Determining Patient Chest Rotation in Medical Images

Abstract

A computer-implemented method for determining rotation of a patient's thorax in a medical image is provided, the method comprising receiving a medical image of the patient, processing the medical image to determine scapular space data related to the patient's scapula, and determining a rotation of the patient's thorax relative to at least one reference axis using the scapular space data.

Description

[Technical field]

[0001] The present invention relates to a method and system for determining the rotation of a patient's chest in a medical image. [Background technology]

[0002] Ensuring non-rotated positioning of the patient is an important quality aspect in posteroanterior (PA) chest radiography because otherwise the location of relevant anatomical structures in the image may appear misaligned in appearance or be obstructed from view. Summary of the Invention [Problem to be solved by the invention]

[0003] For example, the lung parenchyma to be analyzed may be shadowed by thoracic structures if not acquired at the correct angle. Conventional solutions involve locating landmarks, particularly the clavicle tips and spinous processes, to estimate the rotation angle and measuring the deviation from a symmetric constellation. In some cases (e.g., scoliosis), this is not a good indicator of global thoracic rotation. [Means for solving the problem]

[0004] To better address one or more of these concerns, in a first aspect of the present invention, a computer-implemented method for determining rotation of a patient's chest in a medical image is provided, the method comprising the steps of receiving a medical image of a patient, processing the medical image to determine scapular space data relating to the patient's scapula, and determining a rotation of the patient's chest relative to at least one reference axis using the scapular space data.

[0005] Thus, the approach to patient rotation detection proposed herein allows the rotation of the patient's chest (i.e., rib cage), and in particular the rotation of internal organs such as the heart and lungs, to be accurately reflected, even in severe cases such as scoliosis.

[0006] The method may further include processing the scapular space data to check the symmetry of the patient's scapula, i.e., the symmetry of the position of the scapula with respect to external rotation of the scapula, prior to determining the rotation. This may be performed as a form of "health check." Checking the symmetry of the patient's scapula may include identifying the acromion and coracoid process of the scapula and determining their lengths, calculating a ratio of the length of the coracoid process to the length of the acromion for each of the left and right scapula, and checking the symmetry of the scapula based on the calculated ratio. Checking the symmetry of the scapula based on the calculated ratio may include determining that the ratio of the left scapula matches the ratio of the right scapula within an acceptable range. In some examples, a similarity or distance metric may be used to compare the ratios and thereby check the symmetry or otherwise. Additionally or alternatively, checking the symmetry of the scapula may be based on one or more other parameters selected from the group consisting of the inferior angle of the scapula, the angle between the clavicle and the acromion, the angle between the lateral edge of the scapula and the humerus, and any combination thereof. Thus, in general terms, the method may include checking scapular symmetry based on comparing values ​​for the left and right scapula of one or more scapular position parameters, in particular parameters relating to the relative positions of bony structures within or around the scapula. Stated differently, processing the scapular space data to check scapular symmetry of the patient may include checking for the absence of scapular external rotation movement in both scapulae, or the absence of unequal scapular external rotation movement.

[0007] Processing the medical image to determine scapular space data may include segmenting the medical image to determine contours (i.e., boundaries or borders) of the patient's scapula. The method may include using a convolutional neural network trained on the annotated sample images to perform image segmentation. The method may further include, prior to segmenting the medical image, processing the image to locate one or more landmarks to facilitate image segmentation.

[0008] The patient rotation can be determined geometrically. Thus, determining the patient's thoracic rotation can include using the scapular space data to calculate a scapular line connecting corresponding points on the patient's scapula and determining the rotation using a displacement between the scapular line and at least one reference axis. In one example, the scapular line connects two points on the contour of each scapula, each point being the inferior corner of the border of the respective scapula, in other words the most inferior vertex relative to the patient. In other examples, other suitable anatomical reference points are selected relative to the scapular line. Determining the rotation can include determining the patient rotation in the image plane and / or determining the patient rotation out of the image plane. If the in-plane patient rotation is determined, the at least one reference axis includes the detector plane axis, the displacement includes an angular displacement between the scapular line and the detector plane axis, and determining the rotation includes determining the in-plane rotation of the patient using the angular displacement. The detector plane axis may be, for example, horizontal to the image or detector, but it will be understood that any orientation relative to the detector plane axis can be selected, including at least a vertical axis. Where out-of-plane patient rotation is determined, the at least one reference axis can, for example, include a medial axis of the patient's body, the displacement comprising a linear displacement between a midpoint of the scapular line and said medial axis, and determining the rotation comprising determining the out-of-plane rotation of the patient using said linear displacement.

[0009] In addition to or instead of a geometric determination of the patient's rotation, the patient's rotation (and / or scapular symmetry) may be determined empirically, for example using machine learning. Thus, the method can include using a trained machine learning model to determine the patient's thoracic rotation relative to at least one reference axis using the scapular space data.

[0010] Thus, in a second aspect, a method for training the machine learning model of the first aspect is provided. Training data for use in training the model can include medical images showing various degrees of patient rotation and appropriately labeled by a qualified radiographer. The training data can further include medical images of different age groups, gender groups, and different geographies to make the model more robust. The training data can also include such data that has been artificially modified to look similar to the original patient data or synthetic data.

[0011] As used herein, the term "scapula space data" refers to data that can be used to determine the position and / or shape of the scapula. In particular, scapula space data can include data indicating the location of the scapula contours resulting from, for example, a process of image segmentation.

[0012] The method of the first and / or second aspect may be computer implemented.

[0013] According to a third aspect, there is provided a computing system configured to perform the method of the first and / or second aspect.

[0014] According to a fourth aspect there is provided a computer program product comprising instructions which, when executed by a computing system, cause the computing system to perform the method of the first and / or second aspects.

[0015] According to a fifth aspect, there is provided a computer readable medium having instructions having instructions, which when executed by a computing system, cause the computing system to perform the method of the first and / or the aspects. The invention may include one or more aspects, examples or features, either alone or in combination, whether or not that combination or alone is specifically disclosed. Any feature or sub-aspect of one of the above aspects applies to any of the other aspects, as appropriate.

[0016] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter.

[0017] A detailed description will now be given, by way of example only, with reference to the accompanying drawings, in which: [Brief description of the drawings]

[0018] [Figure 1] FIG. 1 illustrates image segmentation to locate anatomical structures including a patient's scapula. [Diagram 2] Diagram showing determining the relative lengths of the acromion and coracoid processes of both scapulae to ensure symmetry of scapular position [Figure 3A] FIG. 13 illustrates the determination of a patient's in-plane and out-of-plane rotation using scapular space data. [Figure 3B] FIG. 13 illustrates the determination of a patient's in-plane and out-of-plane rotation using scapular space data. [Figure 4] FIG. 1 illustrates a computing system that can be used in accordance with the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] According to the present disclosure, rotation of a patient's chest in a medical image is determined by processing the medical image to determine scapular space data related to the patient's scapula and using the scapular space data to determine a rotation of the patient's chest relative to at least one reference axis.

[0020] FIG. 1 illustrates the use of image segmentation to locate anatomical structures, including a patient's scapula, in a medical image 100, thereby obtaining scapular spatial data. As shown, the result of image segmentation is the localization of the scapula and their contours 102. In addition, the contours of the lungs 104 and clavicle 106 may be located. The output of image segmentation is spatial data including at least scapular spatial data indicating the shape and location of the scapula. The image 100 can be segmented using a convolutional neural network trained on annotated sample images. To achieve better performance with a comparable amount of annotation effort, landmarks can be first located before determining the contours, as described, for example, in VON BERG, J. et al. Robust chest x-ray quality assessment using convolutional neural networks and atlas regularization. In: Medical Imaging 2020: Image Processing, International Society for Optics and Photonics. 2020. p.113131L.

[0021] Using the scapular space data, the rotation of the patient's thorax relative to at least one reference axis can be determined. Using the shape and positioning of the scapula and optionally the lung fields to assess the patient's rotation provides a better reflection of the rotation of the entire thorax. However, the scapula has more mobility compared to the clavicle and spine, allowing more freedom of internal rotation, external rotation, abduction, adduction, etc. Thus, the present disclosure contemplates performing an optional sanity check to ensure that the scapula is symmetrical with respect to the patient (i.e., no external rotation movement of the scapula) before using the scapular space data to determine the patient's rotation.

[0022] FIG. 2 shows the confirmation of scapular position symmetry. The acromion and coracoid process of the scapula are identified on both sides, and segmentation is used to identify their endpoints and determine their lengths. Shown in FIG. 2 are the length of the left acromion 202L, the length of the left coracoid process 204L, the length of the right acromion 202R, and the length of the right coracoid process 204R. Ratio, R L = Length of left coracoid process / length of left acromion and R R = Length of right coracoid process / length of right acromion is calculated. In the absence of scapular movement in the external rotation direction, the ratio R L is the right-hand side ratio R within a given accuracy. R It will be appreciated that other parameters may additionally or alternatively be utilized to assess scapular symmetry, i.e., the inferior angle of the scapula, the angle between the clavicle and the acromion, the angle between the lateral border of the scapula and the humerus, etc.

[0023] 3A and 3B show the determination of a patient's in-plane and out-of-plane rotation using scapular space data.

[0024] The out-of-plane rotation of the patient can be determined about a suitable reference axis (in this case the central axis of the patient's body, shown as line A in FIG. 3A). Such an axis of rotation can be robustly determined by registering a chest atlas to images annotated with such axes, similar to the technique described in VON BERG, J. et al. Robust chest x-ray quality assessment using convolutional neural networks and atlas regularization. In: Medical Imaging 2020: Image Processing, International Society for Optics and Photonics. Calculate the lowest point [P1, P2] on the inferior border of the scapula. The scapular line, shown as line B in FIG. 3A, connects points P1 and P2. The midpoint of line B is found, which is shown as point P3 in FIG. 3A. The intersection of line A and line B is found, which is shown as point P4. The distance between P3 and P4 corresponds to the component of the out-of-plane patient rotation. This is also shown in FIG. 3B, which shows the phenomenon with respect to the source 302 and detector 304 of the x-ray system 300. The more the patient rotates, the greater the linear displacement between points P3 and P4.

[0025] The in-plane rotation of the patient can be determined with respect to the detector plane axis. Figure 3A, for example, shows line C as the horizontal axis in the image or detector plane. The angular displacement between lines B and C corresponds to the in-plane rotation of the patient.

[0026] Such image information can be used to train a recurrent convolutional neural network to improve the accuracy of patient rotation determination. For example, the trained recurrent convolutional neural network can be used to estimate the rotation angle from the contour 102 represented by the scapular space data. This approach can follow the method described in KRONKE et al. CNN-based pose-estimation of musculoskeletal X-ray images. Philips ocupai conference. 2021, which has been successfully applied to a similar task of estimating ankle pose and flexion from a single radiographic image.

[0027] Numerous variations on the above-described systems and methods are contemplated by this disclosure. For example, the patient rotation determination methods disclosed herein can be used independently or in combination with existing methods that use the clavicle and spine. In another example, the scapular space data used by the systems and methods disclosed herein results from a process used to locate both scapulae and analyze them for shape and positioning for purposes of a separate "scapular checker." Lung fields may also be located for this purpose. In another example, the ratio R L and R R may be plotted. If the calculated value for the patient under examination appears on the nomogram, then it can be assumed that the scapula is symmetrical with respect to the patient. In yet another example, thoracic rotation is estimated based on redundant x-ray contours in addition to or instead of the scapular contours.

[0028] The systems and methods disclosed herein can be applied in the context of x-ray quality assessment.

[0029] 4 illustrates an exemplary computing system 800 that can be used in accordance with the systems and methods disclosed herein. The computing system 800 can form part of or include any desktop, laptop, server, or cloud-based computing system. The computing system 800 has at least one processor 802 that executes instructions stored in a memory 804. The instructions can be, for example, instructions for implementing functions described as being performed by one or more components described above, or instructions for implementing one or more of the methods described above. The processor 802 can access the memory via a system bus 806. In addition to storing executable instructions, the memory 804 can also store conversational inputs, scores assigned to the conversational inputs, and the like.

[0030] Computing system 800 further includes a data store 808 accessible by processor 802 via system bus 806. Data store 808 may include executable instructions, log data, etc. Computing system 800 also includes an input interface 810 that allows external devices to communicate with computing system 800. For example, input interface 810 may be used to receive instructions from an external computer device, a user, etc. Computing system 800 also includes an output interface 812 that interfaces computing system 800 with one or more external devices. For example, computing system 800 may display text, images, etc. via output interface 812.

[0031] It is contemplated that external devices communicating with computing system 800 via input interface 810 and output interface 812 can be included in the environment to provide virtually any type of user interface with which a user can interact. Examples of user interface types include graphical user interfaces, native user interfaces, and the like. For example, a graphical user interface can accept input from a user using an input device such as a keyboard, mouse, remote control, and provide output on an output device such as a display. Furthermore, a native user interface can enable a user to interact with computing system 800 in a manner that is not subject to the constraints imposed by input devices such as a keyboard, mouse, remote control, and the like. Rather, a natural user interface can rely on voice recognition, touch and stylus recognition, on-screen and adjacent-screen gesture recognition, air gestures, head and eye tracking, voice and speech, vision, touch, gestures, machine intelligence, and the like.

[0032] Further, although computing system 800 is shown as a single system, it should be understood that it may be a distributed system, such that, for example, several devices may communicate over network connections and collectively perform the tasks described as being performed by computing system 800.

[0033] Various functions described herein can be implemented in hardware, software, or any combination thereof. If implemented in software, the functions described above can be stored on or transmitted through a computer-readable medium as one or more instructions or codes. A computer-readable medium includes a computer-readable storage medium. A computer-readable storage medium can be any available storage medium that can be accessed by a computer. By way of example and not limitation, such computer-readable storage media can include flash storage media, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. Disk and disc, as used herein, include compact discs (CDs), laser discs, optical discs, digital versatile discs (DVDs), floppy disks, and Blu-ray discs (BDs), where a disk typically reproduces data magnetically and a disc typically reproduces data optically using a laser. Furthermore, propagated signals are not included within the scope of computer-readable storage media. Computer-readable media also includes communication media, including any medium that facilitates transfer of a computer program from one place to another. A connection may be, for example, a communication medium. For example, if software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included within the definition of communication media. Combinations of the above should also be included within the scope of computer-readable media.

[0034] Alternatively, or in addition, the functions described herein may be performed, at least in part, by one or more hardware logic components. By way of example and not limitation, exemplary types of hardware logic components that may be used include Field Programmable Gate Arrays (FPGAs), Program-specific Integrated Circuits (ASICs), Program-specific Standard Products (ASSPs), System-on-a-Chip (SOCs), Complex Programmable Logic Devices (CPLDs), and the like.

[0035] It will be appreciated that the circuits described above may have other functions in addition to those described above, and that these functions may be performed by the same circuitry.

[0036] The applicant discloses each individual feature described herein and any combination of two or more such features solely to the extent that such feature or combination can be implemented based on the specification as a whole in light of the common general knowledge of a person skilled in the art, regardless of whether such feature or combination solves any problem disclosed herein, and without being limited by the scope of the claims. The applicant indicates that aspects of the invention may consist of any such individual feature or combination of features.

[0037] It should be noted that embodiments of the present invention are described with reference to different categories. In particular, some examples are described with reference to methods, and other examples are described with reference to devices. However, a person skilled in the art will understand from the description that, unless otherwise specified, any combination of features belonging to one category, as well as any combination between features relating to different categories, is also considered to be disclosed by the present application. However, all features combined can provide a synergistic effect that is greater than the simple sum of the features.

[0038] While the invention has been illustrated and described in detail in the drawings and the foregoing description, such illustration and description are to be considered exemplary or explanatory and not restrictive. The invention is not limited to the disclosed embodiments. Other variations to the disclosed embodiments can be understood and effected by those skilled in the art, from a study of the drawings, the disclosure, and the appended claims.

[0039] The term "comprising" does not exclude other elements or steps.

[0040] The indefinite article "a" or "an" does not exclude a plurality. In addition, the articles "a" and "an" as used herein should generally be construed to mean "one or more" unless otherwise specified or unless it is clear from the context that a singular form is intended.

[0041] A single processor or other unit may fulfill the functions of several items recited in the claims.

[0042] The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0043] The computer program may be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state medium, supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless communication systems.

[0044] Any reference signs in the claims should not be construed as limiting the scope.

[0045] Unless otherwise stated or clear from context, as used herein, the phrases "one or more of A, B, and C," "at least one of A, B, and C," and "A, B, and / or C" are intended to mean all possible permutations of one or more of the listed items. That is, the phrase "X has A and / or B" is satisfied by any of the following instances: X has A; X has B; or X has both A and B.

Claims

1. 1. A computer-implemented method for determining the rotation of a patient's chest in a medical image, comprising: receiving the medical images of the patient; processing the medical images to determine scapular space data relating to the patient's scapula; using the scapular space data to determine a rotation of the patient's chest relative to at least one reference axis; and and performing a sanity check by processing the scapular spatial data to verify symmetry of the patient's scapula before determining the rotation. method.

2. the step of verifying scapular symmetry of the patient comprises: Identifying the acromions and coracoid processes of the left and right scapulae and determining their lengths; Calculating the ratio of the length of the coracoid to the length of the acromion for each of the left and right scapula; determining scapular symmetry based on the calculated ratio; 2. The method of claim 1, comprising:

3. 3. The method of claim 2, wherein the step of verifying scapular symmetry based on the calculated ratio comprises determining that the ratio of a left scapula matches the ratio of a right scapula within an acceptable range.

4. 2. The method of claim 1, wherein the step of verifying scapular symmetry is based on one or more parameters selected from the group consisting of the inferior angle of the scapula, the angle between the clavicle and the acromion, the angle between the lateral border of the scapula and the humerus, and any combination thereof.

5. The method of claim 4 , wherein the step of verifying scapular symmetry is based on comparing values ​​of one or more of the parameters for the left and right scapula.

6. The method of claim 1 , wherein the step of processing the medical image to determine the scapula space data comprises segmenting the medical image to determine a contour of the patient's scapula.

7. 7. The method of claim 6, further comprising using a convolutional neural network trained on annotated sample images to perform the segmentation.

8. 7. The method of claim 6, further comprising, prior to segmenting the medical image to determine the contour of the patient's scapula, processing the image to locate one or more landmarks to facilitate the segmentation.

9. The step of determining the rotation of the patient's chest comprises: using the scapular space data to calculate a scapular line connecting corresponding points on the patient's left and right scapula; determining the rotation using a displacement between the scapular line and the at least one reference axis; 2. The method of claim 1, comprising:

10. 10. The method of claim 9, wherein the scapular line is a line connecting two points on the contours of the left and right scapulae, each of the two points being a vertex on the contour of the inferior corner of each scapula.

11. 10. The method of claim 9, wherein the at least one reference axis includes a detector plane axis, the displacement includes an angular displacement between the scapular line and the detector plane axis, and the step of determining the rotation includes determining an in-plane rotation of the patient using the angular displacement.

12. 10. The method of claim 9, wherein the at least one reference axis comprises a central axis of the patient's body, the displacement comprises a linear displacement between a midpoint of the scapular line and the central axis, and the step of determining the rotation comprises determining an out-of-plane rotation of the patient using the linear displacement.

13. 2. The method of claim 1, wherein determining a rotation of the patient's thorax relative to the at least one reference axis using the scapular space data comprises determining the rotation using a trained machine learning model.

14. A computing system configured to perform the method of any one of claims 1 to 13.

15. A computer program comprising instructions which, when executed by a computing system, cause the computing system to carry out the method of any one of claims 1 to 13.

Images