Method for automatically determining setting parameters for an imaging system

Abstract

The present invention relates to a computer-implemented method for automatically determining setting parameters, in particular scanning parameters, for an imaging system, in which in particular image data is generated by solving an inverse problem, such as a tomography system, in particular a computed tomography (CT) scanner, magnetic resonance (MRI) scanner or the like, comprising: determining, in particular extracting, on the basis of initial measurement data which is indicative of at least one characteristic of an object to be scanned, at least one object attribute which is assigned to the object to be scanned, and generating at least one setting parameter for the imaging system on the basis of the at least one object attribute.

Description

[0001] Method for automatically determining setting parameters for an imaging system

[0002] The present invention relates to a computer-implemented method for automatically determining at least one or more setting parameters for an imaging system.

[0003] Especially in non-destructive testing (NDT), for example with imaging systems and / or using tomography techniques such as computed tomography (CT), the accuracy of the results can depend heavily on the operator's expertise. The correct setting of various parameters, such as tube voltage, current, detector sensitivity, and / or scan speed, can be crucial, particularly for ensuring optimal image quality.

[0004] An inexperienced operator who does not set these parameters correctly or adequately may overlook and / or misinterpret potential errors and / or inaccuracies, which could lead to serious safety risks and / or hazards. In medicine, for example, a similar lack of expertise could lead to incorrect diagnoses. The absence of in-depth understanding or expertise and / or the correct handling / operation of such parameters can pose a significant challenge, potentially limiting the successful use of imaging systems and / or tomography devices in non-destructive testing (NDT).

[0005] Determining the optimal parameters for, for example, computed tomography scanners in non-destructive testing requires in-depth and / or comprehensive expert knowledge. Specialists who have undergone years of training and / or acquired extensive expertise through diverse experience are generally capable of performing this task. They are accustomed to considering various variables, parameters, and / or modifications to ensure ideal image quality and / or at least achieve the best possible result.

[0006] However, this very high level of expertise poses a problem in times of skilled labor shortages: there are only a few qualified individuals who can make such precise adjustments. These specialists are often heavily burdened by the complex nature of the task, which translates into a considerable time investment for calibration and / or optimization. The lack of sufficiently qualified specialists, combined with the significant time commitment and necessary expertise, can considerably hinder the efficient and / or widespread use of imaging systems and / or tomography procedures, particularly in NDT.

[0007] The shortage of qualified specialists can also lead to increased training costs. Companies must invest time and / or resources to train employees accordingly, which incurs additional costs and / or can temporarily impair productivity. Furthermore, the shortage of skilled workers can lead to higher employee turnover. New employees must first familiarize themselves with the specific requirements and / or settings of the tomography equipment, which can affect the continuity and / or efficiency of the examination process.

[0008] Even experienced professionals can make mistakes when operating tomography equipment, whether due to inattention, inadequate training, and / or human error. These errors can lead to inaccurate results, pose safety risks, create hazards, and / or compromise the reliability of measurements. In addition to operator error, misinterpretation of tomography results can also be problematic. If staff lack sufficient expertise and / or experience, they may, for example, overlook material defects and / or mistakenly classify them as harmless, creating potential safety risks.

[0009] Furthermore, technical problems can occur that impair image quality and / or disrupt the entire measurement or inspection process. These can range from simple equipment failures to complex software errors requiring extensive troubleshooting. Such obstacles can delay the inspection process and require additional resources, particularly additional expertise from the operator, to resolve them, further impacting the efficiency and / or reliability of the tomography-based method.

[0010] Another problem is the cost structure associated with the use of imaging systems and / or tomography equipment in NDT. The acquisition and operating costs for high-quality tomography equipment are often considerable, making this technology difficult for smaller companies and / or facilities to access. Added to this are the expenses for appropriately trained personnel and / or their training. This can force companies to resort to less precise and / or less expensive testing methods, which in turn can compromise the quality of the test results.

[0011] The object of the present invention is to overcome the disadvantages of the known prior art and, in particular, to provide an improved computer-implemented method for automatically determining setting parameters for an imaging system.

[0012] The problem is solved by the features of the independent claims. The dependent claims describe preferred embodiments. Further aspects, advantages, and features become apparent from the dependent claims, the description, and the accompanying drawings.

[0013] One aspect of the invention relates to a computer-implemented method for automatically determining setting parameters, in particular scan parameters, for an imaging system, in which image data are generated in particular by solving an inverse problem, such as a tomography system, in particular a computed tomography (CT) scanner, magnetic resonance imaging (MRI) scanner or the like, comprising:

[0014] Determine, in particular extract, based on initial measurement data that are indicative of at least one property of an object to be scanned, at least one object attribute that is assigned to the object to be scanned, and

[0015] Generating at least one setting parameter for the imaging system based on at least one object attribute.

[0016] One aspect of the invention relates to a computer program product comprising instructions which, when the program is executed by a computer, cause it to execute the method according to the invention according to one or more of the aspects and / or embodiments described herein.

[0017] One aspect of the invention relates to a computer-readable storage medium comprising instructions which, when executed by a computer, cause the computer to execute the method according to the invention according to one or more of the aspects and / or embodiments described herein.

[0018] One aspect of the invention relates to a training dataset for training, in particular by means of the method according to one or more of the aspects and / or embodiments described herein, at least one AI evaluation system and / or as input therefor, comprising: initial measurement data that are indicative of at least one property of an object to be scanned; initial scan parameters, which are in particular assigned to the initial measurement data; at least one object attribute that is assigned to the object to be scanned; at least one setting parameter for the imaging system; at least one setting parameter that is optimal with respect to the object to be scanned; and / or at least one preliminary, in particular estimated, setting parameter for the imaging system.

[0019] One aspect of the invention relates to an imaging system in which image data are generated, in particular by solving an inverse problem, such as a tomography system, in particular a computed tomography (CT) scanner, magnetic resonance imaging (MRI) scanner or the like, which implements the method according to one or more of the aspects and / or embodiments described herein.

[0020] One aspect of the invention relates to generating a training data set according to one or more of the aspects and / or embodiments described herein for training the at least one AI evaluation system.

[0021] One aspect of the invention relates to generating a training data set, in particular according to one or more of the aspects and / or embodiments described herein, for training at least one AI evaluation system using the method according to one or more of the aspects and / or embodiments described herein.

[0022] One aspect of the invention relates to training at least one AI evaluation system using a training data set according to one or more of the aspects and / or embodiments described herein and / or using the method according to one or more of the aspects and / or embodiments described herein.

[0023] One aspect of the invention relates to a computer system comprising a computer evaluation system and / or a computer feedback system, in particular a trained computer evaluation system and / or a trained computer feedback system that was trained using the method according to one or more of the aspects and / or embodiments described herein; the training data set according to one or more of the aspects and / or embodiments described herein; the training set generated according to one or more of the aspects and / or embodiments described herein; and / or the training according to one or more of the aspects and / or embodiments described herein.

[0024] The invention relates to a computer-implemented method for automatically determining setting parameters, in particular scan parameters, for an imaging system, in which image data are generated in particular by solving an inverse problem, such as a tomography system, in particular a computed tomography (CT) scanner, magnetic resonance imaging (MRI) scanner or the like.

[0025] An "imaging system" can be understood as a device, instrument, system, and / or facility designed to create, generate, construct, and / or output images and / or pictures of objects and / or (internal or external) structures, for example, to obtain information about these objects. Imaging systems can use various physical principles and / or techniques to capture, process, display, generate, and / or produce the images and / or pictures. Imaging systems can be used in medicine, science, industry, and / or other fields.

[0026] An "inverse problem" can be understood as inferring the underlying (internal or external) structure and / or properties of an object under investigation from, for example, measured input data. In other words, instead of directly knowing the object's structure and generating images from it, the attempt is made to deduce the object's structure from the obtained images, information, and / or data. An inverse problem arises, for example, in imaging techniques such as computed tomography (CT) and / or magnetic resonance imaging (MRI), where measurements are taken of signals or radiation that pass through the object under investigation, particularly at least partially. It can describe a problem such as: determining an unknown cause based on a known effect. Based on these measurements, internal or external properties can then be deduced.External structures and / or properties of the object are reconstructed and / or visualized or displayed. An imaging system can be a tomography system, e.g., a tomography method and / or a device or apparatus for carrying out the tomography method, in particular a tomograph. Examples of imaging systems to which the invention described herein can be applied and / or in which an inverse problem is solved include one or more of the following:

[0027] - Computed tomography (CT): in which radiation, especially X-rays, is sent through a patient's body and measured by sensors on the opposite side. By solving an inverse problem, computers can then generate images, such as cross-sectional images of the body's interior, which can provide information about tissue types, tumors, organs, muscles, bones, and / or other structures;

[0028] Magnetic resonance imaging (MRI): In MRI, measured magnetic resonance signals can be used to create detailed images of, for example, tissue structures in the body;

[0029] Positron emission tomography (PET): In PET, an image of the distribution of a radioactive tracer in the body can be created from the measured gamma radiation of positron decay;

[0030] Single-photon emission computed tomography (SPECT): Similar to PET, SPECT can reconstruct an image of the distribution of a radioactive tracer in the body from the measured gamma radiation data;

[0031] - Optical coherence tomography (OCT): OCT can be used to generate detailed images, such as cross-sectional images of biological tissue, from the measured interference data;

[0032] Ultrasound tomography (also sonography): In sonography, measured ultrasound signals can be directly converted into real-time images;

[0033] Electrical impedance tomography (EIT): In EIT, images of, for example, the electrical conductivity in an object under investigation can be created from the measured electrical impedance data;

[0034] Magnetic resonance spectroscopy (MRS): In MRS, information about, for example, the chemical composition and / or the molecular structures in the investigated area can be extracted from the measured magnetic resonance spectra;

[0035] X-ray radiography: This involves using X-rays to create images of the inside of the body, for example;

[0036] Fluoroscopy: can be a dynamic form of X-ray radiography, in which, for example, continuous images are generated in real time; Mammography: for example, a specialized X-ray examination for the early detection of breast cancer in women;

[0037] Infrared imaging: Uses infrared radiation to visualize, for example, temperature differences in the body;

[0038] Magnetic resonance elastography (MRE): Combines magnetic resonance imaging (MRI) with mechanical waves to quantify the stiffness of tissues such as the liver or brain;

[0039] Photoacoustic imaging: Can combine light and ultrasound techniques to produce images of tissues, for example, with high resolution and / or depth;

[0040] Multispectral imaging: This allows images to be acquired across multiple wavelength ranges of the electromagnetic spectrum in order to examine different properties of, for example, tissues; and / or other or similar properties.

[0041] A "setting parameter" can be a parameter, variable, characteristic, size, data, configuration(s), and / or value that can be configured to control, regulate, influence, and / or adjust the imaging system or the procedure performed with it. Setting parameters can, for example, affect the scanning process and / or the acquisition or image and / or image quality. There can be at least one setting parameter, or several, or a plurality of setting parameters, such as at least two or more. Setting parameters or scan parameters can be used in one or more imaging systems (as described above) and can include, for example, one or more of the following:

[0042] - a resolution or image resolution, which defines, for example, the number and / or size of pixels and / or can influence the level of detail;

[0043] - a scan time, which indicates, for example, how long the scan takes to capture the entire image or scan area;

[0044] - an exposure time, which indicates, for example, how long individual projections are recorded;

[0045] - a radiation intensity or energy of the radiation used, which determines the penetrating ability of the radiation and / or can influence the image quality;

[0046] - at least one scan position and / or at least one scan direction, which can define, for example, a view and / or perspective of the image;

[0047] Contrast and / or brightness settings, for example to adjust the representation of different structures, such as tissue types, in an image; - an echo time and / or repetition time, which can, for example, influence image contrast;

[0048] - a tube voltage, which can indicate, for example, the energy of the generated radiation, especially X-rays;

[0049] - a current intensity, which can indicate, for example, the amount of radiation produced;

[0050] Detector sensitivity, which indicates, for example, how well the imaging system's detector responds to the detected radiation and / or other signals. Higher sensitivity can help detect weak signals and improve image quality.

[0051] Scan speed, which indicates, for example, how quickly the scanner moves across the scan area or how quickly data is captured. A higher scan speed can help reduce motion artifacts; focus size, for example, the size of the beam focus, to affect, for example, the sharpness and / or detail of the image;

[0052] Image acquisition sequences, e.g. to capture different structures, such as T1 weights, T2 weights, FLAIR sequences, etc.;

[0053] A number and / or arrangement of measuring sensors, e.g. electrodes in electrical impedance tomography

[0054] - a polarization direction, e.g., in optical coherence tomography, and / or other or similar directions.

[0055] These setting parameters can, for example, influence how the imaging system collects and processes data and generates and / or displays or visualizes one or more images. At least one setting parameter for the imaging system can at least partially influence, and in particular control, the quality, especially the image quality, of an image of the object to be scanned that can be generated by the imaging system. The quality of the generated image can relate to its clarity, sharpness, distinctness, clarity, precision, exactness, and / or accuracy. It can also include aspects such as contrast, resolution, brightness, intensity, exposure, and / or noise, or the like. By adjusting, changing, and / or influencing the setting parameter, certain characteristics of the image can be controlled to achieve better image quality.For example, the setting parameter can be used to improve the sharpness of the image, reduce noise, and / or adjust the contrast in order to display the information in the image more clearly, distinctly, or in greater detail. The method according to the invention can therefore also include the step of influencing, in particular controlling, a quality, especially image quality, of an image of the object to be scanned that can be generated and / or produced by the imaging system, in particular at least partially, based on the at least one setting parameter generated for the imaging system.

[0056] The method according to the invention may include:

[0057] Determine, in particular extract, on the basis of initial measurement data indicative of at least one property of an object to be scanned, at least one object attribute that is assigned to the object to be scanned.

[0058] "Initial measurement data" can refer to values, parameters, information, and / or data that are assigned to an object, or to an object being scanned or examined, and are present initially, at the beginning, and / or as a starting point, for example, even before the scanning process of the object by the imaging system has begun. The initial measurement data can therefore be referred to as raw data or raw measurement data. The initial measurement data can correspond to minimal measurement data, which can mean that it only includes minimal information about an object being scanned, for example, just one property. A scanning process can be understood as the acquisition and / or collection of information and / or data, for example, using radiation. The initial measurement data can be stored, for example, in a memory or database and used, for example, when needed. Storage, e.g.,Following a scanning process, initial scan parameters can be assigned to or linked with the initial measurement data, and these initial scan parameters can also be stored in the memory. The initial measurement data can be acquired by at least one sensor and / or have been acquired and / or be acquirable. Alternatively or additionally, the initial measurement data can comprise at least one image of the object to be scanned. The method according to the invention can optionally also include: acquiring initial measurement data of the object to be scanned and / or acquiring at least one image of the object to be scanned, in particular by at least one sensor. Furthermore, the method can optionally include: storing, in particular saving, the initial measurement data and / or the at least one image in a memory, in particular as training data.

[0059] The initial measurement data can comprise at least two, and in particular at least partially different, images of the object to be scanned. These at least two images of the object to be scanned can be taken from different positions of at least one sensor and / or using different sensors. For example, the initial measurement data can include or show multiple views, angles, and / or perspectives of the object to be scanned, e.g., from multiple directions. The different sensors can, for example, have different characteristics or respond to different types of signals, which can lead to different information about the object.

[0060] The initial measurement data and / or at least one image of the object to be scanned may include one or more of:

[0061] Media data, e.g., one or more of especially image data (e.g., visual representations (2D, 3D, etc.), such as images, e.g., X-ray images, MRI scans, etc.), video data (e.g., in the form of moving images) and / or audio data (acoustic signals and / or sound signals, e.g., using sound waves).

[0062] Time series data, e.g., data collected over a specific period and indicating changes over time, e.g., ECG signals and / or EEG signals,

[0063] Structural data, such as 3D models and / or computer-aided design (CAD) data, which can be indicative of, for example, the shape, structure and / or geometry of an object;

[0064] Text data, e.g., information that is in textual form, such as descriptions;

[0065] Measurement data in a sequence of time and / or position data as well as one or more additional pieces of information (e.g. indices, relative assignments in time and / or space) from one or more sensors (e.g. PET / SPECT data, which are available in particular in list mode format, and include, for example, timestamps, position information and possibly further information such as time-of-flight (e.g. relative timestamps of second signals to each other)) etc.;

[0066] Point clouds (or models derived from them), i.e., datasets representing a large number of points in three-dimensional space and derived using laser scans or other surveying techniques, can be used, for example, to create 3D models and / or to survey surfaces.

[0067] Sensory data, i.e., information or data that can be acquired by means of at least one sensor, and physical properties, such as dimensions, temperature, weight, or the like; chemical properties, such as composition, concentration, or the like; and / or biological properties, such as blood glucose level, or the like, of the object under investigation; and / or other / similar data. The initial measurement data can be obtained from at least one physical interaction, such as absorption, scattering, and / or reflection of radiation, in particular electromagnetic waves and / or particles, with the object to be scanned. Radiation can refer to the propagation of energy, e.g., in the form of electromagnetic waves and / or particles, such as light.Visible light in a wavelength range of approximately 400 nm to approximately 700 nm; ultraviolet radiation, in a wavelength range of approximately 10 nm to approximately 400 nm, e.g., UV-A (400 nm to 315 nm), UV-B (315 nm to 280 nm), UV-C (280 nm to 10 nm); infrared radiation (IR), in a wavelength range of approximately 700 nm to approximately 1 mm, e.g., short-wave IR / SWIR (700 nm to 3 micrometers), medium-wave IR / MWIR (3 micrometers to 8 micrometers), long-wave IR / LWIR (8 micrometers to 15 micrometers); X-rays, in a wavelength range of approximately 0.1 nm to approximately 10 nm; gamma radiation, in a wavelength range of less than approximately 0.1 nm; Radio and / or microwaves, in a wavelength range of approximately a few millimeters to approximately a few meters, i.e. frequencies from approximately 30 kHz to several GHz; sound waves, e.g. ultrasound between approximately 20 kHz and approximately 20 MHz; particle radiation, such as alpha, beta and / or neutron radiation.The type of radiation to be used may depend on the corresponding imaging system.

[0068] "Absorption" can be understood as the absorption or capture of a portion of the radiation striking the object being scanned, and / or its penetration, particularly at least partially. Absorption can depend on the properties of the object and / or the type of radiation. For example, the absorption of X-rays by tissues of varying density can be used to generate images of the body's interior. "Scattering" (e.g., elastic / inelastic) can refer to the deflection, distribution, and / or redirection of radiation from the object being scanned, for example, in one or more different directions. The type and / or extent of scattering can provide information about the object's structure or composition. "Reflection" can describe the backscattering, bouncing, reflecting, and / or mirroring of a portion of the radiation, for example, from a surface of the object being scanned.Reflection can provide information about the surface texture and / or the optical properties of the object.

[0069] The initial measurement data can be indicative of at least one property of the object to be scanned. In other words, the initial measurement data can, in particular directly or indirectly, provide an indication of at least one property of the object, or at least one property can be included, shown, represented, and / or embedded in the initial measurement data. The at least one property can include one or more of the following:

[0070] - a geometry (e.g. a geometric shape, spatial arrangement, contours, surfaces, etc.)

[0071] - a structure (e.g., an arrangement and / or sequence of elements; internal / external structure);

[0072] - a shape (e.g., external form, shape and / or appearance of the object, such as essentially round, conical, cylindrical, wooden-cylindrical, cuboid, star-shaped, diamond-shaped, oblong, wedge-shaped, trough-shaped, pot-shaped, square, cube-shaped, spherical, pyramid-shaped, trapezoidal, pin- or bolt-shaped, or the like, or a combination thereof; and / or, for example, including incisions, tapers, widenings, holes, steps, bevels, inclinations, depressions, protrusions, grooves, projections, changes in wall thickness, textured surfaces, points, steps, notches, edges, wedges, angles, notches, noses, ridges, bridges, grooves, inclinations, recesses, or the like, or a combination thereof);

[0073] - a dimension (e.g. length, width, height, (inner / outer) diameter, thickness, etc.);

[0074] - a material distribution (e.g. density, homogeneous / heterogeneous distribution and / or arrangement of materials, components and / or substances);

[0075] - a material composition (e.g., the type and / or composition of the materials from which the object is made, including the chemical elements or compounds that determine its physical and / or chemical properties; and / or

[0076] - other properties such as hardness, elasticity, porosity or the like;

[0077] - a surface finish, e.g. transparent, semi-transparent or opaque, a color, opacity, and / or a texture, such as a pattern, roughness or smoothness, reflectivity, etc.;

[0078] - a combination of the above and / or other / similar ones.

[0079] The object to be scanned initially exhibits essentially indeterminate, undefined, and / or unknown properties. In other words, before or at the start of the scanning process, the object may possess several of the properties described herein that are not fully known, for example, to a user of the imaging system, but are nevertheless included in the initial measurement data. An "object attribute" can be a property, characteristic piece of information, and / or feature of the object that can actually be assigned, associated, and / or ascribed to the object being scanned. The object attribute can be derived, deduced, inferred, extracted, taken, constructed, generated, produced, obtained, and / or obtained from the initial measurement data, for example, by processing the initial measurement data.The method according to the invention can therefore comprise: processing the initial measurement data to determine, in particular extract, at least one object attribute that is assigned to the object to be scanned. The object attribute can, for example, correspond to at least one property of the object to be scanned, or to one or more other properties of the object to be scanned. The at least one object attribute can, in particular, be generated by reconstruction from the initial measurement data.

[0080] The method according to the invention may include:

[0081] Generating at least one setting parameter for the imaging system based on at least one object attribute.

[0082] At least one setting parameter for the imaging system can be a preliminary, in particular (e.g., temporary) estimated, setting parameter for the imaging system. The preliminary setting parameter can be an approximation of an optimal setting parameter with respect to the object to be scanned. In other words, the generated preliminary setting parameter cannot yet correspond to the final or most optimal setting parameter. It can, for example, correspond to an approximate value that serves to achieve acceptable image quality and / or efficient execution of the scanning process until, for example, more accurate and / or optimal setting parameters are determined. The generation, as described below, can therefore be done, for example, iteratively or through training and / or feedback. After, for example, a preliminary setting parameter has been estimated and the scanning process has been carried out, the obtained data (stored, e.g., in a file) can be used to determine the final, optimal setting parameter.The data (e.g., in the memory mentioned herein) can be analyzed to, for example, evaluate the image quality. Based on this analysis, the preliminary setting parameters can be adjusted and / or reused to, for example, perform an improved scan or obtain improved image quality. This process can be repeated until the desired results are achieved and / or until the parameters are optimized. The inventive method described herein, or the steps of the inventive method, among other things, reduce the need for skilled personnel, because the automation of parameter estimation significantly reduces the need for highly qualified specialists and / or their in-depth expertise. This helps to overcome the challenges of the skilled labor shortage and reduce the associated high costs.Furthermore, this can accelerate the imaging process, as an automated system enables faster parameter estimation. This saves valuable time resources and / or reduces throughput times. The inventive method described herein, or the steps of the inventive method, also allow for adaptation to individual scenarios, as the system offers the possibility of adapting and / or calibrating the automated parameter estimation to individual, as yet unknown scenarios. This enables flexible adaptation to various application areas and improves the adaptability of the method and / or the system as such. Furthermore, the combination of automated parameter estimation and a reduced need for expert knowledge can lower the barrier to using the imaging method.This makes the technology accessible for a wider range of applications and allows it to be used effectively in various fields.

[0083] Determining, in particular extracting, at least one object attribute and / or generating at least one setting parameter for the imaging system can involve processing using at least one artificial intelligence (AI) system, in particular an AI evaluation system. The AI ​​system can be understood as a computer system capable of performing tasks that would normally require human intelligence. Such a system can, for example, recognize patterns, make decisions, solve problems, be trained, and / or learn from experience without, for example, needing to be explicitly programmed. The AI ​​system can, for example, be capable of essentially automatically determining at least one object attribute based on or from the initial measurement data. The AI ​​system can, for example,be able to generate at least one setting parameter essentially automatically based on at least one object attribute.

[0084] The AI ​​system, or AI evaluation system, can comprise at least one neural network. The AI ​​system can include a deep learning system, such as a convolutional neural network (CNN), recurrent neural networks (RNNs), transformer models, or similar architectures. A neural network can be a model, particularly an algorithmic one, inspired by or modeled on the workings of the human brain. The neural network can comprise a plurality and / or collection of interconnected neurons, particularly artificial ones, that process and transmit information and / or recognize and / or learn patterns or relationships. By training on large datasets, a neural network can learn complex tasks such as image recognition, segmentation, natural language processing, and / or decision-making.The AI ​​system can be pre-trained, for example, to determine object attributes and / or to generate setting parameters, or for other purposes. The neural network can be, for example, a CNN that can use specific layers, such as convolutional layers and / or pooling layers, to extract features and / or properties from images and to recognize patterns and / or relationships.

[0085] The initial measurement data, which are indicative of at least one property of an object to be scanned; the initial scan parameters, which are specifically assigned to the initial measurement data; at least one object attribute assigned to the object to be scanned; at least one setting parameter for the imaging system; at least one setting parameter that is optimal with respect to the object to be scanned; and / or at least one preliminary, in particular estimated, setting parameter for the imaging system can be provided to the AI ​​system, in particular the AI ​​evaluation system or the at least one neural network, as input, or also for training purposes. It is also conceivable that the AI ​​system is pre-trained on one or more of these values ​​or data.

[0086] The AI ​​system or AI evaluation system can comprise at least one first neural network and at least one second neural network. As mentioned above, the AI ​​system or AI evaluation system can comprise at least one neural network; that is, the first and second neural networks can also be combined into a single neural network, or only a single neural network can be used. The use of two networks is therefore purely optional. The at least one first neural network and the at least one second neural network can be arranged sequentially, in particular, connected one after the other, e.g., in series, in a series, and / or consecutively. For example, the first and second neural networks can be arranged such that an output of a preceding network (e.g., the first or second network) serves as input to a subsequent network (e.g., the corresponding second or first network).Input) is provided. For example, at least one neural network can process data and generate an output that is then used as input for the second neural network. The sequential arrangement allows the processing steps to be carried out step by step, with the output of one network serving as input for the next. It should be noted that a third and / or fourth network can also be used. For example, the initial measurement data and / or the associated initial scan parameters can be provided as input to one of the first or second neural networks, and, in particular, the resulting output, at least one object attribute, can be provided as input to another of the first or second neural networks.

[0087] At least one object attribute and / or at least one setting parameter for the imaging system can also be provided as input to a feedback system, in particular an artificial intelligence (AI) feedback system. The AI ​​feedback system can be designed in principle like the AI ​​evaluation system, e.g., comprising at least one (e.g., third) neural network, or a third and fourth neural network, the latter being connected in parallel or operating in parallel. This parallel arrangement can make it possible, for example, to determine an error, which can then be propagated back, e.g., to the AI ​​evaluation system. The AI ​​feedback system can be used, in particular, to configure the AI ​​evaluation system, for example, for feedback, and / or for calibration, adjustment, and / or training, especially retraining the AI ​​evaluation system and / or for backpropagation.The AI ​​feedback system allows the AI ​​evaluation system to be adapted, modified, calibrated, adjusted, and / or improved. For example, the AI ​​feedback system can be provided by the customer, e.g., after delivery. It can enable continuous improvement and / or adaptation of the AI ​​system based on customer requirements and / or feedback. It should be noted that adjusting the AI ​​evaluation system, especially retraining, can alternatively or additionally be done through human feedback. Therefore, the provision / use of an AI feedback system is purely optional. For example, a user can provide feedback to the AI ​​feedback system, or the user can provide feedback to the AI ​​evaluation system, particularly directly.

[0088] BRIEF DESCRIPTION OF THE DRAWINGS Exemplary embodiments of the invention are now described with reference to the accompanying drawings. To ensure a detailed understanding of the features of the present disclosure mentioned above, a more detailed description of the disclosure, which has been briefly summarized above, can be obtained by referring to exemplary embodiments. The accompanying drawings relate to embodiments of the disclosure and are described below:

[0089] Fig. 1 shows an embodiment of a method according to the invention using at least one artificial intelligence (AI) evaluation system, in particular using at least one neural network;

[0090] Fig. 2 shows a further embodiment of a method according to the invention, which can be combined in particular with the previous embodiment, using at least two neural networks, which are in particular connected in series; and

[0091] Fig. 3 shows an embodiment of a method according to the invention using at least one artificial intelligence, AI, evaluation system and an AI feedback system.

[0092] DESCRIPTION OF PREFERRED EXECUTION FORMS

[0093] The invention will now be explained in more detail with reference to embodiments shown in the drawings, wherein in all drawings essentially functionally identical elements have the same reference numerals.

[0094] The drawings are schematic and not to scale. Some elements in the drawings may have exaggerated dimensions to emphasize aspects of the present disclosure and / or for greater clarity of presentation. For the sake of simplicity, identical reference numerals are used to denote identical elements common to all drawings. It is intended that elements and features of one embodiment may be advantageously incorporated into other embodiments without further mention. In general, only the differences between individual embodiments are described. Each embodiment serves to illustrate the disclosure and should not be understood as limiting the disclosure. Furthermore, features shown or described as part of one embodiment may be used in conjunction with other embodiments to create yet another embodiment.The description is intended to include such modifications and variations.

[0095] Fig. 1 shows an embodiment of a method 100 according to the invention using at least one artificial intelligence, AI, evaluation system 110, in particular using at least one neural network 112.

[0096] As can be seen in Fig. 1, an imaging system 102, in which image data are generated in particular by solving an inverse problem, such as a tomography system 102, in particular a computed tomography (CT) scanner, magnetic resonance imaging (MRI) scanner or the like (as described herein, for example), is provided.

[0097] As indicated by arrow 103, the imaging system 102 can be connected to a memory 104. Experiences of the imaging system 102 can be stored 103 in the memory 104 or may already be contained therein. The method 100 can therefore optionally include: storing 103, in particular saving 103, the initial measurement data 101 and / or at least one initial scan parameter 101 associated with it in a memory 104, in particular as training data and / or for training purposes. The saving 103 can, for example, take place at regular intervals, e.g., once per scan, once per object, and / or once per unit of time, e.g., per hour, day, etc. It should be noted that the saving step 103 is to be considered optional for the present invention, since data from a memory 104 that is already stored therein can also be used. The memory 104 can, for example,initial measurement data 101 and / or at least one initial scan parameter 101 assigned to it of one or more objects, which may be indicative of at least one property of an object to be scanned (not shown).

[0098] The initial measurement data 101 can be acquired by at least one sensor and / or have been acquired and / or can be acquired. Alternatively or additionally, the initial measurement data 101 can comprise at least one image of the object to be scanned. The initial measurement data 101 can comprise at least two, in particular at least partially different, images of the object to be scanned. The at least two images of the object to be scanned can be acquired from different positions of at least one sensor and / or by means of different sensors. For example, the initial measurement data 101 can comprise or show several views, angles, and / or perspectives of the object to be scanned, e.g., from several directions. The different sensors can, for example, have different properties or react to different types of signals, which, for example,which can lead to different information about the object.

[0099] The initial measurement data 101 may include one or more of:

[0100] Media data, e.g., one or more of especially image data (e.g., visual representations (2D, 3D, etc.), such as images, e.g., X-ray images, MRI scans, etc.), video data (e.g., in the form of moving images) and / or audio data (acoustic signals and / or sound signals, e.g., using sound waves).

[0101] Time series data, e.g., data collected over a specific period and indicating changes over time, e.g., ECG signals and / or EEG signals,

[0102] Structural data, such as 3D models and / or computer-aided design (CAD) data, which can be indicative of, for example, the shape, structure and / or geometry of an object;

[0103] Text data, e.g., information that is in textual form, such as descriptions, etc.;

[0104] Measurement data in a sequence of time and / or position data as well as one or more additional pieces of information (e.g. indices, relative assignments in time and / or space) from one or more sensors (e.g. PET / SPECT data, which are available in particular in list mode format, and include, for example, timestamps, position information and possibly further information such as time-of-flight (e.g. relative timestamps of second signals to each other)) etc.;

[0105] Point clouds (or models derived from them), i.e., datasets representing a large number of points in three-dimensional space and derived using laser scans or other surveying techniques, can be used, for example, to create 3D models and / or to survey surfaces, and / or sensor data.

[0106] As can be seen further in Fig. 1, the initial measurement data 101 can be provided 105 to a computer evaluation system 110, e.g., as input 105. The method 100 according to the invention can comprise: Determining 106, in particular extracting 106, especially by means of the computer evaluation system 110 or the at least one neural network 112, based on initial measurement data 101 that are indicative of at least one property of an object to be scanned, of at least one object attribute 107 that is assigned to the object to be scanned. As can be seen in Fig. 1, the at least one object attribute 107 (output) can be provided (again) as input to the computer evaluation system 110 or the at least one neural network 112. The method 100 can further comprise: Generating 108, especially by means of the computer evaluation system 110 or the at least one neural network 112.the at least one neural network 112, of at least one setting parameter 109 for the imaging system 102 based on the at least one object attribute 107. The at least one generated setting parameter 109 can then be used to influence, control, regulate and / or adjust the imaging system 102 and / or, as further described, be stored 103, as shown in Fig. 1.

[0107] The at least one setting parameter 109 for the imaging system 102 can be a preliminary, in particular (e.g., temporarily) estimated, setting parameter 109 for the imaging system 102. The preliminary setting parameter 109 can be an approximation of an optimal setting parameter 109 with respect to the object to be scanned. In other words, the generated preliminary setting parameter 109 cannot yet correspond to the final or most optimal setting parameter. It can, for example, correspond to an approximate value that serves to achieve acceptable image quality and / or efficient execution of the scanning process until, for example, more accurate and / or optimal setting parameters are determined. The generation, as described below, can therefore be carried out, for example, iteratively or through training and / or feedback. After, for example, a preliminary setting parameter 109 has been determined, orOnce the estimated image quality has been determined and the scanning process has been performed, the obtained data 101 / 109 (stored 103, e.g., in memory 104) can be analyzed to evaluate, for example, the image quality. Based on this analysis, the preliminary setting parameters 109 can be adjusted and / or reused to, for example, perform an improved scan or obtain improved image quality. This process can be repeated until the desired results are achieved and / or until the parameters are optimized. As shown in Fig. 1, at least one setting parameter 109 for the imaging system 102 can also be provided as input to the AI ​​evaluation system 110, e.g., for training purposes.

[0108] The method 100 described herein, or the steps of the method according to the invention, among other things, reduce the need for skilled personnel, because the automation of parameter estimation significantly reduces the need for highly qualified specialists and / or their in-depth specialist knowledge. This helps to overcome the challenges of the shortage of skilled personnel and to reduce the associated high costs. Furthermore, the imaging process can be accelerated, since an automated system enables faster parameter estimation. This saves valuable time resources and / or reduces throughput times. The method according to the invention described herein, orThe steps of the method according to the invention also allow for adaptation to individual scenarios, as the system offers the possibility of adapting and / or calibrating the automated parameter estimation to individual, as yet unknown scenarios. This enables flexible adaptation to various application areas and improves the adaptability of the method and / or the system as such. Furthermore, the combination of automated parameter estimation and a reduced need for expert knowledge lowers the barrier to using the imaging method. This makes the technology accessible to a wider range of applications and allows it to be used effectively in various fields.

[0109] Determining 106, in particular extracting 106, the at least one object attribute 107 and / or generating 108 the at least one setting parameter 109 for the imaging system 102 can include processing by means of at least one artificial intelligence (AI) system 110, in particular an AI evaluation system 110. The AI ​​system 110 can, for example, be able to determine 106 the at least one object attribute 107 essentially automatically based on or from the initial measurement data 101. The AI ​​system 110 can, for example, be able to generate 106 the at least one setting parameter 109 essentially automatically based on the at least one object attribute 107.

[0110] As shown in Fig. 1, for the method 100 according to the invention, at least one, in particular a single, AI system 110 or AI evaluation system 110 or at least one, in particular a single, neural network 112 is sufficient. The AI ​​evaluation system 110 or the network 112 can be pre-trained, for example to determine 106 object attributes 107 and / or to generate 108 setting parameters 109.

[0111] As can be further seen in Fig. 1, the initial measurement data 101, which are indicative of at least one property of an object to be scanned; the initial scan parameters, which are in particular assigned to the initial measurement data; at least one object attribute 107, which is assigned to the object to be scanned; at least one setting parameter 109 for the imaging system 102; at least one setting parameter that is optimal with respect to the object to be scanned; and / or at least one preliminary, in particular estimated, setting parameter 109 for the imaging system 102, can be provided to the AI ​​system 110, in particular to the AI ​​evaluation system 110 or to the at least one neural network 112, as input (cf. 105 or 113), or for training purposes. It is also conceivable that the AI ​​system is already pre-trained on these values ​​or data.

[0112] Fig. 2 shows a further embodiment of a method 100 according to the invention, which can be combined in particular with the previous embodiment shown in Fig. 1. In contrast to Fig. 1, the embodiment shown in Fig. 2 uses at least two neural networks instead of one neural network 112, in particular a first neural network 112A and at least a second neural network 112B. This is indicated in Fig. 2 by the dashed line, which refers to the AI ​​evaluation system 110 from Fig. 1. In this respect, the explanations, examples, and / or features that apply to Fig. 1 can equally apply to the embodiment shown in Fig. 2, and vice versa. Therefore, for the sake of clarity, these explanations, examples, and / or features are not repeated here.

[0113] As shown in Fig. 2, the at least one first neural network 112A and the at least one second neural network 112B can be arranged sequentially, in particular connected one after the other, e.g., in series, in a series, and / or consecutively. For example, the first 112A and second 112B neural networks can be arranged such that an output (cf. 107) of a preceding network (e.g., the first network 112A) is provided as input (cf. 107) to a subsequent network (e.g., the corresponding second network 112B). For example, the at least one first neural network 112A can process data and generate an output 107, which is then used as input for the second neural network 112B. The sequential arrangement makes it possible for the processing steps to be carried out step by step, with the output of one network serving as input for the next network. For example, as shown in Fig.To recognize that the first neural network 112 receives the initial measurement data 101 and / or the initial scan parameters associated with it as input, and in particular the at least one object attribute 107 output therefrom can be provided as input to the second neural network 112, which in turn can generate the at least one setting parameter 109 for the imaging system 108. The inventive method 100 described herein, or the steps of the inventive method, among other things, reduce the need for skilled workers, because the automation of parameter estimation significantly reduces the need for highly qualified specialists and / or their in-depth specialist knowledge. This helps to overcome the challenges of the shortage of skilled workers and to reduce the associated high costs.Furthermore, this can accelerate the imaging process, as an automated system enables faster parameter estimation. This saves valuable time resources and / or reduces throughput times. The inventive method described herein, or the steps of the inventive method, also allow for adaptation to individual scenarios, as the system offers the possibility of adapting and / or calibrating the automated parameter estimation to individual, as yet unknown scenarios. This enables flexible adaptation to various application areas and improves the adaptability of the method and / or the system as such. Furthermore, the combination of automated parameter estimation and a reduced need for expert knowledge can lower the barrier to using the imaging method.This makes the technology accessible for a wider range of applications and allows it to be used effectively in various fields.

[0114] Determining 106, in particular extracting 106, the at least one object attribute 107 can include processing by means of the at least one first neural network 112A and / or generating 108 the at least one setting parameter 109 for the imaging system 102 can include processing by means of the at least one second neural network 112B. The at least one first neural network 112A can, for example, be able to determine 106 the at least one object attribute 107 essentially automatically based on or from the initial measurement data 101. The at least one second neural network 112B can, for example, be able to generate 106 the at least one setting parameter 109 essentially automatically based on the at least one object attribute 107.

[0115] Fig. 3 shows an embodiment of a method according to the invention using at least one artificial intelligence (AI) evaluation system and an AI feedback system 114. In contrast to Figs. 1 and 2, Fig. 3 additionally includes an AI feedback system 114, which can, for example, be used by the customer or be located at a customer's site. In this respect, the explanations, examples, and / or features that apply to Figs. 1 and / or 2 can equally apply to the embodiment of Fig. 3, and vice versa. Therefore, for the sake of clarity, these explanations, examples, and / or features are not repeated here.

[0116] In Fig. 3, the AI ​​evaluation system 110 comprises at least two neural networks 112A and 112B, respectively. However, it is also possible that the AI ​​evaluation system 110, e.g., based on Fig. 1, comprises only at least one neural network 112. Any input-feedback steps (cf. 113 in Fig. 1) are not shown in Fig. 3 for the sake of clarity. The same applies analogously to the AI ​​feedback system 114, which comprises two neural networks 114A and 114B, respectively. The networks 112, 112A-B and 114, 114A-B can be essentially identical in design and / or structure, i.e., they represent or comprise the same neural network.

[0117] The AI ​​feedback system 114 can be used, in particular, to configure the AI ​​evaluation system 110, for example, for feedback, calibration, adjustment, and / or training, especially retraining the AI ​​evaluation system and / or for feedback propagation. The AI ​​feedback system 114 thus allows the AI ​​evaluation system 110 to be adapted, modified, calibrated, adjusted, and / or improved. For example, the AI ​​feedback system 114 can be provided by the customer, e.g., after subsequent delivery. It can enable continuous improvement and / or adaptation of the AI ​​evaluation system 110 based on customer requirements and / or feedback.

[0118] As can be seen in Fig. 3, the at least one object attribute 107 can also be provided as input to the feedback system 114, in particular to the artificial intelligence (AI) feedback system 114. For example, the object attribute 107 can be split after output from the first neural network 112A of the AI ​​evaluation system 110, and provided once to the (e.g., third) neural network 114A as input and once to the (e.g., fourth) neural network 114B as input, wherein the third / fourth neural networks, as can be seen in Fig. 3, can be connected in parallel or operate in parallel. Alternatively or additionally, the at least one setting parameter 109 for the imaging system 102 can be split, for example, after output from the second neural network 112B of the AI ​​evaluation system 110, and provided once to the (e.g.,The third neural network 114A and the fourth neural network 114B can each receive the object attribute 107 and at least one setting parameter 109 for the imaging system 102 as input. The third neural network 114A can generate a first value or reward 115A from this, and the fourth neural network 114B can generate a second value or reward 115A, which can be evaluated, for example, compared with each other, to determine an error 116. This error 116 can be propagated back to the AI ​​evaluation system 110, as indicated by the dashed arrow in Fig. 3. Through backpropagation, the Kl evaluation system 110 can therefore be adapted, modified, calibrated, adjusted and / or improved.

[0119] It should be noted that the configuration of the AI ​​evaluation system 110, in particular retraining, can alternatively or additionally also be carried out by human feedback 119 / 120. For example, an output 118 of the AI ​​feedback system 114 can be provided to a user 119. The user 119 can then, in particular directly, give feedback 120 to the AI ​​feedback system 114, in particular manual feedback, e.g., regarding 114, 114A and / or 114B.

[0120] Although the foregoing is directed to embodiments of the disclosure, other and further embodiments of the disclosure can be developed without deviation from the basic scope thereof, the scope being determined by the following claims.

[0121] Reference symbol list

[0122] 100 procedures

[0123] 101 Initial measurement data

[0124] 102 Imaging System

[0125] 103 Save

[0126] 104 storage spaces

[0127] 105 Input

[0128] 106 Determine / Extract

[0129] 107 Object attribute / Output

[0130] 108 Produce

[0131] 109 Settings parameters / Output

[0132] 110 Kl-System

[0133] 112 Neural network

[0134] 112A,B First / second neural network 113 Input

[0135] 114 KL feedback system

[0136] 114A,B Third / fourth neural network

[0137] 115A, B Output value / Reward 116 Error

[0138] 117 Backpropagation

[0139] 118 Output

[0140] 119 users

[0141] 120 Manual Feedback

Claims

PATENT CLAIMS 1. Computer-implemented method for automatically determining setting parameters, in particular scan parameters, for an imaging system, in which image data are generated in particular by solving an inverse problem, such as a tomography system, in particular a computed tomography (CT) scanner, magnetic resonance imaging (MRI) scanner or the like, comprising: Determine, in particular extract, based on initial measurement data that are indicative of at least one property of an object to be scanned, at least one object attribute that is assigned to the object to be scanned, and Generating at least one setting parameter for the imaging system based on at least one object attribute.

2. Computer-implemented method according to claim 1, wherein the object to be scanned initially has essentially indeterminate, undefined and / or unknown properties.

3. Computer-implemented method according to one of the preceding claims, wherein the at least one object attribute is generated by reconstruction from the initial measurement data.

4. Computer-implemented method according to one of the preceding claims, wherein the at least one property comprises a geometry, structure, shape, dimension, material distribution and / or material composition.

5. Computer-implemented method according to one of the preceding claims, wherein the initial measurement data are acquired by at least one sensor and / or comprise at least one recording of the object to be scanned.

6. Computer-implemented method according to one of the preceding claims, wherein the initial measurement data are obtained from at least one physical interaction, such as absorption, scattering and / or reflection of radiation, in particular electromagnetic waves and / or particles, with the object to be scanned.

7. Computer-implemented method according to one of the preceding claims, wherein the initial measurement data and / or the at least one recording of the object to be scanned comprise one or more of: Media data, in particular image data, video data and / or audio data, time series data, Structural data, such as 3D models and / or CAD data, Text data, Point clouds, and / or sensor data.

8. Computer-implemented method according to one of the preceding claims, wherein the initial measurement data comprise at least two, in particular at least partially different, images of the object to be scanned, wherein in particular the at least two images of the object to be scanned were taken from different recording positions of a sensor and / or by means of different sensors.

9. Computer-implemented method according to one of the preceding claims, wherein the at least one setting parameter for the imaging system is a preliminary, in particular estimated, setting parameter for the imaging system, wherein in particular the preliminary setting parameter is an approximation of a setting parameter that is optimal with respect to the object to be scanned.

10. Computer-implemented method according to one of the preceding claims, wherein the at least one setting parameter for the imaging system at least partially influences, in particular controls, the quality, in particular the image quality, of an image of the object to be scanned that can be generated from the imaging system.

11. Computer-implemented method according to one of the preceding claims, wherein determining, in particular extracting, the at least one object attribute and / or generating the at least one setting parameter for the imaging system comprises processing by means of at least one artificial intelligence (AI) evaluation system, wherein the initial measurement data, in particular the initial scan parameters and / or the at least one object attribute, are provided to the at least one AI evaluation system as input.

12. Computer-implemented method according to claim 11, wherein the AI ​​evaluation system comprises at least one neural network.

13. Computer-implemented method according to claim 11, wherein the AI ​​evaluation system comprises at least a first neural network and at least a second neural network.

14. Computer-implemented method according to claim 13, wherein the at least one first neural network and the at least one second neural network are arranged sequentially, in particular are connected one after the other, wherein in particular an output of a preceding network is provided as input to a subsequent network.

15. Computer-implemented method according to one of claims 13-14, wherein the initial measurement data and / or the initial scan parameters assigned thereto are provided as input to one of the first or second neural networks and, in particular, the at least one object attribute output therefrom is provided as input to another of the first or second neural networks.

16. Computer-implemented method according to one of the preceding claims, wherein the at least one object attribute and / or the at least one setting parameter for the imaging system are provided as input to a feedback system, in particular an artificial intelligence (AI) feedback system, which serves in particular to set the AI ​​evaluation system.

17. Imaging system in which image data is generated in particular by solving an inverse problem, such as a tomography system, in particular a computed tomography (CT) scanner, magnetic resonance imaging (MRI) scanner or the like, which implements the method according to any one of claims 1-16.

18. Computer program product comprising instructions which, when the program is executed by a computer, cause it to execute the method according to any one of claims 1-16.

19. Computer-readable storage medium comprising instructions which, when executed by a computer, cause it to execute the method according to any one of claims 1-16.

20. Training data set for training, in particular by means of the method according to one of claims 1-16, at least one AI evaluation system and / or as input therefor, comprising: Initial measurement data that are indicative of at least one property of an object to be scanned; initial scan parameters, which are specifically assigned to the initial measurement data; at least one object attribute assigned to the object to be scanned; at least one setting parameter for the imaging system; at least one optimal setting with respect to the object to be scanned Setting parameters; and / or at least one preliminary, in particular estimated, setting parameter for the imaging system.

21. Generating a training data set according to claim 20 for training the at least one Kl evaluation system.

22. Generating a training data set, in particular according to claim 20, for training at least one AI evaluation system using the method according to one of claims 1-16.

23. Training at least one AI evaluation system using a training data set according to claim 20 and / or using the method according to any one of claims 1-16.

24. AI system comprising an AI evaluation system and / or an AI feedback system, in particular a trained AI evaluation system and / or a trained AI feedback system that was trained by the method according to any one of claims 1-16; the training data set according to claim 20; the training set generated according to any one of claims 21-22; and / or the training according to claim 23.

Images