Providing attitude information for x-ray projection images

Abstract

A computer-implemented method of providing attitude information for an X-ray projection image is provided. The method comprises segmenting the X-ray projection image (120) to identify a plurality of projection sub-regions (1501. M) of the anatomical structure, generating a pose metric for the X-ray projection image (120) based on relative dimensions of two or more of the projection sub-regions (1501. M) in the segmented X-ray projection image (120), and outputting the attitude metric to provide attitude information for the X-ray projection image (120).

Description

Technical Field

[0001] The present disclosure relates to providing pose information for X-ray projection images. A computer-implemented method, computer program product, and system are disclosed. Background Art

[0002] A projection X-ray imaging system includes an X-ray source and an X-ray detector. The X-ray source and the X-ray detector are separated by an inspection region. Anatomical structures such as an ankle, a leg or another part of a subject's body can be arranged in the inspection region so as to generate an X-ray projection image of the anatomical structure. The X-ray projection image is collected by a projection X-ray imaging system having a defined posture with respect to the anatomical structure. The X-ray projection image is collected using a single posture of the X-ray imaging system with respect to the anatomical structure, and therefore it is important that the posture used provides the desired information in the resulting X-ray projection image. For example, the diagnosis of a fractured ankle bone may require imaging the bone using a posture in which the fracture is not obscured by other bones in the ankle. If an inappropriate posture is used, repeated images may need to be collected.

[0003] Positioning of the anatomical structure with respect to the projection X-ray imaging system is typically performed manually and according to a protocol for the anatomical structure. The radiotherapist locates the anatomical structure by eye and generates an initial image. If this initial image is not acceptable, the radiotherapist repositions the anatomical structure based on their experience and generates another image. This process may be repeated multiple times until an acceptable image is obtained, sometimes with only minor improvements in the resulting image. As a result, this method increases the amount of X-ray dose delivered to the subject and also hinders workflow.

[0004] Document US2019 / 183438A1 describes a method for ensuring correct positioning for radiographic recording. The method comprises: providing an inspection request for a body region; pre-positioning the body region in a radiographic system for radiographic recording; pre-positioning a recording unit of the radiographic system and at least one of an image detector of the radiographic system for radiographic recording; generating a positioning record of the body region via the radiographic system, the radiographic system switching to a fluoroscopic mode, and the positioning record being a fluoroscopic record; generating positioning information based on the positioning record; and outputting the positioning information.

[0005] Another document WO2020 / 038917A1 relates to determining the imaging direction based on 2D projection images. Based on deep neural networks and possibly on active shape model methods, the complete contour of the anatomy can be determined and classified. Based on certain features, the algorithm can evaluate whether the viewing angle of the C-arm is sufficient. In another step, the algorithm can estimate how far the current viewing angle is from the desired viewing angle, and can provide guidance on how to adjust the position of the C-arm to achieve the desired viewing angle.

[0006] However, there remains a need for improved positioning of anatomical structures with respect to projection X-ray imaging systems. Summary of the invention

[0007] According to one aspect of the present disclosure, a computer-implemented method for providing posture information for an X-ray projection image is provided. The method includes: receiving X-ray projection data, the X-ray projection data including an X-ray projection image representing an anatomical structure, the X-ray projection data being acquired by a projection X-ray imaging system having a corresponding posture with respect to the anatomical structure; segmenting the X-ray projection image to identify a plurality of projection sub-regions of the anatomical structure; generating a posture metric for the X-ray projection image based on relative sizes of two or more projection sub-regions in the segmented X-ray projection image; and outputting the posture metric to provide posture information for the X-ray projection image.

[0008] In the above method, a posture metric is generated for an X-ray projection image. The posture metric is generated based on the relative sizes of two or more projection sub-regions in a segmented X-ray projection image. The inventors have observed that the relative size serves as a reliable metric for the posture of the projection X-ray imaging system with respect to the anatomical structure. The operator can use the posture metric for various purposes, such as, for example, determining the posture of the projection X-ray imaging system with respect to the anatomical structure, or evaluating the suitability of the posture for acquiring X-ray projection images, or determining how to adjust the projection X-ray imaging system posture in order to acquire improved X-ray projection images of the anatomical structure. Therefore, the method helps to reduce the amount of X-ray dose delivered to the object by reducing the number of repeated X-ray image acquisitions. The method also helps to improve the workflow.

[0009] Other aspects, features and advantages of the present disclosure will become apparent from the following description of examples with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 is an example of an X-ray projection image 120 representing an anatomical structure according to some aspects of the present disclosure.

[0011] Figure 2 is a flow chart illustrating an example of a method of providing pose information for an X-ray projection image according to some aspects of the present disclosure.

[0012] Figure 3 is a schematic diagram illustrating an example of a system 200 for providing pose information for an X-ray projection image according to some aspects of the present disclosure.

[0013] Figure 4 is a schematic diagram illustrating an example of a projection X-ray imaging system 140 having a pose P with respect to an anatomical structure 130 according to some aspects of the present disclosure.

[0014] Figure 5 is an example of a segmented X-ray projection image 120 according to some aspects of the present disclosure and includes a projection sub-region 150 of an anatomical structure. 1..m .

[0015] Figure 6 A segmented X-ray projection image 120 is shown according to some aspects of the present disclosure. 1..3 Three examples of projection sub-regions 150 of anatomical structures are shown in FIG. 1..m and have been derived from different poses P of the projection X-ray imaging system with respect to the anatomical structure. 1 ( Figure 6 a) P 2 ( Figure 6 b) and P 3 ( Figure 6 c) Collection.

[0016] Figure 7 is an example of a segmented X-ray projection image 120 according to some aspects of the present disclosure and includes a projection sub-region 150 of an anatomical structure. 1..m Examples and multiple anatomical landmarks 160 1..n Example of location.

[0017] Figure 8 is a schematic diagram showing an example of a method for providing posture information for an X-ray projection image according to some aspects of the present disclosure, and wherein X-ray projection data 110 including the X-ray projection image is input into a first neural network NN1 for segmenting the X-ray projection image, and wherein segmented image data representing the segmented image is input into a second neural network NN2 for generating a posture metric for the X-ray projection image. DETAILED DESCRIPTION

[0018] Various examples of the present disclosure are provided with reference to the following description and accompanying drawings. In this specification, for the purpose of explanation, a lot of specific details of certain examples are set forth. References to "examples," "implementations," or similar language in the specification mean that the features, structures, or characteristics described in conjunction with the examples are included in at least that example. It should also be understood that features described with respect to one example may also be used in another example, and for the sake of brevity, all features are not necessarily repeated in each example. For example, features described with respect to computer-implemented methods may be implemented in systems and computer program products in a corresponding manner.

[0019] In the following description, reference is made to an example of a method in which a projection X-ray imaging system is used to acquire an X-ray projection image representing an anatomical structure. As some examples, the projection X-ray imaging system may be a DigitalDiagnost C90, or a Philips Azurion 7, or a MobileDiagnost M50 mobile digital X-ray system, all of which are sold by Philips Healthcare, Best, Netherlands. However, it should be understood that these are used only as examples, and the projection X-ray imaging system may generally be provided by any type of projection X-ray imaging system.

[0020] In some examples, an arrangement of an X-ray imaging system is described in which an X-ray source of a projection X-ray imaging system is mounted to a ceiling via a rack, and a corresponding X-ray detector is mounted to a bracket and maintained in a vertical position. This type of arrangement can be used by the above-mentioned DigitalDiagnost C90 imaging system. However, it should be understood that this arrangement is used only as an example, and the X-ray source and X-ray detector of the projection X-ray imaging system can be arranged in a different manner instead. For example, the X-ray source can be mounted to a bracket or an articulated arm, and the X-ray detector can be mounted to a ceiling or an articulated arm and maintained in different positions. An arrangement can be used alternatively in which the X-ray source and the X-ray detector are mounted to a common support structure. This arrangement is used in the above-mentioned Philips Azurion7 projection X-ray imaging system. In this example, the support structure is a so-called C-arm. Other arrangements can be used alternatively in which the X-ray source and the X-ray detector are mounted to a common support structure having a different shape from the C-arm, including an arrangement in which the support structure is provided by a so-called "O-arm". Other arrangements may alternatively be used in which the X-ray source and X-ray detector are mounted or supported in different ways, including portable X-ray imaging systems such as the MobileDiagnost M50 mobile digital X-ray system described above.

[0021] In the following description, reference is made to examples of methods involving acquisition of X-ray projection images representing anatomical structures. In some examples, the anatomical structure is an ankle. However, it should be understood that the ankle is used only as an example, and in general the methods disclosed herein can be used to acquire X-ray projection images representing any anatomical structure within a subject. It should also be understood that although the following description refers to various examples of methods in which pose metrics are generated for a single X-ray projection image, these methods can be similarly used to generate pose metrics for each of a plurality of images. The method can be used, for example, to provide pose metrics for a time series of images, and wherein the pose metrics change in response to changes in the pose of the projection X-ray imaging system during the sequence.

[0022] In the following description, reference is made to various methods implemented by a computer (i.e., by a processor). Note that the computer-implemented methods disclosed herein may be provided as a non-transitory computer-readable storage medium including computer-readable instructions stored thereon, which, when executed by at least one processor, cause the at least one processor to perform the method. In other words, the computer-implemented method may be implemented as a computer program product. The computer program product can be provided by dedicated hardware or hardware capable of running software associated with appropriate software. When provided by a processor, the functionality of the method features can be provided by a single dedicated processor, or by a single shared processor, or by a plurality of separate processors, some of which can be shared. The explicit use of the term "processor" or "controller" should not be interpreted as exclusively referring to hardware capable of running software, and can implicitly include, but are not limited to, digital signal processor "DSP" hardware, read-only memory "ROM" for storing software, random access memory "RAM", non-volatile storage devices, etc. In addition, the examples of the present disclosure can take the form of a computer program product accessible from a computer-usable storage medium or a computer-readable storage medium, which provides program code used by or in conjunction with a computer or any instruction execution system. For the purposes of this specification, a computer usable storage medium or computer readable storage medium can be any apparatus that can contain, store, communicate, propagate or transport a program for use by or in conjunction with an instruction execution system, apparatus or device. The medium can be an electronic, magnetic, optical, electromagnetic, infrared or semiconductor system or device or a propagation medium. Examples of computer readable media include semiconductor or solid-state memory, magnetic tape, removable computer disk, random access memory "RAM", read-only memory "ROM", rigid magnetic disks and optical disks. Current examples of optical disks include compact disk read-only memory "CD-ROM", compact disk read / write "CD-R / W", Blu-Ray TM and DVD.

[0023] As mentioned above, there remains a need for improved positioning of anatomical structures with respect to projection X-ray imaging systems.

[0024] Figure 1 is an example of an X-ray projection image 120 representing an anatomical structure according to some aspects of the present disclosure. For example, during a clinical investigation for a bone fracture in the ankle, a radiologist may acquire a bone fracture such as Figure 1 However, as mentioned above, the disadvantage of such projection X-ray images is that they need to be acquired using a projection X-ray imaging system that has a suitable posture with respect to the anatomical structure. This is the result of generating the projection X-ray image by projecting the attenuation caused by the three-dimensional object onto the surface of the two-dimensional X-ray detector. Figure 1 The projection X-ray imaging system of the image shown has an improper posture with respect to the ankle, and the fracture may be obscured by other bones in the ankle. This may result in the need to acquire another image, which increases the amount of X-ray dose delivered to the subject and also hinders the workflow.

[0025] Figure 2 is a flow chart illustrating an example of a method of providing pose information for an X-ray projection image according to some aspects of the present disclosure. Figure 3 is a schematic diagram illustrating an example of a system 200 for providing pose information for an X-ray projection image according to some aspects of the present disclosure. The system 200 includes one or more processors 210. Figure 2 The operations described in the method shown can also be performed by Figure 3 Similarly, the operations described with respect to the one or more processors 210 of the system 200 may also be performed in reference to Figure 2 The method described is executed. Figure 2 , a computer-implemented method for providing posture information for an X-ray projection image comprises:

[0026] Receiving S110 X-ray projection data 110 , the X-ray projection data comprising an X-ray projection image 120 representing an anatomical structure 130 , the X-ray projection data being acquired by a projection X-ray imaging system 140 having a corresponding pose P with respect to the anatomical structure 130 ;

[0027] Segmenting S120 the X-ray projection image 120 to identify a plurality of projection sub-regions 150 of the anatomical structure 130 1..m ;

[0028] Based on two or more projection sub-regions 150 in the segmented X-ray projection image 120 1..m to generate S130 a pose metric for the X-ray projection image 120 ; and

[0029] Output S140 the posture metric to provide posture information for the X-ray projection image 120

[0030] In the above method, a posture metric is generated for an X-ray projection image. The posture metric is generated based on the relative sizes of two or more projection sub-regions in a segmented X-ray projection image. The inventors have observed that the relative size serves as a reliable metric for the posture of the projection X-ray imaging system with respect to the anatomical structure. The operator can use the posture metric for various purposes, such as, for example, determining the posture of the projection X-ray imaging system with respect to the anatomical structure, or evaluating the suitability of the posture for acquiring X-ray projection images, or determining how to adjust the projection X-ray imaging system posture in order to acquire improved X-ray projection images of the anatomical structure. Therefore, the method promotes a reduction in the amount of X-ray dose delivered to the object by reducing the number of times the X-ray image acquisition is repeated. The method also contributes to improvements in workflow.

[0031] refer to Figure 2 In the illustrated method, in operation S110, X-ray projection data 110 are received. The X-ray projection data include an X-ray projection image 120 representing an anatomical structure 130. The X-ray projection data are acquired by a projection X-ray imaging system 140 having a corresponding pose P with respect to the anatomical structure 130.

[0032] A projection X-ray imaging system includes an X-ray source and an X-ray detector. During the acquisition of X-ray projection data, the X-ray source and the X-ray detector remain in a static position with respect to the anatomical structure. During the acquisition of X-ray projection data, it can be said that the projection X-ray imaging system has a posture with respect to the anatomical structure. X-ray projection data is acquired using a single posture of the projection X-ray imaging system with respect to the anatomical structure. The X-ray projection data can be used to generate an X-ray projection image, i.e., a two-dimensional image, representing the anatomical structure. The X-ray projection data acquired by the projection X-ray imaging system is contrasted with the volumetric X-ray data generated by a computed tomography "CT" imaging system. In a CT imaging system, the X-ray source and the X-ray detector rotate around the anatomical structure so as to acquire volumetric X-ray data from each of a plurality of postures of the CT imaging system with respect to the anatomical structure. The volumetric X-ray data is then reconstructed into a three-dimensional or volumetric image of the anatomical structure.

[0033] The X-ray projection data 110 received in operation S110 may be received from various sources. For example, the X-ray projection data 110 may be received from a projection X-ray imaging system such as Figure 3 In this example, the X-ray projection data 110 may be received by a projection X-ray imaging system 140 as shown. Figure 3The X-ray projection data 110 is received by one or more processors 210 of the system 200 shown. Instead of receiving the X-ray projection data 110 from the projection X-ray imaging system, the X-ray projection data 110 may be received from another source, such as a computer-readable storage medium or the Internet or the cloud. Therefore, the X-ray projection data 110 may have been acquired by the projection X-ray imaging system at an earlier point in time and then received from the computer-readable storage medium or the Internet or the cloud in operation S110. Generally, the X-ray projection data 110 received in operation S110 may be received via any form of data communication, including wired, optical, and wireless communications. As some examples, when wired or optical communications are used, communication may be performed via signals transmitted on cables or optical cables, and when wireless communications are used, communication may be performed via RF or optical signals.

[0034] As described above, the X-ray projection data 110 is acquired by the projection X-ray imaging system 140 having a corresponding pose P with respect to the anatomical structure 130 . Figure 4 is a schematic diagram illustrating an example of a projection X-ray imaging system 140 having a pose P with respect to an anatomical structure 130 according to some aspects of the present disclosure. Figure 4 The projection X-ray imaging system 140 shown corresponds to Figure 3 The projection X-ray imaging system 140 shown includes an X-ray source 140 S and X-ray detector 140 D Examples of anatomical structures are also available at Figure 3 and Figure 4 In the figure, it is shown in the form of an ankle. Figure 3 and Figure 4 In both cases, the X-ray detector 140 D Detection by X-ray source 140 S The X-ray projection data of the ankle is acquired by emitting X-ray radiation that has passed through the ankle. The X-ray projection data represents the X-ray radiation along the X-ray source 140 S and X-ray detector 140 D Line integral of ankle X-ray attenuation between paths.

[0035] The X-ray imaging system 140 measures the posture P of the ankle via Figure 4 The thick black arrow in FIG. 1 is shown with respect to the orientation of the ankle. In the example shown, the posture P is defined by the orientation of the central ray of the projection X-ray imaging system 140 with respect to the anatomical structure. Figure 4The figure within the dashed outline box on the upper right side of shows a generalized pose P of the projection X-ray imaging system 140 with respect to the anatomical structure 130. In the illustration, the pose P is defined by angles α and β with respect to the coordinate system of the anatomical structure 130. As an example, the angle α may be defined as the rotation angle of the central ray of the projection X-ray imaging system 140 around the longitudinal axis of the object, and the angle β may be defined as the tilt angle of the central ray of the projection X-ray imaging system 140 with respect to the cranial-caudal axis of the object.

[0036] Return to Figure 2 In operation S120, the X-ray projection image 120 is segmented to identify a plurality of projection sub-regions 150 of the anatomical structure 130. 1..m . Figure 5 is an example of a segmented X-ray projection image 120 according to some aspects of the present disclosure and includes a projection sub-region 150 of an anatomical structure. 1..m In the example shown, the segmentation has resulted in a projected sub-region 150 for the ankle. 1..m These include: a first projection area 150 of the tibia 1 , the second projection area of ​​the tibia 150 2 , Projected joint space between tibia and talus 150 3 , first articular surface of the talus 150 4 , overlap between fibula and talus 150 5 and the second articular surface 150 of the talus 6 .

[0037] Typically, the projected sub-region 150 of the anatomical structure 130 identified by the segmentation operation 120 1..m Represents one or more of the following: a portion of a bone 150 1..2 , 150 4..6 and the space between the two bones 150 3 Typically, a portion of a bone may include a portion of the diaphysis, i.e., the shaft, of a bone, or a portion of the metaphysis, or a portion of the epiphysis. Examples of portions of a bone that may be identified as projection subregions in operation S120 include projections of a boundary of a bone, and projections of facets of a bone, such as a projection of an articular surface of a bone. Figure 5 In the example shown, the projection sub-area 150 1..m The part representing the tibia, i.e. 150 1..2 , the fibula, i.e. 150 5 , and the talus part, which is 150 4 and 150 6 . Figure 5 Items in 150 4 and 150 6 is the articular surface of the bone (ie, the first articular surface 150 of the talus 4 and the second articular surface of the talus 150 6 As described above, the projection sub-region may alternatively represent the space 150 between two bones. 3 . The space between two bones is also referred to herein as a gap, or more specifically a joint space. A joint space is a gap associated with a joint between two adjacent bones. For example, the joint can be a synovial joint. In this case, at least a portion of the joint space represented in the projection sub-region represents a portion of the synovial cavity. The portion of the joint space represented in the projection sub-region can also represent a decrease in cartilage due to degenerative joint disease or wear and tear. The joint space provides relatively lower X-ray attenuation than bones, and therefore produces a relatively lower intensity region in the X-ray projection image than the bones on either side of the gap. Figure 5 Items in 150 3 is an example of a projection of the space between two bones, i.e., the projected joint space 150 between the tibia and the talus 3 .

[0038] Projection sub-area 150 1..m In one example, at least one of the two or more projection sub-areas is at least partially defined by the X-ray projection image 150. 1..6 The boundary definition of at least one part of the bone in . For example, Figure 5 In the projection sub-area 150 1 The projection sub-region 150 is defined at least at its superior, left and inferior ends by the boundaries of the facets of the bone, more specifically the articular surface of the tibia. 2 The projection sub-region is defined at least at its superior, right and inferior ends by the boundaries of the facets of the bone, more specifically the articular surface of the tibia. In another example, at least one of the two or more projection sub-regions is defined at least in part by the intersections between multiple bones in the X-ray projection image. For example, in Figure 5 In the projection sub-area 150 1 At least at its right end it is defined by the intersection between the fibula and the tibia. Similarly, the projection sub-region 150 2 At least at its left end, it is defined by the intersection between the fibula and the tibia. In another example, at least one of the two or more projection sub-regions is at least partially defined by the space 150 between the two bones. 3 and another bone in the X-ray projection image 120. Figure 5It is not explicitly shown in the figure, but according to this example the projection sub-area may be composed of the joint gap 150 3 The projection sub-region can also be defined by a combination of these examples.

[0039] Various techniques may be used to perform operation S120 , in which the X-ray projection image 120 is segmented to identify a plurality of projection sub-regions 150 of the anatomical structure 130 . 1..m One example technique involves applying a segmentation algorithm to the received X-ray projection data 110. Segmentation algorithms such as model-based segmentation, watershed-based segmentation, region growing, level set, or graph cuts may be used for this purpose. Another example technique involves inputting the received X-ray projection data 110 into a neural network that is trained to segment X-ray projection images representing anatomical structures 130. Examples of such neural networks are described later in Figure 8 An example of a neural network that can be used to segment the X-ray projection image in operation S120 is shown in FIG. et al., “CNN-based pose estimation for assessing quality of ankle-joint X-ray image”, Proc. SPIE12032, Medical Imaging 2022: Medical Processing, 120321A (2022).

[0040] Return to reference Figure 2 In operation S130, based on the two or more projection sub-areas 150 in the segmented X-ray projection image 120 1..m The relative size of the X-ray projection image 120 is used to generate a pose metric for the X-ray projection image 120. As described above, the inventors have observed that this relative size serves as a reliable metric for the pose of the projection X-ray imaging system with respect to the anatomical structure. Figure 6 Describing this principle, a segmented X-ray projection image 120 is shown according to some aspects of the present disclosure. 1..3 Three examples of projection sub-regions 150 of anatomical structures are shown in FIG. 1..m and have been derived from different poses P of the projection X-ray imaging system with respect to the anatomical structure. 1 ( Figure 6 a) P 2 ( Figure 6 b) and P 3 ( Figure 6 c) Collection.

[0041] exist Figure 6In the upper part of the diagram, the figure within the dashed outline box shows the posture P of the projection X-ray imaging system with respect to the anatomical structure 130 by a thick black arrow. 1..3 . Posture P 1..3 Used to collect Figure 6 The corresponding X-ray projection image 120 is shown in the lower part of 1..3 The posture P of the projection X-ray imaging system 140 1..3 Refer to the above Figure 4 In other words, the posture P 1..3 The pose P is defined by the orientation of the central ray of the projection X-ray imaging system with respect to the anatomical structure. 1..3 It can be defined by angles α and β about the coordinate system of the anatomical structure 130. The angle α can be defined as the rotation angle of the central ray of the projection X-ray imaging system around the longitudinal axis of the object, and the angle β can be defined as the tilt angle of the central ray of the projection X-ray imaging system about the cranial-caudal axis of the object.

[0042] Initial reference Figure 6 b; At this position, the posture P 2 This is optimal for generating an X-ray projection image representing the ankle. With this optimal posture, the tip of the fibula is centered with respect to the tibia, and the projected joint gap 150 between the tibia and the talus is 3 The articular surface of the tibia is continuously visible along its length. The following projection sub-areas are Figure 6 b with the best posture P 2 Visible: First projection area 150 of the tibia 1 , the second projection area of ​​the tibia 150 2 , Projected joint space between tibia and talus 150 3 , and overlap between the fibula and talus 150 5 . A size may be calculated for each of these projection sub-regions. The size may be an area or a distance measurement for each projection sub-region. As some examples, the area may be calculated as the total area of ​​the projection sub-regions, or as an area fitted to a shape of the projection sub-regions, and the distance measurement may be calculated as a maximum, or minimum, or average distance of the projection sub-regions, or a distance measurement fitted to a shape of the projection sub-regions. The size may be calculated in units such as (square) millimeters, (square) centimeters, etc., or may be calculated based on another reference unit such as image pixels.

[0043] Reference now Figure 6 a; In this illustration, the projection X-ray imaging system is rotated counterclockwise about the y-axis of the anatomy by approximately 5 degrees from position P 2 Get posture P 1 . Posture P 1It is suboptimal for generating X-ray projection images representing the ankle. This is in part because the joint space between the tibia and talus is 150 3 The articular surface is no longer continuous along the length of the tibia. Figure 6 Compared with b, Figure 6 In a, the first projection area 150 of the tibia 1 The size of the second projection area 150 of the tibia is increased. 2 The overlap between the fibula and talus is 150 5 The size of is also reduced. Therefore, the projection X-ray imaging system is rotated counterclockwise around the y-axis of the anatomical structure to obtain the posture P 2 has resulted in a projection sub-area of ​​150 1 , 150 2 , 150 3 and 150 5 The size of the object changes.

[0044] Reference now Figure 6 c; In this illustration, the projection X-ray imaging system is rotated approximately 5 degrees clockwise around the y-axis of the anatomy from position P 2 Get posture P 3 . Posture P 3 It is suboptimal for generating X-ray projection images representing the ankle. This is in part because the joint space between the tibia and talus is 150 3 The articular surface is no longer continuous along the length of the tibia. Figure 6 Compared with b, Figure 6 In a, the first projection area 150 of the tibia 1 The size of the second projection area 150 of the tibia is reduced, and 2 Increased size. Overlap between fibula and talus 150 5 The size of is also reduced. Therefore, the projection X-ray imaging system is rotated clockwise around the y-axis of the anatomical structure to obtain the posture P 3 has resulted in a projection sub-area of ​​150 1 , 150 2 , 150 3 and 150 5 The size of the object changes.

[0045] It can be understood from the above explanation that as the rotation around the y-axis, the projection sub-area 150 1..m The result of the size change is that two or more projection sub-regions 150 in the segmented X-ray projection image 120 are 1..m The relative size of the projection X-ray imaging system can be used to determine the pose P of the anatomical structure 130. 1..3 For example, the projection sub-area 150 1The projection sub-area 150 2 The relative size of compared to the size of provides a measure of the rotation of the ankle about the y-axis of the anatomy. Projected sub-region 150 4 The projection sub-area 150 3 The relative size of the size of the projection X-ray imaging system compared to the size of the projection sub-area 150 also provides a measure of the rotation of the ankle about the y-axis of the anatomical structure. Other poses of the projection X-ray imaging system with respect to the anatomical structure 130 can be obtained in a similar manner from the projection sub-area 150. 1..m is determined by the relative size of the

[0046] Various examples of operation S130 are described below, in which two or more projection sub-areas 150 1..m The relative sizes of the projection subregions are used to generate the pose metric. These include a method of calculating the size of the projection subregions from the X-ray projection image, and another method of inputting the X-ray projection image 120 into a neural network, and training the neural network to generate the pose metric using ground truth values ​​for the pose metric, the ground truth values ​​being estimated based on the relative sizes of two or more projection subregions in the segmented X-ray projection image.

[0047] In some examples, the posture metric represents a posture P of the projection X-ray imaging system 140 with respect to the anatomical structure 130. In some examples, the posture metric represents the suitability of the projection X-ray imaging system posture P for acquiring the X-ray projection image 120. In some examples, the posture metric represents feedback for adjusting the projection X-ray imaging system posture P in order to acquire subsequent X-ray projection images representing the anatomical structure 130. The posture metric may also represent a combination of these examples. Subsequent X-ray projection images may be referred to herein as improved or suitable or clinically acceptable X-ray projection images.

[0048] Return to reference Figure 2 The method shown; in operation S140, the posture metric is output to provide posture information for the X-ray projection image 120.

[0049] The posture metrics may be output in a variety of ways, including graphically and audibly. For example, the posture metrics may be graphically output to a display of a monitor, tablet, or another device. As some examples, the posture of the projection X-ray imaging system 140 may be graphically output as a display similar to Figure 6. Feedback for adjusting the posture of the projection X-ray imaging system can be output in a similar manner by displaying an arrow indicating how the posture should be adjusted in order to obtain an improved image of the anatomical structure. Feedback for adjusting the posture of the projection X-ray imaging system can alternatively be output auditorily via a speaker, headphones, etc. For example, a voice message such as "rotate the X-ray source ten degrees to the left" or "rotate the ankle 5 degrees to the left" can be output. As some additional examples, the suitability of the projection X-ray imaging system posture P for acquiring the X-ray projection image 120 can be output by outputting a color or another visual indication indicating whether the posture is suitable. For example, the X-ray projection image 120 can be provided with a red outline indicating that the posture is not suitable and a green outline indicating that the posture is indeed suitable.

[0050] In one example, the posture metric represents feedback for adjusting the projection X-ray imaging system posture P in order to acquire subsequent X-ray projection images representing the anatomical structure 130, and the posture metric is used to automatically adjust the projection X-ray imaging system posture P. In this example, the feedback is output to a control system that generates a control signal for controlling one or more motors or actuators to adjust the posture of the projection X-ray imaging system that generates the X-ray projection image 120. The projection X-ray imaging system may include various sensors (such as rotary encoders and position sensors) that measure the posture of the projection X-ray imaging system, and it operates in combination with one or more motors or actuators to provide an adjusted posture. In this example, the feedback can also be output to the user as a recommended posture, so that the posture is automatically adjusted in response to the user accepting the recommended posture. This example has the following benefits: the need to manually reposition the projection X-ray imaging system and / or the patient in order to acquire subsequent X-ray projection images is avoided, thereby saving time and improving workflow.

[0051] In one example, the posture metrics can be output and stored with the X-ray projection image 120. The X-ray projection image 120 can be stored in a standardized format, such as a DICOM format. For example, the posture metrics can be stored with the X-ray projection image 120 in a so-called "Secondary capture" DICOM format or in a "presentation state" DICOM format. The posture metrics and the X-ray projection image 120 can then be transmitted to a PACS viewing station in a secondary capture DICOM format or a presentation state DICOM format for viewing by a radiologist. For example, the posture metrics and the X-ray projection image 120 can be displayed on a PACS viewing station as a secondary capture image or as a presentation state image or as a structured report.

[0052] As described above, in operation S130, two or more projection sub-areas 1501..m The relative sizes of are used to generate pose metrics for the X-ray projection images.

[0053] In one method, the size of the projection sub-region is calculated based on the X-ray projection image. In this method, image processing techniques are used to calculate the size of the projection sub-region. As an example, the above reference Figure 6 Description of the projection sub-area 150 1 and 150 2 The relative size of can be measured and used as a measure of the rotation of the projection X-ray imaging system about the y-axis of the anatomical structure 130, and thus as a measure of the posture of the projection X-ray imaging system with respect to the anatomical structure. Figure 6 a- Figure 6 The sequence of c rotates clockwise around the y axis, projecting sub-area 150 1 The size and projection sub-area 150 2 The relative sizes of the other projection sub-areas can be used to enhance and thus improve the projection sub-area 150. 1 and 150 2 For example, the projection sub-area 150 5 The size of the joint space between the tibia and talus with a projection of 150 3 The ratio also runs through Figure 6 a- Figure 6 The sequence of c changes and is also rotated around the x-axis, so this can also improve the projection sub-area 150 1 and 150 2 Determine the accuracy of the pose.

[0054] As described above, in one example, the pose metric represents the pose P of the projection X-ray imaging system 140 with respect to the anatomical structure 130. This example can be achieved by measuring, for example, Figure 6 Multiple projection sub-regions 150 of the reference anatomical structure in multiple poses are shown 1..m The method can be implemented by determining the size of the X-ray image and generating a functional relationship between the relative size and the corresponding pose. For example, the functional relationship can be provided by a map or a lookup table. The map or the lookup table can then be queried to determine the pose of the new projected X-ray image representing the anatomical structure.

[0055] As described above, in another example, the posture metric represents the suitability of the projection X-ray imaging system posture P for acquiring the X-ray projection image 120. This example can be implemented in a similar manner by determining the posture P as described above and by checking the posture against the posture range that has been functionally marked as suitable for acquiring the X-ray projection image. In this example, the posture metric represents the suitability of the projection X-ray imaging system posture P for acquiring the X-ray projection image 120, and the operation S130 for generating the posture metric of the X-ray projection image 120 includes dividing the two or more projection sub-areas 150 into the projection sub-areas 150. 1..m The relative size of the X-ray projection image 120 is compared to at least one threshold. In this example, the threshold may define a range of postures suitable for acquiring the X-ray projection image 120. The radiologist may set the threshold to define an image that is clinically acceptable for imaging the anatomical structure.

[0056] As described above, in another example, the pose metric represents feedback for adjusting the projection X-ray imaging system pose P in order to acquire subsequent X-ray projection images representing the anatomical structure 130. This example can be implemented by calculating the pose for the new projection X-ray image as described above, and also calculating the pose adjustment as the difference between the calculated pose and the reference pose used to acquire the X-ray projection image representing the anatomical structure 130. Figure 6 , in this example, the calculation may result in a posture adjustment that is calculated as a posture value P 1 and P 2 The difference between them, or is calculated as the attitude value P 3 and P 2 The difference between 2 It is the best posture for the anatomy.

[0057] In one example, the pose metric of the X-ray projection image 120 is further determined based on anatomical landmarks in the X-ray projection image. This is based on two or more projection sub-regions 150 in the segmented X-ray projection image 120. 1..m The relative size of the X-ray projection image is used to determine the pose metric to provide redundancy. This example is useful for improving the accuracy of the pose metric in situations where the X-ray projection image does not allow accurate segmentation in operation S130. For example, bone contours may be difficult to see in some poses and under some X-ray imaging settings. For example, anatomical structure variability and implants in the anatomy can cause obstruction of image features in the X-ray projection image, thereby confusing the segmentation operation. In this example, reference Figure 1 The methods described include:

[0058] Identifying one or more anatomical landmarks 160 in the x-ray projection image 120 1..n location; and

[0059] The pose metric generated S130 for the X-ray projection image 120 is further based on one or more anatomical landmarks 160 1..n The identified location.

[0060] refer to Figure 7 Describe the example, Figure 7 is an example of a segmented X-ray projection image 120 according to some aspects of the present disclosure and includes a projection sub-region 150 of an anatomical structure. 1..m Examples and multiple anatomical landmarks 160 1..n Example of location. Figure 7 The segmented X-ray projection image 120 shown in FIG. 1 shows a region around the fibula tip 160 1 of Figure 5 , and includes the location of the following anatomical landmarks: fibula tip 160 1 and the tip of the tibia and ankle 160 2 Understandably, Figure 7 The marker 160 in the segmented X-ray projection image 120 is shown 1..n The location is based on Figure 5 The projection X-ray imaging system shown changes with respect to changes in the posture of the ankle. For example, as the posture changes, the position of a single anatomical landmark in the projection X-ray image 120 changes with respect to the projection sub-area 150. 1..m As a result, the X-ray projection image 120 may be based on one or more anatomical landmarks 160 1..n and two or more projection sub-regions 150 in the segmented X-ray projection image 120 1..m The relative size of the projection sub-region 150 is used to generate a pose metric for the X-ray projection image 120. Similarly, if multiple anatomical landmarks are identified in the projection X-ray image 120, the pose metric can also be generated with respect to one or more projection sub-regions 150. 1..m The positions of these marks are determined by their positions.

[0061] If multiple landmarks are identified in the X-ray projection image 120, further information may also be derived from the X-ray projection image 120 and used to generate posture metrics. For example, in one example, multiple anatomical landmarks 160 may be measured. 1..n The distance between the positions. Figure 7 In this example, multiple distances 170 between different anatomical landmarks have been calculated. 1..i For example, a distance of 170 1..3 The radius of the circular approximation representing the two talar and tibial condyles. Distance 170 4 Indicates the distance between the tip of the fibula and the tip of the tibial condyle. Distance 170 5Represents the posterior joint space, which is represented by the distance between the circular approximation of the tibial condyle and the proximal talar condyle. Distance 170 6 Represents the central joint space represented by the distance between the circular approximation of the tibial condyle and the more proximal talar condyle. Distance 170 8 Represents the anterior joint space represented by the distance between the tibial condyle and the circular approximation of the more proximal talar condyle. Distance 170 7 represents the anterior distance between the fibula and tibia at defined important locations. As can be appreciated, these distances are based on the distances used to generate Figure 7 The X-ray projection images 120 shown vary depending on the pose of the X-ray imaging system, and thus they can be used to characterize the pose of the X-ray imaging system. In another example, the angles between the trajectories defined by multiple anatomical landmarks can be measured, and these angles can be used in a similar manner to characterize the pose of the X-ray imaging system. Figure 7 An X-ray projection image 120 is shown showing the pose of the X-ray imaging system.

[0062] These two examples can be applied to different anatomical structures and are stated more formally as follows: Let ..., where ... · represents image coordinates, and ·:... represents a grayscale image. Define the mapping It assigns an anatomical region to each pixel ·, where · is the number of anatomical regions. In addition, an additional geometric metric is calculated where · is the number of additional metrics. These metrics can be Figure 7 In the example where the pose metric represents feedback for adjusting the pose of the projection X-ray imaging system, in the second step, by mapping ·(·(·), ·)·· · to transform these anatomical segments, where · is the dimension of the feedback to correct the pose of the projection X-ray imaging system with respect to the anatomical structure. The mapping · is a linear or nonlinear mapping from the anatomical structure and the additional geometrical metrics to the pose feedback.

[0063] In another example, the anatomical landmark 160 in the X-ray projection image 120 is 1..n The position of the anatomical landmarks 160 is mapped to the corresponding landmark in the reference image, and based on the plurality of anatomical landmarks 160 1..nThe pose metric of the X-ray projection image is determined with respect to the mapped positions of the corresponding landmarks in the reference image. In this example, a 3D model representing the anatomical structure is projected using a projection X-ray imaging system at different poses with respect to the anatomical structure to obtain a reference image. The 3D model is projected in a virtual sense using a model of a projection X-ray imaging system, wherein an X-ray source and an X-ray detector are arranged in the same manner as in an imaging system used to generate the X-ray projection image. The projected landmarks in the reference image appear in positions that are characteristic of the pose of the model of the projection X-ray imaging system used to generate the reference image. Thus, by mapping the anatomical landmarks 160 in the X-ray projection image 120 to the reference image, the 3D model is projected in a virtual sense using a model of a projection X-ray imaging system, wherein an X-ray source and an X-ray detector are arranged in the same manner as in an imaging system used to generate the X-ray projection image. 1..n The position of the image can be matched with the corresponding landmarks in the reference image, and the pose metric can also be determined.

[0064] Therefore, in these examples, reference Figure 2 The method described includes identifying a plurality of anatomical landmarks 160 in an X-ray projection image 120 1..n , and:

[0065] Measure multiple anatomical landmarks160 1..n One or more distances between the locations 170 i..1 ; and / or

[0066] measuring angles of one or more trajectories defined by a plurality of anatomical landmarks; and / or

[0067] Multiple anatomical landmarks 160 1..n The positions are mapped to a plurality of corresponding landmarks in a reference image;

[0068] and wherein generating S130 the pose metric for the X-ray projection image 120 is further based on the measured one or more distances 170 1..i and / or one or more trajectories of measured angles and / or multiple anatomical landmarks 160 1..n The location of the map with respect to the corresponding landmark in the reference image.

[0069] In these examples, various techniques may be used to identify anatomical landmarks. For example, a feature detector or edge detector or model-based segmentation or neural network may be used to identify anatomical landmarks in the x-ray projection image 120.

[0070] In the above example, two or more projection sub-regions 150 in the segmented X-ray projection image 120 may be determined. 1..mThe X-ray projection image 120 is scaled before the relative size of the X-ray projection image 120 is determined. This reduces the influence of factors such as differences in object size and differences in the field of view of the projection X-ray imaging system on the pose metric. Therefore, in one example, the operation of generating S130 a pose metric for the X-ray projection image 120 includes scaling the X-ray projection image 120, and determining two or more projection sub-regions 150 from the scaled X-ray projection image. 1..m The relative size of .

[0071] In this example, an area of ​​the X-ray projection image 120 may be scaled based on a measurement of the area in the X-ray projection image 120. Alternatively, an area of ​​the X-ray projection image 120 may be scaled based on a measurement of a distance in the X-ray projection image 120. An area of ​​the X-ray projection image may be scaled based on a measurement of the area in the X-ray projection image 120 with respect to a reference area. For example, the reference area may be defined as a projected area of ​​a portion of a bone having a known pose. Similarly, an area of ​​the X-ray projection image 120 may be scaled based on a measurement of a distance in the X-ray projection image 120 with respect to a reference distance. For example, the reference distance may be defined as the length of the bone, or the width of the bone at a predetermined position along its length.

[0072] In another example, the anatomical structure 130 in the X-ray projection image 120 is registered to an anatomical atlas image representing the anatomical structure 130 to provide a scaling factor for the X-ray projection image, and a region of the X-ray projection image 120 is scaled using the scaling factor.

[0073] In another method, an X-ray projection image is input into a neural network, and the neural network is trained using a ground truth for a pose metric to generate a pose metric, the ground truth being estimated based on the relative sizes of two or more projection sub-regions in the segmented X-ray projection image. In this method, the pose metric may also represent: a pose P of the projection X-ray imaging system 140 with respect to the anatomical structure 130, and / or the suitability of the projection X-ray imaging system pose P for acquiring the X-ray projection image 120, and / or feedback for adjusting the projection X-ray imaging system pose P to acquire subsequent X-ray projection images representing the anatomical structure 130.

[0074] In the method, the operation of generating S130 a posture metric for the X-ray projection image 120 includes:

[0075] inputting the segmented image data representing the segmented X-ray projection image into the second neural network NN2; and

[0076] generating the pose metric using the second neural network NN2 in response to the input; and

[0077] Wherein the second neural network NN2 is trained using X-ray projection image training data to generate posture metrics from the segmented image data, the X-ray projection image training data comprising a plurality of segmented X-ray projection images representing the anatomical structure 130 and corresponding ground truth values ​​for the posture metrics, wherein the ground truth values ​​for the posture metrics are estimated based on the relative sizes of two or more projection sub-regions in the segmented X-ray projection images.

[0078] refer to Figure 8 Describe the method, Figure 8 is a schematic diagram showing an example of a method for providing posture information of an X-ray projection image according to some aspects of the present disclosure, and wherein X-ray projection data 110 including the X-ray projection image is input into a first neural network NN1 so as to segment the X-ray projection image, and wherein segmented image data representing the segmented image is input into a second neural network NN2 so as to generate a posture metric for the X-ray projection image.

[0079] refer to Figure 8 In the method, segmented image data representing segmented X-ray projection images are input into a second neural network NN2. Figure 8 In the example shown, the segmented image data is obtained by inputting the X-ray projection data 110 into the first neural network NN1. The first neural network NN1 is trained to segment the X-ray projection image representing the anatomical structure 130. As described above, an example of a neural network that can be used to segment the X-ray projection image in operation S120 is et al., “CNN-based pose estimation for assessing quality of ankle-joint X-ray image” Proc. SPIE 12032, Medical imaging 2022: Image processing, 120321A (2022). The neural network NN1 can be provided by various architectures, including, for example, a convolutional neural network “CNN”, an encoder-decoder based model, a recurrent neural network “RNN” based model, and a generative model with adversarial training “GAN”. It should be noted that, for example, as described above, instead of using a neural network (i.e., NN1) to obtain segmented image data, a segmentation algorithm can be used to obtain segmented image data, such as model-based segmentation, watershed-based segmentation, region growing, level set, or graph cutting.

[0080] The segmented image data is then input into the second neural network NN2. The second neural network NN2 can be provided by a variety of different architectures, including, for example, CNN, ResNet, U-Net, and encoder-decoder architectures.

[0081] The second neural network NN2 is trained using the X-ray projection image training data to generate posture metrics from the segmented image data. The X-ray projection image training data includes a plurality of segmented X-ray projection images representing the anatomical structure 130 and corresponding ground truth values ​​of the posture metrics. The X-ray projection image training data may include dozens, hundreds, or thousands or even more segmented X-ray projection images representing the anatomical structure 130. The X-ray projection image training data may include data from objects of various ages, body mass index "BMI" values, different genders, and different pathologies. The segmented X-ray projection images used in the training data may be obtained by segmenting the X-ray projection images. The segmentation may be performed manually by an expert, or using various segmentation algorithms such as those described above. The ground truth value of the posture metric is estimated based on the relative sizes of two or more projection sub-regions in the segmented X-ray projection images. In this regard, the relative sizes may be calculated using the above-described image processing techniques and may be used to determine values ​​for one or more of: a pose P of the projection X-ray imaging system 140 with respect to the anatomical structure 130, the suitability of the projection X-ray imaging system pose P for acquiring the X-ray projection image 120, and feedback for adjusting the projection X-ray imaging system pose P for acquiring subsequent X-ray projection images representing the anatomical structure 130. The pose metric may be calculated using the above-described techniques. During training, the ground truth of the pose metric and, optionally, the relative sizes of two or more projection sub-regions in the segmented X-ray projection images used to derive the pose metric are input into the neural network.

[0082] Training of a neural network involves inputting a training data set into the neural network and iteratively adjusting the parameters of the neural network until the trained neural network provides accurate outputs. Training is typically performed using a graphics processing unit "GPU" or a dedicated neural processor (such as a neural processing unit "NPU" or a tensor processing unit "TPU"). Training typically employs a centralized approach, where cloud-based or mainframe-based neural processors are used to train the neural network. After training with the training data set, the trained neural network can be deployed to a device for analyzing new input data during inference. The processing requirements during inference are significantly less than those required during training, allowing the neural network to be deployed to a variety of systems, such as laptops, tablets, mobile phones, etc. Inference can be performed, for example, by a central processing unit "CPU", GPU, NPU, TPU on a server or in the cloud.

[0083] Therefore, the process of training the neural network NN2 described above includes adjusting its parameters. Parameters, or more specifically weights and biases, control the operation of activation functions in the neural network. In supervised learning, the training process automatically adjusts weights and biases so that when input data is presented, the neural network accurately provides the corresponding expected output data. To this end, the value of the loss function or error is calculated based on the difference between the predicted output data and the expected output data. The value of the loss function can be calculated using functions such as negative log-likelihood loss, mean absolute error (or L1 norm), mean square error, root mean square error (or L2 norm), Huber loss or (binary) cross entropy loss. During training, the value of the loss function is usually minimized, and the training is terminated when the value of the loss function meets the stopping criteria. Sometimes, when the value of the loss function meets one or more of a plurality of criteria, the training is terminated.

[0084] Various methods for solving the loss minimization problem are known, such as gradient descent, quasi-Newton method, etc. Various algorithms have been developed to implement these methods and their variants, including but not limited to stochastic gradient descent "SGD", batch gradient descent, mini-batch gradient descent, Gauss-Newton, Levenberg Marquardt, Momentum, Adam, Nadam, Adagrad, Adadelta, RMSProp and Adamex "optimizers". These algorithms use the chain rule to calculate the derivatives of the loss with respect to the model parameters. This process is called back propagation because the derivatives are calculated from the last layer or output layer and move toward the first layer or input layer. These derivatives inform the algorithm how the model parameters must be adjusted in order to minimize the error function. That is, starting from the output layer and working backwards in the network until the input layer is reached, the model parameters are adjusted. In the first training iteration, the initial weights and biases are usually random. The neural network then predicts the output data, which is also random. Back propagation is then used to adjust the weights and biases. The training process is iteratively performed by adjusting the weights and biases in each iteration. When the error or difference between the predicted output data and the expected output data is within an acceptable range for the training data or some validation data, the training is terminated. Subsequently, the neural network can be deployed and the trained neural network uses the trained values ​​of its parameters to make predictions on new input data. If the training process is successful, the trained neural network accurately predicts the expected output data from the new input data.

[0085] As described above, in one example, the pose metric represents the suitability of the projection X-ray imaging system pose P for acquiring X-ray projection images. In this example, the second neural network NN2 is trained to generate the pose metric from the segmented X-ray projection images in the following manner:

[0086] receiving X-ray projection image training data;

[0087] Inputting the X-ray projection image training data into the second neural network NN2;

[0088] For each of the plurality of segmented X-ray projection images:

[0089] Predicting the value of the posture metric using the second neural network NN2;

[0090] Adjusting parameters of the second neural network NN2 based on a difference between the predicted value of the pose metric and the true value of the pose metric; and

[0091] The predicting and the adjusting are repeated until a stopping criterion is met.

[0092] These operations can be performed according to the back-propagation technique described above. In this example, the ground truth of the pose metric, i.e., suitability, can be automatically or manually evaluated by an expert reviewer based on the relative sizes of two or more projection sub-regions in the segmented X-ray projection image. The ground truth of the pose metric can also be determined based on one or more additional criteria.

[0093] As described above, in another example, the pose metric represents feedback for adjusting the pose P of the projection X-ray imaging system so as to acquire subsequent X-ray projection images representing the anatomical structure 130. In this example, the ground truth of the pose metric includes one or more pose adjustments for adjusting the pose of the projection X-ray imaging system so as to acquire an X-ray projection image representing the anatomical structure having a target pose with respect to the anatomical structure. The second neural network is trained to generate the pose metric from the segmented X-ray projection images by:

[0094] receiving X-ray projection image training data;

[0095] inputting the X-ray projection image training data into the second neural network; and

[0096] For each of the plurality of segmented X-ray projection images:

[0097] predicting, using the second neural network, a value of a pose metric, the value of the pose metric comprising one or more pose adjustments for adjusting a pose of the projection X-ray imaging system so as to acquire an X-ray projection image representing the anatomical structure having a target pose with respect to the anatomical structure;

[0098] adjusting parameters of the second neural network based on a difference between the predicted value of the pose metric and the true value of the pose metric; and

[0099] The predicting and the adjusting are repeated until a stopping criterion is met.

[0100] These operations can be performed according to the above-mentioned back-propagation technique. In this example, the ground truth of the posture metric, i.e., feedback, can be determined automatically or manually by an expert reviewer based on the relative sizes of two or more projection sub-regions in the segmented X-ray projection image. The ground truth of the posture metric can also be determined based on one or more additional criteria. Therefore, given the actual size of the projection sub-region, calculate how to adjust the current posture to acquire subsequent X-ray projection images representing the anatomical structure 130, where the relative size of the projection sub-region is within a predetermined range for the anatomical region. A predetermined range can be set for subsequent X-ray projection images so that the resulting X-ray projection images provide preferred or clinically acceptable images.

[0101] In another example, the above reference Figure 2 The method described also includes generating a value of a second posture metric. The value of the second posture metric is used to adjust the value of the posture metric. In this example, the above reference Figure 2 Both the described posture metric and the second posture metric represent the suitability of the projection X-ray imaging system posture P for acquiring the X-ray projection image 120. However, the second posture metric is evaluated based on the size of the gap between two bones in the X-ray projection image or the segmented X-ray projection image. The second posture metric is based on the same as described above with respect to Figure 1 The described posture metrics are evaluated using different criteria and can be used to override the evaluation provided by the posture metrics.

[0102] In this example, the pose metric represents the suitability of the X-ray imaging system pose P for acquiring the X-ray projection image 120 and is referenced to Figure 1 The methods described include:

[0103] Based on the gap 150 between the two bones in the X-ray projection image 120 3 The size of the X-ray projection image 120 is used to calculate a value of a second posture metric for the X-ray projection image 120, where the value of the second posture metric indicates suitability of the X-ray imaging system posture P for acquiring the X-ray projection image 120;

[0104] And wherein generating S130 a pose metric for the X-ray projection image 120 is further based on a value of a second pose metric.

[0105] In this example, the value of the second posture metric is used to adjust the value of the first posture metric. In this regard, if the value of the second posture metric indicates that the X-ray imaging system posture P is indeed suitable for the gap 150 between the two bones in the X-ray projection image 120, 3If the X-ray projection image 120 is acquired with a size of , the posture metric output in operation S140 is adjusted from "unsuitable" to "suitable". Therefore, if the value of the second posture metric indicates that the posture is suitable for the X-ray projection image, it overrides the original evaluation of the posture metric.

[0106] An example of a gap between two bones in the X-ray projection image 120 is Figure 5 The gap 150 between the tibia and the talus in the X-ray projection image 120 is shown. 3 . Clearance 150 3 The size of the gap can be calculated as an area or distance measurement of the gap. As described above, the projection sub-area 150 can be 1..m The size of the gap is calculated in a similar manner.

[0107] The value of the second posture metric can be calculated using a method similar to that described above for the posture metric. Thus, in one method, image processing techniques are used to calculate the size of the gap between the two bones in the X-ray projection image. Alternatively, in another method described in more detail below, a neural network is trained to generate the value of the second posture metric from the X-ray projection data.

[0108] In the image processing method, image processing techniques may be used to calculate the gap 150 between two bones in the X-ray projection image 120. 3 The size of the second posture metric can be calculated based on the size in various ways. Generally, larger size values ​​correspond to higher levels of posture fitness of the X-ray projection image. Therefore, in one example, the size can be directly used as an analog value of the second posture metric. In another example, the size can be digitized based on one or more thresholds and used to generate discrete categories of values ​​for the second posture metric. In this example, the discrete categories correspond to different fitness levels.

[0109] As described above, in another method, a neural network is trained to generate a value of a second posture metric from the X-ray projection data. In this method, the operation of calculating the value of the second posture metric includes:

[0110] Inputting the X-ray projection data representing the X-ray projection image 120 or the segmented X-ray projection image into the third neural network NN3; and

[0111] In response to the input, generating a value of a second posture metric using a third neural network NN3; and

[0112] The third neural network NN3 is trained using X-ray projection image training data and corresponding ground truth values ​​for the second posture metric to generate a value of the second posture metric from the X-ray projection data, the X-ray projection image training data comprising a plurality of X-ray projection images or segmented X-ray projection images representing the anatomical structure 130, and the ground truth value of the second posture metric is estimated based on the size of a gap between two bones in the X-ray projection image or the segmented X-ray projection image.

[0113] refer to Figure 8 In the lower part of the example, the X-ray projection data representing the X-ray projection image 120 or the segmented X-ray projection image is input into the third neural network NN3, as shown in FIG. Figure 8 As shown by the dotted line in . As described above, the segmented X-ray projection images can be generated using the neural network NN1 or alternatively using various segmentation algorithms.

[0114] The third neural network NN3 can be provided by various architectures, including, for example, convolutional neural networks "CNN", ResNet, U-Net, and encoder-decoder architectures. Examples of neural networks that can be trained for this purpose are disclosed in the literature "An AI-based Framework for Diagnostic Quality Assessment of Ankle Radiographs" by Mairhofer, D. et al., Proceedings of Machine Learning Research 143: 484-496, 2021.

[0115] The third neural network NN3 is trained using the X-ray projection image training data to generate a second posture metric from the X-ray projection data. The X-ray projection image training data includes a plurality of X-ray projection images or segmented X-ray projection images representing the anatomical structure 130, and a corresponding ground truth value for the second posture metric. The X-ray projection image training data may include dozens, hundreds, or thousands or more segmented X-ray projection images representing the anatomical structure 130. The X-ray projection image training data may include data from objects of various ages, body mass index "BMI" values, different genders, and different pathologies. The ground truth value for the second posture metric is estimated based on the size of the gap between two bones in the X-ray projection image or the segmented X-ray projection image. The ground truth value for the second posture metric is estimated by determining the size of the gap between two bones in the X-ray projection image or the segmented X-ray projection image using the above-mentioned image processing method. In this regard, the size of the gap is calculated and used to determine the value of the suitability of the X-ray imaging system posture P for acquiring the X-ray projection image 120. The above-mentioned technique can be used to calculate the second posture metric for the (segmented) X-ray projection image. Thus, the size can be used directly as an analog value for the second posture metric, or the size can be digitized based on one or more thresholds and used to generate discrete categories of values ​​for the second posture metric. During training, the ground truth of the posture metric and optionally also the relative sizes of two or more projection sub-regions in the segmented X-ray projection images used to derive the posture metric are input into the third neural network NN3.

[0116] In this example, the third neural network NN3 is trained to generate values ​​of the second pose metric from the X-ray projection data in the following manner:

[0117] receiving X-ray projection image training data;

[0118] Inputting the X-ray projection image training data into the third network NN3;

[0119] For each of the plurality of X-ray projection images or the segmented X-ray projection images:

[0120] Predicting a value of a second posture metric using a third neural network NN3;

[0121] adjusting parameters of the third neural network NN3 based on a difference between a predicted value of the second posture metric and a true value of the second posture metric;

[0122] The predicting and the adjusting are repeated until a stopping criterion is met.

[0123] These operations can be performed according to the back-propagation technique described above.

[0124] In another example, a computer program product is provided. The computer program product includes instructions that, when executed by one or more processors, cause the one or more processors to perform a method of providing posture information for an X-ray projection image. The method includes:

[0125] Receiving S110 X-ray projection data 110 , the X-ray projection data comprising an X-ray projection image 120 representing an anatomical structure 130 , the X-ray projection data being acquired by a projection X-ray imaging system 140 having a corresponding pose P with respect to the anatomical structure 130 ;

[0126] Segmenting S120 the X-ray projection image 120 to identify a plurality of projection sub-regions 150 of the anatomical structure 130 1..m ;

[0127] Based on two or more projection sub-regions 150 in the segmented X-ray projection image 120 1..m to generate S130 a pose metric for the X-ray projection image 120 ; and

[0128] Output S140 pose metrics to provide pose information for the X-ray projection image 120

[0129] In another example, a system 200 for providing posture information for an X-ray projection image is provided. The system includes one or more processors 210 configured to:

[0130] Receiving S110 X-ray projection data 110 , the X-ray projection data comprising an X-ray projection image 120 representing an anatomical structure 130 , the X-ray projection data being acquired by a projection X-ray imaging system 140 having a corresponding pose P with respect to the anatomical structure 130 ;

[0131] Segmenting S120 the X-ray projection image 120 to identify a plurality of projection sub-regions 150 of the anatomical structure 130 1..m ;

[0132] Based on two or more projection sub-regions 150 in the segmented X-ray projection image 120 1..m to generate S130 a pose metric for the X-ray projection image 120 ; and

[0133] Output S140 pose metrics to provide pose information for the X-ray projection image 120

[0134] Figure 3 An example of the system 200 is shown in FIG. 2. Note that the system 200 may also include one or more of the following: a projection X-ray imaging system 140 for generating X-ray projection data 110; a user interface device ( Figure 3), which is used to receive user input related to the operations performed by one or more processors 210, such as a keyboard, a mouse, a touch screen, etc.; a bed ( Figure 3 ); and a monitor ( Figure 3 ), which is used to display the X-ray projection image 120, the output posture metrics, and other data related to the operations performed by one or more processors 210.

[0135] In the above example, a value of a second posture metric is generated, and the value of the posture metric is adjusted using the value of the second posture metric. In this example, the second posture metric represents the suitability of the projection X-ray imaging system posture P for acquiring the X-ray projection image 120, and the second posture metric is estimated based on the size of the gap between two bones in the X-ray projection image or the segmented X-ray projection image. The second posture metric is described as being used to override the assessment provided by the posture metric. In yet another example, the second posture metric can simply be output instead of being used to adjust the value of the posture metric. In the example described below, the second posture metric is simply referenced as the posture metric. In this example, a computer-implemented method for providing posture information for an X-ray projection image includes:

[0136] Receiving S110 X-ray projection data 110 , the X-ray projection data comprising an X-ray projection image 120 representing an anatomical structure 130 , the X-ray projection data being acquired by a projection X-ray imaging system 140 having a corresponding pose P with respect to the anatomical structure 130 ;

[0137] Segmenting S120 the X-ray projection image 120 to identify a plurality of projection sub-regions 150 of the anatomical structure 130 1..m ;

[0138] Generate S130 a posture metric for the X-ray projection image 120, wherein the value of the posture metric represents the suitability of the X-ray imaging system posture P for acquiring the X-ray projection image 120, and wherein the value of the posture metric is based on the projection sub-area 150 1..m The dimensions of one or more projected sub-regions in are generated, including:

[0139] The gap 150 between the two bones in the X-ray projection image 120 3 size, and / or

[0140] At least in part, the X-ray projection image 150 1 , 150 2 The size of the projection sub-region defined by the intersection points between the multiple bones in; and / or

[0141] At least partially by the gap 150 between the two bones 3The size of the projection sub-region defined by the intersection with another bone in the X-ray projection image 120; and

[0142] The pose metrics are output S140 to provide pose information for the X-ray projection image 120 .

[0143] This example can be implemented in the same manner as described above for the second posture metric. Thus, an example of a gap between two bones in the X-ray projection image 120 according to this example is Figure 5 The gap 150 between the tibia and the talus in the X-ray projection image 120 is shown. 3 . Clearance 150 3 The size of the gap can be calculated as an area or distance measurement of the gap. 1..m In one example, the gap 150 between two bones in the X-ray projection image 120 is estimated by measuring the spacing between the bones at multiple locations along a portion of the gap. 3 The size of the gap is determined, and a pose metric is generated for the x-ray projection image 120 based on the variation in the spacing between the bones along a portion of the gap. The variation in the spacing between the bones along the gap can be measured using a variety of metrics, including, for example, by evaluating the variance or standard deviation of the spacing between the bones along a portion of the gap. Examples of such gaps that can be measured in this manner are Figure 5 The gap 150 between the tibia and the talus in the X-ray projection image 120 is shown. 3 A variation value indicating a relatively high degree of uniformity (e.g., a relatively low variance) in the measured intervals between bones along a portion of the gap may be associated with a suitable X-ray imaging system posture P for acquiring the X-ray projection image 120, while a variation value indicating a relatively low degree of uniformity (e.g., a relatively high variance) in the measured intervals between bones along a portion of the gap may be associated with an unsuitable X-ray imaging system posture P for acquiring the X-ray projection image 120. For example, this example may be used to determine a posture metric for an anterior-posterior “AP” image of an ankle.

[0144] Similarly, the value of the posture metric may be generated based on the size of the projection sub-region in the segmented X-ray projection image 120, the projection sub-region 150 1..m At least in part, the X-ray projection image 150 1 , 150 2 The intersection between multiple bones in the or the gap between two bones 150 3 The intersection point with another bone in the X-ray projection image 120 is defined in the same manner as described above.

[0145] A method similar to the method described above may be used to calculate the value of the pose metric. Thus, in one method, image processing techniques are used to calculate the size of the gap between two bones in the X-ray projection image. Alternatively, in another method, a neural network is trained to generate the value of the pose metric from the X-ray projection data. In the image processing method, the image processing technique may be used to calculate the gap 150 between two bones in the X-ray projection image 120. 3 The size of the posture metric. The value of the posture metric can be calculated based on the size in various ways. Generally, larger size values ​​correspond to higher levels of suitability of the posture for the X-ray projection image. Therefore, in one example, the size can be directly used as an analog value of the posture metric. In another example, the size can be digitized based on one or more thresholds and used to generate discrete categories of values ​​of the posture metric. In this example, the discrete categories correspond to different levels of suitability. As described above, in another method, a neural network is trained to generate values ​​of the posture metric from the X-ray projection data. In this method, the operation of calculating the value of the posture metric includes:

[0146] Inputting the X-ray projection data representing the X-ray projection image 120 or the segmented X-ray projection image into the third neural network NN3; and

[0147] In response to the input, generating a value of a posture metric using a third neural network NN3; and

[0148] The third neural network NN3 is trained using X-ray projection image training data to generate values ​​of posture metrics from the X-ray projection data, the X-ray projection image training data comprising a plurality of X-ray projection images or segmented X-ray projection images representing the anatomical structure 130, and corresponding ground truth values ​​of the posture metrics, the ground truth values ​​of the posture metrics being estimated based on the size of a gap between two bones in the X-ray projection images or the segmented X-ray projection images.

[0149] In this example, the third neural network NN3 is trained to generate values ​​of pose metrics from X-ray projection data in the following way:

[0150] receiving X-ray projection image training data;

[0151] Inputting the X-ray projection image training data into the third network NN3;

[0152] For each of the plurality of X-ray projection images or the segmented X-ray projection images:

[0153] Predicting the value of the posture metric using a third neural network NN3;

[0154] adjusting parameters of the third neural network NN3 based on a difference between the predicted value of the pose metric and the true value of the pose metric; and

[0155] The predicting and the adjusting are repeated until a stopping criterion is met.

[0156] These operations can be performed according to the back-propagation technique described above.

[0157] According to this example, a corresponding computer program product and a corresponding system are also provided.

[0158] The computer program product comprises instructions which, when executed by one or more processors, cause the one or more processors to perform a method for providing posture information for an X-ray projection image. The method comprises:

[0159] Receiving S110 X-ray projection data 110 , the X-ray projection data comprising an X-ray projection image 120 representing an anatomical structure 130 , the X-ray projection data being acquired by a projection X-ray imaging system 140 having a corresponding pose P with respect to the anatomical structure 130 ;

[0160] Segmenting S120 the X-ray projection image 120 to identify a plurality of projection sub-regions 150 of the anatomical structure 130 1..m ;

[0161] Generate S130 a posture metric for the X-ray projection image 120, wherein the value of the posture metric represents the suitability of the X-ray imaging system posture P for acquiring the X-ray projection image 120, and wherein the value of the posture metric is based on the projection sub-area 150 1..m The dimensions of one or more projected sub-regions in are generated, including:

[0162] The gap 150 between the two bones in the X-ray projection image 120 3 size, and / or

[0163] At least in part, the X-ray projection image 150 1 , 150 2 The size of the projection sub-region defined by the intersection points between the multiple bones in; and / or

[0164] At least partially by the gap 150 between the two bones 3 The size of the projection sub-region defined by the intersection with another bone in the X-ray projection image 120; and

[0165] Output S140 pose metrics to provide pose information for the X-ray projection image 120

[0166] The corresponding system includes one or more processors configured to:

[0167] Receiving S110 X-ray projection data 110 , the X-ray projection data comprising an X-ray projection image 120 representing an anatomical structure 130 , the X-ray projection data being acquired by a projection X-ray imaging system 140 having a corresponding pose P with respect to the anatomical structure 130 ;

[0168] Segmenting S120 the X-ray projection image 120 to identify a plurality of projection sub-regions 150 of the anatomical structure 130 1..m ;

[0169] Generate S130 a posture metric for the X-ray projection image 120, wherein the value of the posture metric represents the suitability of the X-ray imaging system posture P for acquiring the X-ray projection image 120, and wherein the value of the posture metric is based on the projection sub-area 150 1..m The size of one or more projected sub-regions in is generated, including:

[0170] The gap 150 between the two bones in the X-ray projection image 120 3 size, and / or

[0171] At least in part, the X-ray projection image 150 1 , 150 2 The size of the projection sub-region defined by the intersection points between the multiple bones in; and / or

[0172] At least partially by the gap 150 between the two bones 3 The size of the projection sub-region defined by the intersection with another bone in the X-ray projection image 120; and

[0173] Output S140 pose metrics to provide pose information for the X-ray projection image 120

[0174] In the relevant example, a gap of 150 is used. 3 The normalized value of the size of is used to perform the operation of generating S130 the posture metric for the X-ray projection image 120. 3 The normalized value of the size is calculated as follows:

[0175] i) Scaling the gap 150 based on the size of the anatomical features in the X-ray projection image 3 or the size of the projected subregion, or

[0176] ii) registering the anatomical structure in the X-ray projection image 120 to the anatomical atlas image representing the anatomical structure 130 to provide a scaling factor for the X-ray projection image, scaling a region of the X-ray projection image 120 using the scaling factor to provide a scaled X-ray projection image, and measuring a gap 150 in the scaled X-ray projection image 3The size of the projection or the size of the projected sub-region.

[0177] In this example, the anatomical feature used to generate the normalized value of the size of the gap can be, for example, a dimension of a bone, such as the length of the bone or the width of the bone. Alternatively, an anatomical atlas image can be used. Normalization provides compensation for differences in the sizes of the features used to determine the pose in the projection image, and therefore provides a more reliable pose metric.

[0178] An enumerated list of examples of the present disclosure is provided below.

[0179] Example 1: A computer-implemented method of providing pose information for an X-ray projection image, the method comprising:

[0180] Receiving (S110) X-ray projection data (110), the X-ray projection data comprising an X-ray projection image (120) representing an anatomical structure (130), the X-ray projection data being acquired by a projection X-ray imaging system (140) having a corresponding pose (P) with respect to the anatomical structure (130);

[0181] Segmenting (S120) the X-ray projection image (120) to identify a plurality of projection sub-regions (150) of the anatomical structure (130) 1..m );

[0182] Based on the projection sub-region (150) in the segmented X-ray projection image (120) 1..m ) to generate (S130) a pose metric for the X-ray projection image (120) based on the relative sizes of two or more projection sub-regions in the X-ray projection image (120); and

[0183] The pose metric is output (S140) to provide the pose information for the X-ray projection image (120).

[0184] Example 2: The computer-implemented method of Example 1, wherein the projected sub-region (150) of the anatomical structure (130) 1..m ) means one or more of the following: a part of a bone (150 1..2 , 150 4..6 ), and the space between the two bones (150 3 ).

[0185] Example 3: The computer-implemented method of example 1 or example 2, wherein the anatomical structure (130) comprises a plurality of bones, and wherein at least one of the two or more projection sub-regions is defined at least in part by:

[0186] X-ray projection image (150 1..6) in the boundary of a portion of at least one bone; and / or

[0187] X-ray projection image (150 1 , 150 2 ) between multiple bones; and / or

[0188] The space between two bones (150 3 ) and the intersection between another bone in the X-ray projection image (120).

[0189] Example 4: A computer-implemented method according to any one of Examples 1-3, wherein segmenting (S120) the X-ray projection image (120) to identify a plurality of projection sub-regions (150) of the anatomical structure (130) 1..m )include:

[0190] applying a segmentation algorithm to the received X-ray projection data (110); or

[0191] The received X-ray projection data (110) is input into a first neural network (NN1); and wherein the first neural network (NN1) is trained to segment X-ray projection images representing anatomical structures (130).

[0192] Example 5: A computer-implemented method according to Example 1, wherein the posture metric represents a posture (P) of the projection X-ray imaging system (140) with respect to the anatomical structure (130), and / or wherein the posture metric represents the suitability of the projection X-ray imaging system posture (P) for acquiring an X-ray projection image (120), and / or wherein the posture metric represents feedback for adjusting the projection X-ray imaging system posture (P) in order to acquire subsequent X-ray projection images representing the anatomical structure (130).

[0193] Example 6: The computer-implemented method of any previous example, wherein the pose metric represents suitability of a projection X-ray imaging system pose (P) for acquiring an X-ray projection image (120); and

[0194] The generating (S130) of the posture metric for the X-ray projection image (120) includes converting two or more projection sub-areas (150 1..m ) is compared with at least one threshold.

[0195] Example 7: A computer-implemented method according to any of the preceding examples, wherein the method further comprises:

[0196] Identify one or more anatomical landmarks (160 1..n ) in the X-ray projection image (120); and

[0197] The generating (S130) of the pose metric for the X-ray projection image (120) is further based on one or more anatomical landmarks (160 1..n )’s identified location.

[0198] Example 8: The computer-implemented method of Example 7, wherein the method comprises identifying a plurality of anatomical landmarks (160 1..n ) in the X-ray projection image (120); and wherein the method further comprises:

[0199] Measurement of multiple anatomical landmarks (160 1..n ) between one or more distances (170 i..1 ); and / or

[0200] measuring angles of one or more trajectories defined by a plurality of anatomical landmarks; and / or

[0201] Multiple anatomical landmarks (160 1..n ) are mapped to a plurality of corresponding landmarks in a reference image;

[0202] and wherein generating (S130) a pose metric for the X-ray projection image (120) is further based on the measured one or more distances (170 1..i ) and / or one or more trajectories of measured angles and / or multiple anatomical landmarks (160 1..n ) with respect to the location of the mapping of the corresponding landmark in the reference image.

[0203] Example 9: The computer-implemented method according to any previous example, wherein generating (S130) a pose metric for the X-ray projection image (120) comprises scaling the X-ray projection image (120); and

[0204] The projection sub-region (150) is determined from the scaled X-ray projection image. 1..m ) in a graph.

[0205] Example 10: The computer-implemented method of Example 9, wherein scaling the X-ray projection image comprises:

[0206] scaling a region of the x-ray projection image based on a measurement of the region in the x-ray projection image (120); or

[0207] A region of the X-ray projection image (120) is scaled based on the measured value of the distance in the X-ray projection image (120).

[0208] Example 11: The computer-implemented method of Example 1, wherein generating (S130) a pose metric for the X-ray projection image (120) comprises:

[0209] inputting segmented image data representing the segmented X-ray projection image into a second neural network (NN2); and

[0210] generating the pose metric using the second neural network (NN2) in response to the input; and

[0211] A second neural network (NN2) is trained using X-ray projection image training data to generate pose metrics from the segmented image data, the X-ray projection image training data comprising a plurality of segmented X-ray projection images (130) representing anatomical structures and corresponding ground truth values ​​for the pose metrics, the ground truth values ​​for the pose metrics being estimated based on relative sizes of two or more of the projection sub-regions in the segmented X-ray projection images.

[0212] Example 12: A computer-implemented method according to Example 11, wherein the pose metric represents the suitability of the projection X-ray imaging system pose (P) for acquiring the X-ray projection images, and wherein the second neural network (NN2) is trained to generate the pose metric from the segmented X-ray projection images by:

[0213] receiving X-ray projection image training data;

[0214] Inputting the X-ray projection image training data into the second neural network NN2; and

[0215] For each of the plurality of segmented X-ray projection images:

[0216] Predicting the value of the pose metric using a second neural network (NN2);

[0217] adjusting parameters of a second neural network (NN2) based on a difference between the predicted value of the pose metric and the ground truth value of the pose metric; and

[0218] The predicting and the adjusting are repeated until a stopping criterion is met.

[0219] Example 13: A computer-implemented method according to Example 11, wherein the pose metric represents feedback (130) for adjusting a pose (P) of a projection X-ray imaging system so as to acquire subsequent X-ray projection images representing the anatomical structure, and wherein the ground truth of the pose metric includes one or more pose adjustments for adjusting the pose of the projection X-ray imaging system so as to acquire the X-ray projection image representing the anatomical structure using a target pose with respect to the anatomical structure; and

[0220] The second neural network is trained to generate pose metrics from the segmented X-ray projection images by:

[0221] receiving X-ray projection image training data;

[0222] Inputting the X-ray projection image training data into the second neural network; and for each of a plurality of segmented X-ray projection images:

[0223] predicting, using the second neural network, values ​​of pose metrics, the values ​​of pose metrics comprising one or more pose adjustments for adjusting a pose of the projection X-ray imaging system so as to acquire an X-ray projection image representing the anatomical structure using a target pose with respect to the anatomical structure;

[0224] adjusting parameters of the second neural network based on a difference between the predicted value of the pose metric and the ground truth value of the pose metric; and

[0225] The predicting and the adjusting are repeated until a stopping criterion is met.

[0226] Example 14: The computer-implemented method of Example 1, wherein the pose metric represents suitability of the X-ray imaging system pose (P) for acquiring the X-ray projection image (120), and wherein the method further comprises:

[0227] Based on the gap (150) between the two bones in the X-ray projection image (120) 3 ) to calculate a value of a second posture metric for the X-ray projection image (120), the value of the second posture metric representing suitability of the X-ray imaging system posture (P) for acquiring the X-ray projection image (120);

[0228] And wherein generating (S130) a pose metric for the X-ray projection image (120) is further based on a value of a second pose metric.

[0229] Example 15: The computer-implemented method of Example 14, wherein calculating the value of the second posture metric comprises:

[0230] inputting the X-ray projection data representing the X-ray projection image (120) or the segmented X-ray projection image into a third neural network (NN3); and

[0231] generating, in response to the input, a value of a second posture metric using a third neural network (NN3); and

[0232] A third neural network (NN3) is trained using X-ray projection image training data to generate values ​​of a second posture metric from the X-ray projection data, the X-ray projection image training data comprising a plurality of X-ray projection images or segmented X-ray projection images representing an anatomical structure (130), and corresponding ground truth values ​​for the second posture metric, the ground truth values ​​for the second posture metric being estimated based on a size of a gap between two bones in the X-ray projection images or segmented X-ray projection images.

[0233] The above examples should be understood to illustrate the present disclosure, rather than to be restrictive. Other examples are also contemplated. For example, the examples described about the computer-implemented method may also be provided in a corresponding manner by a computer program product or by a computer-readable storage medium or by an X-ray imaging system. It should be understood that the features described about any one example may be used alone, or in combination with the features described with other features, and may be used in combination with one or more features of another example, or in combination with a combination of other examples. In addition, equivalents and modifications not described above may also be adopted without departing from the scope of the present invention defined in the appended claims. In the claims, the word "including" does not exclude other elements or operations, and the indefinite article "one" or "an" does not exclude multiple. The fact that certain features are recorded in mutually different dependent claims does not indicate that the combination of these features cannot be used advantageously. Any figure mark in the claims should not be interpreted as limiting its scope.

Claims

1. A computer-implemented method for providing posture information for an X-ray projection image, the method comprising: Receiving (S110) X-ray projection data (110), the X-ray projection data comprising an X-ray projection image (120) representing an anatomical structure (130) comprising a plurality of bones, the X-ray projection data being acquired by a projection X-ray imaging system (140), the projection X-ray imaging system having a corresponding pose (P) with respect to the anatomical structure (130); Segmenting (S120) the X-ray projection image (120) to identify a plurality of projection sub-regions (150) of the anatomical structure (130) 1..m ); Based on the projection sub-region (150) in the segmented X-ray projection image (120), 1..m ) to generate (S130) a posture metric for the X-ray projection image (120) based on the relative sizes of two or more projection sub-regions in the projection image (120), wherein the posture metric represents: the posture (P) of the projection X-ray imaging system (140) with respect to the anatomical structure (130) and / or the suitability of the projection X-ray imaging system posture (P) for acquiring the X-ray projection image (120) and / or feedback for adjusting the projection X-ray imaging system posture (P) so as to acquire subsequent X-ray projection images representing the anatomical structure (130); and Wherein, the projection sub-area (150 1..m ) represents a gap (1503) between two bones in the skeleton; and / or Wherein, the projection sub-area (150 1..m ) is defined at least in part by: intersection points between the plurality of bones (1501, 1502) in the X-ray projection image; and / or an intersection point between a gap (1503) between two bones and another bone in the X-ray projection image (120); and The pose metric is output (S140) to provide the pose information for the X-ray projection image (120).

2. The computer-implemented method of claim 1 , wherein: The projection sub-area (150 1..m ) represents a portion of the bone (150 1..2 , 150 4..6 ).

3. A computer-implemented method according to claim 1 or claim 2, wherein: At least one of the projection sub-areas is defined at least in part by: A portion of at least one of the bones in the X-ray projection image (150 1..6 )’s boundary.

4. The computer-implemented method according to any one of claims 1 to 3, wherein: The segmentation (S120) of the X-ray projection image (120) to identify a plurality of projection sub-regions (150) of the anatomical structure (130) 1..m )include: applying a segmentation algorithm to the received X-ray projection data (110); or The received X-ray projection data (110) is input into a first neural network (NN1); and wherein the first neural network (NN1) is trained to segment the X-ray projection image representing the anatomical structure (130).

5. A computer-implemented method according to any preceding claim, wherein: The pose metric represents the suitability of the projection X-ray imaging system pose (P) for acquiring the X-ray projection image (120); and The generating (S130) of the posture metric for the X-ray projection image (120) comprises: 1..m ) is compared with at least one threshold.

6. A computer-implemented method according to any preceding claim, wherein: The method further comprises: Identify one or more anatomical landmarks (160 1..n ) in the X-ray projection image (120); and The generating (S130) of the posture metric for the X-ray projection image (120) is further based on the identified one or more anatomical landmarks (160 1..n )’s location.

7. The computer-implemented method of claim 6, wherein: The method includes identifying a plurality of anatomical landmarks (160 1..n ) in the X-ray projection image (120); and wherein the method further comprises: Measuring the plurality of anatomical landmarks (160 1..n ) between one or more distances (170 i..1 ); and / or measuring angles of one or more trajectories defined by the plurality of anatomical landmarks; and / or The plurality of anatomical landmarks (160 1..n ) is mapped to a plurality of corresponding landmarks in a reference image; And wherein the generating (S130) of the posture metric for the X-ray projection image (120) is further based on the measured one or more distances (170 1..i ) and / or the angle of the one or more trajectories and / or the plurality of anatomical landmarks (160 1..n ) are respectively related to the mapping positions of the corresponding marks in the reference image.

8. A computer-implemented method according to any preceding claim, wherein: The generating (S130) of the pose metric for the X-ray projection image (120) includes scaling the X-ray projection image (120); and The projection sub-area (150 1..m ) in a graph.

9. The computer-implemented method of claim 8, wherein: The step of scaling the X-ray projection image comprises: scaling a region of the X-ray projection image (120) based on a measurement value of the region in the X-ray projection image (120); or scaling a region of the X-ray projection image (120) based on a measurement of a distance in the X-ray projection image (120); or The anatomical structure (130) in the X-ray projection image (120) is registered to an anatomical atlas image representing the anatomical structure (130) to provide a scaling factor for the X-ray projection image, and a region of the X-ray projection image (120) is scaled using the scaling factor.

10. The computer-implemented method of claim 1, wherein: The generating (S130) of the posture metric for the X-ray projection image (120) comprises: inputting segmented image data representing the segmented X-ray projection image into a second neural network (NN2); and generating the pose metric using the second neural network (NN2) in response to the input; and Wherein, the second neural network (NN2) is trained using X-ray projection image training data to generate the posture metric from the segmented image data, the X-ray projection image training data comprising a plurality of segmented X-ray projection images representing the anatomical structure (130) and corresponding ground truth values ​​for the posture metric, the ground truth values ​​for the posture metric being estimated based on relative sizes of two or more of the projection sub-regions in the segmented X-ray projection images.

11. The computer-implemented method of claim 10, wherein: The pose metric represents the suitability of the projection X-ray imaging system pose (P) for acquiring the X-ray projection images, and wherein the second neural network (NN2) is trained to generate the pose metric from the segmented X-ray projection images by: receiving the X-ray projection image training data; inputting the X-ray projection image training data into the second neural network (NN2); and For each of the plurality of segmented X-ray projection images: predicting a value of the posture metric using the second neural network (NN2); adjusting parameters of the second neural network (NN2) based on a difference between the predicted value of the pose metric and the ground truth value of the pose metric; and The predicting and the adjusting are repeated until a stopping criterion is met.

12. The computer-implemented method of claim 10, wherein: The pose metric represents feedback for adjusting a pose (P) of the projection X-ray imaging system in order to acquire subsequent X-ray projection images representing the anatomical structure (130), and wherein the ground truth of the pose metric comprises one or more pose adjustments for adjusting the pose of the projection X-ray imaging system in order to acquire an X-ray projection image representing the anatomical structure using a target pose with respect to the anatomical structure; and wherein the second neural network is trained to generate the pose metric from the segmented X-ray projection image in the following manner: receiving the X-ray projection image training data; Inputting the X-ray projection image training data into the second neural network; and for each of the plurality of segmented X-ray projection images: predicting, using the second neural network, a value of the pose metric, the value of the pose metric comprising one or more pose adjustments for adjusting the pose of the projection X-ray imaging system so as to acquire an X-ray projection image representing the anatomical structure using a target pose with respect to the anatomical structure; adjusting parameters of the second neural network based on a difference between a predicted value of the pose metric and the ground truth value of the pose metric; and The predicting and the adjusting are repeated until a stopping criterion is met.

13. The computer-implemented method of claim 1, wherein: The pose metric represents suitability of the X-ray imaging system pose (P) for acquiring the X-ray projection image (120), and wherein the method further comprises: Calculating a value of a second posture metric for the X-ray projection image (120) based on a size of a gap (1503) between two bones in the X-ray projection image (120), the value of the second posture metric representing suitability of the X-ray imaging system posture (P) for acquiring the X-ray projection image (120); And wherein, the generating (S130) of the posture metric for the X-ray projection image (120) is further based on the value of the second posture metric.

14. The computer-implemented method of claim 13, wherein: The calculating of the value of the second posture metric comprises: inputting the X-ray projection data representing the X-ray projection image (120) or the segmented X-ray projection image into a third neural network (NN3); and generating, in response to the input, a value of the second posture metric using the third neural network (NN3); and Wherein, the third neural network (NN3) is trained using X-ray projection image training data to generate the value of the second posture metric from the X-ray projection data, the X-ray projection image training data comprising a plurality of X-ray projection images or segmented X-ray projection images representing the anatomical structure (130), and a corresponding ground truth value for the second posture metric, the ground truth value for the second posture metric being estimated based on the size of a gap between two bones in the X-ray projection image or the segmented X-ray projection image.

15. A computer-implemented method for providing pose information for an X-ray projection image, the method comprising: Receiving (S110) X-ray projection data (110), the X-ray projection data comprising an X-ray projection image (120) representing an anatomical structure (130), the X-ray projection data being acquired by a projection X-ray imaging system (140), the projection X-ray imaging system having a corresponding posture (P) with respect to the anatomical structure (130); Segmenting (S120) the X-ray projection image (120) to identify a plurality of projection sub-regions (150) of the anatomical structure (130) 1..m ); Generating (S130) a posture metric for the X-ray projection image (120), wherein the value of the posture metric represents the suitability of the posture (P) of the X-ray imaging system for acquiring the X-ray projection image (120), and wherein the value of the posture metric is based on the projection sub-area (150 1..m ), comprising: the size of the gap (1503) between the two bones in the X-ray projection image (120), and / or a size of a projection sub-region defined at least in part by intersections between a plurality of bones in the x-ray projection images (1501, 1502); and / or a size of a projection sub-region defined at least in part by an intersection between a gap (1503) between two bones and another bone in the X-ray projection image (120); and The pose metric is output (S140) to provide the pose information for the X-ray projection image (120).

16. The computer-implemented method of claim 15, wherein: The generating (S130) of the pose metric for the X-ray projection image (120) is performed using a normalized value of the size of the gap (1503), and wherein the normalized value of the size of the gap (1503) is calculated by: i) scaling the size of the gap (1503) or the size of the projection sub-area based on the size of a feature in the X-ray projection image, or ii) aligning the anatomical structure in the X-ray projection image (120) to an anatomical atlas image representing the anatomical structure (130) to provide a scaling factor for the X-ray projection image, scaling a region of the X-ray projection image (120) using the scaling factor to provide a scaled X-ray projection image, and measuring the size of the gap (1503) or the size of the projection sub-region in the scaled X-ray projection image.

Images