Method and apparatus for imaging in computed tomography

Abstract

The invention relates to a method for imaging in computed tomography, the method comprising the following steps: - providing a CT data set (D) of an object (O), wherein the CT data set (D) has been recorded by means of a multispectral recording method; - suppressing contrast caused by tissue types (G1, G2) and generating a contrast-suppressed data set (K); - analyzing the CT data set (D) and / or the contrast-suppressed data set (K) by means of a machine learning algorithm, wherein the analysis is configured to identify at least one change within the tissue types (G1, G2). The invention also relates to a corresponding device (6), a control device (10) for a computed tomography system (1) or a diagnostic system, as well as a diagnostic system and a computed tomography system (1).

Description

Technical Field

[0001] The invention relates to a method and an apparatus for imaging in computed tomography, in particular to a method for identifying and quantitatively evaluating different tissue components by using a combination of multispectral (photon counting) data recording with iterative data reconstruction and artificial intelligence, in particular to predict the benefit of various treatment methods using cerebral ischemia as an example. Here, the invention is used for imaging, which means that the invention can be applied directly in the context of image reconstruction, the invention can be used to process reconstructed images, or additional data can be generated in addition to or in relation to the original data set in preparation for reconstruction. Background Art

[0002] Until now, it has not been possible to satisfactorily represent all areas of the human body with appropriate measurement times for all possible medical indications. For example, in the case of ischemic stroke, the criterion for treatment decisions has always been the time that has passed since the stroke occurred, because more precise examinations cannot be effectively achieved with the aid of imaging.

[0003] Intravenous drugs (r-tPA) should be administered within a time window of up to 4.5 hours after the occurrence of ischemic stroke and no more than a certain degree of damage to brain tissue, which is intended to dissolve vascular occlusions (clots). Since the successful completion of various studies on mechanical reopening of large blood vessels (mechanical thrombectomy) in 2014, this treatment is also allowed within up to 6 hours after the occurrence of vascular occlusion. The latest studies on mechanical thrombectomy show that mechanical reopening of blood vessels in a specific subgroup of patients is still expected to succeed even within 24 hours after the infarction. The criteria for this are evidence of occlusion of the large cerebral arteries and so-called "tissues at risk" (i.e., brain tissue with insufficient blood supply (minderdurchblutet) that is dying). At the same time, the volume of the infarct core (Infarktkern) (i.e., irreversibly damaged brain tissue) cannot be too large compared to the "tissue at risk". On this basis, recent studies have proposed the following hypothesis: drug treatment of vascular occlusion after exceeding the 4.5-hour time window can also be expected to succeed. In particular, this applies to the patient group of so-called "slow progressors", i.e. patients in whom the volume of dead brain tissue increases only slowly over time, because there is residual perfusion through so-called collateral vessels. Currently, current studies are investigating the benefit of perfusion imaging for patient selection, but without taking into account any other pathophysiological parameters.

[0004] In treatment, especially in drug thrombolytic therapy, the main risk for patients is bleeding into the infarct core. This is because the blood vessel wall in the area of ​​insufficient blood supply will become brittle, because cells die and lose adhesion due to reduced oxygen supply. The proteolytic enzymes released during cell death will aggravate this effect. Therefore, the blood vessel wall can no longer withstand the hypertension that occurs after reperfusion therapy. As time goes by, this risk will increase. Therefore, it is desirable to have an algorithm as a decision-making aid, which allows to make a statement (Aussage) about whether the blood vessel wall has become brittle or tends to become brittle. Therefore, it is possible to determine the integrity of the blood vessel wall and identify the increased bleeding risk early. The increased bleeding risk, for example, constitutes a contraindication to intravenous thrombolytic therapy after 4.5 hours. In this regard, changes in the blood vessel wall that manifest as edema, fractures or bleeding and analysis and quantification of different tissue components help to make decisions.

[0005] Another decision criterion is the development of cerebral edema and its spread in the area of ​​insufficient blood supply. Due to the different locations of the ischemic areas and the associated different distribution of gray and white matter, the spread of cerebral edema is not always the same. Therefore, it is difficult to compare and state for individual patients.

[0006] In addition to detecting the occlusion of the large arteries supplying the brain as the basis for mechanical thrombectomy, the success of treatment is currently only evaluated by detecting certain parameters. These parameters are the so-called "tissue at risk" or penumbra (insufficient blood supply to tissue without causing cell death), the infarct core and the brain tissue that is regularly supplied with blood. Only when there is a sufficient number of tissues that are insufficiently supplied with blood but are normally supplied after the blood supply is restored and the infarct core is small enough at the same time, treatment is performed in the form of drug thrombolysis or mechanical thrombectomy. Since the risk of hemorrhage in the infarct core increases with time, the "onset to treatment time" of more than 4.5 hours is currently a limiting factor for intravenous thrombolysis with r-tPA. Therefore, although treatment has been successfully carried out even if the time window exceeds 4.5 hours in individual cases, there are still not many patients who do not receive intravenous thrombolysis within this time window, and some patients are also accompanied by so-called "wake-up" strokes.

[0007] No further information is available about the integrity of the vessel wall. Therefore, if thrombolysis is given beyond the 4.5-hour window, it is not possible to reliably predict whether a patient will bleed into the infarct area. Previous studies have shown that there is a higher risk of bleeding in this setting.

[0008] This example represents a medical event in which imaging is currently not possible or can be performed only to a limited extent, but in which parameters must nevertheless be derived from the clinical images in order to define an optimal therapy. Summary of the invention

[0009] The object of the present invention is to provide a method and a device for imaging in computed tomography, in particular a method and a device for controlling a computed tomography system or a diagnostic system, with which the disadvantages of the prior art are overcome and with which improved imaging is achieved with regard to improved possibilities for tissue characterization. In particular, the object of the present invention is to separate (substantially all) tissue structures with small density differences, preferably tissue structures in brain tissue.

[0010] This object is achieved by a method, a device, a control device, a diagnostic system and a computed tomography system according to the invention.

[0011] The method according to the invention relates to an improvement in CT imaging and is used for better differentiation of different tissue types of an object. In particular, diseases with changes in the water content (formation of edema) can be identified in areas with different tissue types in close proximity to each other. Here, a disease can spread to both tissue types and have different therapeutic or diagnostic significance at the two tissue types, for example for tumor diseases or bone fractures.

[0012] The object can be an inanimate element, but is preferably a part of a (human or animal) patient. The object to be examined can be understood as a specific area, such as a part or organ in the body, which is also often referred to as a region of interest (Region of Interest, RoI). The tissue types (or "tissue types", such as gray matter and white matter) to be distinguished should be understood as tissue characteristics of interest (also called "compartments", such as the consistency of white matter or gray matter or the consistency of the liver). These are usually water content and fat content. In some cases, such as in the ventricular system, the region of interest and the tissue type (compartment) can be the same.

[0013] The method according to the invention for imaging in computed tomography or for controlling a computed tomography system or a diagnostic system comprises the following steps:

[0014] - A CT data set of the object is provided, wherein the CT data set has been recorded by means of a multispectral recording method, for example by means of a dual-spectral or multispectral recording method, in particular by means of a dual-source dual-energy method or a multi-source / dual-source multi-energy method, particularly preferably by means of a photon counting method. Here, the computed tomography data set comprises raw data or (reconstructed) image data of the object to be examined. The possibilities of multispectral recording methods are known to the person skilled in the art. However, it should be noted that the spectral and spatial resolution as well as the dose efficiency are sufficient.

[0015] The data sets can be provided with the aid of a radiology information system (RIS), wherein the data sets can already be directly recorded or can be called up from a PACS (Picture Archiving and Communication System).

[0016] -Suppress the contrast caused by (at least) one tissue type, preferably suppress the contrast caused by two different tissue types, and generate a data set with suppressed contrast. For example, when examining the brain, suppress the contrast of gray and white, that is, the contrast of gray matter and white matter. In order to obtain the best display effect, the contrast of all known and healthy tissue types present in the RoI should actually be suppressed. However, this is not absolutely necessary, because known tissue types can also be identified by corresponding markings (and then ignored in subsequent analysis). However, it is important that the image signal of the following healthy tissue should be suppressed, and relevant examinations should be performed in such healthy tissue.

[0017] Since the image information of healthy tissue is hidden, changes (e.g. edema) in the examined tissue area can be visualized by contrast suppression. Even if processing of the (reconstructed) image data is preferred in this context, this step can still be performed within the preprocessing of the raw data before image reconstruction. When applied to different areas of the body, different contrasts usually need to be suppressed. In addition to one tissue type, contrast suppression can also be based on two, three or more tissue types. Contrast suppression provides advantageous contrast information that reflects changes in tissue types in the data set, such as edema content (intensity and affected area).

[0018] Pathologies are often covered by contrast and are "hidden" by contrast. For example, edema (water) in brain tissue is difficult to identify because the gray-white contrast overlaps the water / brain contrast, making the water / brain contrast difficult to identify. In addition, the degree of edema in gray matter and white matter is different and has different diagnostic significance. In general, for suppressing contrast, in the sense of the present invention, contrast that characterizes the corresponding tissue type but not the changes (to be examined) of the tissue type is suppressed when examining the tissue type.

[0019] The contrast suppression should preferably be based on the energy-dependent attenuation of radiation by different materials, so that these materials can be distinguished based on their energy-dependent attenuation. This makes it possible, for example, to distinguish between gray matter and white matter and to visualize the water content. For this purpose, it is preferred to distinguish between different tissue types (e.g., gray matter from white matter), which will be explained in more detail below in the context of iterative data processing.

[0020] If the method is used for different objects and pathologies (e.g. different areas of the human body), the CT values ​​(Hounsfieldwerte) of the corresponding tissue types should preferably be taken into account in order to suppress the contrast. For example, the same parameters can be used for brain tumors, but for liver tumors, site-specific thresholds must be taken into account. For example, the gray matter and white matter in the brain have energy-specific properties, which can be utilized in the context of contrast suppression. These properties are used to amplify or attenuate information selected from the corresponding raw data or image data (e.g. voxels from 3D images). Thus, by contrast suppression, for example, only information about the water content that is "hidden" in the gray / white noise can be seen. Therefore, it is generally preferred that the contrast suppression is based on the contrast between two (or more) different energy attenuations.

[0021] - analyzing at least the contrast suppressed data set or a data set based on the contrast suppressed data set (e.g. a synthetic data set referred to below) by means of a machine learning algorithm, wherein the analysis is configured to identify at least one change in a tissue type (or two or more different tissue types). A machine learning algorithm is an algorithm based on the principles of machine learning, which is sometimes also referred to as "artificial intelligence". However, the English name is often used in German.

[0022] Preferably, the analysis is configured so that a statement can be made about the likelihood of a predetermined symptom occurring due to the change or the likelihood of its course, such as the likelihood of a hemorrhage in brain tissue. Alternatively or additionally, the analysis is preferably configured so that a statement can be made about the likelihood of satisfaction with different treatment options and / or about the clinical prognosis. Thus, the method can assist in treatment decisions, which advantageously improves the reliability of diagnosis and treatment. This helps reduce complications and makes treatment more cost-effective.

[0023] Typically, the analysis is configured to obtain the properties of the highlighted area after contrast suppression. However, this does not necessarily refer to all areas that are still visible (e.g., the skull in the examination of brain tissue), but mainly refers to all areas that should be examined conventionally, that is, areas that should actually have other tissue properties. Relevant attributes are, for example, the shape, size, characteristics (e.g., Hounsfield units) or position of the relevant area. Preferably, the area to be examined is defined before the analysis or, for example, image information of tissue types that should not be examined is eliminated from the CT data set by contrast suppression, or, for example, at least these areas are marked by marking or masking. The analysis can of course also use the original CT data set or a data set derived therefrom in addition.

[0024] In the analysis of the structure and properties of the vessel wall, bleeding, swelling and changes in the tissue structure are preferably taken into account. An important piece of information obtained from this information is a statement about the probability of a rupture of the vessel after the blood circulation has been restored. Thus, for example, the probability of a life-threatening bleeding, in particular in the brain tissue, can be determined. In particular in the case of a stroke, this also helps to make the diagnosis and treatment more reliable.

[0025] Here, a "machine learning" method or artificial intelligence system may refer to a system for artificially generating knowledge from experience. Here, machine learning refers to a collection of mathematical methods for pattern recognition. These methods can achieve pattern recognition, for example, by decomposing the database as best as possible into a hierarchical structure (decision tree) or by obtaining similarities between data sets through vectors (e.g., k-nearest neighbors) and others.

[0026] A special form of machine learning is the so-called deep learning machine learning using artificial neural networks. Instead of the term "neural network", the term "neural net" can also be used. In particular, the trained function can also be a deep artificial neural network (English: deep neural network, deep artificial neural network) and / or a convolutional neural network (CNN). The structure of the artificial neural network used here is similar to a nervous system in which neurons are connected to each other in a mesh. The first layer of the neural network, the input layer, processes the input of the input data, such as a single pixel or a single voxel of an image or an image data set or an image parameter (e.g., grayscale value) of a group of pixels or a group of voxels. The first layer passes its output to the next layer. The second layer processes the information of the previous layer and also passes the result. The next layer receives the information of the second layer and further processes this information. These layers are called hidden layers. The features contained in these hidden layers become more and more abstract. The results are output in the output layer. In order to "learn" to process the input data, the artificial neural network is trained using a training data set. Corresponding methods (e.g., supervised learning, unsupervised learning, reinforcement learning) are known to those skilled in the art.

[0027] If the aim of the method is to determine the probability of bleeding (e.g. in brain tissue in the case of a stroke), in particular after administration of the relevant drug (e.g. r-tPA), the relevant machine learning algorithm is trained with the aid of a training data set that has been selected for predicting the probability of bleeding. Alternatively or additionally, however, training data sets of other treatment methods and / or predictions of the clinical course made by a physician can also be used.

[0028] In the case of stroke, according to current knowledge, the water content / edema content or its changes seem to have an influence on the probability of bleeding. Therefore, the water content in the different compartments is first highlighted by suppressing the contrast and then determined by analysis. From the acquired data, the probability of bleeding can also be determined within the scope of the analysis with the support of artificial intelligence and a treatment decision can be made.

[0029] Using artificial intelligence ("AI", machine learning algorithms) that are appropriately trained based on raw data or image data, the data set can be evaluated and, for example, compared with a patient set, the data of which have been used to train the algorithm. For this purpose, the clinical course and / or the imaging results should be known for this patient set. For example, a comparison of the pattern of edema characteristics, quantification of different tissue components and / or vessel wall properties in the current patient with a base patient set allows statements to be made about the likelihood of bleeding and the clinical outcome. Based on this, the attending physician can indicate an appropriate treatment method.

[0030] The advantage of the method is that a higher spatial resolution can be achieved with the aid of this imaging technique as a basis for improved image evaluation. Based on this basis and information about material properties that can be particularly well achieved as a result of photon counting techniques, it is possible to more accurately determine, for example, the distribution of water in the tissue (edema) and (for example, in terms of hemorrhage) the consistency and properties of the vessel wall, the fat content or calcium content, inflammation, porosity, etc. In addition, the individual components of the tissue can be quantified, which can be used to make decisions. In addition, different tissue types can be characterized and visualized (particularly through information obtained by contrast suppression). Thus, in terms of analysis, changes in brain tissue, such as the diffusion pattern of edema, can be analyzed (by differentially observing white matter and gray matter with different diffusion characteristics), and the analysis is performed in a manner comparable to a patient set whose distribution pattern and clinical course are known.

[0031] The quantification of the results is preferably performed by a combination of two values, namely, spatial extent and tissue properties, wherein the tissue properties are characterized by HU distribution (HU: Houndsfield Unit). Thus, the contrast suppression of the tissue structure leads to changes that can be normalized. Quantification allows comparison with a control group, so that the probability can be determined.

[0032] As will be described in detail below, for the present method, a multispectral photon counting recording method is preferably combined with iterative data reconstruction and in particular simultaneous phantom measurements. This particular combination allows for an extended analysis of certain pathologies, in particular an optimal separation of white and gray matter.

[0033] In photon counting computed tomography (photon counting CT), individual photons incident on a pixel detector are counted. Photon counting CT systems use photon counting detectors (PCDs), whose detector pixels can record and count individual photons. Additionally, photons can be counted within pre-set energy thresholds ("energy bins"). Thus, the individual detector pixels record the energy spectrum, which ultimately results in a spectrally or energy-resolved image recording. In contrast, conventional CT scanners are based on an energy-integrated recording technique, in which the total energy (usually a large number of photons and electronic noise) integrated in one pixel by the device over a set time period is measured. Thus, conventional CT scanners only measure the photon intensity, similar to black and white photography, while in photon counting CT spectral information is also recorded, similar to color photography. A particular advantage of photon counting CT is that the noise component is very low.

[0034] As a final step, the display of data is preferably provided on a display device (e.g., a screen). Here, preferably different compartments are displayed in different ways, such as with different colors or different textures, that is, regions with different tissue characteristics have been identified during processing. It is conceivable that information is displayed as edema content superimposed on color on the gray-white image recorded by brain CT. This superposition advantageously provides the main information about edema assessment, because the significance (Bedeutung) of edema in two substances is significantly different. Preferably, a 3D model of an object (e.g., a single brain tissue) and its regions (e.g., gray matter and white matter) is also used for visual presentation. As described above, the main information can be visualized in a color-coded manner and can also be superimposed if necessary.

[0035] The device according to the invention for imaging in computed tomography or for controlling a computed tomography system or a diagnostic system is particularly preferably configured to carry out the method according to the invention and comprises the following components:

[0036] A data interface configured to receive a CT dataset of the object, wherein the CT dataset has been recorded by means of a multispectral recording method.

[0037] - Optionally, a data interface configured to receive a comparison data set of the object, wherein the comparison data set has been recorded by means of the multispectral recording method described above. This may be the same data interface as described above.

[0038] - a contrast suppression unit configured to suppress contrast that has been caused by one tissue type (or by two or more different tissue types) and to generate a contrast suppressed data set.

[0039] - an analysis unit configured to analyze at least the contrast-suppressed dataset or at least a dataset based on the contrast-suppressed dataset (e.g., a synthetic dataset mentioned below) with the help of a machine learning algorithm, wherein the analysis is configured to identify at least one change in a tissue type (or two or more different tissue types).

[0040] It should be noted that the present invention is not primarily intended for displaying information, but primarily for processing CT data sets. This processing is not an easy task and, as is common with all iterative data processing methods in the field of image processing, cannot be performed simply with the aid of manual calculations. With the aid of the present invention, qualitative and quantitative analyses of tissues can be performed.

[0041] A control device according to the invention for controlling a computed tomography system or a diagnostic system is configured to carry out the method according to the invention and / or comprises an apparatus according to the invention.

[0042] However, the device for imaging in computed tomography according to the invention can itself also be designed as a device for controlling a CT system.Alternatively or additionally, the device for imaging in computed tomography according to the invention can be designed as a device for controlling a diagnostic system.

[0043] The device according to the present invention can also be connected to a CT device, for example, in particular via a network. In an alternative embodiment, the device can also be designed as a part of a CT device. In particular, the individual modules or components of the device can also be arranged in a network or functionally connected via a network.

[0044] The network may be a local area network ("LAN" for short) or a wide area network ("WAN" for short). An example of a local area network is an intranet, an example of a wide area network is the Internet. In particular, the network may also be implemented wirelessly, in particular as a WLAN ("Wireless LAN", often referred to by the English abbreviation "WiFi") or as a Bluetooth connection. The network may also be implemented as a combination of the above examples.

[0045] The computer tomography system according to the invention is designed for recording CT data sets by means of a multispectral recording method, wherein a photon counting computer tomography system is preferred, which comprises a control device according to the invention.

[0046] The diagnostic system according to the invention comprises a control device according to the invention. The method described above is particularly advantageous when used in a corresponding diagnostic system. Therefore, alternatively, the diagnostic system can be configured to perform the method according to the invention. The diagnostic system can be an independent system, which can, for example, access a PACS or can be directly integrated in a CT system (e.g., a CT scanner). It is also preferred that there is interactive data communication between the CT system and the diagnostic system, in particular in order to be able to optimally implement computationally intensive steps of raw data preprocessing (e.g., flexible binning).

[0047] A diagnostic system is a system or component that enables analysis and / or processing of data detected or recorded by a CT system (so-called CT data). In particular, the analysis and / or processing of CT data or CT data sets may include reconstruction of image data from raw data, post-processing of image data or raw data, analysis of image data or raw data, segmentation of structures in the image data, identification or quantification of anatomical structures, in particular identification or quantification of pathologically altered anatomical structures in the image data, and similar steps of image processing and data processing.

[0048] Although fat and water or gray matter and white matter are often mentioned as examples in the context of this description, other materials such as calcium, iodine, blood, cholesterol, fat, iron, etc. can also be distinguished with the invention and changes can be determined. With the invention, all possible tissue types can be examined (differentiated, characterized and quantified) based on the distinguishability of materials (such as the above substances) in terms of contrast. For example, clots in blood vessels or fragile blood vessel walls can be identified.

[0049] In the processor of the corresponding device or control device, most of the above-mentioned components of the device or control device can be implemented in whole or in part in the form of software modules. The advantage of the implementation based mainly on software is that even previously used control devices can be upgraded in a simple manner by software updates so that they can work in accordance with the method according to the invention. In this regard, the object of the present invention can also be achieved by a corresponding computer program product with a computer program, which can be directly loaded into the computing system (for creating the device) or storage device of the control device of the diagnostic system or the computer tomography system, and the computer program has program segments so that when the program is run in the computing system or the control device, all steps of the method according to the invention are performed. In addition to the computer program, such a computer program product can also include additional components such as documents and / or additional components and hardware components, such as hardware keys (encryption dongles, etc.) for using the software.

[0050] For transmission to a computing system or control device and / or for storage at or in a computing system or control device, a computer-readable medium, such as a memory stick, a hard disk or other portable or fixedly mounted data carrier, can be used, on which program segments of a computer program are stored, which can be read and executed by a computing unit of the computing system or control device. For this purpose, the computing unit can, for example, have one or more cooperating microprocessors or the like.

[0051] Particularly advantageous further designs and improvements of the invention are derived from the claims and the following description, wherein claims of one claim category may also be improved similarly to claims and parts of the description of another claim category, and in particular, individual features of different embodiments or variants may also be combined into new embodiments or variants.

[0052] According to a preferred method, the provision of the CT data set of the object comprises a recording within the scope of a multispectral recording method, wherein the provided CT data set is recorded using a multispectrally resolving detector, preferably a photon counting detector.

[0053] Preferably, a detector that allows flexible binning is used. Optimization of spectral data or reduction of measurement errors can be achieved through flexible binning. Flexible binning is known to those skilled in the art. "Binning" can be understood as the classification of measured values ​​during data acquisition or data processing. Here, the target number of measured values ​​is divided into intervals, so-called boxes (English: Bins), in ascending order according to size. Then, the measured value is replaced by a representative of the interval in which the measured value is located. In flexible binning, the width of the box, i.e. the size of the interval, can be changed (if necessary, individually).

[0054] According to a preferred method, in the analysis, the location or size or intensity of a region (e.g., a brain region) with changes (e.g., edema in gray and white matter) or the characteristics of the changes (e.g., characteristics of the edema) within the tissue type examined there are additionally determined. In addition, a prediction of the clinical outcome and / or treatability is preferably obtained. This advantageously helps in making decisions about treatment strategies. It is also preferred that during the examination, the degree of disease progression is also obtained in a subsequent examination.

[0055] According to a preferred method, the analysis is performed with the aid of a machine learning algorithm based on a plurality of models from the group consisting of Markov models, neural networks, convolution algorithms and 3-D mesh models, in particular based on a 3-D reconstruction of the image data.

[0056] A preferred method comprises the following additional steps:

[0057] - providing a comparison data set (particularly based on a phantom), wherein the comparison data set has been recorded by means of the aforementioned multispectral recording method. The comparison data set thus comprises data that can be directly compared with the CT data set, since the same (particularly the same) recording method was used respectively.

[0058] The comparison data set comprises data on the expected tissue types, in particular on the tissue types altered by the pathology. The comparison data set preferably comprises CT data of a standardized phantom having defined tissue types to be measured conventionally in the subject, such as tissue types of different ventricles (e.g. gray matter and white matter), and also comparative data sets related thereto, depending on the question and the expected pathology. However, in principle, the method can also be performed with a comparison data set that is not based on a phantom, for example the comparison data set can comprise a scanner-specific calibration curve.

[0059] The comparison data set can be provided with the aid of the RIS, wherein the comparison data set can have been directly recorded or can be called up from the PACS.

[0060] - Processing one of the data sets by means of iterative data reconstruction, wherein the dependent data set is optimized based on the comparison data set. In iterative data processing, a known reference on which the comparison data set is based can be reconstructed until the reference is satisfactorily present, and then the corresponding reconstruction steps can be applied to the dependent data set.

[0061] Iterative data reconstruction uses iterative algorithms for reconstructing 2D and 3D images. In computed tomography, images must usually be reconstructed from projections of the object. Here, iterative reconstruction techniques are a better but more computationally intensive alternative to the conventional Filtered-Back-Projection Methode (FBP), which allows the image to be directly calculated in a single reconstruction step. However, there are several feasible solutions for iterative reconstruction to achieve extremely fast calculations and large-scale parallelization, which makes iterative reconstruction practical for commercial use. Reconstructing an image from the detected data is an inverse problem. It is usually not possible to solve the inverse problem directly in an exact manner. In this case, the direct algorithm will have to calculate the solution in an approximate manner, which may result in visible reconstruction artifacts in the image. Iterative algorithms use multiple iterative steps to approach the correct solution, so better reconstruction can be obtained at the expense of longer computing time.

[0062] There are multiple algorithms, however, each of which starts with an assumed image (here, the image is a pre-known benchmark on which the comparison data set is based), computes projections from the image, compares the raw projection data, and updates the image based on the difference between the computed projections and the actual projections. Preferred algorithms are those for algebraic reconstruction, those for iterative sparse asymptotic minimum variance (SAMV), those for statistical reconstruction, and those for learned iterative reconstruction (erlernte Iterative Rekonstruktion). Algebraic reconstruction techniques (ART) were the first iterative reconstruction techniques used for computed tomography by Hounsfield. The iterative sparse asymptotic minimum variance algorithm is an iterative, parameter-free, high-resolution tomography reconstruction method inspired by compressed sensing. In learned iterative reconstruction, the update algorithm is learned from training data using machine learning techniques (such as neural convolutional networks), while still including an image formation model. This typically results in faster and higher quality reconstructions.

[0063] Iterative data processing enables optimal selection of the best data points obtained from multispectral imaging and enables optimal display of selected compartments (regions of the same tissue type, such as gray matter). This is done by optimizing the object signal behavior expected in the comparison data based on a priori knowledge.

[0064] The comparison data set is iteratively processed in a nested manner, wherein the distinguishability of the benchmark and each tissue type is preferably checked separately and a quantitative distinguishing value is assigned to the distinguishability. This distinguishing value is then maximized in order to achieve an optimal distinguishability. At the same time, the parameters for data processing are applied to the CT data set and in particular another distinguishing value is assigned to its result (e.g., by manual marking of representative tissue regions).

[0065] The iterative approach forms an advantageous algorithm for optimally selecting the best data points obtained from multispectral imaging and enables an optimal display of the selected compartment (e.g., gray matter). This is done by optimizing the signal behavior expected based on a priori knowledge, as measured in a comparison data set (e.g., because the phantom is known), and the signal behavior measured in the patient. The prior knowledge improves the accuracy of the reconstruction, thereby improving the reliability of the results.

[0066] Iterative data processing (if necessary, as image reconstruction) can be applied to the CT data set after its recording (as raw data or as reconstructed image data). Alternatively or additionally, iterative data processing can be applied to contrast suppressed (e.g., to improve gray suppression) and / or contrast suppressed data sets and / or synthetic data sets (e.g., to improve edema display). In particular, iterative data processing is performed within the scope of an analysis, wherein in this case, according to the invention, a machine learning algorithm is involved in the iterative data processing.

[0067] Iterative reconstruction is very advantageous, but not absolutely necessary, and iterative reconstruction can better pre-process the data set during or for the analysis, thereby improving or even enabling a qualitative or quantitative assessment. For example, the distinction between gray matter and white matter can be further improved with the aid of iterative methods. In particular, the data set can be better pre-processed for the analysis (or during the analysis) with the aid of iterative reconstruction, thereby improving the qualitative or quantitative assessment.

[0068] Preferably, the comparison data set is a CT data set of a phantom, which phantom includes samples corresponding to the tissue to be examined on the subject, i.e. different tissue types present in the compartment. Here, the provision of the comparison data set comprises a recording of the phantom within the scope of a multispectral recording method, wherein the recording of the phantom is preferably performed synchronously (i.e. in parallel in time) with the recording of the subject.

[0069] The phantom is preferably fixed to the device (e.g. to the headrest) or, depending on the type of examination, is preferably attached in a modular manner to other structures or the patient and introduced into the examination channel. In order to simplify the automated evaluation including quality assurance, a machine-readable identification of the phantom is also preferred. Measuring synchronously with the phantom can be used to improve the results and exclude differences between different scanners (e.g. differences in measurement accuracy or comparability between different devices) or to standardize the measurement results. The advantage of synchronous measurement is that individual differences in the scans, such as artifacts, eccentric support, can be better presented.

[0070] A preferred method comprises the additional step of providing a database, which contains other comparative data sets for different tissue types and preferably also contains other comparative data sets for pathology. The database may also be referred to as an "atlas" and preferably comprises data about anatomical signal behavior and / or pathological signal behavior, and in particular data about differences in signal behavior. The database preferably comprises comparative data sets for tissue types of different body parts or compartments. The comparative data sets present in the database have preferably been created using the method according to the invention and therefore contain information about relevant aberrations. The atlas advantageously contributes to improving the accuracy of the assessment.

[0071] The database is preferably integrated in the algorithm for data evaluation. Particularly preferably, the database is used for training the aforementioned machine learning algorithm (the machine learning algorithm used in the analysis and / or the machine learning algorithm used within the scope of iterative data reconstruction).

[0072] As already indicated above, the iterative data reconstruction preferably comprises an algorithm based on machine learning principles, preferably a deep learning algorithm, wherein preferably a comparative evaluation of different regions of the object is performed, particularly preferably a lateral comparative evaluation (seitenvergleichende ). In a side comparison assessment, different sides of an object are compared to each other, such as the two hemispheres if the object is a brain. Here, the symmetry between the two sides is exploited to obtain the structure of one of the two sides.

[0073] Multivariate solutions may emerge in the processing, depending on the number of compartments to be examined (e.g., optimized datasets for gray or white matter, calcifications, hemorrhages, etc.), as well as the generation of new datasets using iterative methods to improve the display of desired features. For optimal processing of the resulting amount of data, the use of deep learning is highly advantageous.

[0074] In the context of machine learning methods, preference is given to devices in which the processing unit comprises a plurality of machine learning algorithms, preferably two or more machine learning algorithms which perform data processing in a manner which is functionally intertwined with one another.

[0075] In the iterative data reconstruction, there is preferably a combination of two or more different evaluation algorithms that are functionally linked to one another. Here, the algorithms preferably implement different kinds of data processing and are particularly preferably linked to one another by means of quantitative distinguishing values.

[0076] According to a preferred method, within the context of contrast suppression, the contrast involved is suppressed to more than 50%, preferably to more than 75% (or even 90%), particularly preferably to 100%. Since work is particularly performed with image data that simplify the final examination of the physician, the contrast-suppressed data set preferably includes reconstructed image data with suppressed contrast.

[0077] According to a preferred method, during or before the analysis, image data are reconstructed from a data set, preferably from a CT data set. In this case, the subsequent steps of the method (i.e. contrast suppression and / or analysis) are preferably no longer performed with the original data, but with the reconstructed image data. In the context of reconstruction, three-dimensional image data are particularly preferably reconstructed. This is advantageous because treatment decisions are usually based on volume information. In this case, the optional superposition of the contrast-suppressed data set with the CT data set can preferably be performed completely based on 3D image data, in particular using surface contours from the 3D image (i.e. the three-dimensional image data).

[0078] According to a preferred method, tissue types are identified based on the CT data set. As described above, this is preferably done by means of iterative data processing. This step is preferably performed before contrast suppression or at least before analysis. The identification is preferably based on the water content and / or fat content in the CT data set. For example, after analyzing the water content and fat content of the image reconstructed from the CT data set, an analysis using artificial intelligence will be performed to identify the area affected by the stroke.

[0079] Preferably, the identification is performed by using a database, which can also be referred to here as an "atlas", since it also contains comparison data sets. Here, a part of the CT data set (including also the complete CT data set) is compared with an atlas of previously known data about tissue types, wherein the atlas preferably includes differences in the CT signals measured for different tissue types (of the object) and differences in the reconstruction of these CT signals, in particular for recorded CT signals based on tissue types that are changed by pathology. Thus, the CT data set is compared with the data of the database. Here, differences that cannot be reliably identified by an expert can also be displayed in a specific data set and in a comparison with a large number of similar groups (groβeKohorte). This applies to intra-individual and inter-individual differences within a fixed time point or interval.

[0080] The atlas preferably includes not only anatomical information, but also information about the pathological behavior (e.g. edema) and related information about the likelihood of an indication (e.g. the risk of cerebral hemorrhage), in particular in the form of quantitative information or formulas which together with the analysis results provide quantitative information about the likelihood.

[0081] According to a preferred method, a contrast-suppressed data set and a CT data set (i.e. a data set without contrast suppression) are superimposed to form a composite data set. Thus, a composite data set is created which comprises a superposition of the CT data set and the contrast-suppressed data set. Here, for example, in the case of working with raw data, the marked data can be superimposed, or (if working with reconstructed image data) a visual superposition can be performed, for example based on a "contour-fit" model. Preferably, images (in particular 3D images) from the CT data set are registered with images from the contrast-suppressed data set. This superposition makes it possible to make statements about changes in tissue types, for example different edema contents of gray and white matter.

[0082] For superimposition, a preferred device comprises a superimposition unit, which is configured to superimpose the contrast-suppressed data set with the CT data set to form a composite data set. Here, the superimposition unit can be part of the contrast suppression unit.

[0083] A preferred device comprises a recognition unit, which is configured to automatically recognize a part of the body for which a CT data set has been recorded. The recognition unit can, for example, comprise a body atlas and obtain the part concerned by comparing the recorded image data with the body atlas. Preferably, the recognized part is also automatically named, or based on the recognition, the CT data set receives a corresponding label. Here, the recognition unit can be part of the contrast suppression unit.

[0084] According to a preferred method, the CT data set is a multispectral recording of the brain, in particular of the brain of a stroke patient. In the context of contrast suppression, the gray-white contrast of gray matter and white matter is suppressed, and the contrast-suppressed data set contains information based on the gray-white contrast suppression. The analysis is configured to identify at least one change in the white matter and / or gray matter, in particular edema.

[0085] A preferred device comprises a data interface, which is configured to control a computed tomography system (1), which is configured to record CT data sets (D) by means of a multispectral recording method, the data interface is particularly configured to control a dual-source dual-energy computed tomography system or a multi-source / dual-source multi-energy computed tomography system, the data interface being particularly preferably configured to control a computed tomography system configured to measure based on a photon counting method. The data interface mentioned here can be identical to the following data interface, via which the device receives data, provided that such a data interface is suitable for bidirectional communication.

[0086] The method may also include elements of "cloud computing". In "cloud computing", IT infrastructure, such as storage space or computing power and / or application software, is provided via a network. Here, communication between the user and the "cloud" takes place by means of data interfaces and / or data transmission protocols.

[0087] In the context of "cloud computing", in a preferred embodiment of the method according to the invention, data is provided to the "cloud" via a data channel (e.g., a network). The "cloud" includes a (remote) computing system, such as a computer cluster, which usually does not include a local computer of the user. In particular, the cloud can be provided by a medical institution that also provides a medical technology system. In particular, the data of the CT data set are sent to the (remote) computer system ("cloud") via a RIS or PACS. Preferably, the computer system, the network and the associated CT system of the "cloud" constitute an interconnection in the sense of data technology. Here, the method according to the invention can be implemented in a network by means of a combination of instructions. Preferably, the data or results calculated in the cloud are sent to the local computer of the user via a data channel (e.g., via a network). For example, the data of the provided CT data set are processed by the computer system of the hospital according to the method according to the invention, and the results of the processing are sent back to the user via a RIS or PACS.

[0088] In the context of a preferred embodiment of the device according to the invention, the components relevant to the invention are located on the "cloud" side. In addition to such a device, a preferred system also includes a local computing unit, which is connected to the device via a data channel (for example, a network designed in particular as a RIS or PACS). The local computing unit includes at least one data receiving interface in order to receive data. Furthermore, it is preferred that the local computer also has a sending interface in order to send data to the device.

[0089] The invention has the advantage that the clinical application of CT imaging is expanded, in particular the clinical application of the original presentation of the brain and other body parts, the semi-quantitative presentation of pathological processes (e.g. inflammation, atrophy, edema or tumors) is improved, and the diagnosis is standardized, thereby achieving a higher diagnostic reliability and the presentation of tissue changes that have not been detected or not detected sufficiently so far. Another advantage is that CT can be used for patients with contraindications to MRT, and image data can be acquired faster, especially in the emergency and for uncooperative patients, and the quality assurance of CT can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] The invention is further explained in detail below with reference to the drawings by means of embodiments. In the different drawings, the same components are provided with the same reference numerals. The drawings are generally not drawn to scale. In particular:

[0091] Figure 1 A schematic diagram of a computed tomography system is shown with an exemplary embodiment of a control device having an apparatus according to the invention for carrying out the method according to the invention;

[0092] Figure 2 A flow chart showing a possible process of the method according to the present invention;

[0093] Figure 3 A detailed flow chart of a possible procedure of the method according to the present invention is shown using a stroke examination as an example. DETAILED DESCRIPTION

[0094] Figure 1 A computer tomography system 1 is schematically shown, which has a control device 10 for carrying out the method according to the invention. Here, the CT system 1 is configured as a photon counting system and has a scanner 2 with a gantry in which an X-ray source 3 rotates, which X-ray sources 3 respectively irradiate a patient who is pushed into the measurement space of the gantry by means of a couch 5 so that the radiation is incident on a detector 4 which is arranged opposite the X-ray source 3 and is a photon counting detector. It should be explicitly pointed out that according to Figure 1 The embodiment of is only one example for CT and the invention can also be used for any CT configuration for multispectral recording, for example a CT configuration with an annular fixed X-ray detector and / or multiple X-ray sources.

[0095] For the control device 10, only the components which are essential for explaining the invention are shown. In principle, such CT systems and the associated control devices are known to the person skilled in the art and therefore do not require a detailed explanation.

[0096] Here, the core component of the control device 10 is a processor 11, on which various components are implemented in the form of software modules. The control device 10 also has a terminal interface 14, to which a terminal 20 is connected, and an operator can operate the control device 10 through the terminal 20 to operate the computer tomography system 1. Another interface 15 is a network interface for connecting to a data bus 21, so as to establish a connection with a RIS (Radiology Information System) or a PACS (Picture Archiving and Communication System).

[0097] The control device 10 can control the scanner 2 via a control interface 13, i.e., for example, the rotation speed of the gantry, the movement of the patient couch 5 and the X-ray source 3 itself. Raw data RD are read from the detector 4 via an acquisition interface 12. The control device 10 also has a storage unit 16. The storage unit 16 may include a database, i.e., an atlas, which contains comparative data sets V of tissue types of different body parts or compartments.

[0098] In particular, the measurement control unit 17 is implemented as a software component on the processor 11 . The measurement control unit 17 controls the scanner 2 via the control interface 13 based on one or more selected measurement protocols, which may be modified by the operator via the terminal 20 , to perform measurements and acquire data.

[0099] Another component of the processor 11 is an image data reconstruction unit 6, with which desired image data are reconstructed from the raw data RD obtained via the data acquisition interface 12. The image data reconstruction unit 6 is configured as a device 6 according to the invention.

[0100] The device 6 comprises a data interface 7, which is configured to receive a CT data set D of the object and a comparison data set V, which are both transmitted here in the form of raw data RD. The data interface 7 is configured here as a bidirectional data interface, which can both receive raw data and send control data, for example for controlling the recording of the object O and the phantom P.

[0101] The apparatus further comprises a contrast suppression unit 8, which is configured to suppress contrast already caused by one tissue type (or by two or more different tissue types) and is used to generate a contrast-suppressed dataset K. The suppression unit 8 comprises a superposition unit 8a, which is configured to create a synthetic dataset R, which comprises a superposition of the contrast-suppressed dataset K and the CT dataset D. The apparatus 6 further comprises an analysis unit 9, which is configured to analyze the contrast-suppressed dataset R by means of a machine learning algorithm, wherein the analysis is configured to identify at least one change within a tissue type (or two or more different tissue types).

[0102] In this example, in addition to the analysis, the analysis unit 9 is also configured to reconstruct the image. Alternatively, the image can also be reconstructed before the contrast is suppressed. Preferably, a 3D image is reconstructed.

[0103] Figure 2 A flow chart showing a possible sequence of the method according to the invention is shown.

[0104] In step I, a multispectral recording of a CT data set D and a comparison data set V is performed in the form of raw data RD, in particular with the aid of a photon counting CT system. Here, the Figure 1 The CT system shown.

[0105] In step II, a CT data set D of the object which has been extracted from the raw data and a comparison data set V are provided. Here, the comparison data set V is a CT data set D of a phantom P which includes a sample corresponding to the tissue to be examined on the object O, wherein the provision of the comparison data set V includes recording the phantom P in the context of a multispectral recording method and is performed synchronously with the recording of the object O.

[0106] In step III, the CT data set D is processed together with the comparison data set V by means of iterative data reconstruction, wherein with the aid of the comparison data set V an identification of the tissue types G1 , G2 is carried out based on the CT data set.

[0107] In step IV, the contrast that has been caused by the two different tissue types is suppressed and a contrast-suppressed data set K is generated.

[0108] In step V, the contrast suppressed data set K is analyzed by means of a machine learning algorithm, wherein the analysis is configured to identify at least one change within the two different tissue types G1 , G2 .

[0109] In step VI, the results are displayed, wherein preferably different compartments are displayed in different ways, for example in different colors.

[0110] Figure 3 A detailed flow chart of a possible process of the method according to the invention is shown using a stroke examination as an example. In this example, the CT data set D includes a CT recording of the brain, and the tissue types G1, G2 are gray matter and white matter.

[0111] In a combined step I / II, a CT data set D is recorded by means of a photon counting CT system and a CT data set D is provided which is a multispectral recording of the brain of a stroke patient.

[0112] In step III, the CT data set D is processed, in which step III 3D image data are reconstructed from the CT data set. In this example, the following steps use the reconstructed image data instead of the original data.

[0113] In step IV, the gray-white contrast of gray matter and white matter is suppressed (in particular completely), and the result of the suppression is stored in a contrast-suppressed data set K.

[0114] In step IVa, a synthetic data set R is created, which comprises a superposition of the contrast suppressed data set K and the CT data set D. Step IVa may be a part of step IV.

[0115] In step V, an analysis is again performed, which here is particularly configured to identify at least one change in the white matter and / or the grey matter, in particular edema. Here, the analysis can be based on the synthetic data set R.

[0116] In step VI, the result is displayed again, here in the form of a 3D image, wherein preferably different compartments are displayed in different ways, for example in different colors.

[0117] Finally, it should be pointed out again that the method described in detail above and the illustrated computer tomography system 1 are only embodiments, which can be modified in various ways by a person skilled in the art without departing from the scope of the present invention. In addition, the use of the indefinite article "a" or "an" does not exclude the presence of a plurality of related features. Similarly, the terms "unit" and "module" do not exclude that the related components are composed of a plurality of cooperating subcomponents that can also be distributed spatially if necessary.

Claims

1. A method for imaging in computed tomography, comprising the following steps: - providing a CT data set (D) of an object (O), wherein the CT data set (D) has been recorded by means of a multispectral recording method; - suppressing the contrast already caused by the tissue types (G1, G2) and generating a contrast-suppressed data set (K) based on the CT data set (D); - analyzing at least the contrast-suppressed data set (K) or at least a data set based on the contrast-suppressed data set (K) with the aid of a machine learning algorithm, wherein the analysis is configured to identify at least one change in the tissue type (G1, G2).

2. The method according to claim 1, wherein the provision of the CT data set (D) of the object (O) comprises recording within the scope of a multispectral recording method, wherein a multispectrally resolving detector (4) is used.

3. The method according to claim 2, wherein a photon counting detector is used to record the provided CT data set (D). The method of claim 2 , wherein the multi-spectral resolving detector allows for flexible binning.

5. The method according to claim 1 , wherein the position of a region with a change within the tissue type (G1, G2) and / or the size of the region and / or the characteristics of the change and / or the progression of the disease are additionally determined in the analysis; and / or The analysis is performed with the aid of a machine learning algorithm based on a plurality of models from the group consisting of a Markov model, a neural network, a convolution algorithm and a 3-D mesh model.

6. The method of claim 5, wherein a prediction of clinical outcome and / or treatability is obtained.

7. The method according to claim 5, wherein the analysis is performed with the aid of a machine learning algorithm based on a 3-D reconstruction of the image data.

8. The method according to claim 1, comprising the following additional steps: - providing a comparison data set (V), wherein the comparison data set (V) has been recorded by means of the multispectral recording method; - processing the CT data set (D) together with the comparison data set (V) by means of iterative data reconstruction, wherein an optimization of the CT data set (D) is performed based on the comparison data set (V); The comparison data set (V) is a CT data set (D) of a phantom (P), the phantom comprising a sample corresponding to the tissue to be examined on the object (O), wherein the provision of the comparison data set (V) comprises recording the phantom (P) within the scope of the multispectral recording method.

9. Method according to claim 8, wherein the recording of the phantom (P) is performed synchronously with the recording of the object (O).

10. The method according to claim 1, wherein in the context of contrast suppression, the contrast is suppressed to more than 50%.

11. A method according to claim 10, wherein the contrast involved is suppressed to more than 75%.

12. The method according to claim 11, wherein the contrast involved is suppressed to 100%.

13. The method according to claim 10, wherein the contrast suppressed data set (K) comprises reconstructed image data with suppressed contrast.

14. The method of claim 1, wherein during or prior to the analysis, image data are reconstructed from the data set. The method according to claim 14 , wherein image data is reconstructed from the CT data set. The method according to claim 14 , wherein the contrast suppression and / or the analysis are performed using the reconstructed image data. The method of claim 14 , wherein three-dimensional image data is reconstructed.

18. The method according to claim 1, wherein the tissue type (G1, G2) is identified based on the CT data set (D).

19. The method according to claim 18, wherein said identifying precedes said contrast suppression or at least precedes said analyzing.

20. The method of claim 18, wherein the identifying is by means of iterative data processing.

21. The method of claim 18, wherein the identifying is based on water content and / or fat content in the CT dataset.

22. The method according to claim 1, comprising the following additional steps: - creating a synthetic data set (R) comprising a superposition of the contrast suppressed data set (K) and the CT data set (D).

23. The method according to claim 22, wherein the analysis is based on the synthetic data set (R).

24. The method of claim 1, wherein - the CT dataset (D) is a multispectral recording of the brain; - the gray-white contrast of gray matter and white matter is suppressed, and the contrast-suppressed data set (K) contains information based on the suppression of the gray-white contrast; - The analysis is configured to identify at least one change within the white matter and / or the grey matter.

25. The method according to claim 24, wherein the CT data set (D) is a multispectral recording of the brain of a stroke patient.

26. The method of claim 24, wherein the analysis is configured to identify edema within the white matter and / or the gray matter.

27. An apparatus (6) for imaging in computed tomography, the apparatus comprising: a data interface (7) configured to receive a CT data set (D) of an object (O), wherein the CT data set (D) has been recorded by means of a multispectral recording method; - a contrast suppression unit (8) configured to suppress contrast already caused by the tissue types (G1, G2) and to generate a contrast suppressed data set (K) based on the CT data set (D); - an analysis unit (9) configured to analyze at least the contrast-suppressed data set (K) or at least a data set based on the contrast-suppressed data set (K) with the aid of a machine learning algorithm, wherein the analysis is configured to identify at least one change within the tissue type (G1, G2).

28. The apparatus according to claim 27, wherein the apparatus further comprises: A superposition unit (8a) is configured to superimpose the contrast-suppressed data set (K) and the CT data set (D) into a composite data set (R).

29. The apparatus of claim 27, wherein the apparatus further comprises: A recognition unit is configured to automatically recognize a part of the body for which a CT data set has been recorded.

30. The apparatus of claim 27, wherein the apparatus further comprises: A data interface (7) is configured to control a computed tomography system (1) which is configured to record CT data sets (D) by means of a multispectral recording method.

31. The apparatus of claim 30, wherein the data interface is configured to control a dual-source dual-energy computed tomography system or a multi-source / dual-source multi-energy computed tomography system.

32. The apparatus of claim 31, wherein the data interface is configured to control a computed tomography system configured to perform measurements based on a photon counting method.

33. A control device (10) for controlling a computed tomography system (1) or a diagnostic system, the control device comprising an apparatus (6) according to claim 27, and / or the control device being configured to perform a method according to any one of claims 1 to 26.

34. A diagnostic system, comprising the control device according to claim 33, and / or configured to perform the method according to any one of claims 1 to 26.

35. A computer tomography system (1) configured to record a CT data set (D) by means of a multispectral recording method, the computer tomography system comprising a control device (10) according to claim 33.

36. The computed tomography system of claim 35, wherein the computed tomography system is a photon counting computed tomography system.

37. A computer program product comprising a computer program that can be directly loaded into a storage device of a control device (10), the computer program having program segments so that when the computer program is run in the control device (10), all steps of the method according to any one of claims 1 to 26 are performed.

38. A computer-readable medium having stored thereon a program segment that can be read and executed by a computing unit, so that when the computing unit executes the program segment, all steps of the method according to any one of claims 1 to 26 are executed.

Images