Medical image recording device, method for operating a medical image recording device, computer program, electronically readable data carrier and method for designing a medical image recording device

Abstract

The invention relates to a medical image recording device, having: - a recording arrangement (3) for recording raw image data with a raw image data resolution and a control device (9) comprising a reconstruction unit (13) for reconstructing an image data set from the raw image data, wherein the image data set has a first spatial resolution in at least one reference plane when reconstructed from the raw image data with the raw image data resolution, wherein the control device (9) also has an image processing unit (14) for carrying out a trained resolution enhancement function for enhancing the spatial resolution to a second spatial resolution which is higher than the first spatial resolution.

Description

[0001] Description

[0002] Medical imaging device, method for operating a medical imaging device, computer program, electronically readable data carrier and method for designing a medical imaging device

[0003] The invention relates to a medical image acquisition device which comprises:

[0004] - a recording setup for capturing raw image data at a raw image data resolution and

[0005] - a control device with a reconstruction unit for reconstructing an image data set from the raw image data, wherein the image data set, when reconstructed from the raw image data, has a first spatial resolution in at least one reference plane.

[0006] In addition, the invention relates to a computer-implemented method for operating such an image acquisition device, a computer program, an electronically readable data carrier and a method for designing a medical image acquisition device.

[0007] A key quality characteristic of medical imaging devices, such as X-ray and magnetic resonance imaging (MRI) systems, is the spatial resolution of the reconstructed image data set resulting from an imaging procedure. The corresponding imaging setup, as well as the data transmission and processing capabilities, must be significantly larger and more complex to achieve increasing spatial resolution.

[0008] For example, in computed tomography (CT), configurations are typically used in which the cross-sectional images of a reconstructed image stack have a spatial resolution of 512 x 512 pixels. Depending on the selected field of view size, this results in various specific resolutions (e.g., in mm). To achieve the desired spatial resolution, the X-ray detector, the X-ray source, and the data transmission system—specifically its data transmission equipment—must meet certain design parameters. These parameters might include, for example, a minimum resolution and the number of pixels from the X-ray detector that can be read out. The resulting raw image data must also be transmitted, ideally in real time, to a control unit of the CT scanner.Design parameters for the X-ray detector can include, for example, a target size of the detector pixels and a number of electronic channels; design parameters for the X-ray source include a target size of the focal point; and design parameters for data transmission means, in particular a sliding contact from the X-ray detector rotating in the gantry to the control unit, include a minimum data transmission rate. Due to these limitations, computed tomography systems remain extremely complex and expensive at the desired spatial resolutions. Furthermore, a lower limit exists regarding further cost reduction measures due to the hardware requirements, especially the design parameters based on the target spatial resolution in the generated image datasets. On the other hand, the demand for more affordable entry-level models is growing, both in computed tomography and in other medical imaging modalities.

[0009] The invention is therefore based on the objective of providing a way to design an image acquisition device, in particular a computed tomography device, in a less complex and in particular more cost-effective manner, while maintaining a spatial target resolution for reconstructed image data sets.

[0010] This problem is solved according to the invention by a medical image acquisition device, a computer-implemented method, a computer program, an electronically readable data carrier, and a method for designing an image acquisition device according to the dependent claims. Advantageous embodiments are described in the sub-claims.

[0011] In an image acquisition device of the type mentioned at the outset, the invention provides that the control unit further comprises an image processing unit for executing a trained resolution enhancement function to increase the spatial resolution of the image data set to a second spatial resolution that is higher than the first spatial resolution.

[0012] The trained resolution enhancement function can be, in particular, a super-resolution function. It is therefore proposed to implement a cost-effective image acquisition device, especially a computed tomography system, by designing the hardware components with a lower initial resolution, but implementing a trained resolution enhancement function, preferably always-on, in an image processing unit to obtain a higher secondary resolution, particularly as the target resolution, in the image dataset. Algorithms for increasing the spatial resolution in image data, also known as super-resolution algorithms (SR algorithms), increase the spatial resolution, especially the pixel resolution, of images in computer image processing.Their primary application over the past few decades has been image and video enhancement in fields such as aerial and space reconnaissance, for example, of satellite imagery. More recently, resolution enhancement functions, especially those based on machine learning techniques, i.e., artificial intelligence (trained resolution enhancement functions), have also become increasingly common in the gaming industry. For example, graphics cards for gaming software can achieve higher resolutions and higher frame rates simultaneously.

[0013] For example, leading graphics card manufacturers have developed proprietary trained super resolution functions, such as Nvidia's "Deep Learning Super Sampling" (DLSS) and AMD's "FidelityFX Super Resolution" (FSR). See, for example, Sven Bauduin's article in PC World, "AMD FSR 3 vs. Nvidia DLSS 3: Similarities and Differences," November 2, 2023. Trained resolution upscaling functions, particularly with regard to real-time capability, are also the subject of research in other areas. Resolution improvements of factors of 1.5 to 3 per dimension (so-called upscaling factors) can already be achieved.

[0014] According to this disclosure, these developments are now also used to achieve high target resolutions in image datasets even with less complex and, in particular, less expensive hardware for image acquisition devices. Due to the now widespread availability of resolution enhancement functions, especially super-resolution functions, no cost increases for the control electronics are expected as a result of implementing the resolution enhancement function. Thus, a particularly advantageous embodiment of the present invention can provide that the image processing unit, at least for the portion executing the resolution enhancement function, is implemented on a graphics card of the control device. Graphics cards are particularly suitable for executing image processing algorithms of artificial intelligence, especially due to their specially designed processors (GPUs).In particular, graphics cards are already available on which trained resolution enhancement functions are implemented, which can also be used at least partially within the image processing unit of the control device.

[0015] In general, a trained function replicates cognitive functions that people associate with other human brains. Through training based on training data (machine learning), the trained function is able to adapt to new circumstances and detect and extrapolate patterns. Another term for "trained function" is "trained machine learning model."

[0016] Generally speaking, the parameters of a trained function can be adjusted through training. Specifically, supervised learning, semi-supervised learning, unsupervised learning, reinforcement learning, and / or active learning can be used. Furthermore, representational learning (also known as feature learning) can be employed. The parameters of the trained function can be adjusted iteratively through multiple training steps. In particular, a specific cost function can be minimized during training. For example, the backpropagation algorithm can be used when training a neural network.

[0017] A trained function can, for example, comprise a neural network, a support vector machine (SVM), a decision tree, and / or a Bayesian network, and / or the trained function can be based on k-means clustering, Q-learning, genetic algorithms, and / or assignment rules. Specifically, a neural network can be a deep neural network, a convolutional neural network (CNN), or a deep CNN. Furthermore, the neural network can be an adversarial network, a deep adversarial network, and / or a generative adversarial network (GAN). For the trained resolution enhancement function, the use of neural networks, especially CNNs, is preferred.

[0018] From a technical perspective, a wide variety of architectures exist for achieving super-resolution, an increase in resolution beyond that permitted by the recording setup and data transmission device. In a particularly advantageous embodiment of the present invention, the trained resolution enhancement function can be based on a SwinIR architecture. This open-source algorithm is described, for example, in an article by Jingyun Liang, "SwinIR: Image Restoration Using Swin Transformer," 2021, arXiv:2108.10257v1, and combines a neural network for implementing a vision transformer with CNNs. Thus, the architecture offers good accessibility. Tests have shown that the present invention provides excellent performance.

[0019] When implementing and training the resolution enhancement function, a loss function or loss model can be used to guide the model optimization. A loss function is preferred that allows for improvements in both the peak SNR and the perceived quality of the image dataset.

[0020] In specific configurations, the image processing unit may be configured to apply the trained resolution enhancement function to the image data set reconstructed by the reconstruction unit, and / or the image processing unit may be configured to apply the resolution enhancement function to the raw image data to increase the spatial resolution of the image data set. Applying the function to the raw image data is particularly advantageous when it comprises a temporal sequence of several raw images, such as projection images from X-ray imaging, preferably computed tomography. For example, in computed tomography (CT) or related techniques, projection images of an area of ​​an object under investigation are acquired in rapid succession using different projection geometries that are, however, only slightly different in pairs.This allows the use of temporal super-resolution techniques because overlapping image content is present. In other words, when dealing with a temporal sequence of raw images showing at least partially overlapping content, the resolution enhancement function can be designed to take into account at least one temporally adjacent raw image when improving the resolution of a specific raw image. This allows for an improvement in the spatial resolution of the raw images, particularly by enabling the recovery of image content based on the additional information in temporally adjacent raw images, or by allowing it to be determined despite the reduced raw image data resolution.

[0021] It should be noted that such temporal super-resolution techniques can, of course, also be advantageously applied when reconstructing a temporal sequence of image datasets, for example, in surveillance imaging, in medical technology for monitoring minimally invasive procedures, or the like. In such cases, it can be provided that, when dealing with a temporal sequence of image datasets showing at least partially overlapping image content, the resolution enhancement function is designed to take into account at least one temporally adjacent image dataset when improving the resolution of a specific image dataset.

[0022] Applying this technique directly to the image dataset after it has been reconstructed at the first resolution is advantageous because it typically involves a larger, already statistically merged dataset, thus reducing the probability of image errors by increasing the spatial resolution. For example, single-frame super-resolution techniques, which do not require temporal considerations, can be used here.

[0023] It should be noted here that incorporating background knowledge is not only useful when considering temporal aspects, such as the projection geometries of temporally adjacent raw images, but can also be used in other ways to improve the resolution enhancement achieved by the resolution enhancement function and to obtain the highest possible image quality. This can also include static background knowledge, such as pre-image datasets of the same investigation area.

[0024] A fundamental idea of ​​the present invention is to employ techniques for computationally increasing spatial resolution, i.e., super-resolution techniques, in order to significantly reduce the complexity and cost of the main components of the image acquisition device, in particular the camera assembly and a data transmission device to the control unit. This achieves a reduction in hardware costs without increasing, or excessively increasing, the demands placed on the control unit, especially with regard to computing power, particularly when the image processing unit is at least partially implemented on a graphics card. A cost-effective entry-level device for the respective imaging modality can be provided by compensating for any reduction in resolution due to the use of inexpensive and / or less complex components through software.By using common, mature, trained resolution enhancement functions, a significantly better perceptible image dataset can be provided without loss of image information, especially with regard to diagnostics, whereby, particularly with temporal super resolution techniques, image information can also be recognizably recovered in the image dataset.

[0025] In other words, the present invention offers the particular advantage of allowing at least one design parameter of at least one component of the image processing device to be selected for less effort, particularly for obtaining the first spatial resolution, than for directly obtaining the second spatial resolution through the reconstruction unit. Specifically, the design parameter can relate to a third spatial resolution of a detector of the imaging arrangement, for example, described by the size of detector pixels or a number of electronic channels, and / or, in the case of transmission imaging, the focus size of a radiation source and / or a data transmission rate of a data transmission means between the imaging arrangement and the control unit.It is particularly advantageous to provide that the image acquisition device is a computed tomography system comprising a gantry in which an X-ray source and an X-ray detector of the imaging arrangement are rotatably mounted in a rotational plane, wherein the reconstruction unit for reconstructing the image data set, particularly three-dimensional image data, especially as a stack of cross-sectional images, is designed from two-dimensional projection images taken at different projection angles. In this case, it can therefore be specifically provided that the raw image data is acquired as a temporal sequence of projection images.

[0026] Especially with regard to computed tomography equipment, the concept proposed according to the invention can be used expediently, since detector pixel target sizes for certain target resolutions, just like X-ray emitters with sufficiently small foci as well as data transmission means from the rotating part to the stationary part for high data transmission rates, represent enormous effort and cost factors.

[0027] In a computed tomography system, it may be specifically designed that the first and second resolutions relate at least to the plane of rotation, in particular as the image plane of the slice images in the image stack. However, it is also possible for the first and second resolutions to relate to all spatial directions of the image data set, meaning that the resolution can also be increased perpendicular to the image plane of the slice images, for example by increasing the number of slice images in the image stack.

[0028] In a specific embodiment of the present invention, the second spatial resolution (or target resolution) may comprise 512 by 512 pixels in the plane of rotation, while the first spatial resolution is 256 by 256 pixels in the plane of rotation. These spatial resolutions thus relate in particular to the number of pixels to be reconstructed in the slice images of the slice image stack.

[0029] In the classic design of computed tomography (CT) systems, the cost of the X-ray detector scales with the cost per unit area of ​​the detector and the total detector area. The integrated electronics, the configuration of a photon-counting direct converter, and the application-specific integrated circuits (ASICs) have been optimized for high-end CT systems. For low-end or entry-level CT systems, the most cost-effective solution for the X-ray detector is a simple photosensor with a small number of electronic channels provided by a cost-effective ASIC. For example, if a spatial resolution of only 256 x 256 is targeted instead of 512 x 512 in the image plane of the slices (rotation plane), the number of electronic channels can be reduced by at least a factor of 2.If the resolution increase is also performed perpendicular to the plane of rotation using the trained resolution increase function, an additional factor of 2 can be achieved to reduce the number of channels.

[0030] If, for example, the spatial resolution of the X-ray detector (and thus the raw image data resolution) is reduced by a factor of 2, as described above, the size of the focal point of the X-ray source can also be reduced by a corresponding factor. The explanations regarding the increase in resolution perpendicular to the plane of rotation can be continued analogously for the X-ray source. Reducing the size of the focal point has a particularly advantageous, non-linear effect on the design of the X-ray source, since the cooling requirements can be reduced and the demands placed on it by the rotation are also lowered.

[0031] Reducing the resolution to the first spatial resolution also allows for a reduction in the required data transmission rates, particularly by a factor of 2 to 4. This also permits the use of wireless data transmission means, optionally with buffering in a storage medium if necessary. Therefore, an advantageous embodiment of the present invention with regard to the computed tomography device can provide for a data transmission device for transmitting the raw image data from the imaging arrangement to the control unit. This data transmission device is a wireless data transmission means for transmitting data between a rotating portion of the computed tomography device comprising the imaging arrangement and a stationary portion of the computed tomography device comprising the control unit. A slip ring or the like can thus be omitted.

[0032] Generally speaking, the control unit preferably comprises at least one processor and at least one storage medium. Functional units, particularly for carrying out steps of an operating procedure, are implemented by hardware and / or software. In addition to the reconstruction unit and the image processing unit, the control unit may, for example, also include a recording unit for controlling the recording operation of the image acquisition device. Besides the medical image acquisition device, the invention also relates to a method for operating such a medical image acquisition device, which comprises:

[0033] - a recording setup for capturing raw image data at a raw image data resolution and

[0034] - a control device with a reconstruction unit for reconstructing an image data set from the raw image data, wherein the image data set, when reconstructed from the raw image data, has a first spatial resolution in at least one reference plane.

[0035] The method involves using a trained resolution enhancement function to increase the spatial resolution of the image data set to a second spatial resolution that is higher than the first spatial resolution.

[0036] All statements relating to the image acquisition device according to the invention can be applied analogously to the operating method according to the invention and vice versa, so that the advantages already mentioned can also be obtained with the operating method.

[0037] A computer program according to the invention can be directly loaded into a control unit of an image acquisition device and comprises program means such that, when the computer program is executed in the control unit, the latter is caused to perform the steps of the operating procedure according to the invention. The computer program can be stored on an electronically readable data carrier according to the invention, which therefore includes control information stored thereon, comprising at least one computer program according to the invention and designed such that, when the data carrier is used in a control unit of an image acquisition device, the latter is configured to carry out the operating procedure according to the invention. The data carrier can, in particular, be a non-transient data carrier, for example, a CD-ROM.

[0038] Finally, the invention also relates to a method for designing a medical image acquisition device, comprising the following design steps:

[0039] - Defining a spatial target resolution for image data sets acquired with the image acquisition device in at least one reference plane,

[0040] - Defining a first spatial resolution to be used as a design target for the image data sets, at least in the reference plane, wherein the first spatial resolution is lower than the target resolution,

[0041] - Designing a recording setup for recording raw image data at a raw image data resolution and / or at least a data transmission means for the raw image data such that, when reconstructing an image data set from raw image data of the raw image data resolution, this has the first spatial resolution,

[0042] - Designing a control unit of the image acquisition device such that it includes an image processing unit for executing a trained resolution enhancement function to increase the spatial resolution to a second spatial resolution corresponding to the target resolution.

[0043] All statements regarding the image acquisition device, the operating procedure, the computer program, and the electronically readable data carrier also apply accordingly to the design procedure. In a manufacturing process for producing an image acquisition device, the design determined in the design steps of the design procedure can be used to manufacture the image acquisition device.

[0044] Further advantages and details of the present invention will become apparent from the exemplary embodiments described below and from the drawings. These show:

[0045] Fig. 1 shows a schematic diagram of an image acquisition device according to the invention.

[0046] Fig. 2 shows a flowchart of a first embodiment of an operating method according to the invention,

[0047] Fig. 3 shows a flowchart of a second embodiment of an operating method according to the invention, and

[0048] Fig. 4 shows a flowchart of an exemplary embodiment of a design method according to the invention.

[0049] Fig. 1 shows a schematic diagram of an embodiment of an image acquisition device 1 according to the invention, which in this case is configured as a computed tomography device. The computed tomography device comprises a gantry 2 in which an imaging assembly 3, comprising an X-ray tube 4 and an X-ray detector 5, is rotatably mounted about a through-hole 6. A patient 7 can be positioned within the gantry 2 for image acquisition by means of a patient table 8.

[0050] The X-ray detector 5, more precisely the number of individual channels and thus the available spatial raw image data resolution for projection images as raw image data, and the X-ray emitter 4, in this case the size of the focal point, are designed such that an image data set reconstructed from raw image data of the raw image data resolution has a first spatial resolution in the plane of rotation, which is the image plane of the sliced ​​images of the sliced ​​image stack reconstructed as an image data set. This first spatial resolution is lower than a target resolution of the image acquisition device 1, for example 256 by 256 pixels as opposed to 512 by 512 pixels. The first spatial resolution can also be lower than the target resolution perpendicular to the plane of rotation.

[0051] Accordingly, the X-ray emitter 4 and the X-ray detector 5 are of simple design and the required data transmission rate to a control unit 9 controlling the operation of the image acquisition device 1 is low, so that a wireless data transmission means 10, for example a radio means, is used as the data transmission means of a data transmission device from the acquisition arrangement 3 to the control unit 9 between the rotating part of the gantry 2 and the stationary part of the gantry 2.

[0052] The control unit 9 includes, in addition to a storage medium 11, a recording unit 12 (which is generally known and controls the recording operation), and a reconstruction unit 13 designed to reconstruct the image data set from the raw image data, an image processing unit 14, which is designed to apply a trained resolution enhancement function, which in this case has a SwinIR architecture, to increase the resolution of the image data set to a second spatial resolution corresponding to the target resolution.

[0053] It is conceivable to apply the trained resolution enhancement function to the raw image data—in this case, the temporal sequence of projection images—thus increasing the spatial resolution of the raw image data. Alternatively, the trained resolution enhancement function can be applied to the image dataset of the first spatial resolution, which was reconstructed from the raw image data of the first spatial resolution, in order to obtain the second spatial resolution. Particularly in the first case, the trained resolution enhancement function can also take into account temporally contiguous projection images that, although they have a slightly different projection angle, nevertheless contain overlapping image content. In this way, a temporal super-resolution technique can be employed. This is also possible when applying it to the image dataset if a temporal sequence of image datasets is acquired, for example, during the monitoring of a minimally invasive procedure.In the present case, the trained resolution enhancement function is implemented on a graphics card 15 of the control unit 9, which is only indicated here, and which uses at least one GPU of the graphics card.

[0054] The image acquisition device 1 is therefore designed to be cost-effective and inexpensive in terms of its hardware components, in particular the recording arrangement 3 and the data transmission devices, for the first spatial resolution, while the actually desired target resolution is provided as the second spatial resolution by means of the trained resolution enhancement function of the image processing unit 14.

[0055] Fig. 2 shows a flowchart of a first embodiment of a method for operating the image acquisition device 1. First, in step S1, for example along a revolution or partial revolution of the recording arrangement 3 around an examination area of ​​the patient, the raw image data of the raw image data resolution are acquired as projection images of different projection angles, controlled by the recording unit 12. In step S2, the reconstruction unit 13 is used to reconstruct an image data set of the first spatial resolution from the raw image data, for example by means of filtered back projection. In step S3, the trained resolution enhancement function is then applied to the image data set of the first spatial resolution by means of the image processing unit 14 to obtain the image data set of the second spatial resolution.

[0056] Fig. 3 shows a flowchart of a preferred, second embodiment of the operating procedure. In contrast to Fig. 1, after the acquisition of the raw image data at the raw image data resolution in step S3', the trained resolution enhancement function is applied to the raw image data, here the projection images, by means of the image processing unit 14 in order to obtain a spatial resolution of the raw image data that is increased compared to the raw image data resolution. These raw image data with increased spatial resolution are then used in step S2' by the reconstruction unit 13 to reconstruct the image data set of the second spatial resolution.Applying this to projection images, particularly when using background information about the respective projection angles, allows the use of temporal super-resolution approaches, i.e., taking into account temporally contemporaneous projection images when increasing the resolution of a specific projection image. Since image processing only takes place after transmission to the control unit 9, projection images acquired later in time can also be used.The two embodiments can also be combined, for example when using two trained resolution enhancement functions, so that in partial steps the spatial resolution of the raw image data is first increased in order to reconstruct an image data set of an intermediate spatial resolution lying between the first spatial resolution and the second spatial resolution, after which the spatial resolution of the image data set is increased from the intermediate resolution to the second spatial resolution by means of a second trained resolution enhancement function.

[0057] Fig. 4 shows an embodiment of a method for designing an image acquisition device 1. In step S4, the desired target resolution is defined, for example, 512 by 512 pixels in the image plane of the cross-sectional images (rotation plane). It should be noted here that the resolution does not necessarily have to refer to a number of pixels, as described, but can of course also be described in other ways, for example, in mm and / or as lp / mm (line pairs per mm). However, computed tomography systems are usually designed for specific pixel resolutions.

[0058] In step S5, the first spatial resolution, which determines the choice of design parameters for the components of the image acquisition device, in particular the recording arrangement and the data transmission device, is then chosen to be lower than the target resolution.

[0059] In step S6, the components are then designed, in particular the design parameters are selected, such that when reconstructing an image dataset from raw image data of the imaging setup, only the first spatial resolution would be obtained. For example, the number of electronic channels of the X-ray detector, generally speaking its third spatial resolution, can be specified, as can the size of the focal point of the X-ray tube. The amount of raw image data to be transmitted also follows from these design parameters, so that the design parameters of the data transmission device, especially the wireless data transmission medium, can be selected accordingly.

[0060] In step S7, the control unit is then designed to provide the trained resolution enhancement function to the image processing unit in such a way that image data sets of the second spatial resolution, corresponding to the target resolution, can be determined. The design result can then be provided and used in a manufacturing process to produce the image acquisition device 1. Regardless of the grammatical gender of a particular term, persons of male, female, or other gender identities are included.

Claims

Patent claims 1. Medical imaging device (1) comprising: - a recording arrangement (3) for recording raw image data at a raw image data resolution and - a control device (9) with a reconstruction unit (13) for reconstructing an image data set from the raw image data, wherein the image data set, when reconstructed from the raw image data, has a first spatial resolution in at least one reference plane, characterized in that the control device (9) further comprises an image processing unit (14) for executing a trained resolution enhancement function to increase the spatial resolution to a second spatial resolution which is higher than the first spatial resolution.

2. Image acquisition device according to claim 1, characterized in that the trained resolution enhancement function is a Super Resolution function and / or is based on a SwinlR architecture and / or the image processing unit (14) is implemented at least for the part executing the resolution enhancement function on a graphics card (15) of the control device (9).

3. Image acquisition device according to claim 1 or 2, characterized in that the image processing unit (14) is configured to apply the resolution enhancement function to the image data set reconstructed by the reconstruction unit (13) and / or that the image processing unit (14) is configured to apply the resolution enhancement function to the raw image data in order to increase the spatial resolution of the image data set.

4. Image acquisition device according to claim 3, characterized in that, in the case of a temporal sequence of raw images showing at least partially overlapping image content as raw image data, the trained resolution enhancement function is designed to take into account at least one temporally adjacent raw image in the sequence when improving the resolution of a specific raw image.

5. Image acquisition device according to one of the preceding claims, characterized in that at least one design parameter of at least one component of the image processing device (14) is selected for a lower effort than for the direct obtaining of the second spatial resolution by the reconstruction unit (13).

6. Image acquisition device according to claim 5, characterized in that the design parameter relates to a third spatial resolution of a detector (5) of the recording arrangement (3) and / or, in the case of radiographic imaging, the focus size of a radiation source and / or a data transmission rate of a data transmission means (10) between the recording arrangement (3) and the control device (9).

7. Image acquisition device according to one of the preceding claims, characterized in that it is a computed tomography device which has a gantry (2) in which an X-ray source (4) and an X-ray detector (5) of the acquisition arrangement (3) are rotatably mounted in a plane of rotation, wherein the reconstruction unit (13) for the reconstruction of the image data set, in particular three-dimensional, in particular as a stack of cross-sectional images, is formed from two-dimensional projection images of different projection angles.

8. Image acquisition device according to claim 7, characterized in that the first and the second resolution relate at least to the plane of rotation, in particular as the image plane of the cross-sectional images of the cross-sectional image stack.

9. Image acquisition device according to claim 8, characterized in that the first and the second resolutions relate to all spatial directions of the image data set.

10. Computer-implemented method for operating a medical image acquisition device (1) comprising: - a recording arrangement (3) for recording raw image data at a raw image data resolution and - a control device (9) with a reconstruction unit (13) for reconstructing an image data set on the raw image data, wherein the image data set, when reconstructed from the raw image data, has a first spatial resolution in at least one reference plane, characterized in that the spatial resolution of the image data set is increased to a second spatial resolution, which is higher than the first spatial resolution, by means of a trained resolution enhancement function.

11. Computer program which, when executed on a control device (9) of a medical imaging device (1), causes the latter to perform the steps of a method according to claim 10.

12. Electronically readable data carrier on which a computer program according to claim 11 is stored.

13. Procedure for the design of a medical imaging device (1), comprising the following design steps: - Defining a spatial target resolution for image data sets acquired with the image acquisition device (1) in at least one reference plane, - Defining a first spatial resolution to be used as a design target for the image data sets, at least in the reference plane, which is lower than the target resolution, - Designing a recording arrangement (3) for recording raw image data at a raw image data resolution and / or at least one data transmission means (10) for the raw image data such that when reconstructing an image data set from raw image data of the raw image data resolution, this image data set has the first spatial resolution, - Designing a control unit (9) of the image acquisition device (1) such that it includes an image processing unit (14) for executing a trained resolution enhancement function to increase the spatial resolution to a second spatial resolution corresponding to the target resolution.

Images