How to get a mobile x-ray of your patient

Abstract

The present invention relates to a method, system, software module, and use of markers for mobile x-ray acquisition of a patient, comprising the steps of detecting a marker in an x-ray image using deep learning methods, the marker having a plate and a rod of known geometry, determining the position of the marker's rod, analyzing the projection of the rod in the x-ray image, and determining the position of an x-ray source above a patient bed based on the analyzed projection of the rod.

Description

[Technical Field]

[0001] The present invention relates to the field of X-ray imaging of patients, and more particularly to a method for mobile X-ray acquisition of a patient, a system for mobile X-ray acquisition, a software module for mobile X-ray acquisition, and the use of markers for use in the method and / or system. [Background technology]

[0002] Mobile X-ray systems are often used in emergency departments and intensive care units because they can be the only imaging modality for patients at the bedside. The quality of mobile X-ray images is typically lower than that of stationary X-ray systems. Furthermore, the relative position of the emitter and detector is flexible in mobile X-ray systems to accommodate life support equipment. In X-ray images acquired by mobile X-ray devices, objects visible in the X-ray image may appear different depending on how the X-ray source is positioned relative to the patient. For example, if a technician is required to position the X-ray source over the patient's head, the object will have a different silhouette than if the X-ray source were positioned over the patient's abdomen. These distortions complicate physical measurements. Summary of the Invention [Problem to be solved by the invention]

[0003] Therefore, there is a need to optimize mobile X-ray acquisition of patients, and more specifically to reduce projection distortions caused by different positions of the X-ray source on the patient, and to be able to perform physical measurements on the X-ray images. [Means for solving the problem]

[0004] It is an object of the present invention to provide improved methods, systems, and software modules and markers for use in mobile X-ray acquisition, which may in particular reduce projection distortion to enable physical measurements to be performed on X-ray images.

[0005] The object of the present invention is solved by the subject matter of the independent claims, further embodiments are incorporated in the dependent claims.

[0006] It should be noted that features, functions, and / or elements described below with reference to methods apply equally to systems and / or software modules, and vice versa. Thus, any feature, function, step, and / or element described below with reference to one aspect of the disclosure applies equally to any other aspect of the disclosure.

[0007] According to a first aspect of the present invention, a method for evaluating a mobile X-ray acquisition of a patient is described. The method may include receiving an X-ray image and detecting a marker in the X-ray image using a deep learning method, the marker comprising a plate and a rod of known geometry. The method may further include determining a position of the rod of the marker, analyzing a projection of the rod in the X-ray image, and determining a position of an X-ray source above a patient bed based on the analyzed projection of the rod.

[0008] In the context of the present invention, the term "mobile X-ray" should be understood to describe an X-ray device (system) often used in emergency departments and / or intensive care units, which may be used to X-ray a patient at the bedside. The positions of the emitter, X-ray source, and X-ray detector are flexible and therefore not fixed relative to one another.

[0009] In the context of the present invention, the term "known geometry" should be understood to describe that the geometry, and therefore the width, length, height, etc., of the marker is known for the method. It is not important what the exact geometry is, meaning that it is not important whether the marker is 10 cm or 20 cm tall. Instead, it is not important whether the marker has a particular geometry, but only that the geometry is known and provided.

[0010] In other words, a method for inferring the position of an X-ray source relative to a patient by analyzing the projection distortion of a marker of known geometry is described. The silhouette, and therefore the projection, of the marker on the X-ray image can be located using deep learning-based detection methods. Furthermore, knowledge of the X-ray source's position relative to the patient can be used to recalculate the relative positions of important image structures. The problem of projection distortion in mobile X-ray images is solved. Thus, accurate physical measurements on mobile X-ray images can be enabled, such as measurements from the tip of an endotracheal tube inserted into a patient to the patient's tracheal carina and / or measurements from the tip of a central venous catheter to the superior vena cava region. This method can also be useful for measuring the size of pulmonary nodules, cardiothoracic index, etc.

[0011] In particular, methods, systems, software modules, and the use of at least one marker are described for the evaluation of mobile X-ray images and their refinement to enable the determination of the X-ray source above the patient, which allows reducing the effects of projection distortion and allows the relative positions of important image structures to be recalculated.

[0012] In chest X-ray acquisitions, it is standard procedure to place markers indicating the left or right side of the patient. The markers used are the letters "L" and "R" for each side of the patient. The marker geometry can be important for strain assessment. Therefore, the marker geometry needs to be known; for example, standardized markers with standardized dimensions may be used. The standardized markers used are described below in further embodiments of this specification.

[0013] This method (and the use of the system, software module, and markers) allows for overcoming the problem of projection distortion by adding specific physical markers of known geometry to the patient for X-ray imaging, and these markers can be visible in the X-ray image. These markers that are visible in the X-ray image can be used to calculate distortion errors. The resulting errors can be used to estimate corrections to physical measurements for some application tasks, such as X-ray imaging.

[0014] The processing unit may be configured to perform the above-mentioned method steps, where the processing unit may be part of a computer. The computer may be part of the X-ray imaging system or may be an external computer that receives all information, in particular the X-ray images and instructions for performing the methods as described by the different embodiments herein. The processing unit may be configured to perform the method steps one by one, or to perform only some of them simultaneously, or to perform all of them simultaneously.

[0015] According to an exemplary embodiment of the present invention, analyzing the projection of the rod may include at least one of analyzing the projection of the rod along the image X-axis, analyzing the projection of the rod along the image Y-axis, and analyzing the projection of the rod along the rod axis. Analyzing the projection of the rod may include one or more of the above steps, or any combination thereof. In particular, analyzing the projection of the rod may include analysis along the image axis to obtain horizontal displacements on both axes, which can be used to determine the position of the X-ray source above the patient's bed.

[0016] According to an exemplary embodiment of the present invention, the step of analyzing the projection of the rod along the image X-axis may include measuring the projection of the rod in the X-ray image and / or determining the radius and height of the rod from the known geometry of the marker. Further details for analyzing the projection of the rod are described in FIGS. 5 to 7. The known geometry of the marker may be stored in a memory of a processing unit used to perform the method. Thus, when using standardized markers, the dimensions of the marker and rod can be provided from memory. Alternatively, the geometry and dimensions of the marker and rod may be provided by a user for each X-ray acquisition. Therefore, it is not important from where the dimensions and shape of the marker and rod are received, they only need to be known to the method.

[0017] According to an exemplary embodiment of the present invention, the step of analyzing the projection of the rod along the image Y axis may include measuring the projection of the rod in the X-ray image and / or determining the radius and height of the rod from the known geometry of the marker.

[0018] According to an exemplary embodiment of the present invention, the step of analyzing the projection of the rod along its axis may include determining a smaller rod radius at the base of the rod and a larger rod radius at the tip of the rod. It may also be possible to use diameters instead of radii, thus using smaller and larger diameters at the base and tip of the rod, respectively. The determination of the smaller and larger rod radii may be performed in the projection of the rod along the rod direction, rather than along the image x- or y-axis.

[0019] According to an exemplary embodiment of the present invention, the method may further comprise determining the distance from the rod to a line passing through the X-ray source perpendicular to the patient's bed and / or determining the distance between the patient's bed and the detector. The distance determination may be performed using formulas such as those described in the embodiments of Figures 5 to 7.

[0020] According to an exemplary embodiment of the present invention, the projection of the rod, the smaller radius of the rod, and the larger radius of the rod may be measured by pixel measurements on the X-ray image and converted to physical dimensions using the resolution stored in the DICOM tag.

[0021] According to an exemplary embodiment of the present invention, the step of detecting the marker may include detecting letters arranged on the marker. In particular, the step may include detecting at least two letters arranged on the marker. Preferably, the step may include detecting all letters, for example, four letters, arranged on the marker. The more letters detected, the better the determination of the rod's location can be performed based on letter detection. For example, in one embodiment, the marker includes four letters arranged in a square pattern on one side of the marker plate. Thus, the letters do not naturally occur in the body, and therefore letter detection is proposed to find the marker. This provides the following advantages: If the letters are hidden (or two letters, in which only two of the four letters can be detected), the marker location can still be inferred using the remaining three or two letters. If the patient may have an orthopedic implant, the implant will not be mistaken for a marker rod because there are no metal letters around the implant.

[0022] According to an exemplary embodiment of the present invention, the marker may include four letters arranged on one side of the marker plate, and wherein detecting the marker may include detecting at least two letters of the marker to determine the position of the rod. If all letters are detected, the position of the rod may be determined. For example, if all four letters are detected, the position of the rod may be determined by the following equation (1):

number

[0023] Rx, Ry are the base positions of the rods at the center of the marker plate, and Lxi and Lyi are the x and y coordinates of the letters in the X-ray image. If one or two letters are hidden and therefore cannot be detected, additional rules must be added to evaluate the positions of the letters within the square and infer the rod base positions. For example, if three letters are detected and one letter is hidden, the three letters form a triangle. The longest distance between the letters must be determined, and that longest distance forms the hypotenuse of the triangle. The base position of the rod can be found in the middle of the hypotenuse of the triangle formed by the letters.

[0024] According to an exemplary embodiment of the present invention, a deep learning method may use a deep learning network trained on a dataset of characters on x-rays, where the dataset is trained with characters of different sizes and rotations. In particular, the dataset may be trained with characters from the entire alphabet, and thus the characters on the markers are not limited to specific characters. Nevertheless, preferred characters may be "L" to indicate the patient's left side in the x-ray image and "R" to indicate the patient's right side. Furthermore, the dataset may be trained with characters as uppercase and lowercase. The deep learning method may also be trained with rotated characters so that all alignments of characters in x-ray images can be detected. Thus, the deep learning method for marker detection alleviates the need for technicians to align the markers with the major axes of the image.

[0025] According to an exemplary embodiment of the present invention, the method may further include calculating the relative positions of image structures based on the determined position of the X-ray source relative to the patient. For example, a simple trigonometric relationship can be used to determine the actual relative positions of important image structures. For example, the trachea and endotracheal tube may be segmented, and a projection of the distance between the endotracheal tube tip and the carina can be measured on the X-ray. A statistical atlas can then be used to obtain the typical distance between the trachea and the patient bed (patient's back), and the actual physical distance between the endotracheal tube tip and the carina can be recalculated, and therefore whether endotracheal tube placement can be confirmed.

[0026] According to an exemplary embodiment of the present invention, the marker plate may have a square shape, the rod may be positioned in the center of the plate, and at least two letters may be positioned on the marker plate, with the letters being equal and each letter being positioned on or within the edge of the square-shaped plate. The marker may consist of a square plate or a named base, which may be made of plastic. Furthermore, the marker may have four inlaid flat letters that identify the marker. The letters may be made of metal. The plate may be made of plastic to minimize visibility on received X-ray images. For example, the dimensions of the plate may be 80 mm wide, 80 mm long, and 3 mm high. The marker rod may be positioned in the center of the marker plate, and the rod may be made of metal and have a cylindrical shape with a known height and diameter, e.g., 80 mm long and 2.5 mm in diameter. Furthermore, a space may be provided within the plate that can be used to receive the rod, which can be clipped by an elastic element, thus facilitating transportation to the patient bed. The rod may be attached to the plate by a plastic thread. Furthermore, the plastic portion of the marker may be realized with nylon. The nylon fiber structure may present a structure that is harmless to the patient when placed near the patient. Additional material properties may be present for the plastic parts, radiolucent plastic may be used for the metal parts, and radiopaque materials such as stainless steel and / or titanium, aluminum, or tungsten may be used. The exact marker dimensions may not be critical but may depend on the ease of handling of the marker. Therefore, they should be large enough to be carried in one hand.

[0027] According to an exemplary embodiment of the present invention, the step of determining markers using deep learning methods for object detection may include, inter alia, Faster RCNN or YOLO. Characters may be detected on X-ray images using deep learning techniques, for example, as YOLOv4.

[0028] According to a second aspect of the present invention, a system for evaluating mobile X-ray acquisition is described using the method according to any one of the above-mentioned embodiments. The system may have an X-ray source configured to be positioned above a patient and a detector configured to detect X-ray radiation emitted from the X-ray source through the patient. All embodiments described with this method may also be applied to a system for mobile X-ray measurement. The X-ray source may be positioned anywhere above the patient and is therefore preferably a mobile X-ray system. The patient may be positioned between the detector and the X-ray source.

[0029] According to a third aspect of the present invention, a software module is described that can cause a computer system to execute a method for evaluating mobile X-ray acquisition, the software module causing the computer to use deep learning methods to detect markers in an X-ray image, where the markers have plates and rods of known geometry, determine the position of the marker's rods, and analyze the projections of the rods in the X-ray image to determine the position of an X-ray source above a patient bed based on the analyzed projections of the rods. The X-ray images may be received by the computer. Alternatively, the X-ray images may already be contained in the computer's memory, or the X-ray images may be provided to the software module from the cloud or directly from a mobile X-ray system.

[0030] The software module may be part of a computer and / or computer program, but may also be an entire program on its own, for example, the software module may be used to update an already existing computer program to achieve the present invention.

[0031] The software modules may be stored on a computer readable medium, which may be considered as a storage medium, such as for example a USB stick, a CD, a DVD, a data storage device, a hard disk, or any other medium on which such software modules can be stored.

[0032] According to a fourth aspect of the present invention, the use of markers for evaluating mobile X-ray acquisition is described, which uses deep learning methods to detect markers in X-ray images, where the markers have plates and rods of known geometry, and analyzes the projections of the rods in the X-ray images to determine the position of the marker's rods and determine the position of the X-ray source above the patient bed based on the analyzed projections of the rods.

[0033] According to various embodiments of the present disclosure, the methods described herein may be implemented using a hardware computer system executing a software program. Furthermore, in exemplary, non-limiting embodiments, implementations may include distributed processing, component / object distributed processing, and parallel processing. A virtual computer system process may implement one or more of the methods or functions described herein, and the processors described herein may be used to support a virtual processing environment.

[0034] It should be noted that embodiments of the present invention are described with reference to different subject matters. In particular, some embodiments are described with reference to apparatus / system type claims, while other embodiments are described with reference to method type claims. However, those skilled in the art will infer from the above and following description that, unless otherwise indicated, any combination of features belonging to one type of subject matter, as well as any combination between features relating to different subject matters, in particular any combination between features of an apparatus type claim and a feature of a method type claim, is considered to be disclosed in the present application.

[0035] The above-defined aspects and further aspects of the present invention will be apparent from and will be explained with reference to the example embodiments described hereinafter. The present invention is explained in more detail below with reference to examples, to which the invention is not limited. [Brief explanation of the drawings]

[0036] [Figure 1]1 illustrates a flow diagram of a method according to an exemplary embodiment of the present invention. [Figure 2] 1 shows various projections of an object depending on the position of the X-ray source. [Figure 3] A single X-ray source is used to show different projections of different objects. [Figure 4] 1 illustrates a marker according to an exemplary embodiment of the present invention. [Figure 5] 10 illustrates an analysis of a projection of a rod according to an embodiment of the present invention. [Figure 6] 10 illustrates an analysis of the projection of a rod along the image axis according to an embodiment of the present invention. [Figure 7] 10 illustrates an analysis of the projection of a rod along the rod axis according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0037] The figures in the drawings are schematic. It should be noted that in different figures, similar or identical elements are provided with the same reference signs.

[0038] FIG. 1 shows a flow diagram with method steps according to one embodiment of the present invention. The method includes step S1 of detecting a marker in an X-ray image using a deep learning method, the marker comprising a plate and a rod of known geometry. The method further includes step S2 of determining the position of the marker's rod, step S3 of analyzing the projection of the rod in the X-ray image, and step S4 of determining the position of the X-ray source above the patient bed based on the analyzed projection of the rod. The method begins with marker detection S1, where an X-ray image of the patient is received before the marker detection begins, to enable the marker to be detected. This may be included in step S1. Step S3 of analyzing the projection of the rod may further include at least one of analyzing the projection of the rod along the image X-axis, analyzing the projection of the rod along the image Y-axis, or analyzing the projection of the rod along the rod's axis. These steps may be substeps of step S3 or may be included as additional method steps between steps S3 and S4. Furthermore, step S3 of analyzing the projection of the rod along the image X-axis may comprise measuring the projection of the rod in the X-ray image and / or determining the radius and height of the rod from the known geometry of the markers, which may also be substeps of step S3 or added as further steps of the method between steps S3 and S4, and the substep S3 may be further divided into determining a smaller rod radius at the base of the rod and a larger rod radius at the tip of the rod.

[0039] Additional steps that may be added between steps S3 and S4 may be determining the distance from the rod to a line passing through the X-ray source perpendicular to the patient's bed, and determining the distance between the patient's bed and the detector.

[0040] Furthermore, detecting the marker may include detecting characters disposed on the marker, which may include detecting at least two characters disposed on the marker. Determining the marker using a deep learning method for object detection may include Faster RCNN or YOLO, among others.

[0041] Additionally, the method may include the additional step of calculating the relative position of the image structures based on the determined position of the x-ray source relative to the patient.

[0042] FIG. 2 shows different projections of an object 107 depending on the position of the X-ray source 101. FIG. 2 illustrates how the relative position of an endotracheal tube 107 in the trachea appears different depending on how the X-ray source 101 is positioned relative to the patient 103. The patient 103 lies on the detector 102, and the X-ray source 101 must be positioned above the patient, where the position of the X-ray source 101 can change depending on the external environment. If the technician must position the X-ray source above the patient's head, as shown on the left side of FIG. 2, this results in a silhouette 106 that is significantly shorter than if the technician had to position the X-ray source above the patient's abdomen, as shown on the right side of FIG. 2. The resulting silhouette is indicated at 104 when the X-ray source is positioned above the abdomen. As can be seen in FIG. 2, depending on the position of the X-ray source 101, the resulting projection of the endotracheal tube can change. The respective resulting projections are indicated as projection 104, projection 105, and projection 106. These distortions caused by X-ray source placement complicate physical measurements. In the case of an endotracheal tube, the physical distance from the tracheal carina 108 to the endotracheal tube tip 107 is an indicator of the correct position of the tube within the patient 103. Therefore, there is a need for a mobile X-ray system to eliminate or at least reduce these distortions caused by different projections 104-106 due to the position of the X-ray source 101 on the patient 103. Using the methods described in the exemplary embodiments herein, this problem can be solved.

[0043] FIG. 3 shows different projections 213, 214 of different objects 211, 212 using one X-ray source 101. In particular, FIG. 3 shows that a low-profile object, such as object 212, has slight projection distortion 214 even when positioned further away from the X-ray source 101. Also, a larger object, such as object 211, has significant projection distortion 213 even when positioned closer to the X-ray source 101 than the low-profile object 212. Therefore, it is important to know the marker geometry used in the X-ray image for distortion assessment. Therefore, standardized markers with standardized dimensions can be used to facilitate the determination of the marker geometry. Such markers are shown in FIG. 4 below. Nevertheless, the method may also be possible using only known marker geometries, in a manner in which the marker dimensions must be provided to the system for the method prior to each method execution. This means that it is not essential that standardized markers with standardized dimensions be used; it is only necessary that the dimensions be known and provided to the method.

[0044] FIG. 4 illustrates a marker 300 according to an exemplary embodiment of the present invention. The marker 300 is composed of a square plastic plate 331 and a metal rod 332 to minimize the resulting X-ray visibility. The rod 332 is located on one side of the plate 331. The dimensions of the plate 331 can be set to 80 x 80 x 3 mm. The metal cylindrical rod 332 is located in the center of the plate 331 of the marker 300, and the rod 332 has a known height and diameter, for example, 80 x 2.5 mm. The rod 332 extends perpendicularly away from the plate 331. A space (not shown) can be created within the plate 331 to which the rod 332 can be clipped by an elastic element, thus facilitating transportation to the patient bed. In FIG. 3, the marker 300 includes four letters 333 of the letter "L." Other letters 333 of the alphabet can also be placed on the marker 300 in a similar manner. The letters 333 are placed on the side of the plate 331 opposite the side on which the rods 332 are placed on the plate. The dimensions of the marker 300 that must be known for this method are at least one of the following: marker height 334, marker length 335, marker plate 331 thickness 339, and letter lengths 336, 337 (e.g., for each letter, different letters will have the same dimensions). Additionally, of the rod base 338, at least one of the rod length 340, rod diameter 341, and rod base diameter 342 should be known. Each of the four letters 333 is placed at one corner of the square plate 331, the same distance to the edge of the plate 331.

[0045] It should be noted that the above dimensions are exemplary only and other dimensions may be possible.

[0046] FIG. 5 illustrates an analysis of the projection of a rod 451 according to an embodiment of the present invention. In particular, FIG. 5 illustrates the appearance of a marker on an X-ray image. The projection of the rod 551 can be decomposed onto two axes of the X-ray image: x (proj x) on the x-axis and y (proj y) on the y-axis. For easier understanding, all the following explanations are based on the x-axis projection; similar reasoning can be realized for the y-axis. The projection of the rod on the X-ray image is denoted 551, which can be divided into projections on two axes (x, y). Furthermore, FIG. 5 illustrates the projection of the rod along its direction 552. The method may use the projection of the rod 551 onto the x-axis and / or y-axis to determine the position of the rod and, therefore, the position of the X-ray source relative to the patient.

[0047] 6 shows an analysis of the projection of a rod along the image axis according to one embodiment of the present invention. In particular, FIG. 6 shows the X-axis projection, where p(proj x) is the measured length of the projection of the rod 551, r and h are the radius and height of the rod, respectively, d is the distance from the rod to a line perpendicular to the patient bed passing through the X-ray source 101, a is the position of the X-ray source 101 above the patient bed, and b is the distance between the patient bed and the detector 102. The projection of the rod 551 can be measured by pixel analysis, and therefore the pixel size and number of pixels measured. Using Thales' theorem leads to the following equation:

number

[0048] Figure 7 shows an analysis of the projection of a rod 551 along the rod axis in accordance with an embodiment of the present invention. In particular, Figure 7 analyzes the rod axis projection with r-low as the small rod radius at the base and r-high as the large rod radius at the tip. Due to the projection, the radius at the tip is not equal to the radius at the base of the rod. Again using Thales' theorem here yields the following equation:

number

[0049] Similarly, for larger radii r,

number

[0050] In equations (2), (3), and (4), the unknown variables are a, b, and d. The physical dimensions h and r are given by the dimensions of the rod, and therefore the rod structure p, r-low, and r-high are the physical lengths of the rod projection obtained from the x-ray image using pixel measurements on the x-ray image, which are converted to physical dimensions using the resolution stored in DICOM tag (0018,1050). Thus, equations (2), (3), and (4) form a system of equations that can be solved for a, b, and d.

[0051] If multiple markers are used, this allows several values ​​to be obtained for the position of the X-ray source and the distance between the patient bed and the detector, and the uncertainty introduced by pixel measurements of the rod projection can be averaged out.

[0052] The method described in the exemplary embodiments herein may be able to use an equation to determine the position of the X-ray source relative to the patient, and therefore may perform the step of calculating the above-mentioned equation.

[0053] While the present invention has been illustrated and described in detail in the drawings and the foregoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive. The present invention is not limited to the disclosed embodiments. Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims. It should be noted that the term "comprising" does not exclude other elements or steps, and "a" or "an" does not exclude a plurality. Also, elements described in association with different embodiments may be combined. It should also be noted that reference signs in the claims should not be construed as limiting the scope of the claims. [Explanation of symbols]

[0054] 101 X-ray source 102 detector 103 patients 104 Projection 105 Projection 106 Projection 107 tubes 108 Carina 211 Large Objects 212 Small Objects 213 Large projection distortion 214 Small projection distortion 300 markers 331 Plate 332 Rod 333 characters 334 height 335 length 336 length 337 length 338 Rod base 339 Marker Thickness 340 rod length 341 Rod diameter 342 Rod base diameter 551 Rod Projection 552 Projection along the rod direction

Claims

1. 1. A method for evaluating a mobile x-ray acquisition of a patient, comprising: receiving an x-ray image of the patient; detecting markers in the x-ray images using deep learning methods; the marker comprises plates and rods of known geometry; determining the position of the rod of the marker; analyzing the projection of the rod in the X-ray image; determining a position of the X-ray source above the patient bed based on the analyzed projections of the rods; A method having the following.

2. Analyzing the projection of the rods includes: analyzing the projection of the rod along the X axis of the image; analyzing the projection of the rod along the Y axis of the image; analyzing the projection of the rod along the axis of the rod; The method has at least one of the steps of The method of claim 1.

3. Analyzing the projection of the rod along the X axis of the image comprises: measuring the projection of the rod in the X-ray image; determining a radius and a height of the rod from the known geometry of the marker; having The method of claim 2.

4. Analyzing the projection of the rod along the axis of the rod includes: determining a smaller rod radius at a base of the rod; determining a larger rod radius at a tip of the rod; having The method of claim 2.

5. determining a distance from the rod to a line passing through the x-ray source and perpendicular to the patient bed; determining a distance between the patient's bed and a detector; The method of any one of claims 1 to 4, further comprising:

6. the projection of the rod, the smaller radius of the rod, and the larger radius of the rod are measured by pixel measurement on the x-ray image and converted to physical dimensions using a resolution stored in a DICOM tag; 6. The method according to any one of claims 1 to 5.

7. the step of detecting the marker includes detecting a character placed on the marker; 7. The method according to any one of claims 1 to 6.

8. the marker includes four letters disposed on the plate of the marker; the step of detecting the marker includes detecting at least two characters of the marker to determine the position of the rod; 8. The method according to any one of claims 1 to 7.

9. The deep learning method uses a deep learning network trained on a dataset of X-ray characters, the dataset of characters including characters of different sizes and rotations.

9. The method according to any one of claims 1 to 8.

10. calculating a relative position of an image structure based on the determined position of the x-ray source relative to the patient; 10. The method of claim 1, further comprising:

11. the plate of the marker has a square shape; The rod is disposed in the center of the plate; At least two characters are disposed on the plate of the marker; The letters are equal and each letter is located on one edge of the square-shaped plate.

11. The method according to any one of claims 1 to 10.

12. The step of determining the markers using a deep learning method for object detection includes, in particular, Faster RCNN or YOLO.

12. The method according to any one of claims 1 to 11.

13. 13. A system for evaluating mobile X-ray acquisitions using the method of any one of claims 1 to 12, comprising: an x-ray source configured to be positioned over a patient; a detector configured to detect x-ray radiation emitted from the x-ray source through the patient; A system having:

14. 1. A software module for causing a computer system to execute a method for evaluating mobile X-ray acquisition, the software module including: Deep learning methods are used to detect markers in X-ray images, wherein the marker comprises a plate and a rod of known geometry; determining the position of the rod on the marker; analyzing a projection of the rod within the x-ray image; determining a position of an X-ray source on a patient bed based on the analyzed projections of the rods; Software module.

15. 1. A method for using markers to evaluate mobile X-ray acquisitions, comprising: detecting a marker in an X-ray image using a deep learning method, the marker comprising a plate and a rod of known geometry; determining a position of the rod at the marker; and analyzing a projection of the rod in the X-ray image and determining a position of an X-ray source above a patient bed based on the analyzed projection of the rod.

Images