Generating additional views in X-ray imaging of body parts.

Abstract

The present invention relates to X-ray imaging. To improve the X-ray imaging workflow, an image processing device is proposed, which includes an input unit, a processor (14), and an output unit. The input unit is configured to receive a first X-ray image acquired in an image acquisition. The first X-ray image has a first view of a body part of a patient. The processor (14) is configured to generate a second X-ray image having a second view of the body part of the patient based on the received first X-ray image using a pre-trained machine learning model. The second view is different from the first view. The output unit is configured to output the generated second X-ray image.

Description

[Technical field]

[0001] The present invention relates to medical imaging, and in particular to an image processing apparatus, an imaging processing method, a computer program element and a computer readable data carrier. [Background technology]

[0002] X-rays are used in medical imaging. Projection radiographs are useful in detecting pathologies of the skeletal system, as well as detecting some disease processes in soft tissues. Different views (also known as projections) of a body part can be obtained by changing the relative orientation of the body and the direction of the X-ray beam.

[0003] For example, chest X-ray is the most used type of medical imaging study. During the examination, different views may be acquired to facilitate diagnosis. Depending on the indication and clinical guidelines, this may include posteroanterior (PA) or anterior-posterior (AP), lateral (LAT) and other views. However, although some studies suggest the value of LAT images, they are not always available when needed. Therefore, early indications may suggest only acquiring PA images for various reasons, e.g., due to dosage concerns or due to the presumed condition of the patient. Another reason is that LAT images may also be difficult to acquire (e.g., in bedside imaging).

[0004] Thus, in some cases, a new exam must be scheduled for the patient to obtain additional views of the body part for diagnostic purposes later in the workflow, which can cause significant organizational overhead and a negative patient experience. Summary of the Invention [Problem to be solved by the invention]

[0005] Therefore, there may be a need to improve x-ray imaging workflow. [Means for solving the problem]

[0006] The invention is defined by the independent claims. The dependent claims define advantageous embodiments.

[0007] According to a first aspect of the present invention, there is provided an image processing device. The image processing device comprises an input unit, a processor, and an output unit. The input unit is configured to receive a first X-ray image acquired in an image acquisition. The first X-ray image has a first view of a body part of a patient. The processor is configured to generate a second X-ray image having a second view of the body part of the patient using a pre-trained machine learning model based on the received first X-ray image. The second view is different from the first view. The output unit is configured to output the generated second X-ray image.

[0008] The machine learning based approach described herein can generate additional synthetic views from acquired images. For example, a synthetic lateral view of the chest may be generated from an acquired X-ray image with a PA view. Although such an approach may not be able to predict pathologies obscured in the original view, the synthetic additional view may provide the most likely representation based on a large amount of clinical data. Thus, a synthetic X-ray image with different views of a body part may allow a radiologist to further evaluate pathologies that may be present in the acquired image without performing a new examination for the patient. This can reduce the dosage without compromising (or even below) diagnosis and treatment.

[0009] The first X-ray image is the X-ray image acquired during image acquisition. The first X-ray image may also be referred to as the actual X-ray image or the acquired X-ray image.

[0010] The second X-ray image is an X-ray image that is generated from the first X-ray image. The second X-ray image is sometimes called a synthetic X-ray image or a generated X-ray image.

[0011] Examples of body parts include, but are not limited to, joints, neck, spine, chest, limbs, and other parts of the body.

[0012] As used herein, a "machine learning model" may refer to a statistical method that allows a machine to "learn" a task from data without being explicitly programmed, instead relying on patterns in the data. For example, the machine learning model may be a deep learning model. Deep learning is a subset of machine learning loosely modeled on the neural pathways of the human brain. Deep refers to multiple layers between the input and output layers. In deep learning, the algorithm automatically learns which features are useful. For a general introduction to machine learning and corresponding software frameworks, see "machine Learning and Deep Learning frameworks and libraries for larges a survey (Artificial Intelligence Review; Giang Nguyen et al., June 2019, Volume 52, Issue 1, pp 77-124)". The machine learning model may be a machine learning generative model for reproducing a mapping between a first X-ray image having a first view of a body part and a second X-ray image having a second view of the body part. The machine learning generative model may be used, for example, to generate additional views in the formation of synthetic data from the first X-ray image (e.g., PA data).

[0013] According to an embodiment of the invention, the input is further configured to receive non-image patient data of the patient, and the pre-trained machine learning model is further configured to apply (use) the received non-image patient data to generate a second X-ray image.

[0014] In other words, in addition to the first x-ray image, non-image patient data can be included in the training and inference process to improve prediction.

[0015] Examples of non-imaging patient data include, but are not limited to, the patient's age, the patient's gender, the presence of an implant, medication records associated with the patient, or any combination thereof.

[0016] According to an embodiment of the invention, the input is further configured to receive system data of an X-ray imaging device for acquiring a first X-ray image of the patient, and the pre-trained machine learning model is further configured to apply (use) the received system data to generate a second X-ray image.

[0017] In addition to the first X-ray image data, system data may also be included in the training and generation process to improve predictions.

[0018] The system data may include, for example, view position, one or more acquisition parameters (such as kVp), and any other system data.

[0019] According to one embodiment of the present invention, the pre-trained machine learning model includes an encoder-decoder architecture.

[0020] An exemplary encoder-decoder architecture is shown in FIG.

[0021] According to an embodiment of the present invention, the pre-trained machine learning model comprises a generator component and a discriminator component. The generator component comprises an encoder-decoder architecture configured to map a first X-ray image to a second X-ray image. The discriminator component comprises a discriminator trained using concatenated image pairs to discriminate between a first X-ray image pair including a first and a second X-ray image acquired by an X-ray imaging device, and a second X-ray image pair including the first X-ray image acquired by the X-ray imaging device and a second X-ray image generated by an image processor.

[0022] This is explained in more detail below, particularly with reference to the example shown in FIG.

[0023] According to an embodiment of the invention, the processor is further configured to detect the presence of one or more pathologies in the second X-ray image.

[0024] According to an embodiment of the invention, the processor is further configured to provide a probability score of the one or more detected pathologies in the second X-ray image.

[0025] Therefore, detection of pathology in additional views can facilitate the diagnostic process. For example, LAT images can be beneficial for pneumonia detection and other pathologies.

[0026] According to one embodiment of the present invention, the processor is configured to detect the presence of one or more pathologies in the first X-ray image, provide a probability score of the one or more detected pathologies in the first X-ray image, and determine whether to generate a second X-ray image having a second view of the body part to further evaluate the one or more detected pathologies based on the probability score.

[0027] In other words, a second X-ray image may be generated based on the probability of the presence of one or more suspicious pathologies in the first X-ray image, allowing the radiologist to further evaluate the suspicious pathologies from another perspective.

[0028] According to a second aspect of the present invention there is provided an X-ray imaging system comprising an X-ray imaging device for acquiring an X-ray image of a patient and an image processing device according to the first aspect and any associated examples.

[0029] According to a third aspect of the present invention, there is provided an image processing method.

[0030] The image processing method is receiving a first x-ray image obtained in an image acquisition, the first x-ray image having a first view of a body part of a patient; generating a second x-ray image having a second view of the patient's body part using a trained machine learning model based on the received first x-ray image, the second view being different from the first view; outputting the generated second X-ray image; has.

[0031] According to an embodiment of the invention, the method further comprises detecting the presence of one or more pathologies in the second X-ray image.

[0032] According to one embodiment of the invention, the method further comprises providing a probability score of one or more detected pathologies in the second X-ray image.

[0033] According to one embodiment of the present invention, the method further includes detecting the presence of one or more pathologies in the first X-ray image, providing a probability score of the one or more detected pathologies in the first X-ray image, and determining whether to generate a second X-ray image having a second view of the body part to further evaluate the one or more detected pathologies based on the probability score.

[0034] The image processing method may be at least partially computer-implemented, or may be implemented in software or hardware, or in software and hardware. Furthermore, the method may be performed by computer program instructions executed on a means providing data processing functionality. The data processing means may be a distributed computer system, or may be any suitable computing means such as an electronic control module. The data processing means or computer may each comprise one or more processors, memories, data interfaces, etc.

[0035] According to a fourth aspect of the present invention there is provided a computer program product comprising instructions which, when executed by one or more processors, cause the one or more processors to perform the method steps of the third aspect and any associated examples.

[0036] According to a further aspect of the invention there is provided a computer readable data carrier having a computer program product stored thereon.

[0037] It should be noted that the above embodiments may be combined with each other regardless of the associated aspects, thus methods may be combined with structural features, and similarly, devices and systems may be combined with features discussed above with respect to methods.

[0038] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter.

[0039] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. [Brief description of the drawings]

[0040] [Figure 1] 1 shows a flow chart illustrating an exemplary image processing method. [Diagram 2] 1A-1C are schematic diagrams illustrating exemplary PA and LAT chest X-ray images. [Diagram 3] An example of a machine learning model is shown below. [Figure 4] Further examples of machine learning models are given below. [Diagram 5] 1 shows a block diagram of an exemplary image processing device. [Figure 6] 1 illustrates an example of a medical imaging system. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0041] The drawings are merely schematic representations and serve only to illustrate embodiments of the invention, and identical or similar elements are in principle provided with the same reference signs.

[0042] The approach is described below with respect to chest x-ray images for illustrative purposes. However, those skilled in the art will appreciate that the methods and apparatus described above and below can be adapted to other body parts, including but not limited to joints, neck, spine, extremities, and other parts of the body. Accordingly, the examples described below are provided without any loss of generality to, and without any limitation of, the claimed invention.

[0043] FIG. 1 shows a flow chart illustrating an exemplary image processing method 100. The image processing method 100 may be implemented as a device, module, or associated component in configurable logic, such as a programmable logic array (PLA), a field programmable gate array (FPGA), a complex programmable logic device (CPLD), in fixed function hardware logic using circuit technologies, such as an application specific integrated circuit (ASIC), complementary metal oxide semiconductor (CMOS) or transistor transistor logic (TTL) technology, in a set of logic instructions stored in a non-transitory machine-readable or computer-readable storage medium, such as a random access memory (RAM), a read only memory (ROM), a programmable ROM (PROM), firmware, flash memory, etc. For example, computer program code for performing the operations shown in the method 100 may be written in any combination of one or more programming languages, including object-oriented programming languages ​​and conventional procedural programming languages, such as JAVA, SMALLTALK, C++, Python, or "C" programming languages ​​or similar programming languages. For example, the exemplary image processing method may be implemented in an image processing device 10 shown in FIG. 5.

[0044] At block 110, a first x-ray image having a first view of a patient's body part is received. The first x-ray image may be received from an x-ray imaging device or from a database such as a Picture Archiving and Communication System (PACS). Examples of the patient's body part include, but are not limited to, a joint, neck, spine, limb, or other part of the body. For purposes of explanation, the following approach is described with reference to the patient's chest.

[0045] For a chest x-ray image, the first view of the chest may be selected from a PA view, an AP view, a LAT view, a supine view, a lordotic view, an expiratory view, an inspiratory view, and an oblique view. The first x-ray image may be a two-dimensional image including image pixel data.

[0046] At block 120, a second x-ray image having a second view of the patient's body part is generated using the trained machine learning model based on the received first x-ray image, the second view being different from the first view.

[0047] In one example, if the first X-ray image has a PA view of the chest, the second X-ray image may have, for example, a LAT view, a supine view, a lordotic view, an expiratory view, or an oblique view. For example, FIG. 2 shows a schematic of exemplary PA and LAT chest X-ray images.

[0048] In another example, if the first x-ray image has a LAT view, the second x-ray image may have an AP view, a PA view, a supine view, a lordotic view, an expiratory view, or an oblique view.

[0049] The machine learning model may be a machine learning generative model for reproducing a relationship between a first X-ray image having a first view of a body part and a second X-ray image having a second view of the body part. The machine learning generative model may be used, for example, to generate additional views in the formation of a composite X-ray image from the first X-ray image (e.g., PA data). In one example, the machine learning generative model may include a generative model of image synthesis using a probabilistic framework. In another example, the machine learning generative model may include a deep encoder-decoder network architecture that can synthesize an X-ray image having a view of a body part into an X-ray image having a different view of the body part.

[0050] FIG. 3 illustrates an example of a machine learning model with an encoder-decoder network architecture. The architecture relies on finding a mapping between an input image and a desired image. The input image is an acquired x-ray image having a first view of a body part, and the desired image is a synthetic x-ray image having a second view of the body part. In the example of FIG. 3, an exemplary encoder-decoder network architecture is used to perform image-to-image mapping for PA and LAT chest x-ray images.

[0051] The objective of the encoder-decoder network architecture is to find relations that relate representations of X-ray images with different views of the same body part. In particular, the input layer L1 receives a first X-ray image with a first view of the body part, such as the X-ray image with a PA view of the chest shown in FIG. 2. The layers L2 and L3 then aim to encode the input. The subsequent layers L4 and L5 essentially decode the information coming from the previous layers and provide at the output the desired second X-ray image with a second view of the body part, such as the X-ray image with a LAT view of the chest shown in FIG. 2. The weights per layer may be pre-trained based on a training dataset that includes multiple pairs of acquired X-ray images with the first and second views of the body part. The pre-training may rely on unsupervised learning.

[0052] In the inference phase, an X-ray image with a first view of the body part is fed to layer L1 and is fed to the entire network. The activations on layer L5 are the output of the network and represent the actual synthetic X-ray image with a second view of the body part.

[0053] Optionally, as shown in Figure 3, in addition to the X-ray image, additional data may be included in the training and inference process, for example, by concatenating this information with the output of a convolution channel in the encoder or decoder channel. The additional data may include non-image patient data, including, but not limited to, the patient's age, the patient's gender, the presence of an implant, medication records associated with the patient, or any combination thereof. Additionally or alternatively, the additional data may include system data of the X-ray imaging system for acquiring the first X-ray image, which may include, but is not limited to, view position and peak kilovoltage (kVp).

[0054] FIG. 4 shows a further example of a machine learning model with an additional classifier that is used to ensure the generation of realistic synthetic X-ray images.

[0055] That is, the machine learning model shown in FIG. 4 comprises a generator and a classifier. The generator component comprises an encoder-decoder architecture configured to map a first X-ray image to a second X-ray image. An exemplary encoder-decoder architecture is shown in FIG. 3. The classifier component comprises a classifier trained using concatenated image pairs to classify both. Each concatenated image pair includes a first X-ray image pair including a first and a second X-ray image acquired by an X-ray imaging device, and a second X-ray image pair including the first X-ray image acquired by an X-ray imaging device and a second X-ray image generated by an image processing device.

[0056] The classifier can be pre-trained on concatenated image pairs such as (real PA, real LAT) and (real PA, synthetic LAT) to try to distinguish between them and derive the generation of more realistic synthetic X-ray images with different views, where real PA and real LAT represent acquired X-ray images with PA and LAT views, and synthetic LAT represents an X-ray image with LAT view generated by an image processor.

[0057] Similarly, additional data, such as non-imaging patient data and system data, can be included in the training and prediction process, for example, by concatenating this information with the output of a convolutional channel in the encoder, decoder, or classifier.

[0058] 1, in block 130, the generated second x-ray image is provided to an electronic display at the PACS / diagnostic workstation, such as a display on a system console. Alternatively or additionally, the generated second x-ray image may be provided to a storage medium.

[0059] In some examples, the presence of one or more pathologies can be detected in a second X-ray image, i.e., a synthetic X-ray image with a different view. For example, an AI model such as a deep convolutional neural network (CNN) architecture trained on large population data can be used to detect subtle signs of a particular pathology on the synthetic X-ray image. The AI-based detection can be performed on the synthetic virtual projection. The original projection in the first X-ray image can also be used as an additional input.

[0060] There may be a dedicated selection of pathologies, which are typically diagnosed on a different projection image than the one at hand, for example on a lateral projection. The AI ​​model can provide a probability score for one or more of such pathologies in the first X-ray image. High probability scores can be used as a recommender system to create another projection image, i.e., a second image, from the same patient and evaluate the pathology.

[0061] 5 shows an exemplary image processing apparatus 10. The image processing apparatus 10 comprises an input 12, one or more processors 14, and an output 16.

[0062] In general, the image processing device 10 may comprise various physical and / or logical components for communicating and manipulating information, which may be implemented as hardware components (e.g., computing devices, processors, logic devices), executable computer program instructions (e.g., firmware, software) executed by the various hardware components, or any combination thereof, as desired for a given set of design parameters or performance constraints. While Fig. 5 may show a limited number of components as an example, it may be understood that for a given implementation, a greater or lesser number of components may be used.

[0063] In some implementations, the imaging device 10 may be embodied as or in a device or apparatus, such as a server, a workstation, or a mobile device. The imaging device 10 may include one or more microprocessors or computers that execute appropriate software. The processor 14 of the imaging device 10 may be embodied by one or more of these microprocessors or computers. The software may be downloaded and / or stored in a corresponding memory, for example, a volatile memory such as RAM, or a non-volatile memory such as Flash. The software may comprise instructions that configure one or more processors to perform the functions described herein.

[0064] The image processing device 10 may be realized using a processor, may be realized without a processor, or may be realized as a combination of dedicated hardware for performing some functions and a processor (e.g., one or more programmed microprocessors and associated circuits) for performing other functions. For example, the functional units of the image processing device 10, such as the input unit 12, one or more processors 14, and the output unit 16, may be implemented in a device or apparatus in the form of programmable logic, for example, as a Field-Programmable Gate Array. In general, each functional unit of an apparatus may be implemented in the form of a circuit.

[0065] In some implementations, the image processing device 10 may also be implemented in a distributed manner. For example, some or all of the units of the image processing device 10 may be arranged as separate modules in a distributed architecture and connected to an appropriate communication network, such as a 3rd Generation Partnership Project (3GPP) network, a Long Term Evolution (LTE) network, the Internet, a LAN (Local Area Network), a wireless LAN (Local Area Network), a WAN (Wide Area Network), etc.

[0066] The processor 14 may execute instructions to perform the image processing methods described herein, which are described in detail with respect to the embodiment shown in FIG.

[0067] The input 12 and output 14 may include hardware and / or software to enable imaging device 10 to receive data input and communicate with other devices and / or networks. The input 12 may receive data input via a wired connection or via a wireless connection. The output 16 may also provide cellular telephone and / or other data communications for imaging device 10.

[0068] 6 shows a schematic and exemplary example of an X-ray imaging system 200. In this example, the exemplary X-ray imaging system is a chest X-ray imaging system 200. The chest X-ray imaging system 200 includes an image processing device 10, an X-ray imaging device 210, and a system console 220.

[0069] The X-ray imaging system 210 includes an X-ray source 212 and an X-ray detector 214. The X-ray detector 214 is spaced from the X-ray source 212 to accommodate the patient PAT being imaged.

[0070] In general, during image acquisition, a collimated x-ray beam (indicated by arrow P) is emitted from the x-ray source 212, passes through the patient PAT in the region of interest (ROI), experiences attenuation due to interaction with the material therein, and then the attenuated beam strikes the surface of the x-ray detector 214. The density of the organic material that makes up the ROI determines the level of attenuation; for example, the rib cage and lung tissue in a chest x-ray imaging study. High density materials (such as bone) cause higher attenuation than less dense materials (such as lung tissue). The registered digital values ​​for the x-rays are then integrated into an array of digital values ​​that form an x-ray projection image for a given acquisition time and projection direction.

[0071] The overall operation of the X-ray imaging system 210 may be controlled by an operator from a system console 220. The system console 220 may be coupled to a screen or monitor 230 on which acquired X-ray images or imager settings may be viewed or reviewed. An operator, such as a medical laboratory technician, may control the image acquisition that is performed via the system console 220, for example, by releasing individual X-ray exposures by actuating a joystick or pedal or other suitable input means coupled to the system console 220.

[0072] 6, the patient PAT is standing facing a flat surface behind the X-ray detector 214. According to another example (not shown), the X-ray imaging system 210 is of a C-arm type and the patient PAT is actually lying on an examination table instead of standing.

[0073] As mentioned above, the image processing device 10 may be any computing device, including desktop and laptop computers, smartphones, tablets, etc. The image processing device 10 may be a general-purpose device or may have dedicated units of equipment suitable for providing the functionality described herein. In the example of FIG. 6, the components of the image processing device 10 are shown integrated into one unit. However, in alternative examples, some or all components may be configured as separate modules in a distributed architecture and connected in a suitable communication network. The image processing device 10 and its components may be configured as dedicated FPGAs or as hardwired standalone chips such as the image processing device shown in FIG. 6. In some examples (not shown), the image processing device 10 or some of its components may reside in a console 230 that operates as software routines.

[0074] During operation, x-ray images acquired by the x-ray imager 210 are provided to the image processor 10. The acquired x-ray images may include a PA view of the patient PAT chest. The image processor 10 may generate additional views (e.g., LAT images) from the PA data. Results may be displayed directly on the system console 220 or on a PACS / diagnostic workstation (not shown) to allow a radiologist to evaluate pathology in different views.

[0075] In another exemplary embodiment of the invention, a computer program or a computer program element is provided, characterized in that it is adapted to execute, on a suitable system, the method steps of the method according to one of the previous embodiments.

[0076] Thus, a computer program element may be stored on a computer which may be part of an embodiment of the present invention. This computer may be adapted to perform or trigger the execution of the steps of the above-mentioned method. Furthermore, it may be adapted to operate the components of the above-mentioned apparatus. The computer may be adapted to operate automatically and / or to execute the instructions of a user. The computer program may be loaded into the working memory of a data processor. The data processor may thus be equipped to perform the method of the present invention.

[0077] This exemplary embodiment of the invention encompasses both computer programs that use the invention from the beginning, and computer programs that, through updates, turn existing programs into programs that use the invention.

[0078] Moreover, the computer program element may provide all the steps required to fulfill the steps of the exemplary embodiments of the procedures described above.

[0079] According to a further exemplary embodiment of the present invention, a computer readable medium, such as a CDROM, is presented, having stored thereon a computer program element, the computer program element being as described in the previous section.

[0080] The computer program may be stored and / or distributed on a suitable medium, such as an optical storage medium or a solid-state medium, provided together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless communication systems.

[0081] However, the computer program may also be presented via a network such as the World Wide Web and downloaded from such a network into the working memory of a data processor. According to a further exemplary embodiment of the invention, a medium for making available a computer program element for downloading is provided, this computer program element being configured to perform a method according to one of the aforementioned embodiments of the invention.

[0082] It should be noted that the embodiments of the present invention are described with reference to different subject matters. In particular, some embodiments are described with reference to method type claims, and other embodiments are described with reference to device type claims. However, a person skilled in the art will gather from the above and following description that, unless otherwise notified, any combination of features belonging to one type of subject matter, as well as any combination between features related to different subject matters, are considered to be disclosed in this application. However, all features can be combined to provide synergistic effects that are more than the simple sum of the features.

[0083] While the invention has been illustrated and described in detail in the drawings and the foregoing description, such illustration and description are to be considered as illustrative or exemplary and not restrictive. The invention is not limited to the disclosed embodiments. Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the dependent claims.

[0084] In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. Measures recited in mutually different dependent claims may be combined to advantage. Any reference signs in the claims should not be interpreted as limiting the scope.

Claims

1. An image processing device, an input configured to receive a first X-ray image acquired in an image acquisition, the first X-ray image having a first view of a body-part of a patient; 1. A processor, comprising: Detecting the presence of one or more pathologies in the first x-ray image; providing a probability score for the detected one or more pathologies in the first x-ray image; determining, based on the probability score, whether to generate a second X-ray image having a second view of the body part based on the received first X-ray image using a pre-trained machine learning model to further evaluate the detected one or more pathologies; wherein the second view is different from the first view; and an output configured to provide the generated second X-ray image; An image processing device comprising:

2. the input is further configured to receive non-image patient data for the patient; the pre-trained machine learning model is further configured to apply the received non-image patient data to generate the second X-ray image. The image processing device according to claim 1 .

3. the input unit is further configured to receive system data of an X-ray imaging device for acquiring a first X-ray image of the patient; the pre-trained machine learning model is further configured to apply the received system data to generate the second X-ray image.

3. The image processing device according to claim 1 or 2.

4. The image processing device according to claim 1 or 2, wherein the pre-trained machine learning model comprises an encoder-decoder architecture.

5. the pre-trained machine learning model has a generator component and a discriminator component; a generator component having the encoder-decoder architecture configured to map the first X-ray image to the second X-ray image; the classifier component includes a classifier trained on concatenated image pairs to distinguish between: a first X-ray image pair having a first and a second X-ray image acquired by an X-ray imaging device; and a second X-ray image pair having a first X-ray image acquired by the X-ray imaging device and a second X-ray image generated by the image processing device. The image processing device according to claim 4.

6. The image processing device of claim 1 or 2, wherein the processor is further configured to detect the presence of one or more pathologies in the second X-ray image.

7. The image processing apparatus of claim 6 , wherein the processor is further configured to provide a probability score of the detected one or more pathologies in the second x-ray image.

8. An X-ray imaging system, comprising: an x-ray imaging device for acquiring x-ray images of a patient; The image processing device according to claim 1 or 2; An X-ray imaging system comprising:

9. 1. A method for image processing, the method comprising: receiving a first x-ray image obtained in an image acquisition, the first x-ray image having a first view of a body-part of a patient; detecting the presence of one or more pathologies in the first x-ray image; providing a probability score for the detected one or more pathologies in the first x-ray image; determining, based on the probability score, whether to generate a second X-ray image having a second view of the body part based on the received first X-ray image using a pre-trained machine learning model to further evaluate the detected one or more pathologies, wherein the second view is different from the first view; providing the generated second X-ray image; An image processing method comprising:

10. receiving non-image patient data for the patient; 10. The image processing method of claim 9, wherein the pre-trained machine learning model is further configured to generate the second x-ray image using the received non-image patient data.

11. receiving system data of an X-ray imaging device for acquiring the first X-ray image of the patient; 11. The image processing method of claim 9 or 10, wherein the pre-trained machine learning model is further configured to generate the second X-ray image using the received system data.

12. detecting the presence of one or more pathologies in the second x-ray image.

11. The image processing method according to claim 9 or 10, further comprising:

13. providing a probability score for one or more detected pathologies in the second x-ray image. The image processing method of claim 12 further comprising:

14. A computer program having instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the method according to claim 9 or 10.

15. A computer readable data carrier storing a computer program according to claim 14.

Images