Prediction of representations of an investigation area of ​​an investigation object after application of different amounts of a contrast agent

Description

[0001] The present invention relates to the technical field of radiology, in particular to supporting radiologists in radiological examinations using artificial intelligence methods. The present invention relates to training a machine learning model and using the trained model to predict representations of an examination area after the application of different amounts of a contrast agent.

[0002] WO2021 / 044153A1 discloses a method and a device for improving magnetic resonance images to produce contrast-enhanced images without the need to administer a contrast agent to a patient.

[0003] US2020294288A1 discloses methods for reconstructing an image of a contrast-enhanced CT output image from a non-enhanced CT input image, for reconstructing a non-enhanced CT output image from a contrast-enhanced CT image, for reconstructing a contrast-enhanced dual-energy CT output image from a contrast-enhanced CT single image, and for reconstructing a contrast-enhanced CT full-dose image from a contrast-enhanced CT low-dose image.

[0004] EP3739522A1 discloses systems for training a machine learning model for image processing, an image enhancement system, a method for training a machine learning model, an imaging arrangement, a computer program element and a computer-readable medium.

[0005] WO2019 / 074938A1 discloses a method for reducing the amount of contrast agent used in the production of radiological images using an artificial neural network.

[0006] In a first step, a training dataset is generated. The training dataset includes, for each person in a large number of people, i) a native radiological image ( zero-contrast image), ii) a radiological image after the administration of a small amount of contrast medium ( low-contrast image ) and iii) a radiological image after the administration of a standard amount of contrast medium ( full-contrast image ) .

[0007] In a second step, an artificial neural network is trained to predict, for each person in the training dataset, an artificial radiological image showing an image area after the application of the standard amount of contrast agent, based on the native image and the image after administration of a small amount of contrast agent. The measured radiological image after the application of a standard amount of contrast agent serves as the reference during training. ground truth ) .

[0008] In a third step, the trained artificial neural network can be used to generate an artificial radiological image for a new person based on a native image and a radiological image after the application of a small amount of contrast agent, which shows the recorded area as it would look if a standard amount of contrast agent had been applied.

[0009] The method disclosed in WO2019 / 074938A1 involves training an artificial neural network to map the native radiological image and the radiological image after the application of a small amount of contrast agent to the radiological image after the application of a standard amount of contrast agent. mapping). However, the artificial neural network is not trained to learn about the increasing influence of contrast agent on a radiological image and therefore cannot be used to generate radiological images that show an examination area after the application of different amounts of contrast agent.

[0010] In particular, the artificial neural network described in WO2019 / 074938A1 is not suitable for generating radiological images that show an area of ​​investigation after the application of a larger quantity than the standard quantity of contrast agent.

[0011] Furthermore, in the case of the procedure described in WO2019 / 074938A1 for producing a radiological image showing an area under investigation after application of the standard amount of contrast medium, at least two radiological images (a native radiological image and a radiological image after application of a small amount of contrast medium) are always required.

[0012] Based on the described state of the art, the task was to provide a machine learning model capable of generating artificial radiological images that depict an examination area after the application of different amounts of a contrast agent. Furthermore, the model should, in principle, be able to generate artificial radiological images after the application of different amounts of contrast agent based on a single metrologically generated image.

[0013] These problems are solved by the subject matter of the independent claims. Preferred embodiments are found in the dependent claims, the present description, and the drawings.

[0014] A first object of the present invention is a computer-implemented method for training and using a machine learning model. The training method comprises: Receiving training data, wherein the training data for a variety of objects under investigation comprise representations T< R 1 to T< R n of an area of ​​investigation, wherein n is an integer greater than two, where each representation T < R i the area under investigation after application of a quantity a i a contrast agent is represented, whereby i an integer in the range of 1 to n is, where the quantities a 1 to a n forming a sequence of preferably increasing sets, training the machine learning model, whereby the model is trained starting from the representation T< R 1 successively representations T< R 2 *< until T< R n *< to generate, wherein the generation of the representation T< R 2 *< is at least partially based on the representation T< R 1 and the generation of each further representation T< R k *< at least partially based on the previously generated representation T< R k -1 *< is done, whereby k an index that represents whole numbers from 3 to n passes through, where the representation T< R k * the area of ​​investigation after application of the quantity a k of the contrast agent and the representation T< R k -1 * the area of ​​investigation after application of the quantity a k-1 of the contrast agent is represented, where deviations between the generated representations T< R are accounted for using an error function. j * and the representations T< R j be quantified, whereby j an index that represents the numbers from 2 to n goes through, minimizing deviations by modifying model parameters of the machine learning model in an optimization procedure, using the machine learning model for prediction.

[0015] Another aspect of the present invention is a computer-implemented method for predicting a representation of an area of ​​investigation of a test object. The prediction method comprises: Receiving a representation R p of the area under investigation, where the representation R p the examination area without contrast agent or after application of a quantity a p a contrast agent is represented, whereby p an integer that is less than n is, whereby n an integer greater than 2, supplying the representation R p a trained machine learning model, where the machine learning model has been trained on the basis of training data, starting from a first representation T< R 1 successively a number n-1 to generate representations T< R 2 * to T< R n *, where the first representation T< R 1 represents the examination area without contrast agent or after application of a first amount a 1 of the contrast agent is represented and each generated representation T< R j * the area of ​​investigation after application of a quantity a j of the contrast agent, whereby j an index that represents whole numbers from 2 to n passes through, with the quantities a 1 to a n a sequence of preferably increasing sets, wherein the generation of the representation T< R 2 * is at least partially based on the representation T< R 1 and the generation of each further representation T< R k * is at least partially based on the previously generated representation T< R k-1 *, wherein k an index that represents whole numbers from 3 to n passes through, receiving one or more representations R p + q *< of the domain of investigation of the machine learning model, where each of the one or more representations R p + q *< the area of ​​investigation after application of the quantity a p+q of the contrast agent, where q is an index representing the numbers from 1 to m runs through, whereby m an integer that is less than n-1 is or equals n-1 is or greater than n-1is, output and / or storage and / or transmission of one or more representations of R p+q *< .

[0016] Another object of the present invention is a computer system comprising an input unit, a control and calculation unit, and an output unit; where the control and computing unit is configured, to cause the input unit to create a representation R p to receive a survey area of ​​a survey object, where the representation R p the examination area without contrast agent or after application of a quantity a p a contrast agent is represented, whereby p an integer that is less than n is, whereby n If an integer is greater than 2, the representation R p to feed a trained machine learning model, where the machine learning model has been trained on the basis of training data, starting from a first representation T< R 1, successively a number n-1 from representations T< R 2 *< to T< R n *< to generate, where the first representation T< R 1 is the area under investigation without contrast agent or after application of an initial amount a 1 of the contrast agent is represented and each generated representation T< R j *< the area of ​​investigation after application of a quantity a j of the contrast agent, whereby j an index that represents whole numbers from 2 to n passes through, with the quantities a 1 to a n a sequence of preferably increasing sets, wherein the generation of the representation T< R 2 *< is at least partially based on the representation T< R 1 and the generation of each further representation T< R k *< at least partially based on the previously generated representation T< R k -1 *< is done, whereby k an index that represents whole numbers from 3 to n runs through one or more representations of the machine learning model R p + q *< of the investigation area to receive, each of the one or more representations R p+q *< the area of ​​investigation after application of the quantity a p+q of the contrast agent, where q is an index representing the numbers from 1 to m runs through, whereby m an integer that is less than n-1 is or equals n-1 is or greater than n-1 is to cause the output unit to display one or more representations of R p + q *< to save and / or output and / or transmit to a separate computer system.

[0017] Another object of the present invention is a computer program product comprising a data storage device in which a computer program is stored that can be loaded into the main memory of a computer system and causes the computer system to perform the following steps: Receiving a representation R p of an investigation area of ​​an investigation object, where the representation R p the examination area without contrast agent or after application of a quantity a p a contrast agent is represented, whereby p an integer that is less than n is, whereby n an integer greater than 2, supplying the representation R p a trained machine learning model, where the machine learning model has been trained on the basis of training data, starting from a first representation T< R 1 successively a number n-1from representations T< R 2 *< to T< R n *< to generate, where the first representation T< R 1 is the area under investigation without contrast agent or after application of an initial amount a 1 of the contrast agent is represented and each generated representation T< R j *< the area of ​​investigation after application of a quantity a j of the contrast agent, whereby j an index that represents whole numbers from 2 to n passes through, with the quantities a 1 to a n a sequence of preferably increasing sets, wherein the generation of the representation T< R 2 *< is at least partially based on the representation T< R 1 and the generation of each further representation T< R k *< at least partially based on the previously generated representation T< R k -1 *< is done, whereby k an index that represents whole numbers from 3 to npasses through, receiving one or more representations R p + q *< of the domain of investigation of the machine learning model, where each of the one or more representations R p + q *< the area of ​​investigation after application of the quantity a p+q of the contrast agent, where q is an index representing the numbers from 1 to m runs through, whereby m an integer that is less than n-1 is or equals n-1 is or greater than n-1 is, output and / or storage and / or transmission of one or more representations of R p+q *< .

[0018] Another object of the present invention is the use of a contrast agent in a radiological examination procedure, wherein the radiological examination procedure comprises the following steps: Receiving a representation R p of an investigation area of ​​an investigation object, where the representation R p the examination area without contrast agent or after application of a quantity a p of the contrast agent, whereby p an integer that is less than n is, whereby n an integer greater than 2, supplying the representation R p a trained machine learning model, where the machine learning model has been trained on the basis of training data, starting from a first representation T< R 1 successively a number n-1 from representations T< R 2 *< to T< R n *< to generate, where the first representation T< R 1 is the area under investigation without contrast agent or after application of an initial amount a 1 of the contrast agent is represented and each generated representation T< R j *< the area of ​​investigation after application of a quantity a i of the contrast agent, whereby j an index that represents whole numbers from 2 to n passes through, with the quantities a 1 to a n a sequence of preferably increasing sets, wherein the generation of the representation T< R 2 *< is at least partially based on the representation T< R 1 and the generation of each further representation T< R k *< at least partially based on the previously generated representation T< R k -1 *< is done, whereby k an index that represents whole numbers from 3 to n passes through, receiving one or more representations R p + q *< of the domain of investigation of the machine learning model, where each of the one or more representations R p + q *< the area of ​​investigation after application of the quantity a p+q of the contrast agent, where q is an index representing the numbers from 1 tom runs through, whereby m an integer that is less than n-1 is or equals n-1 is or greater than n-1 is, output and / or storage and / or transmission of one or more representations of R p+q *< .

[0019] Another object of the present invention is a kit comprising a contrast agent and a computer program product, wherein the computer program product comprises a computer program that can be loaded into a working memory of a computer system and causes the computer system to perform the following steps: Receiving a representation R p of an investigation area of ​​an investigation object, where the representation R p the examination area without contrast agent or after application of a quantity a p of the contrast agent, whereby p an integer that is less thann is, whereby n an integer greater than 2, supplying the representation R p a trained machine learning model, where the machine learning model has been trained on the basis of training data, starting from a first representation T< R 1 successively a number n-1 from representations T< R 2 *< to T< R n *< to generate, where the first representation T< R 1 is the area under investigation without contrast agent or after application of an initial amount a 1 of the contrast agent is represented and each generated representation T< R j *< the area of ​​investigation after application of a quantity a i of the contrast agent, whereby j an index that represents whole numbers from 2 to n passes through, with the quantities a 1 to a n a sequence of preferably increasing sets, wherein the generation of the representation T< R 2 *< is at least partially based on the representation T< R 1 and the generation of each further representation T< R k *< at least partially based on the previously generated representation T< R k -1 *< is done, whereby k an index that represents whole numbers from 3 to n passes through, receiving one or more representations R p + q *< of the domain of investigation of the machine learning model, where each of the one or more representations R p + q *< the area of ​​investigation after application of the quantity a p+q of the contrast agent, where q is an index representing the numbers from 1 to m runs through, whereby m an integer that is less than n-1 is or equals n-1 is or greater than n-1is, output and / or storage and / or transmission of one or more representations of R p+q *< .

[0020] The invention is explained in more detail below without distinguishing between the subject matter of the invention (training method, prediction method, computer system, computer program product, use, kit). Rather, the following explanations are intended to apply analogously to all subject matter of the invention, regardless of the context in which they are made (training method, prediction method, computer system, computer program product, use, kit). If steps are mentioned in a sequence in the present description or in the claims, this does not necessarily mean that the invention is limited to the stated sequence. It is conceivable that the steps could also be carried out in a different sequence or even in parallel to one another; unless one step builds upon another step, which necessarily requires that the building step be carried out subsequently (which will be clear in the specific case).The aforementioned sequences thus represent preferred embodiments of the invention.

[0021] With the help of the present invention, representations of an investigation area of ​​an investigation object can be predicted.

[0022] The "object of study" is usually a living being, preferably a mammal, and most preferably a human being.

[0023] The "area of ​​investigation" is a part of the object being examined, for example an organ or part of an organ such as the liver, brain, heart, kidney, lung, stomach, intestine, or bladder, or a part of the aforementioned organs, or several organs, or another part of the body.

[0024] In a preferred embodiment of the present invention, the area of ​​investigation is the liver or a part of the liver of a human being.

[0025] The examination area, also called the recording volume (English: field of view, The field of view (FOV) is a volume that is depicted in radiological images. The area under examination is typically selected by a radiologist, for example, on a panoramic radiograph (also known as an overview radiograph). localizer ). Alternatively or additionally, the scope of investigation can also be defined automatically, for example based on a selected protocol.

[0026] A "representation of the area under investigation" is preferably a medical photograph (English: medical image). A "representation of the area under investigation" is preferably the result of a radiological examination.

[0027] Radiology is the branch of medicine that deals with the application of primarily electromagnetic radiation and (including, for example, ultrasound diagnostics) mechanical waves for diagnostic, therapeutic, and / or scientific purposes. In addition to X-rays, other ionizing radiation such as gamma rays or electrons are also used. Since a key application is imaging, other imaging techniques such as sonography and magnetic resonance imaging (MRI) are also considered part of radiology, even though these techniques do not use ionizing radiation. The term "radiology" as used in the present invention therefore includes, in particular, the following examination methods: computed tomography, magnetic resonance imaging, and sonography.

[0028] In a preferred embodiment of the present invention, the radiological examination is a computed tomography examination or a magnetic resonance tomography examination.

[0029] Computed tomography (CT) is an X-ray procedure used to create cross-sectional images of the human body. In contrast to a conventional X-ray, which typically only shows rough structures and bones, CT scans also capture soft tissue with subtle contrast differences in detail. An X-ray tube generates a fan-shaped X-ray beam that penetrates the body and is attenuated to varying degrees by different structures, such as organs and bones. Receiving detectors opposite the X-ray source pick up these signals of varying intensity and transmit them to a computer, which uses the received data to create cross-sectional images of the body. Computed tomography images (CT scans) can be viewed in 2D or 3D. To better distinguish structures within the human body (e.g.,(vessels), a contrast agent can be injected into a vein, for example, before CT scans are taken.

[0030] Magnetic resonance imaging, abbreviated MRI (English: MRI: Magnetic Resonance Imaging), is an imaging technique used primarily in medical diagnostics to depict the structure and function of tissues and organs in the human or animal body.

[0031] In magnetic resonance imaging (MRI), the magnetic moments of protons in a sample are aligned in a baseline magnetic field, resulting in macroscopic magnetization along a longitudinal axis. This magnetization is then displaced from its resting position by the application of radiofrequency (RF) pulses (excitation). The return of the excited states to their resting positions (relaxation), or the magnetization dynamics, is subsequently detected as relaxation signals using one or more RF receiver coils.

[0032] For spatial encoding, rapidly switched magnetic gradient fields are superimposed on the fundamental magnetic field. The acquired relaxation signals or the detected MR data initially exist as raw data in the frequency domain and can be transformed into the spatial domain (image space) by subsequent inverse Fourier transformation.

[0033] Contrast agents can also be used in magnetic resonance imaging to enhance contrast.

[0034] "Contrast agents" are substances or mixtures of substances that improve the visualization of the body's structures and functions in radiological imaging procedures.

[0035] Beispiele für Kontrastmittel sind in der Literatur zu finden (siehe z.B. A. S. L. Jascinth et al.: Contrast Agents in computed tomography: A Review, Journal of Applied Dental and Medical Sciences, 2016, Vol. 2, Issue 2, 143 - 149; H. Lusic et al.: X-ray-Computed Tomography Contrast Agents, Chem. Rev. 2013, 113, 3, 1641-1666; https: / / www.radiology.wisc.edu / wp-content / uploads / 2017 / 10 / contrast-agents-tutorial.pdf, M. R. Nough et al.: Radiographic and magnetic resonances contrast agents: Essentials and tips for safe practices, World J Radiol. 2017 Sep 28; 9(9): 339-349; L. C. Abonyi et al.: Intravascular Contrast Media in Radiography: Historical Development & Review of Risk Factors for Adverse Reactions, South American Journal of Clinical Research, 2016, Vol. 3, Issue 1, 1-10; ACR Manual on Contrast Media, 2020, ISBN: 978-1-55903-012-0; A. Ignee et al.: Ultrasound contrast agents, Endosc Ultrasound. 2016 Nov-Dec; 5(6): 355-362).

[0036] A representation of the area of ​​investigation within the meaning of the present invention can be a computed tomography scan, an MRI scan, an ultrasound image or the like.

[0037] A representation of the investigation area within the meaning of the present invention can be a representation in spatial space (image space), a representation in frequency space, or another representation. Preferably, representations of the investigation area are in a spatial representation or in a form that can be converted (transformed) into a spatial representation. A spatial representation is often also referred to as a pictorial representation or as an image (English: image) designated.

[0038] In a spatial representation, also referred to in this description as a spatial representation, the examination area is typically represented by a multitude of image elements (pixels or voxels), which may be arranged in a grid, for example. Each image element represents a portion of the examination area, and each image element may be assigned a color or grayscale value. A widely used format in radiology for storing and processing spatial representations is the DICOM format. DICOM (Digital Imaging and Communications in Medicine) is an open standard for storing and exchanging information in medical image data management.

[0039] A representation in spatial space can be transformed into a representation in frequency space, for example, by a Fourier transform. Conversely, a representation in frequency space can be transformed into a representation in spatial space, for example, by an inverse Fourier transform.

[0040] In a frequency-space representation, also referred to in this description as a frequency-space plot or frequency-space representation, the area under investigation is represented by a superposition of fundamental oscillations. For example, the area under investigation can be represented by a sum of sine and / or cosine functions with different amplitudes, frequencies, and phases. The amplitudes and phases can be plotted as a function of the frequencies, for example, in a two- or three-dimensional representation. Typically, the lowest frequency (origin) is placed at the center. The further one moves away from this center, the higher the frequencies. Each frequency can be assigned an amplitude, with which the frequency is represented in the frequency-space plot, and a phase, which indicates how far the respective oscillation is shifted relative to a sine or cosine oscillation.

[0041] Details about spatial representations and frequency-space representations and their respective conversion into one another are described in numerous publications, see e.g.: https: / / see.stanford.edu / materials / lsoftaee261 / book-fall-07.pdf.

[0042] A representation within the meaning of the present invention represents the area under investigation before or after the application of a quantity of a contrast agent.

[0043] The term "quantity" can refer to the absolute amount administered to a subject of investigation (e.g., measured in kilograms, moles, or liters); it can also refer to an administered dose, such as the amount of contrast agent (e.g., measured in kilograms, moles, or liters) per kilogram of the subject's body weight. Furthermore, "quantity" can refer to a concentration that is established in the subject of investigation, at least for a defined period after the administration of the contrast agent. Other definitions of quantity are conceivable.

[0044] Preferably, the contrast agent is an MR contrast agent. MR contrast agents exert their effect by altering the relaxation times of the structures that absorb the contrast agent. Two groups of substances can be distinguished: para- and superparamagnetic substances. Both groups possess unpaired electrons that induce a magnetic field around the individual atoms or molecules. Superparamagnetic contrast agents lead to a predominantly T2 shortening, while paramagnetic contrast agents lead primarily to a T1 shortening. The effect of these contrast agents is indirect, since the contrast agent itself does not emit a signal, but only influences the signal intensity in its surroundings. An example of a superparamagnetic contrast agent is iron oxide nanoparticles (SPIOs). superparamagnetic iron oxide ) .Examples of paramagnetic contrast agents include gadolinium chelates such as gadopentetate dimeglumine (trade name: Magnevist ®< etc.), gadoteric acid (Dotarem ®< , Dotagita ®< , Cyclolux ®< ), gadodiamide (Omniscan ®< ), gadoteridol (ProHance ®< ) and gadobutrol (Gadovist ®< ).

[0045] Preferably, the MRI contrast agent is a hepatobiliary contrast agent. A hepatobiliary contrast agent is characterized by its specific uptake by liver cells (hepatocytes), its accumulation in functional tissue (parenchyma), and its enhancement of contrast in healthy liver tissue. An example of a hepatobiliary contrast agent is the disodium salt of gadoxetic acid (Gd-EOB-DTPA disodium), described in US Patent No. 6,039,931A, which is commercially available under the brand names Primovist® and Eovist®.

[0046] The contrast agent can be administered in a weight-adjusted form as a bolus injection, for example into a vein in the arm.

[0047] The present invention enables the prediction of representations of an examination area after the application of different amounts of a contrast agent. In principle, the prediction can be based on a single representation that depicts the examination area without contrast agent or after the application of a first amount of contrast agent.

[0048] The prediction is made using a machine learning model.

[0049] A "machine learning model" can be understood as a computer-implemented data processing architecture. The model can receive input data and deliver output data based on this input data and model parameters. Through training, the model can learn a relationship between the input data and the output data. During training, model parameters can be adjusted to produce a desired output for a given input.

[0050] When training such a model, it is presented with training data from which it can learn. The trained machine learning model is the result of the training process. In addition to input data, the training data includes the correct output data (target data) that the model is to generate based on the input data. During training, patterns are recognized that map the input data to the target data.

[0051] During the training process, the input data for the training data is fed into the model, and the model generates output data. This output data is then compared to the target data. Model parameters are adjusted to reduce the deviations between the output data and the target data to a (defined) minimum.

[0052] The deviations can be analyzed using an error function (English: loss function ) can be quantified. Such an error function can be used to determine an error value (English: loss value The goal of the training process is to calculate the error value for a given pair of output and target data. This can involve modifying (adjusting) the parameters of the machine learning model to reduce the error value for all pairs in the training dataset to a (defined) minimum.

[0053] If the output and target data are numbers, for example, the error function can be the absolute difference between these numbers. In this case, a high absolute error value may mean that one or more model parameters need to be changed significantly.

[0054] For output data in the form of vectors, difference metrics between vectors such as the mean squared error, a cosine distance, a norm of the difference vector such as a Euclidean distance, a Chebyshev distance, an Lp norm of a difference vector, a weighted norm, or any other type of difference metric of two vectors can be chosen as the error function.

[0055] For higher-dimensional outputs, such as two-dimensional, three-dimensional, or even higher-dimensional outputs, an element-wise difference metric can be used. Alternatively or additionally, the output data can be transformed before calculating an error value, for example, into a one-dimensional vector.

[0056] In the present case, the machine learning model is trained using training data, starting from an initial representation representing an examination area without contrast agent or after the application of an initial amount of contrast agent, to predict a sequence of representations representing the examination area after the application of different amounts of contrast agent.

[0057] The varying amounts of contrast agent create a sequence in which the contrast agent quantities can, for example, increase or decrease. This can be demonstrated by... Fig. 1 explained.

[0058] Fig. 1 Figure 1 shows four exemplary and schematic spatial representations of an examination area after the application of different amounts of a contrast agent. The examination area is the lung of a human. The four representations T< R 1 , T< R 2 , T< R 3 and T< R 4 are arranged along an axis, where the axis represents the amount of contrast agent. a The representation T< R 1 indicates the contrast agent that was administered to the subject. Thus, the representation T< R 1 shows the area under investigation after the application of an initial amount. a 1 of the contrast agent; the representation T< R 2 shows the examination area after application of a second amount a 2 of the contrast agent; the representation T< R 3 shows the examination area after application of a third amount a 3 of the contrast agent; the representation T< R 4 shows the examination area after application of a fourth amount a 4 of the contrast agent. The first amount a 1 can be greater than or equal to zero, the second set a 2 , the third set a 3 and the fourth set a 4 In this example, each value is greater than zero. In this example, the second set a 2 is larger than the first set. a 1, the third set a 3 larger than the second set a 2 and the fourth set a 4 greater than the third set a 3 : 0 ≤ a 1 < a 2 < a 3 < a 4

[0059] In the Fig. 1 In the example shown, the amounts of contrast agent are therefore a consequence of increasing amounts of contrast agent.

[0060] This is evident in Fig. 1 in the representations T< R 1 to T< R 4 by an increasing contrast enhancement of the blood vessels compared to the surrounding tissue.

[0061] In the Fig. 1 (a) In the example shown, the quantity increases by a constant amount, i.e., the difference between the quantities of two immediately consecutive sets is the same for all immediately consecutive sets: a 2 - a 1 = a 3 - a 2 = a 4 - a 3 However, in a sequence of contrast agent doses, the amounts can also increase by different amounts. In the case of the Fig. 1 (b) In the example shown, the quantity increases by an increasing amount: a 2 - a 1 < a 3 - a 2 < a 4 - a 3

[0062] However, it is also possible that the quantities increase by a decreasing amount: a 2 - a 1 > a 3 - a 2 > a 4 - a 3 However, it is also possible that the representations are distributed differently along the set axis.

[0063] Furthermore, it is possible that the contrast agent amounts form a sequence of decreasing amounts or some other sequence.

[0064] It should be noted that the different amounts of contrast agent are not usually administered immediately sequentially. An additive effect of the different amounts is generally to be prevented. In order to generate a dataset for a study object that represents two or more representations of a study area after the administration of different amounts of contrast agent, the different amounts are usually administered at a time interval sufficient to ensure that the amount of contrast agent present in the study area from an earlier application is so small at the time of a later application of a different amount of contrast agent that it has no measurable influence on the representation of the study area.In other words, the expression, "the representation represents the area of ​​investigation after a quantity of contrast agent" preferably means "the representation represents the area of ​​investigation with a quantity of contrast agent".

[0065] It should be noted that representations of the study area are marked with the letter R in this description. The letter R may be followed by an index, such as R1, R2, R3, or R3. i The appended index indicates what the respective representation represents. The representation R 1 represents the domain of investigation of an object of study after the application of the set. a1 of a contrast agent; representation R 2 represents the area of ​​investigation of an object after the application of the amount a 2 of a contrast agent; representation R 3 represents the area of ​​investigation of an object after the application of the amount a 3 a contrast agent, and so on. In general, the representation R represents i the area of ​​investigation of an object of investigation after the application of the quantity a i a contrast agent, whereby i an index that represents the whole numbers of i until n runs through, whereby n an integer greater than two. The expression "an index i runs through all the numbers of a until b" means that i one after the other the values ​​of a until b assumes, so first the value a (i=a), then the value a +1 ( i=a +1) and so on, until ithe value b is reached ( i=b ) . The expression "every generated representation R j *< represents the scope of investigation after application of a set a j "of the contrast agent" means that the representation R 1 *< the examination area after application of a quantity a 1 of the contrast agent, the representation R 2 *< the examination area after application of a quantity a 2 of the contrast agent is represented, and so on. Representations used for training and representations generated during training are specifically marked in the claims and drawings with a preceding T<, as in the case of T< R 1 , T< R 2 and T< R kRepresentations generated by the machine learning model are indicated, in particular in the patent claims and the drawings, by a superscript asterisk symbol *, as in the case of R 1 *< , R 2 *< and T< R k -1 *< . The designations described here serve only for clarification, in particular to avoid objections regarding clarity in the patent granting procedure.

[0066] The machine learning model can be trained using training data to predict, starting with the first representation T< R 1, the representations T< R 2, T< R 3, and T< R 4 in succession. In other words, the machine learning model can learn the sequence of contrast agent quantities and then apply it to new representations.

[0067] For the sake of simplicity, the training will first be explained using three representations. This will then be extended to an arbitrary number.n representations, whereby n a whole number greater than two.

[0068] The following explanation of the training based on three representations will refer to Fig. 2 Reference made to.

[0069] Fig. 2 shows three representations, a first representation T< R 1 , a second representation T< R 2 and a third representation T< R 3 .

[0070] The first representation T< R 1 represents an area of ​​investigation of a subject without contrast medium or after an application of a first amount a 1 of a contrast agent, the second representation T< R 2 represents the examination area after an application of a second amount a 2 of the contrast agent and the third representation T< R 3 represents the examination area after application of a third amount a 3 of the contrast agent. The amounts of contrast agent a 1 ,a 2 and a 3 can, for example, form a consequence of increasing amounts of contrast agent (0 ≤ a 1 < a 2 < a 3) .

[0071] The three representations T< R 1 , T< R 2 , and T< R 3 form a dataset within training data TD. The training data TD comprises a multitude of such datasets. The term multitude preferably means more than 100. Each dataset includes at least two, usually three or more, representations of the area under investigation after the application of the respective amounts of contrast agent (which, in the case of a 1, can also be zero). The area under investigation is the same for all representations; only the object under investigation can vary; usually, each dataset originates from a different object under investigation. The statements made in this paragraph apply generally and not only with regard to the [reference to be added]. Fig. 2 Example shown.

[0072] In a first step (A), the first representation T< R 1 is fed into a machine learning model M. The machine learning model M is configured to generate an output T< R 2 *< based on the first representation T< R 1 and on the basis of model parameters MP (step (B)).

[0073] The output T< R 2 *< should approximate the second representation T< R 2 as closely as possible; ideally, the output T< R 2 *< should be indistinguishable from the second representation T< R 2. In other words, the output T< R 2 *< is a predicted second representation. The output T< R 2 *< is compared with the (actual) second representation T< R 2, and the deviations are quantified using an error function LF 2. In the case of multiple objects under investigation, an error value LV 2 can be calculated for each pair of output T< R 2 *< and second representation T< R 2 using the error function LF 2.

[0074] Examples of error functions that are general (not limited to the example in Fig. 2 ) can be used to carry out the present invention, are L1 fault function ( L1 loss), L2 fault function (L2 loss), Lp loss function, structural similarity index measure ( structural similarity index measure (SSIM)), VGG error function (VGG loss), perceptual error function, or a combination of the above functions. Further details on error functions can be found, for example, in the scientific literature (see, e.g., R. Mechrez et al.: The Contextual Loss for Image Transformation with Non-Aligned Data, 2018, arXiv:1803.02077v4; H. Zhao et al.: Loss Functions for Image Restoration with Neural Networks, 2018, arXiv:1511.08861v3).

[0075] The output T< R 2 *< is fed into the machine learning model M in step (C). Even if Fig. 2 This might give a different impression: the machine learning model M, which in step (A) receives the first representation T< R 1 and in step (C) the generated representation T< R 2 *<, is the same model. That is, the machine learning model M is not only configured to generate a representation T< R 2 *< based on the first representation T< R 1, but it is also configured to generate an output T< R 3 *< from the representation T< R 2 *< that is as close as possible to the third representation T< R 3. The output T< R 3 *< is a predicted third representation. The output T< R 3 *< is then compared to the third representation T< R 3.Using an error function LF 3, the deviation between the output T< R 3 *< and the third representation T< R 3 can be quantified; in the case of multiple objects under investigation, an error value LV 3 can be determined for each pair of a third representation T< R 3 and a generated representation T< R 3 *<.

[0076] Preferably, the machine learning model is end-to-end (English: end-to-end). end-to-end ) trained. This means that the machine learning model is trained simultaneously to generate a predicted second representation based on the first representation and a predicted third representation based on the predicted second representation. Preferably, an error function is used that takes into account both the deviations between the predicted second representation and the second representation, as well as the deviations between the predicted third representation and the third representation.

[0077] It is possible to quantify the deviations between the representation T< R 2 and the generated representation T< R 2 *< using the error function LF 2, and to quantify the deviations between the third representation T< R 3 and the generated representation T< R 3 *< using the error function LF 3. An error function LF that takes both deviations into account can, for example, be the sum of the individual error functions: LF = LF 2 + LF 3. It is also possible to weight the contributions that the individual error functions LF 2 and LF 3 have to the overall error function LF differently: LF = w 2 · LF 2 + w 3 · LF 3, where w 2 and w 3 are weighting factors that can take values ​​between 0 and 1, for example. A weighting factor of zero can be used, for instance, if a representation is missing in a dataset (more on this can be found in the description later).The weight factors can also be changed during training.

[0078] The in Fig. 2 The training method described is a preferred embodiment of the present disclosure and comprises the following steps: Receiving training data, wherein the training data for a variety of investigation objects comprise representations T< R 1 to T< R 3 of an investigation area, where the representation T< R 1 is the investigation area without contrast medium or after application of a quantity a 1 represents a contrast agent, the representation T< R 2 represents the examination area after application of a quantity a 2 of the contrast agent, and the representation T< R 3 represents the examination area after application of a quantity a 3 of the contrast agent, with the amounts a 1 to a 3 a sequence of preferably increasing sets, training a machine learning model, wherein the training for each of the multitude of objects under investigation comprises: feeding the representation T< R 1 to the machine learning model, wherein the machine learning model is configured to generate a representation T< R 2 *< at least partially based on the representation T< R 1 and on the basis of model parameters, feeding the representation T< R 2 *< to the machine learning model, wherein the machine learning model is further configured to generate a representation T< R 3 *< at least partially based on the representation T< R 2 *< and on the basis of model parameters, quantifying deviations between the representation T< R 2 and the generated representation T< R 2 *< as well as between the representation T< R 3 and the generated representation T< R 3 *< ,Minimize deviations by modifying the model parameters, saving the trained machine learning model, and / or using the machine learning model for prediction.

[0079] When the trained machine learning model is used for prediction (later), the goal may be to predict a third representation based on a first representation. In principle, a model could be trained for this purpose to generate the third representation directly based on the first ("direct training"). However, according to the invention, the machine learning model is trained not to generate the third representation directly (in one step) based on the first representation, but in two steps. In the first step, a second representation is predicted, and the third representation is then predicted based on this second representation. The advantage of the iterative approach according to the invention compared to the aforementioned "direct training" is, among other things, that additional training data (e.g.,Second representations, representing the study area in the second state, can be used, thus achieving higher predictive accuracy. Furthermore, the model does not learn a mapping from one representation to another (or, as in the case of WO2019 / 074938A1, a mapping from multiple representations to one representation), but rather it learns the influence of contrast agent on the representations of the study area, specifically how increasing or decreasing amounts of contrast agent affect the representations. The more amounts of contrast agent the training data covers, the more accurately the model can learn the influence of different amounts.

[0080] Is that in Fig. 2 The machine learning model shown can be trained and used for prediction. This is demonstrated in Fig. 3 As shown, in a first step (A), an initial representation R1 is fed to the trained machine learning model M. The first representation R1 represents the domain of investigation of a new object of study after the application of the first set. a 1 of a contrast agent (where the first amount a(1 can also be zero). The term "new" can mean that no representations of the object of study were used when training the machine learning model M. In step (B), the model M generates a second representation R2*< based on the first representation R1 and the model parameters MP. The second representation R2*< represents the domain of the object of study after the application of the second set a2. In a third step (C), the generated second representation R2*< is fed into the machine learning model M. In step (D), the model M generates a third representation R3*< based on the second representation R2*< and the model parameters MP. The third representation R3*< represents the domain of the object of study after the application of the third set. a 3 contrast agent.

[0081] The in Fig. 3 presented prediction method, which is based on the in Fig. 2 The training method described is a preferred embodiment of the present disclosure and comprises the following steps: Receiving a representation R 1 of an investigation area, wherein the representation R 1 represents the investigation area without contrast medium or after application of an initial amount a 1 of a contrast agent, generating a representation R 2 *< of the investigation area using a trained machine learning model at least partially based on the representation R 1 , wherein the generated representation R 2 *< represents the investigation area after application of a second quantity a 2 of the contrast agent, wherein the second quantity a 2 is preferably greater than zero and preferably larger than the first quantity a1 is, generating a representation R 3 *< of the domain of investigation using the machine learning model at least partially based on the generated representation R 2 *< , where the generated representation R 3 *< represents the domain of investigation after application of a third set a 3 of the contrast agent, with the third amount a 3 preferably larger than the second quantity a 2 is, output and / or storage and / or transmission of the representation R 3 *< , wherein the machine learning model was trained on the basis of training data, wherein the training data comprises a large number of datasets, each dataset being an initial representation T R 1 , a second representation T R 2 and a third representation T R 3 includes, with the first representation T R 1 the examination area without contrast medium or after application of the first amount a 1 the contrast agent represents the second representation T R 2 the area under investigation after application of the second quantity a 2 of the contrast agent and the third representation T R 3 the area under investigation after application of the third quantity a 3 of the contrast agent, whereby the machine learning model was trained, at least partially, on the basis of the first representation T R 1 a representation T R 2 * to generate and at least partially based on the generated representation T R 2 * a representation T R 3 * to generate, whereby deviations between the representations occur during training. T R 2 and T R 2 * as well as between the representations T R 3 and T R 3 * They are quantified and model parameters are modified to minimize the deviations.

[0082] In general, the machine learning model can be trained by successively applying a number, starting from an initial representation T< R 1. n-1 from representations T< R 2 *< to T< R n *< to generate, whereby nan integer greater than 2, where the first representation T< R 1 is the examination area without contrast agent or after application of a first amount a 1 represents a contrast agent and each generated representation T< R j *< the area of ​​investigation after application of a quantity a j of the contrast agent, whereby j an index that represents whole numbers from 2 to n passes through, with the quantities a 1 to a n a sequence of preferably increasing sets, wherein the generation of the representation T< R 2 *< is at least partially based on the representation T< R 1 and the generation of each further representation T< R k *< at least partially based on the previously generated representation T< R k -1 *< is done, whereby k an index that represents whole numbers from 3 to n goes through.

[0083] This can be illustrated by the example of Fig. 4 explained in more detail. Fig. 4 can be considered an extension of the in Fig. 2 The shown scheme of three representations can be applied to any number (at least within defined limits). n are understood as representations, whereby n an integer greater than 2. Preferably the number n in the range of 3 to 100. The number n However, it can also be greater than 100.

[0084] Fig. 4 This shows a machine learning model M. The machine learning model M is shown three times; however, it is always the same model. The machine learning model M is trained to successively process a number n-1 to predict representations. In the present example, nAn integer greater than 3. Model M is fed a first representation T< R 1 as input data. The first representation T< R 1 represents an area of ​​investigation of a test object without contrast agent or after the application of a first amount a 1 of contrast agent. The machine learning model M is configured to generate an output T< R 2 *< based at least partially on the first representation T< R 1 and on model parameters MP. The output T< R 2 *< represents a predicted second representation that represents the area of ​​investigation after the application of a second amount a 2 of the contrast agent. The model is further configured to generate an output T< R 3 *< based at least partially on the output T< R 2 *< and model parameters MP. The output T< R 3 *< represents a predicted third representation that represents the area of ​​investigation after the application of a third amount a 3 of the contrast agent. The model is further configured to generate, at least partially, another output based on the output T< R 3 *< and model parameters MP, which represents a predicted representation of the examination area after the application of a further amount of the contrast agent. This scheme is used up to an output T< R n *< continued. The output T< R n *< represents a predicted nth representation that defines the domain of investigation after applying an nth set a n of the contrast agent. The amounts of contrast agent a 1 , a 2 , a 3 , ... , a n form a sequence. This sequence could, for example, be a sequence of increasing amounts of contrast agent. In other words, preferably the second quantity a₂ is greater than zero and greater than the first quantity a₁, the third quantity a 3 greater than the second amount a 2 ,the fourth set a 4 greater than the third set a 3 and so on. In general terms: the j-th set a j is preferably larger than the ( j -1)-th quantity a j -1 , where j an index that represents the numbers from 2 to n goes through.

[0085] In the case of the in Fig. 4 In the example shown, besides the first representation T< R, there are 1 further (real, metrologically generated) representations T< R j (with j = 2 to n) of the area under investigation after the application of the quantities a j (with j = 2 to n) of the contrast agent, which serve as target data ( ground truth) can be used to train the machine learning model. This allows the deviation between an output T < R to be determined. j *< (the one predicted j-th representation of the domain of investigation after the application of a set a j of the contrast agent) and the representation T< R j with an error function LF j can be quantified: an error function LF 2 can quantify the deviation between the representation T< R 2 and the generated representation T< R 2 *<, an error function LF 3 can quantify the deviation between the representation T< R 3 and the generated representation T< R 3 *<, and so on.

[0086] In end-to-end training, a total error function (LF) can be used that takes all individual deviations into account; this could, for example, be the sum of the individual deviations: LF = LF 2 + LF 3 + … + LF n = ∑ j = 2 n LF j

[0087] However, it could also be a weighted sum: LF = w 2 ⋅ LF 2 + w 3 ⋅ LF 3 + … + w n ⋅ LF n = ∑ j = 2 n w j ⋅ LF j where the weight factors w i for example, they can take values ​​from 0 to 1.

[0088] The weighting has the advantage that different weights are assigned to the representations fed into the machine learning model when generating a predicted representation.

[0089] For example, it is possible that the weight factors w j in the series w 2 , w 3 , ..., w n with the amount of contrast agent a 2 , a 3 , ..., a n increase and thus representations whose amount of contrast agent is closer to the amount a n The increase can be weighted more heavily. This increase can be logarithmic, linear, quadratic, cubic, exponential, or some other type of increase. It is also conceivable to assign less weight to representations with a larger amount of contrast agent by adjusting the weighting factors in the series w₂, w₃, ..., w₁. n with increasing amount of contrast agent a 2 , a 3 , ..., a n decrease, thereby assigning more weight to representations that are closer to the initial representation R1. Such a decrease can also be logarithmic, linear, quadratic, cubic, exponential, or some other type of decrease.

[0090] It is also conceivable that the training data includes incomplete datasets. This means that some datasets of the objects under investigation do not include all representations T < R 1 to T < R n. As explained later in the description, even incomplete datasets can be used to train the machine learning model. If a representation T < R j If the weight factor w is missing, it can be used. j , which gives the error function to the representation T< R j and the generated representation T< R j *< is to be weighted, set to zero, where j an integer that has the values ​​2 to n can assume.

[0091] It is also conceivable that random numbers 1 ≤ were used in the training. j < k ≤ n and the associated sub-sequences of contrast agent quantities a j , a j+ 1 , ..., a k -1 , a k The learning problem described above can then be solved on these (optionally varying from time to time) subsequences. For example, it is conceivable that a random initial amount of contrast agent could be used. a j (1 ≤ j ≤ n- 2) is determined and the model on the basis of which always the representations of the following two contrast agent quantities a j+ 1 , a j +2 should synthesize.

[0092] In Fig. 4 It is also schematically shown that to generate a representation T< R j *< , which defines the scope of investigation after the application of a quantity a j of the contrast agent, in addition to the representation T< R j-1 *< , which defines the scope of investigation after the application of a quantity a j- 1 represents, also further representation T< R j -2 *< and / or T< R j -3 *< and / or ... to T< R 1 , which define the scope of investigation after the application of the further quantities a j -2 , a j -3 ... to a 1 represent, can be used, whereby j an integer that has the values ​​2 to n This can be assumed. This is expressed by the dashed arrows. Thus, to generate the representation T< R 3 *<, in addition to the representation T< R 2 *<, the representation T< R 1 can also be fed into the machine learning model. To generate the representation T< R 4 *<, in addition to the representation T< R 3 *<, the representation T< R 2 *< and / or the representation T< R 1 can also be fed into the machine learning model.

[0093] Additional information can be incorporated into the machine learning model to predict representations, such as information about the amount of contrast agent administered in each case. In other words, in addition to the first representation, information about the amount of contrast agent administered, represented by the first representation, can be used to predict the second representation. Similarly, the predicted third representation can be generated based on the predicted second representation and information about the second amount of contrast agent. If information about the amount of contrast agent is provided with a representation, the machine learning model "knows" its position within the sequence of contrast agent quantities.

[0094] Further information that can be used to train and later predict the machine learning model includes, for example, information about the area under investigation, the object under investigation, the conditions that prevailed when the representation was created, information about the contrast agent, information about the amount of contrast agent applied to the object under investigation, and other / further information.

[0095] The machine learning model of the present disclosure can be understood as a transformation that can be applied to a representation of a domain of investigation in one state of a sequence of states to predict a representation of the domain of investigation in a subsequent state of the sequence of states. Each state is represented by an amount of applied contrast agent. The transformation can be applied simply (once) to predict a representation of the domain of investigation in the immediately subsequent state, or multiple times (multiple times) to predict a representation of the domain of investigation in a state further down the sequence of states.

[0096] Is there a consequence of n States Z1 to Z n present, and is available for every state Z i a representation R i before, which the investigation area in state Z i If represented, then the machine learning model M can be applied q-fold starting from the representation R 1 to create a representation R 1+ q to generate the investigation area in the state Z 1+ q represents, where q represents the values ​​1 to n -1 can be assumed: M q R 1 = R 1 + q * q = 1 : M R 1 = R 2 * q = 2 : M R 2 * = M M R 1 = M 2 R 1 = R 3 * q = 3 : M R 3 * = M M R 2 * = M M M R 1 = M 3 R 1 = R 4 * … q = n − 1 : M R n − 1 * = M M R n − 2 * = … = M n − 1 R 1 = R n *

[0097] This also applies analogously to the training procedure, where the q-fold transformation M can be described by the following formula: M q R <mprescripts / > <none / > T 1 = R <mprescripts / > <none / > T 1 + q *

[0098] The respective state indicates the amount of contrast agent that has been administered. A sequence of states can therefore, for example, characterize the area under examination after the application of a succession of increasing amounts of contrast agent.

[0099] An error function that accounts for all deviations between predicted representations T < R q *< and real (measurably generated) representations T< R q Quantified, for example, can be expressed by the following formula: LV = w 2 ⋅ d M R 1 <none / > <mprescripts / > <none / > T , R 2 <none / > <mprescripts / > <none / > T + w 3 ⋅ d M 2 R 1 <none / > <mprescripts / > <none / > T , R 3 <none / > <mprescripts / > <none / > T + … + w n ⋅ d M n − 1 R 1 <none / > <mprescripts / > <none / > T , R n <none / > <mprescripts / > <none / > T

[0100] LV is the error value that results for a data set encompassing the (real, metrologically generated) representations T< R 1 , T< R 2 , ..., T< R n. d is an error function that measures the deviations between a predicted representation M ( T< R q -1 ) and the representation T< R q quantified. As described earlier, this can be, for example, one of the following error functions: L1 loss, L2 loss, Lp loss, structural similarity index measure (SSIM), VGG loss, perceptual loss or a combination thereof.

[0101] w 2 , w 3 , ..., w n These are weight factors that have already been described earlier in the description.

[0102] nindicates the number of states (the number of different amounts of contrast agent).

[0103] It is also conceivable that the error value is the maximum deviation calculated for a data set: LV = max w 2 ⋅ d M R 1 <none / > <mprescripts / > <none / > T , R 2 <none / > <mprescripts / > <none / > T ; w 3 ⋅ d M 2 R 1 <none / > <mprescripts / > <none / > T , R 3 <none / > <mprescripts / > <none / > T ; … ; w n ⋅ d M n − 1 R 1 <none / > <mprescripts / > <none / > T , R n <none / > <mprescripts / > <none / > T where, in this formula as well, different weights can be assigned to the individual deviations through the weighting factors.

[0104] As mentioned earlier in the description, it should be noted that for training the machine learning model, it is not necessary for each training dataset to contain representations of all contrast agent amounts that the model is to learn. In principle, two representations of the examination area after the application of two different amounts of contrast agent are sufficient. This will be illustrated with an example. Let us assume that the machine learning model is to be trained with representations of an examination area with increasing amounts of contrast agent. a 1 to a 6 to predict. It is assumed that training data comprising 10 datasets from 10 test subjects is sufficient to train the machine learning model. Each dataset includes representations of the test area after application of different amounts of contrast agent, e.g.: Data set 1: T< R 1 , T< R 3 , T< R 4 , T< R 5 , T< R 6 Data set 2: T< R 1 , T< R 2 , T< R 4 , T< R 6 Data set 3: T< R 1 , T< R 2 , T< R 3 , T< R 4 , T< R 5 Data set 4: T< R 1 , T< R 2 , T< R 3 , T< R 5 , T< R 6 Data set 5: T< R 2 , T< R 3 , T< R 5 , T< R 6 Data set 6: T< R 2 , T< R 3 , T< R 4 , T< R 5 , T< R 6 Data set 7: T< R 2 , T< R 3 , T< R 5 , T< R 6 Data set 8: T < R 3 , T< R 5 , T< R 6 Data set 9: T< R 3 , T< R 4 , T< R 5 , T< R 6 Data set 10: T< R 3 , T< R 4 , T< R 6

[0105] In this example, there is no single "complete" dataset, that is, no dataset that includes all possible representations T < R 1, T < R 2, T < R 3, T < R 4, T < R 5, and T < R 6. Nevertheless, it is possible to train the machine learning model based on such training data, creating a representation of the domain of study after applying any set a 2 to a6. This is a further advantage of the present invention over the "direct training" described above.

[0106] Once the machine learning model is trained, new (i.e., not used in training) representations of an area of ​​investigation can be fed to the model after application of a quantity of the contrast agent, and the model can predict (generate) one or more representations of the area of ​​investigation after application of another (larger) quantity.

[0107] This is an example and a schematic representation in Fig. 5 depicted. In the Fig. 5 The example shown starts from a representation R p that defines the domain of investigation after the application of the set a p a contrast agent represents a sequence of representations R p +1 *< , R p +2 *< , R p +3 *< and R p+4 *< generates the investigation area after the application of the quantities a p +1 , a p +2 , a p +3 and a p +4 represent the contrast agent amounts a p +1 , a p +2 , a p +3 and a p +4 preferably form a sequence of increasing quantities: a p +1 < a p +2 < a p +3 < a p +4 .

[0108] In step (A), the machine learning model M is assigned the representation R. p supplied. In addition to the representation R p The machine learning model M can also be given information about the amount of contrast agent. a p and / or further / other information will be provided, as described earlier in this description.

[0109] In step (B), the machine learning model M generates the representation R p+1 *< based on the input data, which defines the domain of investigation after the application of the set a p +1 represents.

[0110] In step (C), the previously generated representation R is added to the machine learning model M. p +1 *< supplied. In addition to the representation R p +1 *< can also provide the machine learning model M with information about the amount of contrast agent. a p +1 and / or further information can be added. Furthermore, the machine learning model can also be given the representation R. p and / or information on the amount of contrast agent a p be supplied.

[0111] Preferably, the machine learning model M comprises a memory S that stores input data (and preferably also output data), so that input data and / or generated output data do not need to be received and / or entered again, but are already available to the machine learning model. This applies not only to the use of the trained machine learning model for prediction described in this example, but also to the training of the machine learning model according to the invention and generally to all other embodiments.

[0112] In step (D), the machine learning model M generates the representation R p+2 *< based on the input data, which defines the examination area after the application of the contrast agent quantity. a p +2 represents.

[0113] In step (E), the previously generated representation R is added to the machine learning model M. p +2 * supplied. In addition to the representation R p +2 *< can also provide the machine learning model M with information about the amount of contrast agent a p +2 and / or further information can be added. Furthermore, the machine learning model can also be given the representation R. p and / or the representation R p +1 *< and / or information on the contrast agent quantities a p and / or a p +1 will be added.

[0114] In step (F), the machine learning model M generates the representation R based on the input data. p +3 *< , which define the examination area after the application of the contrast agent quantity a p +3 represents.

[0115] In step (G), the previously generated representation R is added to the machine learning model M. p +3 * supplied. In addition to the representation R p+3 *<, information about the amount of contrast agent can also be added to the machine learning model M. a p +3 and / or further information can be added. Furthermore, the machine learning model can also be given the representation R. p and / or the representation R p +1 *< and / or the representation R p +2 *< and / or information on the contrast agent quantities a p and / or a p +1 and / or a p-2 are supplied.

[0116] In step (H), the machine learning model M generates the representation R based on the input data. p +4 *< , which define the examination area after the application of the contrast agent quantity a p +4 represents.

[0117] The generated representations R p+1 *< , R p+2 *< , R p+3 *< and / or R p+4 *< can be output (e.g. displayed on a monitor and / or printed with a printer) and / or stored in a data storage device and / or transmitted to a (separate) computer system.

[0118] The inventive machine learning model was trained starting from a first representation T< R 1, which defines the investigation domain after the application of a first set a 1 of the contrast agent represents a sequence of predicted representations T< R 2 *< , T< R 3 *< , ..., T< R n *< to generate, where each predicted representation T< R j *< the area of ​​investigation after the application of a quantity a j of the contrast agent, whereby j an index that represents whole numbers from 2 to n If the process runs through the model, then such a model can be assigned a new representation R. p are supplied, which the area under investigation after application of the quantity a p of the contrast agent, and the machine learning model can successively generate the predicted representations R p +1 *< , R p +2 *< , ... , R n *< generate, where p a number that has the values ​​2 to n can assume.

[0119] This means that the trained model does not necessarily receive a new representation R 1 that defines the domain of investigation after the application of the first set. a 1 represents, must be supplied, and the trained machine learning model is not limited to only representing R n , which the investigation area after the application of the (preferably largest) quantity a n represented, predict. Instead, one can predict the sequence of contrast agent amounts. a 1 to a n "Enter" and "exit" at any point, i.e., based on any representation of the sequence of contrast agent quantities, predict any other representation of a subsequent concentration quantity in the sequence of concentration quantities.

[0120] For example, if a machine learning model has been trained to detect the effect of an increasing amount of contrast agent in a sequence of increasing contrast agent amounts a 1 , a 2 , a 3 , a 4. To learn, then the trained machine learning model can, for example, a representation R 1 is fed to generate a representation R 2 *< and / or R 3 *< and / or R 4 *<, or a representation R 2 is fed to generate a representation R 3 *< and / or R 4 *<, or a representation R 3 is fed to generate a representation R 4 *<, where the representation R 1 the area under investigation after application of the quantity a 1 represented, the representations R 2 and R 2 * the area under investigation after application of quantity a 2 represented, the representations R 3 and R 3 * the area under investigation after application of the quantity a 3 represented and the representations R 4 and R 4 * the area under investigation after application of the quantity a 4 represented.

[0121] It is even possible to generate representations that depict the examination area after the application of a contrast agent quantity that was not the subject of the training. Thus, instead of focusing on the representation R n *< (e.g., in the example above, R 4 *< ) can also be a representation R n +1 * (e.g. R 5 *< ), R n +2 *< (e.g., R 6 *< ) and so on can be generated. The trained machine learning model can therefore be used to continue the learned effect of a preferably increasing amount of contrast agent and to calculate representations that have never been generated by measurement. In this respect, the trained machine learning model can be used to extrapolate to very high amounts of contrast agent that would never be applied to a subject under investigation.

[0122] Furthermore, the predictions are not limited to increasing amounts of contrast agent. It is also possible to predict representations with decreasing amounts of contrast agent. Firstly, the machine learning model can, in principle, be trained in both directions: towards increasing amounts of contrast agent and towards decreasing amounts. Secondly, the machine learning model performs a transformation on a representation input into the model, a transformation that can, in principle, be reversed. By analyzing the learned mathematical functions of the model that transform an input representation into an output representation, inverse functions can be determined that reverse the process and transform the previous output representation back into the previous input representation.The inverse functions can then be used to predict representations with decreasing amounts of contrast agent, even if the model has been trained to predict representations with increasing amounts of contrast agent, and vice versa.

[0123] Fig. 6 shows an extension of the in Fig. 5 shown scheme. While in Fig. 5 based on a representation R p, which defines the examination area after the application of a contrast agent quantity a p represents, successively the representations R p +1 *< , R p +2 *< , R p +3 *< and R p +4 *< are generated, which define the scope of investigation after the application of the quantities a p +1 , a p+2 , a p +3 and a p +4 of the contrast agent represents, is in Fig. 6 shown how, based on the representation R p , which examine the area after the application of the contrast agent a p represented, one after the other, the representations R p +1 *< to R p + q *< are generated, whereby p an integer that has the values ​​2 to n (or, as described, even larger values ​​than n ) can assume, and q is an integer greater than 1. In Fig. 6 The iterative nature of the machine learning model M becomes particularly evident. The machine learning model M is developed starting from the representation R. p applied q times, whereby (q-1) times the output data is fed back into the machine model.

[0124] It should be noted that the (trained or untrained) machine learning model does not have to be applied to a complete radiological image (e.g., an MRI scan, a CT scan, or the like). It is possible to apply the machine learning model to only a portion of a radiological image. For example, it is possible to first segment a radiological image to identify an area of ​​interest. region of interest) to identify / select. The model can then be applied exclusively to the area of ​​interest, for example.

[0125] Furthermore, the application of the machine learning model can include one or more preprocessing and / or postprocessing steps. For example, it is conceivable to first subject a received representation of the domain of investigation to one or more transformations, such as motion correction, color space conversion, normalization, segmentation, a Fourier transform (e.g., to convert from a space-space representation to a frequency-space representation), an inverse Fourier transform (e.g., to convert from a frequency-space representation to a space-space representation), and / or the like.In a further step, the transformed representation can be fed into the machine learning model, which then goes through a series of iterations (cycles) to generate a series of further (subsequent) representations of the domain of investigation in a series of further (subsequent) states, starting from the transformed representation.

[0126] The machine learning model according to the invention may, for example, be an artificial neural network, or it may comprise one.

[0127] An artificial neural network comprises at least three layers of processing elements: a first layer with input neurons (nodes), an N-th layer with at least one output neuron (node), and N-2 inner layers, where N is a natural number and greater than 2.

[0128] The input neurons are used to receive the representations. Typically, there is one input neuron for each pixel or voxel of a representation if the representation is a spatial representation in the form of a raster graphic. Additional input neurons may be present for additional input values ​​(e.g., information about the study area, the object under investigation, the conditions that prevailed during the creation of the representation, information about the contrast agent, information about the amount of contrast agent applied to the object, and other / additional information).

[0129] The output neurons can be used to output a predicted representation.

[0130] The processing elements of the layers between the input neurons and the output neurons are interconnected in a predetermined pattern with predetermined connection weights.

[0131] Preferably, the artificial neural network is a so-called Convolutional Neural Network (CNN) or it includes one.

[0132] A CNN typically consists essentially of filters (convolutional layer) and aggregation layers (pooling layer) that alternate and repeat, and ultimately of one or more layers of fully connected neurons (dense / fully connected layer).

[0133] Training a neural network can be performed, for example, using a backpropagation method. The goal is for the network to predict the dynamics of the domain under investigation as reliably as possible, from a given state through at least one intermediate state to a final state. The quality of the prediction is described by an error function. The aim is to minimize this error function. In the backpropagation method, the artificial neural network is trained by changing the connection weights.

[0134] In the trained state, the link weights between the processing elements contain information regarding the effect of an increasing or decreasing amount of contrast agent on the examination area, which can be used to predict, based on an initial representation representing the examination area after the application of an initial amount of contrast agent, one or more representations representing the examination area after the application of a different amount of contrast agent.

[0135] A cross-validation method can be used to split the data into training and validation datasets. The training dataset is used for backpropagation training of the network weights. The validation dataset is used to verify the prediction accuracy of the trained network when applied to unknown data.

[0136] The artificial neural network can have an autoencoder architecture; for example, the artificial neural network can have an architecture like the U-Net (see, for example, O. Ronneberger et al.: U-net: Convolutional networks for biomedical image segmentation, International Conference on Medical image computing and computer-assisted intervention, pages 234-241, Springer, 2015, https: / / doi.org / 10.1007 / 978-3-319-24574-4_28).

[0137] The artificial neural network can be a Generative Adversarial Network (GAN).

[0138] The artificial neural network can be a recurrent neural network or comprise one. Recurrent or feedback neural networks are neural networks that, unlike feedforward networks, are characterized by connections between neurons of one layer and neurons of the same or a preceding layer. The artificial neural network can, for example, be a Long short-term memory (LSTM) (see e.g. Y. Gao et al.: Fully convolutional structured LSTM networks for joint 4D medical image segmentation, DOI: 10.1109 / ISBI.2018.8363764).

[0139] The artificial neural network can be a transformer network (see e.g. D. Karimi et al.: Convolution-Free Medical Image Segmentation using Transformers, arXiv:2102.13645 [eess.IV]).

[0140] The present invention can be used, for example, to reduce the amount of contrast agent in a radiological examination.

[0141] For each contrast agent, there is a recommended amount that can be administered to a subject for a defined purpose. Contrast agents typically have a marketing authorization for a specific purpose, and this authorization includes information about the amount to be administered. An amount recommended by a manufacturer and / or distributor of a contrast agent, or an amount prescribed by a marketing authorization, is referred to in this description as the standard amount.

[0142] For example, the standard amount of Gd-EOB-DTPA disodium is 0.025 mmol / kg body weight.

[0143] The present invention can be used, for example, to predict a representation of the examination area after the application of a standard amount of contrast medium, based on a representation that represents an examination area after the application of a smaller amount of contrast medium than the standard amount.

[0144] For this purpose, a machine learning model can be trained on training data to learn the influence of an increasing amount of contrast agent on the representation of the examination area.

[0145] The training dataset can, for example, include not only a representation representing the examination area after the application of the standard amount of contrast agent, but also multiple representations representing the examination area after the application of different amounts of contrast agent, where the amounts are typically smaller than the standard amount. For instance, the training data might include one or more of six representations for each of a multitude of examination objects: a first representation, a second representation, a third representation, a fourth representation, a fifth representation, and a sixth representation.

[0146] The first representation can depict the area examined without contrast agent. The second representation can depict the area examined after the administration of, for example, 20% of the standard amount of contrast agent. The third representation can depict the area examined after the administration of, for example, 40% of the standard amount of contrast agent. The fourth representation can depict the area examined after the administration of, for example, 60% of the standard amount of contrast agent. The fifth representation can depict the area examined after the administration of, for example, 80% of the standard amount of contrast agent. The sixth representation can depict the area examined after the administration of, for example, 100% of the standard amount of contrast agent.

[0147] The machine learning model can be trained on the basis of the training data, starting from a representation that represents the area of ​​investigation after the application of a quantity of the contrast agent (which can also be zero), by applying the machine learning model one or more times to predict another representation that represents the area of ​​investigation after the application of a different quantity of the contrast agent.

[0148] Once the model is trained, it can be used for prediction. To do this, the trained machine learning model is provided with a representation of the examination area (of a different object not considered during training). This representation depicts the examination area after the application of a quantity of contrast agent that represents 0%, 20%, 40%, 60%, or 80% of the standard amount of contrast agent. Using the model, a representation of the examination area that corresponds to the standard amount of contrast agent can be predicted by applying the trained machine learning model five times (in the case of 0%), four times (in the case of 20%), three times (in the case of 40%), twice (in the case of 60%), or once (in the case of 80%).

[0149] In another application example, the present invention can be used to predict a representation that represents the investigation area after the application of a larger quantity than the standard quantity, without having to apply such a large quantity to an investigation object in the training.

[0150] To continue with the previously mentioned example, the trained machine learning model can be repeatedly applied to predict a representation of the examination area that reflects the area after applying 120%, 140%, 160%, 180%, or 200% of the standard amount. Values ​​exceeding 200% are also possible. While verification that a predicted representation would actually appear that way in real-world scenarios would be impossible, as applying such large amounts might be ethically unacceptable, such a predicted representation can be beneficial if a radiologist identifies functional impairments, lesions, or other abnormalities in such images that they might not detect or would be less certain of detecting in other images.

[0151] In a particularly preferred embodiment, the present invention is used to achieve artificial contrast enhancement in a CT scan produced using an MR contrast agent.

[0152] MRI contrast agents can, in principle, also be used in computed tomography (CT). However, the contrast enhancement in CT is lower than in magnetic resonance imaging (MRI) because the contrast of MRI contrast agents in MRI is based on a different physical effect than the contrast of MRI contrast agents in CT. Furthermore, MRI contrast agents result in less contrast enhancement in CT than conventional CT contrast agents because they exhibit lower X-ray absorption. Nevertheless, using an MRI contrast agent in CT can be advantageous because some MRI contrast agents are specifically absorbed by or bind to certain body cells, while no comparable CT contrast agents exist. Therefore, using an MRI contrast agent in CT can make visible features that were previously not visible, or not reliably visible, in CT scans.

[0153] For magnetic resonance imaging (MRI), intracellular contrast agents are available that are specifically taken up by certain body cells. One example is the disodium salt of gadoxetic acid (Gd-EOB-DTPA disodium), described in US patent no. 6,039,931A, which is commercially available under the brand names Primovist® and Eovist®. This contrast agent is a so-called hepatobiliary contrast agent, which is specifically taken up by liver cells (hepatocytes), accumulates in functional tissue (parenchyma), and enhances contrast in healthy liver tissue. It is approved for use in MRI. No comparable hepatobiliary contrast agent is known for computed tomography (CT).This also applies to gadofosveset, a gadolinium-based intravascular MRI contrast agent that binds to serum albumin in the blood, resulting in a long residence time of the contrast agent in the bloodstream (half-life in the blood approximately 17 hours). Another example of an intravascular contrast agent used in magnetic resonance imaging, but for which there is no comparable equivalent in computed tomography, is ferumoxytol, a colloidal iron-carbohydrate complex. Ferumoxytol can be administered as an intravenous injection and is available as a solution for intravenous injection under the brand names Rienso® or Ferahme®. The iron-carbohydrate complex exhibits superparamagnetic properties and can therefore be used (off-label) for contrast enhancement in MRI examinations (see, e.g., LP Smits et al.).: Evaluation of ultrasmall superparamagnetic iron-oxide (USPIO) enhanced MRI with ferumoxytol to quantify arterial wall inflammation, Atherosclerosis 2017, 263: 211-218).

[0154] A machine learning model can be trained to learn the effect of increasing amounts of an MR contrast agent on a representation of an examination area in a computed tomography scan.

[0155] The machine learning model can be presented with representations of the examination area of ​​a multitude of objects, representing the examination area in a CT scan after the application of different amounts of an MR contrast agent. The maximum amount applied could, for example, be the standard amount of MR contrast agent as recommended or approved for use in MRI.

[0156] Once the machine learning model is trained, representations can be predicted that represent the examination area in a CT scan after the application of a larger amount of MR contrast agent than the standard amount.

[0157] The present invention can be carried out wholly or partly with the aid of a computer system.

[0158] Fig. 7 This shows an example and schematic of such a computer system.

[0159] A "computer system" is a system for electronic data processing that processes data using programmable instructions. Such a system typically comprises a "computer," the unit containing a processor for performing logical operations, as well as peripherals.

[0160] In computer technology, "peripherals" refers to all devices connected to a computer that are used to control the computer and / or as input and output devices. Examples include monitors (screens), printers, scanners, mice, keyboards, drives, cameras, microphones, speakers, etc. Internal ports and expansion cards are also considered peripherals in computer technology.

[0161] The in Fig. 7 The computer system shown (1) comprises an input unit (10), a control and arithmetic unit (20) and an output unit (30).

[0162] The control and computing unit (20) serves to control the computer system (1), to coordinate the data flows between the units of the computer system (1) and to perform calculations.

[0163] The control and computing unit (20) is configured to cause the input unit (10) to produce a representation R p to receive a survey area of ​​a survey object, where the representation R p the examination area without contrast agent or after application of a quantity a p a contrast agent is represented, whereby p an integer that is less than n is, whereby n If an integer is greater than 2, the representation R p to feed a trained machine learning model, where the machine learning model has been trained on the basis of training data, starting from a first representation T< R 1, successively a number n-1 from representations T< R 2 *< to T< R n *< to generate, where the first representation T< R 1 is the area under investigation without contrast agent or after application of an initial amount a 1 of the contrast agent is represented and each generated representation T< R j *< the area of ​​investigation after application of a quantity a j of the contrast agent, whereby j an index that represents whole numbers from 2 to n passes through, with the quantities a 1 to a n a sequence of preferably increasing sets, wherein the generation of the representation T< R 2 *< is at least partially based on the representation T< R 1 and the generation of each further representation T< R k *< at least partially based on the previously generated representation T< R k -1 *< is done, whereby k an index that represents whole numbers from 3 to n runs through one or more representations of the machine learning model R p + q *< of the investigation area to receive, each of the one or more representations R p + q *< the area of ​​investigation after application of the quantity a p+q of the contrast agent, where q is an index representing the numbers from 1 to m runs through, whereby m an integer that is less than n-1 is or equals n-1 is or greater than n-1 is to cause the output unit (30) to display one or more representations R p + q *< to save and / or output and / or transmit to a separate computer system.

[0164] Fig. 8 Figure 1 shows, by way of example and schematically, a further embodiment of the computer system according to the invention. The computer system (1) comprises a processing unit (21) which is connected to a memory (22). The processing unit (21) and the memory (22) form a control and arithmetic unit, as described in Fig. 7 shown.

[0165] The processing unit (21) (English: processing unit The processing unit (21) may comprise one or more processors alone or in combination with one or more memories. The processing unit (21) may be ordinary computer hardware capable of processing information such as digital images (e.g., representations of the examination area), computer programs, and / or other digital information. The processing unit (21) typically consists of an arrangement of electronic circuits, some of which may be implemented as an integrated circuit or as several interconnected integrated circuits (an integrated circuit is sometimes referred to as a "chip"). The processing unit (21) may be configured to execute computer programs, which may be stored in a working memory of the processing unit (21) or in the memory (22) of the same or another computer system.

[0166] The memory (22) can be ordinary computer hardware capable of storing information such as digital images (e.g., representations of the study area), data, computer programs, and / or other digital information, either temporarily and / or permanently. The memory (22) can be volatile and / or non-volatile and can be permanently installed or removable. Examples of suitable memory include RAM (Random Access Memory), ROM (Read-Only Memory), a hard disk, flash memory, a removable computer disk, an optical disc, a magnetic tape, or a combination of the above. Optical discs can include read-only compact discs (CD-ROM), read / write compact discs (CD-R / W), DVDs, Blu-ray discs, and similar media.

[0167] In addition to the memory (22), the processing unit (21) can also be connected to one or more interfaces (11, 12, 31, 32, 33) to display, transmit, and / or receive information. The interfaces can include one or more communication interfaces (32, 33) and / or one or more user interfaces (11, 12, 31). The one or more communication interfaces (32, 33) can be configured to send and / or receive information, for example, to and / or from an MRI scanner, a CT scanner, an ultrasound camera, other computer systems, networks, data storage devices, or the like. The one or more communication interfaces (32, 33) can be configured to transmit and / or receive information via physical (wired) and / or wireless communication links.The one or more communication interfaces (32, 33) may include one or more interfaces for connecting to a network, e.g., using technologies such as cellular, Wi-Fi, satellite, cable, DSL, fiber optic, and / or the like. In some examples, the one or more communication interfaces (32, 33) may include one or more near-field communication interfaces configured to connect devices using near-field communication technologies such as NFC, RFID, Bluetooth, Bluetooth LE, ZigBee, infrared (e.g., IrDA), or the like.

[0168] The user interfaces (11, 12, 31) may include a display (31). A display (31) may be configured to show information to a user. Suitable examples include a liquid crystal display (LCD), a light-emitting diode (LED) display, a plasma display (PDP), or the like. The user input interface(s) (11, 12) may be wired or wireless and may be configured to receive information from a user into the computer system (1), for example, for processing, storage, and / or display. Suitable examples of user input interfaces (11, 12) include a microphone, an image or video recording device (e.g., a camera), a keyboard or keypad, a joystick, a touch-sensitive surface (separate from or integrated into a touchscreen), or the like.In some examples, the user interfaces (11, 12, 31) may include automatic identification and data capture (AIDC) technology for machine-readable information. This may include barcodes, radio frequency identification (RFID), magnetic stripes, optical character recognition (OCR), integrated circuit cards (ICC), and the like. The user interfaces (11, 12, 31) may also include one or more interfaces for communication with peripheral devices such as printers and the like.

[0169] One or more computer programs (40) can be stored in memory (22) and executed by the processing unit (21), which is programmed to perform the functions described in this description. The retrieval, loading, and execution of instructions from the computer program (40) can be sequential, with one instruction being retrieved, loaded, and executed at a time. However, the retrieval, loading, and / or execution can also be performed in parallel.

[0170] The memory (22) can also store the machine learning model according to the invention.

[0171] The computer system of the present disclosure can be implemented as a laptop, notebook, netbook and / or tablet PC; it can also be a component of an MRI scanner, a CT scanner or an ultrasound diagnostic device.

[0172] Fig. 9 Figure 100 schematically shows, in the form of a flowchart, an embodiment of the training method according to the invention. The training method comprises the following steps: (110) Receiving training data, wherein the training data comprise representations T< R 1 to T< R n of a domain of investigation for a multitude of objects of investigation, wherein n is an integer greater than two, where each representation T< R 1 is the domain of investigation after application of a set a i a contrast agent is represented, whereby i an integer in the range of 1 to n is, where the quantities a 1 to a n (120) train the machine learning model, training the model starting from representation T< R 1 successively representations T< R 2 *< until T< R n *< to generate, wherein the generation of the representation T< R 2 *< is at least partially based on the representation T< R 1 and the generation of each further representation T< R k *< at least partially based on the previously generated representation T< R k -1 *< is done, whereby k an index that represents whole numbers from 3 to n passes through, where the representation T< R k *< the area of ​​investigation after application of the quantity a k of the contrast agent and the representation T< R k -1 * the area of ​​investigation after application of the quantity a k- 1 of the contrast agent is represented, where deviations between the generated representations T< R are accounted for using an error function. j *< and the representations T< R 1 are quantified, where j an index that represents the numbers from 2 to n(130) uses the machine learning model to make predictions.

[0173] Fig. 10 Figure 200 schematically shows, in the form of a flowchart, an embodiment of the prediction method according to the invention. The prediction method comprises the following steps: (210) Receiving a representation R p of the area under investigation, where the representation R p the examination area without contrast agent or after application of a quantity a p a contrast agent is represented, whereby p an integer that is less than n is, whereby n an integer greater than 2, (220) Feeding the representation R p a trained machine learning model, where the machine learning model has been trained on the basis of training data, starting from a first representation T< R 1 successively a number n-1 from representations T< R 2 *< to T< R n *< to generate, where the first representation T< R 1 is the area under investigation without contrast agent or after application of an initial amount a 1 of the contrast agent is represented and each generated representation T< R j *< the area of ​​investigation after application of a quantity a i of the contrast agent, whereby j an index that represents whole numbers from 2 to n passes through, with the quantities a 1 to a n a sequence of preferably increasing sets, wherein the generation of the representation T< R 2 *< is at least partially based on the representation T< R 1 and the generation of each further representation T< R k *< at least partially based on the previously generated representation T< R k -1 *< is done, whereby k an index that represents whole numbers from 3 to n (230)Receiving one or more representations R p + q *< of the domain of investigation of the machine learning model, where each of the one or more representations R p + q *< the area of ​​investigation after application of the quantity a p+q of the contrast agent, where q is an index representing the numbers from 1 to m runs through, whereby m an integer that is less than n-1 is or equals n-1 is or greater than n-1 is, (240) Output and / or store and / or transmit one or more representations R p+q *< .

Claims

1. Computer-implemented method for training and using a machine-learning model (M), wherein the method comprises: - receiving training data, • where the training data comprise representations TR1 to TRn of an examination region for a multiplicity of examination objects, where n is an integer greater than two, • where each representation TRi represents the examination region after administration of an amount ai of a contrast agent, where i is an integer in the range from 1 to n, where the amounts a1 to an form a sequence of preferably increasing amounts, • where each examination object is preferably a human and the examination region is preferably part of the human, - training the machine-learning model (M), • where the model (M) is trained to generate, starting from the representation TR1, representations TR2* to TRn* one after the other, • where the representation TR2* is generated at least partly on the basis of the representation TR1 and each further representation TRk* is generated at least partly on the basis of the respective previously generated representation TRk-1*, where k is an index passing through integers from 3 to n, where the representation TRk* represents the examination region after administration of the amount ak of the contrast agent and the representation TRk-1* represents the examination region after administration of the amount ak-1 of the contrast agent, • where differences between the generated representations TRj* and the representations TRj are quantified by means of a loss function, where j is an index passing through the numbers from 2 to n, • where the differences are minimized by modifying model parameters (MP) of the machine-learning model (M), - using the machine-learning model (M) for prediction.

2. Method according to Claim 1, further comprising the steps of: - for each generated representation TRj*: calculating a loss value for each pair composed of a representation TRj and the generated representation TRj* with the aid of a loss function, where j is an index passing through the numbers from 2 to n, - calculating a total loss value with the aid of a total loss function, where the total loss function is a function of the loss values, - minimizing the total loss value by modifying the model parameters (MP).

3. Method according to Claim 2, wherein the total loss function has the following formula: LV = ∑ j = 2 n w j ⋅ LF j where LV is the total loss value, where LFj are the loss functions for calculating the loss values for the differences between the representation TRj and the generated representation TRj*, and wj are weight factors, where j is an index passing through the numbers from 2 to n.

4. Computer-implemented method for predicting a representation of an examination region of an examination object, comprising: - receiving a representation Rp of the examination region, • where the representation Rp represents the examination region without contrast agent or after administration of an amount ap of a contrast agent, where p is an integer less than n, where n is an integer greater than 2, where the examination object is preferably a human and the examination region is preferably part of the human, - feeding the representation Rp to a trained machine-learning model (M), • where the machine-learning model (M) has been trained on the basis of training data to generate, starting from a first representation TR1, a number n-1 of representations TR2* to TRn* one after the other, • where the first representation TR1 represents the examination region without contrast agent or after administration of a first amount a1 of the contrast agent and each generated representation TRj* represents the examination region after administration of an amount aj of the contrast agent, where j is an index passing through integers from 2 to n, where the amounts a1 to an form a sequence of preferably increasing amounts, • where the representation TR2* is generated at least partly on the basis of the representation TR1 and each further representation TRk* is generated at least partly on the basis of the respective previously generated representation TRk-1*, where k is an index passing through integers from 3 to n, - receiving one or more representations Rp+q* of the examination region from the machine-learning model (M), • where each of the one or more representations Rp+q* represents the examination region after administration of the amount ap+q of the contrast agent, where q is an index passing through the numbers from 1 to m, where m is an integer less than n-1 or equal to n-1 or greater than n-1, - outputting and / or storing and / or transmitting the one or more representations Rp+q*.

5. Method according to Claim 4, wherein the method comprises: - receiving the representation Rp of the examination region, where the representation Rp represents the examination region without contrast agent or after administration of the amount ap of the contrast agent, where p is an integer less than n, - feeding the representation Rp to the trained machine-learning model (M), where the machine-learning model (M) has been trained in a method according to any of Claims 1 to 3, - receiving the one or more representations Rp+q* of the examination region from the machine-learning model (M), where each of the one or more representations Rp+q* represents the examination region after administration of the amount ap+q of the contrast agent, where q is an index passing through the numbers from 1 to m, where m is an integer less than n-1 or equal to n-1 or greater than n-1, - outputting and / or storing and / or transmitting the one or more representations Rp+q*.

6. Method according to any of Claims 1 to 5, wherein each representation TRp+q* and / or Rp+q* generated by the machine-learning model (M) is returned to the trained machine-learning model (M) in order to generate a representation TRp+q+1* and / or Rp+q+1*.

7. Method according to any of Claims 4 to 6, wherein m is in the range from n to n+10.

8. Method according to any of Claims 1 to 7, wherein each generated representation is calculated according to the following formula: M g R 1 = R 1 + q * where M represents a transformation which is applied by the machine-learning model (M) to input data of the machine-learning model (M), where Mq means the q-times application of the transformation, where in a first step the transformation is applied to a first representation R1, in a second step the transformation is applied to the result of the application in the first step in that the result is passed back into the machine-learning model (M) and the procedure is repeated with each further result of a transformation until the transformation has been applied a total of q times, where q is an integer which may assume the values of 1 to m, where m is an integer less than n-1 or equal to n-1 or greater than n-1.

9. Method according to any of Claims 1 to 8, wherein the examination region is the human liver or is part of the human liver.

10. Method according to any of Claims 1 to 9, wherein the received representations are CT images, MRI images or ultrasound images.

11. Method according to any of Claims 1 to 9, wherein the received representations are CT images and wherein the contrast agent is an MRI contrast agent.

12. Computer system (1) comprising • an input unit (10), • a control and calculation unit (20) and • an output unit (30), wherein the control and calculation unit (20) is configured - to cause the input unit (10) to receive a representation Rp of an examination region of an examination object, where the representation Rp represents the examination region without contrast agent or after administration of an amount ap of a contrast agent, where p is an integer less than n, where n is an integer greater than 2, where the examination object is preferably a human and the examination region is preferably part of the human, - to feed the representation Rp to a trained machine-learning model (M), where the machine-learning model (M) has been trained on the basis of training data to generate, starting from a first representation TR1, a number n-1 of representations TR2* to TRn* one after the other, where the first representation TR1 represents the examination region without contrast agent or after administration of a first amount a1 of the contrast agent and each generated representation TRj* represents the examination region after administration of an amount aj of the contrast agent, where j is an index passing through integers from 2 to n, where the amounts a1 to an form a sequence of preferably increasing amounts, where the representation TR2* is generated at least partly on the basis of the representation TR1 and each further representation TRk* is generated at least partly on the basis of the respective previously generated representation TRk-1*, where k is an index passing through integers from 3 to n, - to receive from the machine-learning model (M) one or more representations Rp+q* of the examination region, where each of the one or more representations Rp+q* represents the examination region after administration of the amount ap+q of the contrast agent, where q is an index passing through the numbers from 1 to m, where m is an integer less than n-1 or equal to n-1 or greater than n-1, - to cause the output unit (30) to store the one or more representations Rp+q* and / or to output them and / or to transmit them to a separate computer system.

13. Computer program product comprising a computer program (40) that can be loaded into a working memory (22) of a computer system (1), where it causes the computer system (1) to execute the following steps: - receiving a representation Rp of an examination region of an examination object, where the representation Rp represents the examination region without contrast agent or after administration of an amount ap of a contrast agent, where p is an integer less than n, where n is an integer greater than 2, where the examination object is preferably a human and the examination region is preferably part of the human, - feeding the representation Rp to a trained machine-learning model (M), where the machine-learning model (M) has been trained on the basis of training data to generate, starting from a first representation TR1, a number n-1 of representations TR2* to TRn* one after the other, where the first representation TR1 represents the examination region without contrast agent or after administration of a first amount a1 of the contrast agent and each generated representation TRj* represents the examination region after an amount aj of the contrast agent, where j is an index passing through integers from 2 to n, where the amounts a1 to an form a sequence of preferably increasing amounts, where the representation TR2* is generated at least partly on the basis of the representation TR1 and each further representation TRk* is generated at least partly on the basis of the respective previously generated representation TRk-1*, where k is an index passing through integers from 3 to n, - receiving one or more representations Rp+q* of the examination region from the machine-learning model (M), where each of the one or more representations Rp+q* represents the examination region after administration of the amount ap+q of the contrast agent, where q is an index passing through the numbers from 1 to m, where m is an integer less than n-1 or equal to n-1 or greater than n-1, - outputting and / or storing and / or transmitting the one or more representations Rp+q*.

14. Use of a contrast agent in a radiological examination method, wherein the radiological examination method comprises the following steps: - receiving a representation Rp of an examination region of an examination object, where the representation Rp represents the examination region without contrast agent or after administration of an amount ap of the contrast agent, where p is an integer less than n, where n is an integer greater than 2, - feeding the representation Rp to a trained machine-learning model (M), where the machine-learning model (M) has been trained on the basis of training data to generate, starting from a first representation TR1, a number n-1 of representations TR2* to TRn* one after the other, where the first representation TR1 represents the examination region without contrast agent or after administration of a first amount a1 of the contrast agent and each generated representation TRj* represents the examination region after administration of an amount aj of the contrast agent, where j is an index passing through integers from 2 to n, where the amounts a1 to an form a sequence of preferably increasing amounts, where the representation TR2* is generated at least partly on the basis of the representation TR1 and each further representation TRk* is generated at least partly on the basis of the respective previously generated representation TRk-1*, where k is an index passing through integers from 3 to n, - receiving one or more representations Rp+q* of the examination region from the machine-learning model (M), where each of the one or more representations Rp+q* represents the examination region after administration of the amount ap+q of the contrast agent, where q is an index passing through the numbers from 1 to m, where m is an integer less than n-1 or equal to n-1 or greater than n-1, - outputting and / or storing and / or transmitting the one or more representations Rp+q*.

15. Kit comprising a contrast agent and a computer program product, wherein the computer program product comprises a computer program (40) that can be loaded into a working memory (22) of a computer system (1), where it causes the computer system (1) to execute the following steps: - receiving a representation Rp of an examination region of an examination object, where the representation Rp represents the examination region without contrast agent or after administration of an amount ap of the contrast agent, where p is an integer less than n, where n is an integer greater than 2, - feeding the representation Rp to a trained machine-learning model (M), where the machine-learning model (M) has been trained on the basis of training data to generate, starting from a first representation TR1, a number n-1 of representations TR2* to TRn* one after the other, where the first representation TR1 represents the examination region without contrast agent or after administration of a first amount a1 of the contrast agent and each generated representation TRj* represents the examination region after administration of an amount aj of the contrast agent, where j is an index passing through integers from 2 to n, where the amounts a1 to an form a sequence of preferably increasing amounts, where the representation TR2* is generated at least partly on the basis of the representation TR1 and each further representation TRk* is generated at least partly on the basis of the respective previously generated representation TRk-1*, where k is an index passing through integers from 3 to n, - receiving one or more representations Rp+q* of the examination region from the machine-learning model (M), where each of the one or more representations Rp+q* represents the examination region after administration of the amount ap+q of the contrast agent, where q is an index passing through the numbers from 1 to m, where m is an integer less than n-1 or equal to n-1 or greater than n-1, - outputting and / or storing and / or transmitting the one or more representations Rp+q*.