providing a differential image data set and providing a training function

Abstract

This invention relates to a computer-implemented method for providing a differential image data set, the method comprising: determining a first set of real image data for an examination volume with respect to a first X-ray energy; and determining a multi-energy set of real image data for an examination volume with respect to the first X-ray energy and a second X-ray energy, wherein the second X-ray energy differs from the first X-ray energy. The method further comprises: determining a differential image data set for an examination volume by applying a training function to input data, wherein the input data is based on the first set of real image data and the multi-energy set of real image data; and providing the differential image data set. The invention also relates to a computer-implemented method for providing a training function, a providing system, a training system, an X-ray system, a computer program product, and a computer-readable storage medium.

Description

[0001] This invention application is a divisional application of the invention patent application filed on January 13, 2020, with application number "202010031634.9" and invention title "Providing Differential Image Data Sets and Providing Training Functions". Background Technology

[0002] In digital subtraction angiography (DSA), one or more blood vessels in an examination volume are recorded using X-rays. To suppress other structures in the examination volume, the recording of the vessels without contrast agent (so-called masked recording) is combined with the recording of the vessels containing contrast agent. Here, contrast agent is introduced into the blood vessels during the examination to determine parameters, particularly the hydrodynamic parameters of the fluid flowing within the vessels.

[0003] In 4D DSA, a time-resolved series of 3D DSA image data is provided using image reconstruction methods. Here, a normalized 2D X-ray projection of the examined volume is back-projected onto the volume element along with temporal information. The 2D X-ray projection is typically derived from the rotational recording of a C-arm X-ray machine.

[0004] Digital subtraction angiography records not only vessels without contrast agent but also vessels containing contrast agent, subjecting the examined volume to a high X-ray load. Recording vessels without contrast agent is also known as masked recording.

[0005] Below, an image data set can be referred to as a true image data set if it depicts the actual distribution of values ​​and / or intensities (e.g., Henle units, X-ray attenuation coefficients) within the examined volume. An image data set can also be referred to as a difference image data set if it depicts the difference between the actual distributions of values ​​and / or intensities within the examined volume. However, a difference image data set is not necessarily determined by subtracting two true image data sets. An image data set can also be referred to as a subtraction image data set if it is determined by subtracting two image data sets, particularly by subtracting two true image data sets. Therefore, each subtraction image data set can be understood as a difference image data set, but not every difference image data set can be understood as a subtraction image data set.

[0006] As known from the unpublished patent application EP18182251, a differential image dataset is determined by applying a training function to a set of real image data without performing additional masking recording. However, because structures in the examination area, such as bone structures, metallic structures (e.g., implants), or calcified structures (calcifications in blood vessels), have X-ray absorption similar to that of contrast agents, such structures in the examination area can lead to errors in determining the differential image dataset. Summary of the Invention

[0007] Therefore, the object of the present invention is to achieve more accurate and less error-insensitive determination of differential image data sets.

[0008] The objective, according to the independent claims, is achieved by a method for providing a differential image data set, a method for providing a training function, a determination system, a training system, a computer program product, and a computer-readable storage medium. Advantageous improvements are given in the dependent claims and in the specification.

[0009] The solution according to the invention described below relates not only to the claimed device but also to the claimed method. The features, advantages, or alternative embodiments mentioned herein can also be applied to other claimed subjects, and vice versa. In other words, the entity claims (for example, those concerning the device) can also be improved by combining the described method or the claimed features. Here, the corresponding functional features of the method are constituted by corresponding entity modules.

[0010] Furthermore, the solution according to the invention for the stated objective relates not only to methods and apparatus for providing differential image datasets, but also to methods and apparatus for providing training functions. Here, features and alternative implementations of data structures and / or functions in the methods and apparatus for providing differential image datasets can be adapted to similar data structures and / or functions in the methods and apparatus for providing training functions. Here, a feature of similar data structures is particularly the use of the prefix "training". Furthermore, the training functions used in the methods and apparatus for providing differential image datasets can be adapted and / or provided by the methods and apparatus for providing training functions.

[0011] In a first aspect, the present invention relates to a computer-implemented method for providing a differential image data set, the method comprising: determining a first set of true image data for an examination volume with respect to a first X-ray energy; and determining a multi-energy set of true image data for an examination volume with respect to the first X-ray energy and a second X-ray energy, wherein the second X-ray energy differs from the first X-ray energy. The method further comprises: determining the differential image data set for the examination volume by applying a training function to input data, wherein the input data is based on the first set of true image data and the multi-energy set of true image data; and providing the differential image data set.

[0012] Here, the first real image data group can be determined by receiving the first real image data group. Furthermore, the multi-functional real image data group can be determined by receiving the multi-functional real image data group. Here, the first real image data group can be determined particularly by means of an interface and / or a computing unit. Here, the differential image data group can be determined particularly by means of a computing unit. Here, the differential image data group can be provided particularly by means of an interface. Here, providing the differential image data group can particularly include storing, transmitting, and / or displaying the differential image data group.

[0013] Here, the first X-ray energy and the second X-ray energy correspond to the accelerating voltage of the X-ray tube or the energy of the X-ray photons. The terms "first X-ray energy" and "second X-ray energy" can also describe, in particular, the first X-ray spectrum and the second X-ray spectrum, where the X-ray spectrum corresponds to the intensity distribution of different wavelengths or energies of the X-ray radiation. The accelerating voltage used to generate the radiation, its energy, or its spectrum corresponding to the X-ray energy, is characterized, in particular, by the X-ray energy.

[0014] The image data set specifically includes multiple pixels or multiple voxels. Here, each pixel or voxel is associated with an intensity value. In the X-ray image data set, each pixel or voxel is particularly associated with an X-ray intensity value, which is a measure of the X-ray intensity incident on the pixel or voxel or the X-ray absorption coefficient of the pixel or voxel. The incident X-ray intensity is related to the number, size, shape, and material of the object within the examination volume and penetrated by the X-ray radiation. The image data set may include other data, particularly metadata about the imaging examination, especially X-ray examination.

[0015] Here, the two-dimensional image data set includes at least one two-dimensional representation of the examined volume. Here, the three-dimensional image data set includes at least one three-dimensional representation of the examined volume, and the three-dimensional image data set may also additionally include one or more two-dimensional representations of the examined volume.

[0016] A set of true image data of an examination volume with respect to X-ray energy can be a set of X-ray image data of the examination volume recorded using X-ray radiation with said X-ray energy. A set of true image data of an examination volume with respect to X-ray energy can also be based on a set of X-ray image data of the examination volume recorded using X-ray radiation with said X-ray energy. A set of multi-energy true image data of an examination volume with respect to a first X-ray energy and a second X-ray energy can include a set of X-ray image data of the examination volume recorded using X-ray radiation characterized by the first X-ray energy and a set of X-ray image data of the examination volume recorded using X-ray radiation characterized by the second X-ray energy. A set of multi-energy true image data of an examination volume can also be based on a set of X-ray image data of the examination volume recorded using X-ray radiation characterized by the first X-ray energy and a set of X-ray image data of the examination volume recorded using X-ray radiation characterized by the second X-ray energy.

[0017] A training function maps input data to output data. Furthermore, the output data can be related to one or more parameters of the training function. Training can determine and / or adjust one or more parameters of the training function. The determination and / or adjustment of one or more parameters of the training function can be based, in particular, on pairs consisting of training input data and corresponding training output data, wherein the training function used to generate the training mapping data is applied to the training input data. The determination and / or adjustment can also be based, in particular, on a comparison between the training mapping data and the training output data. Generally, a trainable function, that is, a function with one or more parameters that have not yet been adjusted, is also called a training function.

[0018] Other terms used for training functions include training mapping rules, mapping rules with training parameters, functions with training parameters, artificial intelligence-based algorithms, and machine learning algorithms. An example of a training function is an artificial neural network, where the edge weights of the artificial neural network correspond to the parameters of the training function. The term "neural network" can also be used instead of "neural network." Training functions can also be, in particular, deep artificial neural networks (the technical term is "deep neural network" or "deep artificial neural network"). Another example of a training function is a "Support Vector Machine," and other machine learning algorithms can also be used as training functions.

[0019] The first set of real image datasets and the multi-dimensional set of real image datasets share a common dimension. The first set of real image datasets is particularly a two-dimensional image dataset, a three-dimensional image dataset, or a four-dimensional image dataset.

[0020] The first set of real image data and the multi-functional real image data set specifically depict the examination volume; in other words, the image region of the first set of real image data corresponds to the image region of the multi-functional real image data set. Compared to the first set of real image data, the multi-functional real image data set has a particularly higher spatial resolution. With respect to at least one dimension, the spatial expansion of the multi-functional real image data set measured in pixels or voxels is particularly greater than that of the first set of real image data sets measured in pixels or voxels.

[0021] Multidimensional real image datasets and differential image datasets, in particular, share the same dimensionality. Therefore, differential image datasets are especially characterized by two-dimensional, three-dimensional, or four-dimensional differential image datasets.

[0022] The multi-dimensional real image dataset and the differential image dataset specifically depict the examination volume; in other words, the image region of the multi-dimensional real image dataset corresponds to the image region of the differential image dataset. The multi-dimensional real image dataset and the differential image dataset particularly have the same spatial resolution. With respect to each dimension, the spatial expansion measured in pixels or voxels in the multi-dimensional real image dataset is particularly the same as the spatial expansion measured in pixels or voxels in the differential image dataset.

[0023] The inventors have recognized that by using not only a first set of real image data but also a multi-energy set of real image data, the characteristics of the examination volume can be provided more accurately as input data for training the function. By using first and second X-ray energies, it is particularly possible to distinguish materials that have similar X-ray absorption values ​​at one of the two X-ray energies, thus making them difficult or barely distinguishable when using only one X-ray energy. Therefore, by means of the training function, better distinction can be achieved, in particular, between different materials in the examination area, and a more accurate and, especially, less error-prone set of differential image data can be determined.

[0024] The inventors have also recognized that the high resolution of a multi-energy real image dataset is sufficient to achieve the corresponding resolution of a differential image dataset. Different materials can be distinguished based on different X-ray energies based on a first real image dataset with a lower resolution. Therefore, what is particularly unnecessary for improved distinction is subjecting the examination volume to a higher X-ray dose, since the lower resolution of the first real image dataset and therefore less raw data are sufficient for the first real image dataset.

[0025] According to another aspect of the invention, the method for determining the differential image data set further includes determining a second true image data set of the examination volume with respect to the second X-ray energy, wherein the input data is also based on the second true image data set.

[0026] The second real image data group can be determined by receiving the second real image data group. In particular, the second real image data group can be determined by means of an interface and / or a computing unit.

[0027] Secondary real image datasets and multi-dimensional real image datasets, in particular, share the same dimensionality. Secondary real image datasets are especially two-dimensional, three-dimensional, or four-dimensional image datasets.

[0028] The second set of real image data and the multi-functional real image data set specifically describe the examination volume; in other words, the image region of the second set of real image data corresponds to the image region of the multi-functional real image data set. Compared to the first set of real image data, the multi-functional real image data set has a particularly higher spatial resolution. With respect to at least one dimension, the spatial expansion of the multi-functional real image data set, particularly measured in pixels or voxels, is greater than the spatial expansion measured in pixels or voxels of the first set of real image data.

[0029] The inventors have realized that by using a second set of real image data in the input data, improved differentiation between different materials in the inspection area can be achieved, since information about the X-ray absorption of the region of the inspection volume with respect to the second X-ray energy can also exist and be processed separately.

[0030] According to another feasible aspect of the invention, the first set of true image data is based on the first output data, and the multi-energy true image data set is based on the second output data and the first output data regarding the second X-ray energy. The spatial resolution of the multi-energy true image data set is particularly higher than that of the first true image data set. The spatial resolution of the multi-energy true image data set is particularly similar to that of the differential image data set. According to another feasible aspect of the invention, the second set of true image data is based on the second output data. The spatial resolution of the multi-energy true image data set is particularly higher than that of the second true image data set.

[0031] The inventors have recognized that, under the premise that the first set of real image data is based on the first output data, and the optional second set of real image data is based on the second output data, while the multi-energy real image data set is based on both the first and second output data, it is unnecessary, compared to the prior art, to subject the examination volume to a higher X-ray dose. This also applies when using the second set of real image data. The spatial resolution of the differential image data set can therefore be based on the multi-energy real image data set, and therefore, X-ray energy-related distinctions between different materials can be made based on the lower resolution first and second sets of real image data.

[0032] According to another aspect of the invention, the method for determining differential image data sets further includes: receiving a first X-ray projection of the examination volume with respect to a first X-ray energy; and receiving a second X-ray projection of the examination volume with respect to a second X-ray energy; wherein the first true image data set is based on and / or includes the first X-ray projection; and / or wherein the second true image data set is based on and / or includes the second X-ray projection; and / or wherein the multi-energy true image data set is based on and / or includes the first X-ray projection and the second X-ray projection.

[0033] Specifically, a first X-ray projection can be received using an interface. Specifically, a second X-ray projection can be received using an interface. Specifically, the first X-ray projection corresponds to first output data, and the second X-ray projection corresponds to second output data.

[0034] First X-ray projections with first X-ray energy, especially X-ray projections recorded using X-ray radiation of first X-ray energy. Second X-ray projections with second X-ray energy, especially X-ray projections recorded using X-ray radiation of second X-ray energy.

[0035] The inventors have recognized that by using multiple two-dimensional X-ray projections, it is particularly possible to examine changes in volume over time, especially time-varying contrast agent concentrations. Consequently, the training function can be based on more data and therefore more accurately and less susceptible to errors in determining the differential image dataset.

[0036] According to another aspect of the invention, the first set of real image data is a reconstruction of at least three dimensions of the first X-ray projection; and / or the second set of real image data is a reconstruction of at least three dimensions of the second X-ray projection; and / or the multi-energy real image data set is a reconstruction of at least three dimensions of the first X-ray projection and the second X-ray projection. In particular, the reconstruction of the first X-ray projection and the second X-ray projection is independent of the first X-ray energy and the second X-ray energy, or the reconstruction is a reconstruction without considering the X-ray energy.

[0037] Generally, reconstruction refers to determining an n-dimensional image data set based on multiple m-dimensional image data sets, where m < n. Here, the multiple m-dimensional image data sets should specifically be projected onto an n-dimensional volume described by the n-dimensional image data sets. Reconstruction can also, in particular, refer to determining a three-dimensional image data set based on multiple two-dimensional image data sets. For example, such reconstruction can be based on filtered back projection; alternatively, iterative reconstruction methods or the Feldkamp algorithm are known to those skilled in the art.

[0038] The inventors have recognized that sets of three-dimensional real image data are particularly suitable for illustrating the characteristics of an inspection volume. In particular, almost complete information about the inspection volume can be obtained through sets of three-dimensional real image data.

[0039] According to another feasible aspect of the invention, the method for determining a differential image data set further includes: receiving a first X-ray projection of the examination volume with respect to a first X-ray energy, wherein the first true image data set is based on and / or includes the first X-ray projection. Unlike the aspects described above, in these aspects of the invention, a second X-ray projection is not necessarily received.

[0040] According to another aspect of the invention, the examination volume includes a contrast agent during the recording of a first X-ray projection, and / or during the recording of a second X-ray projection. In particular, the examination volume includes a contrast agent if one or more blood vessels within it contain the contrast agent. The concentration of the contrast agent is particularly variable over time. The inventors have recognized that a training function can extract vascular structures particularly well if the vascular structures are highlighted by the presence of the contrast agent.

[0041] According to another aspect of the invention, the first X-ray projection and the second X-ray projection are recorded simultaneously. Here, if the first X-ray projection and the second X-ray projection are recorded within a time interval of 30 seconds or less, especially within a time interval of 20 seconds or less, especially within a time interval of 10 seconds or less, especially within a time interval of 5 seconds or less, then the first X-ray projection and the second X-ray projection are referred to as being recorded simultaneously.

[0042] The inventors have recognized that when a first X-ray projection and a second X-ray projection are recorded simultaneously, the first X-ray projection and the second X-ray projection depict the same or similar temporal state of the examined volume, respectively. Thus, the first set of real image data and the second set of real image data, or the multi-energy set of real image data, also describe the same or similar state of the examined volume, and information about not only the first X-ray energy but also the second X-ray energy can be provided through said state.

[0043] According to another aspect of the invention, the first X-ray projection is a record of the first X-ray source and the first X-ray detector, and the second X-ray projection is a record of the second X-ray source and the second X-ray detector.

[0044] The first X-ray source differs in particular from the second X-ray source, and the second X-ray detector differs in particular from the first X-ray detector. The first and second X-ray sources have the same structure and / or the same type, and / or the first and second X-ray detectors have the same structure and / or the same type.

[0045] The inventors have realized that a first X-ray projection and a second X-ray projection can be recorded independently of each other using two X-ray sources and two X-ray detectors. Therefore, especially if the first and second X-ray projections are recorded alternately, the necessary movement between recording the first and second X-ray projections can be significantly reduced. Furthermore, the first X-ray source can be driven by a first X-ray energy, and the second X-ray source can be driven by a second X-ray energy, thus eliminating the need for switching between the first and second X-ray energies.

[0046] According to another aspect of the invention, a dual-plane X-ray apparatus includes a first X-ray source, a second X-ray source, a first X-ray detector, and a second X-ray detector. The inventors have recognized that by using a dual-plane X-ray apparatus, the first X-ray source and the first X-ray detector can be better coordinated with the second X-ray source and the second X-ray detector. In particular, registration of the first X-ray source and the first X-ray detector with respect to the second X-ray source and the second X-ray detector can be eliminated because the relative positions of the X-ray source and the X-ray detector in the dual-plane X-ray apparatus are known.

[0047] According to another aspect of the invention, each of the first X-ray projections in the first X-ray projection is an X-ray projection of the examination volume with respect to a projection direction from a first projection angle range, and each of the second X-ray projections in the second X-ray projection is an X-ray projection of the examination volume with respect to a projection direction from a second projection angle range, wherein the first projection angle range is different from the second projection angle range.

[0048] The projection direction of an X-ray projection, specifically the direction from the position of the X-ray source to the position of the X-ray detector at the moment the X-ray projection is recorded, is the direction from the X-ray source to the X-ray detector. The projection direction can be understood in particular as a vector or a straight line in space.

[0049] The projection angle range includes multiple projection directions. The projection angle range can also be understood, in particular, as the spatial angle range with respect to a point on the inspection volume, and especially with respect to the midpoint of the inspection volume. In this case, the multiple projection directions with respect to the point on the inspection volume are particularly within the spatial angle range. Alternatively, the projection angle range can also be understood as the trajectory of the X-ray detector during X-ray projection recording. The projection angle range can also be understood, in particular, as an arc. The projection angle range can also be understood, in particular, as a convex envelope of multiple projection directions.

[0050] The inventors have recognized that (assuming the first and second projection angle ranges have fixed sizes) first and second X-ray projections from different projection angle ranges include more spatial information about the area being examined compared to first and second X-ray projections from the same projection angle range. Here, better spatial information can relate not only to the size of the covered angle range but also to the angular resolution.

[0051] According to another aspect of the invention, the first projection angle range and the second projection angle range are non-intersecting. The first and second projection angle ranges are particularly non-intersecting if the projection direction of the first projection angle range is not included in the second projection angle range, and if the projection direction of the second projection angle range is not included in the first projection angle range. The inventors have recognized that a particularly large angular range can be covered by means of non-intersecting projection angle ranges.

[0052] According to another aspect of the invention, the first projection angle range includes the second projection angle range, or the second projection angle range includes the first projection angle range. The inventors have recognized that by using overlapping projection angle ranges, complete angular information is provided not only for the first X-ray energy but also for the second X-ray energy, and therefore, especially for the entire angle range, it is possible to distinguish between different materials using a trained function.

[0053] According to another aspect of the invention, the output data of the training function includes a set of probability data, wherein the differential image data set is based on the set of probability data.

[0054] Here, the probabilistic data set specifically associates one or more pixels or voxels of the multi-potential real-image data set with probability values. The probability values ​​can be assigned to all pixels or voxels of the multi-potential real-image data set, in which case the probabilistic data set can be understood as a probabilistic image data set. The probability values ​​are, in particular, numbers greater than or equal to 0 and less than or equal to 1. The probability value associated with a voxel can specifically relate to the probability that the voxel is included in an image of a blood vessel within the examination volume. Alternatively, the probability value associated with a voxel can specifically relate to the probability that the voxel is not included in an image of a blood vessel within the examination volume.

[0055] The probability values ​​can also be binary, meaning they can either have a value of 0 or a value of 1. In this case, the probabilistic image data set can also be understood as a segment of the multipotent real image data set, specifically as a segment of images of blood vessels within the multipotent real image data set.

[0056] Here, the probabilistic data set has the same dimensions as the multipotent real image data set and / or the difference image data set. Furthermore, particularly with respect to each dimension, the probabilistic data set has the same extension as the multipotent real image data set and / or the difference image data set, wherein the extension is measured, in particular, in the number of pixels or in the number of voxels.

[0057] The inventors have recognized that, when applying a training function to input data, probability values ​​can be determined particularly simply for a specific voxel in a set of multipotent real image data corresponding to a blood vessel included in the examination volume. Here, in a broader sense, image processing is involved, where the training function can achieve good results for image processing in a known manner.

[0058] According to another aspect of the invention, a method for determining a differential image data set particularly includes: receiving a transfer function, and modifying the probability data set based on the transfer function.

[0059] Here, in particular, the interface is used to receive the transfer function, and in particular, the computation unit is used to modify the probability data set.

[0060] A transfer function is, in particular, a function that maps probability values ​​to probability values. Therefore, a transfer function is, in particular, a function that maps the interval [0;1] to the interval [0;1]. The transfer function T can be, in particular, a monotonically increasing function, that is, for x < y, T(x) ≤ T(y), and the transfer function T can also be, in particular, a strictly monotonically increasing function, that is, for x < y, T(x) < T(y). Advantageously, the transfer function is a continuous and / or differentiable function. Advantageously, for the transfer function T, the relations T(0) = 0 and T(1) = 1 apply.

[0061] The transfer function can be specified by the user. Alternatively, the transfer function can be selected from several available transfer functions, such as based on the type of real image data set, based on the recording parameters used for the first real image data set, the second real image data set, and / or the multi-energy real image data set, or for the first X-ray projection and / or the second X-ray projection, based on the location of the examination volume in the patient's body, and / or based on the blood vessels included in the examination volume.

[0062] Modifying a set of probability data may in particular include applying a transfer function to each probability value of the set of probability data. For each probability value of the set of probability data, the modified probability value is determined, in particular, by applying the transfer function to the probability value, and the modified set of probability data includes, in particular, the modified probability value.

[0063] The inventors have recognized that, by using an appropriate transfer function, the strength of the image structure or image structure corresponding to the background can be enhanced or weakened. For example, if T(x) = x γUsing γ as the transfer function, for 0 < γ < 1, it enhances the image structure corresponding to the background, and for γ > 1, it weakens the image structure corresponding to the background.

[0064] According to another aspect of the invention, the differential image data set is based on the multiplication of a probability data set with a first real image data set and / or with a second real image data set and / or with a multi-potential real image data set. The differential image data set specifically corresponds to the multiplication of the probability data set with the multi-potential real image data set.

[0065] The inventors have recognized that a three-dimensional differential image data set can be effectively generated by multiplying a set of probability data with a set of real image data, since the intensity values ​​of image regions with low probability values ​​are masked by multiplication, and the image regions correspond precisely to regions of the examination volume that do not correspond to blood vessels included in the examination volume.

[0066] In another feasible aspect, the present invention relates to a method for determining a differential image data set for an examination volume, the method comprising: receiving a two-dimensional first X-ray projection of the examination volume with respect to a first X-ray energy; receiving a two-dimensional second X-ray projection of the examination volume with respect to a second X-ray energy, wherein the first X-ray energy and the second X-ray energy are different; determining a three-dimensional first true image data set based on the two-dimensional first X-ray projection; determining a three-dimensional multi-energy true image data set based on the two-dimensional first X-ray projection and the two-dimensional second X-ray projection; determining a differential image data set by applying a training function to input data, wherein the input data includes the three-dimensional first true image data set and the three-dimensional multi-energy true image data set; and providing the three-dimensional differential image data set.

[0067] In a second aspect, the present invention relates to a computer-implemented method for providing a training function, the method comprising: determining a first set of training real image data with respect to a first training X-ray energy; and determining a multi-energy training real image data set with respect to a first training X-ray energy and a second training X-ray energy, wherein the second training X-ray energy differs from the first training X-ray energy. The method further comprises: determining a comparative differential image data set of the training check volume, and determining the training differential image data set of the training check volume by applying the training function to input data, wherein the input data is based on the first set of training real image data and the multi-energy training real image data set. The method further comprises: adjusting the training function based on a comparison of the training differential image data set and the comparative differential image data set, and providing the training function.

[0068] Here, the first training real image data set can be determined by receiving the first training real image data set. Here, the multi-potential training real image data set can be determined by receiving the multi-potential training real image data set. Here, the comparison difference image data set can be determined by receiving the comparison difference image data set. In particular, the training function can be provided by displaying, transmitting, and / or storing the training function.

[0069] In particular, the first training real image dataset can be determined using a training interface and / or a training computation unit. In particular, the multi-potential training real image dataset can be determined using a training interface and / or a training computation unit. In particular, the comparative difference image dataset can be determined using a training interface and / or a training computation unit. In particular, the training difference image dataset can be determined using a training computation unit. In particular, the training function can be tuned using a training computation unit. In particular, the training function can be provided using a training interface.

[0070] The first training set of real image data can, in particular, possess all the characteristics of a reference set of real image data used to provide a method description for the difference set of image data. The first training set of real image data is, in particular, a set of real image data. The multi-capable training set of real image data can, in particular, possess all the characteristics of a reference set of multi-capable real image data used to provide a method description for the difference set of image data. The multi-capable training set of real image data is, in particular, a set of real image data. The training and comparison sets of difference image data can possess all the characteristics of a reference set of difference image data used to provide a method description for the difference set of image data.

[0071] The inventors have recognized that the described method can provide a training function that can be used in methods for providing differential image data sets.

[0072] According to another feasible aspect of the invention, the method for providing the training function further includes: determining a second set of training real image data with respect to a second training X-ray energy, wherein the input data is further based on the second set of training real image data. Here, the second set of training real image data can be determined by receiving the second set of training real image data. In particular, the second set of training real image data can be determined by means of a training interface and / or by means of a training computation unit. The second set of training real image data can, in particular, have all the characteristics described in the method for providing a differential image data set. The second set of training real image data is, in particular, a set of real image data.

[0073] According to another aspect of the invention, a method for providing a training function includes: determining a set of masked image data for a training examination volume, wherein the set of differential image data is determined by digital subtraction angiography based on the masked image data set and a first set of training real image data, or by digital subtraction angiography based on the masked image data set and a multi-potential training real image data set. The comparison of the differential image data set can also be particularly determined by digital subtraction angiography based on the masked image data set and a second set of training real image data.

[0074] In particular, the mask image data set can be determined by receiving the mask image data set. Alternatively, the mask image data set can be determined based on the mask X-ray projection of the training examination volume, wherein the training examination volume does not include the contrast agent at the time of recording the mask X-ray projection. In particular, the mask image data set can be determined by means of a training interface and / or a training computation unit.

[0075] In particular, the difference image dataset can be compared by subtracting the masked image dataset from the real image dataset. In this case, the difference image dataset is compared, especially the subtracted image dataset.

[0076] The inventors have recognized that by training a set of differential image data and comparing it with a set of comparative differential image data based on digital subtraction angiography (DSA), the output data of the training function corresponds particularly well to the results of DSA. In particular, the output data of the training function can be used to replace the results of DSA.

[0077] According to another aspect of the invention, the method for providing a training function further includes: receiving a first three-dimensional material model of a training check volume, wherein a first set of training real image data and / or a set of multi-energy training real image data are based on simulations of the interaction between X-ray radiation and the first three-dimensional material model.

[0078] Material models, in particular, assign material properties to a set of spatial locations. These spatial locations can be given, in particular, by voxels. Material properties can be, in particular, X-ray absorption coefficients, or functions describing X-ray absorption coefficients in relation to X-ray energy.

[0079] In particular, training with real image datasets can be simulated by simulating the spatial distribution of X-ray radiation with respect to the projection direction and material properties, using the first and / or second X-ray energies. Monte Carlo simulations can be employed to perform this simulation.

[0080] The inventors have realized that by using a material model to adjust the parameters of a training function, it is possible to discard as much real image data as possible. This allows for the generation of any amount of training data without subjecting patients to unnecessary radiation doses through X-ray recordings.

[0081] According to another aspect of the invention, the method for providing a training function further includes: receiving a second three-dimensional material model of a training examination volume, wherein the first three-dimensional material model is a material model of the training examination volume including a contrast agent, wherein the second three-dimensional material model is a material model of the training examination volume without a contrast agent, and wherein the masked image data set is based on a simulation of the interaction between X-ray radiation and the second three-dimensional material model.

[0082] The inventors have recognized that digital subtraction angiography can be fully described using a first material model with contrast agent and a second material model without contrast agent. This allows for the generation of training data for a training function through simulation, without the need to record actual training data. Furthermore, it allows for the determination of the difference between the masked image data set and the training difference image data set, in particular, without the need to record actual training data or subject the patient to radiation exposure.

[0083] In a third aspect, the present invention relates to a system for providing differential image data sets for examining volumes, the system comprising an interface and a computing unit.

[0084] - Wherein the interface and / or computing unit constitute a first set of true image data for determining the examination volume with respect to a first X-ray energy.

[0085] - The interface and / or computing unit further constitute a set of multi-energy true image data for determining the examination volume with respect to a first X-ray energy and a second X-ray energy, wherein the second X-ray energy differs from the first X-ray energy.

[0086] - The computational unit further comprises a differential image data set for determining the examination volume by applying a training function to the input data, wherein the input data is based on a first set of real image data and a multi-real image data set, and

[0087] - The interface also constitutes a set of differential image data.

[0088] This providing unit can be configured to implement the aforementioned method and aspects thereof for providing differential image data sets according to the present invention. The providing unit is configured to implement the method and aspects thereof in such a way that an interface and a computing unit are configured to implement the corresponding method steps.

[0089] In a fourth aspect, the present invention relates to an X-ray apparatus comprising a system according to the invention. The X-ray apparatus particularly includes a first X-ray source, a second X-ray source, a first X-ray detector, and a second X-ray detector. The first X-ray source and the first X-ray detector are particularly configured to rotate simultaneously around an examination volume. The second X-ray source and the second X-ray detector are also particularly configured to rotate simultaneously around an examination volume. The X-ray apparatus particularly relates to a dual-source C-arm X-ray system or a dual-source computed tomography device.

[0090] In a fifth aspect, the present invention relates to a training system for providing a training function, the training system comprising a training interface and a training computation unit.

[0091] - Wherein the training interface and / or training computation unit constitute a first set of training real image data for determining the training check volume with respect to the first training X-ray energy.

[0092] - The training interface and / or training computation unit also constitute a multi-energy training real image data set for determining the training check volume with respect to the first training X-ray energy and the second training X-ray energy.

[0093] The energy of the second training X-ray is different from that of the first training X-ray.

[0094] - The training interface and / or training computation unit also constitute a set of comparative differential image data for determining the training check volume.

[0095] - The training computation unit also constitutes a training differential image data set for determining the training check volume by applying a training function to the input data.

[0096] The input data is based on the first set of real training image data and the multi-potential training image data set.

[0097] - The training computation unit also constitutes a function for adjusting the training function based on a comparison between the training differential image data set and the comparison differential image data set.

[0098] - The training interface also constitutes the function used to provide training functions.

[0099] This training system is specifically configured to implement the aforementioned method and aspects thereof for providing training functions according to the present invention. The training system is configured to implement the method and aspects thereof by means of a training interface and training units configured to implement the corresponding method steps.

[0100] In a sixth aspect, the present invention relates to a computer program product having a computer program that can be directly loaded into the memory of a providing system, the computer program having program segments so as to implement all steps of a method for providing a differential image data set or all steps of an aspect of the method when the program segments are run by the providing system; and / or, the computer program product can be directly loaded into the training memory of a training system, the computer program having program segments so as to implement all steps of a method for providing a training function or all steps of an aspect of the method when the program segments are run by the training system.

[0101] In a feasible seventh aspect, the present invention relates to a computer program product having a computer program that can be directly loaded into the memory of a providing system, the computer program having program segments so as to implement all steps of a method for providing a differential image data set or all steps of aspects of the method when the program segments are run by the providing system.

[0102] In a feasible eighth aspect, the present invention relates to a computer program product having a computer program that can be directly loaded into the training memory of a training system, the computer program having program segments so that, when the program segments are run by the training system, all steps of the method for providing training functions or all steps of an aspect of the method are implemented.

[0103] In a ninth aspect, the present invention relates to a computer-readable storage medium storing program segments readable and operable by a providing system, such that when the program segments are run by the providing system, all steps of a method for providing a differential data set or all steps of an aspect of the method are implemented; and / or, the computer-readable storage medium storing program segments readable and operable by a training system, such that when the program segments are run by the training system, all steps of a method for providing a training function or all steps of an aspect of the method are implemented.

[0104] In a feasible tenth aspect, the present invention relates to a computer-readable storage medium on which program segments readable and operable by a providing system are stored, so that when the program segments are run by the providing system, all steps of a method for providing a differential image data set or all steps of an aspect of the method are implemented.

[0105] In a feasible eleventh aspect, the present invention relates to a computer-readable storage medium on which program segments readable and operable by a training system are stored, so that when the program segments are run by the training system, all steps of the method for providing a training function or all steps of an aspect of the method are implemented.

[0106] In a twelfth aspect, the present invention relates to a computer program or a computer-readable storage medium comprising a training function provided by a method for providing a training function or an aspect thereof.

[0107] The software implementation scheme has the following advantages: it can also easily be upgraded to include existing supply units and / or training systems to operate in accordance with the invention. In addition to the computer program, such a computer program product may include additional components, such as assemblies and / or additional parts, and hardware components, such as hardware keys (software dongles, etc.) for using the software. Attached Figure Description

[0108] The above-described features, characteristics, and advantages of the present invention, as well as the ways and methods of achieving said features, characteristics, and advantages, will become clearer and more apparent in conjunction with the following description of the embodiments, which are explained in more detail with reference to the accompanying drawings. The invention is not limited to the embodiments described herein. In the different drawings, the same parts are given the same reference numerals. The drawings are generally not to scale. The drawings show:

[0109] Figure 1 The examination volume and three-dimensional differential image data set with blood vessels are shown.

[0110] Figure 2 The first two-dimensional X-ray projection of the examined volume is shown.

[0111] Figure 3 A two-dimensional second X-ray projection of the examined volume is shown.

[0112] Figure 4 The first X-ray spectrum and the second X-ray spectrum are shown.

[0113] Figure 5 A first embodiment of the data stream for determining differential image data groups is shown.

[0114] Figure 6 A second embodiment of the data stream for determining differential image data groups is shown.

[0115] Figure 7 A third embodiment of the data stream for determining differential image data groups is shown.

[0116] Figure 8 A first embodiment showing the first projection angle range and the second projection angle range is illustrated.

[0117] Figure 9 The possible locations of the X-ray source in the first embodiment for the first and second projection angle ranges are shown.

[0118] Figure 10 A second embodiment showing the first projection angle range and the second projection angle range is illustrated.

[0119] Figure 11 The possible locations of the X-ray source in the second embodiment for the first and second projection angle ranges are shown.

[0120] Figure 12 A flowchart illustrating a first embodiment of a method for providing differential image data sets is shown.

[0121] Figure 13 A flowchart illustrating a second embodiment of a method for providing differential image data sets is shown.

[0122] Figure 14 A flowchart illustrating a first embodiment of a method for providing a training function is shown.

[0123] Figure 15 A flowchart illustrating a second embodiment of the method for providing a training function is shown.

[0124] Figure 16 An embodiment of the system is shown.

[0125] Figure 17 An embodiment of the training system is shown.

[0126] Figure 18 An X-ray device is shown. Detailed Implementation

[0127] Figure 1An examination volume VOL with two blood vessels VES.1 and VES.2 is shown, along with a three-dimensional differential image data set DD. Here, the image range of the differential image data set DDS corresponds to the examination volume VOL. In the illustrated embodiment, the examination volume VOL includes a first blood vessel VES.1 and a second blood vessel VES.2, wherein the first blood vessel VES.1 branches into two branches within the examination volume VOL. It is also possible that the examination volume VOL does not include blood vessels VES.1 and VES.2, includes exactly one blood vessel VES.1 and VES.2, or includes more than two blood vessels VES.1 and VES.2. In addition to blood vessels VES.1 and VES.2, the examination volume VOL includes other structures OS.1 and OS.2, which are not depicted in the three-dimensional first differential image data set DDS because they are attributed to the background and therefore not depicted in the three-dimensional first differential image data set.

[0128] In the illustrated embodiment, the volume VOL and the differential image data set DDS are examined about the first direction x, the second direction y, and the third direction z. Here, the first direction x, the second direction y, and the third direction z are orthogonal in pairs.

[0129] Figure 2 The diagram shows multiple first X-ray projections XP.1a, ..., XP.1d of the examination volume VOL with respect to the first X-ray energy. Figure 3 Multiple second X-ray projections XP.2a, ..., XP.2d are shown for the examination volume VOL with respect to a second X-ray energy, wherein the second X-ray energy differs from the first X-ray energy. In the illustrated embodiment, the first X-ray projections XP.1a, ..., XP.1d form a first true image data set RD.1, and the second X-ray projections XP.2a, ..., XP.2d form a second true image data set RD.2. Alternatively, the first true image data set RD.1 may be determined based on a three-dimensional reconstruction of the first X-ray projections XP.1a, ..., XP.1d, and / or the second true image data set RD.2 may be determined based on a three-dimensional reconstruction of the second X-ray projections XP.2a, ..., XP.2d.

[0130] In the illustrated embodiment, four two-dimensional X-ray projections XP.1a, ..., XP.1d, XP.2a, ..., XP.2d are shown respectively. More or fewer two-dimensional X-ray projections XP.1a, ..., XP.1d, XP.2a, ..., XP.2d may also be present or used.

[0131] Here, each of the two-dimensional X-ray projections XP.1a, ..., XP.1d, XP.2a, ..., XP.2d is an X-ray projection of the examination volume VOL with respect to the following projection directions: XP.1a and XP.2a are X-ray projections of the examination volume VOL with respect to the following projection directions, wherein the projection directions are parallel to a first direction x. XP.1b and XP.2b are X-ray projections of the examination volume VOL with respect to the following projection directions, wherein the projection directions are parallel to a second direction y. XP.1c and XP.2c are X-ray projections of the examination volume VOL with respect to the following projection directions, wherein the projection directions are parallel to the first direction x. XP.1d and XP.2d are X-ray projections of the examination volume VOL with respect to the following projection directions, wherein the projection directions are parallel to the second direction y.

[0132] Furthermore, each of the two-dimensional X-ray projections XP.1a, ..., XP.1d, XP.2a, ..., XP.2d is associated with a time, wherein in the embodiment described, the time corresponds to the time at which the corresponding X-ray projection is recorded.

[0133] In the illustrated embodiment, each of the two-dimensional X-ray projections XP.1a, ..., XP.1d, XP.2a, ..., XP.2d depicts the blood vessels VES.1, VES.2 contained within the examination volume VOL. Furthermore, other structures OS.1, OS.2 within the examination volume VOL are depicted by the two-dimensional X-ray projections XP.1a, ..., XP.1d, XP.2a, ..., XP.2d.

[0134] At different times during the recording of two-dimensional X-ray projections XP.1a, ..., XP.1d, XP.2a, ..., XP.2d, the blood vessels VES.1 and VES.2 include contrast agent concentrations CA.1, ..., CA.4 that are variable over time. Specifically, during the recording of X-ray projections XP.1a and XP.2a, the blood vessels VES.1 and VES.2 have a contrast agent concentration of CA.1. Furthermore, during the recording of X-ray projections XP.1b and XP.2b, the blood vessels VES.1 and VES.2 have a contrast agent concentration of CA.2. Additionally, during the recording of X-ray projections XP.1c and XP.2c, the blood vessels VES.1 and VES.2 have a contrast agent concentration of CA.3. Furthermore, during the recording of X-ray projections XP.1d and XP.2d, the blood vessels VES.1 and VES.2 have a contrast agent concentration of CA.4. Here, the contrast agent involves an X-ray contrast agent, such that the corresponding contrast agent concentrations CA.1, ..., CA.4 can be determined by X-ray projection. The contrast agent concentrations CA.1, ..., CA.4 vary over time through static or dynamic fluid flow in blood vessels VES.1, VES.2. In the illustrated embodiment, the fluid involves blood.

[0135] exist Figure 2 In the records of the first X-ray projections XP.1a, ..., XP.1d with the first X-ray energy shown, the contrast agent and the first other structure OS.1 (e.g., a bone structure) have similar X-ray absorption. Therefore, it is almost impossible to distinguish the contrast agent and the first other structure OS.1 based on the first X-ray projections XP.1a, ..., XP.1d. However, the contrast agent and the second other structure OS.2 (e.g., a metallic structure) have different X-ray absorption and are therefore easily distinguishable.

[0136] exist Figure 3 In the records of the second X-ray projections XP.2a, ..., XP.2d with the second X-ray energy shown, the contrast agent and the second other structure OS.2 (e.g., a metallic structure) have similar X-ray absorption. Therefore, it is almost impossible to distinguish the contrast agent and the second other structure OS.2 based on the second X-ray projections XP.2a, ..., XP.2d. However, the contrast agent and the first other structure OS.1 (e.g., a bone structure) have different X-ray absorption and are therefore easily distinguishable.

[0137] Therefore, the contrast agent can be precisely distinguished from other structures OS.1 and OS.2 by means of the first real image data set RD.1 and the multi-functional real image data set RD.M.

[0138] Figure 4The diagram shows a first X-ray spectrum SP.1 and a second X-ray spectrum SP.2, generated by X-ray tubes SRC.1 and SRC.2 as X-ray sources. Here, the first X-ray spectrum SP.1 corresponds to a first X-ray energy E1 or a first accelerating voltage U1 = E1 / e (where e corresponds to the elementary charge), and the second X-ray spectrum corresponds to a second X-ray energy E2 or a second accelerating voltage U2 = E2 / e, where the first X-ray energy E1 or the first accelerating voltage U1 is greater than the second X-ray energy E2 or the second accelerating voltage U2. In the diagram, the intensity I(λ) of the X-ray radiation is illustrated in relation to the wavelength λ of the X-ray radiation. Here, the intensity I(λ) is proportional to the number of X-ray photons of wavelength λ generated by the X-ray sources SRC.1 and SRC.2.

[0139] According to Duane-Hent's law, X-ray spectra SP.1 and SP.2 have a minimum wavelength λ. (min) = hc / eU (where c represents the speed of light and h represents Planck's constant), such that the minimum wavelength λ in this first X-ray spectrum SP.1 (min) 1 is less than the minimum wavelength λ of the second X-ray spectrum (min) 2. Furthermore, according to Cramer's law, the X-ray spectrum at λ... 1 / 2 =2λ (min) 1 / 2 It has the maximum relative intensity at a wavelength.

[0140] Furthermore, the first X-ray spectrum and the second X-ray spectra SP.1 and SP.2 exhibit one or more characteristic wavelengths λ. (c1) , λ (c2) The peak of characteristic X-ray radiation is located at this point. Here, the characteristic wavelength λ... (c1) , λ (c2) The characteristic X-ray radiation is not related to the accelerating voltages U1 and U2 or the X-ray energies E1 and E2, but rather to the anode material of the X-ray tube. It is generated by transitions between energy levels within the electron shells of the anode material.

[0141] Figure 5A first embodiment of the data flow for determining differential image data sets DD is shown. In the illustrated embodiment, the first true image data set RD.1 is a two-dimensional X-ray projection of the examination volume VOL with respect to a first X-ray energy E1, and the second true image data set RD.2 is a two-dimensional X-ray projection of the examination volume VOL with respect to a second X-ray energy E2. Advantageously, the first true image data set RD.1 and the second true image data set RD.2 are recorded with respect to the same projection direction. In this embodiment, the multi-energy true image data set RD.M includes both X-ray projections with respect to the first X-ray energy E1 and X-ray projections with respect to the second X-ray energy E2. In particular, in this embodiment, not only the first true image data set RD.1, but also the second true image data set RD.2 and the multi-energy true image data set RD.M are two-dimensional image data sets.

[0142] In the embodiment described, the training function TF receives a first set of real image data RD.1 and a second set of real image data RD.2 as input data. Therefore, in this embodiment, the input data of the training function TF also includes a multi-set of real image data RD.M.

[0143] Furthermore, in this embodiment, the output data of the training function TF corresponds to the differential image data set DD, which is also a two-dimensional image data set in this embodiment. Therefore, in this embodiment, the training function TF is a function that maps a two-dimensional image data set onto another two-dimensional image data set.

[0144] In the embodiment described, the first real image data set RD.1 and the second real image data set RD.2 have the same pixel-measured expansion with respect to both dimensions, and in the same embodiment, the differential image data set DD has the same pixel-measured expansion with respect to both dimensions as the first real image data set RD.1 and the second real image data set RD.2. For example, the first real image data set RD.1, the second real image data set RD.2, and the differential image data set may have an expansion of 512 pixels with respect to the first dimension, and may also have an expansion of 512 pixels with respect to the second dimension.

[0145] Figure 6 A second embodiment of the data flow for determining a differential image data set DD is shown. In this embodiment, the first real image data set RD.1, the second real image data set RD.2, and the multi-functional real image data set RD.M are three-dimensional image data sets for examining the volume volume (VOL), and the differential image data set DD is also a three-dimensional image data set for examining the volume volume (VOL).

[0146] In the described embodiment, the first real image data set RD.1 is a three-dimensional reconstruction of the first X-ray projection XP.1, wherein the first X-ray projection XP.1 is an X-ray projection of the examination volume VOL with respect to the first X-ray energy E1. Furthermore, the second real image data set RD.2 is a three-dimensional reconstruction of the second X-ray projection XP.2, wherein the second X-ray projection XP.2 is an X-ray projection of the examination volume VOL with respect to the second X-ray energy E2. Additionally, the multi-energy real image data set RD.M is a three-dimensional reconstruction of the first X-ray projection XP.1 and the second X-ray projection XP.2. The first X-ray projection XP.1 and the second X-ray projection XP.2 are particularly two-dimensional X-ray projections of the examination volume VOL, especially two-dimensional X-ray projections with respect to multiple projection directions, respectively.

[0147] In the described embodiment, the first real image data set RD.1 and the second real image data set RD.2 have the same expansion measured in voxels with respect to each dimension. For example, the first real image data set RD.1 and the second real image data set RD.2 may have an expansion of 256 voxels with respect to the first dimension, an expansion of 256 voxels with respect to the second dimension, and an expansion of 256 voxels with respect to the third dimension (totaling approximately 17.10). 6 (Volls). Furthermore, in the described embodiment, the multi-potential real image dataset RD.M has a higher voxel-measured expansion per dimension than the first real image dataset RD.1. For example, the multi-potential real image dataset RD.M may have a 512 voxel expansion per first dimension, a 512 voxel expansion per second dimension, and a 512 voxel expansion per third dimension (totaling approximately 134.10). 6 (Volumetrics).

[0148] In the described embodiment, the training function TF receives a first set of real image data RD.1, a second set of real image data RD.2, and a multi-functional set of real image data RD.M as input data. Alternatively, the training function TF may receive only the first set of real image data RD.1 and the multi-functional set of real image data RD.M as input data. Furthermore, the training function TF generates a differential image data set DD as output data, wherein the differential image data set DD is particularly a three-dimensional image data set DD. The training function TF is therefore, in particular, a function that maps the three sets of three-dimensional image data as input data to a single set of three-dimensional image data as output data, wherein the three sets of three-dimensional image data may also have different extensions.

[0149] Alternatively, the training function TF can generate a set of probability data as output data, and the differential image data set DD can be determined by multiplying the set of probability data with the set of multipotential real image data RD.M on a voxel-by-voxel basis.

[0150] In the described embodiment, the differential image data set DD has the same voxel-measured extension with respect to each dimension as the multipotent real image set RD.M. In the described alternative, the probability data set is particularly a three-dimensional probability data set, which has the same voxel-measured extension with respect to each dimension as the multipotent real image set RD.M. For example, the differential image data set DD or the probability data set can have an extension of 512 voxels with respect to the first dimension, 512 voxels with respect to the second dimension, and 512 voxels with respect to the third dimension (totaling approximately 134.10). 6 (Volumetrics).

[0151] Figure 7 A third embodiment of the data flow for determining a differential image data set DD is shown. In this embodiment, a first real image data set RD.1 includes a plurality of first X-ray projections XD.1, and a second real image data set RD.2 includes a plurality of second X-ray projections XP.2. A multi-energy real image data set RD.M also includes a plurality of first X-ray projections XP.1 and a plurality of second X-ray projections XP.2. In particular, here, the first X-ray projection XP.1 is an X-ray projection of the examination volume VOL with respect to a first X-ray energy E1, and the second X-ray projection XP.2 is an X-ray projection of the examination volume VOL with respect to a second X-ray energy E2.

[0152] In the described embodiment, the training function TF receives a first set of real image data RD.1 and a second set of real image data RD.2 as input data. Thus, the input data of the training function TF is implicitly also based on a multi-dimensional set of real image data RD.M. Furthermore, the training function TF generates a differential image data set DD as output data, wherein the differential image data set DD is particularly a three-dimensional image data set DD. The training function TF is therefore, in particular, a function that maps multiple first two-dimensional X-ray projections and multiple second two-dimensional X-ray projections, which are input data, onto the three-dimensional image data set, which is output data.

[0153] Figure 8 A first embodiment of a first projection angle range PA.1 and a second projection angle range PA.2 is shown. The projection angle ranges PA.1 and PA.2 shown can be used in particular for: recording a first X-ray projection XP.1 and / or a second X-ray projection XP.2, a first real image data set RD.1 and / or a second real image data set RD.2 and / or a multi-energy real image data set RD.M based on the first X-ray projection and / or the second X-ray projection.

[0154] The projection angle ranges PA.1 and PA.2 describe the projection directions of the X-ray projections XP.1 and XP.2 of the examination volume VOL. Here, the examination volume VOL is part of the patient PAT, which is mounted on the patient support device PPOS. Here, the first projection angle range PA.1 shows the possible position of the first X-ray source SRC.1, in particular, when recording the first X-ray projection XP.1. Here, the associated first X-ray detector DTC.1 is positioned on the side of the first X-ray source SRC.1 opposite to the examination volume VOL. Furthermore, the second projection angle range PA.2 shows the possible position of the first X-ray source SRC.1 or the second X-ray source SRC.2 when recording the second X-ray projection XP.2. Here, the associated first X-ray detector DTC.1 or the associated second X-ray detector DTC.2 is positioned on the side of the first X-ray source SRC.1 or the second X-ray source SRC.2 opposite to the examination volume VOL. The projection angle ranges PA.1 and PA.2 can also be interpreted as a set of projection directions.

[0155] In particular, the first projection angle range PA.1 can also be understood as the trajectory of the first X-ray source SRC.1 when recording the first X-ray projection XP.1. If the second X-ray projection XP.2 is recorded using the same X-ray source SRC.1 as the first X-ray projection XP.1, then the second projection angle range PA.2 can also be understood as the trajectory of the first X-ray source SRC.1 when recording the second X-ray projection XP.2. Alternatively, if the second X-ray projection XP.2 is recorded using a second X-ray source SRC.2 that is different from the first X-ray source SRC.1, then the second projection angle range PA.2 can be understood as the trajectory of the second X-ray source SRC.2.

[0156] The first projection angle range PA.1 can also be determined by means of circular rotation of the first X-ray source SRC.1 around the inspection volume VOL, wherein the first X-ray source SRC.1 describes an arc having an angle α / 2. Similarly, the second projection angle range PA.2 can be determined by means of circular rotation of the second X-ray source SRC.2 around the inspection volume VOL, wherein the second X-ray source SRC.2 also describes an arc having an angle α / 2. Alternatively to circular rotation and arc, elliptical rotation or elliptical arc of the first or second X-ray source SRC.1, SRC.2, or other at least partially concave movements are also possible. Here, the angle α is particularly greater than 180°, and the angle α particularly corresponds to the sum of 180° and the opening angle of the X-ray radiation originating from the first or second X-ray source SRC.1, SRC.2. Therefore, in the described embodiment, the angle α particularly corresponds to 200°.

[0157] exist Figure 8The figure shows a first projection angle range PA.1 and a second projection angle range PA.2 with different radii. The different radii are chosen for the purpose of overview in the figure and do not imply that the first X-ray source SRC.1 or the second X-ray source SRC.2 has a different spacing from the examination volume VOL or the center of rotation when recording the first X-ray projection XP.1 and the second X-ray projection XP.2.

[0158] Figure 9 For in Figure 8 The first embodiment shown in the diagram illustrates the first projection angle range PA.1 and the second projection angle range PA.2, indicating the possible locations POS.1 of the X-ray sources SRC.1 and SRC.2. t(1) 1), ..., POS.2 ( t(2) 3).

[0159] Here, POS.1(t) represents the position of the first X-ray source SRC.1 at time t, and POS.2(t) represents the position of the second X-ray source SRC.2 at time t. The first X-ray source SRC.1 and the second X-ray source SRC.2 are different, and the first X-ray source SRC.1 records the first X-ray projection XP.1 with respect to the first X-ray energy E1, while the second X-ray source SRC.2 records the second X-ray projection XP.2 with respect to the second X-ray energy E2.

[0160] In the illustrated embodiment, at time t (1) i Record the i-th X-ray projection of the first X-ray projection XP.1, where for i < j, t (1) i < t (1) j Furthermore, in the illustrated embodiment, at time t (2) i Record the i-th X-ray projection of the second X-ray projection XP.2, where for i < j, t (2) i < t (2) j Furthermore, in the illustrated embodiment, t (1) i <t (2) i < t (1) i+1 However, alternatively, other time series of the first X-ray projection and the second X-ray projection XP.1 and XP.2 can be used. Time t (1) i In particular, it can be included in the first X-ray projection XP.1, especially as metadata, and furthermore, at time t (2)i In particular, it can be included in the second X-ray projection XP.2.

[0161] At time t, the first X-ray source SRC.1 (1) i (here t) (1) 1. t (1) 2. t (1) 3) Position POS(t) (1) i )(exist Figure 9 Depicting the position POS(t) (1) 1) POS(t) (1) 2) POS(t) (1) At one of the locations in 3), the first X-ray source SRC.1 records information about the first X-ray energy E1 with respect to the projection direction v. (1) i (here v) (1) 1. v (1) 2. v (1) 3) The first X-ray projection XP.1 is a first X-ray projection. At the first X-ray source SRC.1 at time t (2) i (here t) (2) 1. t (2) 2. t (2) 3) Position POS(t) (2) i )(exist Figure 9 Depicting the position POS(t) (2) 1) POS(t) (2) 2) POS(t) (2) At position 3), the first X-ray source SRC.1 generally does not record the X-ray projection (except for a pair of i, j, t). (1) i = t (2) i (in the case of)

[0162] At time t, the second X-ray source SRC.2 (2) i (here t) (2) 1. t (2) 2. t (2) 3) Position POS.2(t (2) i )(exist Figure 9 The position POS.2(t) is depicted in the middle. (2) 1) POS.2(t (2) 2) POS.2(t (2) At one of the locations in 3), the second X-ray source SRC.2 records information about the second X-ray energy E2 with respect to the projection direction v.(2) i (here v) (2) 1. v (2) 2. v (2) 3) A second X-ray projection in XP.2. At time t, the second X-ray source SRC.2... (1) i (here t) (1) 1. t (1) 2. t (1) 3) Position POS.2(t (1) i )(exist Figure 9 The position POS.2(t) is depicted in the middle. (1) 1) POS.2(t (1) 2) POS.2(t (1) At position 3), the second X-ray source SRC.2 generally does not record the X-ray projection (except for a pair of i, j, t). (1) i = t (2) i (in the case of)

[0163] exist Figure 9 In this diagram, for overview purposes, only the positions of the first X-ray source SRC.1 and the second X-ray source SRC.2 are shown for the three X-ray projections XP.1 and XP.2, respectively. Generally, significantly more of the first X-ray projections XP.1 and XP.2 are used, and the positions of the first X-ray detector DTC.1 and the second X-ray detector DTC.2 are along the projection direction v. (1) 1. v (1) 2. v (1) 3. v (2) 1. v (2) 2. v (2) 3 is located on the side opposite to the first X-ray source or the second X-ray source SRC.1, SRC.2 of the inspection volume VOL.

[0164] exist Figure 9 The diagram shows a first projection angle range PA.1 and a second projection angle range PA.2 with different radii, and correspondingly, the position POS.1(t) (1) 1), ..., POS.2(t) (2) 3) It also has a different distance from the inspection volume VOL. The different radii or different distances are chosen especially for the sake of the overview of the figures, and especially do not imply that the first X-ray source SRC.1 or the second X-ray source SRC.2 has a different distance from the inspection volume VOL or the center of rotation when recording the first X-ray projection XP.1 and the second X-ray projection.

[0165] Figure 10 A second embodiment is shown, illustrating a first projection angle range PA.1 and a second projection angle range PA.2. The shown projection angle ranges PA.1 and PA.2 are particularly useful for: recording a first X-ray projection XP.1 and / or a second X-ray projection XP.2, a first real image data set RD.1 and / or a second real image data set RD.2 and / or a multi-energy real image data set RD.M based on the first X-ray projection and / or the second X-ray projection. The significance of the projection angle ranges PA.1 and PA.2 relative to the positions of the first X-ray source SRC.1, the first X-ray detector DTC.1, the second X-ray source SRC.2, and the second X-ray detector DTC.2 corresponds to the position of the first X-ray source SRC.1, the first X-ray detector DTC.1, the second X-ray source SRC.2, and the second X-ray detector DTC.2. Figure 8 The significance of the description.

[0166] In the second embodiment shown, the first projection angle range PA.1 can also be determined by means of circular rotation of the first X-ray source SRC.1 around the inspection volume VOL, wherein the first X-ray source SRC.1 describes an arc having an angle α+β. Furthermore, the second projection angle range PA.2 can be determined, in particular, by means of circular rotation of the second X-ray source SRC.2 around the inspection volume VOL, wherein the second X-ray source SRC.2 also describes an arc having an angle α+β. Alternatively to circular rotation and arc, elliptical rotation or elliptical arc of the first or second X-ray source SRC.1, SRC.2, or other at least partially concave movements, are also possible. Here, angle α is particularly greater than 180°, and angle α particularly corresponds to the sum of 180° and the opening angle of the X-ray originating from the first or second X-ray source SRC.1, SRC.2. Therefore, in the embodiment described, angle α particularly corresponds to 200°. Angle β can specifically correspond to the minimum angle between the direction from the first X-ray source SRC.1 to the first X-ray detector DTC.1 and the direction from the second X-ray source SRC.2 to the second X-ray detector DTC.2. Angle β is therefore lowered, particularly by the extension and geometry of the X-ray sources SRC.1, SRC.2 and the X-ray detectors DTC.1, DTC.2.

[0167] Figure 11 For in Figure 10 The second embodiment shown in the diagram illustrates the possible locations POS1(t) of the X-ray sources SRC.1 and SRC.2, including the first projection angle range PA.1 and the second projection angle range PA.2. (1) 1), ..., POS.2(t) (2) 3). Regarding the objects shown, refer to... Figure 9 The description.

[0168] In the second embodiment shown, the first X-ray detector DTC.1 and the second X-ray detector DTC.2 have a constant, especially minimal, spacing, and / or the first X-ray source SRC.1 and the second X-ray source SRC.2 have a constant, especially minimal, spacing.

[0169] exist Figure 10 The projection angle ranges PA.1, PA.2 and PA.2 shown in the figure are... Figure 11 The positions shown can also be used as a basis for recording using only one X-ray source SRC.1 and only one X-ray detector DTC.1, where the X-ray source SRC.1 can switch between a first X-ray energy and a second X-ray energy. Here, the first X-ray source is at position POS.1(t (1) 1) POS.1(t (1) 2) POS.1(t (1) 3) Record the first X-ray projection XP.1 with the first X-ray energy at position POS.2 (t (2) 1) POS.2(t (2) 2) POS.2(t (2) 3) Record the second X-ray projection XP.2 with the second X-ray energy. The remaining X-ray projections are not important.

[0170] Figure 12 A flowchart illustrating a first embodiment for providing differential image data sets (DDs) is shown.

[0171] The first step of the first embodiment shown is to determine a first real image data set RD.1 with respect to the first X-ray energy E1, which is the DET-RD.1 examination volume VOL. Here, DET-RD.1 is determined by receiving the first real image data set RD.1 via interface IF. Another step of the first embodiment shown is to determine a multi-energy real image data set RD.M with respect to the first X-ray energy E1 and the second X-ray energy E2, which is the DET-RD.M examination volume VOL. Here, DET-RD.M is determined by receiving a second real image data set RD.2 via interface IF.

[0172] Optionally, in the first embodiment, a second real image data set RD.2 of the DET-RD.2 examination volume VOL with respect to the second X-ray energy E2 is also determined. Here, DET-RD.2 is determined by receiving the second real image data set RD.2 via interface IF.

[0173] As another step, in the illustrated embodiment, a differential image dataset DD for examining the volume VOL is determined by applying a training function TF to the input data, wherein the input data is based on a first real image dataset RD.1 and a multi-real image dataset RD.M. Optionally, the input data may also be based on a second real image dataset RD.2.

[0174] In a first variant of the first embodiment, the first real image data group RD.1, the multi-functional real image data group RD.M, and the second real image data group RD.2 are all two-dimensional image data groups, and the differential image data group DD is a two-dimensional differential image data group. Here, the multi-functional real image data group RD.M includes the first real image data group and the second real image data groups RD.1 and RD.2. For example, this corresponds to... Figure 5 The data stream is shown below. Applicable is d = f1(b (1) , b (2) (where d represents the two-dimensional difference image data DD, b) (1) Represents the two-dimensional first real image data RD.1,b (2) Let f1 represent the second real image data RD.2, and f1 represent the training function TF. Here, the input data of the training function TF is implicitly based on the multi-real image data set RD.M in the following way: the input data includes the first real image data set and the second real image data sets RD.1 and RD.2.

[0175] In a second variation of the first embodiment, the first real image data group RD.1, the multi-functional real image data group RD.M, and the optional second real image data group RD.2 are all three-dimensional image data groups, and the differential image data group DD is a three-dimensional differential image data group. For example, this corresponds to... Figure 6 The data stream is shown below. Applicable is D = f2(B (1) B (2) B (m) ) or D = f2(B (2) B (m) (where D represents the three-dimensional difference image data set DD, B) (1) Represents the first three-dimensional real image data set RD.1, B (2) Let RD.2 represent the three-dimensional second real image data set, and B (m) Let f2 represent the three-dimensional multi-functional real image dataset RD.M, where f2 represents the training function TF.

[0176] In a third variation of the first embodiment, the first real image data group RD.1, the multi-energy real image data group RD.M, and the second real image data group RD.2 are all two-dimensional image data groups, and the differential image data group DD is a three-dimensional differential image data group. Here, the first real image data group RD.1 includes a plurality of first X-ray projections XP.1 with respect to a first X-ray energy E1, and the second real image data group RD.2 includes a plurality of second X-ray projections XP.2 with respect to a second X-ray energy E2. Furthermore, the multi-energy real image data group RD.M includes first X-ray projections and second X-ray projections XP.1 and XP.2, which in turn include the first real image data group and the second real image data group RD.1 and RD.2. For example, this corresponds to... Figure 7 The data stream is shown below. Applicable is D = f3(b (1) 1, …, b (1) m , b (2) 1, …, b (2) n (where D represents the three-dimensional differential image data set DD, b) (1) 1, …, b (1) m This represents the first two-dimensional real image data set RD.2 or the first X-ray projection XP.2, and b (2) 1, …, b (2) n Let f3 represent the second differential image data set RD.2 or the second X-ray projection XP.2, and f3 represent the training function TF. Here, m is the number of first X-ray projections XP.1, and n is the number of second X-ray projections XP.2. In particular, n = m is applicable, but n and m can also be different numbers. Here, the input data of the training function TF is implicitly based on the multi-energy real image data set RD.M in such a way that the input data is based on the first and second real image data sets RD.1 and RD.2.

[0177] Here, the training function TF is a neural network, especially a convolutional neural network, or a network that includes convolutional layers. Neural networks can, in particular, have an architecture known as the "U-Net" as described in O. Ronneberger, P. Fischer, and T. Brox's "U-Net: Convolutional Networks for Biomedical Image Segmentation," MICCAI, 2015.

[0178] The final step in the illustrated embodiment is to provide the PROV-DD differential image data set DD, which is provided here by means of the interface IF. Providing the PROV-DD differential image data set may in particular include displaying, storing, and / or transmitting the differential image data set DD.

[0179] Figure 13 A flowchart illustrating a second embodiment of a method for providing differential image data sets (DDs) is shown. The second embodiment follows... Figure 6 The data stream shown in the image.

[0180] The first step of the second embodiment is to receive REC-XP.1 via interface IF to check the first X-ray projection XP.1 of volume VOL with respect to the first X-ray energy E1, and to receive REC-XP.2 via interface IF to check the second X-ray projection XP.2 of volume VOL with respect to the second X-ray energy E2, where the first X-ray energy E1 and the second X-ray energy E2 are different.

[0181] Another step in the second exemplary embodiment is to determine a first true image data set RD.1 of the DET-RD.1 examination volume VOL with respect to the first X-ray energy E1, and to determine a second true image data set RD.2 of the DET-RD.2 examination volume VOL with respect to the second X-ray energy E2. Here, the first true image data set RD.1 of DET-RD.1 is determined by performing three-dimensional reconstruction of the first X-ray projection XP.1 using the computing unit CU, and the second true image data set RD.2 of DET-RD.2 is determined by performing three-dimensional reconstruction of the second X-ray projection XP.2 using the computing unit CU.

[0182] In the illustrated embodiment, 3D reconstruction is performed using filtered back projection. Alternatively, iterative reconstruction or reconstruction based on the Feldkamp algorithm is known.

[0183] In the second embodiment, in mathematical notation, the three-dimensional first real image data group RD.1 is represented by B. (1) = R(b (1) 1, …, b (1) m The second real image data set RD.2 is given by B. (2) = R(b (2) 1, …, b (2) n The function is given by ). Here, R represents the reconstruction function, and b (1) i Let represent the i-th (out of a total of m) first X-ray projection XP.1, and b (2) iLet XP.2 represent the i-th (out of a total of n) second X-ray projection.

[0184] Another step in the second embodiment is to determine the DET-RD.M examination volume VOL as a multi-energy true image data set RD.M with respect to the first X-ray energy E1 and the second X-ray energy E2 by means of a computing unit CU. Here, the DET-RD.M multi-energy true image data set RD.M is determined by three-dimensional reconstruction of the first X-ray projection XP.1 and the second X-ray projection XP.2. In the illustrated embodiment, three-dimensional reconstruction is performed by means of filtered back projection. Alternatively, iterative reconstruction or reconstruction based on the Feldkamp algorithm is known.

[0185] In the second embodiment, in mathematical notation, the three-dimensional multi-functional real image data set RD.M is represented by B. (m) = R(b (1) 1, …, b (1) m , b (2) 1, …, b (2) n (This is given.)

[0186] In the illustrated embodiment, the training function TF is applied to the three-dimensional first real image data set RD.1, the three-dimensional second real image data set RD.2, and the three-dimensional multi-real image data set RD.M, which serve as input data, and generates a three-dimensional probability data set as output data. In mathematical notation, W = f(B) (1) B (2) B (m) The three-dimensional probabilistic data set has the same voxel-measured extension as the three-dimensional multipotential real image data set RD.M in each of the three dimensions, thereby associating each voxel in the three-dimensional multipotential real image data set RD.M with a probability value. The probability values ​​associated with the voxels of the three-dimensional multipotential real image data set RD.M are specifically quantities for probabilities of: the contrast agent in the voxel-depicted examination volume VOL, or the blood vessels VES.1, VES.2 in the voxel-depicted examination volume VOL.

[0187] Another step in the second embodiment is to receive the REC-TRF transfer function via the interface IF, and to modify the probability data set of at least three dimensions of the MOD based on the transfer function via the computing unit CU. Here, the transfer function T: [0, 1] → [0, 1] is a function that maps probability values ​​to probability values, and in particular, a monotonically increasing function. Modifying the MOD by applying the transfer function voxel-by-voxel to the probability data set is applicable, where W' ijk = T(W) ijk = T(Wijk ), where W' is the modified probability data set.

[0188] Furthermore, in the aforementioned embodiment, the DET-DD differential image data set DD is determined by voxel-by-voxel multiplication of the modified probability data set with the multipotential real image data set RD.M; that is, where applicable, D = T(W)·B. (m) = T(f(B (1) B (2) B (m) ))·B (m) Alternatively, the modified MOD of the probability data set can be discarded. In this case, the differential image data set can be determined by multiplying the probability data set with the multipotential real image data set RD.M, i.e., by D = W·B. (m) =f(B (1) B (2) B (m) )·B (m) .

[0189] Figure 14 A flowchart illustrating a first embodiment of a method for providing a training function is shown. A first step of the first embodiment is to determine a first set of training real image data with respect to a first training X-ray energy, and to determine a multi-energy training real image data set with respect to both the first and second training X-ray energies, wherein the second training X-ray energy differs from the first training X-ray energy. Furthermore, the first embodiment includes an optional step of determining a second set of training real image data with respect to the second training X-ray energy. In the first embodiment, not only the first set of training real image data, but also the second set of training real image data and the multi-energy training real image data set are received via the training interface TIF.

[0190] Alternatively, the training interface can also receive a first training X-ray projection of the training check volume with respect to a first training X-ray energy, and the training computing unit can be used to determine a first training real image data set as a three-dimensional reconstruction of the first training X-ray projection. Furthermore, the training interface can also receive a second training X-ray projection of the training check volume with respect to a second training X-ray energy, and the training computing unit can be used to determine a second training real image data set as a three-dimensional reconstruction of the second training X-ray projection.

[0191] Another step in the first embodiment is to determine the comparative differential image data set for the DET-CDD training inspection volume. Here, the comparative differential image data set is specifically the differential image data set for the training inspection volume, and forms the ground truth of the training method. In the first embodiment, the comparative differential image data set is received via a training interface.

[0192] Another step in the first embodiment is to determine a training differential image data set for the DET-TDD training check volume by applying a training function to the input data, wherein the input data is based on a first training real image data set and a multi-potential training real image data set. Optionally, the input data is also based on a second training real image data set.

[0193] Here, the training function TF is a neural network, especially a convolutional neural network, or a network that includes convolutional layers. Neural networks can, in particular, have an architecture known as the "U-Net" as described in O. Ronneberger, P. Fischer, and T. Brox's "U-Net: Convolutional Networks for Biomedical Image Segmentation," MICCAI, 2015.

[0194] Here, the training differential image data set and the comparison differential image data set have the same dimensions, and the expansion of the training differential image data set measured in pixels or voxels is the same with respect to each dimension as the expansion of the comparison differential image data set measured in pixels or voxels.

[0195] Another step in the first embodiment is to adjust the ADJ-TF training function based on a comparison of the training differential image data set and the comparison differential image data set. Specifically, the adjustment is based on a cost function that evaluates the deviation between the training differential image data set and the comparison differential image data set. The cost function can, in particular, be the sum of the squared differences of the individual pixels or voxels of the training differential image data set and the comparison differential image data set. In this embodiment, the training function is an artificial neural network, and adjusting the artificial neural network includes adjusting at least one edge weight of the artificial neural network, and the adjustment is based on the backpropagation algorithm.

[0196] The final step in the first embodiment shown is to provide a PROV-TF training function. In the illustrated embodiment, the training function is stored; alternatively, the training function (or one or more parameters of the training function) may be displayed or transmitted for further processing.

[0197] In the first embodiment, the first training real image data set, the second training real image data set, the training difference image data set, and the comparison difference image data set can be two-dimensional image data sets. Here, the data structure is similar to... Figure 5 In this case, the multi-faceted training real image data set includes a first training real image data set and a second training real image data set.

[0198] Alternatively, the first training real image data set, the second training real image data set, the multi-potential training real image data set, the training difference image data set, and the comparison difference image data set can be three-dimensional image data sets. Here, the data structure is similar to... Figure 6 .

[0199] Alternatively, the first and second training real image data sets can each include multiple two-dimensional X-ray projections, and the training difference image data set and the comparison difference image data set are three-dimensional image data sets. Here, the data structure is similar to... Figure 6 The multi-functional training real image dataset includes, in particular, a first training real image dataset and a second training real image dataset.

[0200] Figure 15 A flowchart illustrating a second embodiment of a method for providing a training function is shown. The second embodiment has... Figure 14 All steps of the first embodiment shown herein, and in particular, advantageous implementations and improvements described therein are also possible.

[0201] The illustrated embodiment further includes receiving a first three-dimensional material model of the DET-MM.1 training examination volume and a second three-dimensional material model of the DET-MM.2 training examination volume, specifically received via the training interface TIF. Here, the first three-dimensional material model is a material model including the training examination volume of the contrast agent, and the second three-dimensional material model is a material model without the contrast agent in the training examination volume.

[0202] In the second embodiment, the material model describes the three-dimensional spatial distribution of the energy-related X-ray absorption coefficient μ(x, E). In this embodiment, the material model is continuous, meaning it is a function of the three-dimensional spatial coordinate x, particularly a continuous function or, particularly, a differentiable function of the spatial coordinate x. Alternatively, the material model can also be spatially discrete, meaning it comprises a set of voxels, each associated with an energy-related X-ray absorption coefficient μ(E). Therefore, particularly in the regularization of voxels, the energy-related X-ray absorption coefficient μ(E) can be used to... ijk(E) is used to describe the material model. Furthermore, the material model can be defined for any number of X-ray energies E, but it is sufficient to describe the material model only for the first training X-ray energy and the second training X-ray energy, i.e., for the first training X-ray energy μ. (1) (x) or μ (1) ijk And for the second training X-ray energy of μ (2) (x) or μ (2) ijk .

[0203] In the second embodiment shown, the first training real image data TRD.1 and the second training real image data TRD.2 are respectively sets of three-dimensional image data including 256·256·256 voxels for the training check volume, and the multi-functional training real image data TRD.M is a set of three-dimensional image data including 512·512·512 voxels for the check volume, and the first material model also includes 512·512·512 voxels. Therefore, the first training real image data set TRD.1 and the second training real image data set TRD.2 serve as... Computation, and multi-functional training of real image datasets TRD.M as calculate.

[0204] Alternatively, the two-dimensional first training X-ray projection and the two-dimensional second training X-ray projection can be determined based on the first material model, wherein the first training X-ray projection corresponds to the X-ray projection of the training examination volume with respect to the first training X-ray energy, and wherein the second training X-ray projection corresponds to the X-ray projection of the training examination volume with respect to the second training X-ray energy. Therefore, in particular, a three-dimensional first training real image dataset can be reconstructed based on the first training X-ray projection, a three-dimensional second training real image dataset can be reconstructed based on the second training X-ray projection, and a multi-energy training real image dataset can be reconstructed based on both the first and second training X-ray projections. In this case, the training X-ray projection is determined by the equation...

[0205]

[0206] It is derived that, if the projection direction corresponds to angle v, then Γ(y, v) is the path from the X-ray source to the X-ray detector at the y-coordinate. In this case, the first material model can also be modeled in a time-variable manner to simulate the time-varying density of the contrast agent in the training examination volume.

[0207] The second embodiment shown further includes: determining a set of masked image data for the DET-MD training inspection volume. In this case, the set of masked image data is determined based on a second three-dimensional material model, for example, as... (Here, ν represents the second material model). Alternatively, masked image data sets can be received directly via the training interface.

[0208] In particular, the mask image data set can also be determined from the three-dimensional reconstruction of a two-dimensional X-ray projection, wherein the mask image data set can be determined by...

[0209]

[0210] Confirmed. Here, the masked image data set can be based solely on the X-ray projection m with respect to the first training X-ray energy. (1) Based solely on the X-ray projection m with respect to the second X-ray training energy (2) Or not only based on X-ray projections of the energy of the first training X-rays. (1) Furthermore, it is also based on the X-ray projection m of the second training X-ray energy. (2) To determine.

[0211] In the illustrated embodiment, the DET-CDD comparison differential image data set is then determined using digital subtraction angiography based on the masked image data set and the multi-potential training real image data set, i.e., via B... (m) ijk – M ijk Alternatively, the differential image data set can be determined by digital subtraction angiography based on a masked image data set and a first training real image data set, or by digital subtraction angiography based on a masked image data set and a second training real image data set.

[0212] Figure 16 The system PRVS is shown. Figure 17 The training system TRS is shown. The provided system PRVS shown is configured to implement the method for providing differential image datasets DD according to the present invention. The training system shown is configured to implement the method for providing training functions TF according to the present invention. The provided system PRVS includes an interface IF, a computation unit CU, and a storage unit MU, and the training system TRS includes a training interface TIF, a training computation unit TCU, and a training storage unit TMU.

[0213] The PRVS and / or TRS provided can be, in particular, computers, microcontrollers, or integrated circuits. Alternatively, the PRVS and / or TRS provided can be a real or virtual cluster of computers (the technical term for a real cluster is "Cluster," and the technical term for a virtual cluster is "Cloud"). The PRVS and / or TRS provided can also be configured as a virtual system running on a real computer or a real or virtual cluster of computers (the technical term is "Virtualization").

[0214] The interface IF and / or training interface TIF can be hardware or software interfaces (e.g., PCI bus, USB, or FireWire). The compute unit CU and / or training compute unit TCU can have hardware or software components, such as a microprocessor or a so-called FPGA (Field Programmable Gate Array). The storage unit MU and / or training storage unit TMU can be implemented as non-persistent working memory (Random Access Memory, or RAM) or as persistent mass storage (hard disk, USB stick, SD card, solid-state drive).

[0215] The interface IF and / or training interface TIF can in particular include multiple sub-interfaces that implement different steps of the corresponding method. In other words, the interface IF and / or training interface TIF can also be understood as multiple interface IFs or multiple training interface TIFs. The computation unit CU and / or training computation unit TCU can in particular include multiple sub-computation units that implement different steps of the corresponding method. In other words, the computation unit CU and / or training computation unit TCU can also be understood as multiple computation units CU or multiple training computation units TCU.

[0216] Figure 18 An embodiment of the X-ray device XSYS is shown. Here, the X-ray device XSYS is configured as a dual C-arm X-ray device. The X-ray device includes a first C-arm CA.1, with a first X-ray source SRC.1 disposed at a first end of the first C-arm CA.1, and a first X-ray detector DTC.1 disposed at a second end of the first C-arm CA.1. The X-ray device also includes a second C-arm CA.2, with a second X-ray source SRC.2 disposed at a first end of the second C-arm CA.2, and a second X-ray detector disposed at a second end of the second C-arm CA.2. The first C-arm CA.1 is disposed at a first suspension MNT.1, wherein the first suspension is configured as a multi-axis articulated robot. The second C-arm CA.2 is disposed at a second suspension MNT.2, wherein the second suspension includes a cover fixing member.

[0217] The first X-ray source SRC.1 and the second X-ray source SRC.2 are, in particular, X-ray tubes having the same anode material. The first X-ray detector DTC.1 and the second X-ray detector are, in particular, flat panel detectors.

[0218] Here, X-ray sources SRC.1 and SRC.2 and X-ray detectors DTC.1 and DTC.2 are configured for rotation about the imaging axis IA, particularly circular rotation about the imaging axis IA. Here, the imaging axis IA intersects specifically with the examination volume VOL. During rotation about the imaging axis, the X-ray sources SRC.1 and SRC.2 and the X-ray detectors move in the imaging plane IP, which is orthogonal to the imaging axis IA. The X-ray sources SRC.1 and SRC.2 and the X-ray detectors DTC.1 and DTC.2 are specifically configured for rotation about the imaging axis IA in such a way that the C-arms CA.1 and CA.2 are configured for rotation about the imaging axis IA.

[0219] The X-ray equipment XSYS also includes a patient support device PPOS, which is configured to support the patient PAT. The patient PAT can be moved along the imaging axis IA by means of the patient support device.

[0220] Where none are explicitly stated but are meaningful and within the scope of the invention, various embodiments, their sub-aspects, or features may be combined or substituted with each other without departing from the scope of the invention. The advantages described in the reference embodiments of the invention also apply to other embodiments where applicable, unless explicitly enumerated.

[0221] According to embodiments of this disclosure, the following notes are also disclosed:

[0222] 1. A computer-implemented method for providing a differential image data set (DD) of the examination volume (VOL),

[0223] - Determine the first true image data set (RD.1) of the examination volume (VOL) with respect to the first X-ray energy (DET-RD.1).

[0224] - Determine (DET-RD.M) the examination volume (VOL) with respect to the first X-ray energy and the second X-ray energy of the multi-energy true image data set (RD.M).

[0225] The energy of the second X-ray is different from that of the first X-ray.

[0226] - The differential image dataset (DD) for determining the examination volume (VOL) is determined by applying a training function (TF) to the input data, wherein the input data is based on the first real image dataset (RD.1) and the multi-real image dataset (RD.M), and

[0227] - Provide (PROV-DD) the differential image data set (DD).

[0228] 2. The method according to Appendix 1, further comprising:

[0229] - Determine the second true image data set (RD.2) of the examination volume (VOL) with respect to the second X-ray energy (DET-RD.2);

[0230] The input data is also based on the second set of real image data (RD.2).

[0231] 3. The method according to Appendix 1 or 2, further comprising:

[0232] - Receive (REC-XP.1) the first X-ray projection (XP.1) of the examination volume (VOL) with respect to the first X-ray energy.

[0233] - Receive (REC-XP.2) the second X-ray projection of the examination volume (VOL) with respect to the second X-ray energy (XP.2);

[0234] The first real image data set (RD.1) is based on the first X-ray projection (XP.1) and / or includes the first X-ray projection; and / or

[0235] The second real image data set (RD.2) is based on the second X-ray projection (XP.2) and / or includes the second X-ray projection; and / or

[0236] The multi-energy real image data set (RD.M) is based on the first X-ray projection (XP.1) and the second X-ray projection (XP.2) and / or includes the first X-ray projection and the second X-ray projection.

[0237] 4. According to the method described in Appendix 3,

[0238] Wherein the first real image data set (RD.1) is at least a three-dimensional reconstruction of the first X-ray projection (XP.1); and / or

[0239] The second set of real image data (RD.2) is at least a three-dimensional reconstruction of the second X-ray projection (XP.2); and / or

[0240] The multi-energy real image data set (RD.M) is at least a three-dimensional reconstruction of the first X-ray projection (XP.1) and the second X-ray projection (XP.2).

[0241] 5. According to the method described in Appendix 3 or 4,

[0242] When recording the first X-ray projection (XP.1), the examination volume (VOL) includes a contrast agent, and / or, when recording the second X-ray projection (XP.2), the examination volume (VOL) includes a contrast agent.

[0243] 6. The method according to any one of Appendices 3 to 5,

[0244] The first X-ray projection (XP.1) and the second X-ray projection (XP.2) are recorded simultaneously.

[0245] 7. The method according to any one of Appendices 3 to 6,

[0246] The first X-ray projection (XP.1) is a record of the first X-ray source (SRC.1) and the first X-ray detector (DTC.1), and the second X-ray projection (XP.2) is a record of the second X-ray source (SRC.2) and the second X-ray detector (DTC.2).

[0247] 8. According to the method described in Appendix 7,

[0248] The dual-plane X-ray device (XSYS) includes the first X-ray source (SRC.1), the second X-ray source (SRC.2), the first X-ray detector (DTC.1), and the second X-ray detector (DTC.2).

[0249] 9. The method according to any one of Appendices 3 to 8,

[0250] Each of the first X-ray projections (XP.1) is an X-ray projection of the examination volume (VOL) with respect to the projection directions (v(1)1, v(1)2, v(1)3) within the first projection angle range (PA.1).

[0251] Each of the second X-ray projections (XP.2) is an X-ray projection of the examination volume (VOL) with respect to the projection directions (v(2)1, v(2)2, v(2)3) within the second projection angle range (PA.2).

[0252] Furthermore, the first projection angle range (PA.1) is different from the second projection angle range (PA.2).

[0253] 10. According to the method described in Appendix 9,

[0254] The first projection angle range (PA.1) and the second projection angle range (PA.2) do not intersect.

[0255] 11. According to the method described in Appendix 10,

[0256] The overlap between the first projection angle range and the second projection angle range includes at least 50% of the first projection angle range and / or the second projection angle range, particularly at least 75% of the first projection angle range and / or the second projection angle range, and particularly at least 90% of the first projection angle range and / or the second projection angle range.

[0257] 12. The method according to any one of the foregoing notes,

[0258] The output data of the training function (TF) includes a set of probability data, and the differential image data set (DD) is based on the set of probability data.

[0259] 13. The method according to Appendix 12, further comprising:

[0260] - Receive (REC-TRF) transfer function,

[0261] - Modify (MOD) at least three-dimensional probabilistic data sets based on the transfer function.

[0262] 14. According to the method described in Appendix 12 or 13,

[0263] The differential image data set (DD) is based on the product of the probability data set and the first real image data set (RD.1) and / or the second real image data set (RD.2) and / or the multi-potential real image data set (RD.M).

[0264] 15. A computer-implemented method for providing a training function, the method comprising:

[0265] - Determine (DET-TRD.1) the training check volume with respect to the first training X-ray energy of the first set of real image data.

[0266] - Determine (DET-TRD.M) the training check volume with respect to the multi-energy training real image data set with respect to the first training X-ray energy and the second training X-ray energy.

[0267] The energy of the second training X-ray is different from that of the first training X-ray.

[0268] - Determine the comparative differential image data set for the training check volume described in (DET-CDD).

[0269] - The training differential image data set for determining the training check volume (DET-TDD) is determined by applying the training function to the input data, wherein the input data is based on the first training real image data set and the multi-potential training real image data set.

[0270] - Adjust the training function (TF) based on the comparison of the training differential image data set and the comparison differential image data set (ADJ-TF).

[0271] - Provide (PROV-TF) the training function (TF).

[0272] 16. The method according to Appendix 15, further comprising:

[0273] - Determine (DET-MD) the masked image data set for the training inspection volume;

[0274] The comparative differential image data set is determined by digital subtraction angiography based on the mask image data set and the first training real image data set, or by digital subtraction angiography based on the mask image data set and the multipotential training real image data set.

[0275] 17. The method according to Appendix 15 or 16, further comprising:

[0276] - Receive the first 3D material model of the training check volume as described in (REC-MM.1).

[0277] The first training real image data set and / or the multi-energy training real image data set are simulations based on the interaction between X-ray radiation and the first three-dimensional material model.

[0278] 18. The method according to Appendix 17, further comprising:

[0279] - Receive the second three-dimensional material model of the training check volume described in (REC-MM.2).

[0280] The first three-dimensional material model is a material model of the training examination volume, including the contrast agent.

[0281] The second three-dimensional material model is a material model of the training examination volume without contrast agent.

[0282] The masked image data set is based on a simulation of the interaction between X-ray radiation and the second three-dimensional material model.

[0283] 19. A system for providing differential image data sets (DDs) of an examination volume (VOL), the system comprising an interface (IF) and a computing unit (CU).

[0284] - The interface (IF) and / or the computing unit (CU) constitute a first set of true image data (RD.1) for determining the examination volume (VOL) with respect to the first X-ray energy (DET-RD.1).

[0285] - The interface (IF) and / or the computing unit (CU) further constitute a multi-energy real image data set (RD.M) for determining the examination volume (VOL) with respect to the first X-ray energy and the second X-ray energy (DET-RD.M).

[0286] The energy of the second X-ray is different from that of the first X-ray.

[0287] - The computational unit (CU) further constitutes the differential image dataset (DD) for determining the examination volume (VOL) by applying a training function (TF) to the input data (DET-DID), wherein the input data is based on the first real image dataset (RD.1) and the multi-real image dataset (RD.M), and

[0288] - The interface (IF) therein also constitutes a means for providing (PROV-DD) the differential image data set (DD).

[0289] 20. An X-ray apparatus (XSYS) comprising a provisioning system as described in Appendix 19.

[0290] 21. A training system for providing a training function, the training system comprising a training interface (TIF) and a training computation unit (TCU).

[0291] - The training interface (TIF) and / or the training computation unit (TCU) constitute a first set of training real image data for determining (DET-TRD.1) the training check volume with respect to the first training X-ray energy.

[0292] - The training interface (TIF) and / or the training computation unit (TCU) further constitute a multi-energy training real image data set for determining (DET-TRD.M) the training check volume with respect to the first training X-ray energy and the second training X-ray energy.

[0293] The energy of the second training X-ray is different from that of the first training X-ray.

[0294] - The training interface (TIF) and / or the training computation unit (TCU) further constitute a set of comparative differential image data for determining the (DET-CDD) training check volume.

[0295] - The training computation unit (TCU) further constitutes a training differential image data set for determining the training check volume (DET-TDD) by applying the training function (TF) to the input data.

[0296] The input data is based on the first set of real training image data and the multi-potential training real data set.

[0297] - The training computation unit (TCU) further comprises a function (TF) configured to adjust (ADJ-TF) the training function (TF) based on a comparison of the training differential image data set with the comparison differential image data set.

[0298] - The training interface described therein also constitutes a training function (TF) for providing (PROV-TF).

[0299] 22. A computer program product comprising a computer program capable of being directly loaded into the memory (MU) of a providing system (PRVS), the computer program having program segments for implementing all steps of the method according to any one of Appendices 1 to 14 when the program segments are run by the providing system (PRVS); and / or the computer program capable of being directly loaded into the training memory (TMU) of a training system (TRS), the computer program having program segments for implementing all steps of the method according to any one of Appendices 15 to 18 when the program segments are run by the training system (TRS).

[0300] 23. A computer-readable storage medium storing on the computer-readable storage medium program segments readable and operable by a providing system (PRVS) so that, when the providing system (PRVS) runs the program segments, all steps of the method according to any one of Appendices 1 to 14 are implemented; and / or storing on the computer-readable storage medium program segments readable and operable by a training system (TRS) so that, when the training system (TRS) runs the program segments, all steps of the method according to any one of Appendices 15 to 18 are implemented.

[0301] 24. A computer program or computer-readable storage medium comprising a training function (TF) provided by means of the method described in accordance with Appendices 15 to 18.

Claims

1. A computer-implemented method for providing a differential image data set (DD) of the examination volume (VOL), - Receive (REC-XP.1) the first X-ray projection (XP.1) of the examination volume (VOL) with respect to the first X-ray energy. - Receive (REC-XP.2) the second X-ray projection of the examination volume (VOL) with respect to the second X-ray energy (XP.2); - Determine (DET-RD.1) the first set of true image data (RD.1) of the examination volume (VOL) with respect to the first X-ray energy, wherein the first set of true image data (RD.1) is based on the first X-ray projection (XP.1) and / or includes the first X-ray projection. - Determine (DET-RD.M) the examination volume (VOL) with respect to the first X-ray energy and the second X-ray energy as a set of multi-energy true image data (RD.M), wherein the set of multi-energy true image data (RD.M) is based on the first X-ray projection (XP.1) and the second X-ray projection (XP.2) and / or includes the first X-ray projection and the second X-ray projection. The energy of the second X-ray is different from that of the first X-ray. - The differential image dataset (DD) for determining the examination volume (VOL) is determined by applying a training function (TF) to the input data, wherein the input data is based on the first real image dataset (RD.1) and the multi-real image dataset (RD.M), and - Provide (PROV-DD) the differential image data set (DD). The first X-ray projection (XP.1) is a record of the first X-ray source (SRC.1) and the first X-ray detector (DTC.1), and the second X-ray projection (XP.2) is a record of the second X-ray source (SRC.2) and the second X-ray detector (DTC.2). The first X-ray projection (XP.1) and the second X-ray projection (XP.2) are recorded simultaneously. The dual-plane X-ray device (XSYS) includes the first X-ray source (SRC.1), the second X-ray source (SRC.2), the first X-ray detector (DTC.1), and the second X-ray detector (DTC.2). wherein each of the first X-ray projections (XP.1) is an X-ray projection of the examination volume (VOL) about a projection direction (v (1) 1、v (1) 2、v (1) 3) of a first projection angle range (PA.1), wherein each of the second X-ray projections (XP.2) is an X-ray projection of the examination volume (VOL) about a projection direction (v (2) 1、v (2) 2、v (2) 3) of a second range of projection angles (PA.2), Furthermore, the first projection angle range (PA.1) is different from the second projection angle range (PA.2).

2. The method according to claim 1, further comprising: - Determine (DET-RD.2) the second true image data set (RD.2) of the examination volume (VOL) with respect to the second X-ray energy; The input data is also based on the second set of real image data (RD.2).

3. The method according to claim 1 or 2, The second real image data set (RD.2) is based on the second X-ray projection (XP.2) and / or includes the second X-ray projection.

4. The method according to claim 3, Wherein the first real image data set (RD.1) is at least a three-dimensional reconstruction of the first X-ray projection (XP.1); and / or The second set of real image data (RD.2) is at least a three-dimensional reconstruction of the second X-ray projection (XP.2); and / or The multi-energy real image data set (RD.M) is at least a three-dimensional reconstruction of the first X-ray projection (XP.1) and the second X-ray projection (XP.2).

5. The method according to claim 1 or 2, When recording the first X-ray projection (XP.1), the examination volume (VOL) includes a contrast agent, and / or, when recording the second X-ray projection (XP.2), the examination volume (VOL) includes a contrast agent.

6. The method according to claim 1 or 2, The first projection angle range (PA.1) and the second projection angle range (PA.2) do not intersect.

7. The method according to claim 6, The overlap between the first projection angle range and the second projection angle range includes at least 50% of the first projection angle range and / or the second projection angle range.

8. The method according to claim 6, The overlap between the first projection angle range and the second projection angle range includes at least 75% of the first projection angle range and / or the second projection angle range.

9. The method according to claim 6, The overlap between the first projection angle range and the second projection angle range includes at least 90% of the first projection angle range and / or the second projection angle range.

10. The method according to claim 1 or 2, The output data of the training function (TF) includes a set of probability data, and the differential image data set (DD) is based on the set of probability data.

11. The method according to claim 10, further comprising: - Receive (REC-TRF) transfer function, - Modify (MOD) at least three-dimensional probabilistic data sets based on the transfer function.

12. The method according to claim 10, The differential image data set (DD) is based on the product of the probability data set and the first real image data set (RD.1) and / or the second real image data set (RD.2) and / or the multi-potential real image data set (RD.M).

13. A computer-implemented method for providing a training function, the method comprising: - Determine (DET-TRD.1) a first set of training real image data with respect to a first training X-ray energy, wherein the first set of training real image data is based on and / or includes the first training X-ray projection of the training check volume with respect to the first training X-ray energy, wherein the first training X-ray projection is a record of the first X-ray source and the first X-ray detector. - Determine (DET-TRD.M) a set of multi-energy training real image data with respect to the first training X-ray energy and the second training X-ray energy, wherein the set of multi-energy training real image data is based on the first training X-ray projection and the second training X-ray projection of the training examination volume with respect to the second training X-ray energy, and / or includes the first training X-ray projection and the second training X-ray projection, wherein the second training X-ray projection is a record of the second X-ray source and the second X-ray detector. The energy of the second training X-ray is different from that of the first training X-ray. - Determine the comparative differential image data set for the training check volume described in (DET-CDD). - The training differential image data set for determining the training check volume (DET-TDD) is determined by applying the training function to the input data, wherein the input data is based on the first training real image data set and the multi-potential training real image data set. - Adjust the training function (TF) based on the comparison of the training differential image data set and the comparison differential image data set (ADJ-TF). - Provide the training function (TF) described in (PROV-TF). The first training X-ray projection and the second training X-ray projection are recorded simultaneously. The dual-plane X-ray device includes a first X-ray source, a second X-ray source, a first X-ray detector, and a second X-ray detector. Each of the first training X-ray projections is a training X-ray projection of the training examination volume with respect to the training projection direction within the range of first training projection angles. Each of the second training X-ray projections is a training X-ray projection of the training examination volume with respect to the training projection direction within the range of second training projection angles. Furthermore, the first training projection angle range is different from the second training projection angle range.

14. The method according to claim 13, further comprising: - Determine (DET-MD) the masked image data set for the training inspection volume; The comparative differential image data set is determined by digital subtraction angiography based on the mask image data set and the first training real image data set, or by digital subtraction angiography based on the mask image data set and the multipotential training real image data set.

15. The method according to claim 13 or 14, further comprising: - Receive the first 3D material model of the training check volume as described in (REC-MM.1). The first training real image data set and / or the multi-energy training real image data set are simulations based on the interaction between X-ray radiation and the first three-dimensional material model.

16. The method according to claim 15, further comprising: - Receive the second three-dimensional material model of the training check volume described in (REC-MM.2). The first three-dimensional material model is a material model of the training examination volume, including the contrast agent. The second three-dimensional material model is a material model of the training examination volume without contrast agent. The masked image data set is based on a simulation of the interaction between X-ray radiation and the second three-dimensional material model.

17. A system for providing differential image data sets (DDs) of inspection volume (VOL), the system comprising an interface (IF) and a computing unit (CU). - The interface (IF) and / or the computing unit (CU) therein constitute a first X-ray projection (XP.1) of the examination volume (VOL) with respect to the first X-ray energy. - The interface (IF) and / or the computing unit (CU) therein constitute a second X-ray projection (XP.2) of the examination volume (VOL) with respect to the second X-ray energy. - The interface (IF) and / or the computing unit (CU) constitute a first true image data set (RD.1) for determining the examination volume (VOL) with respect to the first X-ray energy (DET-RD.1), wherein the first true image data set (RD.1) is based on the first X-ray projection (XP.1) and / or includes the first X-ray projection. - The interface (IF) and / or the computing unit (CU) further constitute a multi-energy true image data set (RD.M) for determining the examination volume (VOL) with respect to the first X-ray energy and the second X-ray energy (DET-RD.M), wherein the multi-energy true image data set (RD.M) is based on the first X-ray projection (XP.1) and the second X-ray projection (XP.2) and / or includes the first X-ray projection and the second X-ray projection. The energy of the second X-ray is different from that of the first X-ray. - The computational unit (CU) further constitutes the differential image dataset (DD) for determining the examination volume (VOL) by applying a training function (TF) to the input data (DET-DID), wherein the input data is based on the first real image dataset (RD.1) and the multi-real image dataset (RD.M), and - The interface (IF) described therein also constitutes a means for providing (PROV-DD) the differential image data set (DD). - Wherein the first X-ray projection (XP.1) is a record of the first X-ray source (SRC.1) and the first X-ray detector (DTC.1), and wherein the second X-ray projection (XP.2) is a record of the second X-ray source (SRC.2) and the second X-ray detector (DTC.2), The first X-ray projection (XP.1) and the second X-ray projection (XP.2) are recorded simultaneously. The dual-plane X-ray device (XSYS) includes the first X-ray source (SRC.1), the second X-ray source (SRC.2), the first X-ray detector (DTC.1), and the second X-ray detector (DTC.2). Each of the first X-ray projections (XP.1) is a projection direction (v) of the examination volume (VOL) with respect to the first projection angle range (PA.1). (1) 1. v (1) 2. v (1) 3) X-ray projection, Each of the second X-ray projections (XP.2) is a projection direction (v) of the examination volume (VOL) with respect to the second projection angle range (PA.2). (2) 1. v (2) 2. v (2) 3) X-ray projection, Furthermore, the first projection angle range (PA.1) is different from the second projection angle range (PA.2).

18. An X-ray apparatus (XSYS) comprising a supply system according to claim 17, a first X-ray source (SRC.1), a first X-ray detector (DTC.1), a second X-ray source (SRC.2), and a second X-ray detector (DTC.2).

19. A training system for providing a training function, the training system comprising a training interface (TIF) and a training computation unit (TCU). - The training interface (TIF) and / or the training computation unit (TCU) constitute a first set of training real image data for determining (DET-TRD.1) a training inspection volume with respect to a first training X-ray energy, wherein the first set of training real image data is based on and / or includes the first training X-ray projection of the training inspection volume with respect to the first training X-ray energy, wherein the first training X-ray projection is a record of the first X-ray source and the first X-ray detector. - The training interface (TIF) and / or the training computation unit (TCU) further constitute a multi-energy training real image data set for determining (DET-TRD.M) the training examination volume with respect to the first training X-ray energy and the second training X-ray energy, wherein the multi-energy training real image data set is based on the first training X-ray projection and the second training X-ray projection of the training examination volume with respect to the second training X-ray energy, and / or includes the first training X-ray projection and the second training X-ray projection, wherein the second training X-ray projection is a record of the second X-ray source and the second X-ray detector. The energy of the second training X-ray is different from that of the first training X-ray. - The training interface (TIF) and / or the training computation unit (TCU) further constitute a set of comparative differential image data for determining the (DET-CDD) training check volume. - The training computation unit (TCU) further constitutes a training differential image data set for determining the training check volume (DET-TDD) by applying the training function (TF) to the input data. The input data is based on the first set of real training image data and the multi-potential training real data set. - The training computation unit (TCU) further comprises a function (TF) configured to adjust (ADJ-TF) the training function (TF) based on a comparison of the training differential image data set with the comparison differential image data set. - The training interface described therein also constitutes a function (TF) for providing (PROV-TF). The first training X-ray projection and the second training X-ray projection are recorded simultaneously. The dual-plane X-ray device includes a first X-ray source, a second X-ray source, a first X-ray detector, and a second X-ray detector. Each of the first training X-ray projections is a training X-ray projection of the training examination volume with respect to the training projection direction within the range of first training projection angles. Each of the second training X-ray projections is a training X-ray projection of the training examination volume with respect to the training projection direction within the range of second training projection angles. Furthermore, the first training projection angle range is different from the second training projection angle range.

20. A computer program product having a computer program that can be directly loaded into the memory (MU) of a providing system (PRVS), the computer program having program segments to implement all steps of the method according to any one of claims 1 to 12 when the program segments are run by the providing system (PRVS); and / or the computer program being directly loaded into the training memory (TMU) of a training system (TRS), the computer program having program segments to implement all steps of the method according to any one of claims 13 to 16 when the program segments are run by the training system (TRS).

21. A computer-readable storage medium storing on the computer-readable storage medium a program segment readable and operable by a providing system (PRVS) so that, when the providing system (PRVS) runs the program segment, all steps of the method according to any one of claims 1 to 12 are implemented; and / or storing on the computer-readable storage medium a program segment readable and operable by a training system (TRS) so that, when the training system (TRS) runs the program segment, all steps of the method according to any one of claims 13 to 16 are implemented.

22. A computer program or computer-readable storage medium comprising a training function (TF) provided by the method according to claims 13 to 16.

Images