Anonymization of medical image data

Abstract

A computer-implemented method for providing classified image features, comprising: - Receiving (REC-BD) medical image data (BD), wherein the medical image data (BD) is acquired using at least one medical imaging device; - Identifying multiple image features in the medical image data (BD) and classifying the multiple image features into patient-specific image features (pBM) and patient-nonspecific image features (uBM) by applying a trained image feature identification and classification function (TF-IDCL-BM) to input data, wherein the multiple image features include geometric and / or anatomical image features and / or statistical image information; wherein the input data is based on the medical image data (BD); and wherein at least one parameter of the trained image feature identification and classification function (TF-IDCL-BM) is based on a comparison of training identification parameters (TIDP-pTBM).TIDP-uTBM) with comparison identification parameters (VIDP) and a comparison of training diagnostic parameters (TDIAGP-pTBM, TDIAGP-uTBM) with comparison diagnostic parameters (VDIAGP), - Classifying patient-specific image features (pBM) into phenotypically expressed patient-specific image features (paBM) and non-phenotypically expressed patient-specific image features (naBM) by applying a trained function for classifying patient-specific image features (TF-CL-pBM) to input data, wherein the input data is based on the patient-specific image features (pBM),wherein at least one parameter of the trained function for classifying patient-specific image features (TF-CL-pBM) is based on a comparison of phenotypically pronounced patient-specific training image features (paTBM) with phenotypically pronounced patient-specific comparison image features (paVBM) and a comparison of non-phenotypically pronounced patient-specific training image features (naTBM) with non-phenotypically pronounced patient-specific comparison image features (naVBM),- providing (PROV-pBM) the classified patient-specific image features (paBM, naBM) and the patient-non-specific image features (uBM).

Description

The invention relates to a computer-implemented method for providing classified image features, a computer-implemented method for providing synthetic medical image data, a computer-implemented method for providing a trained function for identifying and classifying image features, a computer-implemented method for providing a trained function for classifying patient-specific image features, a computer-implemented method for providing a trained function for generating synthetic medical image data, a computer-implemented method for providing a further trained function for generating synthetic medical image data, a computer-implemented method for providing a further trained function for classifying patient-specific image features, and a provisioning unit for providing classified image features.a delivery unit for providing synthetic medical image data, a medical imaging device, a training unit, a computer program product, and a computer-readable storage medium. Patient data and / or measurement data, especially medical image data, which may be processed and / or released by hospitals and / or medical practices, should be securely and as completely anonymized as possible. Previously, it was often sufficient to remove data, particularly text data and / or metadata, describing the patient, such as name and date of birth, from the measurement data. The measurement data may, for example, be in DICOM format, with the text data and / or metadata describing the patient often contained in the DICOM header. With increasing measurement accuracy, modern 3D imaging techniques, and improved reconstruction algorithms, it has become possible to reconstruct a patient's phenotypic characteristics based on measurement data. For example, magnetic resonance imaging (MRI) scans and / or X-ray images can be used to reconstruct a skull, face, and / or other phenotypic features of the patient. These characteristics, which are suitable for identifying a patient, can be considered biometric features. The features described above may be obvious to a person skilled in the art, so attempts are often made to prevent such reconstruction. With the ever-increasing need for medical image data, particularly clinical data, for training machine learning (ML) algorithms, the removal of biometric features from medical image data has become significantly more important. Known ML algorithms can extract many additional biometric features from medical image data that are not immediately apparent to a specialist. The invention is therefore based on the objective of enabling secure anonymization of medical image data while maintaining diagnosability. The problem is solved according to the invention by the respective subject matter of the independent claims. Advantageous embodiments with expedient further developments are the subject matter of the dependent claims. The inventive solution to the problem is described below with regard to methods and devices for providing classified image features and / or synthetic medical image data, as well as with regard to methods and devices for providing trained functions. Features, advantages, and alternative embodiments of data structures and / or functions in methods and devices for providing classified image features and / or synthetic medical image data can be transferred to analogous data structures and / or functions in methods and devices for providing trained functions. Analogous data structures can be characterized, in particular, by the use of the prefix "training".Furthermore, the trained functions used in methods and devices for providing classified image features and / or synthetic medical image data may have been adapted and / or provided, in particular, by methods and devices for providing trained functions. The invention relates, in a first aspect, to a computer-implemented method for providing classified image features. In a first step, medical image data is received. By applying a trained function for identifying and classifying image features to the input data, several image features in the medical image data are identified and classified into patient-specific and patient-nonspecific image features. The input data is based on the medical image data. Furthermore, at least one parameter of the trained function for identifying and classifying image features is based on a comparison of training identification parameters with comparison identification parameters and a comparison of training diagnostic parameters with comparison diagnostic parameters. The classified image features are then provided in a further step. Receiving medical image data can include, in particular, capturing and / or reading data from a computer-readable data storage device and / or receiving data from a data storage unit, such as a database. Furthermore, the medical image data can be provided by a provisioning unit of a medical imaging device for the purpose of receiving the medical image data. Medical image data can, for example, consist of two-dimensional and / or three-dimensional images, comprising multiple image points, in particular pixels and / or voxels. Furthermore, the medical image data can depict at least one area of ​​an object under investigation. This object can, for example, be a human and / or animal patient. In addition, the medical image data can depict a temporal progression, such as a change in the area of ​​the object under investigation. Finally, the medical image data can be acquired by one or more, in particular different, medical imaging devices.In this context, one or at least one of the several medical imaging devices may be designed as an X-ray machine and / or C-arm X-ray machine and / or magnetic resonance imaging (MRI) system and / or computed tomography (CT) system and / or sonography system and / or positron emission tomography (PET) system. Furthermore, medical image data can advantageously include metadata. This metadata can contain information about acquisition parameters and / or operating parameters of the medical imaging device used to acquire the medical image data. By applying the trained function for identifying and classifying image features to the received medical image data, multiple image features can be identified. Furthermore, the identified image features can be classified into patient-specific and patient-nonspecific features. The multiple image features in the medical image data can include, for example, geometric and / or anatomical image features. Furthermore, the multiple image features can include image information, particularly statistical information, that depicts a distribution of image values ​​within the medical image data, such as a histogram. Identifying the multiple image features by applying the trained function can, in particular, involve localizing and / or segmenting the multiple image features within the medical image data. The classification of the identified multiple image features can further include a differentiation and / or grouping of the multiple image features into patient-specific and patient-nonspecific image features. Patient-specific image features can include, in particular, those image features that enable a clear and unambiguous representation of the object under investigation. Furthermore, patient-specific image features can include, for example, biometric and / or diagnostic image features that allow for a conclusion and / or a clear and unambiguous identification of the object under investigation. In addition, patient-nonspecific image features can include, for example, diagnostic and / or other anatomical and / or geometric image features that do not allow for a conclusion and / or identification of the object under investigation.For example, contrast, particularly a ratio of image values, can be classified as a patient-nonspecific image feature. Furthermore, a spatial contrast profile, such as an edge along an anatomical structure, can be identified as an anatomical image feature and classified as a patient-specific image feature. In particular, patient-specific image features can include all biometric image features identified in the medical image data. Biometric image features can include, for example, spatial position information, spatial arrangement information, and / or shape information of at least one anatomical image feature.For example, a skull shape and / or a tumor surface and / or an organ surface and / or a spatial arrangement of several anatomical image features relative to each other can be classified as a patient-specific image feature, in particular as a biometric image feature. The advantages and / or properties of a trained function described below essentially correspond to the advantages of the proposed trained function for identifying and classifying image features. Features, advantages, or alternative embodiments mentioned here can also be applied to the other proposed trained functions, and vice versa. The trained function can advantageously be trained using a machine learning method. In particular, the trained function can be a neural network, especially a convolutional neural network (CNN) or a network comprising a convolutional layer. The trained function maps input data to output data. The output data may, in particular, still depend on one or more parameters of the trained function. These one or more parameters of the trained function can be determined and / or adjusted through training. Determining and / or adjusting the one or more parameters of the trained function can, in particular, be based on a pair of training input data and corresponding training output data, where the trained function is applied to the training input data to generate training mapping data. Specifically, determining and / or adjusting can be based on a comparison of the training mapping data and the training output data. In general, a trainable function, i.e., a function with one or more parameters that have not yet been adjusted, is also referred to as a trained function. Other terms for a trained function include trained mapping rule, mapping rule with trained parameters, function with trained parameters, artificial intelligence-based algorithm, and machine learning algorithm. An example of a trained function is an artificial neural network, where the edge weights of the artificial neural network correspond to the parameters of the trained function. The term "neural network" can also be used instead of "neural network." In particular, a trained function can also be a deep artificial neural network. Another example of a trained function is a support vector machine; furthermore, other machine learning algorithms can also be used as trained functions. The trained function can be trained, in particular, by backpropagation. First, training mapping data can be determined by applying the trained function to training input data. Then, a deviation between the training mapping data and the training output data can be determined by applying an error function to both. Furthermore, at least one parameter, in particular a weight, of the trained function, especially of the neural network, can be iteratively adjusted based on a gradient of the error function with respect to the at least one parameter of the trained function. This advantageously minimizes the deviation between the training mapping data and the training output data during the training of the trained function. Advantageously, the trained function, particularly the neural network, has an input layer and an output layer. The input layer can be configured to receive input data. Furthermore, the output layer can be configured to provide mapping data. Both the input layer and / or the output layer can each comprise multiple channels, particularly neurons. Preferably, at least one parameter of the trained function for identifying and classifying image features can be based on a comparison of training identification parameters with comparison identification parameters and a comparison of training diagnostic parameters with comparison diagnostic parameters. The training identification parameters, the training diagnostic parameters, the comparison identification parameters, and / or the comparison diagnostic parameters can be determined as part of a proposed computer-implemented method for providing a trained function for identifying and classifying image features, which will be explained later in this description. Furthermore, providing the classified image features may include, in particular, storing them on a computer-readable storage medium and / or displaying them on a display unit and / or transmitting them to a delivery unit. This enables a particularly robust and reliable identification and classification of image features in medical image data. In a further advantageous embodiment of the proposed computer-implemented method for providing classified image features, the patient-specific image features can be classified into phenotypically expressed patient-specific image features and non-phenotypically expressed patient-specific image features by applying a trained function for classifying patient-specific image features to input data. The input data can be based on the patient-specific image features.Furthermore, at least one parameter of the trained function for classifying patient-specific image features can be based on a comparison of phenotypically pronounced patient-specific training image features with phenotypically pronounced patient-specific comparison image features, and a comparison of non-phenotypically pronounced patient-specific training image features with non-phenotypically pronounced patient-specific comparison image features. The classified patient-specific image features can also be provided. The advantages and / or properties of a trained function described above essentially correspond to the advantages of the proposed trained function for classifying patient-specific image features. Features, advantages, or alternative embodiments mentioned herein can be transferred to the proposed trained function for classifying patient-specific image features, and vice versa. Advantageously, the classification of patient-specific image features can include a distinction and / or grouping of these features into phenotypically expressed and non-phenotypically expressed patient-specific image features. Phenotypically expressed patient-specific image features can include, in particular, all patient-specific image features that allow for the identification, especially unambiguous identification, and / or inference of the object of study by comparing the patient-specific image feature with another image feature that can be detected by external observation of the object. Furthermore, non-phenotypically expressed patient-specific image features can include, in particular, all patient-specific image features that cannot be detected by external observation of the object of study.The phenotypically expressed patient-specific image features can, for example, include information about at least part of a face and / or body shape of the subject. Furthermore, the non-phenotypically expressed patient-specific image features can, for example, include shape information about an internal organ of the subject. Preferably, at least one parameter of the trained function for classifying patient-specific image features can be based on a comparison of phenotypically pronounced patient-specific training image features with phenotypically pronounced patient-specific comparison image features and a comparison of non-phenotypically pronounced patient-specific training image features with non-phenotypically pronounced patient-specific comparison image features.The phenotypically pronounced patient-specific training image features, the phenotypically pronounced patient-specific comparison image features, the non-phenotypically pronounced patient-specific training image features and / or the non-phenotypically pronounced patient-specific comparison image features can be determined as part of a proposed computer-implemented procedure for providing a trained function for classifying patient-specific image features, which will be explained further in the description. Furthermore, providing the classified patient-specific image features may include, in particular, storing them on a computer-readable storage medium and / or displaying them on a display unit and / or transmitting them to a delivery unit. This allows the classification of patient-specific image features to be advantageously extended to include phenotypically expressed patient-specific image features, particularly those detectable through external observation of the subject. Furthermore, non-phenotypically expressed image features, especially patient-specific non-phenotypically expressed diagnostically relevant image features, can be classified with particular reliability. In a second aspect, the invention relates to a computer-implemented method for providing synthetic medical image data. In a first step, medical image data is received. In a second step, classified image features are received by applying an embodiment of the proposed computer-implemented method for providing classified image features to the medical image data. Furthermore, in a third step, synthetic medical image data is generated by applying a trained function for generating synthetic medical image data to input data. The input data is based on patient-specific image features. At least one parameter of the trained function for generating synthetic medical image data is based on a comparison of synthetic medical training image data with synthetic medical reference image data.In a further step, the synthetic medical image data will be provided. Receiving medical image data can include, in particular, capturing and / or reading data from a computer-readable data storage device and / or receiving data from a data storage unit, such as a database. Furthermore, the medical image data can be provided by a provisioning unit of the medical imaging device for the purpose of receiving the medical image data. Similarly, the classified image features, which are provided by applying an embodiment of the proposed method for providing classified image features, can be received. Receiving the classified image features can include acquiring and / or reading a computer-readable data storage device and / or receiving them from a data storage unit, such as a database. The received classified image features can preferably include patient-specific and patient-nonspecific image features. Furthermore, the received patient-specific image features can be further classified into phenotypically expressed patient-specific image features and non-phenotypically expressed patient-specific image features. The advantages and / or properties of a trained function described above essentially correspond to the advantages of the proposed trained function for generating synthetic medical image data. Features, advantages, or alternative embodiments mentioned herein can be transferred to the proposed trained function for generating synthetic medical image data, and vice versa. By applying the trained function for generating synthetic medical image data to the received patient-specific image features, the synthetic medical image data can be generated. At least one parameter of the trained function for generating synthetic medical image data can be based on a comparison of synthetic medical training image data with synthetic medical reference image data. The synthetic medical training image data and the synthetic medical reference image data can be determined as part of a proposed computer-implemented method for providing a trained function for generating synthetic medical image data, which will be explained later in this description. Advantageously, the synthetic medical image data exhibits all patient-specific image features. The synthetic medical image data advantageously includes an image of at least a portion of the examination area of ​​the patient. Furthermore, the synthetic medical image data can advantageously correspond to the received medical image data in its image properties, such as contrast, dimensionality, and / or image geometry. Additionally, the synthetic medical image data can be generated based on at least one acquisition parameter of the medical imaging device used to acquire the medical image data or of another medical imaging device. Furthermore, the provision of the synthetic medical image data may include, in particular, storage on a computer-readable storage medium and / or display on a display unit and / or transmission to a delivery unit. This can advantageously enable an improved assessment of the phenotypic expression of the patient-specific image features contained in the synthetic medical image data. In a third aspect, the invention relates to a further computer-implemented method for providing synthetic medical image data. In a first step, medical image data is received. Furthermore, in a second step, classified image features are received by applying an embodiment of the proposed computer-implemented method for providing classified image features to the medical image data. In a third step, the synthetic medical image data is generated by applying a further trained function for generating synthetic image data to input data. The input data is based on patient-nonspecific image features and / or non-phenotypically expressed patient-specific image features.Furthermore, at least one parameter of the additional trained function for generating synthetic medical image data is based on a comparison of synthetic medical training image data with synthetic medical reference image data. In a further step, the synthetic medical image data is provided. Receiving medical image data can include, in particular, capturing and / or reading data from a computer-readable data storage device and / or receiving data from a data storage unit, such as a database. Furthermore, the medical image data can be provided by a provisioning unit of the medical imaging device for the purpose of receiving the medical image data. Similarly, the classified image features, which are provided by applying an embodiment of the proposed method for providing classified image features, can be received. Receiving the classified image features can include acquiring and / or reading a computer-readable data storage device and / or receiving them from a data storage unit, such as a database. The received classified image features can preferably include patient-specific and patient-nonspecific image features. Furthermore, the received patient-specific image features can be further classified into phenotypically expressed patient-specific image features and non-phenotypically expressed patient-specific image features. The advantages and / or properties of a trained function described above essentially correspond to the advantages of the proposed additional trained function for generating synthetic medical image data. Features, advantages, or alternative embodiments mentioned herein can be transferred to the proposed additional trained function for generating synthetic medical image data, and vice versa. By applying the further trained function for generating synthetic medical image data to the received patient-nonspecific image features and / or the non-phenotypically expressed patient-specific image features, the synthetic medical image data can be generated. At least one parameter of the further trained function for generating synthetic medical image data can be based on a comparison of synthetic medical training image data with synthetic medical reference image data. The synthetic medical training image data and the synthetic medical reference image data can be determined as part of a proposed computer-implemented procedure for providing the further trained function for generating synthetic medical image data, which will be explained later in this description. Advantageously, the synthetic medical image data exhibit all non-phenotypically expressed patient-specific and / or patient-nonspecific image features. The synthetic medical image data advantageously include an image of at least a portion of the examination area of ​​the subject. Advantageously, the synthetic medical image data can correspond to the received medical image data in its image properties, for example, in contrast, dimensionality, and / or image geometry. Furthermore, the synthetic medical image data can be generated based on at least one acquisition parameter of the medical imaging device used to acquire the medical image data or of another medical imaging device. Furthermore, the provision of the synthetic medical image data may include, in particular, storage on a computer-readable storage medium and / or display on a display unit and / or transmission to a delivery unit. This allows for the advantageous provision of particularly secure anonymized synthetic medical image data, which can be used as input data for further image processing algorithms and / or for training neural networks. In a further advantageous embodiment of the proposed computer-implemented method for providing classified image features, synthetic medical image data can be received by applying a proposed computer-implemented method for providing synthetic medical image data to the medical image data. In a further step, the patient-specific image features can be classified into phenotypically expressed patient-specific image features and non-phenotypically expressed patient-specific image features by applying another trained function for classifying patient-specific image features to input data. The input data can be based on the patient-specific image features and the synthetic medical image data.Furthermore, at least one parameter of the additional trained function for classifying patient-specific image features can be based on a comparison of phenotypically pronounced patient-specific training image features with phenotypically pronounced patient-specific comparison image features, and a comparison of non-phenotypically pronounced patient-specific training image features with non-phenotypically pronounced patient-specific comparison image features. In a further step, the classified patient-specific image features can be provided. Receiving the synthetic medical image data, which is provided by applying an embodiment of the proposed method for providing synthetic medical image data, can in particular include acquiring and / or reading a computer-readable data storage device and / or receiving data from a data storage unit, for example, a database. Furthermore, the medical image data can be provided by a provisioning unit of the medical imaging device for receiving the medical image data. The advantages and / or properties of a trained function described above essentially correspond to the advantages of the proposed additional trained function for classifying patient-specific image features. Features, advantages, or alternative embodiments mentioned herein can be transferred to the proposed additional trained function for classifying patient-specific image features, and vice versa. Advantageously, the classification of patient-specific image features can include, by applying the further trained function for classifying patient-specific image features to the input data, a differentiation and / or grouping of the patient-specific image features into phenotypically pronounced patient-specific image features and non-phenotypically pronounced patient-specific image features. Because the input data of the further trained function for classifying patient-specific image features is based on the patient-specific image features and the synthetic image data, the classification of the patient-specific image features can advantageously be made particularly precise and / or take into account a characteristic in the synthetic medical image data. Preferably, at least one parameter of the further trained function for classifying patient-specific image features can be based on a comparison of phenotypically pronounced patient-specific training image features with phenotypically pronounced patient-specific comparison image features and a comparison of non-phenotypically pronounced patient-specific training image features with non-phenotypically pronounced patient-specific comparison image features.The phenotypically pronounced patient-specific training image features, the phenotypically pronounced patient-specific comparison image features, the non-phenotypically pronounced patient-specific training image features and / or the non-phenotypically pronounced patient-specific comparison image features can be determined as part of a proposed computer-implemented procedure for providing a further trained function for classifying patient-specific image features, which will be explained further in the description. Furthermore, providing the classified patient-specific image features may include, in particular, storing them on a computer-readable storage medium and / or displaying them on a display unit and / or transmitting them to a delivery unit. This enables a particularly reliable and secure classification of patient-specific image features according to their phenotypic expression in the synthetic medical image data. In this way, the phenotypic expression of patient-specific image features can be advantageously considered, especially for other medical imaging modalities. In a fourth aspect, the invention relates to a computer-implemented method for providing a trained function for identifying and classifying image features. In a first step, medical training image data from several examination subjects are received. By applying the trained function for identifying and classifying image features to the input data, several training image features are identified in the medical training image data in a second step, and these multiple training image features are classified into patient-specific training image features and patient-nonspecific training image features. The input data is based on the medical training image data. Furthermore, in a third step, training identification parameters and training diagnostic parameters are determined based on the classified training image features.In this process, one training identification parameter and one training diagnostic parameter are determined for each of the classified training image features and / or for a combination of classified training image features. Furthermore, in a fourth step, one comparison identification parameter and one comparison diagnostic parameter are received for each of the objects under investigation. Each comparison identification parameter contains identification information for one of the objects under investigation. Each comparison diagnostic parameter also contains diagnostic information for one of the objects under investigation.In a fifth step, at least one parameter of the trained function for identifying and classifying image features is adjusted based on a comparison between the training identification parameters and the comparison identification parameters, and between the training diagnostic parameters and the comparison diagnostic parameters. Furthermore, in a sixth step, the trained function for identifying and classifying image features is made available. Receiving medical training image data from multiple subjects can, in particular, involve capturing and / or reading data from a computer-readable data storage device and / or receiving data from a data storage unit, such as a database. Furthermore, the medical training image data can be provided by a provisioning unit of at least one medical imaging device. Advantageously, the medical training image data depicts multiple, particularly different, subjects. Additionally, the medical training image data can depict multiple, particularly different, examination areas of the respective subjects. Advantageously, the medical training image data can be acquired by one and / or more, particularly different, medical imaging devices.The medical training image data may be acquired from several medical imaging devices using different imaging modalities and / or imaging procedures. The medical training image data can, in particular, exhibit all the properties of the medical image data that have been described in relation to the computer-implemented procedure for providing classified image features, and vice versa. Specifically, the medical training image data can be medical image data. Furthermore, the medical training image data can be simulated. By applying the trained function for identification and classification to the input data, which is based on the medical training image data, the multiple training image features in the medical training image data can advantageously be identified. Furthermore, the multiple training image features identified in this way can be classified into patient-specific training image features and patient-non-specific training image features. For each of the classified training image features and / or combinations thereof, a training identification parameter and a training diagnostic parameter can be determined. Advantageously, the training identification parameters can be determined by applying an identification function, such as a biometric and / or anatomical identification function, to the classified training image features. Advantageously, each training identification parameter can include identification information, such as a biometric parameter, suitable for identifying one of the multiple study objects. Furthermore, the training diagnostic parameters can be determined, for example, by determining a deviation of the classified training image features from an anatomy atlas and / or based on artificial intelligence.Advantageously, each of the training diagnostic parameters can include diagnostic information on the respective classified training feature and / or combination of classified training features. This diagnostic information can, for example, include probability information and / or severity information for a disease pattern and / or an anatomical deviation from a healthy anatomy. Furthermore, the training identification parameters and the training diagnostic parameters can be determined semi-automatically, for example by annotating the classified training image features. In particular, annotated classified training image features can be received. Receiving the one comparison identification parameter and the one comparison diagnostic parameter for each of the objects under investigation may in particular include capturing and / or reading from a computer-readable data storage device and / or receiving from a data storage unit, for example a database. The comparative identification parameters can advantageously include identification information for each of the objects under investigation. For example, the identification information can include biometric information and / or an image, particularly a photograph, of the respective object under investigation. Furthermore, the comparative diagnostic parameters can advantageously include diagnostic information for each of the objects under investigation. This diagnostic information can include, for example, probability information and / or information on the severity of a disease in the respective object under investigation and / or information on an anatomical deviation of the respective object under investigation from a healthy anatomy. Furthermore, at least one parameter of the trained function for identifying and classifying image features can be adjusted based on a comparison between the training identification parameters and the comparison identification parameters, and a comparison between the training diagnostic parameters and the comparison diagnostic parameters. In particular, each of the training identification parameters can be compared with each of the comparison identification parameters. Furthermore, each of the training diagnostic parameters can be compared with each of the comparison diagnostic parameters. The comparison between the training identification parameters and the comparison identification parameters, and / or the comparison between the training diagnostic parameters and the comparison diagnostic parameters, can advantageously be based on a pattern recognition algorithm. Since each of the training identification parameters and each of the training diagnostic parameters corresponds to one of the classified training image features and / or to a combination of classified training image features, comparing the training identification parameters with the comparison identification parameters and the training diagnostic parameters with the comparison diagnostic parameters advantageously allows for the exclusion of those classified training image features that do not enable the identification of one of the objects under investigation and / or diagnostic support. This can advantageously improve the identification of training image features by applying the trained function for identifying and classifying image features to the input data. Furthermore, the classification of training image features into patient-specific and patient-nonspecific training image features can be advantageously improved by applying the trained function to identify and classify image features, in particular by comparing the training identification parameters with the comparison identification parameters. Providing the trained function for identifying and classifying image features may include, in particular, storing it on a computer-readable storage medium and / or transferring it to a delivery unit. Advantageously, the proposed method for providing a trained function for identifying and classifying image features can provide a trained function for identifying and classifying image features that can be used in the computer-implemented method for providing classified image features. According to a further advantageous embodiment of the computer-implemented method for providing a trained function for classifying patient-specific image features, at least one training acquisition parameter can be determined based on the classified training image features. Furthermore, at least one comparison acquisition parameter can be received for the medical training image data of multiple subjects. Advantageously, the comparison acquisition parameter can include information on an operating parameter of the medical imaging device used to acquire the medical training image data and / or information on the acquisition geometry of the medical training image data.In this context, at least one parameter of the trained function for identifying and classifying image features can additionally be based on a comparison of at least one training acquisition parameter with at least one comparison acquisition parameter. This advantageously allows the training image features to be classified as patient-nonspecific training image features, which are caused by an acquisition parameter that is unique, particularly within the medical training image data of multiple subjects. In a fifth aspect, the invention relates to a computer-implemented method for providing a trained function for classifying patient-specific image features. In a first step, medical training image data from several examination subjects are received. In a second step, classified training image features are received by applying a proposed computer-implemented method for providing classified image features to the medical training image data. The classified image features are provided as the classified training image features, and the patient-specific image features are provided as patient-specific training image features.In a third step, the patient-specific training image features are classified into phenotypically pronounced comparison image features and non-phenotypically pronounced comparison image features by applying an identification function, particularly a biometric one, to the patient-specific training image features. Furthermore, in a fourth step, the patient-specific training image features are classified into phenotypically pronounced patient-specific training image features and non-phenotypically pronounced training image features by applying the trained function for classifying patient-specific image features to input data. The input data is based on the patient-specific training image features. Furthermore, in a fifth step, at least one parameter of the trained function for classifying patient-specific image features is adjusted based on a comparison of the phenotypically expressed patient-specific training image features with the phenotypically expressed patient-specific comparison image features and a comparison of the non-phenotypically expressed patient-specific training image features with the non-phenotypically expressed patient-specific comparison image features. In a sixth step, the trained function for classifying patient-specific image features is made available. The medical training image data can, in particular, exhibit all the properties of the medical training image data that have been described in relation to the computer-implemented procedure for providing a trained function for identifying and classifying image features, and vice versa. Specifically, the medical training image data can be medical image data. Receiving medical training image data from multiple subjects can include, in particular, capturing and / or reading data from a computer-readable data storage device and / or receiving data from a data storage unit, such as a database. Furthermore, the medical training image data can be provided by a provisioning unit of at least one medical imaging device for receiving the medical training image data. Additionally, the medical training image data can be simulated. Similarly, the classified training image features, which are provided by applying an embodiment of the proposed method for providing classified image features, can be received. Receiving the classified training image features can include acquiring and / or reading from a computer-readable data storage device and / or receiving from a data storage unit, such as a database. The classified training image features can, in particular, exhibit all the properties of the classified image features described in relation to the computer-implemented method for providing classified image features, and vice versa. Specifically, the classified training image features can be classified image features. The received classified training image features can preferably include both patient-specific and patient-nonspecific training image features.Advantageously, the classified image features are provided as classified training image features. Patient-specific image features can also be advantageously provided as patient-specific training image features. Advantageously, patient-specific training image features can be classified into phenotypically pronounced and non-phenotypically pronounced patient-specific comparison image features by applying an identification function, particularly a biometric one, to these features. The classification of patient-specific training image features can advantageously include a differentiation and / or grouping of these features into phenotypically pronounced and non-phenotypically pronounced patient-specific comparison image features. Furthermore, the patient-specific training image features can be classified semi-automatically, for example, by annotating them. In particular, annotated patient-specific training image features can be received. In particular, by applying the identification function to the patient-specific training image features, a probability value can be determined for each patient-specific training image feature and / or a combination of patient-specific training image features, which evaluates a phenotypic expression of the respective patient-specific training image feature. In this context, a spatial positioning, such as a spatial position and / or orientation, of the respective patient-specific training image feature based on the medical training image data can be taken into account. Furthermore, tissue parameters of the tissue surrounding the respective patient-specific training image feature, such as density information, can be advantageously considered in the classification of the patient-specific training image features.In particular, the identification function can be applied to the patient-specific training image features and additionally to the medical training image data. For example, an external view of the respective object of study can be simulated by applying the identification function to the patient-specific training image features. This allows, for instance, the application of prior art pattern recognition algorithms and / or biometric identification algorithms, particularly facial recognition and / or algorithms based on artificial intelligence, to capture the respective patient-specific training image feature through the simulated external view of the respective object of study. If a patient-specific training image feature can be captured, particularly through a simulated external view of the respective object of study, the patient-specific training image feature can be classified by the identification function as a phenotypically pronounced patient-specific comparison image feature.In particular, the classification of patient-specific training image features by applying the identification function can include a comparison of the patient-specific training image features with, especially known, phenotypically expressed biometric features. If a patient-specific training image feature is identified as a phenotypically expressed biometric feature, the patient-specific training image feature can be classified as a phenotypically expressed patient-specific comparison image feature. Furthermore, a higher probability value can be assigned to a patient-specific training image feature that can be detected, or could be detected, under certain circumstances by external observation of the respective object of study than to a patient-specific training image feature that cannot be detected by external observation of the respective object of study. For example, patient-specific training image features may only be detectable by observing the object of study using a camera system, particularly within a specific wavelength range of light. Moreover, patient-specific training image features detectable by a camera system can enable the identification of the respective object of study, for example, by applying artificial intelligence to the detected patient-specific training image features.For this purpose, the identification function can assign a high probability value to the respective patient-specific training image feature, so that this patient-specific training image feature can be classified as a phenotypically pronounced comparison image feature. The classification of patient-specific training image features by applying the identification function can advantageously be based on a threshold value regarding the probability value regarding the detectability of the respective patient-specific training image feature by external observation of the respective object of investigation. Furthermore, the classification of training image features into phenotypically pronounced patient-specific training image features and non-phenotypically pronounced patient-specific training image features can be advantageously improved by applying the trained function for classifying patient-specific image features. This involves comparing the phenotypically pronounced patient-specific training image features with the phenotypically pronounced patient-specific comparison image features and comparing the non-phenotypically pronounced patient-specific training image features with the non-phenotypically pronounced patient-specific comparison image features. In particular, each of the phenotypically pronounced patient-specific training image features can be compared with each of the phenotypically pronounced and non-phenotypically pronounced comparison image features.Furthermore, each of the non-phenotypically expressed patient-specific training image features can be compared with each of the phenotypically expressed and non-phenotypically expressed comparison image features. Providing the trained function for classifying patient-specific image features may include, in particular, storing it on a computer-readable storage medium and / or transferring it to a delivery unit. Advantageously, the proposed method for providing a trained function for classifying patient-specific image features can provide a trained function for classifying patient-specific image features that can be used in the computer-implemented method for providing classified image features. In a sixth aspect, the invention relates to a computer-implemented method for providing a trained function for generating synthetic medical image data. In a first step, medical training image data from several examination subjects are received. In a second step, classified training image features are received by applying a proposed computer-implemented method for providing classified image features to the medical training image data. The classified image features are provided as the classified training image features, and the patient-specific image features are provided as patient-specific training image features. Furthermore, in a third step, synthetic medical comparison image data are generated by applying a reconstruction function to the patient-specific training image features.In a fourth step, synthetic medical training image data is generated by applying the trained function for generating synthetic medical image data to input data. This input data is based on patient-specific training image features. In a fifth step, at least one parameter of the trained function for generating synthetic medical image data is adjusted based on a comparison of the synthetic medical reference image data with the synthetic medical training image data. In a sixth step, the trained function for generating synthetic medical image data is deployed. The medical training image data can, in particular, exhibit all the properties of the medical training image data that have been described in relation to the computer-implemented procedure for providing a trained function for identifying and classifying image features, and vice versa. Specifically, the medical training image data can be medical image data. Receiving medical training image data from multiple subjects can include, in particular, capturing and / or reading data from a computer-readable data storage device and / or receiving data from a data storage unit, such as a database. Furthermore, the medical training image data can be provided by a provisioning unit of at least one medical imaging device for receiving the medical training image data. Additionally, the medical training image data can be simulated. Similarly, the classified training image features, which are provided by applying an embodiment of the proposed method for providing classified image features, can be received. Receiving the classified training image features can include acquiring and / or reading from a computer-readable data storage device and / or receiving from a data storage unit, such as a database. The classified training image features can, in particular, exhibit all the properties of the classified image features described in relation to the computer-implemented method for providing classified image features, and vice versa. Specifically, the classified training image features can be classified image features. The received classified training image features can preferably include both patient-specific and patient-nonspecific training image features.Advantageously, the classified image features are provided as classified training image features. Patient-specific image features can also be advantageously provided as patient-specific training image features. Advantageously, the synthetic medical comparison image data can be generated, in particular reconstructed, by applying the reconstruction function to the patient-specific training image features. The reconstruction function can advantageously be configured to generate the synthetic medical comparison image data based on the patient-specific training image features. Advantageously, the synthetic medical comparison image data can comprise a synthetic medical comparison image for each of the examination objects. These synthetic medical comparison images can be, for example, two-dimensional and / or three-dimensional. Advantageously, the synthetic medical comparison images can comprise a representation, in particular a two-dimensional and / or three-dimensional representation, of at least a section of the examination area of ​​the respective examination object.Furthermore, the reconstruction function can include a transformation rule, for example, for a Fourier transform and / or a Radon transform, and / or an interpolation rule and / or an extrapolation rule for reconstructing the synthetic medical reference image data. Advantageously, the reconstruction function can include a rule for reconstruction based on incomplete input data, in particular the patient-specific training image features. Furthermore, the generation of the synthetic medical reference image data by applying the reconstruction function can include interpolation and / or extrapolation and / or transformation of the patient-specific training image features, in particular based on anatomical information and / or an acquisition parameter.Furthermore, the synthetic medical comparison image data can additionally be generated based on at least one acquisition parameter of the medical imaging device used to acquire the medical training image data or another medical imaging device. Advantageously, the synthetic medical reference image data exhibit all patient-specific training image characteristics. Furthermore, the synthetic medical reference image data can advantageously match the received medical training image data in their image properties, for example, in contrast and / or dimensionality and / or image geometry. Advantageously, the synthetic medical training image data, generated by applying the trained function for generating synthetic medical image data to input data based on patient-specific training image features, can be improved by comparison with the synthetic medical reference image data. Advantageously, the synthetic medical training image data can comprise a single synthetic medical training image for each of the study subjects. These synthetic medical training images can be, for example, two-dimensional and / or three-dimensional. Furthermore, the comparison between the synthetic medical training image data and the synthetic medical reference image data can be performed pixel-by-pixel, and / or voxel-by-voxel.Furthermore, a comparison can be made between the synthetic medical training images and the synthetic medical comparison images that correspond to a common subject of investigation. Providing the trained function for generating synthetic medical image data may include, in particular, storing it on a computer-readable storage medium and / or transferring it to a delivery unit. Advantageously, the proposed method for providing a trained function for generating synthetic medical image data can provide a trained function for generating synthetic medical image data that can be used in the computer-implemented method for providing synthetic medical image data. The invention relates in a seventh aspect to a computer-implemented method for providing a further trained function for generating synthetic medical image data. In a first step, medical training image data from several examination subjects are received. Furthermore, in a second step, classified training image features are received by applying a proposed computer-implemented method for providing classified image features to the medical training image data. The classified image features are provided as the classified training image features, the patient-nonspecific image features as patient-nonspecific training image features, and / or the non-phenotypically expressed patient-specific image features as non-phenotypically expressed training image features.Furthermore, in a third step, synthetic medical comparison image data are generated by applying a reconstruction function to the non-phenotypically pronounced patient-specific training image features and / or the patient-unspecific training image features. Furthermore, in a fourth step, synthetic medical training image data is generated by applying the additional trained function for generating synthetic medical image data to input data. This input data is based on the non-phenotypically expressed patient-specific training image features and / or the patient-nonspecific training image features. In a fifth step, at least one parameter of the additional trained function for generating synthetic medical image data is adjusted based on a comparison of the synthetic medical reference image data with the synthetic medical training image data. In a sixth step, the additional trained function for generating synthetic medical image data is deployed. The medical training image data may, in particular, exhibit all the properties of the medical training image data that have been described in relation to the computer-implemented procedure for providing a trained function for identifying and classifying image features, and vice versa. Receiving medical training image data from multiple subjects can include, in particular, capturing and / or reading data from a computer-readable data storage device and / or receiving data from a data storage unit, such as a database. Furthermore, the medical training image data can be provided by a provisioning unit of at least one medical imaging device for receiving the medical training image data. Additionally, the medical training image data can be simulated. Similarly, the classified training image features, which are provided by applying an embodiment of the proposed method for providing classified image features, can be received. Receiving the classified training image features can include acquiring and / or reading from a computer-readable data storage device and / or receiving from a data storage unit, such as a database. In particular, the classified training image features can exhibit all the properties of the classified image features described in relation to the computer-implemented method for providing classified image features, and vice versa.Furthermore, the classified training image features can exhibit all properties of the classified patient-specific image features that have been described in relation to the computer-implemented procedure for providing classified patient-specific image features, and vice versa. Furthermore, the classified training image features can be classified image features. The received classified training image features can preferably include patient-specific and patient-nonspecific training image features. Advantageously, the classified image features are provided as classified training image features. The patient-specific image features can advantageously be provided as patient-specific training image features. Furthermore, the patient-specific image features can be classified into phenotypically expressed patient-specific image features and non-phenotypically expressed patient-specific image features. The non-phenotypically expressed patient-specific image features can be provided as non-phenotypically expressed patient-specific training image features. Advantageously, the synthetic medical comparison image data can be generated, and in particular reconstructed, by applying the reconstruction function to the non-phenotypically patient-specific training image features. The reconstruction function can advantageously be configured to generate the synthetic medical comparison image data based on the patient-nonspecific training image features and / or the non-phenotypically pronounced training image features. Advantageously, the synthetic medical comparison image data can comprise a single synthetic medical comparison image for each of the subjects under investigation. These synthetic medical comparison images can be, for example, two-dimensional and / or three-dimensional.The synthetic medical comparison images can advantageously include an image, in particular a two-dimensional and / or three-dimensional image, of at least a section of the examination area of ​​the respective object under investigation. Furthermore, the reconstruction function can include a transformation rule, for example, for a Fourier transform and / or a Radon transform, and / or an interpolation rule and / or an extrapolation rule for reconstructing the synthetic medical reference image data. Advantageously, the reconstruction function can include a rule for reconstruction based on incomplete input data, in particular the patient-nonspecific training image features and / or the non-phenotypically expressed patient-specific training image features.Furthermore, the generation of synthetic medical reference image data by applying the reconstruction function can include interpolation and / or extrapolation and / or transformation of patient-nonspecific training image features and / or non-phenotypically expressed patient-specific training image features, particularly based on anatomical information and / or an acquisition parameter. Additionally, the synthetic medical reference image data can be generated based on at least one acquisition parameter of the medical imaging device used to acquire the medical training image data or of another medical imaging device. Advantageously, the synthetic medical reference image data exhibit all patient-nonspecific training image features and / or all patient-specific training image features that are not phenotypically pronounced. Furthermore, the synthetic medical reference image data can advantageously correspond to the received medical training image data in their image properties, for example, in contrast and / or dimensionality and / or image geometry. Advantageously, the synthetic medical training image data, generated by applying the further trained function for generating synthetic medical image data to input data based on patient-nonspecific training image features and / or non-phenotypically expressed patient-specific training image features, can be improved by comparison with the synthetic medical reference image data. Advantageously, the synthetic medical training image data can comprise a single synthetic medical training image for each of the study subjects. These synthetic medical training images can be, for example, two-dimensional and / or three-dimensional. Furthermore, the comparison between the synthetic medical training image data and the synthetic medical reference image data can be performed pixel-wise, in particular pixel-wise and / or voxel-wise.Furthermore, a comparison can be made between the synthetic medical training images and the synthetic medical comparison images that correspond to a common subject of investigation. Providing the further trained function for generating synthetic medical image data may in particular include storing it on a computer-readable storage medium and / or transferring it to a delivery unit. Advantageously, the proposed method for providing another trained function for generating synthetic medical image data can provide another trained function for generating synthetic medical image data that can be used in the computer-implemented method for providing synthetic medical image data. The invention relates in an eighth aspect to a computer-implemented method for providing a further trained function for classifying patient-specific image features. In a first step, medical training image data of several examination subjects are received. Furthermore, synthetic medical training image data are received by applying a proposed computer-implemented method for providing synthetic medical image data to the medical training image data. The synthetic medical image data are provided as the synthetic medical training image data, and the patient-specific image features are provided as patient-specific training image features.In a second step, the patient-specific training image features are classified into phenotypically pronounced patient-specific comparison image features and non-phenotypically pronounced patient-specific comparison image features by applying a further, in particular biometric, identification function to the patient-specific training image features and the synthetic medical training image data. In a third step, the patient-specific training image features are classified into phenotypically pronounced patient-specific training image features and non-phenotypically pronounced training image features by applying the further trained function for classifying patient-specific image features to input data. The input data is based on the patient-specific training image features and the synthetic medical training image data.In a fourth step, at least one parameter of the further trained function for classifying patient-specific image features is adjusted based on a comparison of the phenotypically expressed patient-specific training image features with the phenotypically expressed patient-specific comparison image features and a comparison of the non-phenotypically expressed patient-specific training image features with the non-phenotypically expressed patient-specific comparison image features. In a fifth step, the further trained function for classifying patient-specific image features is made available. The medical training image data can, in particular, exhibit all the properties of the medical training image data that have been described in relation to the computer-implemented procedure for providing a trained function for identifying and classifying image features, and vice versa. Specifically, the medical training image data can be medical image data. Receiving medical training image data from multiple subjects can include, in particular, capturing and / or reading data from a computer-readable data storage device and / or receiving data from a data storage unit, such as a database. Furthermore, the medical training image data can be provided by a provisioning unit of at least one medical imaging device for receiving the medical training image data. Additionally, the medical training image data can be simulated. Similarly, the synthetic medical training image data, which is provided to the medical training image data by applying an embodiment of the proposed method for providing synthetic medical image data, can be received. Receiving the synthetic medical training image data can include acquiring and / or reading from a computer-readable data storage device and / or receiving from a data storage unit, such as a database. In particular, the synthetic medical training image data can have all the properties of the synthetic medical image data that were defined in relation to the computer-implemented method for providing synthetic medical image data, and vice versa. Specifically, the synthetic medical training image data can be synthetic medical image data.Advantageously, the synthetic medical training image data for each of the multiple examination objects can include a single training image, in particular a two-dimensional and / or three-dimensional one. Furthermore, the classified image features obtained when applying the proposed method for providing synthetic medical image data can be provided as classified training image features. These classified training image features can preferably include patient-specific and patient-nonspecific training image features. Furthermore, the patient-specific image features can advantageously be provided as patient-specific training image features. Advantageously, the patient-specific training image features can be classified into phenotypically pronounced patient-specific comparison image features and non-phenotypically pronounced patient-specific comparison image features by applying further identification functions, particularly biometric ones, to the patient-specific training image features and the synthetic medical training image data. Advantageously, the classification of the patient-specific training image features can include a differentiation and / or grouping of the patient-specific training image features into phenotypically pronounced patient-specific comparison image features and non-phenotypically pronounced patient-specific comparison image features. Furthermore, the patient-specific training image features can be classified semi-automatically, for example, by annotating the patient-specific training image features.In particular, annotated patient-specific training image features can be received. Advantageously, the phenotypic expression of these patient-specific training image features in the synthetic medical training image data can be evaluated, especially through external observation. In particular, by applying the further identification function to the patient-specific training image features and the synthetic medical training image data, a probability value can be determined for each of the patient-specific training image features and / or for a combination of patient-specific training image features. This value evaluates the phenotypic expression of the respective patient-specific training image feature in the synthetic medical training image data. In this context, a spatial positioning, such as a spatial position and / or orientation, of the respective patient-specific training image feature in the synthetic medical training image data can be taken into account.Furthermore, tissue parameters of the tissue surrounding the respective patient-specific training image feature, for example density information, can be advantageously taken into account when classifying the patient-specific training image features. For example, an external view of the respective object of study can be simulated by applying the further identification function to the patient-specific training image features and the synthetic medical training image data. This allows, for example, the application of prior art pattern recognition algorithms and / or biometric identification algorithms, particularly for facial recognition and / or based on artificial intelligence, to capture the respective patient-specific training image feature through the simulated external view of the respective object of study in the training image of the synthetic medical training image data.If a patient-specific training image feature can be captured in the synthetic medical training image data, particularly through a simulated external view of the respective subject, the patient-specific training image feature can be classified as a phenotypically pronounced patient-specific comparison image feature by the further identification function. The classification of patient-specific training image features by applying the further identification function can advantageously be based on a threshold value regarding the probability value regarding the detectability of the respective patient-specific training image feature by external observation of the synthetic medical training image data. Furthermore, the classification of training image features into phenotypically pronounced patient-specific training image features and non-phenotypically pronounced patient-specific training image features can be advantageously improved by applying the additional trained function for classifying patient-specific image features. This involves comparing the phenotypically pronounced patient-specific training image features with the phenotypically pronounced patient-specific comparison image features and comparing the non-phenotypically pronounced patient-specific training image features with the non-phenotypically pronounced patient-specific comparison image features. In particular, each of the phenotypically pronounced patient-specific training image features can be compared with each of the phenotypically pronounced and non-phenotypically pronounced comparison image features.Furthermore, each of the non-phenotypically expressed patient-specific training image features can be compared with each of the phenotypically expressed and non-phenotypically expressed comparison image features. Providing the further trained function for classifying patient-specific image features may in particular include saving it to a computer-readable storage medium and / or transferring it to a delivery unit. Advantageously, the proposed method for providing an additional trained function for classifying patient-specific image features can provide an additional trained function for classifying patient-specific image features, which can be used in the computer-implemented method for providing classified image features. In a ninth aspect, the invention relates to a provisioning unit for providing classified image features, comprising a processing unit and an interface. The interface is configured to receive medical image data. Furthermore, the processing unit is configured to identify multiple image features in the medical image data and to classify these multiple image features into patient-specific and patient-nonspecific image features by applying a trained function for identifying and classifying image features to input data. The input data is based on the medical image data. At least one parameter of the trained function for identifying and classifying image features is based on a comparison of training identification parameters with comparison identification parameters and a comparison of training diagnostic parameters with comparison diagnostic parameters.Furthermore, the interface for providing the classified image features is designed. Such a provisioning unit for providing classified image features can, in particular, be configured to execute the previously described methods according to the invention for providing classified image features and their aspects. The provisioning unit for providing classified image features is configured to execute these methods and their aspects by providing the interface and the processing unit to perform the corresponding process steps. The advantages of the proposed provisioning unit for providing classified image features are essentially the same as the advantages of the proposed computer-implemented method for providing classified image features. Features, advantages, or alternative embodiments mentioned here can also be transferred to the other claimed subject matter and vice versa. In a tenth aspect, the invention relates to a provisioning unit for providing synthetic medical image data, comprising a computing unit and an interface. The interface is configured for receiving medical image data. Furthermore, the interface is configured for receiving classified image features by applying a proposed computer-implemented method for providing classified image features to the medical image data. Finally, the computing unit is configured for generating the synthetic medical image data by applying a trained function for generating synthetic medical image data to input data. The input data is based on patient-specific image features.Furthermore, at least one parameter of the trained function for generating synthetic medical image data is based on a comparison of synthetic medical training image data with synthetic medical reference image data. In addition, the interface for providing the synthetic medical image data is implemented. Such a provisioning unit for providing synthetic medical image data can, in particular, be configured to execute the previously described methods according to the invention for providing synthetic medical image data and their aspects. The provisioning unit for providing synthetic medical image data is configured to execute these methods and their aspects by providing the interface and the processing unit to perform the corresponding process steps. The advantages of the proposed delivery unit for providing synthetic medical image data are essentially the same as the advantages of the proposed computer-implemented method for providing synthetic medical image data. Features, advantages, or alternative embodiments mentioned here can likewise be transferred to the other claimed subject matter and vice versa. In an eleventh aspect, the invention relates to a medical imaging device comprising a proposed provisioning unit for providing classified image features and / or synthetic medical image data. The medical imaging device, and in particular the proposed provisioning unit, is configured to execute a proposed computer-implemented method for providing classified image features and / or synthetic medical image data. The medical imaging device can be, for example, a medical X-ray device, in particular a C-arm X-ray device, and / or a computed tomography (CT) scanner, and / or a magnetic resonance imaging (MRI) scanner, and / or an ultrasound scanner. Furthermore, the medical imaging device can be configured to acquire and / or receive and / or provide the medical image data. The medical imaging device may, in particular, include a display unit, such as a display and / or a monitor, which is configured to display information and / or graphical representations of information from the medical imaging device and / or the delivery unit and / or other components. In particular, the display unit may be configured to display a graphical representation of the medical image data and / or the classified image features and / or the synthetic medical image data. The advantages of the proposed medical imaging device essentially correspond to the advantages of the proposed computer-implemented methods for providing classified image features and / or for providing synthetic medical image data. Features, advantages, or alternative embodiments mentioned herein can likewise be transferred to the other claimed subject matter and vice versa. In a twelfth aspect, the invention relates to a training unit configured to execute the previously described computer-implemented methods according to the invention for providing a trained function for identifying and classifying image features, and / or for providing a trained function for classifying patient-specific image features, and / or for providing a trained function for generating synthetic medical image data, and / or for providing a further trained function for generating synthetic medical image data, and / or for providing a further trained function for classifying patient-specific image features, and their aspects. The training unit advantageously comprises a training interface and a training computing unit.The training unit is configured to execute these procedures and their aspects by means of a training interface and a training processing unit that perform the corresponding procedural steps. Specifically, the training interface can be configured to receive medical training image data and / or classified training image features and / or synthetic medical training image data. Furthermore, the training interface can be configured to provide the trained function. In a thirteenth aspect, the invention relates to a computer program product comprising a computer program that can be directly loaded into a memory of a provisioning unit, with program sections to execute all steps of the computer-implemented method for providing classified image features and / or for providing synthetic medical image data when the program sections are executed by the provisioning unit;and / or which is directly loadable into a training memory of a training unit, with program sections to execute all steps of the proposed procedure for providing a trained function for identifying and classifying image features and / or providing a trained function for classifying patient-specific image features and / or providing a trained function for generating synthetic medical image data and / or providing another trained function for generating synthetic medical image data and / or providing another trained function for classifying patient-specific image features and / or any aspect thereof, when the program sections are executed by the training unit. In a fourteenth aspect, the invention relates to a computer-readable storage medium on which program sections readable and executable by a provisioning unit are stored in order to execute all steps of the computer-implemented method for providing classified image features and / or for providing synthetic medical image data when the program sections are executed by the provisioning unit;and / or on which program sections readable and executable by a training unit are stored to execute all steps of the procedure for providing a trained function for identifying and classifying image features and / or providing a trained function for classifying patient-specific image features and / or providing a trained function for generating synthetic medical image data and / or providing another trained function for generating synthetic medical image data and / or providing another trained function for classifying patient-specific image features and / or any aspect thereof, when the program sections are executed by the training unit. In a fifteenth aspect, the invention relates to a computer program or computer-readable storage medium comprising a trained function for identifying and classifying image features and / or for classifying patient-specific image features and / or for generating synthetic medical image data and / or a further trained function for generating synthetic medical image data and / or for classifying patient-specific image features provided by a proposed computer-implemented method or one of its aspects. A largely software-based implementation has the advantage that existing deployment units and / or training units can be easily retrofitted via a software update to operate according to the invention. Such a computer program product may, in addition to the computer program itself, optionally include additional components such as documentation and / or other components, as well as hardware components such as hardware keys (dongles, etc.) for using the software. Exemplary embodiments of the invention are illustrated in the drawings and are described in more detail below. The same reference numerals are used for identical features in different figures. Figures 1 and 2 show schematic representations of various embodiments of a proposed computer-implemented method for providing classified image features, Figures 3 and 4 show schematic representations of various embodiments of a proposed method for generating synthetic medical image data, Figure 5 shows a schematic representation of a further embodiment of the proposed computer-implemented method for providing classified image features, and Figure 6 shows a schematic representation of an embodiment of a proposed computer-implemented method for providing a trained function for identifying and classifying image features.Fig. 7 a schematic representation of an embodiment of a proposed computer-implemented method for providing a trained function for classifying patient-specific image features, Fig. 8 a schematic representation of an embodiment of a proposed computer-implemented method for providing a trained function for generating synthetic medical image data, Fig. 9 a schematic representation of an embodiment of a proposed computer-implemented method for providing a further trained function for generating synthetic medical image data, Fig. 10 a schematic representation of an embodiment of a proposed computer-implemented method for providing a further trained function for classifying patient-specific image features, Fig. 11 a schematic representation of a proposed provisioning unit, Fig.Fig. 12 a schematic representation of a proposed training unit, Fig. 13 a schematic representation of a medical C-arm X-ray unit as an example of a proposed medical imaging device. Figure 1 schematically illustrates an embodiment of the proposed computer-implemented method for providing classified image features. In a first step, REC-BD medical image data (BD) can be received. By applying a trained function for identifying and classifying image features (TF-IDCL-BM) to input data based on the medical image data (BD), multiple image features can be identified within the medical image data (BD). These multiple image features can then be further classified into patient-specific image features (pBM) and patient-nonspecific image features (uBM).Advantageously, at least one parameter of the trained function for identifying and classifying image features TF-IDCL-BM can be based on a comparison of training identification parameters with comparison identification parameters and a comparison of training diagnostic parameters with comparison diagnostic parameters. In a further step, PROV-BM, the classified image features pBM and uBM can be provided. Figure 2 schematically illustrates another embodiment of the proposed computer-implemented method for providing classified image features. In this method, patient-specific image features (pBM) can be classified into phenotypically expressed patient-specific image features (paBM) and non-phenotypically expressed patient-specific image features (naBM) by applying a trained function for classifying patient-specific image features (TF-CL-pBM) to input data based on the patient-specific image features (pBM).At least one parameter of the trained function for classifying patient-specific image features TF-CL-pBM can be based on a comparison of phenotypically pronounced patient-specific training image features with phenotypically pronounced patient-specific comparison image features, and a comparison of non-phenotypically pronounced training image features with non-phenotypically pronounced comparison image features. In a further step, PROV-pBM can provide the classified patient-specific image features paBM and naBM. Figure 3 schematically illustrates an embodiment of a proposed computer-implemented method for generating synthetic medical image data. In a first step, REC-BD medical image data (BD) can be received. Furthermore, in a second step, REC-BM classified image features (pBM) and uBM can be received by applying a proposed computer-implemented method for providing classified image features to the medical image data (BD). In a subsequent step, the synthetic medical image data (SBD) can be generated by applying a trained function (TF-SBD) to input data. Advantageously, the input data can be based on patient-specific image features (pBM).Furthermore, at least one parameter of the trained function for generating synthetic medical image data (TF-SBD) can be based on a comparison of synthetic medical training image data with synthetic medical reference image data. In a further step, PROV-SBD, the synthetic medical image data (SBD) can be provided. Figure 4 schematically illustrates another embodiment of a proposed computer-implemented method for generating synthetic medical image data. In a first step, REC-BD medical image data (BD) can be received. Furthermore, classified image features can be received by applying a proposed computer-implemented method for providing classified image features to the medical image data (REC-BM). The received classified image features can be categorized into patient-specific image features (pBM) and patient-nonspecific image features (uBM). In addition, the patient-specific image features (pBM) can be further subdivided into phenotypically expressed patient-specific image features (paBM) and non-phenotypically expressed patient-specific image features (naBM). In a further step, synthetic medical image data (SBD) can be generated by applying another trained function for generating synthetic medical image data, TF2-SBD, to input data. Advantageously, the input data can be based on patient-nonspecific image features (uBM) and non-phenotypically expressed patient-specific image features (naBM). Furthermore, at least one parameter of the further trained function for generating synthetic medical image data, TF2-SBD, can be based on a comparison of synthetic medical training image data with synthetic medical reference image data. In a further step, PROV-SBD, the synthetic medical image data (SBD) can be provided. Figure 5 schematically illustrates another embodiment of a proposed computer-implemented method for providing classified image features. In a further step, synthetic medical image data (SBD) can be received from REC-SBD by applying a proposed computer-implemented method for generating synthetic medical image data to the medical image data (BD). Furthermore, the patient-specific image features (pBM) can be classified into phenotypically expressed patient-specific image features (paBM) and non-phenotypically expressed patient-specific image features (naBM) by applying another trained function for classifying patient-specific image features (TF2-CL-BM) to input data. Advantageously, the input data can be based on the patient-specific image features (pBM) and the synthetic medical image data (SBD).Furthermore, at least one parameter of the additional trained function for classifying patient-specific image features, TF2-CL-BM, can be based on a comparison of phenotypically pronounced patient-specific training image features with phenotypically pronounced patient-specific comparison image features, and a comparison of non-phenotypically pronounced training image features with non-phenotypically pronounced comparison image features. In a further step, PROV-pBM can provide the classified patient-specific image features paBM and naBM. Figure 6 schematically illustrates an embodiment of a proposed computer-implemented method for providing a trained function for identifying and classifying image features TF-IDCL-BM. In a first step, REC-TBD medical training image data (TBD) from multiple examination subjects can be received. In a second step, multiple training image features within the medical training image data (TBD) can be identified and classified by applying the trained function TF-IDCL-BM to input data. The input data can be based on the medical training image data (TBD). Furthermore, the multiple image features can advantageously be classified into patient-specific training image features (pTBM) and patient-unspecific training image features (uTBM).According to this, for each of the classified training image features pBM, uBM and / or for a combination of classified training image features pBM, uBM, a training identification parameter TIDP-pTBM, TIDP-uTBM and a training diagnostic parameter TDIAGP-pTBM, TDIAGP-uTBM can be determined DET-IDDIAGP. Furthermore, one comparison identification parameter (VIDP) and one comparison diagnostic parameter (VDIAGP) can be received for each of the objects under investigation (REC-VIDDIAGP). Each comparison identification parameter (VIDP) can contain identification information for one of the objects under investigation. Additionally, each comparison diagnostic parameter (VDIAGP) can contain diagnostic information for one of the objects under investigation. In a further step, at least one parameter of the trained function for identifying and classifying image features, TF-IDCL-BM, can be adjusted based on a comparison between the training identification parameters TIDP-pTBM and TIDP-uTBM with the comparison identification parameters VIDP, and a comparison between the training diagnostic parameters TDIAGP-pTBM and TDIAGP-uTBM with the comparison diagnostic parameters VDIAGP. Following this, the trained function for identifying and classifying image features, TF-IDCL-BM, can be deployed as PROV-TF-IDCL-BM. Figure 7 schematically illustrates an embodiment of a proposed computer-implemented method for providing a trained function for classifying patient-specific image features TF-CL-pBM. In a first step, REC-TBD medical training image data TBD from multiple examination subjects can be received. Furthermore, classified training image features pTBM and uTBM can be received by applying a proposed computer-implemented method for providing classified image features to the medical training image data TBD REC-TBM. The classified image features uBM and pBM can then be provided as the classified training image features uTBM and pTBM.By applying an identification function, particularly a biometric one, CL-pTBM to the patient-specific training image features pTBM, the patient-specific training image features pTBM can be classified into phenotypically expressed patient-specific comparison image features paVBM and non-phenotypically expressed patient-specific comparison image features naVBM. Furthermore, by applying the trained function for classifying patient-specific image features TF-CL-pBM to input data, the patient-specific training image features pTBM can be classified into phenotypically expressed patient-specific training image features paTBM and non-phenotypically expressed patient-specific training image features naTBM. Advantageously, the input data can be based on the patient-specific training image features pTBM. In a further step, at least one parameter of the trained function for classifying patient-specific image features, TF-CL-pBM, can be adjusted based on a comparison of the phenotypically expressed patient-specific training image features (paTBM) with the phenotypically expressed patient-specific comparison image features (paVBM) and a comparison between the non-phenotypically expressed patient-specific training image features (naTBM) with the non-phenotypically expressed patient-specific comparison image features (naVBM). Following this, the trained function for classifying patient-specific image features, TF-CL-pBM, can be provided as PROV-TF-CL-pBM. Figure 8 schematically illustrates an embodiment of the proposed computer-implemented method for providing a trained function for generating synthetic medical image data (TF-SBD). In a first step, REC-TBD medical training image data (TBD) from multiple subjects can be received. In a second step, REC-TBM classified training image features (pTBM) and uTBM are received by applying a proposed computer-implemented method for providing classified image features to the training image data (TBD). The classified image features (pBM) and uBM are then provided as the classified training image features (pTBM) and uTBM, respectively. In a third step, synthetic medical comparison image data (SVBD) can be generated by applying a reconstruction function (GEN-SVBD) to the patient-specific training image features (pTBM).Furthermore, synthetic medical training image data (STBD) can be generated by applying the trained function for generating synthetic medical image data (TF-SBD) to input data based on patient-specific training image features (pTBM). Subsequently, at least one parameter of the trained function for generating synthetic medical image data (TF-SBD) can be adjusted (ADJ-TF-SBD) based on a comparison of the synthetic medical reference image data (SVBD) with the synthetic medical training image data (STBD). In a further step (PROV-TF-SBD), the trained function for generating synthetic medical image data (TF-SBD) can be deployed. Figure 9 schematically illustrates an embodiment of the proposed computer-implemented method for providing an additional trained function for generating synthetic medical image data TF2-SBD. In a first step, REC-TBD medical training image data TBD from multiple subjects can be received. Furthermore, in a second step, REC-TBM classified training image features can be received by applying a proposed computer-implemented method for providing classified image features to the medical training image data TBD. The classified image features pBM and uBM can be provided as the classified training image features pTBM and uTBM, respectively. Additionally, the phenotypically expressed patient-specific image features paBM can be provided as the phenotypically expressed patient-specific training image features paTBM.Similarly, the non-phenotypically expressed patient-specific image features (naBM) can be provided as the non-phenotypically expressed patient-specific training image features (naTBM). In a further step, GEN-SVBD allows synthetic medical comparison image data (SVBD) to be generated by applying an additional reconstruction function to the patient-nonspecific training image features (uTBM) and the non-phenotypically expressed patient-specific training image features (naTBM). Furthermore, synthetic medical training image data (STBD) can be generated by applying the additional trained function for generating synthetic medical image data (TF2-SBD) to input data based on the patient-nonspecific training image features (uTBM) and the non-phenotypically expressed patient-specific training image features (naTBM). Following this, at least one parameter of the further trained function for generating synthetic medical image data TF2-SBD can be adjusted based on a comparison of the synthetic medical training image data STBD with the synthetic medical comparison image data SVBD. In a further step PROV-TF2-SBD, the further trained function for generating synthetic medical image data TF2-SBD can be deployed. Figure 10 schematically illustrates an embodiment of the proposed computer-implemented method for providing an additional trained function for classifying patient-specific image features TF2-CL-pBM. In a first step, REC-BD medical training image data (TBD) from multiple subjects can be received. In a second step, REC-STBD, synthetic medical training image data (SBD) can be received by applying a proposed computer-implemented method for generating synthetic medical image data to the medical training image data (TBD), particularly to the patient-specific training image features (pTBM). The synthetic medical image data (SBD) can then be provided as the synthetic medical training image data (STBD), and the patient-specific image features (pBM) as the patient-specific training image features (pTBM).In a third step, the patient-specific training image features (pTBM) can be classified into phenotypically pronounced patient-specific comparison image features (paVBM) and non-phenotypically pronounced patient-specific comparison image features (naVBM) by applying a further, particularly biometric, identification function CL2-pTBM to the patient-specific training image features (pTBM) and the synthetic medical training image data (STBD). Furthermore, the patient-specific training image features (pTBM) can be classified into phenotypically pronounced patient-specific training image features (paTBM) and non-phenotypically pronounced patient-specific training image features (naTBM) by applying the further trained function for classifying patient-specific image features (TF2-CL-pBM) to input data.The input data can advantageously be based on the patient-specific training image features pTBM and the synthetic medical training image data STBD. Following this, at least one parameter of the further trained function for classifying patient-specific image features TF-CL-pBM can be adapted based on a comparison of the phenotypically expressed patient-specific training image features paTBM with the phenotypically expressed patient-specific comparison image features paVBM and a comparison between the non-phenotypically expressed patient-specific training image features naTBM with the non-phenotypically expressed patient-specific comparison image features naVBM (ADJ-TF2-CL-pBM). In a further step, PROV-TF2-CL-pBM, the further trained function for classifying patient-specific image features TF2-CL-pBM can be provided. Figure 11 shows a provisioning unit PRVS comprising an interface IF, a processing unit CU, and a storage unit MU. The provisioning unit PRVS can be configured to provide classified image features PROV-BM and / or PROV-pBM. The interface IF can be configured to receive medical image data BM. Furthermore, the processing unit CU can be configured to identify multiple image features in the medical image data BD and to classify these features into patient-specific image features pBM and patient-nonspecific image features uBM by applying a trained function TF-IDCL-BM to input data. The input data can be based on the medical image data BD. Finally, the interface IF can be configured to provide the classified image features PROV-BM and / or PROV-pBM. Such a provisioning unit PRVS for providing classified image features PROV-BM and / or PROV-pBM can, in particular, be configured to execute the previously described methods according to the invention for providing classified image features and their aspects. The provisioning unit PRVS for providing classified image features PROV-BM and / or PROV-pBM can be configured to execute these methods and their aspects by configuring the interface IF and the computing unit CU to perform the corresponding process steps. Furthermore, the PRVS provisioning unit can be configured to provide synthetic medical image data PROV-SBD. Here, the IF interface can be configured to receive classified image features by applying a proposed computer-implemented method for providing classified image features to the medical image data BD. Additionally, the CU processing unit can be configured to generate the synthetic medical image data SBD by applying the trained function TF-SBD to input data. This input data can be based on patient-specific image features. Furthermore, the IF interface can be configured to provide the synthetic medical image data SBD. Such a provisioning unit PRVS for providing synthetic medical image data PROV-SBD can, in particular, be configured to execute the previously described methods according to the invention for providing synthetic medical image data and their aspects. The provisioning unit PRVS for providing synthetic medical image data PROV-SBD can be configured to execute these methods and their aspects by having the interface IF and the computing unit CU configured to perform the corresponding process steps. Fig. 12 shows a training unit TRS comprising a training interface TIF, a training computing unit TCU, and a training storage unit TMU. The training unit TRS can advantageously be configured to execute the previously described computer-implemented methods according to the invention for providing a trained function for identifying and classifying image features, and / or for providing a trained function for classifying patient-specific image features, and / or for providing a trained function for generating synthetic medical image data, and / or for providing a further trained function for generating synthetic medical image data, and / or for providing a further trained function for classifying patient-specific image features, and their aspects.The training unit (TRS) can be trained to execute these procedures and their aspects by training the training interface (TIF) and the training computational unit (TCU) to perform the corresponding procedure steps. Specifically, the training interface (TIF) can be trained to receive medical training image data (TBD) and / or classified training image features and / or synthetic medical training image data (TSBD). Furthermore, the training interface (TIF) can be trained to provide the trained function. The deployment unit PRVS and / or the training unit TRS can be, in particular, a computer, a microcontroller, or an integrated circuit. Alternatively, the deployment unit PRVS and / or the training unit TRS can be a real or virtual cluster of computers (a real cluster is called a "cluster," and a virtual cloud is called a "cloud"). The deployment unit PRVS and / or the training unit TRS can also be configured as a virtual system running on a real computer or a real or virtual cluster of computers (virtualization). An interface (IF) and / or a training interface (TIF) can be a hardware or software interface (e.g., PCI bus, USB, or FireWire). A processing unit (CU) and / or a training processing unit (TCU) can consist of hardware or software elements, such as a microprocessor or an FPGA (Field Programmable Gate Array). A memory unit (MU) and / or a training memory unit (TMU) can be implemented as random access memory (RAM) or as permanent mass storage (hard drive, USB flash drive, SD card, solid-state drive). The interface IF and / or the training interface TIF can, in particular, comprise multiple sub-interfaces that execute different steps of the respective procedures. In other words, the interface IF and / or the training interface TIF can also be understood as a plurality of interfaces IF or a plurality of training interfaces TIF, respectively. The processing unit CU and / or the training processing unit TCU can, in particular, comprise multiple sub-processing units that execute different steps of the respective procedures. In other words, the processing unit CU and / or the training processing unit TCU can also be understood as a plurality of processing units CU or a plurality of training processing units TCU, respectively. Figure 13 schematically depicts an exemplary medical C-arm X-ray unit 37 as a proposed medical imaging device. The medical C-arm X-ray unit 37 advantageously includes a proposed PRVS delivery unit for providing classified image features and / or synthetic medical image data. The medical imaging device 37, and in particular the proposed PRVS delivery unit, is configured to execute a proposed computer-implemented method for providing classified image features and / or synthetic medical image data. The medical C-arm X-ray unit 37 also comprises a detector unit 34 and an X-ray source 33. For the acquisition of medical image data BD, in particular at least one projection X-ray image, the arm 38 of the C-arm X-ray unit 37 can be movably mounted about one or more axes. Furthermore, the medical C-arm X-ray unit 37 can include a movement device 39, which enables movement of the C-arm X-ray unit 37 in space. To acquire medical image data BD of an examination area to be imaged of an object 31 arranged on a patient positioning device 32, the PRVS delivery unit can send a signal 24 to the X-ray source 33. The X-ray source 33 can then emit an X-ray beam, in particular a conical beam and / or fan beam and / or parallel beam. Upon the X-ray beam striking a surface of the detector unit 34 after interacting with the area to be imaged of the object 31, the detector unit 34 can send a signal 21 to the PRVS delivery unit. The PRVS delivery unit can, for example, receive the medical image data BD based on the signal 21. Furthermore, the medical C-arm X-ray device 37 can comprise an input unit 41, for example a keyboard, and / or a display unit 42, for example a monitor and / or display. The input unit 41 can preferably be integrated into the display unit 42, for example in the case of a capacitive input display. In this way, the medical C-arm X-ray device 37 can be controlled by an operator entering data at the input unit 41. For example, a graphical representation of the medical image data BD and / or the classified image features and / or the synthetic medical image data SBD can be displayed on the display unit 42. The schematic representations contained in the described figures do not depict any scale or size ratio. Finally, it should be noted once again that the methods described in detail above and the devices shown are merely exemplary embodiments which can be modified in various ways by a person skilled in the art without departing from the scope of the invention. Furthermore, the use of the indefinite articles "a" or "an" does not preclude the possibility that the features in question may be present multiple times. Likewise, the terms "unit" and "element" do not preclude the possibility that the components in question consist of several interacting sub-components, which may also be spatially distributed.

Claims

A computer-implemented method for providing classified image features, comprising: - Receiving (REC-BD) medical image data (BD), wherein the medical image data (BD) is acquired using at least one medical imaging device; - Identifying multiple image features in the medical image data (BD) and classifying the multiple image features into patient-specific image features (pBM) and patient-nonspecific image features (uBM) by applying a trained image feature identification and classification function (TF-IDCL-BM) to input data, wherein the multiple image features include geometric and / or anatomical image features and / or statistical image information; wherein the input data is based on the medical image data (BD); and wherein at least one parameter of the trained image feature identification and classification function (TF-IDCL-BM) is based on a comparison of training identification parameters (TIDP-pTBM).TIDP-uTBM) with comparison identification parameters (VIDP) and a comparison of training diagnostic parameters (TDIAGP-pTBM, TDIAGP-uTBM) with comparison diagnostic parameters (VDIAGP), - Classifying patient-specific image features (pBM) into phenotypically expressed patient-specific image features (paBM) and non-phenotypically expressed patient-specific image features (naBM) by applying a trained function for classifying patient-specific image features (TF-CL-pBM) to input data, wherein the input data is based on the patient-specific image features (pBM),wherein at least one parameter of the trained function for classifying patient-specific image features (TF-CL-pBM) is based on a comparison of phenotypically pronounced patient-specific training image features (paTBM) with phenotypically pronounced patient-specific comparison image features (paVBM) and a comparison of non-phenotypically pronounced patient-specific training image features (naTBM) with non-phenotypically pronounced patient-specific comparison image features (naVBM),- providing (PROV-pBM) the classified patient-specific image features (paBM, naBM) and the patient-non-specific image features (uBM). A computer-implemented method for providing synthetic medical image data, comprising: - Receiving (REC-BD) medical image data (BD), wherein the medical image data (BD) is acquired using at least one medical imaging device; - Receiving (REC-BM) classified image features by applying a computer-implemented method according to claim 1 to the medical image data (BD); or - Identifying multiple image features in the medical image data (BD) and classifying the multiple image features into patient-specific image features (pBM) and patient-nonspecific image features (uBM) by applying a trained image feature identification and classification function (TF-IDCL-BM) to input data, wherein the multiple image features comprise geometric and / or anatomical image features and / or statistical image information, wherein the input data is based on the medical image data (BD).wherein at least one parameter of the trained function for identifying and classifying image features (TF-IDCL-BM) is based on a comparison of training identification parameters (TIDP-pTBM, TIDP-uTBM) with comparison identification parameters (VIDP) and a comparison of training diagnostic parameters (TDIAGP-pTBM, TDIAGP-uTBM) with comparison diagnostic parameters (VDIAGP), providing (PROV-BM) the classified image features (pBM, uBM), generating the synthetic medical image data (SBD) by applying a trained function for generating synthetic medical image data (TF-SBD) to input data, wherein the input data is based on the patient-specific image features (pBM), wherein at least one parameter of the trained function for generating synthetic medical image data (TF-SBD) is based on a comparison of synthetic medical training image data (STBD) with synthetic medical comparison image data (SVBD),- Provision (PROV-SBD) of synthetic medical image data (SBD). A computer-implemented method for providing synthetic medical image data, comprising: - Receiving (REC-BD) medical image data (BD), wherein the medical image data (BD) is acquired using at least one medical imaging device; - Receiving (REC-BM) classified image features by applying a computer-implemented method according to claim 1 to the medical image data; or - Identifying multiple image features in the medical image data (BD) and classifying the multiple image features into patient-specific image features (pBM) and patient-nonspecific image features (uBM) by applying a trained image feature identification and classification function (TF-IDCL-BM) to input data, wherein the multiple image features comprise geometric and / or anatomical image features and / or statistical image information, wherein the input data is based on the medical image data (BD).wherein at least one parameter of the trained function for identifying and classifying image features (TF-IDCL-BM) is based on a comparison of training identification parameters (TIDP-pTBM, TIDP-uTBM) with comparison identification parameters (VIDP) and a comparison of training diagnostic parameters (TDIAGP-pTBM, TDIAGP-uTBM) with comparison diagnostic parameters (VDIAGP), provision (PROV-BM) of the classified image features (pBM, uBM), generation of the synthetic medical image data by applying another trained function for generating synthetic medical image data (TF2-SBD) to input data, wherein the input data is based on the patient-nonspecific image features (uBM) and / or the non-phenotypically expressed patient-specific image features (naBM),wherein at least one parameter of the further trained function for generating synthetic medical image data (TF2-SBD) is based on a comparison of synthetic medical training image data (STBD) with synthetic medical comparison image data (SVBD),- Provision (PVOV-SBD) of the synthetic medical image data (SBD)., A computer-implemented method for providing classified image features, comprising: - Receiving (REC-BD) medical image data (BD), wherein the medical image data (BD) is acquired using at least one medical imaging device, - Receiving (REC-SBD) synthetic medical image data (SBD) by applying a computer-implemented method according to claim 2 to the medical image data (BD), - Classifying the patient-specific image features (pBM) into phenotypically expressed patient-specific image features (paBM) and non-phenotypically expressed patient-specific image features (naBM) by applying a further trained function for classifying patient-specific image features (TF2-CL-pBM) to input data, wherein the input data is based on the patient-specific image features (pBM) and the synthetic medical image data (SBD).wherein at least one parameter of the further trained function for classifying patient-specific image features (TF2-CL-pBM) is based on a comparison of phenotypically pronounced patient-specific training image features (paTBM) with phenotypically pronounced patient-specific comparison image features (paVBM) and a comparison of non-phenotypically pronounced patient-specific training image features (naTBM) with non-phenotypically pronounced patient-specific comparison image features (naVBM),- Provision (PROV-pBM) of the classified patient-specific image features (paBM, naBM). A computer-implemented method for providing a trained function for identifying and classifying image features (TF-IDCL-BM), comprising: - Receiving (REC-TBD) medical training image data (TBD) from multiple subjects, wherein the medical training image data (TBD) are acquired using at least one medical imaging device; - Identifying multiple training image features in the medical training image data (TBD) and classifying the multiple training image features into patient-specific training image features (pTBM) and patient-non-specific training image features (uTBM) by applying the trained function for identifying and classifying image features (TF-IDCL-BM) to input data, wherein the multiple training image features include geometric and / or anatomical image features and / or statistical image information, and wherein the input data are based on the medical training image data (TBD).- Determining (DET-IDDIAGP) training identification parameters (TIDP-pTBM, TIDP-uTBM) and training diagnostic parameters (TDIAGP-pTBM, TDIAGP-uTBM) based on the classified training image features (pTBM, uTBM), wherein one training identification parameter (TIDP-pTBM, TIDP-uTBM) and one training diagnostic parameter (TDIAGP-pTBM, TDIAGP-uTBM) are determined for each of the classified training image features (pTBM, uTBM) and / or for a combination of classified training image features (pTBM, uTBM), - Receiving (REC-VIDDIAGP) one comparison identification parameter (VIDP) and one comparison diagnostic parameter (VDIAGP) for each of the objects under investigation, wherein one comparison identification parameter (VIDP) comprises identification information for one of the objects under investigation, wherein one Comparative diagnostic parameters (VDIAGP) include diagnostic information for one of the subjects under investigation.- Adapting (ADJ-TF-IDCL-BM) at least one parameter of the trained feature identification and classification function (TF-IDCL-BM) based on a comparison between the training identification parameters (TIDP-pTBM, TIDP-uTBM) with the comparison identification parameters (VIDP) and between the training diagnostic parameters (TDIAGP-pTBM, TDIAGP-uTBM) with the comparison diagnostic parameters (VDIAGP), - Providing (PROV-TF-IDCL-BM) the trained feature identification and classification function (TF-IDCL-BM). A computer-implemented method for providing a trained function for classifying patient-specific image features (TF-CL-pBM), comprising: - Receiving (REC-TBD) medical training image data (TBD) from multiple subjects, wherein the medical training image data (TBD) are acquired using at least one medical imaging device; - Identifying multiple training image features in the medical training image data (TBD) and classifying the multiple training image features into patient-specific training image features (pTBM) and patient-nonspecific training image features (uTBM) by applying a trained function for identifying and classifying image features (TF-IDCL-BM) to input data, wherein the multiple training image features include geometric and / or anatomical image features and / or statistical image information, and wherein the input data are based on the medical training image data (TBD).wherein at least one parameter of the trained function for identifying and classifying image features (TF-IDCL-BM) is based on a comparison of training identification parameters (TIDP-pTBM, TIDP-uTBM) with comparison identification parameters (VIDP) and a comparison of training diagnostic parameters (TDIAGP-pTBM, TDIAGP-uTBM) with comparison diagnostic parameters (VDIAGP), providing the classified training image features (uTBM, pTBM), classifying the patient-specific training image features (pTBM) into phenotypically expressed patient-specific comparison image features (paVBM) and non-phenotypically expressed patient-specific comparison image features (naVBM) by applying an identification function (CL-pTBM), in particular a biometric function, to the patient-specific training image features (pTBM).- Classifying patient-specific training image features (pTBM) into phenotypically expressed patient-specific training image features (paTBM) and non-phenotypically expressed patient-specific training image features (naTBM) by applying the trained function for classifying patient-specific image features (TF-CL-pBM) to input data, where the input data is based on the patient-specific training image features (pTBM).- Adapting (ADJ-TF-CL-pBM) at least one parameter of the trained function for classifying patient-specific image features (TF-CL-pBM) based on a comparison of the phenotypically expressed patient-specific training image features (paTBM) with the phenotypically expressed patient-specific comparison image features (paVBM) and a comparison of the non-phenotypically expressed patient-specific training image features (naTBM) with the non-phenotypically expressed patient-specific comparison image features (naVBM), - Providing (PROV-TF-CL-pBM) the trained function for classifying patient-specific image features (TF-CL-pBM). A computer-implemented method for providing a trained function for generating synthetic medical image data, comprising: - Receiving (REC-TBD) medical training image data (TBD) from multiple subjects, wherein the medical training image data (TBD) are acquired using at least one medical imaging device, - Receiving (REC-TBM) classified training image features by applying a computer-implemented method according to claim 1 to the medical training image data, wherein the classified image features (pBM, uBM) are provided as the classified training image features (pTBM, uTBM) and the patient-specific image features (pBM) are provided as patient-specific training image features (pTBM).or identifying multiple training image features in the medical training image data (TBD) and classifying the multiple training image features into patient-specific training image features (pTBM) and patient-unspecific training image features (uTBM) by applying a trained image feature identification and classification function (TF-IDCL-BM) to input data, wherein the multiple training image features include geometric and / or anatomical image features and / or statistical image information, wherein the input data are based on the medical training image data (TBD), and wherein at least one parameter of the trained image feature identification and classification function (TF-IDCL-BM) is based on a comparison of training identification parameters (TIDP-pTBM, TIDP-uTBM) with comparison identification parameters (VIDP) and a comparison of training diagnostic parameters (TDIAGP-pTBM, TDIAGP-uTBM) with comparison diagnostic parameters (VDIAGP).Provision of the classified training image features (uTBM, pTBM), - Generation (GEN-SVBD) of synthetic medical comparison image data (SVBD) by applying a reconstruction function to the patient-specific training image features (pBM), - Generation of synthetic medical training image data (STBD) by applying the trained synthetic medical image data generation function (TF-SBD) to input data, where the input data is based on the patient-specific training image features (pBM), - Adjustment (ADJ-TF-SBD) of at least one parameter of the trained synthetic medical image data generation function (TF-SBD) based on a comparison of the synthetic medical comparison image data (SVBD) with the synthetic medical training image data (STBD), - Provision (PROV-TF-SBD) of the trained synthetic medical image data generation function (TF-SBD). A computer-implemented method for providing a further trained function for generating synthetic medical image data (TF2-SBD), comprising: - Receiving (REC-TBD) medical training image data (TBD) from multiple subjects, wherein the medical training image data (TBD) are acquired using at least one medical imaging device, - Receiving (REC-TBM) classified training image features by applying a computer-implemented method according to claim 1 to the medical training image data (TBD), wherein the classified image features (pBM, uBM, paBM, naBM) are provided as the classified training image features (pTBM, uTBM, paTBM, naTBM), the patient-nonspecific image features (uBM) as patient-nonspecific training image features (uTBM), and / or the non-phenotypically expressed patient-specific image features (naBM) as non-phenotypically expressed training image features (naTBM).or identifying multiple training image features in the medical training image data (TBD) and classifying the multiple training image features into patient-specific training image features (pTBM) and patient-unspecific training image features (uTBM) by applying a trained image feature identification and classification function (TF-IDCL-BM) to input data, wherein the multiple training image features include geometric and / or anatomical image features and / or statistical image information, wherein the input data are based on the medical training image data (TBD), and wherein at least one parameter of the trained image feature identification and classification function (TF-IDCL-BM) is based on a comparison of training identification parameters (TIDP-pTBM, TIDP-uTBM) with comparison identification parameters (VIDP) and a comparison of training diagnostic parameters (TDIAGP-pTBM, TDIAGP-uTBM) with comparison diagnostic parameters (VDIAGP).Providing the classified training image features (uTBM, pTBM), generating (GEN-SVBD) synthetic medical comparison image data (SVBD) by applying a reconstruction function to the patient-nonspecific training image features (uTBM) and / or the non-phenotypically expressed patient-specific training image features (naTBM), generating synthetic medical training image data (STBD) by applying the further trained function for generating synthetic medical image data (TF2-SBD) to input data, wherein the input data is based on the patient-nonspecific training image features (uTBM) and / or the non-phenotypically expressed patient-specific training image features (naTBM).- Adjusting (ADJ-TF2-SBD) at least one parameter of the further trained function for generating synthetic medical image data (TF2-SBD) based on a comparison of the synthetic medical reference image data (SVBD) with the synthetic medical training image data (STBD), - Providing (PROV-TF2-SBD) the further trained function for generating synthetic medical image data (TF2-SBD). A computer-implemented method for providing a further trained function for classifying patient-specific image features (TF2-CL-pBM), comprising: - Receiving (REC-TBD) medical training image data (TBD) of multiple subjects, wherein the medical training image data (TBD) are acquired using at least one medical imaging device, - Receiving (REC-STBD) synthetic medical training image data (STBD) by applying a computer-implemented method according to claim 2 to the medical training image data (TBD), wherein the synthetic medical image data (STBD) are provided as the synthetic medical training image data (STBD) and the patient-specific image features (pBM) are provided as patient-specific training image features (pTBM).- Classifying patient-specific training image features (pTBM) into phenotypically pronounced patient-specific comparison image features (paVBM) and non-phenotypically pronounced patient-specific comparison image features (naVBM) by applying a further, in particular biometric, identification function (CL2-pTBM) to the patient-specific training image features (pTBM) and the synthetic medical training image data (STBD); - Classifying patient-specific training image features (pTBM) into phenotypically pronounced patient-specific training image features (paTBM) and non-phenotypically pronounced patient-specific training image features (naTBM) by applying the further trained function for classifying patient-specific image features (TF2-CL-pBM) to input data, wherein the input data is based on the patient-specific training image features (pTBM) and the synthetic medical training image data (STBD).- Adapting (ADJ-TF2-CL-pBM) at least one parameter of the further trained function for classifying patient-specific image features (TF2-CL-pBM) based on a comparison of the phenotypically expressed patient-specific training image features (paTBM) with the phenotypically expressed patient-specific comparison image features (paVBM) and a comparison of the non-phenotypically expressed patient-specific training image features (naTBM) with the non-phenotypically expressed patient-specific comparison image features (naVBM), - Providing (PROV-TF2-CL-pBM) the further trained function for classifying patient-specific image features (TF2-CL-pBM). Provisioning unit (PRVS) for providing classified image features, comprising a processing unit (CU) and an interface (IF), wherein the interface (IF) is configured to receive (REC-BD) medical image data (BD), wherein the medical image data (BD) is acquired using at least one medical imaging device, and wherein the processing unit (CU) is configured to identify multiple image features in the medical image data (BD) and classify the multiple image features into patient-specific image features (pBM) and patient-nonspecific image features (uBM) by applying a trained image feature identification and classification function (TF-IDCL-BM) to input data, wherein the multiple image features include geometric and / or anatomical image features and / or statistical image information, and wherein the input data is based on the medical image data (BD).wherein at least one parameter of the trained function for identifying and classifying image features (TF-IDCL-BM) is based on a comparison of training identification parameters (TIDP-pTBM, TIDP-uTBM) with comparison identification parameters (VIDP) and a comparison of training diagnostic parameters (TDIAGP-pTBM, TDIAGP-uTBM) with comparison diagnostic parameters (VDIAGP),- wherein the computational unit (CU) for classifying the patient-specific image features (pBM) into phenotypically expressed patient-specific image features (paBM) and non-phenotypically expressed patient-specific image features (naBM) is trained by applying a trained function for classifying patient-specific image features (TF-CL-pBM) to input data, wherein the input data is based on the patient-specific image features (pBM),wherein at least one parameter of the trained function for classifying patient-specific image features (TF-CL-pBM) is based on a comparison of phenotypically pronounced patient-specific training image features (paTBM) with phenotypically pronounced patient-specific comparison image features (paVBM) and a comparison of non-phenotypically pronounced patient-specific training image features (naTBM) with non-phenotypically pronounced patient-specific comparison image features (naVBM), - wherein the interface (IF) is further configured to provide (PROV-BM) the classified patient-specific image features (paBM, naBM) and the patient-nonspecific image features (uBM). Provisioning unit (PRVS) for providing synthetic medical image data, comprising a processing unit (CU) and an interface (IF), wherein the interface (IF) is configured to receive (REC-BD) medical image data (BD), wherein the medical image data (BD) are acquired using at least one medical imaging device, and wherein the interface (IF) is further configured to receive (REC-BM) classified image features (pBM, uBM, paBM,naBM) by applying a computer-implemented method according to claim 1 to the medical image data (BD), or wherein the computing unit (CU) is configured to identify multiple image features in the medical image data (BD) and classify the multiple image features into patient-specific image features (pBM) and patient-nonspecific image features (uBM) by applying a trained function for identifying and classifying image features (TF-IDCL-BM) to input data, wherein the input data is based on the medical image data (BD), wherein at least one parameter of the trained function for identifying and classifying image features (TF-IDCL-BM) is based on a comparison of training identification parameters (TIDP-pTBM, TIDP-uTBM) with comparison identification parameters (VIDP) and a comparison of training diagnostic parameters (TDIAGP-pTBM, TDIAGP-uTBM) with comparison diagnostic parameters (VDIAGP).wherein the interface (IF) is further configured to provide (PROV-BM) the classified image features (pBM, uBM),- wherein the computing unit (CU) is configured to generate the synthetic medical image data (SBD) by applying a trained synthetic medical image data generation function (TF-SBD) to input data, wherein the input data is based on the patient-specific image features (pBM), wherein at least one parameter of the trained synthetic medical image data generation function (TF-SBD) is based on a comparison of synthetic medical training image data (STBD) with synthetic medical comparison image data (SVBD),- wherein the interface (IF) is further configured to provide (PROV-SBD) the synthetic medical image data (SBD). Medical imaging device (37) comprising a provisioning unit (PRVS) according to claim 10 or 11, which is configured to perform a method according to one of claims 1 to 4, wherein the medical imaging device (37) is configured to acquire and / or receive and / or provide the medical image data (BD). Training unit (TRS) configured to perform a computer-implemented method according to any one of claims 5 to 9. A computer program product comprising a computer program that can be directly loaded into a memory of a provisioning unit, with program sections to execute all steps of the computer-implemented method according to any one of claims 1 to 5 when the program sections are executed by the provisioning unit (PRVS); and / or which can be directly loaded into a training memory (TMU) of a training unit (TRS), with program sections to execute all steps of the computer-implemented method according to any one of claims 6 to 9 when the program sections are executed by the training unit (TRS). A computer-readable storage medium on which program sections readable and executable by a provisioning unit (PRVS) are stored to execute all steps of the computer-implemented method according to any one of claims 1 to 5 when the program sections are executed by the provisioning unit (PRVS); and / or on which program sections readable and executable by a training unit (TRS) are stored to execute all steps of the computer-implemented method according to any one of claims 6 to 9 when the program sections are executed by the training unit (TRS).

Images