. The three adapted cross-sectional images were determined based on different absorption coefficients. The object 4, schematically represented in the adapted cross-sectional images, contains various materials, in particular bone 20, a tissue component 22 (e.g., brain), and blood 24. The materials of the object 4 interact differently with the X-ray radiation irradiating them, which is why different absorption coefficients are visible in the cross-sectional image. Different intensities are detected in different areas 12 of the X-ray detector 3. The different materials are characterized by different hatching patterns. As can be seen from a review and comparison of the Fig. Figures 4a to 4c, which are only schematically visualized through the different hatching patterns, show that the medical images exhibit different contrast ratios. These contrast ratios depend primarily on the absorption coefficients used as the basis for the representation.

[0129] The in Fig. The adapted cross-sectional image shown in Figure 4a is based on the photoelectric absorption coefficient µ. photo generated. One based on the photoelectric absorption coefficient µ. photo The determined, adapted cross-sectional image is particularly suitable for differentiating materials such as bone 20 or contrast agent, because good contrast ratios are present in this imaging area.

[0130] The in Fig. The adapted cross-sectional image shown in 4b is based on the scattering absorption coefficient µ. Scgenerated. The figure shows that when generating a fitted cross-sectional image based on the scattering absorption coefficient µ, Sc The contrast between different soft fabric materials is particularly pronounced, making these materials easily distinguishable.

[0131] The in Fig. The adapted cross-sectional image shown in 4c is based on the total absorption coefficient µ. tot generated. Such an adapted cross-sectional image can be used particularly because scatter radiation artifacts are eliminated, resulting in very few image errors.

[0132] The various illustrated logic blocks, modules, circuits, and algorithm steps described in connection with the disclosed embodiments can be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above in general terms with respect to their functionality. Whether this functionality is implemented as hardware or software depends on the specific application imposed on the overall system. Those skilled in the art may implement the described functionality in various ways for any given application, but such implementation decisions should not be construed as a deviation from the scope of this disclosure or the claims.

[0133] Implemented software versions can be written in software, firmware, middleware, microcode, hardware description languages, or any combination thereof. A code segment or machine-readable instruction can represent a procedure, function, subroutine, program, routine, subroutine, module, software package, class, or any combination of instructions, data structures, or program statements. A code segment can be coupled to another code segment or a hardware circuit by passing and / or receiving information, data, arguments, parameters, or memory contents.

[0134] Information, arguments, parameters, data, etc., can be transmitted, forwarded, or transferred, for example, by memory release, message forwarding, token forwarding, network transfer, etc.

[0135] The actual software code or specialized control hardware used to implement these systems and methods is not limiting to the claimed features or this disclosure. Therefore, the operation and behavior of the systems and methods have been described without reference to the specific software code, with the understanding that software and control hardware can be developed to implement the systems and methods based on the description herein.

[0136] When implemented as software, the functions can be stored as one or more instructions or code on a non-volatile, computer-readable or processor-readable storage medium. The steps of a method or algorithm disclosed herein can be embodied in a processor-executable software module, which may reside on a computer-readable or processor-readable storage medium. A non-volatile, computer-readable or processor-readable storage medium includes both computer storage media and tangible storage media that facilitate the transfer of a computer program from one location to another. A non-volatile, processor-readable storage medium can be any available medium accessible to a computer.By way of example, but not limited to, such non-volatile, processor-readable media may include RAM, ROM, EEPROM, CD-ROM or other optical media, magnetic media or other magnetic data storage devices, or any other tangible storage medium capable of storing desired program code in the form of instructions or data structures and accessible to a computer or processor. Disk / Disc, as used here, includes Compact Disc (CD), Laser Disc, Optical Disc, Digital Versatile Disc (DVD), Floppy Disc, and Blu-ray Disc, where disks usually reproduce data magnetically and discs optically using lasers. Combinations of the above media are also included under computer-readable media.Additionally, the operations of a procedure or algorithm can be represented as one or any combination or series of codes and / or instructions on a non-volatile processor-readable and / or computer-readable medium, which may be integrated into a computer program product.

[0137] The foregoing description is intended to enable a person skilled in the art to manufacture and / or use the embodiments and variants thereof described herein. Various modifications of these embodiments will be immediately apparent to those skilled in the art, and the principles defined herein can be applied to other embodiments without departing from the spirit or scope of the subject matter disclosed herein. Therefore, this disclosure is not intended to be limited to the embodiments shown herein, but rather to the broadest possible scope in accordance with the following claims and the principles and novel features disclosed herein.

[0138] While various aspects and embodiments have been disclosed, other aspects and embodiments are also conceived. The various disclosed aspects and embodiments serve only for illustration and are not to be considered limiting, the true scope and spirit being defined by the following claims.

Claims

[1] Computer-implemented method for determining individual scatter radiation images for X-ray image points to be considered in an X-ray image, comprising the steps: - Obtaining (S1) an X-ray image, wherein the X-ray image is recorded by means of an X-ray detector (3) and comprises a plurality of X-ray image points, each of which is assigned to a detector area (12) of the X-ray detector (3), wherein several or all of the X-ray image points are X-ray image points to be viewed; - Providing (S2) an estimation algorithm which, based on the X-ray image, determines for each of the X-ray image points to be considered a respective individual scatter radiation image assigned to the individual X-ray image point to be considered, wherein the respective individual scatter radiation image has a plurality of scatter radiation image points, each of which is assigned to a detector area (12) of the X-ray detector (3), wherein image values ​​of the scatter radiation image points of the individual scatter radiation image correlate with the intensity of the scatter radiation (SR) which is scattered onto the detector area (12) assigned to the respective scatter radiation image point, starting from the primary radiation (PR) incident on a detector area (12) assigned to the respective X-ray image point to be considered; - Determining (S3) a single scatter radiation image for each of the X-ray image points to be considered by applying the estimation algorithm to the X-ray image. [2] The method of claim 1, wherein the method further comprises the step of: - Determine (S4), for at least one X-ray image point and based on the individual scattered radiation images, at least one overall absorption characteristic, in particular one overall absorption coefficient (µ) tot ), which describes the total absorption of the primary radiation (PR) on the path from the X-ray source (2) to the detector area of ​​the X-ray detector (3) assigned to the respective X-ray image point. [3] Method according to claim 1 or 2, wherein the method further comprises the step: - Determine (S5) at least one adapted medical image based on the determined individual scatter radiation images and / or based on the determined total absorption characteristics, in particular on the determined total radiation coefficient (µ) tot ). [4] A method according to any of the preceding claims, wherein the method further comprises the step of: - Determining an overall scatter radiation image based on the individual scatter radiation images of the respective X-ray image points, whereby, optionally, an adapted medical image is determined based on the overall scatter radiation image. [5] A method according to any one of the preceding claims, wherein the method further comprises: - Determine (S6), for at least one X-ray image point and based on the individual scattered radiation images, at least one photoeffect absorption characteristic, in particular a photoeffect absorption coefficient (µ) photo ), which describes the absorption of the primary radiation (PR) by photoelectric absorption on the path from the X-ray source (2) to the detector area of ​​the X-ray detector (3) assigned to the respective X-ray image point, and, optionally, - Determining at least one adapted X-ray image based on the determined photoeffect absorption parameters, in particular based on the determined photoeffect absorption coefficients (µ). photo ). [6] A method according to any one of the preceding claims, wherein the method further comprises: - Determine (S7), based on the determined individual scatter radiation images, based on the determined total absorption characteristics, in particular the total absorption coefficient (µ) tot ), and / or based on the determined photoelectric absorption parameters, in particular the photoelectric absorption coefficient (µ) photo ), a scattering absorption characteristic, in particular a scattering absorption coefficient (µ) Sc ), for primary radiation (PR), and, optionally, - Determine at least one adapted medical image based on the determined dispersion absorption parameters, in particular based on the dispersion absorption coefficients (µ) Sc ). [7] Method according to any one of claims 3 to 6, wherein the method further comprises the step: - Determining (S8) a further adapted medical image, wherein the adapted medical images are based on different absorption characteristics, in particular based on different absorption coefficients (µ) tot , µ photo , µ Sc ). will be determined. [8] Method according to one of the preceding claims, wherein the estimation algorithm comprises a trained model, wherein the model is trained to determine, based on an X-ray image with a plurality of X-ray image points, for each X-ray image point to be considered, a single scattered radiation image with a plurality of scattered radiation image points assigned to the respective X-ray image point to be considered, wherein image values ​​of the scattered radiation image points correlate with the scattered radiation (SR) which is scattered onto the detector area (12) assigned to the respective scattered radiation image point, starting from the primary radiation (PR) incident on a detector area (12) assigned to the respective X-ray image point to be considered. [9] A method according to any of the preceding claims, wherein the method further comprises the step: - Compressing the determined individual scatter radiation images by approximating the image values ​​of the scatter radiation image points with a parametric function. [10] Method according to one of the preceding claims, wherein when determining the individual scattering radiation images only first-order scattering is taken into account or scattering of a higher order than first order is taken into account. [11] Computer-implemented method for training at least one trained model for use in an estimation algorithm in the computer-implemented method according to any of the preceding claims, wherein the training method comprises the following steps: - Obtaining (ST1) training data, where the training data includes multiple training datasets of source data and target results, wherein the source data each comprise an X-ray image with a multitude of X-ray image points, and wherein the respective target result for each X-ray image point to be considered comprises a single scatter radiation image assigned to the individual X-ray image point to be considered, wherein the single scatter radiation image is to be provided by the trained model when processing the source data of the respective training data set, wherein image values ​​of scatter radiation image points of the single scatter radiation image correlate with the scatter radiation (SR) which is scattered onto the detector area (12) assigned to the respective scatter radiation image point, starting from the primary radiation (PR) incident on a detector area (12) assigned to the respective X-ray image point to be considered, - Training (ST2) the trained model through supervised learning based on the training datasets, - Deploying (ST3) the trained model. [12] Method according to claim 11, wherein the respective target result further comprises an overall scatter radiation image which was determined on the basis of the individual scatter radiation images which are assigned to the respective X-ray image points. [13] Data processing equipment, wherein the data processing equipment comprises means for carrying out the computer-implemented method according to any one of claims 1 to 10. [14] Computer program with instructions configured to carry out the computer-implemented method according to any one of claims 1 to 10 when the computer program is executed by a data processing device. [15] Computer-readable data carrier with instructions configured to carry out the computer-implemented method according to any one of claims 1 to 10 when executed by a data processing device.

Images